3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define yychar PL_yychar
27 #define yylval PL_yylval
29 static char ident_too_long[] = "Identifier too long";
30 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
31 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
33 static void restore_rsfp(pTHX_ void *f);
34 #ifndef PERL_NO_UTF16_FILTER
35 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
36 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 #define XFAKEBRACK 128
42 #ifdef USE_UTF8_SCRIPTS
43 # define UTF (!IN_BYTES)
45 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
48 /* In variables named $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* On MacOS, respect nonbreaking spaces */
53 #ifdef MACOS_TRADITIONAL
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
56 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
59 /* LEX_* are values for PL_lex_state, the state of the lexer.
60 * They are arranged oddly so that the guard on the switch statement
61 * can get by with a single comparison (if the compiler is smart enough).
64 /* #define LEX_NOTPARSING 11 is done in perl.h. */
67 #define LEX_INTERPNORMAL 9
68 #define LEX_INTERPCASEMOD 8
69 #define LEX_INTERPPUSH 7
70 #define LEX_INTERPSTART 6
71 #define LEX_INTERPEND 5
72 #define LEX_INTERPENDMAYBE 4
73 #define LEX_INTERPCONCAT 3
74 #define LEX_INTERPCONST 2
75 #define LEX_FORMLINE 1
76 #define LEX_KNOWNEXT 0
84 # define YYMAXLEVEL 100
86 YYSTYPE* yylval_pointer[YYMAXLEVEL];
87 int* yychar_pointer[YYMAXLEVEL];
91 # define yylval (*yylval_pointer[yyactlevel])
92 # define yychar (*yychar_pointer[yyactlevel])
93 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
95 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 /* Note that REPORT() and REPORT2() will be expressions that supply
136 * their own trailing comma, not suitable for statements as such. */
137 #ifdef DEBUGGING /* Serve -DT. */
138 # define REPORT(x,retval) tokereport(x,s,(int)retval),
139 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
141 # define REPORT(x,retval)
142 # define REPORT2(x,retval)
145 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
146 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
147 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
148 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
149 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
150 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
151 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
152 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
153 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
154 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
155 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
156 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
157 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
158 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
159 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
160 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
161 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
162 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
163 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
164 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
166 /* This bit of chicanery makes a unary function followed by
167 * a parenthesis into a function with one argument, highest precedence.
169 #define UNI(f) return(yylval.ival = f, \
173 PL_last_uni = PL_oldbufptr, \
174 PL_last_lop_op = f, \
175 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
177 #define UNIBRACK(f) return(yylval.ival = f, \
180 PL_last_uni = PL_oldbufptr, \
181 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
183 /* grandfather return to old style */
184 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
189 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
192 SV* report = newSVpv(thing, 0);
193 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
196 if (s - PL_bufptr > 0)
197 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
199 if (PL_oldbufptr && *PL_oldbufptr)
200 sv_catpv(report, PL_tokenbuf);
202 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
211 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
212 * into an OP_ANDASSIGN or OP_ORASSIGN
216 S_ao(pTHX_ int toketype)
218 if (*PL_bufptr == '=') {
220 if (toketype == ANDAND)
221 yylval.ival = OP_ANDASSIGN;
222 else if (toketype == OROR)
223 yylval.ival = OP_ORASSIGN;
231 * When Perl expects an operator and finds something else, no_op
232 * prints the warning. It always prints "<something> found where
233 * operator expected. It prints "Missing semicolon on previous line?"
234 * if the surprise occurs at the start of the line. "do you need to
235 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
236 * where the compiler doesn't know if foo is a method call or a function.
237 * It prints "Missing operator before end of line" if there's nothing
238 * after the missing operator, or "... before <...>" if there is something
239 * after the missing operator.
243 S_no_op(pTHX_ char *what, char *s)
245 char *oldbp = PL_bufptr;
246 bool is_first = (PL_oldbufptr == PL_linestart);
252 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
254 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
255 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
257 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
258 if (t < PL_bufptr && isSPACE(*t))
259 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
260 t - PL_oldoldbufptr, PL_oldoldbufptr);
264 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
271 * Complain about missing quote/regexp/heredoc terminator.
272 * If it's called with (char *)NULL then it cauterizes the line buffer.
273 * If we're in a delimited string and the delimiter is a control
274 * character, it's reformatted into a two-char sequence like ^C.
279 S_missingterm(pTHX_ char *s)
284 char *nl = strrchr(s,'\n');
290 iscntrl(PL_multi_close)
292 PL_multi_close < 32 || PL_multi_close == 127
296 tmpbuf[1] = toCTRL(PL_multi_close);
302 *tmpbuf = (char)PL_multi_close;
306 q = strchr(s,'"') ? '\'' : '"';
307 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
315 Perl_deprecate(pTHX_ char *s)
317 if (ckWARN(WARN_DEPRECATED))
318 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
322 Perl_deprecate_old(pTHX_ char *s)
324 /* This function should NOT be called for any new deprecated warnings */
325 /* Use Perl_deprecate instead */
327 /* It is here to maintain backward compatibility with the pre-5.8 */
328 /* warnings category hierarchy. The "deprecated" category used to */
329 /* live under the "syntax" category. It is now a top-level category */
330 /* in its own right. */
332 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
333 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
334 "Use of %s is deprecated", s);
339 * Deprecate a comma-less variable list.
345 deprecate_old("comma-less variable list");
349 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
350 * utf16-to-utf8-reversed.
353 #ifdef PERL_CR_FILTER
357 register char *s = SvPVX(sv);
358 register char *e = s + SvCUR(sv);
359 /* outer loop optimized to do nothing if there are no CR-LFs */
361 if (*s++ == '\r' && *s == '\n') {
362 /* hit a CR-LF, need to copy the rest */
363 register char *d = s - 1;
366 if (*s == '\r' && s[1] == '\n')
377 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
379 I32 count = FILTER_READ(idx+1, sv, maxlen);
380 if (count > 0 && !maxlen)
388 * Initialize variables. Uses the Perl save_stack to save its state (for
389 * recursive calls to the parser).
393 Perl_lex_start(pTHX_ SV *line)
398 SAVEI32(PL_lex_dojoin);
399 SAVEI32(PL_lex_brackets);
400 SAVEI32(PL_lex_casemods);
401 SAVEI32(PL_lex_starts);
402 SAVEI32(PL_lex_state);
403 SAVEVPTR(PL_lex_inpat);
404 SAVEI32(PL_lex_inwhat);
405 if (PL_lex_state == LEX_KNOWNEXT) {
406 I32 toke = PL_nexttoke;
407 while (--toke >= 0) {
408 SAVEI32(PL_nexttype[toke]);
409 SAVEVPTR(PL_nextval[toke]);
411 SAVEI32(PL_nexttoke);
413 SAVECOPLINE(PL_curcop);
416 SAVEPPTR(PL_oldbufptr);
417 SAVEPPTR(PL_oldoldbufptr);
418 SAVEPPTR(PL_last_lop);
419 SAVEPPTR(PL_last_uni);
420 SAVEPPTR(PL_linestart);
421 SAVESPTR(PL_linestr);
422 SAVEGENERICPV(PL_lex_brackstack);
423 SAVEGENERICPV(PL_lex_casestack);
424 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
425 SAVESPTR(PL_lex_stuff);
426 SAVEI32(PL_lex_defer);
427 SAVEI32(PL_sublex_info.sub_inwhat);
428 SAVESPTR(PL_lex_repl);
430 SAVEINT(PL_lex_expect);
432 PL_lex_state = LEX_NORMAL;
436 New(899, PL_lex_brackstack, 120, char);
437 New(899, PL_lex_casestack, 12, char);
439 *PL_lex_casestack = '\0';
442 PL_lex_stuff = Nullsv;
443 PL_lex_repl = Nullsv;
447 PL_sublex_info.sub_inwhat = 0;
449 if (SvREADONLY(PL_linestr))
450 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
451 s = SvPV(PL_linestr, len);
452 if (!len || s[len-1] != ';') {
453 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
454 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
455 sv_catpvn(PL_linestr, "\n;", 2);
457 SvTEMP_off(PL_linestr);
458 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
459 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
460 PL_last_lop = PL_last_uni = Nullch;
466 * Finalizer for lexing operations. Must be called when the parser is
467 * done with the lexer.
473 PL_doextract = FALSE;
478 * This subroutine has nothing to do with tilting, whether at windmills
479 * or pinball tables. Its name is short for "increment line". It
480 * increments the current line number in CopLINE(PL_curcop) and checks
481 * to see whether the line starts with a comment of the form
482 * # line 500 "foo.pm"
483 * If so, it sets the current line number and file to the values in the comment.
487 S_incline(pTHX_ char *s)
494 CopLINE_inc(PL_curcop);
497 while (SPACE_OR_TAB(*s)) s++;
498 if (strnEQ(s, "line", 4))
502 if (SPACE_OR_TAB(*s))
506 while (SPACE_OR_TAB(*s)) s++;
512 while (SPACE_OR_TAB(*s))
514 if (*s == '"' && (t = strchr(s+1, '"'))) {
519 for (t = s; !isSPACE(*t); t++) ;
522 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
524 if (*e != '\n' && *e != '\0')
525 return; /* false alarm */
530 CopFILE_free(PL_curcop);
531 CopFILE_set(PL_curcop, s);
534 CopLINE_set(PL_curcop, atoi(n)-1);
539 * Called to gobble the appropriate amount and type of whitespace.
540 * Skips comments as well.
544 S_skipspace(pTHX_ register char *s)
546 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
547 while (s < PL_bufend && SPACE_OR_TAB(*s))
553 SSize_t oldprevlen, oldoldprevlen;
554 SSize_t oldloplen = 0, oldunilen = 0;
555 while (s < PL_bufend && isSPACE(*s)) {
556 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
561 if (s < PL_bufend && *s == '#') {
562 while (s < PL_bufend && *s != '\n')
566 if (PL_in_eval && !PL_rsfp) {
573 /* only continue to recharge the buffer if we're at the end
574 * of the buffer, we're not reading from a source filter, and
575 * we're in normal lexing mode
577 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
578 PL_lex_state == LEX_FORMLINE)
581 /* try to recharge the buffer */
582 if ((s = filter_gets(PL_linestr, PL_rsfp,
583 (prevlen = SvCUR(PL_linestr)))) == Nullch)
585 /* end of file. Add on the -p or -n magic */
586 if (PL_minus_n || PL_minus_p) {
587 sv_setpv(PL_linestr,PL_minus_p ?
588 ";}continue{print or die qq(-p destination: $!\\n)" :
590 sv_catpv(PL_linestr,";}");
591 PL_minus_n = PL_minus_p = 0;
594 sv_setpv(PL_linestr,";");
596 /* reset variables for next time we lex */
597 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
599 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
600 PL_last_lop = PL_last_uni = Nullch;
602 /* Close the filehandle. Could be from -P preprocessor,
603 * STDIN, or a regular file. If we were reading code from
604 * STDIN (because the commandline held no -e or filename)
605 * then we don't close it, we reset it so the code can
606 * read from STDIN too.
609 if (PL_preprocess && !PL_in_eval)
610 (void)PerlProc_pclose(PL_rsfp);
611 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
612 PerlIO_clearerr(PL_rsfp);
614 (void)PerlIO_close(PL_rsfp);
619 /* not at end of file, so we only read another line */
620 /* make corresponding updates to old pointers, for yyerror() */
621 oldprevlen = PL_oldbufptr - PL_bufend;
622 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
624 oldunilen = PL_last_uni - PL_bufend;
626 oldloplen = PL_last_lop - PL_bufend;
627 PL_linestart = PL_bufptr = s + prevlen;
628 PL_bufend = s + SvCUR(PL_linestr);
630 PL_oldbufptr = s + oldprevlen;
631 PL_oldoldbufptr = s + oldoldprevlen;
633 PL_last_uni = s + oldunilen;
635 PL_last_lop = s + oldloplen;
638 /* debugger active and we're not compiling the debugger code,
639 * so store the line into the debugger's array of lines
641 if (PERLDB_LINE && PL_curstash != PL_debstash) {
642 SV *sv = NEWSV(85,0);
644 sv_upgrade(sv, SVt_PVMG);
645 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
648 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
655 * Check the unary operators to ensure there's no ambiguity in how they're
656 * used. An ambiguous piece of code would be:
658 * This doesn't mean rand() + 5. Because rand() is a unary operator,
659 * the +5 is its argument.
668 if (PL_oldoldbufptr != PL_last_uni)
670 while (isSPACE(*PL_last_uni))
672 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
673 if ((t = strchr(s, '(')) && t < PL_bufptr)
675 if (ckWARN_d(WARN_AMBIGUOUS)){
678 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
679 "Warning: Use of \"%s\" without parens is ambiguous",
686 * LOP : macro to build a list operator. Its behaviour has been replaced
687 * with a subroutine, S_lop() for which LOP is just another name.
690 #define LOP(f,x) return lop(f,x,s)
694 * Build a list operator (or something that might be one). The rules:
695 * - if we have a next token, then it's a list operator [why?]
696 * - if the next thing is an opening paren, then it's a function
697 * - else it's a list operator
701 S_lop(pTHX_ I32 f, int x, char *s)
708 PL_last_lop = PL_oldbufptr;
709 PL_last_lop_op = (OPCODE)f;
723 * When the lexer realizes it knows the next token (for instance,
724 * it is reordering tokens for the parser) then it can call S_force_next
725 * to know what token to return the next time the lexer is called. Caller
726 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
727 * handles the token correctly.
731 S_force_next(pTHX_ I32 type)
733 PL_nexttype[PL_nexttoke] = type;
735 if (PL_lex_state != LEX_KNOWNEXT) {
736 PL_lex_defer = PL_lex_state;
737 PL_lex_expect = PL_expect;
738 PL_lex_state = LEX_KNOWNEXT;
744 * When the lexer knows the next thing is a word (for instance, it has
745 * just seen -> and it knows that the next char is a word char, then
746 * it calls S_force_word to stick the next word into the PL_next lookahead.
749 * char *start : buffer position (must be within PL_linestr)
750 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
751 * int check_keyword : if true, Perl checks to make sure the word isn't
752 * a keyword (do this if the word is a label, e.g. goto FOO)
753 * int allow_pack : if true, : characters will also be allowed (require,
755 * int allow_initial_tick : used by the "sub" lexer only.
759 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
764 start = skipspace(start);
766 if (isIDFIRST_lazy_if(s,UTF) ||
767 (allow_pack && *s == ':') ||
768 (allow_initial_tick && *s == '\'') )
770 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
771 if (check_keyword && keyword(PL_tokenbuf, len))
773 if (token == METHOD) {
778 PL_expect = XOPERATOR;
781 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
782 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
790 * Called when the lexer wants $foo *foo &foo etc, but the program
791 * text only contains the "foo" portion. The first argument is a pointer
792 * to the "foo", and the second argument is the type symbol to prefix.
793 * Forces the next token to be a "WORD".
794 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
798 S_force_ident(pTHX_ register char *s, int kind)
801 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
802 PL_nextval[PL_nexttoke].opval = o;
805 o->op_private = OPpCONST_ENTERED;
806 /* XXX see note in pp_entereval() for why we forgo typo
807 warnings if the symbol must be introduced in an eval.
809 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
810 kind == '$' ? SVt_PV :
811 kind == '@' ? SVt_PVAV :
812 kind == '%' ? SVt_PVHV :
820 Perl_str_to_version(pTHX_ SV *sv)
825 char *start = SvPVx(sv,len);
826 bool utf = SvUTF8(sv) ? TRUE : FALSE;
827 char *end = start + len;
828 while (start < end) {
832 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
837 retval += ((NV)n)/nshift;
846 * Forces the next token to be a version number.
847 * If the next token appears to be an invalid version number, (e.g. "v2b"),
848 * and if "guessing" is TRUE, then no new token is created (and the caller
849 * must use an alternative parsing method).
853 S_force_version(pTHX_ char *s, int guessing)
855 OP *version = Nullop;
864 while (isDIGIT(*d) || *d == '_' || *d == '.')
866 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
868 s = scan_num(s, &yylval);
869 version = yylval.opval;
870 ver = cSVOPx(version)->op_sv;
871 if (SvPOK(ver) && !SvNIOK(ver)) {
872 (void)SvUPGRADE(ver, SVt_PVNV);
873 SvNVX(ver) = str_to_version(ver);
874 SvNOK_on(ver); /* hint that it is a version */
881 /* NOTE: The parser sees the package name and the VERSION swapped */
882 PL_nextval[PL_nexttoke].opval = version;
890 * Tokenize a quoted string passed in as an SV. It finds the next
891 * chunk, up to end of string or a backslash. It may make a new
892 * SV containing that chunk (if HINT_NEW_STRING is on). It also
897 S_tokeq(pTHX_ SV *sv)
908 s = SvPV_force(sv, len);
909 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
912 while (s < send && *s != '\\')
917 if ( PL_hints & HINT_NEW_STRING ) {
918 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
924 if (s + 1 < send && (s[1] == '\\'))
925 s++; /* all that, just for this */
930 SvCUR_set(sv, d - SvPVX(sv));
932 if ( PL_hints & HINT_NEW_STRING )
933 return new_constant(NULL, 0, "q", sv, pv, "q");
938 * Now come three functions related to double-quote context,
939 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
940 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
941 * interact with PL_lex_state, and create fake ( ... ) argument lists
942 * to handle functions and concatenation.
943 * They assume that whoever calls them will be setting up a fake
944 * join call, because each subthing puts a ',' after it. This lets
947 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
949 * (I'm not sure whether the spurious commas at the end of lcfirst's
950 * arguments and join's arguments are created or not).
955 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
957 * Pattern matching will set PL_lex_op to the pattern-matching op to
958 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
960 * OP_CONST and OP_READLINE are easy--just make the new op and return.
962 * Everything else becomes a FUNC.
964 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
965 * had an OP_CONST or OP_READLINE). This just sets us up for a
966 * call to S_sublex_push().
972 register I32 op_type = yylval.ival;
974 if (op_type == OP_NULL) {
975 yylval.opval = PL_lex_op;
979 if (op_type == OP_CONST || op_type == OP_READLINE) {
980 SV *sv = tokeq(PL_lex_stuff);
982 if (SvTYPE(sv) == SVt_PVIV) {
983 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
989 nsv = newSVpvn(p, len);
995 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
996 PL_lex_stuff = Nullsv;
1000 PL_sublex_info.super_state = PL_lex_state;
1001 PL_sublex_info.sub_inwhat = op_type;
1002 PL_sublex_info.sub_op = PL_lex_op;
1003 PL_lex_state = LEX_INTERPPUSH;
1007 yylval.opval = PL_lex_op;
1017 * Create a new scope to save the lexing state. The scope will be
1018 * ended in S_sublex_done. Returns a '(', starting the function arguments
1019 * to the uc, lc, etc. found before.
1020 * Sets PL_lex_state to LEX_INTERPCONCAT.
1028 PL_lex_state = PL_sublex_info.super_state;
1029 SAVEI32(PL_lex_dojoin);
1030 SAVEI32(PL_lex_brackets);
1031 SAVEI32(PL_lex_casemods);
1032 SAVEI32(PL_lex_starts);
1033 SAVEI32(PL_lex_state);
1034 SAVEVPTR(PL_lex_inpat);
1035 SAVEI32(PL_lex_inwhat);
1036 SAVECOPLINE(PL_curcop);
1037 SAVEPPTR(PL_bufptr);
1038 SAVEPPTR(PL_bufend);
1039 SAVEPPTR(PL_oldbufptr);
1040 SAVEPPTR(PL_oldoldbufptr);
1041 SAVEPPTR(PL_last_lop);
1042 SAVEPPTR(PL_last_uni);
1043 SAVEPPTR(PL_linestart);
1044 SAVESPTR(PL_linestr);
1045 SAVEGENERICPV(PL_lex_brackstack);
1046 SAVEGENERICPV(PL_lex_casestack);
1048 PL_linestr = PL_lex_stuff;
1049 PL_lex_stuff = Nullsv;
1051 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1052 = SvPVX(PL_linestr);
1053 PL_bufend += SvCUR(PL_linestr);
1054 PL_last_lop = PL_last_uni = Nullch;
1055 SAVEFREESV(PL_linestr);
1057 PL_lex_dojoin = FALSE;
1058 PL_lex_brackets = 0;
1059 New(899, PL_lex_brackstack, 120, char);
1060 New(899, PL_lex_casestack, 12, char);
1061 PL_lex_casemods = 0;
1062 *PL_lex_casestack = '\0';
1064 PL_lex_state = LEX_INTERPCONCAT;
1065 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1067 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1068 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1069 PL_lex_inpat = PL_sublex_info.sub_op;
1071 PL_lex_inpat = Nullop;
1078 * Restores lexer state after a S_sublex_push.
1084 if (!PL_lex_starts++) {
1085 SV *sv = newSVpvn("",0);
1086 if (SvUTF8(PL_linestr))
1088 PL_expect = XOPERATOR;
1089 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1093 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1094 PL_lex_state = LEX_INTERPCASEMOD;
1098 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1099 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1100 PL_linestr = PL_lex_repl;
1102 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1103 PL_bufend += SvCUR(PL_linestr);
1104 PL_last_lop = PL_last_uni = Nullch;
1105 SAVEFREESV(PL_linestr);
1106 PL_lex_dojoin = FALSE;
1107 PL_lex_brackets = 0;
1108 PL_lex_casemods = 0;
1109 *PL_lex_casestack = '\0';
1111 if (SvEVALED(PL_lex_repl)) {
1112 PL_lex_state = LEX_INTERPNORMAL;
1114 /* we don't clear PL_lex_repl here, so that we can check later
1115 whether this is an evalled subst; that means we rely on the
1116 logic to ensure sublex_done() is called again only via the
1117 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1120 PL_lex_state = LEX_INTERPCONCAT;
1121 PL_lex_repl = Nullsv;
1127 PL_bufend = SvPVX(PL_linestr);
1128 PL_bufend += SvCUR(PL_linestr);
1129 PL_expect = XOPERATOR;
1130 PL_sublex_info.sub_inwhat = 0;
1138 Extracts a pattern, double-quoted string, or transliteration. This
1141 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1142 processing a pattern (PL_lex_inpat is true), a transliteration
1143 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1145 Returns a pointer to the character scanned up to. Iff this is
1146 advanced from the start pointer supplied (ie if anything was
1147 successfully parsed), will leave an OP for the substring scanned
1148 in yylval. Caller must intuit reason for not parsing further
1149 by looking at the next characters herself.
1153 double-quoted style: \r and \n
1154 regexp special ones: \D \s
1156 backrefs: \1 (deprecated in substitution replacements)
1157 case and quoting: \U \Q \E
1158 stops on @ and $, but not for $ as tail anchor
1160 In transliterations:
1161 characters are VERY literal, except for - not at the start or end
1162 of the string, which indicates a range. scan_const expands the
1163 range to the full set of intermediate characters.
1165 In double-quoted strings:
1167 double-quoted style: \r and \n
1169 backrefs: \1 (deprecated)
1170 case and quoting: \U \Q \E
1173 scan_const does *not* construct ops to handle interpolated strings.
1174 It stops processing as soon as it finds an embedded $ or @ variable
1175 and leaves it to the caller to work out what's going on.
1177 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1179 $ in pattern could be $foo or could be tail anchor. Assumption:
1180 it's a tail anchor if $ is the last thing in the string, or if it's
1181 followed by one of ")| \n\t"
1183 \1 (backreferences) are turned into $1
1185 The structure of the code is
1186 while (there's a character to process) {
1187 handle transliteration ranges
1188 skip regexp comments
1189 skip # initiated comments in //x patterns
1190 check for embedded @foo
1191 check for embedded scalars
1193 leave intact backslashes from leave (below)
1194 deprecate \1 in strings and sub replacements
1195 handle string-changing backslashes \l \U \Q \E, etc.
1196 switch (what was escaped) {
1197 handle - in a transliteration (becomes a literal -)
1198 handle \132 octal characters
1199 handle 0x15 hex characters
1200 handle \cV (control V)
1201 handle printf backslashes (\f, \r, \n, etc)
1203 } (end if backslash)
1204 } (end while character to read)
1209 S_scan_const(pTHX_ char *start)
1211 register char *send = PL_bufend; /* end of the constant */
1212 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1213 register char *s = start; /* start of the constant */
1214 register char *d = SvPVX(sv); /* destination for copies */
1215 bool dorange = FALSE; /* are we in a translit range? */
1216 bool didrange = FALSE; /* did we just finish a range? */
1217 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1218 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1221 const char *leaveit = /* set of acceptably-backslashed characters */
1223 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1226 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1227 /* If we are doing a trans and we know we want UTF8 set expectation */
1228 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1229 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1233 while (s < send || dorange) {
1234 /* get transliterations out of the way (they're most literal) */
1235 if (PL_lex_inwhat == OP_TRANS) {
1236 /* expand a range A-Z to the full set of characters. AIE! */
1238 I32 i; /* current expanded character */
1239 I32 min; /* first character in range */
1240 I32 max; /* last character in range */
1243 char *c = (char*)utf8_hop((U8*)d, -1);
1247 *c = (char)UTF_TO_NATIVE(0xff);
1248 /* mark the range as done, and continue */
1254 i = d - SvPVX(sv); /* remember current offset */
1255 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1256 d = SvPVX(sv) + i; /* refresh d after realloc */
1257 d -= 2; /* eat the first char and the - */
1259 min = (U8)*d; /* first char in range */
1260 max = (U8)d[1]; /* last char in range */
1264 "Invalid range \"%c-%c\" in transliteration operator",
1265 (char)min, (char)max);
1269 if ((isLOWER(min) && isLOWER(max)) ||
1270 (isUPPER(min) && isUPPER(max))) {
1272 for (i = min; i <= max; i++)
1274 *d++ = NATIVE_TO_NEED(has_utf8,i);
1276 for (i = min; i <= max; i++)
1278 *d++ = NATIVE_TO_NEED(has_utf8,i);
1283 for (i = min; i <= max; i++)
1286 /* mark the range as done, and continue */
1292 /* range begins (ignore - as first or last char) */
1293 else if (*s == '-' && s+1 < send && s != start) {
1295 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1298 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1310 /* if we get here, we're not doing a transliteration */
1312 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1313 except for the last char, which will be done separately. */
1314 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1316 while (s < send && *s != ')')
1317 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1319 else if (s[2] == '{' /* This should match regcomp.c */
1320 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1323 char *regparse = s + (s[2] == '{' ? 3 : 4);
1326 while (count && (c = *regparse)) {
1327 if (c == '\\' && regparse[1])
1335 if (*regparse != ')') {
1336 regparse--; /* Leave one char for continuation. */
1337 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1339 while (s < regparse)
1340 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1344 /* likewise skip #-initiated comments in //x patterns */
1345 else if (*s == '#' && PL_lex_inpat &&
1346 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1347 while (s+1 < send && *s != '\n')
1348 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1351 /* check for embedded arrays
1352 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1354 else if (*s == '@' && s[1]
1355 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1358 /* check for embedded scalars. only stop if we're sure it's a
1361 else if (*s == '$') {
1362 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1364 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1365 break; /* in regexp, $ might be tail anchor */
1368 /* End of else if chain - OP_TRANS rejoin rest */
1371 if (*s == '\\' && s+1 < send) {
1374 /* some backslashes we leave behind */
1375 if (*leaveit && *s && strchr(leaveit, *s)) {
1376 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1377 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1381 /* deprecate \1 in strings and substitution replacements */
1382 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1383 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1385 if (ckWARN(WARN_SYNTAX))
1386 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1391 /* string-change backslash escapes */
1392 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1397 /* if we get here, it's either a quoted -, or a digit */
1400 /* quoted - in transliterations */
1402 if (PL_lex_inwhat == OP_TRANS) {
1409 if (ckWARN(WARN_MISC) &&
1412 Perl_warner(aTHX_ packWARN(WARN_MISC),
1413 "Unrecognized escape \\%c passed through",
1415 /* default action is to copy the quoted character */
1416 goto default_action;
1419 /* \132 indicates an octal constant */
1420 case '0': case '1': case '2': case '3':
1421 case '4': case '5': case '6': case '7':
1425 uv = grok_oct(s, &len, &flags, NULL);
1428 goto NUM_ESCAPE_INSERT;
1430 /* \x24 indicates a hex constant */
1434 char* e = strchr(s, '}');
1435 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1436 PERL_SCAN_DISALLOW_PREFIX;
1441 yyerror("Missing right brace on \\x{}");
1445 uv = grok_hex(s, &len, &flags, NULL);
1451 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1452 uv = grok_hex(s, &len, &flags, NULL);
1458 /* Insert oct or hex escaped character.
1459 * There will always enough room in sv since such
1460 * escapes will be longer than any UTF-8 sequence
1461 * they can end up as. */
1463 /* We need to map to chars to ASCII before doing the tests
1466 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1467 if (!has_utf8 && uv > 255) {
1468 /* Might need to recode whatever we have
1469 * accumulated so far if it contains any
1472 * (Can't we keep track of that and avoid
1473 * this rescan? --jhi)
1477 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1478 if (!NATIVE_IS_INVARIANT(*c)) {
1483 STRLEN offset = d - SvPVX(sv);
1485 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1489 while (src >= (U8 *)SvPVX(sv)) {
1490 if (!NATIVE_IS_INVARIANT(*src)) {
1491 U8 ch = NATIVE_TO_ASCII(*src);
1492 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1493 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1503 if (has_utf8 || uv > 255) {
1504 d = (char*)uvchr_to_utf8((U8*)d, uv);
1506 if (PL_lex_inwhat == OP_TRANS &&
1507 PL_sublex_info.sub_op) {
1508 PL_sublex_info.sub_op->op_private |=
1509 (PL_lex_repl ? OPpTRANS_FROM_UTF
1522 /* \N{LATIN SMALL LETTER A} is a named character */
1526 char* e = strchr(s, '}');
1532 yyerror("Missing right brace on \\N{}");
1536 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1538 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1539 PERL_SCAN_DISALLOW_PREFIX;
1542 uv = grok_hex(s, &len, &flags, NULL);
1544 goto NUM_ESCAPE_INSERT;
1546 res = newSVpvn(s + 1, e - s - 1);
1547 res = new_constant( Nullch, 0, "charnames",
1548 res, Nullsv, "\\N{...}" );
1550 sv_utf8_upgrade(res);
1551 str = SvPV(res,len);
1552 #ifdef EBCDIC_NEVER_MIND
1553 /* charnames uses pack U and that has been
1554 * recently changed to do the below uni->native
1555 * mapping, so this would be redundant (and wrong,
1556 * the code point would be doubly converted).
1557 * But leave this in just in case the pack U change
1558 * gets revoked, but the semantics is still
1559 * desireable for charnames. --jhi */
1561 UV uv = utf8_to_uvchr((U8*)str, 0);
1564 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1566 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1567 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1568 str = SvPV(res, len);
1572 if (!has_utf8 && SvUTF8(res)) {
1573 char *ostart = SvPVX(sv);
1574 SvCUR_set(sv, d - ostart);
1577 sv_utf8_upgrade(sv);
1578 /* this just broke our allocation above... */
1579 SvGROW(sv, (STRLEN)(send - start));
1580 d = SvPVX(sv) + SvCUR(sv);
1583 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1584 char *odest = SvPVX(sv);
1586 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1587 d = SvPVX(sv) + (d - odest);
1589 Copy(str, d, len, char);
1596 yyerror("Missing braces on \\N{}");
1599 /* \c is a control character */
1608 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1611 yyerror("Missing control char name in \\c");
1615 /* printf-style backslashes, formfeeds, newlines, etc */
1617 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1620 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1623 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1626 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1629 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1632 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1635 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1641 } /* end if (backslash) */
1644 /* If we started with encoded form, or already know we want it
1645 and then encode the next character */
1646 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1648 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1649 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1652 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1653 STRLEN off = d - SvPVX(sv);
1654 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1656 d = (char*)uvchr_to_utf8((U8*)d, uv);
1660 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1662 } /* while loop to process each character */
1664 /* terminate the string and set up the sv */
1666 SvCUR_set(sv, d - SvPVX(sv));
1667 if (SvCUR(sv) >= SvLEN(sv))
1668 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1671 if (PL_encoding && !has_utf8) {
1672 sv_recode_to_utf8(sv, PL_encoding);
1678 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1679 PL_sublex_info.sub_op->op_private |=
1680 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1684 /* shrink the sv if we allocated more than we used */
1685 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1686 SvLEN_set(sv, SvCUR(sv) + 1);
1687 Renew(SvPVX(sv), SvLEN(sv), char);
1690 /* return the substring (via yylval) only if we parsed anything */
1691 if (s > PL_bufptr) {
1692 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1693 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1695 ( PL_lex_inwhat == OP_TRANS
1697 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1700 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1707 * Returns TRUE if there's more to the expression (e.g., a subscript),
1710 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1712 * ->[ and ->{ return TRUE
1713 * { and [ outside a pattern are always subscripts, so return TRUE
1714 * if we're outside a pattern and it's not { or [, then return FALSE
1715 * if we're in a pattern and the first char is a {
1716 * {4,5} (any digits around the comma) returns FALSE
1717 * if we're in a pattern and the first char is a [
1719 * [SOMETHING] has a funky algorithm to decide whether it's a
1720 * character class or not. It has to deal with things like
1721 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1722 * anything else returns TRUE
1725 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1728 S_intuit_more(pTHX_ register char *s)
1730 if (PL_lex_brackets)
1732 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1734 if (*s != '{' && *s != '[')
1739 /* In a pattern, so maybe we have {n,m}. */
1756 /* On the other hand, maybe we have a character class */
1759 if (*s == ']' || *s == '^')
1762 /* this is terrifying, and it works */
1763 int weight = 2; /* let's weigh the evidence */
1765 unsigned char un_char = 255, last_un_char;
1766 char *send = strchr(s,']');
1767 char tmpbuf[sizeof PL_tokenbuf * 4];
1769 if (!send) /* has to be an expression */
1772 Zero(seen,256,char);
1775 else if (isDIGIT(*s)) {
1777 if (isDIGIT(s[1]) && s[2] == ']')
1783 for (; s < send; s++) {
1784 last_un_char = un_char;
1785 un_char = (unsigned char)*s;
1790 weight -= seen[un_char] * 10;
1791 if (isALNUM_lazy_if(s+1,UTF)) {
1792 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1793 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1798 else if (*s == '$' && s[1] &&
1799 strchr("[#!%*<>()-=",s[1])) {
1800 if (/*{*/ strchr("])} =",s[2]))
1809 if (strchr("wds]",s[1]))
1811 else if (seen['\''] || seen['"'])
1813 else if (strchr("rnftbxcav",s[1]))
1815 else if (isDIGIT(s[1])) {
1817 while (s[1] && isDIGIT(s[1]))
1827 if (strchr("aA01! ",last_un_char))
1829 if (strchr("zZ79~",s[1]))
1831 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1832 weight -= 5; /* cope with negative subscript */
1835 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1836 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1841 if (keyword(tmpbuf, d - tmpbuf))
1844 if (un_char == last_un_char + 1)
1846 weight -= seen[un_char];
1851 if (weight >= 0) /* probably a character class */
1861 * Does all the checking to disambiguate
1863 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1864 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1866 * First argument is the stuff after the first token, e.g. "bar".
1868 * Not a method if bar is a filehandle.
1869 * Not a method if foo is a subroutine prototyped to take a filehandle.
1870 * Not a method if it's really "Foo $bar"
1871 * Method if it's "foo $bar"
1872 * Not a method if it's really "print foo $bar"
1873 * Method if it's really "foo package::" (interpreted as package->foo)
1874 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1875 * Not a method if bar is a filehandle or package, but is quoted with
1880 S_intuit_method(pTHX_ char *start, GV *gv)
1882 char *s = start + (*start == '$');
1883 char tmpbuf[sizeof PL_tokenbuf];
1891 if ((cv = GvCVu(gv))) {
1892 char *proto = SvPVX(cv);
1902 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1903 /* start is the beginning of the possible filehandle/object,
1904 * and s is the end of it
1905 * tmpbuf is a copy of it
1908 if (*start == '$') {
1909 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1914 return *s == '(' ? FUNCMETH : METHOD;
1916 if (!keyword(tmpbuf, len)) {
1917 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1922 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1923 if (indirgv && GvCVu(indirgv))
1925 /* filehandle or package name makes it a method */
1926 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1928 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1929 return 0; /* no assumptions -- "=>" quotes bearword */
1931 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1932 newSVpvn(tmpbuf,len));
1933 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1937 return *s == '(' ? FUNCMETH : METHOD;
1945 * Return a string of Perl code to load the debugger. If PERL5DB
1946 * is set, it will return the contents of that, otherwise a
1947 * compile-time require of perl5db.pl.
1954 char *pdb = PerlEnv_getenv("PERL5DB");
1958 SETERRNO(0,SS_NORMAL);
1959 return "BEGIN { require 'perl5db.pl' }";
1965 /* Encoded script support. filter_add() effectively inserts a
1966 * 'pre-processing' function into the current source input stream.
1967 * Note that the filter function only applies to the current source file
1968 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1970 * The datasv parameter (which may be NULL) can be used to pass
1971 * private data to this instance of the filter. The filter function
1972 * can recover the SV using the FILTER_DATA macro and use it to
1973 * store private buffers and state information.
1975 * The supplied datasv parameter is upgraded to a PVIO type
1976 * and the IoDIRP/IoANY field is used to store the function pointer,
1977 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1978 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1979 * private use must be set using malloc'd pointers.
1983 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1988 if (!PL_rsfp_filters)
1989 PL_rsfp_filters = newAV();
1991 datasv = NEWSV(255,0);
1992 if (!SvUPGRADE(datasv, SVt_PVIO))
1993 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1994 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1995 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1996 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1997 (void*)funcp, SvPV_nolen(datasv)));
1998 av_unshift(PL_rsfp_filters, 1);
1999 av_store(PL_rsfp_filters, 0, datasv) ;
2004 /* Delete most recently added instance of this filter function. */
2006 Perl_filter_del(pTHX_ filter_t funcp)
2009 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2010 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2012 /* if filter is on top of stack (usual case) just pop it off */
2013 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2014 if (IoANY(datasv) == (void *)funcp) {
2015 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2016 IoANY(datasv) = (void *)NULL;
2017 sv_free(av_pop(PL_rsfp_filters));
2021 /* we need to search for the correct entry and clear it */
2022 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2026 /* Invoke the n'th filter function for the current rsfp. */
2028 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2031 /* 0 = read one text line */
2036 if (!PL_rsfp_filters)
2038 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2039 /* Provide a default input filter to make life easy. */
2040 /* Note that we append to the line. This is handy. */
2041 DEBUG_P(PerlIO_printf(Perl_debug_log,
2042 "filter_read %d: from rsfp\n", idx));
2046 int old_len = SvCUR(buf_sv) ;
2048 /* ensure buf_sv is large enough */
2049 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2050 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2051 if (PerlIO_error(PL_rsfp))
2052 return -1; /* error */
2054 return 0 ; /* end of file */
2056 SvCUR_set(buf_sv, old_len + len) ;
2059 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2060 if (PerlIO_error(PL_rsfp))
2061 return -1; /* error */
2063 return 0 ; /* end of file */
2066 return SvCUR(buf_sv);
2068 /* Skip this filter slot if filter has been deleted */
2069 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2070 DEBUG_P(PerlIO_printf(Perl_debug_log,
2071 "filter_read %d: skipped (filter deleted)\n",
2073 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2075 /* Get function pointer hidden within datasv */
2076 funcp = (filter_t)IoANY(datasv);
2077 DEBUG_P(PerlIO_printf(Perl_debug_log,
2078 "filter_read %d: via function %p (%s)\n",
2079 idx, (void*)funcp, SvPV_nolen(datasv)));
2080 /* Call function. The function is expected to */
2081 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2082 /* Return: <0:error, =0:eof, >0:not eof */
2083 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2087 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2089 #ifdef PERL_CR_FILTER
2090 if (!PL_rsfp_filters) {
2091 filter_add(S_cr_textfilter,NULL);
2094 if (PL_rsfp_filters) {
2097 SvCUR_set(sv, 0); /* start with empty line */
2098 if (FILTER_READ(0, sv, 0) > 0)
2099 return ( SvPVX(sv) ) ;
2104 return (sv_gets(sv, fp, append));
2108 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2112 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2116 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2117 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2119 return GvHV(gv); /* Foo:: */
2122 /* use constant CLASS => 'MyClass' */
2123 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2125 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2126 pkgname = SvPV_nolen(sv);
2130 return gv_stashpv(pkgname, FALSE);
2134 static char* exp_name[] =
2135 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2136 "ATTRTERM", "TERMBLOCK"
2143 Works out what to call the token just pulled out of the input
2144 stream. The yacc parser takes care of taking the ops we return and
2145 stitching them into a tree.
2151 if read an identifier
2152 if we're in a my declaration
2153 croak if they tried to say my($foo::bar)
2154 build the ops for a my() declaration
2155 if it's an access to a my() variable
2156 are we in a sort block?
2157 croak if my($a); $a <=> $b
2158 build ops for access to a my() variable
2159 if in a dq string, and they've said @foo and we can't find @foo
2161 build ops for a bareword
2162 if we already built the token before, use it.
2165 #ifdef USE_PURE_BISON
2167 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2172 yylval_pointer[yyactlevel] = lvalp;
2173 yychar_pointer[yyactlevel] = lcharp;
2174 if (yyactlevel >= YYMAXLEVEL)
2175 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2177 r = Perl_yylex(aTHX);
2187 #pragma segment Perl_yylex
2199 I32 orig_keyword = 0;
2201 /* check if there's an identifier for us to look at */
2202 if (PL_pending_ident)
2203 return S_pending_ident(aTHX);
2205 /* no identifier pending identification */
2207 switch (PL_lex_state) {
2209 case LEX_NORMAL: /* Some compilers will produce faster */
2210 case LEX_INTERPNORMAL: /* code if we comment these out. */
2214 /* when we've already built the next token, just pull it out of the queue */
2217 yylval = PL_nextval[PL_nexttoke];
2219 PL_lex_state = PL_lex_defer;
2220 PL_expect = PL_lex_expect;
2221 PL_lex_defer = LEX_NORMAL;
2223 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2224 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2225 (IV)PL_nexttype[PL_nexttoke]); });
2227 return(PL_nexttype[PL_nexttoke]);
2229 /* interpolated case modifiers like \L \U, including \Q and \E.
2230 when we get here, PL_bufptr is at the \
2232 case LEX_INTERPCASEMOD:
2234 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2235 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2237 /* handle \E or end of string */
2238 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2242 if (PL_lex_casemods) {
2243 oldmod = PL_lex_casestack[--PL_lex_casemods];
2244 PL_lex_casestack[PL_lex_casemods] = '\0';
2246 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2248 PL_lex_state = LEX_INTERPCONCAT;
2252 if (PL_bufptr != PL_bufend)
2254 PL_lex_state = LEX_INTERPCONCAT;
2258 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2259 "### Saw case modifier at '%s'\n", PL_bufptr); });
2261 if (s[1] == '\\' && s[2] == 'E') {
2263 PL_lex_state = LEX_INTERPCONCAT;
2267 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2268 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2269 if (strchr("LU", *s) &&
2270 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2271 PL_lex_casestack[--PL_lex_casemods] = '\0';
2274 if (PL_lex_casemods > 10)
2275 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2276 PL_lex_casestack[PL_lex_casemods++] = *s;
2277 PL_lex_casestack[PL_lex_casemods] = '\0';
2278 PL_lex_state = LEX_INTERPCONCAT;
2279 PL_nextval[PL_nexttoke].ival = 0;
2282 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2284 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2286 PL_nextval[PL_nexttoke].ival = OP_LC;
2288 PL_nextval[PL_nexttoke].ival = OP_UC;
2290 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2292 Perl_croak(aTHX_ "panic: yylex");
2296 if (PL_lex_starts) {
2305 case LEX_INTERPPUSH:
2306 return sublex_push();
2308 case LEX_INTERPSTART:
2309 if (PL_bufptr == PL_bufend)
2310 return sublex_done();
2311 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2312 "### Interpolated variable at '%s'\n", PL_bufptr); });
2314 PL_lex_dojoin = (*PL_bufptr == '@');
2315 PL_lex_state = LEX_INTERPNORMAL;
2316 if (PL_lex_dojoin) {
2317 PL_nextval[PL_nexttoke].ival = 0;
2319 #ifdef USE_5005THREADS
2320 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2321 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2322 force_next(PRIVATEREF);
2324 force_ident("\"", '$');
2325 #endif /* USE_5005THREADS */
2326 PL_nextval[PL_nexttoke].ival = 0;
2328 PL_nextval[PL_nexttoke].ival = 0;
2330 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2333 if (PL_lex_starts++) {
2339 case LEX_INTERPENDMAYBE:
2340 if (intuit_more(PL_bufptr)) {
2341 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2347 if (PL_lex_dojoin) {
2348 PL_lex_dojoin = FALSE;
2349 PL_lex_state = LEX_INTERPCONCAT;
2352 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2353 && SvEVALED(PL_lex_repl))
2355 if (PL_bufptr != PL_bufend)
2356 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2357 PL_lex_repl = Nullsv;
2360 case LEX_INTERPCONCAT:
2362 if (PL_lex_brackets)
2363 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2365 if (PL_bufptr == PL_bufend)
2366 return sublex_done();
2368 if (SvIVX(PL_linestr) == '\'') {
2369 SV *sv = newSVsv(PL_linestr);
2372 else if ( PL_hints & HINT_NEW_RE )
2373 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2374 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2378 s = scan_const(PL_bufptr);
2380 PL_lex_state = LEX_INTERPCASEMOD;
2382 PL_lex_state = LEX_INTERPSTART;
2385 if (s != PL_bufptr) {
2386 PL_nextval[PL_nexttoke] = yylval;
2389 if (PL_lex_starts++)
2399 PL_lex_state = LEX_NORMAL;
2400 s = scan_formline(PL_bufptr);
2401 if (!PL_lex_formbrack)
2407 PL_oldoldbufptr = PL_oldbufptr;
2410 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2411 exp_name[PL_expect], s);
2417 if (isIDFIRST_lazy_if(s,UTF))
2419 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2422 goto fake_eof; /* emulate EOF on ^D or ^Z */
2427 if (PL_lex_brackets)
2428 yyerror("Missing right curly or square bracket");
2429 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2430 "### Tokener got EOF\n");
2434 if (s++ < PL_bufend)
2435 goto retry; /* ignore stray nulls */
2438 if (!PL_in_eval && !PL_preambled) {
2439 PL_preambled = TRUE;
2440 sv_setpv(PL_linestr,incl_perldb());
2441 if (SvCUR(PL_linestr))
2442 sv_catpv(PL_linestr,";");
2444 while(AvFILLp(PL_preambleav) >= 0) {
2445 SV *tmpsv = av_shift(PL_preambleav);
2446 sv_catsv(PL_linestr, tmpsv);
2447 sv_catpv(PL_linestr, ";");
2450 sv_free((SV*)PL_preambleav);
2451 PL_preambleav = NULL;
2453 if (PL_minus_n || PL_minus_p) {
2454 sv_catpv(PL_linestr, "LINE: while (<>) {");
2456 sv_catpv(PL_linestr,"chomp;");
2459 if (strchr("/'\"", *PL_splitstr)
2460 && strchr(PL_splitstr + 1, *PL_splitstr))
2461 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2464 s = "'~#\200\1'"; /* surely one char is unused...*/
2465 while (s[1] && strchr(PL_splitstr, *s)) s++;
2467 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2468 "q" + (delim == '\''), delim);
2469 for (s = PL_splitstr; *s; s++) {
2471 sv_catpvn(PL_linestr, "\\", 1);
2472 sv_catpvn(PL_linestr, s, 1);
2474 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2478 sv_catpv(PL_linestr,"our @F=split(' ');");
2481 sv_catpv(PL_linestr, "\n");
2482 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2483 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2484 PL_last_lop = PL_last_uni = Nullch;
2485 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2486 SV *sv = NEWSV(85,0);
2488 sv_upgrade(sv, SVt_PVMG);
2489 sv_setsv(sv,PL_linestr);
2492 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2497 bof = PL_rsfp ? TRUE : FALSE;
2498 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2501 if (PL_preprocess && !PL_in_eval)
2502 (void)PerlProc_pclose(PL_rsfp);
2503 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2504 PerlIO_clearerr(PL_rsfp);
2506 (void)PerlIO_close(PL_rsfp);
2508 PL_doextract = FALSE;
2510 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2511 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2512 sv_catpv(PL_linestr,";}");
2513 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2514 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2515 PL_last_lop = PL_last_uni = Nullch;
2516 PL_minus_n = PL_minus_p = 0;
2519 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2520 PL_last_lop = PL_last_uni = Nullch;
2521 sv_setpv(PL_linestr,"");
2522 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2524 /* if it looks like the start of a BOM, check if it in fact is */
2525 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2526 #ifdef PERLIO_IS_STDIO
2527 # ifdef __GNU_LIBRARY__
2528 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2529 # define FTELL_FOR_PIPE_IS_BROKEN
2533 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2534 # define FTELL_FOR_PIPE_IS_BROKEN
2539 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2540 /* This loses the possibility to detect the bof
2541 * situation on perl -P when the libc5 is being used.
2542 * Workaround? Maybe attach some extra state to PL_rsfp?
2545 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2547 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2550 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2551 s = swallow_bom((U8*)s);
2555 /* Incest with pod. */
2556 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2557 sv_setpv(PL_linestr, "");
2558 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2559 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2560 PL_last_lop = PL_last_uni = Nullch;
2561 PL_doextract = FALSE;
2565 } while (PL_doextract);
2566 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2567 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2568 SV *sv = NEWSV(85,0);
2570 sv_upgrade(sv, SVt_PVMG);
2571 sv_setsv(sv,PL_linestr);
2574 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2576 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2577 PL_last_lop = PL_last_uni = Nullch;
2578 if (CopLINE(PL_curcop) == 1) {
2579 while (s < PL_bufend && isSPACE(*s))
2581 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2585 if (*s == '#' && *(s+1) == '!')
2587 #ifdef ALTERNATE_SHEBANG
2589 static char as[] = ALTERNATE_SHEBANG;
2590 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2591 d = s + (sizeof(as) - 1);
2593 #endif /* ALTERNATE_SHEBANG */
2602 while (*d && !isSPACE(*d))
2606 #ifdef ARG_ZERO_IS_SCRIPT
2607 if (ipathend > ipath) {
2609 * HP-UX (at least) sets argv[0] to the script name,
2610 * which makes $^X incorrect. And Digital UNIX and Linux,
2611 * at least, set argv[0] to the basename of the Perl
2612 * interpreter. So, having found "#!", we'll set it right.
2614 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2615 assert(SvPOK(x) || SvGMAGICAL(x));
2616 if (sv_eq(x, CopFILESV(PL_curcop))) {
2617 sv_setpvn(x, ipath, ipathend - ipath);
2623 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2624 char *lstart = SvPV(x,llen);
2626 bstart += blen - llen;
2627 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2628 sv_setpvn(x, ipath, ipathend - ipath);
2633 TAINT_NOT; /* $^X is always tainted, but that's OK */
2635 #endif /* ARG_ZERO_IS_SCRIPT */
2640 d = instr(s,"perl -");
2642 d = instr(s,"perl");
2644 /* avoid getting into infinite loops when shebang
2645 * line contains "Perl" rather than "perl" */
2647 for (d = ipathend-4; d >= ipath; --d) {
2648 if ((*d == 'p' || *d == 'P')
2649 && !ibcmp(d, "perl", 4))
2659 #ifdef ALTERNATE_SHEBANG
2661 * If the ALTERNATE_SHEBANG on this system starts with a
2662 * character that can be part of a Perl expression, then if
2663 * we see it but not "perl", we're probably looking at the
2664 * start of Perl code, not a request to hand off to some
2665 * other interpreter. Similarly, if "perl" is there, but
2666 * not in the first 'word' of the line, we assume the line
2667 * contains the start of the Perl program.
2669 if (d && *s != '#') {
2671 while (*c && !strchr("; \t\r\n\f\v#", *c))
2674 d = Nullch; /* "perl" not in first word; ignore */
2676 *s = '#'; /* Don't try to parse shebang line */
2678 #endif /* ALTERNATE_SHEBANG */
2679 #ifndef MACOS_TRADITIONAL
2684 !instr(s,"indir") &&
2685 instr(PL_origargv[0],"perl"))
2691 while (s < PL_bufend && isSPACE(*s))
2693 if (s < PL_bufend) {
2694 Newz(899,newargv,PL_origargc+3,char*);
2696 while (s < PL_bufend && !isSPACE(*s))
2699 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2702 newargv = PL_origargv;
2704 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2705 Perl_croak(aTHX_ "Can't exec %s", ipath);
2709 U32 oldpdb = PL_perldb;
2710 bool oldn = PL_minus_n;
2711 bool oldp = PL_minus_p;
2713 while (*d && !isSPACE(*d)) d++;
2714 while (SPACE_OR_TAB(*d)) d++;
2717 bool switches_done = PL_doswitches;
2719 if (*d == 'M' || *d == 'm') {
2721 while (*d && !isSPACE(*d)) d++;
2722 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2725 d = moreswitches(d);
2727 if ((PERLDB_LINE && !oldpdb) ||
2728 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2729 /* if we have already added "LINE: while (<>) {",
2730 we must not do it again */
2732 sv_setpv(PL_linestr, "");
2733 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2734 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2735 PL_last_lop = PL_last_uni = Nullch;
2736 PL_preambled = FALSE;
2738 (void)gv_fetchfile(PL_origfilename);
2741 if (PL_doswitches && !switches_done) {
2742 int argc = PL_origargc;
2743 char **argv = PL_origargv;
2746 } while (argc && argv[0][0] == '-' && argv[0][1]);
2747 init_argv_symbols(argc,argv);
2753 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2755 PL_lex_state = LEX_FORMLINE;
2760 #ifdef PERL_STRICT_CR
2761 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2763 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2765 case ' ': case '\t': case '\f': case 013:
2766 #ifdef MACOS_TRADITIONAL
2773 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2774 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2775 /* handle eval qq[#line 1 "foo"\n ...] */
2776 CopLINE_dec(PL_curcop);
2780 while (s < d && *s != '\n')
2784 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2785 Perl_croak(aTHX_ "panic: input overflow");
2787 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2789 PL_lex_state = LEX_FORMLINE;
2799 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2806 while (s < PL_bufend && SPACE_OR_TAB(*s))
2809 if (strnEQ(s,"=>",2)) {
2810 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2811 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2812 "### Saw unary minus before =>, forcing word '%s'\n", s);
2814 OPERATOR('-'); /* unary minus */
2816 PL_last_uni = PL_oldbufptr;
2818 case 'r': ftst = OP_FTEREAD; break;
2819 case 'w': ftst = OP_FTEWRITE; break;
2820 case 'x': ftst = OP_FTEEXEC; break;
2821 case 'o': ftst = OP_FTEOWNED; break;
2822 case 'R': ftst = OP_FTRREAD; break;
2823 case 'W': ftst = OP_FTRWRITE; break;
2824 case 'X': ftst = OP_FTREXEC; break;
2825 case 'O': ftst = OP_FTROWNED; break;
2826 case 'e': ftst = OP_FTIS; break;
2827 case 'z': ftst = OP_FTZERO; break;
2828 case 's': ftst = OP_FTSIZE; break;
2829 case 'f': ftst = OP_FTFILE; break;
2830 case 'd': ftst = OP_FTDIR; break;
2831 case 'l': ftst = OP_FTLINK; break;
2832 case 'p': ftst = OP_FTPIPE; break;
2833 case 'S': ftst = OP_FTSOCK; break;
2834 case 'u': ftst = OP_FTSUID; break;
2835 case 'g': ftst = OP_FTSGID; break;
2836 case 'k': ftst = OP_FTSVTX; break;
2837 case 'b': ftst = OP_FTBLK; break;
2838 case 'c': ftst = OP_FTCHR; break;
2839 case 't': ftst = OP_FTTTY; break;
2840 case 'T': ftst = OP_FTTEXT; break;
2841 case 'B': ftst = OP_FTBINARY; break;
2842 case 'M': case 'A': case 'C':
2843 gv_fetchpv("\024",TRUE, SVt_PV);
2845 case 'M': ftst = OP_FTMTIME; break;
2846 case 'A': ftst = OP_FTATIME; break;
2847 case 'C': ftst = OP_FTCTIME; break;
2855 PL_last_lop_op = (OPCODE)ftst;
2856 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2857 "### Saw file test %c\n", (int)ftst);
2862 /* Assume it was a minus followed by a one-letter named
2863 * subroutine call (or a -bareword), then. */
2864 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2865 "### %c looked like a file test but was not\n",
2874 if (PL_expect == XOPERATOR)
2879 else if (*s == '>') {
2882 if (isIDFIRST_lazy_if(s,UTF)) {
2883 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2891 if (PL_expect == XOPERATOR)
2894 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2896 OPERATOR('-'); /* unary minus */
2903 if (PL_expect == XOPERATOR)
2908 if (PL_expect == XOPERATOR)
2911 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2917 if (PL_expect != XOPERATOR) {
2918 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2919 PL_expect = XOPERATOR;
2920 force_ident(PL_tokenbuf, '*');
2933 if (PL_expect == XOPERATOR) {
2937 PL_tokenbuf[0] = '%';
2938 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2939 if (!PL_tokenbuf[1]) {
2942 PL_pending_ident = '%';
2961 switch (PL_expect) {
2964 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2966 PL_bufptr = s; /* update in case we back off */
2972 PL_expect = XTERMBLOCK;
2976 while (isIDFIRST_lazy_if(s,UTF)) {
2977 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2978 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2979 if (tmp < 0) tmp = -tmp;
2994 d = scan_str(d,TRUE,TRUE);
2996 /* MUST advance bufptr here to avoid bogus
2997 "at end of line" context messages from yyerror().
2999 PL_bufptr = s + len;
3000 yyerror("Unterminated attribute parameter in attribute list");
3003 return 0; /* EOF indicator */
3007 SV *sv = newSVpvn(s, len);
3008 sv_catsv(sv, PL_lex_stuff);
3009 attrs = append_elem(OP_LIST, attrs,
3010 newSVOP(OP_CONST, 0, sv));
3011 SvREFCNT_dec(PL_lex_stuff);
3012 PL_lex_stuff = Nullsv;
3015 /* NOTE: any CV attrs applied here need to be part of
3016 the CVf_BUILTIN_ATTRS define in cv.h! */
3017 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3018 CvLVALUE_on(PL_compcv);
3019 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3020 CvLOCKED_on(PL_compcv);
3021 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3022 CvMETHOD_on(PL_compcv);
3024 else if (PL_in_my == KEY_our && len == 6 &&
3025 strnEQ(s, "unique", len))
3026 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3028 /* After we've set the flags, it could be argued that
3029 we don't need to do the attributes.pm-based setting
3030 process, and shouldn't bother appending recognized
3031 flags. To experiment with that, uncomment the
3032 following "else". (Note that's already been
3033 uncommented. That keeps the above-applied built-in
3034 attributes from being intercepted (and possibly
3035 rejected) by a package's attribute routines, but is
3036 justified by the performance win for the common case
3037 of applying only built-in attributes.) */
3039 attrs = append_elem(OP_LIST, attrs,
3040 newSVOP(OP_CONST, 0,
3044 if (*s == ':' && s[1] != ':')
3047 break; /* require real whitespace or :'s */
3049 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3050 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3051 char q = ((*s == '\'') ? '"' : '\'');
3052 /* If here for an expression, and parsed no attrs, back off. */
3053 if (tmp == '=' && !attrs) {
3057 /* MUST advance bufptr here to avoid bogus "at end of line"
3058 context messages from yyerror().
3062 yyerror("Unterminated attribute list");
3064 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3072 PL_nextval[PL_nexttoke].opval = attrs;
3080 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3081 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3098 if (PL_lex_brackets <= 0)
3099 yyerror("Unmatched right square bracket");
3102 if (PL_lex_state == LEX_INTERPNORMAL) {
3103 if (PL_lex_brackets == 0) {
3104 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3105 PL_lex_state = LEX_INTERPEND;
3112 if (PL_lex_brackets > 100) {
3113 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3115 switch (PL_expect) {
3117 if (PL_lex_formbrack) {
3121 if (PL_oldoldbufptr == PL_last_lop)
3122 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3124 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3125 OPERATOR(HASHBRACK);
3127 while (s < PL_bufend && SPACE_OR_TAB(*s))
3130 PL_tokenbuf[0] = '\0';
3131 if (d < PL_bufend && *d == '-') {
3132 PL_tokenbuf[0] = '-';
3134 while (d < PL_bufend && SPACE_OR_TAB(*d))
3137 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3138 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3140 while (d < PL_bufend && SPACE_OR_TAB(*d))
3143 char minus = (PL_tokenbuf[0] == '-');
3144 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3152 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3157 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3162 if (PL_oldoldbufptr == PL_last_lop)
3163 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3165 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3168 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3170 /* This hack is to get the ${} in the message. */
3172 yyerror("syntax error");
3175 OPERATOR(HASHBRACK);
3177 /* This hack serves to disambiguate a pair of curlies
3178 * as being a block or an anon hash. Normally, expectation
3179 * determines that, but in cases where we're not in a
3180 * position to expect anything in particular (like inside
3181 * eval"") we have to resolve the ambiguity. This code
3182 * covers the case where the first term in the curlies is a
3183 * quoted string. Most other cases need to be explicitly
3184 * disambiguated by prepending a `+' before the opening
3185 * curly in order to force resolution as an anon hash.
3187 * XXX should probably propagate the outer expectation
3188 * into eval"" to rely less on this hack, but that could
3189 * potentially break current behavior of eval"".
3193 if (*s == '\'' || *s == '"' || *s == '`') {
3194 /* common case: get past first string, handling escapes */
3195 for (t++; t < PL_bufend && *t != *s;)
3196 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3200 else if (*s == 'q') {
3203 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3206 /* skip q//-like construct */
3208 char open, close, term;
3211 while (t < PL_bufend && isSPACE(*t))
3213 /* check for q => */
3214 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3215 OPERATOR(HASHBRACK);
3219 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3223 for (t++; t < PL_bufend; t++) {
3224 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3226 else if (*t == open)
3230 for (t++; t < PL_bufend; t++) {
3231 if (*t == '\\' && t+1 < PL_bufend)
3233 else if (*t == close && --brackets <= 0)
3235 else if (*t == open)
3242 /* skip plain q word */
3243 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3246 else if (isALNUM_lazy_if(t,UTF)) {
3248 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3251 while (t < PL_bufend && isSPACE(*t))
3253 /* if comma follows first term, call it an anon hash */
3254 /* XXX it could be a comma expression with loop modifiers */
3255 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3256 || (*t == '=' && t[1] == '>')))
3257 OPERATOR(HASHBRACK);
3258 if (PL_expect == XREF)
3261 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3267 yylval.ival = CopLINE(PL_curcop);
3268 if (isSPACE(*s) || *s == '#')
3269 PL_copline = NOLINE; /* invalidate current command line number */
3274 if (PL_lex_brackets <= 0)
3275 yyerror("Unmatched right curly bracket");
3277 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3278 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3279 PL_lex_formbrack = 0;
3280 if (PL_lex_state == LEX_INTERPNORMAL) {
3281 if (PL_lex_brackets == 0) {
3282 if (PL_expect & XFAKEBRACK) {
3283 PL_expect &= XENUMMASK;
3284 PL_lex_state = LEX_INTERPEND;
3286 return yylex(); /* ignore fake brackets */
3288 if (*s == '-' && s[1] == '>')
3289 PL_lex_state = LEX_INTERPENDMAYBE;
3290 else if (*s != '[' && *s != '{')
3291 PL_lex_state = LEX_INTERPEND;
3294 if (PL_expect & XFAKEBRACK) {
3295 PL_expect &= XENUMMASK;
3297 return yylex(); /* ignore fake brackets */
3307 if (PL_expect == XOPERATOR) {
3308 if (ckWARN(WARN_SEMICOLON)
3309 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3311 CopLINE_dec(PL_curcop);
3312 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3313 CopLINE_inc(PL_curcop);
3318 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3320 PL_expect = XOPERATOR;
3321 force_ident(PL_tokenbuf, '&');
3325 yylval.ival = (OPpENTERSUB_AMPER<<8);
3344 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3345 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3347 if (PL_expect == XSTATE && isALPHA(tmp) &&
3348 (s == PL_linestart+1 || s[-2] == '\n') )
3350 if (PL_in_eval && !PL_rsfp) {
3355 if (strnEQ(s,"=cut",4)) {
3369 PL_doextract = TRUE;
3372 if (PL_lex_brackets < PL_lex_formbrack) {
3374 #ifdef PERL_STRICT_CR
3375 for (t = s; SPACE_OR_TAB(*t); t++) ;
3377 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3379 if (*t == '\n' || *t == '#') {
3397 if (PL_expect != XOPERATOR) {
3398 if (s[1] != '<' && !strchr(s,'>'))
3401 s = scan_heredoc(s);
3403 s = scan_inputsymbol(s);
3404 TERM(sublex_start());
3409 SHop(OP_LEFT_SHIFT);
3423 SHop(OP_RIGHT_SHIFT);
3432 if (PL_expect == XOPERATOR) {
3433 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3436 return ','; /* grandfather non-comma-format format */
3440 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3441 PL_tokenbuf[0] = '@';
3442 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3443 sizeof PL_tokenbuf - 1, FALSE);
3444 if (PL_expect == XOPERATOR)
3445 no_op("Array length", s);
3446 if (!PL_tokenbuf[1])
3448 PL_expect = XOPERATOR;
3449 PL_pending_ident = '#';
3453 PL_tokenbuf[0] = '$';
3454 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3455 sizeof PL_tokenbuf - 1, FALSE);
3456 if (PL_expect == XOPERATOR)
3458 if (!PL_tokenbuf[1]) {
3460 yyerror("Final $ should be \\$ or $name");
3464 /* This kludge not intended to be bulletproof. */
3465 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3466 yylval.opval = newSVOP(OP_CONST, 0,
3467 newSViv(PL_compiling.cop_arybase));
3468 yylval.opval->op_private = OPpCONST_ARYBASE;
3474 if (PL_lex_state == LEX_NORMAL)
3477 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3480 PL_tokenbuf[0] = '@';
3481 if (ckWARN(WARN_SYNTAX)) {
3483 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3486 PL_bufptr = skipspace(PL_bufptr);
3487 while (t < PL_bufend && *t != ']')
3489 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3490 "Multidimensional syntax %.*s not supported",
3491 (t - PL_bufptr) + 1, PL_bufptr);
3495 else if (*s == '{') {
3496 PL_tokenbuf[0] = '%';
3497 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3498 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3500 char tmpbuf[sizeof PL_tokenbuf];
3502 for (t++; isSPACE(*t); t++) ;
3503 if (isIDFIRST_lazy_if(t,UTF)) {
3504 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3505 for (; isSPACE(*t); t++) ;
3506 if (*t == ';' && get_cv(tmpbuf, FALSE))
3507 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3508 "You need to quote \"%s\"", tmpbuf);
3514 PL_expect = XOPERATOR;
3515 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3516 bool islop = (PL_last_lop == PL_oldoldbufptr);
3517 if (!islop || PL_last_lop_op == OP_GREPSTART)
3518 PL_expect = XOPERATOR;
3519 else if (strchr("$@\"'`q", *s))
3520 PL_expect = XTERM; /* e.g. print $fh "foo" */
3521 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3522 PL_expect = XTERM; /* e.g. print $fh &sub */
3523 else if (isIDFIRST_lazy_if(s,UTF)) {
3524 char tmpbuf[sizeof PL_tokenbuf];
3525 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3526 if ((tmp = keyword(tmpbuf, len))) {
3527 /* binary operators exclude handle interpretations */
3539 PL_expect = XTERM; /* e.g. print $fh length() */
3544 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3545 if (gv && GvCVu(gv))
3546 PL_expect = XTERM; /* e.g. print $fh subr() */
3549 else if (isDIGIT(*s))
3550 PL_expect = XTERM; /* e.g. print $fh 3 */
3551 else if (*s == '.' && isDIGIT(s[1]))
3552 PL_expect = XTERM; /* e.g. print $fh .3 */
3553 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3554 PL_expect = XTERM; /* e.g. print $fh -1 */
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]) {
3569 if (PL_lex_state == LEX_NORMAL)
3571 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3573 PL_tokenbuf[0] = '%';
3575 /* Warn about @ where they meant $. */
3576 if (ckWARN(WARN_SYNTAX)) {
3577 if (*s == '[' || *s == '{') {
3579 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3581 if (*t == '}' || *t == ']') {
3583 PL_bufptr = skipspace(PL_bufptr);
3584 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3585 "Scalar value %.*s better written as $%.*s",
3586 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3591 PL_pending_ident = '@';
3594 case '/': /* may either be division or pattern */
3595 case '?': /* may either be conditional or pattern */
3596 if (PL_expect != XOPERATOR) {
3597 /* Disable warning on "study /blah/" */
3598 if (PL_oldoldbufptr == PL_last_uni
3599 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3600 || memNE(PL_last_uni, "study", 5)
3601 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3603 s = scan_pat(s,OP_MATCH);
3604 TERM(sublex_start());
3612 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3613 #ifdef PERL_STRICT_CR
3616 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3618 && (s == PL_linestart || s[-1] == '\n') )
3620 PL_lex_formbrack = 0;
3624 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3630 yylval.ival = OPf_SPECIAL;
3636 if (PL_expect != XOPERATOR)
3641 case '0': case '1': case '2': case '3': case '4':
3642 case '5': case '6': case '7': case '8': case '9':
3643 s = scan_num(s, &yylval);
3644 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3645 "### Saw number in '%s'\n", s);
3647 if (PL_expect == XOPERATOR)
3652 s = scan_str(s,FALSE,FALSE);
3653 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3654 "### Saw string before '%s'\n", s);
3656 if (PL_expect == XOPERATOR) {
3657 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3660 return ','; /* grandfather non-comma-format format */
3666 missingterm((char*)0);
3667 yylval.ival = OP_CONST;
3668 TERM(sublex_start());
3671 s = scan_str(s,FALSE,FALSE);
3672 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3673 "### Saw string before '%s'\n", s);
3675 if (PL_expect == XOPERATOR) {
3676 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3679 return ','; /* grandfather non-comma-format format */
3685 missingterm((char*)0);
3686 yylval.ival = OP_CONST;
3687 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3688 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3689 yylval.ival = OP_STRINGIFY;
3693 TERM(sublex_start());
3696 s = scan_str(s,FALSE,FALSE);
3697 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3698 "### Saw backtick string before '%s'\n", s);
3700 if (PL_expect == XOPERATOR)
3701 no_op("Backticks",s);
3703 missingterm((char*)0);
3704 yylval.ival = OP_BACKTICK;
3706 TERM(sublex_start());
3710 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3711 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3713 if (PL_expect == XOPERATOR)
3714 no_op("Backslash",s);
3718 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3722 while (isDIGIT(*start) || *start == '_')
3724 if (*start == '.' && isDIGIT(start[1])) {
3725 s = scan_num(s, &yylval);
3728 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3729 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) {
3733 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3736 s = scan_num(s, &yylval);
3743 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3783 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3785 /* Some keywords can be followed by any delimiter, including ':' */
3786 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3787 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3788 (PL_tokenbuf[0] == 'q' &&
3789 strchr("qwxr", PL_tokenbuf[1])))));
3791 /* x::* is just a word, unless x is "CORE" */
3792 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3796 while (d < PL_bufend && isSPACE(*d))
3797 d++; /* no comments skipped here, or s### is misparsed */
3799 /* Is this a label? */
3800 if (!tmp && PL_expect == XSTATE
3801 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3803 yylval.pval = savepv(PL_tokenbuf);
3808 /* Check for keywords */
3809 tmp = keyword(PL_tokenbuf, len);
3811 /* Is this a word before a => operator? */
3812 if (*d == '=' && d[1] == '>') {
3814 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3815 yylval.opval->op_private = OPpCONST_BARE;
3816 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3817 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3821 if (tmp < 0) { /* second-class keyword? */
3822 GV *ogv = Nullgv; /* override (winner) */
3823 GV *hgv = Nullgv; /* hidden (loser) */
3824 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3826 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3829 if (GvIMPORTED_CV(gv))
3831 else if (! CvMETHOD(cv))
3835 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3836 (gv = *gvp) != (GV*)&PL_sv_undef &&
3837 GvCVu(gv) && GvIMPORTED_CV(gv))
3844 tmp = 0; /* overridden by import or by GLOBAL */
3847 && -tmp==KEY_lock /* XXX generalizable kludge */
3849 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3851 tmp = 0; /* any sub overrides "weak" keyword */
3853 else { /* no override */
3855 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3856 Perl_warner(aTHX_ packWARN(WARN_MISC),
3857 "dump() better written as CORE::dump()");
3861 if (ckWARN(WARN_AMBIGUOUS) && hgv
3862 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3863 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3864 "Ambiguous call resolved as CORE::%s(), %s",
3865 GvENAME(hgv), "qualify as such or use &");
3872 default: /* not a keyword */
3876 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3878 /* Get the rest if it looks like a package qualifier */
3880 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3882 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3885 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3886 *s == '\'' ? "'" : "::");
3891 if (PL_expect == XOPERATOR) {
3892 if (PL_bufptr == PL_linestart) {
3893 CopLINE_dec(PL_curcop);
3894 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3895 CopLINE_inc(PL_curcop);
3898 no_op("Bareword",s);
3901 /* Look for a subroutine with this name in current package,
3902 unless name is "Foo::", in which case Foo is a bearword
3903 (and a package name). */
3906 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3908 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3909 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3910 "Bareword \"%s\" refers to nonexistent package",
3913 PL_tokenbuf[len] = '\0';
3920 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3923 /* if we saw a global override before, get the right name */
3926 sv = newSVpvn("CORE::GLOBAL::",14);
3927 sv_catpv(sv,PL_tokenbuf);
3930 sv = newSVpv(PL_tokenbuf,0);
3932 /* Presume this is going to be a bareword of some sort. */
3935 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3936 yylval.opval->op_private = OPpCONST_BARE;
3937 /* UTF-8 package name? */
3938 if (UTF && !IN_BYTES &&
3939 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3942 /* And if "Foo::", then that's what it certainly is. */
3947 /* See if it's the indirect object for a list operator. */
3949 if (PL_oldoldbufptr &&
3950 PL_oldoldbufptr < PL_bufptr &&
3951 (PL_oldoldbufptr == PL_last_lop
3952 || PL_oldoldbufptr == PL_last_uni) &&
3953 /* NO SKIPSPACE BEFORE HERE! */
3954 (PL_expect == XREF ||
3955 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3957 bool immediate_paren = *s == '(';
3959 /* (Now we can afford to cross potential line boundary.) */
3962 /* Two barewords in a row may indicate method call. */
3964 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3967 /* If not a declared subroutine, it's an indirect object. */
3968 /* (But it's an indir obj regardless for sort.) */
3970 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3971 ((!gv || !GvCVu(gv)) &&
3972 (PL_last_lop_op != OP_MAPSTART &&
3973 PL_last_lop_op != OP_GREPSTART))))
3975 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3980 PL_expect = XOPERATOR;
3983 /* Is this a word before a => operator? */
3984 if (*s == '=' && s[1] == '>' && !pkgname) {
3986 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3987 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3988 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3992 /* If followed by a paren, it's certainly a subroutine. */
3995 if (gv && GvCVu(gv)) {
3996 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3997 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4002 PL_nextval[PL_nexttoke].opval = yylval.opval;
4003 PL_expect = XOPERATOR;
4009 /* If followed by var or block, call it a method (unless sub) */
4011 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4012 PL_last_lop = PL_oldbufptr;
4013 PL_last_lop_op = OP_METHOD;
4017 /* If followed by a bareword, see if it looks like indir obj. */
4020 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4021 && (tmp = intuit_method(s,gv)))
4024 /* Not a method, so call it a subroutine (if defined) */
4026 if (gv && GvCVu(gv)) {
4028 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4029 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4030 "Ambiguous use of -%s resolved as -&%s()",
4031 PL_tokenbuf, PL_tokenbuf);
4032 /* Check for a constant sub */
4034 if ((sv = cv_const_sv(cv))) {
4036 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4037 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4038 yylval.opval->op_private = 0;
4042 /* Resolve to GV now. */
4043 op_free(yylval.opval);
4044 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4045 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4046 PL_last_lop = PL_oldbufptr;
4047 PL_last_lop_op = OP_ENTERSUB;
4048 /* Is there a prototype? */
4051 char *proto = SvPV((SV*)cv, len);
4054 if (strEQ(proto, "$"))
4056 while (*proto == ';')
4058 if (*proto == '&' && *s == '{') {
4059 sv_setpv(PL_subname, PL_curstash ?
4060 "__ANON__" : "__ANON__::__ANON__");
4064 PL_nextval[PL_nexttoke].opval = yylval.opval;
4070 /* Call it a bare word */
4072 if (PL_hints & HINT_STRICT_SUBS)
4073 yylval.opval->op_private |= OPpCONST_STRICT;
4076 if (ckWARN(WARN_RESERVED)) {
4077 if (lastchar != '-') {
4078 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4079 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4080 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4087 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4088 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4089 "Operator or semicolon missing before %c%s",
4090 lastchar, PL_tokenbuf);
4091 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4092 "Ambiguous use of %c resolved as operator %c",
4093 lastchar, lastchar);
4099 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4100 newSVpv(CopFILE(PL_curcop),0));
4104 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4105 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4108 case KEY___PACKAGE__:
4109 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4111 ? newSVsv(PL_curstname)
4120 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4121 char *pname = "main";
4122 if (PL_tokenbuf[2] == 'D')
4123 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4124 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4127 GvIOp(gv) = newIO();
4128 IoIFP(GvIOp(gv)) = PL_rsfp;
4129 #if defined(HAS_FCNTL) && defined(F_SETFD)
4131 int fd = PerlIO_fileno(PL_rsfp);
4132 fcntl(fd,F_SETFD,fd >= 3);
4135 /* Mark this internal pseudo-handle as clean */
4136 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4138 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4139 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4140 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4142 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4143 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4144 /* if the script was opened in binmode, we need to revert
4145 * it to text mode for compatibility; but only iff it has CRs
4146 * XXX this is a questionable hack at best. */
4147 if (PL_bufend-PL_bufptr > 2
4148 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4151 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4152 loc = PerlIO_tell(PL_rsfp);
4153 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4156 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4158 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4159 #endif /* NETWARE */
4160 #ifdef PERLIO_IS_STDIO /* really? */
4161 # if defined(__BORLANDC__)
4162 /* XXX see note in do_binmode() */
4163 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4167 PerlIO_seek(PL_rsfp, loc, 0);
4171 #ifdef PERLIO_LAYERS
4174 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4175 else if (PL_encoding) {
4182 XPUSHs(PL_encoding);
4184 call_method("name", G_SCALAR);
4188 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4189 Perl_form(aTHX_ ":encoding(%"SVf")",
4207 if (PL_expect == XSTATE) {
4214 if (*s == ':' && s[1] == ':') {
4217 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4218 if (!(tmp = keyword(PL_tokenbuf, len)))
4219 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4233 LOP(OP_ACCEPT,XTERM);
4239 LOP(OP_ATAN2,XTERM);
4245 LOP(OP_BINMODE,XTERM);
4248 LOP(OP_BLESS,XTERM);
4257 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4274 if (!PL_cryptseen) {
4275 PL_cryptseen = TRUE;
4279 LOP(OP_CRYPT,XTERM);
4282 LOP(OP_CHMOD,XTERM);
4285 LOP(OP_CHOWN,XTERM);
4288 LOP(OP_CONNECT,XTERM);
4304 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4308 PL_hints |= HINT_BLOCK_SCOPE;
4318 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4319 LOP(OP_DBMOPEN,XTERM);
4325 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4332 yylval.ival = CopLINE(PL_curcop);
4346 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4347 UNIBRACK(OP_ENTEREVAL);
4362 case KEY_endhostent:
4368 case KEY_endservent:
4371 case KEY_endprotoent:
4382 yylval.ival = CopLINE(PL_curcop);
4384 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4386 if ((PL_bufend - p) >= 3 &&
4387 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4389 else if ((PL_bufend - p) >= 4 &&
4390 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4393 if (isIDFIRST_lazy_if(p,UTF)) {
4394 p = scan_ident(p, PL_bufend,
4395 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4399 Perl_croak(aTHX_ "Missing $ on loop variable");
4404 LOP(OP_FORMLINE,XTERM);
4410 LOP(OP_FCNTL,XTERM);
4416 LOP(OP_FLOCK,XTERM);
4425 LOP(OP_GREPSTART, XREF);
4428 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4443 case KEY_getpriority:
4444 LOP(OP_GETPRIORITY,XTERM);
4446 case KEY_getprotobyname:
4449 case KEY_getprotobynumber:
4450 LOP(OP_GPBYNUMBER,XTERM);
4452 case KEY_getprotoent:
4464 case KEY_getpeername:
4465 UNI(OP_GETPEERNAME);
4467 case KEY_gethostbyname:
4470 case KEY_gethostbyaddr:
4471 LOP(OP_GHBYADDR,XTERM);
4473 case KEY_gethostent:
4476 case KEY_getnetbyname:
4479 case KEY_getnetbyaddr:
4480 LOP(OP_GNBYADDR,XTERM);
4485 case KEY_getservbyname:
4486 LOP(OP_GSBYNAME,XTERM);
4488 case KEY_getservbyport:
4489 LOP(OP_GSBYPORT,XTERM);
4491 case KEY_getservent:
4494 case KEY_getsockname:
4495 UNI(OP_GETSOCKNAME);
4497 case KEY_getsockopt:
4498 LOP(OP_GSOCKOPT,XTERM);
4520 yylval.ival = CopLINE(PL_curcop);
4524 LOP(OP_INDEX,XTERM);
4530 LOP(OP_IOCTL,XTERM);
4542 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4574 LOP(OP_LISTEN,XTERM);
4583 s = scan_pat(s,OP_MATCH);
4584 TERM(sublex_start());
4587 LOP(OP_MAPSTART, XREF);
4590 LOP(OP_MKDIR,XTERM);
4593 LOP(OP_MSGCTL,XTERM);
4596 LOP(OP_MSGGET,XTERM);
4599 LOP(OP_MSGRCV,XTERM);
4602 LOP(OP_MSGSND,XTERM);
4608 if (isIDFIRST_lazy_if(s,UTF)) {
4609 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4610 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4612 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4613 if (!PL_in_my_stash) {
4616 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4624 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4631 if (PL_expect != XSTATE)
4632 yyerror("\"no\" not allowed in expression");
4633 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4634 s = force_version(s, FALSE);
4639 if (*s == '(' || (s = skipspace(s), *s == '('))
4646 if (isIDFIRST_lazy_if(s,UTF)) {
4648 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4650 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4652 && !(t[0] == '=' && t[1] == '>')
4654 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4655 "Precedence problem: open %.*s should be open(%.*s)",
4656 d - s, s, d - s, s);
4662 yylval.ival = OP_OR;
4672 LOP(OP_OPEN_DIR,XTERM);
4675 checkcomma(s,PL_tokenbuf,"filehandle");
4679 checkcomma(s,PL_tokenbuf,"filehandle");
4698 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4702 LOP(OP_PIPE_OP,XTERM);
4705 s = scan_str(s,FALSE,FALSE);
4707 missingterm((char*)0);
4708 yylval.ival = OP_CONST;
4709 TERM(sublex_start());
4715 s = scan_str(s,FALSE,FALSE);
4717 missingterm((char*)0);
4719 if (SvCUR(PL_lex_stuff)) {
4722 d = SvPV_force(PL_lex_stuff, len);
4725 for (; isSPACE(*d) && len; --len, ++d) ;
4728 if (!warned && ckWARN(WARN_QW)) {
4729 for (; !isSPACE(*d) && len; --len, ++d) {
4731 Perl_warner(aTHX_ packWARN(WARN_QW),
4732 "Possible attempt to separate words with commas");
4735 else if (*d == '#') {
4736 Perl_warner(aTHX_ packWARN(WARN_QW),
4737 "Possible attempt to put comments in qw() list");
4743 for (; !isSPACE(*d) && len; --len, ++d) ;
4745 sv = newSVpvn(b, d-b);
4746 if (DO_UTF8(PL_lex_stuff))
4748 words = append_elem(OP_LIST, words,
4749 newSVOP(OP_CONST, 0, tokeq(sv)));
4753 PL_nextval[PL_nexttoke].opval = words;
4758 SvREFCNT_dec(PL_lex_stuff);
4759 PL_lex_stuff = Nullsv;
4765 s = scan_str(s,FALSE,FALSE);
4767 missingterm((char*)0);
4768 yylval.ival = OP_STRINGIFY;
4769 if (SvIVX(PL_lex_stuff) == '\'')
4770 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4771 TERM(sublex_start());
4774 s = scan_pat(s,OP_QR);
4775 TERM(sublex_start());
4778 s = scan_str(s,FALSE,FALSE);
4780 missingterm((char*)0);
4781 yylval.ival = OP_BACKTICK;
4783 TERM(sublex_start());
4791 s = force_version(s, FALSE);
4793 else if (*s != 'v' || !isDIGIT(s[1])
4794 || (s = force_version(s, TRUE), *s == 'v'))
4796 *PL_tokenbuf = '\0';
4797 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4798 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4799 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4801 yyerror("<> should be quotes");
4809 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4813 LOP(OP_RENAME,XTERM);
4822 LOP(OP_RINDEX,XTERM);
4845 LOP(OP_REVERSE,XTERM);
4856 TERM(sublex_start());
4858 TOKEN(1); /* force error */
4867 LOP(OP_SELECT,XTERM);
4873 LOP(OP_SEMCTL,XTERM);
4876 LOP(OP_SEMGET,XTERM);
4879 LOP(OP_SEMOP,XTERM);
4885 LOP(OP_SETPGRP,XTERM);
4887 case KEY_setpriority:
4888 LOP(OP_SETPRIORITY,XTERM);
4890 case KEY_sethostent:
4896 case KEY_setservent:
4899 case KEY_setprotoent:
4909 LOP(OP_SEEKDIR,XTERM);
4911 case KEY_setsockopt:
4912 LOP(OP_SSOCKOPT,XTERM);
4918 LOP(OP_SHMCTL,XTERM);
4921 LOP(OP_SHMGET,XTERM);
4924 LOP(OP_SHMREAD,XTERM);
4927 LOP(OP_SHMWRITE,XTERM);
4930 LOP(OP_SHUTDOWN,XTERM);
4939 LOP(OP_SOCKET,XTERM);
4941 case KEY_socketpair:
4942 LOP(OP_SOCKPAIR,XTERM);
4945 checkcomma(s,PL_tokenbuf,"subroutine name");
4947 if (*s == ';' || *s == ')') /* probably a close */
4948 Perl_croak(aTHX_ "sort is now a reserved word");
4950 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4954 LOP(OP_SPLIT,XTERM);
4957 LOP(OP_SPRINTF,XTERM);
4960 LOP(OP_SPLICE,XTERM);
4975 LOP(OP_SUBSTR,XTERM);
4981 char tmpbuf[sizeof PL_tokenbuf];
4982 SSize_t tboffset = 0;
4983 expectation attrful;
4984 bool have_name, have_proto, bad_proto;
4989 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4990 (*s == ':' && s[1] == ':'))
4993 attrful = XATTRBLOCK;
4994 /* remember buffer pos'n for later force_word */
4995 tboffset = s - PL_oldbufptr;
4996 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4997 if (strchr(tmpbuf, ':'))
4998 sv_setpv(PL_subname, tmpbuf);
5000 sv_setsv(PL_subname,PL_curstname);
5001 sv_catpvn(PL_subname,"::",2);
5002 sv_catpvn(PL_subname,tmpbuf,len);
5009 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5010 PL_expect = XTERMBLOCK;
5011 attrful = XATTRTERM;
5012 sv_setpv(PL_subname,"?");
5016 if (key == KEY_format) {
5018 PL_lex_formbrack = PL_lex_brackets + 1;
5020 (void) force_word(PL_oldbufptr + tboffset, WORD,
5025 /* Look for a prototype */
5029 s = scan_str(s,FALSE,FALSE);
5031 Perl_croak(aTHX_ "Prototype not terminated");
5032 /* strip spaces and check for bad characters */
5033 d = SvPVX(PL_lex_stuff);
5036 for (p = d; *p; ++p) {
5039 if (!strchr("$@%*;[]&\\", *p))
5044 if (bad_proto && ckWARN(WARN_SYNTAX))
5045 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5046 "Illegal character in prototype for %"SVf" : %s",
5048 SvCUR(PL_lex_stuff) = tmp;
5056 if (*s == ':' && s[1] != ':')
5057 PL_expect = attrful;
5058 else if (!have_name && *s != '{' && key == KEY_sub)
5059 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5062 PL_nextval[PL_nexttoke].opval =
5063 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5064 PL_lex_stuff = Nullsv;
5068 sv_setpv(PL_subname,
5069 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5072 (void) force_word(PL_oldbufptr + tboffset, WORD,
5081 LOP(OP_SYSTEM,XREF);
5084 LOP(OP_SYMLINK,XTERM);
5087 LOP(OP_SYSCALL,XTERM);
5090 LOP(OP_SYSOPEN,XTERM);
5093 LOP(OP_SYSSEEK,XTERM);
5096 LOP(OP_SYSREAD,XTERM);
5099 LOP(OP_SYSWRITE,XTERM);
5103 TERM(sublex_start());
5124 LOP(OP_TRUNCATE,XTERM);
5136 yylval.ival = CopLINE(PL_curcop);
5140 yylval.ival = CopLINE(PL_curcop);
5144 LOP(OP_UNLINK,XTERM);
5150 LOP(OP_UNPACK,XTERM);
5153 LOP(OP_UTIME,XTERM);
5159 LOP(OP_UNSHIFT,XTERM);
5162 if (PL_expect != XSTATE)
5163 yyerror("\"use\" not allowed in expression");
5165 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5166 s = force_version(s, TRUE);
5167 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5168 PL_nextval[PL_nexttoke].opval = Nullop;
5171 else if (*s == 'v') {
5172 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5173 s = force_version(s, FALSE);
5177 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5178 s = force_version(s, FALSE);
5190 yylval.ival = CopLINE(PL_curcop);
5194 PL_hints |= HINT_BLOCK_SCOPE;
5201 LOP(OP_WAITPID,XTERM);
5210 ctl_l[0] = toCTRL('L');
5212 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5215 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5220 if (PL_expect == XOPERATOR)
5226 yylval.ival = OP_XOR;
5231 TERM(sublex_start());
5236 #pragma segment Main
5240 S_pending_ident(pTHX)
5243 register I32 tmp = 0;
5244 /* pit holds the identifier we read and pending_ident is reset */
5245 char pit = PL_pending_ident;
5246 PL_pending_ident = 0;
5248 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5249 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5251 /* if we're in a my(), we can't allow dynamics here.
5252 $foo'bar has already been turned into $foo::bar, so
5253 just check for colons.
5255 if it's a legal name, the OP is a PADANY.
5258 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5259 if (strchr(PL_tokenbuf,':'))
5260 yyerror(Perl_form(aTHX_ "No package name allowed for "
5261 "variable %s in \"our\"",
5263 tmp = allocmy(PL_tokenbuf);
5266 if (strchr(PL_tokenbuf,':'))
5267 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5269 yylval.opval = newOP(OP_PADANY, 0);
5270 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5276 build the ops for accesses to a my() variable.
5278 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5279 then used in a comparison. This catches most, but not
5280 all cases. For instance, it catches
5281 sort { my($a); $a <=> $b }
5283 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5284 (although why you'd do that is anyone's guess).
5287 if (!strchr(PL_tokenbuf,':')) {
5288 #ifdef USE_5005THREADS
5289 /* Check for single character per-thread SVs */
5290 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
5291 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
5292 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
5294 yylval.opval = newOP(OP_THREADSV, 0);
5295 yylval.opval->op_targ = tmp;
5298 #endif /* USE_5005THREADS */
5300 tmp = pad_findmy(PL_tokenbuf);
5301 if (tmp != NOT_IN_PAD) {
5302 /* might be an "our" variable" */
5303 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5304 /* build ops for a bareword */
5305 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5306 sv_catpvn(sym, "::", 2);
5307 sv_catpv(sym, PL_tokenbuf+1);
5308 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5309 yylval.opval->op_private = OPpCONST_ENTERED;
5310 gv_fetchpv(SvPVX(sym),
5312 ? (GV_ADDMULTI | GV_ADDINEVAL)
5315 ((PL_tokenbuf[0] == '$') ? SVt_PV
5316 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5321 /* if it's a sort block and they're naming $a or $b */
5322 if (PL_last_lop_op == OP_SORT &&
5323 PL_tokenbuf[0] == '$' &&
5324 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5327 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5328 d < PL_bufend && *d != '\n';
5331 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5332 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5338 yylval.opval = newOP(OP_PADANY, 0);
5339 yylval.opval->op_targ = tmp;
5345 Whine if they've said @foo in a doublequoted string,
5346 and @foo isn't a variable we can find in the symbol
5349 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5350 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5351 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5352 && ckWARN(WARN_AMBIGUOUS))
5354 /* Downgraded from fatal to warning 20000522 mjd */
5355 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5356 "Possible unintended interpolation of %s in string",
5361 /* build ops for a bareword */
5362 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5363 yylval.opval->op_private = OPpCONST_ENTERED;
5364 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5365 ((PL_tokenbuf[0] == '$') ? SVt_PV
5366 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5372 Perl_keyword(pTHX_ register char *d, I32 len)
5377 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5378 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5379 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5380 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5381 if (strEQ(d,"__END__")) return KEY___END__;
5385 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5390 if (strEQ(d,"and")) return -KEY_and;
5391 if (strEQ(d,"abs")) return -KEY_abs;
5394 if (strEQ(d,"alarm")) return -KEY_alarm;
5395 if (strEQ(d,"atan2")) return -KEY_atan2;
5398 if (strEQ(d,"accept")) return -KEY_accept;
5403 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5406 if (strEQ(d,"bless")) return -KEY_bless;
5407 if (strEQ(d,"bind")) return -KEY_bind;
5408 if (strEQ(d,"binmode")) return -KEY_binmode;
5411 if (strEQ(d,"CORE")) return -KEY_CORE;
5412 if (strEQ(d,"CHECK")) return KEY_CHECK;
5417 if (strEQ(d,"cmp")) return -KEY_cmp;
5418 if (strEQ(d,"chr")) return -KEY_chr;
5419 if (strEQ(d,"cos")) return -KEY_cos;
5422 if (strEQ(d,"chop")) return -KEY_chop;
5425 if (strEQ(d,"close")) return -KEY_close;
5426 if (strEQ(d,"chdir")) return -KEY_chdir;
5427 if (strEQ(d,"chomp")) return -KEY_chomp;
5428 if (strEQ(d,"chmod")) return -KEY_chmod;
5429 if (strEQ(d,"chown")) return -KEY_chown;
5430 if (strEQ(d,"crypt")) return -KEY_crypt;
5433 if (strEQ(d,"chroot")) return -KEY_chroot;
5434 if (strEQ(d,"caller")) return -KEY_caller;
5437 if (strEQ(d,"connect")) return -KEY_connect;
5440 if (strEQ(d,"closedir")) return -KEY_closedir;
5441 if (strEQ(d,"continue")) return -KEY_continue;
5446 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5451 if (strEQ(d,"do")) return KEY_do;
5454 if (strEQ(d,"die")) return -KEY_die;
5457 if (strEQ(d,"dump")) return -KEY_dump;
5460 if (strEQ(d,"delete")) return KEY_delete;
5463 if (strEQ(d,"defined")) return KEY_defined;
5464 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5467 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5472 if (strEQ(d,"END")) return KEY_END;
5477 if (strEQ(d,"eq")) return -KEY_eq;
5480 if (strEQ(d,"eof")) return -KEY_eof;
5481 if (strEQ(d,"exp")) return -KEY_exp;
5484 if (strEQ(d,"else")) return KEY_else;
5485 if (strEQ(d,"exit")) return -KEY_exit;
5486 if (strEQ(d,"eval")) return KEY_eval;
5487 if (strEQ(d,"exec")) return -KEY_exec;
5488 if (strEQ(d,"each")) return -KEY_each;
5491 if (strEQ(d,"elsif")) return KEY_elsif;
5494 if (strEQ(d,"exists")) return KEY_exists;
5495 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5498 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5499 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5502 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5505 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5506 if (strEQ(d,"endservent")) return -KEY_endservent;
5509 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5516 if (strEQ(d,"for")) return KEY_for;
5519 if (strEQ(d,"fork")) return -KEY_fork;
5522 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5523 if (strEQ(d,"flock")) return -KEY_flock;
5526 if (strEQ(d,"format")) return KEY_format;
5527 if (strEQ(d,"fileno")) return -KEY_fileno;
5530 if (strEQ(d,"foreach")) return KEY_foreach;
5533 if (strEQ(d,"formline")) return -KEY_formline;
5538 if (strnEQ(d,"get",3)) {
5543 if (strEQ(d,"ppid")) return -KEY_getppid;
5544 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5547 if (strEQ(d,"pwent")) return -KEY_getpwent;
5548 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5549 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5552 if (strEQ(d,"peername")) return -KEY_getpeername;
5553 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5554 if (strEQ(d,"priority")) return -KEY_getpriority;
5557 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5560 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5564 else if (*d == 'h') {
5565 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5566 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5567 if (strEQ(d,"hostent")) return -KEY_gethostent;
5569 else if (*d == 'n') {
5570 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5571 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5572 if (strEQ(d,"netent")) return -KEY_getnetent;
5574 else if (*d == 's') {
5575 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5576 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5577 if (strEQ(d,"servent")) return -KEY_getservent;
5578 if (strEQ(d,"sockname")) return -KEY_getsockname;
5579 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5581 else if (*d == 'g') {
5582 if (strEQ(d,"grent")) return -KEY_getgrent;
5583 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5584 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5586 else if (*d == 'l') {
5587 if (strEQ(d,"login")) return -KEY_getlogin;
5589 else if (strEQ(d,"c")) return -KEY_getc;
5594 if (strEQ(d,"gt")) return -KEY_gt;
5595 if (strEQ(d,"ge")) return -KEY_ge;
5598 if (strEQ(d,"grep")) return KEY_grep;
5599 if (strEQ(d,"goto")) return KEY_goto;
5600 if (strEQ(d,"glob")) return KEY_glob;
5603 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5608 if (strEQ(d,"hex")) return -KEY_hex;
5611 if (strEQ(d,"INIT")) return KEY_INIT;
5616 if (strEQ(d,"if")) return KEY_if;
5619 if (strEQ(d,"int")) return -KEY_int;
5622 if (strEQ(d,"index")) return -KEY_index;
5623 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5628 if (strEQ(d,"join")) return -KEY_join;
5632 if (strEQ(d,"keys")) return -KEY_keys;
5633 if (strEQ(d,"kill")) return -KEY_kill;
5639 if (strEQ(d,"lt")) return -KEY_lt;
5640 if (strEQ(d,"le")) return -KEY_le;
5641 if (strEQ(d,"lc")) return -KEY_lc;
5644 if (strEQ(d,"log")) return -KEY_log;
5647 if (strEQ(d,"last")) return KEY_last;
5648 if (strEQ(d,"link")) return -KEY_link;
5649 if (strEQ(d,"lock")) return -KEY_lock;
5652 if (strEQ(d,"local")) return KEY_local;
5653 if (strEQ(d,"lstat")) return -KEY_lstat;
5656 if (strEQ(d,"length")) return -KEY_length;
5657 if (strEQ(d,"listen")) return -KEY_listen;
5660 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5663 if (strEQ(d,"localtime")) return -KEY_localtime;
5669 case 1: return KEY_m;
5671 if (strEQ(d,"my")) return KEY_my;
5674 if (strEQ(d,"map")) return KEY_map;
5677 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5680 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5681 if (strEQ(d,"msgget")) return -KEY_msgget;
5682 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5683 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5688 if (strEQ(d,"next")) return KEY_next;
5689 if (strEQ(d,"ne")) return -KEY_ne;
5690 if (strEQ(d,"not")) return -KEY_not;
5691 if (strEQ(d,"no")) return KEY_no;
5696 if (strEQ(d,"or")) return -KEY_or;
5699 if (strEQ(d,"ord")) return -KEY_ord;
5700 if (strEQ(d,"oct")) return -KEY_oct;
5701 if (strEQ(d,"our")) return KEY_our;
5704 if (strEQ(d,"open")) return -KEY_open;
5707 if (strEQ(d,"opendir")) return -KEY_opendir;
5714 if (strEQ(d,"pop")) return -KEY_pop;
5715 if (strEQ(d,"pos")) return KEY_pos;
5718 if (strEQ(d,"push")) return -KEY_push;
5719 if (strEQ(d,"pack")) return -KEY_pack;
5720 if (strEQ(d,"pipe")) return -KEY_pipe;
5723 if (strEQ(d,"print")) return KEY_print;
5726 if (strEQ(d,"printf")) return KEY_printf;
5729 if (strEQ(d,"package")) return KEY_package;
5732 if (strEQ(d,"prototype")) return KEY_prototype;
5737 if (strEQ(d,"q")) return KEY_q;
5738 if (strEQ(d,"qr")) return KEY_qr;
5739 if (strEQ(d,"qq")) return KEY_qq;
5740 if (strEQ(d,"qw")) return KEY_qw;
5741 if (strEQ(d,"qx")) return KEY_qx;
5743 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5748 if (strEQ(d,"ref")) return -KEY_ref;
5751 if (strEQ(d,"read")) return -KEY_read;
5752 if (strEQ(d,"rand")) return -KEY_rand;
5753 if (strEQ(d,"recv")) return -KEY_recv;
5754 if (strEQ(d,"redo")) return KEY_redo;
5757 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5758 if (strEQ(d,"reset")) return -KEY_reset;
5761 if (strEQ(d,"return")) return KEY_return;
5762 if (strEQ(d,"rename")) return -KEY_rename;
5763 if (strEQ(d,"rindex")) return -KEY_rindex;
5766 if (strEQ(d,"require")) return KEY_require;
5767 if (strEQ(d,"reverse")) return -KEY_reverse;
5768 if (strEQ(d,"readdir")) return -KEY_readdir;
5771 if (strEQ(d,"readlink")) return -KEY_readlink;
5772 if (strEQ(d,"readline")) return -KEY_readline;
5773 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5776 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5782 case 0: return KEY_s;
5784 if (strEQ(d,"scalar")) return KEY_scalar;
5789 if (strEQ(d,"seek")) return -KEY_seek;
5790 if (strEQ(d,"send")) return -KEY_send;
5793 if (strEQ(d,"semop")) return -KEY_semop;
5796 if (strEQ(d,"select")) return -KEY_select;
5797 if (strEQ(d,"semctl")) return -KEY_semctl;
5798 if (strEQ(d,"semget")) return -KEY_semget;
5801 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5802 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5805 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5806 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5809 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5812 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5813 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5814 if (strEQ(d,"setservent")) return -KEY_setservent;
5817 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5818 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5825 if (strEQ(d,"shift")) return -KEY_shift;
5828 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5829 if (strEQ(d,"shmget")) return -KEY_shmget;
5832 if (strEQ(d,"shmread")) return -KEY_shmread;
5835 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5836 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5841 if (strEQ(d,"sin")) return -KEY_sin;
5844 if (strEQ(d,"sleep")) return -KEY_sleep;
5847 if (strEQ(d,"sort")) return KEY_sort;
5848 if (strEQ(d,"socket")) return -KEY_socket;
5849 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5852 if (strEQ(d,"split")) return KEY_split;
5853 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5854 if (strEQ(d,"splice")) return -KEY_splice;
5857 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5860 if (strEQ(d,"srand")) return -KEY_srand;
5863 if (strEQ(d,"stat")) return -KEY_stat;
5864 if (strEQ(d,"study")) return KEY_study;
5867 if (strEQ(d,"substr")) return -KEY_substr;
5868 if (strEQ(d,"sub")) return KEY_sub;
5873 if (strEQ(d,"system")) return -KEY_system;
5876 if (strEQ(d,"symlink")) return -KEY_symlink;
5877 if (strEQ(d,"syscall")) return -KEY_syscall;
5878 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5879 if (strEQ(d,"sysread")) return -KEY_sysread;
5880 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5883 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5892 if (strEQ(d,"tr")) return KEY_tr;
5895 if (strEQ(d,"tie")) return KEY_tie;
5898 if (strEQ(d,"tell")) return -KEY_tell;
5899 if (strEQ(d,"tied")) return KEY_tied;
5900 if (strEQ(d,"time")) return -KEY_time;
5903 if (strEQ(d,"times")) return -KEY_times;
5906 if (strEQ(d,"telldir")) return -KEY_telldir;
5909 if (strEQ(d,"truncate")) return -KEY_truncate;
5916 if (strEQ(d,"uc")) return -KEY_uc;
5919 if (strEQ(d,"use")) return KEY_use;
5922 if (strEQ(d,"undef")) return KEY_undef;
5923 if (strEQ(d,"until")) return KEY_until;
5924 if (strEQ(d,"untie")) return KEY_untie;
5925 if (strEQ(d,"utime")) return -KEY_utime;
5926 if (strEQ(d,"umask")) return -KEY_umask;
5929 if (strEQ(d,"unless")) return KEY_unless;
5930 if (strEQ(d,"unpack")) return -KEY_unpack;
5931 if (strEQ(d,"unlink")) return -KEY_unlink;
5934 if (strEQ(d,"unshift")) return -KEY_unshift;
5935 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5940 if (strEQ(d,"values")) return -KEY_values;
5941 if (strEQ(d,"vec")) return -KEY_vec;
5946 if (strEQ(d,"warn")) return -KEY_warn;
5947 if (strEQ(d,"wait")) return -KEY_wait;
5950 if (strEQ(d,"while")) return KEY_while;
5951 if (strEQ(d,"write")) return -KEY_write;
5954 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5957 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5962 if (len == 1) return -KEY_x;
5963 if (strEQ(d,"xor")) return -KEY_xor;
5966 if (len == 1) return KEY_y;
5975 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5979 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5980 if (ckWARN(WARN_SYNTAX)) {
5982 for (w = s+2; *w && level; w++) {
5989 for (; *w && isSPACE(*w); w++) ;
5990 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5991 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5992 "%s (...) interpreted as function",name);
5995 while (s < PL_bufend && isSPACE(*s))
5999 while (s < PL_bufend && isSPACE(*s))
6001 if (isIDFIRST_lazy_if(s,UTF)) {
6003 while (isALNUM_lazy_if(s,UTF))
6005 while (s < PL_bufend && isSPACE(*s))
6010 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6014 Perl_croak(aTHX_ "No comma allowed after %s", what);
6019 /* Either returns sv, or mortalizes sv and returns a new SV*.
6020 Best used as sv=new_constant(..., sv, ...).
6021 If s, pv are NULL, calls subroutine with one argument,
6022 and type is used with error messages only. */
6025 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6029 HV *table = GvHV(PL_hintgv); /* ^H */
6033 const char *why1, *why2, *why3;
6035 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6038 why2 = strEQ(key,"charnames")
6039 ? "(possibly a missing \"use charnames ...\")"
6041 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6042 (type ? type: "undef"), why2);
6044 /* This is convoluted and evil ("goto considered harmful")
6045 * but I do not understand the intricacies of all the different
6046 * failure modes of %^H in here. The goal here is to make
6047 * the most probable error message user-friendly. --jhi */
6052 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6053 (type ? type: "undef"), why1, why2, why3);
6055 yyerror(SvPVX(msg));
6059 cvp = hv_fetch(table, key, strlen(key), FALSE);
6060 if (!cvp || !SvOK(*cvp)) {
6063 why3 = "} is not defined";
6066 sv_2mortal(sv); /* Parent created it permanently */
6069 pv = sv_2mortal(newSVpvn(s, len));
6071 typesv = sv_2mortal(newSVpv(type, 0));
6073 typesv = &PL_sv_undef;
6075 PUSHSTACKi(PERLSI_OVERLOAD);
6087 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6091 /* Check the eval first */
6092 if (!PL_in_eval && SvTRUE(ERRSV)) {
6094 sv_catpv(ERRSV, "Propagated");
6095 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6097 res = SvREFCNT_inc(sv);
6101 (void)SvREFCNT_inc(res);
6110 why1 = "Call to &{$^H{";
6112 why3 = "}} did not return a defined value";
6121 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6123 register char *d = dest;
6124 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6127 Perl_croak(aTHX_ ident_too_long);
6128 if (isALNUM(*s)) /* UTF handled below */
6130 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6135 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6139 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6140 char *t = s + UTF8SKIP(s);
6141 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6143 if (d + (t - s) > e)
6144 Perl_croak(aTHX_ ident_too_long);
6145 Copy(s, d, t - s, char);
6158 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6168 e = d + destlen - 3; /* two-character token, ending NUL */
6170 while (isDIGIT(*s)) {
6172 Perl_croak(aTHX_ ident_too_long);
6179 Perl_croak(aTHX_ ident_too_long);
6180 if (isALNUM(*s)) /* UTF handled below */
6182 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6187 else if (*s == ':' && s[1] == ':') {
6191 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6192 char *t = s + UTF8SKIP(s);
6193 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6195 if (d + (t - s) > e)
6196 Perl_croak(aTHX_ ident_too_long);
6197 Copy(s, d, t - s, char);
6208 if (PL_lex_state != LEX_NORMAL)
6209 PL_lex_state = LEX_INTERPENDMAYBE;
6212 if (*s == '$' && s[1] &&
6213 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6226 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6231 if (isSPACE(s[-1])) {
6234 if (!SPACE_OR_TAB(ch)) {
6240 if (isIDFIRST_lazy_if(d,UTF)) {
6244 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6246 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6249 Copy(s, d, e - s, char);
6254 while ((isALNUM(*s) || *s == ':') && d < e)
6257 Perl_croak(aTHX_ ident_too_long);
6260 while (s < send && SPACE_OR_TAB(*s)) s++;
6261 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6262 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6263 const char *brack = *s == '[' ? "[...]" : "{...}";
6264 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6265 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6266 funny, dest, brack, funny, dest, brack);
6269 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6273 /* Handle extended ${^Foo} variables
6274 * 1999-02-27 mjd-perl-patch@plover.com */
6275 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6279 while (isALNUM(*s) && d < e) {
6283 Perl_croak(aTHX_ ident_too_long);
6288 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6289 PL_lex_state = LEX_INTERPEND;
6294 if (PL_lex_state == LEX_NORMAL) {
6295 if (ckWARN(WARN_AMBIGUOUS) &&
6296 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6298 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6299 "Ambiguous use of %c{%s} resolved to %c%s",
6300 funny, dest, funny, dest);
6305 s = bracket; /* let the parser handle it */
6309 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6310 PL_lex_state = LEX_INTERPEND;
6315 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6320 *pmfl |= PMf_GLOBAL;
6322 *pmfl |= PMf_CONTINUE;
6326 *pmfl |= PMf_MULTILINE;
6328 *pmfl |= PMf_SINGLELINE;
6330 *pmfl |= PMf_EXTENDED;
6334 S_scan_pat(pTHX_ char *start, I32 type)
6339 s = scan_str(start,FALSE,FALSE);
6341 Perl_croak(aTHX_ "Search pattern not terminated");
6343 pm = (PMOP*)newPMOP(type, 0);
6344 if (PL_multi_open == '?')
6345 pm->op_pmflags |= PMf_ONCE;
6347 while (*s && strchr("iomsx", *s))
6348 pmflag(&pm->op_pmflags,*s++);
6351 while (*s && strchr("iogcmsx", *s))
6352 pmflag(&pm->op_pmflags,*s++);
6354 /* issue a warning if /c is specified,but /g is not */
6355 if (ckWARN(WARN_REGEXP) &&
6356 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6358 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6361 pm->op_pmpermflags = pm->op_pmflags;
6363 PL_lex_op = (OP*)pm;
6364 yylval.ival = OP_MATCH;
6369 S_scan_subst(pTHX_ char *start)
6376 yylval.ival = OP_NULL;
6378 s = scan_str(start,FALSE,FALSE);
6381 Perl_croak(aTHX_ "Substitution pattern not terminated");
6383 if (s[-1] == PL_multi_open)
6386 first_start = PL_multi_start;
6387 s = scan_str(s,FALSE,FALSE);
6390 SvREFCNT_dec(PL_lex_stuff);
6391 PL_lex_stuff = Nullsv;
6393 Perl_croak(aTHX_ "Substitution replacement not terminated");
6395 PL_multi_start = first_start; /* so whole substitution is taken together */
6397 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6403 else if (strchr("iogcmsx", *s))
6404 pmflag(&pm->op_pmflags,*s++);
6409 /* /c is not meaningful with s/// */
6410 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6412 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6417 PL_sublex_info.super_bufptr = s;
6418 PL_sublex_info.super_bufend = PL_bufend;
6420 pm->op_pmflags |= PMf_EVAL;
6421 repl = newSVpvn("",0);
6423 sv_catpv(repl, es ? "eval " : "do ");
6424 sv_catpvn(repl, "{ ", 2);
6425 sv_catsv(repl, PL_lex_repl);
6426 sv_catpvn(repl, " };", 2);
6428 SvREFCNT_dec(PL_lex_repl);
6432 pm->op_pmpermflags = pm->op_pmflags;
6433 PL_lex_op = (OP*)pm;
6434 yylval.ival = OP_SUBST;
6439 S_scan_trans(pTHX_ char *start)
6448 yylval.ival = OP_NULL;
6450 s = scan_str(start,FALSE,FALSE);
6452 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6453 if (s[-1] == PL_multi_open)
6456 s = scan_str(s,FALSE,FALSE);
6459 SvREFCNT_dec(PL_lex_stuff);
6460 PL_lex_stuff = Nullsv;
6462 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6465 complement = del = squash = 0;
6466 while (strchr("cds", *s)) {
6468 complement = OPpTRANS_COMPLEMENT;
6470 del = OPpTRANS_DELETE;
6472 squash = OPpTRANS_SQUASH;
6476 New(803, tbl, complement&&!del?258:256, short);
6477 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6478 o->op_private = del|squash|complement|
6479 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6480 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6483 yylval.ival = OP_TRANS;
6488 S_scan_heredoc(pTHX_ register char *s)
6491 I32 op_type = OP_SCALAR;
6498 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6502 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6505 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6506 if (*peek && strchr("`'\"",*peek)) {
6509 s = delimcpy(d, e, s, PL_bufend, term, &len);
6519 if (!isALNUM_lazy_if(s,UTF))
6520 deprecate_old("bare << to mean <<\"\"");
6521 for (; isALNUM_lazy_if(s,UTF); s++) {
6526 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6527 Perl_croak(aTHX_ "Delimiter for here document is too long");
6530 len = d - PL_tokenbuf;
6531 #ifndef PERL_STRICT_CR
6532 d = strchr(s, '\r');
6536 while (s < PL_bufend) {
6542 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6551 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6556 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6557 herewas = newSVpvn(s,PL_bufend-s);
6559 s--, herewas = newSVpvn(s,d-s);
6560 s += SvCUR(herewas);
6562 tmpstr = NEWSV(87,79);
6563 sv_upgrade(tmpstr, SVt_PVIV);
6568 else if (term == '`') {
6569 op_type = OP_BACKTICK;
6570 SvIVX(tmpstr) = '\\';
6574 PL_multi_start = CopLINE(PL_curcop);
6575 PL_multi_open = PL_multi_close = '<';
6576 term = *PL_tokenbuf;
6577 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6578 char *bufptr = PL_sublex_info.super_bufptr;
6579 char *bufend = PL_sublex_info.super_bufend;
6580 char *olds = s - SvCUR(herewas);
6581 s = strchr(bufptr, '\n');
6585 while (s < bufend &&
6586 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6588 CopLINE_inc(PL_curcop);
6591 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6592 missingterm(PL_tokenbuf);
6594 sv_setpvn(herewas,bufptr,d-bufptr+1);
6595 sv_setpvn(tmpstr,d+1,s-d);
6597 sv_catpvn(herewas,s,bufend-s);
6598 (void)strcpy(bufptr,SvPVX(herewas));
6605 while (s < PL_bufend &&
6606 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6608 CopLINE_inc(PL_curcop);
6610 if (s >= PL_bufend) {
6611 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6612 missingterm(PL_tokenbuf);
6614 sv_setpvn(tmpstr,d+1,s-d);
6616 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6618 sv_catpvn(herewas,s,PL_bufend-s);
6619 sv_setsv(PL_linestr,herewas);
6620 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6621 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6622 PL_last_lop = PL_last_uni = Nullch;
6625 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6626 while (s >= PL_bufend) { /* multiple line string? */
6628 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6629 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6630 missingterm(PL_tokenbuf);
6632 CopLINE_inc(PL_curcop);
6633 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6634 PL_last_lop = PL_last_uni = Nullch;
6635 #ifndef PERL_STRICT_CR
6636 if (PL_bufend - PL_linestart >= 2) {
6637 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6638 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6640 PL_bufend[-2] = '\n';
6642 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6644 else if (PL_bufend[-1] == '\r')
6645 PL_bufend[-1] = '\n';
6647 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6648 PL_bufend[-1] = '\n';
6650 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6651 SV *sv = NEWSV(88,0);
6653 sv_upgrade(sv, SVt_PVMG);
6654 sv_setsv(sv,PL_linestr);
6657 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6659 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6662 sv_catsv(PL_linestr,herewas);
6663 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6667 sv_catsv(tmpstr,PL_linestr);
6672 PL_multi_end = CopLINE(PL_curcop);
6673 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6674 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6675 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6677 SvREFCNT_dec(herewas);
6679 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6681 else if (PL_encoding)
6682 sv_recode_to_utf8(tmpstr, PL_encoding);
6684 PL_lex_stuff = tmpstr;
6685 yylval.ival = op_type;
6690 takes: current position in input buffer
6691 returns: new position in input buffer
6692 side-effects: yylval and lex_op are set.
6697 <FH> read from filehandle
6698 <pkg::FH> read from package qualified filehandle
6699 <pkg'FH> read from package qualified filehandle
6700 <$fh> read from filehandle in $fh
6706 S_scan_inputsymbol(pTHX_ char *start)
6708 register char *s = start; /* current position in buffer */
6714 d = PL_tokenbuf; /* start of temp holding space */
6715 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6716 end = strchr(s, '\n');
6719 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6721 /* die if we didn't have space for the contents of the <>,
6722 or if it didn't end, or if we see a newline
6725 if (len >= sizeof PL_tokenbuf)
6726 Perl_croak(aTHX_ "Excessively long <> operator");
6728 Perl_croak(aTHX_ "Unterminated <> operator");
6733 Remember, only scalar variables are interpreted as filehandles by
6734 this code. Anything more complex (e.g., <$fh{$num}>) will be
6735 treated as a glob() call.
6736 This code makes use of the fact that except for the $ at the front,
6737 a scalar variable and a filehandle look the same.
6739 if (*d == '$' && d[1]) d++;
6741 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6742 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6745 /* If we've tried to read what we allow filehandles to look like, and
6746 there's still text left, then it must be a glob() and not a getline.
6747 Use scan_str to pull out the stuff between the <> and treat it
6748 as nothing more than a string.
6751 if (d - PL_tokenbuf != len) {
6752 yylval.ival = OP_GLOB;
6754 s = scan_str(start,FALSE,FALSE);
6756 Perl_croak(aTHX_ "Glob not terminated");
6760 bool readline_overriden = FALSE;
6761 GV *gv_readline = Nullgv;
6763 /* we're in a filehandle read situation */
6766 /* turn <> into <ARGV> */
6768 (void)strcpy(d,"ARGV");
6770 /* Check whether readline() is overriden */
6771 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6772 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6774 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6775 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6776 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6777 readline_overriden = TRUE;
6779 /* if <$fh>, create the ops to turn the variable into a
6785 /* try to find it in the pad for this block, otherwise find
6786 add symbol table ops
6788 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6789 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6790 SV *sym = sv_2mortal(
6791 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6792 sv_catpvn(sym, "::", 2);
6798 OP *o = newOP(OP_PADSV, 0);
6800 PL_lex_op = readline_overriden
6801 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6802 append_elem(OP_LIST, o,
6803 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6804 : (OP*)newUNOP(OP_READLINE, 0, o);
6813 ? (GV_ADDMULTI | GV_ADDINEVAL)
6816 PL_lex_op = readline_overriden
6817 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6818 append_elem(OP_LIST,
6819 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6820 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6821 : (OP*)newUNOP(OP_READLINE, 0,
6822 newUNOP(OP_RV2SV, 0,
6823 newGVOP(OP_GV, 0, gv)));
6825 if (!readline_overriden)
6826 PL_lex_op->op_flags |= OPf_SPECIAL;
6827 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6828 yylval.ival = OP_NULL;
6831 /* If it's none of the above, it must be a literal filehandle
6832 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6834 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6835 PL_lex_op = readline_overriden
6836 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6837 append_elem(OP_LIST,
6838 newGVOP(OP_GV, 0, gv),
6839 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6840 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6841 yylval.ival = OP_NULL;
6850 takes: start position in buffer
6851 keep_quoted preserve \ on the embedded delimiter(s)
6852 keep_delims preserve the delimiters around the string
6853 returns: position to continue reading from buffer
6854 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6855 updates the read buffer.
6857 This subroutine pulls a string out of the input. It is called for:
6858 q single quotes q(literal text)
6859 ' single quotes 'literal text'
6860 qq double quotes qq(interpolate $here please)
6861 " double quotes "interpolate $here please"
6862 qx backticks qx(/bin/ls -l)
6863 ` backticks `/bin/ls -l`
6864 qw quote words @EXPORT_OK = qw( func() $spam )
6865 m// regexp match m/this/
6866 s/// regexp substitute s/this/that/
6867 tr/// string transliterate tr/this/that/
6868 y/// string transliterate y/this/that/
6869 ($*@) sub prototypes sub foo ($)
6870 (stuff) sub attr parameters sub foo : attr(stuff)
6871 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6873 In most of these cases (all but <>, patterns and transliterate)
6874 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6875 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6876 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6879 It skips whitespace before the string starts, and treats the first
6880 character as the delimiter. If the delimiter is one of ([{< then
6881 the corresponding "close" character )]}> is used as the closing
6882 delimiter. It allows quoting of delimiters, and if the string has
6883 balanced delimiters ([{<>}]) it allows nesting.
6885 On success, the SV with the resulting string is put into lex_stuff or,
6886 if that is already non-NULL, into lex_repl. The second case occurs only
6887 when parsing the RHS of the special constructs s/// and tr/// (y///).
6888 For convenience, the terminating delimiter character is stuffed into
6893 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6895 SV *sv; /* scalar value: string */
6896 char *tmps; /* temp string, used for delimiter matching */
6897 register char *s = start; /* current position in the buffer */
6898 register char term; /* terminating character */
6899 register char *to; /* current position in the sv's data */
6900 I32 brackets = 1; /* bracket nesting level */
6901 bool has_utf8 = FALSE; /* is there any utf8 content? */
6902 I32 termcode; /* terminating char. code */
6903 U8 termstr[UTF8_MAXLEN]; /* terminating string */
6904 STRLEN termlen; /* length of terminating string */
6905 char *last = NULL; /* last position for nesting bracket */
6907 /* skip space before the delimiter */
6911 /* mark where we are, in case we need to report errors */
6914 /* after skipping whitespace, the next character is the terminator */
6917 termcode = termstr[0] = term;
6921 termcode = utf8_to_uvchr((U8*)s, &termlen);
6922 Copy(s, termstr, termlen, U8);
6923 if (!UTF8_IS_INVARIANT(term))
6927 /* mark where we are */
6928 PL_multi_start = CopLINE(PL_curcop);
6929 PL_multi_open = term;
6931 /* find corresponding closing delimiter */
6932 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6933 termcode = termstr[0] = term = tmps[5];
6935 PL_multi_close = term;
6937 /* create a new SV to hold the contents. 87 is leak category, I'm
6938 assuming. 79 is the SV's initial length. What a random number. */
6940 sv_upgrade(sv, SVt_PVIV);
6941 SvIVX(sv) = termcode;
6942 (void)SvPOK_only(sv); /* validate pointer */
6944 /* move past delimiter and try to read a complete string */
6946 sv_catpvn(sv, s, termlen);
6949 if (PL_encoding && !UTF) {
6953 int offset = s - SvPVX(PL_linestr);
6954 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
6955 &offset, (char*)termstr, termlen);
6956 char *ns = SvPVX(PL_linestr) + offset;
6957 char *svlast = SvEND(sv) - 1;
6959 for (; s < ns; s++) {
6960 if (*s == '\n' && !PL_rsfp)
6961 CopLINE_inc(PL_curcop);
6964 goto read_more_line;
6966 /* handle quoted delimiters */
6967 if (*(svlast-1) == '\\') {
6969 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
6971 if ((svlast-1 - t) % 2) {
6975 SvCUR_set(sv, SvCUR(sv) - 1);
6980 if (PL_multi_open == PL_multi_close) {
6987 for (w = t = last; t < svlast; w++, t++) {
6988 /* At here, all closes are "was quoted" one,
6989 so we don't check PL_multi_close. */
6991 if (!keep_quoted && *(t+1) == PL_multi_open)
6996 else if (*t == PL_multi_open)
7004 SvCUR_set(sv, w - SvPVX(sv));
7007 if (--brackets <= 0)
7013 SvCUR_set(sv, SvCUR(sv) - 1);
7019 /* extend sv if need be */
7020 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7021 /* set 'to' to the next character in the sv's string */
7022 to = SvPVX(sv)+SvCUR(sv);
7024 /* if open delimiter is the close delimiter read unbridle */
7025 if (PL_multi_open == PL_multi_close) {
7026 for (; s < PL_bufend; s++,to++) {
7027 /* embedded newlines increment the current line number */
7028 if (*s == '\n' && !PL_rsfp)
7029 CopLINE_inc(PL_curcop);
7030 /* handle quoted delimiters */
7031 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7032 if (!keep_quoted && s[1] == term)
7034 /* any other quotes are simply copied straight through */
7038 /* terminate when run out of buffer (the for() condition), or
7039 have found the terminator */
7040 else if (*s == term) {
7043 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7046 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7052 /* if the terminator isn't the same as the start character (e.g.,
7053 matched brackets), we have to allow more in the quoting, and
7054 be prepared for nested brackets.
7057 /* read until we run out of string, or we find the terminator */
7058 for (; s < PL_bufend; s++,to++) {
7059 /* embedded newlines increment the line count */
7060 if (*s == '\n' && !PL_rsfp)
7061 CopLINE_inc(PL_curcop);
7062 /* backslashes can escape the open or closing characters */
7063 if (*s == '\\' && s+1 < PL_bufend) {
7065 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7070 /* allow nested opens and closes */
7071 else if (*s == PL_multi_close && --brackets <= 0)
7073 else if (*s == PL_multi_open)
7075 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7080 /* terminate the copied string and update the sv's end-of-string */
7082 SvCUR_set(sv, to - SvPVX(sv));
7085 * this next chunk reads more into the buffer if we're not done yet
7089 break; /* handle case where we are done yet :-) */
7091 #ifndef PERL_STRICT_CR
7092 if (to - SvPVX(sv) >= 2) {
7093 if ((to[-2] == '\r' && to[-1] == '\n') ||
7094 (to[-2] == '\n' && to[-1] == '\r'))
7098 SvCUR_set(sv, to - SvPVX(sv));
7100 else if (to[-1] == '\r')
7103 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7108 /* if we're out of file, or a read fails, bail and reset the current
7109 line marker so we can report where the unterminated string began
7112 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7114 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7117 /* we read a line, so increment our line counter */
7118 CopLINE_inc(PL_curcop);
7120 /* update debugger info */
7121 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7122 SV *sv = NEWSV(88,0);
7124 sv_upgrade(sv, SVt_PVMG);
7125 sv_setsv(sv,PL_linestr);
7128 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7131 /* having changed the buffer, we must update PL_bufend */
7132 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7133 PL_last_lop = PL_last_uni = Nullch;
7136 /* at this point, we have successfully read the delimited string */
7138 if (!PL_encoding || UTF) {
7140 sv_catpvn(sv, s, termlen);
7143 if (has_utf8 || PL_encoding)
7146 PL_multi_end = CopLINE(PL_curcop);
7148 /* if we allocated too much space, give some back */
7149 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7150 SvLEN_set(sv, SvCUR(sv) + 1);
7151 Renew(SvPVX(sv), SvLEN(sv), char);
7154 /* decide whether this is the first or second quoted string we've read
7167 takes: pointer to position in buffer
7168 returns: pointer to new position in buffer
7169 side-effects: builds ops for the constant in yylval.op
7171 Read a number in any of the formats that Perl accepts:
7173 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7174 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7177 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7179 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7182 If it reads a number without a decimal point or an exponent, it will
7183 try converting the number to an integer and see if it can do so
7184 without loss of precision.
7188 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7190 register char *s = start; /* current position in buffer */
7191 register char *d; /* destination in temp buffer */
7192 register char *e; /* end of temp buffer */
7193 NV nv; /* number read, as a double */
7194 SV *sv = Nullsv; /* place to put the converted number */
7195 bool floatit; /* boolean: int or float? */
7196 char *lastub = 0; /* position of last underbar */
7197 static char number_too_long[] = "Number too long";
7199 /* We use the first character to decide what type of number this is */
7203 Perl_croak(aTHX_ "panic: scan_num");
7205 /* if it starts with a 0, it could be an octal number, a decimal in
7206 0.13 disguise, or a hexadecimal number, or a binary number. */
7210 u holds the "number so far"
7211 shift the power of 2 of the base
7212 (hex == 4, octal == 3, binary == 1)
7213 overflowed was the number more than we can hold?
7215 Shift is used when we add a digit. It also serves as an "are
7216 we in octal/hex/binary?" indicator to disallow hex characters
7222 bool overflowed = FALSE;
7223 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7224 static char* bases[5] = { "", "binary", "", "octal",
7226 static char* Bases[5] = { "", "Binary", "", "Octal",
7228 static char *maxima[5] = { "",
7229 "0b11111111111111111111111111111111",
7233 char *base, *Base, *max;
7239 } else if (s[1] == 'b') {
7243 /* check for a decimal in disguise */
7244 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7246 /* so it must be octal */
7253 if (ckWARN(WARN_SYNTAX))
7254 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7255 "Misplaced _ in number");
7259 base = bases[shift];
7260 Base = Bases[shift];
7261 max = maxima[shift];
7263 /* read the rest of the number */
7265 /* x is used in the overflow test,
7266 b is the digit we're adding on. */
7271 /* if we don't mention it, we're done */
7275 /* _ are ignored -- but warned about if consecutive */
7277 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7278 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7279 "Misplaced _ in number");
7283 /* 8 and 9 are not octal */
7286 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7290 case '2': case '3': case '4':
7291 case '5': case '6': case '7':
7293 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7297 b = *s++ & 15; /* ASCII digit -> value of digit */
7301 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7302 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7303 /* make sure they said 0x */
7308 /* Prepare to put the digit we have onto the end
7309 of the number so far. We check for overflows.
7314 x = u << shift; /* make room for the digit */
7316 if ((x >> shift) != u
7317 && !(PL_hints & HINT_NEW_BINARY)) {
7320 if (ckWARN_d(WARN_OVERFLOW))
7321 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7322 "Integer overflow in %s number",
7325 u = x | b; /* add the digit to the end */
7328 n *= nvshift[shift];
7329 /* If an NV has not enough bits in its
7330 * mantissa to represent an UV this summing of
7331 * small low-order numbers is a waste of time
7332 * (because the NV cannot preserve the
7333 * low-order bits anyway): we could just
7334 * remember when did we overflow and in the
7335 * end just multiply n by the right
7343 /* if we get here, we had success: make a scalar value from
7348 /* final misplaced underbar check */
7350 if (ckWARN(WARN_SYNTAX))
7351 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7356 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7357 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7358 "%s number > %s non-portable",
7364 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7365 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7366 "%s number > %s non-portable",
7371 if (PL_hints & HINT_NEW_BINARY)
7372 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7377 handle decimal numbers.
7378 we're also sent here when we read a 0 as the first digit
7380 case '1': case '2': case '3': case '4': case '5':
7381 case '6': case '7': case '8': case '9': case '.':
7384 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7387 /* read next group of digits and _ and copy into d */
7388 while (isDIGIT(*s) || *s == '_') {
7389 /* skip underscores, checking for misplaced ones
7393 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7394 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7395 "Misplaced _ in number");
7399 /* check for end of fixed-length buffer */
7401 Perl_croak(aTHX_ number_too_long);
7402 /* if we're ok, copy the character */
7407 /* final misplaced underbar check */
7408 if (lastub && s == lastub + 1) {
7409 if (ckWARN(WARN_SYNTAX))
7410 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7413 /* read a decimal portion if there is one. avoid
7414 3..5 being interpreted as the number 3. followed
7417 if (*s == '.' && s[1] != '.') {
7422 if (ckWARN(WARN_SYNTAX))
7423 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7424 "Misplaced _ in number");
7428 /* copy, ignoring underbars, until we run out of digits.
7430 for (; isDIGIT(*s) || *s == '_'; s++) {
7431 /* fixed length buffer check */
7433 Perl_croak(aTHX_ number_too_long);
7435 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7436 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7437 "Misplaced _ in number");
7443 /* fractional part ending in underbar? */
7445 if (ckWARN(WARN_SYNTAX))
7446 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7447 "Misplaced _ in number");
7449 if (*s == '.' && isDIGIT(s[1])) {
7450 /* oops, it's really a v-string, but without the "v" */
7456 /* read exponent part, if present */
7457 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7461 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7462 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7464 /* stray preinitial _ */
7466 if (ckWARN(WARN_SYNTAX))
7467 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7468 "Misplaced _ in number");
7472 /* allow positive or negative exponent */
7473 if (*s == '+' || *s == '-')
7476 /* stray initial _ */
7478 if (ckWARN(WARN_SYNTAX))
7479 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7480 "Misplaced _ in number");
7484 /* read digits of exponent */
7485 while (isDIGIT(*s) || *s == '_') {
7488 Perl_croak(aTHX_ number_too_long);
7492 if (ckWARN(WARN_SYNTAX) &&
7493 ((lastub && s == lastub + 1) ||
7494 (!isDIGIT(s[1]) && s[1] != '_')))
7495 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7496 "Misplaced _ in number");
7503 /* make an sv from the string */
7507 We try to do an integer conversion first if no characters
7508 indicating "float" have been found.
7513 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7515 if (flags == IS_NUMBER_IN_UV) {
7517 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7520 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7521 if (uv <= (UV) IV_MIN)
7522 sv_setiv(sv, -(IV)uv);
7529 /* terminate the string */
7531 nv = Atof(PL_tokenbuf);
7535 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7536 (PL_hints & HINT_NEW_INTEGER) )
7537 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7538 (floatit ? "float" : "integer"),
7542 /* if it starts with a v, it could be a v-string */
7545 sv = NEWSV(92,5); /* preallocate storage space */
7546 s = new_vstring(s,sv);
7550 /* make the op for the constant and return */
7553 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7555 lvalp->opval = Nullop;
7561 S_scan_formline(pTHX_ register char *s)
7565 SV *stuff = newSVpvn("",0);
7566 bool needargs = FALSE;
7569 if (*s == '.' || *s == /*{*/'}') {
7571 #ifdef PERL_STRICT_CR
7572 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7574 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7576 if (*t == '\n' || t == PL_bufend)
7579 if (PL_in_eval && !PL_rsfp) {
7580 eol = strchr(s,'\n');
7585 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7587 for (t = s; t < eol; t++) {
7588 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7590 goto enough; /* ~~ must be first line in formline */
7592 if (*t == '@' || *t == '^')
7596 sv_catpvn(stuff, s, eol-s);
7597 #ifndef PERL_STRICT_CR
7598 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7599 char *end = SvPVX(stuff) + SvCUR(stuff);
7611 s = filter_gets(PL_linestr, PL_rsfp, 0);
7612 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7613 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7614 PL_last_lop = PL_last_uni = Nullch;
7617 yyerror("Format not terminated");
7627 PL_lex_state = LEX_NORMAL;
7628 PL_nextval[PL_nexttoke].ival = 0;
7632 PL_lex_state = LEX_FORMLINE;
7634 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7636 else if (PL_encoding)
7637 sv_recode_to_utf8(stuff, PL_encoding);
7639 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7641 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7645 SvREFCNT_dec(stuff);
7646 PL_lex_formbrack = 0;
7657 PL_cshlen = strlen(PL_cshname);
7662 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7664 I32 oldsavestack_ix = PL_savestack_ix;
7665 CV* outsidecv = PL_compcv;
7668 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7670 SAVEI32(PL_subline);
7671 save_item(PL_subname);
7672 SAVESPTR(PL_compcv);
7674 PL_compcv = (CV*)NEWSV(1104,0);
7675 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7676 CvFLAGS(PL_compcv) |= flags;
7678 PL_subline = CopLINE(PL_curcop);
7679 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7680 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7681 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7682 #ifdef USE_5005THREADS
7683 CvOWNER(PL_compcv) = 0;
7684 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7685 MUTEX_INIT(CvMUTEXP(PL_compcv));
7686 #endif /* USE_5005THREADS */
7688 return oldsavestack_ix;
7692 #pragma segment Perl_yylex
7695 Perl_yywarn(pTHX_ char *s)
7697 PL_in_eval |= EVAL_WARNONLY;
7699 PL_in_eval &= ~EVAL_WARNONLY;
7704 Perl_yyerror(pTHX_ char *s)
7707 char *context = NULL;
7711 if (!yychar || (yychar == ';' && !PL_rsfp))
7713 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7714 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7717 The code below is removed for NetWare because it abends/crashes on NetWare
7718 when the script has error such as not having the closing quotes like:
7720 Checking of white spaces is anyway done in NetWare code.
7723 while (isSPACE(*PL_oldoldbufptr))
7726 context = PL_oldoldbufptr;
7727 contlen = PL_bufptr - PL_oldoldbufptr;
7729 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7730 PL_oldbufptr != PL_bufptr) {
7733 The code below is removed for NetWare because it abends/crashes on NetWare
7734 when the script has error such as not having the closing quotes like:
7736 Checking of white spaces is anyway done in NetWare code.
7739 while (isSPACE(*PL_oldbufptr))
7742 context = PL_oldbufptr;
7743 contlen = PL_bufptr - PL_oldbufptr;
7745 else if (yychar > 255)
7746 where = "next token ???";
7747 #ifdef USE_PURE_BISON
7748 /* GNU Bison sets the value -2 */
7749 else if (yychar == -2) {
7751 else if ((yychar & 127) == 127) {
7753 if (PL_lex_state == LEX_NORMAL ||
7754 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7755 where = "at end of line";
7756 else if (PL_lex_inpat)
7757 where = "within pattern";
7759 where = "within string";
7762 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7764 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7765 else if (isPRINT_LC(yychar))
7766 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7768 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7769 where = SvPVX(where_sv);
7771 msg = sv_2mortal(newSVpv(s, 0));
7772 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7773 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7775 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7777 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7778 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7779 Perl_sv_catpvf(aTHX_ msg,
7780 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7781 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7784 if (PL_in_eval & EVAL_WARNONLY)
7785 Perl_warn(aTHX_ "%"SVf, msg);
7788 if (PL_error_count >= 10) {
7789 if (PL_in_eval && SvCUR(ERRSV))
7790 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7791 ERRSV, OutCopFILE(PL_curcop));
7793 Perl_croak(aTHX_ "%s has too many errors.\n",
7794 OutCopFILE(PL_curcop));
7797 PL_in_my_stash = Nullhv;
7801 #pragma segment Main
7805 S_swallow_bom(pTHX_ U8 *s)
7808 slen = SvCUR(PL_linestr);
7812 /* UTF-16 little-endian */
7813 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7814 Perl_croak(aTHX_ "Unsupported script encoding");
7815 #ifndef PERL_NO_UTF16_FILTER
7816 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7818 if (PL_bufend > (char*)s) {
7822 filter_add(utf16rev_textfilter, NULL);
7823 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7824 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7825 PL_bufend - (char*)s - 1,
7827 Copy(news, s, newlen, U8);
7828 SvCUR_set(PL_linestr, newlen);
7829 PL_bufend = SvPVX(PL_linestr) + newlen;
7830 news[newlen++] = '\0';
7834 Perl_croak(aTHX_ "Unsupported script encoding");
7839 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7840 #ifndef PERL_NO_UTF16_FILTER
7841 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7843 if (PL_bufend > (char *)s) {
7847 filter_add(utf16_textfilter, NULL);
7848 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7849 PL_bufend = (char*)utf16_to_utf8(s, news,
7850 PL_bufend - (char*)s,
7852 Copy(news, s, newlen, U8);
7853 SvCUR_set(PL_linestr, newlen);
7854 PL_bufend = SvPVX(PL_linestr) + newlen;
7855 news[newlen++] = '\0';
7859 Perl_croak(aTHX_ "Unsupported script encoding");
7864 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7865 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7870 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7871 s[2] == 0xFE && s[3] == 0xFF)
7873 Perl_croak(aTHX_ "Unsupported script encoding");
7881 * Restore a source filter.
7885 restore_rsfp(pTHX_ void *f)
7887 PerlIO *fp = (PerlIO*)f;
7889 if (PL_rsfp == PerlIO_stdin())
7890 PerlIO_clearerr(PL_rsfp);
7891 else if (PL_rsfp && (PL_rsfp != fp))
7892 PerlIO_close(PL_rsfp);
7896 #ifndef PERL_NO_UTF16_FILTER
7898 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7900 I32 count = FILTER_READ(idx+1, sv, maxlen);
7905 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7906 if (!*SvPV_nolen(sv))
7907 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7910 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7911 sv_usepvn(sv, (char*)tmps, tend - tmps);
7917 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7919 I32 count = FILTER_READ(idx+1, sv, maxlen);
7924 if (!*SvPV_nolen(sv))
7925 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7928 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7929 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7930 sv_usepvn(sv, (char*)tmps, tend - tmps);