3 * Copyright (c) 1991-2001, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
40 /* For now 'use utf8' does not affect tokenizer on EBCDIC */
41 #define UTF (PL_linestr && DO_UTF8(PL_linestr))
43 #define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
46 /* In variables name $^X, these are the legal values for X.
47 * 1999-02-27 mjd-perl-patch@plover.com */
48 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
50 /* On MacOS, respect nonbreaking spaces */
51 #ifdef MACOS_TRADITIONAL
52 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
57 /* LEX_* are values for PL_lex_state, the state of the lexer.
58 * They are arranged oddly so that the guard on the switch statement
59 * can get by with a single comparison (if the compiler is smart enough).
62 /* #define LEX_NOTPARSING 11 is done in perl.h. */
65 #define LEX_INTERPNORMAL 9
66 #define LEX_INTERPCASEMOD 8
67 #define LEX_INTERPPUSH 7
68 #define LEX_INTERPSTART 6
69 #define LEX_INTERPEND 5
70 #define LEX_INTERPENDMAYBE 4
71 #define LEX_INTERPCONCAT 3
72 #define LEX_INTERPCONST 2
73 #define LEX_FORMLINE 1
74 #define LEX_KNOWNEXT 0
82 # define YYMAXLEVEL 100
84 YYSTYPE* yylval_pointer[YYMAXLEVEL];
85 int* yychar_pointer[YYMAXLEVEL];
89 # define yylval (*yylval_pointer[yyactlevel])
90 # define yychar (*yychar_pointer[yyactlevel])
91 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
93 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
98 /* CLINE is a macro that ensures PL_copline has a sane value */
103 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
106 * Convenience functions to return different tokens and prime the
107 * lexer for the next token. They all take an argument.
109 * TOKEN : generic token (used for '(', DOLSHARP, etc)
110 * OPERATOR : generic operator
111 * AOPERATOR : assignment operator
112 * PREBLOCK : beginning the block after an if, while, foreach, ...
113 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
114 * PREREF : *EXPR where EXPR is not a simple identifier
115 * TERM : expression term
116 * LOOPX : loop exiting command (goto, last, dump, etc)
117 * FTST : file test operator
118 * FUN0 : zero-argument function
119 * FUN1 : not used, except for not, which isn't a UNIOP
120 * BOop : bitwise or or xor
122 * SHop : shift operator
123 * PWop : power operator
124 * PMop : pattern-matching operator
125 * Aop : addition-level operator
126 * Mop : multiplication-level operator
127 * Eop : equality-testing operator
128 * Rop : relational operator <= != gt
130 * Also see LOP and lop() below.
133 /* Note that REPORT() and REPORT2() will be expressions that supply
134 * their own trailing comma, not suitable for statements as such. */
135 #ifdef DEBUGGING /* Serve -DT. */
136 # define REPORT(x,retval) tokereport(x,s,(int)retval),
137 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
139 # define REPORT(x,retval)
140 # define REPORT2(x,retval)
143 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
144 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
145 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
146 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
147 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
148 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
149 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
150 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
151 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
152 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
153 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
154 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
155 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
156 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
157 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
158 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
159 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
160 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
161 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
162 #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
164 /* This bit of chicanery makes a unary function followed by
165 * a parenthesis into a function with one argument, highest precedence.
167 #define UNI(f) return(yylval.ival = f, \
171 PL_last_uni = PL_oldbufptr, \
172 PL_last_lop_op = f, \
173 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
175 #define UNIBRACK(f) return(yylval.ival = f, \
178 PL_last_uni = PL_oldbufptr, \
179 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
181 /* grandfather return to old style */
182 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
185 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
189 report = newSVpv(thing, 0);
190 Perl_sv_catpvf(aTHX_ report, ":line %i:%i:", CopLINE(PL_curcop), rv);
192 if (s - PL_bufptr > 0)
193 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
195 if (PL_oldbufptr && *PL_oldbufptr)
196 sv_catpv(report, PL_tokenbuf);
198 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
205 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
206 * into an OP_ANDASSIGN or OP_ORASSIGN
210 S_ao(pTHX_ int toketype)
212 if (*PL_bufptr == '=') {
214 if (toketype == ANDAND)
215 yylval.ival = OP_ANDASSIGN;
216 else if (toketype == OROR)
217 yylval.ival = OP_ORASSIGN;
225 * When Perl expects an operator and finds something else, no_op
226 * prints the warning. It always prints "<something> found where
227 * operator expected. It prints "Missing semicolon on previous line?"
228 * if the surprise occurs at the start of the line. "do you need to
229 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
230 * where the compiler doesn't know if foo is a method call or a function.
231 * It prints "Missing operator before end of line" if there's nothing
232 * after the missing operator, or "... before <...>" if there is something
233 * after the missing operator.
237 S_no_op(pTHX_ char *what, char *s)
239 char *oldbp = PL_bufptr;
240 bool is_first = (PL_oldbufptr == PL_linestart);
246 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
248 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
249 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
251 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
252 if (t < PL_bufptr && isSPACE(*t))
253 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
254 t - PL_oldoldbufptr, PL_oldoldbufptr);
258 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
265 * Complain about missing quote/regexp/heredoc terminator.
266 * If it's called with (char *)NULL then it cauterizes the line buffer.
267 * If we're in a delimited string and the delimiter is a control
268 * character, it's reformatted into a two-char sequence like ^C.
273 S_missingterm(pTHX_ char *s)
278 char *nl = strrchr(s,'\n');
284 iscntrl(PL_multi_close)
286 PL_multi_close < 32 || PL_multi_close == 127
290 tmpbuf[1] = toCTRL(PL_multi_close);
296 *tmpbuf = PL_multi_close;
300 q = strchr(s,'"') ? '\'' : '"';
301 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
309 Perl_deprecate(pTHX_ char *s)
311 if (ckWARN(WARN_DEPRECATED))
312 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
317 * Deprecate a comma-less variable list.
323 deprecate("comma-less variable list");
327 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
328 * utf16-to-utf8-reversed.
331 #ifdef PERL_CR_FILTER
335 register char *s = SvPVX(sv);
336 register char *e = s + SvCUR(sv);
337 /* outer loop optimized to do nothing if there are no CR-LFs */
339 if (*s++ == '\r' && *s == '\n') {
340 /* hit a CR-LF, need to copy the rest */
341 register char *d = s - 1;
344 if (*s == '\r' && s[1] == '\n')
355 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
357 I32 count = FILTER_READ(idx+1, sv, maxlen);
358 if (count > 0 && !maxlen)
366 * Initialize variables. Uses the Perl save_stack to save its state (for
367 * recursive calls to the parser).
371 Perl_lex_start(pTHX_ SV *line)
376 SAVEI32(PL_lex_dojoin);
377 SAVEI32(PL_lex_brackets);
378 SAVEI32(PL_lex_casemods);
379 SAVEI32(PL_lex_starts);
380 SAVEI32(PL_lex_state);
381 SAVEVPTR(PL_lex_inpat);
382 SAVEI32(PL_lex_inwhat);
383 if (PL_lex_state == LEX_KNOWNEXT) {
384 I32 toke = PL_nexttoke;
385 while (--toke >= 0) {
386 SAVEI32(PL_nexttype[toke]);
387 SAVEVPTR(PL_nextval[toke]);
389 SAVEI32(PL_nexttoke);
391 SAVECOPLINE(PL_curcop);
394 SAVEPPTR(PL_oldbufptr);
395 SAVEPPTR(PL_oldoldbufptr);
396 SAVEPPTR(PL_last_lop);
397 SAVEPPTR(PL_last_uni);
398 SAVEPPTR(PL_linestart);
399 SAVESPTR(PL_linestr);
400 SAVEPPTR(PL_lex_brackstack);
401 SAVEPPTR(PL_lex_casestack);
402 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
403 SAVESPTR(PL_lex_stuff);
404 SAVEI32(PL_lex_defer);
405 SAVEI32(PL_sublex_info.sub_inwhat);
406 SAVESPTR(PL_lex_repl);
408 SAVEINT(PL_lex_expect);
410 PL_lex_state = LEX_NORMAL;
414 New(899, PL_lex_brackstack, 120, char);
415 New(899, PL_lex_casestack, 12, char);
416 SAVEFREEPV(PL_lex_brackstack);
417 SAVEFREEPV(PL_lex_casestack);
419 *PL_lex_casestack = '\0';
422 PL_lex_stuff = Nullsv;
423 PL_lex_repl = Nullsv;
427 PL_sublex_info.sub_inwhat = 0;
429 if (SvREADONLY(PL_linestr))
430 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
431 s = SvPV(PL_linestr, len);
432 if (len && s[len-1] != ';') {
433 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
434 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
435 sv_catpvn(PL_linestr, "\n;", 2);
437 SvTEMP_off(PL_linestr);
438 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
439 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
440 PL_last_lop = PL_last_uni = Nullch;
442 PL_rs = newSVpvn("\n", 1);
448 * Finalizer for lexing operations. Must be called when the parser is
449 * done with the lexer.
455 PL_doextract = FALSE;
460 * This subroutine has nothing to do with tilting, whether at windmills
461 * or pinball tables. Its name is short for "increment line". It
462 * increments the current line number in CopLINE(PL_curcop) and checks
463 * to see whether the line starts with a comment of the form
464 * # line 500 "foo.pm"
465 * If so, it sets the current line number and file to the values in the comment.
469 S_incline(pTHX_ char *s)
476 CopLINE_inc(PL_curcop);
479 while (SPACE_OR_TAB(*s)) s++;
480 if (strnEQ(s, "line", 4))
484 if (SPACE_OR_TAB(*s))
488 while (SPACE_OR_TAB(*s)) s++;
494 while (SPACE_OR_TAB(*s))
496 if (*s == '"' && (t = strchr(s+1, '"'))) {
501 for (t = s; !isSPACE(*t); t++) ;
504 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
506 if (*e != '\n' && *e != '\0')
507 return; /* false alarm */
513 Safefree(CopFILE(PL_curcop));
515 SvREFCNT_dec(CopFILEGV(PL_curcop));
517 CopFILE_set(PL_curcop, s);
520 CopLINE_set(PL_curcop, atoi(n)-1);
525 * Called to gobble the appropriate amount and type of whitespace.
526 * Skips comments as well.
530 S_skipspace(pTHX_ register char *s)
532 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
533 while (s < PL_bufend && SPACE_OR_TAB(*s))
539 SSize_t oldprevlen, oldoldprevlen;
540 SSize_t oldloplen, oldunilen;
541 while (s < PL_bufend && isSPACE(*s)) {
542 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
547 if (s < PL_bufend && *s == '#') {
548 while (s < PL_bufend && *s != '\n')
552 if (PL_in_eval && !PL_rsfp) {
559 /* only continue to recharge the buffer if we're at the end
560 * of the buffer, we're not reading from a source filter, and
561 * we're in normal lexing mode
563 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
564 PL_lex_state == LEX_FORMLINE)
567 /* try to recharge the buffer */
568 if ((s = filter_gets(PL_linestr, PL_rsfp,
569 (prevlen = SvCUR(PL_linestr)))) == Nullch)
571 /* end of file. Add on the -p or -n magic */
572 if (PL_minus_n || PL_minus_p) {
573 sv_setpv(PL_linestr,PL_minus_p ?
574 ";}continue{print or die qq(-p destination: $!\\n)" :
576 sv_catpv(PL_linestr,";}");
577 PL_minus_n = PL_minus_p = 0;
580 sv_setpv(PL_linestr,";");
582 /* reset variables for next time we lex */
583 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
585 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
586 PL_last_lop = PL_last_uni = Nullch;
588 /* Close the filehandle. Could be from -P preprocessor,
589 * STDIN, or a regular file. If we were reading code from
590 * STDIN (because the commandline held no -e or filename)
591 * then we don't close it, we reset it so the code can
592 * read from STDIN too.
595 if (PL_preprocess && !PL_in_eval)
596 (void)PerlProc_pclose(PL_rsfp);
597 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
598 PerlIO_clearerr(PL_rsfp);
600 (void)PerlIO_close(PL_rsfp);
605 /* not at end of file, so we only read another line */
606 /* make corresponding updates to old pointers, for yyerror() */
607 oldprevlen = PL_oldbufptr - PL_bufend;
608 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
610 oldunilen = PL_last_uni - PL_bufend;
612 oldloplen = PL_last_lop - PL_bufend;
613 PL_linestart = PL_bufptr = s + prevlen;
614 PL_bufend = s + SvCUR(PL_linestr);
616 PL_oldbufptr = s + oldprevlen;
617 PL_oldoldbufptr = s + oldoldprevlen;
619 PL_last_uni = s + oldunilen;
621 PL_last_lop = s + oldloplen;
624 /* debugger active and we're not compiling the debugger code,
625 * so store the line into the debugger's array of lines
627 if (PERLDB_LINE && PL_curstash != PL_debstash) {
628 SV *sv = NEWSV(85,0);
630 sv_upgrade(sv, SVt_PVMG);
631 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
632 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
639 * Check the unary operators to ensure there's no ambiguity in how they're
640 * used. An ambiguous piece of code would be:
642 * This doesn't mean rand() + 5. Because rand() is a unary operator,
643 * the +5 is its argument.
652 if (PL_oldoldbufptr != PL_last_uni)
654 while (isSPACE(*PL_last_uni))
656 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
657 if ((t = strchr(s, '(')) && t < PL_bufptr)
659 if (ckWARN_d(WARN_AMBIGUOUS)){
662 Perl_warner(aTHX_ WARN_AMBIGUOUS,
663 "Warning: Use of \"%s\" without parens is ambiguous",
669 /* workaround to replace the UNI() macro with a function. Only the
670 * hints/uts.sh file mentions this. Other comments elsewhere in the
671 * source indicate Microport Unix might need it too.
677 #define UNI(f) return uni(f,s)
680 S_uni(pTHX_ I32 f, char *s)
685 PL_last_uni = PL_oldbufptr;
696 #endif /* CRIPPLED_CC */
699 * LOP : macro to build a list operator. Its behaviour has been replaced
700 * with a subroutine, S_lop() for which LOP is just another name.
703 #define LOP(f,x) return lop(f,x,s)
707 * Build a list operator (or something that might be one). The rules:
708 * - if we have a next token, then it's a list operator [why?]
709 * - if the next thing is an opening paren, then it's a function
710 * - else it's a list operator
714 S_lop(pTHX_ I32 f, int x, char *s)
721 PL_last_lop = PL_oldbufptr;
736 * When the lexer realizes it knows the next token (for instance,
737 * it is reordering tokens for the parser) then it can call S_force_next
738 * to know what token to return the next time the lexer is called. Caller
739 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
740 * handles the token correctly.
744 S_force_next(pTHX_ I32 type)
746 PL_nexttype[PL_nexttoke] = type;
748 if (PL_lex_state != LEX_KNOWNEXT) {
749 PL_lex_defer = PL_lex_state;
750 PL_lex_expect = PL_expect;
751 PL_lex_state = LEX_KNOWNEXT;
757 * When the lexer knows the next thing is a word (for instance, it has
758 * just seen -> and it knows that the next char is a word char, then
759 * it calls S_force_word to stick the next word into the PL_next lookahead.
762 * char *start : buffer position (must be within PL_linestr)
763 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
764 * int check_keyword : if true, Perl checks to make sure the word isn't
765 * a keyword (do this if the word is a label, e.g. goto FOO)
766 * int allow_pack : if true, : characters will also be allowed (require,
768 * int allow_initial_tick : used by the "sub" lexer only.
772 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
777 start = skipspace(start);
779 if (isIDFIRST_lazy_if(s,UTF) ||
780 (allow_pack && *s == ':') ||
781 (allow_initial_tick && *s == '\'') )
783 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
784 if (check_keyword && keyword(PL_tokenbuf, len))
786 if (token == METHOD) {
791 PL_expect = XOPERATOR;
794 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
795 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
803 * Called when the lexer wants $foo *foo &foo etc, but the program
804 * text only contains the "foo" portion. The first argument is a pointer
805 * to the "foo", and the second argument is the type symbol to prefix.
806 * Forces the next token to be a "WORD".
807 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
811 S_force_ident(pTHX_ register char *s, int kind)
814 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
815 PL_nextval[PL_nexttoke].opval = o;
818 o->op_private = OPpCONST_ENTERED;
819 /* XXX see note in pp_entereval() for why we forgo typo
820 warnings if the symbol must be introduced in an eval.
822 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
823 kind == '$' ? SVt_PV :
824 kind == '@' ? SVt_PVAV :
825 kind == '%' ? SVt_PVHV :
833 Perl_str_to_version(pTHX_ SV *sv)
838 char *start = SvPVx(sv,len);
839 bool utf = SvUTF8(sv) ? TRUE : FALSE;
840 char *end = start + len;
841 while (start < end) {
845 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
850 retval += ((NV)n)/nshift;
859 * Forces the next token to be a version number.
863 S_force_version(pTHX_ char *s)
865 OP *version = Nullop;
874 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
875 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
877 s = scan_num(s, &yylval);
878 version = yylval.opval;
879 ver = cSVOPx(version)->op_sv;
880 if (SvPOK(ver) && !SvNIOK(ver)) {
881 (void)SvUPGRADE(ver, SVt_PVNV);
882 SvNVX(ver) = str_to_version(ver);
883 SvNOK_on(ver); /* hint that it is a version */
888 /* NOTE: The parser sees the package name and the VERSION swapped */
889 PL_nextval[PL_nexttoke].opval = version;
897 * Tokenize a quoted string passed in as an SV. It finds the next
898 * chunk, up to end of string or a backslash. It may make a new
899 * SV containing that chunk (if HINT_NEW_STRING is on). It also
904 S_tokeq(pTHX_ SV *sv)
915 s = SvPV_force(sv, len);
916 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
919 while (s < send && *s != '\\')
924 if ( PL_hints & HINT_NEW_STRING ) {
925 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
931 if (s + 1 < send && (s[1] == '\\'))
932 s++; /* all that, just for this */
937 SvCUR_set(sv, d - SvPVX(sv));
939 if ( PL_hints & HINT_NEW_STRING )
940 return new_constant(NULL, 0, "q", sv, pv, "q");
945 * Now come three functions related to double-quote context,
946 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
947 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
948 * interact with PL_lex_state, and create fake ( ... ) argument lists
949 * to handle functions and concatenation.
950 * They assume that whoever calls them will be setting up a fake
951 * join call, because each subthing puts a ',' after it. This lets
954 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
956 * (I'm not sure whether the spurious commas at the end of lcfirst's
957 * arguments and join's arguments are created or not).
962 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
964 * Pattern matching will set PL_lex_op to the pattern-matching op to
965 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
967 * OP_CONST and OP_READLINE are easy--just make the new op and return.
969 * Everything else becomes a FUNC.
971 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
972 * had an OP_CONST or OP_READLINE). This just sets us up for a
973 * call to S_sublex_push().
979 register I32 op_type = yylval.ival;
981 if (op_type == OP_NULL) {
982 yylval.opval = PL_lex_op;
986 if (op_type == OP_CONST || op_type == OP_READLINE) {
987 SV *sv = tokeq(PL_lex_stuff);
989 if (SvTYPE(sv) == SVt_PVIV) {
990 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
996 nsv = newSVpvn(p, len);
1002 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1003 PL_lex_stuff = Nullsv;
1007 PL_sublex_info.super_state = PL_lex_state;
1008 PL_sublex_info.sub_inwhat = op_type;
1009 PL_sublex_info.sub_op = PL_lex_op;
1010 PL_lex_state = LEX_INTERPPUSH;
1014 yylval.opval = PL_lex_op;
1024 * Create a new scope to save the lexing state. The scope will be
1025 * ended in S_sublex_done. Returns a '(', starting the function arguments
1026 * to the uc, lc, etc. found before.
1027 * Sets PL_lex_state to LEX_INTERPCONCAT.
1035 PL_lex_state = PL_sublex_info.super_state;
1036 SAVEI32(PL_lex_dojoin);
1037 SAVEI32(PL_lex_brackets);
1038 SAVEI32(PL_lex_casemods);
1039 SAVEI32(PL_lex_starts);
1040 SAVEI32(PL_lex_state);
1041 SAVEVPTR(PL_lex_inpat);
1042 SAVEI32(PL_lex_inwhat);
1043 SAVECOPLINE(PL_curcop);
1044 SAVEPPTR(PL_bufptr);
1045 SAVEPPTR(PL_oldbufptr);
1046 SAVEPPTR(PL_oldoldbufptr);
1047 SAVEPPTR(PL_last_lop);
1048 SAVEPPTR(PL_last_uni);
1049 SAVEPPTR(PL_linestart);
1050 SAVESPTR(PL_linestr);
1051 SAVEPPTR(PL_lex_brackstack);
1052 SAVEPPTR(PL_lex_casestack);
1054 PL_linestr = PL_lex_stuff;
1055 PL_lex_stuff = Nullsv;
1057 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1058 = SvPVX(PL_linestr);
1059 PL_bufend += SvCUR(PL_linestr);
1060 PL_last_lop = PL_last_uni = Nullch;
1061 SAVEFREESV(PL_linestr);
1063 PL_lex_dojoin = FALSE;
1064 PL_lex_brackets = 0;
1065 New(899, PL_lex_brackstack, 120, char);
1066 New(899, PL_lex_casestack, 12, char);
1067 SAVEFREEPV(PL_lex_brackstack);
1068 SAVEFREEPV(PL_lex_casestack);
1069 PL_lex_casemods = 0;
1070 *PL_lex_casestack = '\0';
1072 PL_lex_state = LEX_INTERPCONCAT;
1073 CopLINE_set(PL_curcop, PL_multi_start);
1075 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1076 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1077 PL_lex_inpat = PL_sublex_info.sub_op;
1079 PL_lex_inpat = Nullop;
1086 * Restores lexer state after a S_sublex_push.
1092 if (!PL_lex_starts++) {
1093 SV *sv = newSVpvn("",0);
1094 if (SvUTF8(PL_linestr))
1096 PL_expect = XOPERATOR;
1097 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1101 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1102 PL_lex_state = LEX_INTERPCASEMOD;
1106 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1107 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1108 PL_linestr = PL_lex_repl;
1110 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1111 PL_bufend += SvCUR(PL_linestr);
1112 PL_last_lop = PL_last_uni = Nullch;
1113 SAVEFREESV(PL_linestr);
1114 PL_lex_dojoin = FALSE;
1115 PL_lex_brackets = 0;
1116 PL_lex_casemods = 0;
1117 *PL_lex_casestack = '\0';
1119 if (SvEVALED(PL_lex_repl)) {
1120 PL_lex_state = LEX_INTERPNORMAL;
1122 /* we don't clear PL_lex_repl here, so that we can check later
1123 whether this is an evalled subst; that means we rely on the
1124 logic to ensure sublex_done() is called again only via the
1125 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1128 PL_lex_state = LEX_INTERPCONCAT;
1129 PL_lex_repl = Nullsv;
1135 PL_bufend = SvPVX(PL_linestr);
1136 PL_bufend += SvCUR(PL_linestr);
1137 PL_expect = XOPERATOR;
1138 PL_sublex_info.sub_inwhat = 0;
1146 Extracts a pattern, double-quoted string, or transliteration. This
1149 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1150 processing a pattern (PL_lex_inpat is true), a transliteration
1151 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1153 Returns a pointer to the character scanned up to. Iff this is
1154 advanced from the start pointer supplied (ie if anything was
1155 successfully parsed), will leave an OP for the substring scanned
1156 in yylval. Caller must intuit reason for not parsing further
1157 by looking at the next characters herself.
1161 double-quoted style: \r and \n
1162 regexp special ones: \D \s
1164 backrefs: \1 (deprecated in substitution replacements)
1165 case and quoting: \U \Q \E
1166 stops on @ and $, but not for $ as tail anchor
1168 In transliterations:
1169 characters are VERY literal, except for - not at the start or end
1170 of the string, which indicates a range. scan_const expands the
1171 range to the full set of intermediate characters.
1173 In double-quoted strings:
1175 double-quoted style: \r and \n
1177 backrefs: \1 (deprecated)
1178 case and quoting: \U \Q \E
1181 scan_const does *not* construct ops to handle interpolated strings.
1182 It stops processing as soon as it finds an embedded $ or @ variable
1183 and leaves it to the caller to work out what's going on.
1185 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1187 $ in pattern could be $foo or could be tail anchor. Assumption:
1188 it's a tail anchor if $ is the last thing in the string, or if it's
1189 followed by one of ")| \n\t"
1191 \1 (backreferences) are turned into $1
1193 The structure of the code is
1194 while (there's a character to process) {
1195 handle transliteration ranges
1196 skip regexp comments
1197 skip # initiated comments in //x patterns
1198 check for embedded @foo
1199 check for embedded scalars
1201 leave intact backslashes from leave (below)
1202 deprecate \1 in strings and sub replacements
1203 handle string-changing backslashes \l \U \Q \E, etc.
1204 switch (what was escaped) {
1205 handle - in a transliteration (becomes a literal -)
1206 handle \132 octal characters
1207 handle 0x15 hex characters
1208 handle \cV (control V)
1209 handle printf backslashes (\f, \r, \n, etc)
1211 } (end if backslash)
1212 } (end while character to read)
1217 S_scan_const(pTHX_ char *start)
1219 register char *send = PL_bufend; /* end of the constant */
1220 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1221 register char *s = start; /* start of the constant */
1222 register char *d = SvPVX(sv); /* destination for copies */
1223 bool dorange = FALSE; /* are we in a translit range? */
1224 bool didrange = FALSE; /* did we just finish a range? */
1225 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1226 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1229 const char *leaveit = /* set of acceptably-backslashed characters */
1231 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1234 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1235 /* If we are doing a trans and we know we want UTF8 set expectation */
1236 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1237 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1241 while (s < send || dorange) {
1242 /* get transliterations out of the way (they're most literal) */
1243 if (PL_lex_inwhat == OP_TRANS) {
1244 /* expand a range A-Z to the full set of characters. AIE! */
1246 I32 i; /* current expanded character */
1247 I32 min; /* first character in range */
1248 I32 max; /* last character in range */
1251 char *c = (char*)utf8_hop((U8*)d, -1);
1255 *c = UTF_TO_NATIVE(0xff);
1256 /* mark the range as done, and continue */
1262 i = d - SvPVX(sv); /* remember current offset */
1263 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1264 d = SvPVX(sv) + i; /* refresh d after realloc */
1265 d -= 2; /* eat the first char and the - */
1267 min = (U8)*d; /* first char in range */
1268 max = (U8)d[1]; /* last char in range */
1272 "Invalid [] range \"%c-%c\" in transliteration operator",
1273 (char)min, (char)max);
1277 if ((isLOWER(min) && isLOWER(max)) ||
1278 (isUPPER(min) && isUPPER(max))) {
1280 for (i = min; i <= max; i++)
1282 *d++ = NATIVE_TO_NEED(has_utf8,i);
1284 for (i = min; i <= max; i++)
1286 *d++ = NATIVE_TO_NEED(has_utf8,i);
1291 for (i = min; i <= max; i++)
1294 /* mark the range as done, and continue */
1300 /* range begins (ignore - as first or last char) */
1301 else if (*s == '-' && s+1 < send && s != start) {
1303 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1306 *d++ = UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1318 /* if we get here, we're not doing a transliteration */
1320 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1321 except for the last char, which will be done separately. */
1322 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1324 while (s < send && *s != ')')
1325 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1327 else if (s[2] == '{' /* This should match regcomp.c */
1328 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1331 char *regparse = s + (s[2] == '{' ? 3 : 4);
1334 while (count && (c = *regparse)) {
1335 if (c == '\\' && regparse[1])
1343 if (*regparse != ')') {
1344 regparse--; /* Leave one char for continuation. */
1345 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1347 while (s < regparse)
1348 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1352 /* likewise skip #-initiated comments in //x patterns */
1353 else if (*s == '#' && PL_lex_inpat &&
1354 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1355 while (s+1 < send && *s != '\n')
1356 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1359 /* check for embedded arrays
1360 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1362 else if (*s == '@' && s[1]
1363 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1366 /* check for embedded scalars. only stop if we're sure it's a
1369 else if (*s == '$') {
1370 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1372 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1373 break; /* in regexp, $ might be tail anchor */
1376 /* End of else if chain - OP_TRANS rejoin rest */
1379 if (*s == '\\' && s+1 < send) {
1382 /* some backslashes we leave behind */
1383 if (*leaveit && *s && strchr(leaveit, *s)) {
1384 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1385 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1389 /* deprecate \1 in strings and substitution replacements */
1390 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1391 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1393 if (ckWARN(WARN_SYNTAX))
1394 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1399 /* string-change backslash escapes */
1400 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1405 /* if we get here, it's either a quoted -, or a digit */
1408 /* quoted - in transliterations */
1410 if (PL_lex_inwhat == OP_TRANS) {
1417 if (ckWARN(WARN_MISC) && isALNUM(*s))
1418 Perl_warner(aTHX_ WARN_MISC,
1419 "Unrecognized escape \\%c passed through",
1421 /* default action is to copy the quoted character */
1422 goto default_action;
1425 /* \132 indicates an octal constant */
1426 case '0': case '1': case '2': case '3':
1427 case '4': case '5': case '6': case '7':
1429 STRLEN len = 0; /* disallow underscores */
1430 uv = (UV)scan_oct(s, 3, &len);
1433 goto NUM_ESCAPE_INSERT;
1435 /* \x24 indicates a hex constant */
1439 char* e = strchr(s, '}');
1441 yyerror("Missing right brace on \\x{}");
1445 STRLEN len = 1; /* allow underscores */
1446 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1452 STRLEN len = 0; /* disallow underscores */
1453 uv = (UV)scan_hex(s, 2, &len);
1459 /* Insert oct or hex escaped character.
1460 * There will always enough room in sv since such
1461 * escapes will be longer than any UTF-8 sequence
1462 * they can end up as. */
1464 /* We need to map to chars to ASCII before doing the tests
1467 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1468 if (!has_utf8 && uv > 255) {
1469 /* Might need to recode whatever we have
1470 * accumulated so far if it contains any
1473 * (Can't we keep track of that and avoid
1474 * this rescan? --jhi)
1478 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1479 if (!NATIVE_IS_INVARIANT(*c)) {
1484 STRLEN offset = d - SvPVX(sv);
1486 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1490 while (src >= (U8 *)SvPVX(sv)) {
1491 if (!NATIVE_IS_INVARIANT(*src)) {
1492 U8 ch = NATIVE_TO_ASCII(*src);
1493 *dst-- = UTF8_EIGHT_BIT_LO(ch);
1494 *dst-- = UTF8_EIGHT_BIT_HI(ch);
1504 if (has_utf8 || uv > 255) {
1505 d = (char*)uvchr_to_utf8((U8*)d, uv);
1507 if (PL_lex_inwhat == OP_TRANS &&
1508 PL_sublex_info.sub_op) {
1509 PL_sublex_info.sub_op->op_private |=
1510 (PL_lex_repl ? OPpTRANS_FROM_UTF
1523 /* \N{latin small letter a} is a named character */
1527 char* e = strchr(s, '}');
1533 yyerror("Missing right brace on \\N{}");
1537 res = newSVpvn(s + 1, e - s - 1);
1538 res = new_constant( Nullch, 0, "charnames",
1539 res, Nullsv, "\\N{...}" );
1541 sv_utf8_upgrade(res);
1542 str = SvPV(res,len);
1543 if (!has_utf8 && SvUTF8(res)) {
1544 char *ostart = SvPVX(sv);
1545 SvCUR_set(sv, d - ostart);
1548 sv_utf8_upgrade(sv);
1549 /* this just broke our allocation above... */
1550 SvGROW(sv, send - start);
1551 d = SvPVX(sv) + SvCUR(sv);
1554 if (len > e - s + 4) {
1555 char *odest = SvPVX(sv);
1557 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1558 d = SvPVX(sv) + (d - odest);
1560 Copy(str, d, len, char);
1567 yyerror("Missing braces on \\N{}");
1570 /* \c is a control character */
1579 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1583 /* printf-style backslashes, formfeeds, newlines, etc */
1585 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1588 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1591 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1594 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1597 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1600 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1603 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1609 } /* end if (backslash) */
1612 /* If we started with encoded form, or already know we want it
1613 and then encode the next character */
1614 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1616 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1617 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1620 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1621 STRLEN off = d - SvPVX(sv);
1622 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1624 d = (char*)uvchr_to_utf8((U8*)d, uv);
1628 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1630 } /* while loop to process each character */
1632 /* terminate the string and set up the sv */
1634 SvCUR_set(sv, d - SvPVX(sv));
1635 if (SvCUR(sv) >= SvLEN(sv))
1636 Perl_croak(aTHX_ "panic:constant overflowed allocated space");
1641 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1642 PL_sublex_info.sub_op->op_private |=
1643 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1647 /* shrink the sv if we allocated more than we used */
1648 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1649 SvLEN_set(sv, SvCUR(sv) + 1);
1650 Renew(SvPVX(sv), SvLEN(sv), char);
1653 /* return the substring (via yylval) only if we parsed anything */
1654 if (s > PL_bufptr) {
1655 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1656 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1658 ( PL_lex_inwhat == OP_TRANS
1660 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1663 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1670 * Returns TRUE if there's more to the expression (e.g., a subscript),
1673 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1675 * ->[ and ->{ return TRUE
1676 * { and [ outside a pattern are always subscripts, so return TRUE
1677 * if we're outside a pattern and it's not { or [, then return FALSE
1678 * if we're in a pattern and the first char is a {
1679 * {4,5} (any digits around the comma) returns FALSE
1680 * if we're in a pattern and the first char is a [
1682 * [SOMETHING] has a funky algorithm to decide whether it's a
1683 * character class or not. It has to deal with things like
1684 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1685 * anything else returns TRUE
1688 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1691 S_intuit_more(pTHX_ register char *s)
1693 if (PL_lex_brackets)
1695 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1697 if (*s != '{' && *s != '[')
1702 /* In a pattern, so maybe we have {n,m}. */
1719 /* On the other hand, maybe we have a character class */
1722 if (*s == ']' || *s == '^')
1725 /* this is terrifying, and it works */
1726 int weight = 2; /* let's weigh the evidence */
1728 unsigned char un_char = 255, last_un_char;
1729 char *send = strchr(s,']');
1730 char tmpbuf[sizeof PL_tokenbuf * 4];
1732 if (!send) /* has to be an expression */
1735 Zero(seen,256,char);
1738 else if (isDIGIT(*s)) {
1740 if (isDIGIT(s[1]) && s[2] == ']')
1746 for (; s < send; s++) {
1747 last_un_char = un_char;
1748 un_char = (unsigned char)*s;
1753 weight -= seen[un_char] * 10;
1754 if (isALNUM_lazy_if(s+1,UTF)) {
1755 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1756 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1761 else if (*s == '$' && s[1] &&
1762 strchr("[#!%*<>()-=",s[1])) {
1763 if (/*{*/ strchr("])} =",s[2]))
1772 if (strchr("wds]",s[1]))
1774 else if (seen['\''] || seen['"'])
1776 else if (strchr("rnftbxcav",s[1]))
1778 else if (isDIGIT(s[1])) {
1780 while (s[1] && isDIGIT(s[1]))
1790 if (strchr("aA01! ",last_un_char))
1792 if (strchr("zZ79~",s[1]))
1794 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1795 weight -= 5; /* cope with negative subscript */
1798 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1799 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1804 if (keyword(tmpbuf, d - tmpbuf))
1807 if (un_char == last_un_char + 1)
1809 weight -= seen[un_char];
1814 if (weight >= 0) /* probably a character class */
1824 * Does all the checking to disambiguate
1826 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1827 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1829 * First argument is the stuff after the first token, e.g. "bar".
1831 * Not a method if bar is a filehandle.
1832 * Not a method if foo is a subroutine prototyped to take a filehandle.
1833 * Not a method if it's really "Foo $bar"
1834 * Method if it's "foo $bar"
1835 * Not a method if it's really "print foo $bar"
1836 * Method if it's really "foo package::" (interpreted as package->foo)
1837 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1838 * Not a method if bar is a filehandle or package, but is quoted with
1843 S_intuit_method(pTHX_ char *start, GV *gv)
1845 char *s = start + (*start == '$');
1846 char tmpbuf[sizeof PL_tokenbuf];
1854 if ((cv = GvCVu(gv))) {
1855 char *proto = SvPVX(cv);
1865 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1866 /* start is the beginning of the possible filehandle/object,
1867 * and s is the end of it
1868 * tmpbuf is a copy of it
1871 if (*start == '$') {
1872 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1877 return *s == '(' ? FUNCMETH : METHOD;
1879 if (!keyword(tmpbuf, len)) {
1880 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1885 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1886 if (indirgv && GvCVu(indirgv))
1888 /* filehandle or package name makes it a method */
1889 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1891 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1892 return 0; /* no assumptions -- "=>" quotes bearword */
1894 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1895 newSVpvn(tmpbuf,len));
1896 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1900 return *s == '(' ? FUNCMETH : METHOD;
1908 * Return a string of Perl code to load the debugger. If PERL5DB
1909 * is set, it will return the contents of that, otherwise a
1910 * compile-time require of perl5db.pl.
1917 char *pdb = PerlEnv_getenv("PERL5DB");
1921 SETERRNO(0,SS$_NORMAL);
1922 return "BEGIN { require 'perl5db.pl' }";
1928 /* Encoded script support. filter_add() effectively inserts a
1929 * 'pre-processing' function into the current source input stream.
1930 * Note that the filter function only applies to the current source file
1931 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1933 * The datasv parameter (which may be NULL) can be used to pass
1934 * private data to this instance of the filter. The filter function
1935 * can recover the SV using the FILTER_DATA macro and use it to
1936 * store private buffers and state information.
1938 * The supplied datasv parameter is upgraded to a PVIO type
1939 * and the IoDIRP/IoANY field is used to store the function pointer,
1940 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1941 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1942 * private use must be set using malloc'd pointers.
1946 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1951 if (!PL_rsfp_filters)
1952 PL_rsfp_filters = newAV();
1954 datasv = NEWSV(255,0);
1955 if (!SvUPGRADE(datasv, SVt_PVIO))
1956 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1957 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1958 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1959 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1960 funcp, SvPV_nolen(datasv)));
1961 av_unshift(PL_rsfp_filters, 1);
1962 av_store(PL_rsfp_filters, 0, datasv) ;
1967 /* Delete most recently added instance of this filter function. */
1969 Perl_filter_del(pTHX_ filter_t funcp)
1972 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1973 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1975 /* if filter is on top of stack (usual case) just pop it off */
1976 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1977 if (IoANY(datasv) == (void *)funcp) {
1978 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1979 IoANY(datasv) = (void *)NULL;
1980 sv_free(av_pop(PL_rsfp_filters));
1984 /* we need to search for the correct entry and clear it */
1985 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1989 /* Invoke the n'th filter function for the current rsfp. */
1991 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1994 /* 0 = read one text line */
1999 if (!PL_rsfp_filters)
2001 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2002 /* Provide a default input filter to make life easy. */
2003 /* Note that we append to the line. This is handy. */
2004 DEBUG_P(PerlIO_printf(Perl_debug_log,
2005 "filter_read %d: from rsfp\n", idx));
2009 int old_len = SvCUR(buf_sv) ;
2011 /* ensure buf_sv is large enough */
2012 SvGROW(buf_sv, old_len + maxlen) ;
2013 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2014 if (PerlIO_error(PL_rsfp))
2015 return -1; /* error */
2017 return 0 ; /* end of file */
2019 SvCUR_set(buf_sv, old_len + len) ;
2022 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2023 if (PerlIO_error(PL_rsfp))
2024 return -1; /* error */
2026 return 0 ; /* end of file */
2029 return SvCUR(buf_sv);
2031 /* Skip this filter slot if filter has been deleted */
2032 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2033 DEBUG_P(PerlIO_printf(Perl_debug_log,
2034 "filter_read %d: skipped (filter deleted)\n",
2036 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2038 /* Get function pointer hidden within datasv */
2039 funcp = (filter_t)IoANY(datasv);
2040 DEBUG_P(PerlIO_printf(Perl_debug_log,
2041 "filter_read %d: via function %p (%s)\n",
2042 idx, funcp, SvPV_nolen(datasv)));
2043 /* Call function. The function is expected to */
2044 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2045 /* Return: <0:error, =0:eof, >0:not eof */
2046 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
2050 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2052 #ifdef PERL_CR_FILTER
2053 if (!PL_rsfp_filters) {
2054 filter_add(S_cr_textfilter,NULL);
2057 if (PL_rsfp_filters) {
2060 SvCUR_set(sv, 0); /* start with empty line */
2061 if (FILTER_READ(0, sv, 0) > 0)
2062 return ( SvPVX(sv) ) ;
2067 return (sv_gets(sv, fp, append));
2071 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2075 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2079 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2080 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2082 return GvHV(gv); /* Foo:: */
2085 /* use constant CLASS => 'MyClass' */
2086 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2088 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2089 pkgname = SvPV_nolen(sv);
2093 return gv_stashpv(pkgname, FALSE);
2097 static char* exp_name[] =
2098 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2099 "ATTRTERM", "TERMBLOCK"
2106 Works out what to call the token just pulled out of the input
2107 stream. The yacc parser takes care of taking the ops we return and
2108 stitching them into a tree.
2114 if read an identifier
2115 if we're in a my declaration
2116 croak if they tried to say my($foo::bar)
2117 build the ops for a my() declaration
2118 if it's an access to a my() variable
2119 are we in a sort block?
2120 croak if my($a); $a <=> $b
2121 build ops for access to a my() variable
2122 if in a dq string, and they've said @foo and we can't find @foo
2124 build ops for a bareword
2125 if we already built the token before, use it.
2128 #ifdef USE_PURE_BISON
2130 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2135 yylval_pointer[yyactlevel] = lvalp;
2136 yychar_pointer[yyactlevel] = lcharp;
2137 if (yyactlevel >= YYMAXLEVEL)
2138 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2140 r = Perl_yylex(aTHX);
2150 #pragma segment Perl_yylex
2163 /* check if there's an identifier for us to look at */
2164 if (PL_pending_ident) {
2165 /* pit holds the identifier we read and pending_ident is reset */
2166 char pit = PL_pending_ident;
2167 PL_pending_ident = 0;
2169 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2170 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2172 /* if we're in a my(), we can't allow dynamics here.
2173 $foo'bar has already been turned into $foo::bar, so
2174 just check for colons.
2176 if it's a legal name, the OP is a PADANY.
2179 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2180 if (strchr(PL_tokenbuf,':'))
2181 yyerror(Perl_form(aTHX_ "No package name allowed for "
2182 "variable %s in \"our\"",
2184 tmp = pad_allocmy(PL_tokenbuf);
2187 if (strchr(PL_tokenbuf,':'))
2188 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2190 yylval.opval = newOP(OP_PADANY, 0);
2191 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2197 build the ops for accesses to a my() variable.
2199 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2200 then used in a comparison. This catches most, but not
2201 all cases. For instance, it catches
2202 sort { my($a); $a <=> $b }
2204 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2205 (although why you'd do that is anyone's guess).
2208 if (!strchr(PL_tokenbuf,':')) {
2210 /* Check for single character per-thread SVs */
2211 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2212 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2213 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2215 yylval.opval = newOP(OP_THREADSV, 0);
2216 yylval.opval->op_targ = tmp;
2219 #endif /* USE_THREADS */
2220 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2221 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2222 /* might be an "our" variable" */
2223 if (SvFLAGS(namesv) & SVpad_OUR) {
2224 /* build ops for a bareword */
2225 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2226 sv_catpvn(sym, "::", 2);
2227 sv_catpv(sym, PL_tokenbuf+1);
2228 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2229 yylval.opval->op_private = OPpCONST_ENTERED;
2230 gv_fetchpv(SvPVX(sym),
2232 ? (GV_ADDMULTI | GV_ADDINEVAL)
2235 ((PL_tokenbuf[0] == '$') ? SVt_PV
2236 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2241 /* if it's a sort block and they're naming $a or $b */
2242 if (PL_last_lop_op == OP_SORT &&
2243 PL_tokenbuf[0] == '$' &&
2244 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2247 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2248 d < PL_bufend && *d != '\n';
2251 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2252 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2258 yylval.opval = newOP(OP_PADANY, 0);
2259 yylval.opval->op_targ = tmp;
2265 Whine if they've said @foo in a doublequoted string,
2266 and @foo isn't a variable we can find in the symbol
2269 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2270 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2271 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2272 && ckWARN(WARN_AMBIGUOUS))
2274 /* Downgraded from fatal to warning 20000522 mjd */
2275 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2276 "Possible unintended interpolation of %s in string",
2281 /* build ops for a bareword */
2282 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2283 yylval.opval->op_private = OPpCONST_ENTERED;
2284 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2285 ((PL_tokenbuf[0] == '$') ? SVt_PV
2286 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2291 /* no identifier pending identification */
2293 switch (PL_lex_state) {
2295 case LEX_NORMAL: /* Some compilers will produce faster */
2296 case LEX_INTERPNORMAL: /* code if we comment these out. */
2300 /* when we've already built the next token, just pull it out of the queue */
2303 yylval = PL_nextval[PL_nexttoke];
2305 PL_lex_state = PL_lex_defer;
2306 PL_expect = PL_lex_expect;
2307 PL_lex_defer = LEX_NORMAL;
2309 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2310 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2311 (IV)PL_nexttype[PL_nexttoke]); })
2313 return(PL_nexttype[PL_nexttoke]);
2315 /* interpolated case modifiers like \L \U, including \Q and \E.
2316 when we get here, PL_bufptr is at the \
2318 case LEX_INTERPCASEMOD:
2320 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2321 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2323 /* handle \E or end of string */
2324 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2328 if (PL_lex_casemods) {
2329 oldmod = PL_lex_casestack[--PL_lex_casemods];
2330 PL_lex_casestack[PL_lex_casemods] = '\0';
2332 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2334 PL_lex_state = LEX_INTERPCONCAT;
2338 if (PL_bufptr != PL_bufend)
2340 PL_lex_state = LEX_INTERPCONCAT;
2344 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2345 "### Saw case modifier at '%s'\n", PL_bufptr); })
2347 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2348 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2349 if (strchr("LU", *s) &&
2350 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2352 PL_lex_casestack[--PL_lex_casemods] = '\0';
2355 if (PL_lex_casemods > 10) {
2356 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2357 if (newlb != PL_lex_casestack) {
2359 PL_lex_casestack = newlb;
2362 PL_lex_casestack[PL_lex_casemods++] = *s;
2363 PL_lex_casestack[PL_lex_casemods] = '\0';
2364 PL_lex_state = LEX_INTERPCONCAT;
2365 PL_nextval[PL_nexttoke].ival = 0;
2368 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2370 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2372 PL_nextval[PL_nexttoke].ival = OP_LC;
2374 PL_nextval[PL_nexttoke].ival = OP_UC;
2376 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2378 Perl_croak(aTHX_ "panic: yylex");
2381 if (PL_lex_starts) {
2390 case LEX_INTERPPUSH:
2391 return sublex_push();
2393 case LEX_INTERPSTART:
2394 if (PL_bufptr == PL_bufend)
2395 return sublex_done();
2396 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2397 "### Interpolated variable at '%s'\n", PL_bufptr); })
2399 PL_lex_dojoin = (*PL_bufptr == '@');
2400 PL_lex_state = LEX_INTERPNORMAL;
2401 if (PL_lex_dojoin) {
2402 PL_nextval[PL_nexttoke].ival = 0;
2405 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2406 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2407 force_next(PRIVATEREF);
2409 force_ident("\"", '$');
2410 #endif /* USE_THREADS */
2411 PL_nextval[PL_nexttoke].ival = 0;
2413 PL_nextval[PL_nexttoke].ival = 0;
2415 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2418 if (PL_lex_starts++) {
2424 case LEX_INTERPENDMAYBE:
2425 if (intuit_more(PL_bufptr)) {
2426 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2432 if (PL_lex_dojoin) {
2433 PL_lex_dojoin = FALSE;
2434 PL_lex_state = LEX_INTERPCONCAT;
2437 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2438 && SvEVALED(PL_lex_repl))
2440 if (PL_bufptr != PL_bufend)
2441 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2442 PL_lex_repl = Nullsv;
2445 case LEX_INTERPCONCAT:
2447 if (PL_lex_brackets)
2448 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2450 if (PL_bufptr == PL_bufend)
2451 return sublex_done();
2453 if (SvIVX(PL_linestr) == '\'') {
2454 SV *sv = newSVsv(PL_linestr);
2457 else if ( PL_hints & HINT_NEW_RE )
2458 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2459 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2463 s = scan_const(PL_bufptr);
2465 PL_lex_state = LEX_INTERPCASEMOD;
2467 PL_lex_state = LEX_INTERPSTART;
2470 if (s != PL_bufptr) {
2471 PL_nextval[PL_nexttoke] = yylval;
2474 if (PL_lex_starts++)
2484 PL_lex_state = LEX_NORMAL;
2485 s = scan_formline(PL_bufptr);
2486 if (!PL_lex_formbrack)
2492 PL_oldoldbufptr = PL_oldbufptr;
2495 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2496 exp_name[PL_expect], s);
2502 if (isIDFIRST_lazy_if(s,UTF))
2504 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2507 goto fake_eof; /* emulate EOF on ^D or ^Z */
2512 if (PL_lex_brackets)
2513 yyerror("Missing right curly or square bracket");
2514 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2515 "### Tokener got EOF\n");
2519 if (s++ < PL_bufend)
2520 goto retry; /* ignore stray nulls */
2523 if (!PL_in_eval && !PL_preambled) {
2524 PL_preambled = TRUE;
2525 sv_setpv(PL_linestr,incl_perldb());
2526 if (SvCUR(PL_linestr))
2527 sv_catpv(PL_linestr,";");
2529 while(AvFILLp(PL_preambleav) >= 0) {
2530 SV *tmpsv = av_shift(PL_preambleav);
2531 sv_catsv(PL_linestr, tmpsv);
2532 sv_catpv(PL_linestr, ";");
2535 sv_free((SV*)PL_preambleav);
2536 PL_preambleav = NULL;
2538 if (PL_minus_n || PL_minus_p) {
2539 sv_catpv(PL_linestr, "LINE: while (<>) {");
2541 sv_catpv(PL_linestr,"chomp;");
2543 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2545 GvIMPORTED_AV_on(gv);
2547 if (strchr("/'\"", *PL_splitstr)
2548 && strchr(PL_splitstr + 1, *PL_splitstr))
2549 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2552 s = "'~#\200\1'"; /* surely one char is unused...*/
2553 while (s[1] && strchr(PL_splitstr, *s)) s++;
2555 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2556 "q" + (delim == '\''), delim);
2557 for (s = PL_splitstr; *s; s++) {
2559 sv_catpvn(PL_linestr, "\\", 1);
2560 sv_catpvn(PL_linestr, s, 1);
2562 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2566 sv_catpv(PL_linestr,"@F=split(' ');");
2569 sv_catpv(PL_linestr, "\n");
2570 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2571 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2572 PL_last_lop = PL_last_uni = Nullch;
2573 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2574 SV *sv = NEWSV(85,0);
2576 sv_upgrade(sv, SVt_PVMG);
2577 sv_setsv(sv,PL_linestr);
2578 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2583 bof = PL_rsfp ? TRUE : FALSE;
2584 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2587 if (PL_preprocess && !PL_in_eval)
2588 (void)PerlProc_pclose(PL_rsfp);
2589 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2590 PerlIO_clearerr(PL_rsfp);
2592 (void)PerlIO_close(PL_rsfp);
2594 PL_doextract = FALSE;
2596 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2597 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2598 sv_catpv(PL_linestr,";}");
2599 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2600 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2601 PL_last_lop = PL_last_uni = Nullch;
2602 PL_minus_n = PL_minus_p = 0;
2605 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2606 PL_last_lop = PL_last_uni = Nullch;
2607 sv_setpv(PL_linestr,"");
2608 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2610 /* if it looks like the start of a BOM, check if it in fact is */
2611 else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) {
2612 #ifdef PERLIO_IS_STDIO
2613 # ifdef __GNU_LIBRARY__
2614 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2615 # define FTELL_FOR_PIPE_IS_BROKEN
2619 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2620 # define FTELL_FOR_PIPE_IS_BROKEN
2625 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2626 /* This loses the possibility to detect the bof
2627 * situation on perl -P when the libc5 is being used.
2628 * Workaround? Maybe attach some extra state to PL_rsfp?
2631 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2633 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2636 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2637 s = swallow_bom((U8*)s);
2641 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2642 PL_doextract = FALSE;
2644 /* Incest with pod. */
2645 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2646 sv_setpv(PL_linestr, "");
2647 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2648 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2649 PL_last_lop = PL_last_uni = Nullch;
2650 PL_doextract = FALSE;
2654 } while (PL_doextract);
2655 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2656 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2657 SV *sv = NEWSV(85,0);
2659 sv_upgrade(sv, SVt_PVMG);
2660 sv_setsv(sv,PL_linestr);
2661 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2663 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2664 PL_last_lop = PL_last_uni = Nullch;
2665 if (CopLINE(PL_curcop) == 1) {
2666 while (s < PL_bufend && isSPACE(*s))
2668 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2672 if (*s == '#' && *(s+1) == '!')
2674 #ifdef ALTERNATE_SHEBANG
2676 static char as[] = ALTERNATE_SHEBANG;
2677 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2678 d = s + (sizeof(as) - 1);
2680 #endif /* ALTERNATE_SHEBANG */
2689 while (*d && !isSPACE(*d))
2693 #ifdef ARG_ZERO_IS_SCRIPT
2694 if (ipathend > ipath) {
2696 * HP-UX (at least) sets argv[0] to the script name,
2697 * which makes $^X incorrect. And Digital UNIX and Linux,
2698 * at least, set argv[0] to the basename of the Perl
2699 * interpreter. So, having found "#!", we'll set it right.
2701 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2702 assert(SvPOK(x) || SvGMAGICAL(x));
2703 if (sv_eq(x, CopFILESV(PL_curcop))) {
2704 sv_setpvn(x, ipath, ipathend - ipath);
2707 TAINT_NOT; /* $^X is always tainted, but that's OK */
2709 #endif /* ARG_ZERO_IS_SCRIPT */
2714 d = instr(s,"perl -");
2716 d = instr(s,"perl");
2718 /* avoid getting into infinite loops when shebang
2719 * line contains "Perl" rather than "perl" */
2721 for (d = ipathend-4; d >= ipath; --d) {
2722 if ((*d == 'p' || *d == 'P')
2723 && !ibcmp(d, "perl", 4))
2733 #ifdef ALTERNATE_SHEBANG
2735 * If the ALTERNATE_SHEBANG on this system starts with a
2736 * character that can be part of a Perl expression, then if
2737 * we see it but not "perl", we're probably looking at the
2738 * start of Perl code, not a request to hand off to some
2739 * other interpreter. Similarly, if "perl" is there, but
2740 * not in the first 'word' of the line, we assume the line
2741 * contains the start of the Perl program.
2743 if (d && *s != '#') {
2745 while (*c && !strchr("; \t\r\n\f\v#", *c))
2748 d = Nullch; /* "perl" not in first word; ignore */
2750 *s = '#'; /* Don't try to parse shebang line */
2752 #endif /* ALTERNATE_SHEBANG */
2753 #ifndef MACOS_TRADITIONAL
2758 !instr(s,"indir") &&
2759 instr(PL_origargv[0],"perl"))
2765 while (s < PL_bufend && isSPACE(*s))
2767 if (s < PL_bufend) {
2768 Newz(899,newargv,PL_origargc+3,char*);
2770 while (s < PL_bufend && !isSPACE(*s))
2773 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2776 newargv = PL_origargv;
2778 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2779 Perl_croak(aTHX_ "Can't exec %s", ipath);
2783 U32 oldpdb = PL_perldb;
2784 bool oldn = PL_minus_n;
2785 bool oldp = PL_minus_p;
2787 while (*d && !isSPACE(*d)) d++;
2788 while (SPACE_OR_TAB(*d)) d++;
2792 if (*d == 'M' || *d == 'm') {
2794 while (*d && !isSPACE(*d)) d++;
2795 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2798 d = moreswitches(d);
2800 if ((PERLDB_LINE && !oldpdb) ||
2801 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2802 /* if we have already added "LINE: while (<>) {",
2803 we must not do it again */
2805 sv_setpv(PL_linestr, "");
2806 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2807 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2808 PL_last_lop = PL_last_uni = Nullch;
2809 PL_preambled = FALSE;
2811 (void)gv_fetchfile(PL_origfilename);
2818 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2820 PL_lex_state = LEX_FORMLINE;
2825 #ifdef PERL_STRICT_CR
2826 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2828 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2830 case ' ': case '\t': case '\f': case 013:
2831 #ifdef MACOS_TRADITIONAL
2838 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2839 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2840 /* handle eval qq[#line 1 "foo"\n ...] */
2841 CopLINE_dec(PL_curcop);
2845 while (s < d && *s != '\n')
2850 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2852 PL_lex_state = LEX_FORMLINE;
2862 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2869 while (s < PL_bufend && SPACE_OR_TAB(*s))
2872 if (strnEQ(s,"=>",2)) {
2873 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2874 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2875 "### Saw unary minus before =>, forcing word '%s'\n", s);
2877 OPERATOR('-'); /* unary minus */
2879 PL_last_uni = PL_oldbufptr;
2881 case 'r': ftst = OP_FTEREAD; break;
2882 case 'w': ftst = OP_FTEWRITE; break;
2883 case 'x': ftst = OP_FTEEXEC; break;
2884 case 'o': ftst = OP_FTEOWNED; break;
2885 case 'R': ftst = OP_FTRREAD; break;
2886 case 'W': ftst = OP_FTRWRITE; break;
2887 case 'X': ftst = OP_FTREXEC; break;
2888 case 'O': ftst = OP_FTROWNED; break;
2889 case 'e': ftst = OP_FTIS; break;
2890 case 'z': ftst = OP_FTZERO; break;
2891 case 's': ftst = OP_FTSIZE; break;
2892 case 'f': ftst = OP_FTFILE; break;
2893 case 'd': ftst = OP_FTDIR; break;
2894 case 'l': ftst = OP_FTLINK; break;
2895 case 'p': ftst = OP_FTPIPE; break;
2896 case 'S': ftst = OP_FTSOCK; break;
2897 case 'u': ftst = OP_FTSUID; break;
2898 case 'g': ftst = OP_FTSGID; break;
2899 case 'k': ftst = OP_FTSVTX; break;
2900 case 'b': ftst = OP_FTBLK; break;
2901 case 'c': ftst = OP_FTCHR; break;
2902 case 't': ftst = OP_FTTTY; break;
2903 case 'T': ftst = OP_FTTEXT; break;
2904 case 'B': ftst = OP_FTBINARY; break;
2905 case 'M': case 'A': case 'C':
2906 gv_fetchpv("\024",TRUE, SVt_PV);
2908 case 'M': ftst = OP_FTMTIME; break;
2909 case 'A': ftst = OP_FTATIME; break;
2910 case 'C': ftst = OP_FTCTIME; break;
2918 PL_last_lop_op = ftst;
2919 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2920 "### Saw file test %c\n", (int)ftst);
2925 /* Assume it was a minus followed by a one-letter named
2926 * subroutine call (or a -bareword), then. */
2927 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2928 "### %c looked like a file test but was not\n",
2937 if (PL_expect == XOPERATOR)
2942 else if (*s == '>') {
2945 if (isIDFIRST_lazy_if(s,UTF)) {
2946 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2954 if (PL_expect == XOPERATOR)
2957 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2959 OPERATOR('-'); /* unary minus */
2966 if (PL_expect == XOPERATOR)
2971 if (PL_expect == XOPERATOR)
2974 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2980 if (PL_expect != XOPERATOR) {
2981 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2982 PL_expect = XOPERATOR;
2983 force_ident(PL_tokenbuf, '*');
2996 if (PL_expect == XOPERATOR) {
3000 PL_tokenbuf[0] = '%';
3001 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3002 if (!PL_tokenbuf[1]) {
3004 yyerror("Final % should be \\% or %name");
3007 PL_pending_ident = '%';
3026 switch (PL_expect) {
3029 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3031 PL_bufptr = s; /* update in case we back off */
3037 PL_expect = XTERMBLOCK;
3041 while (isIDFIRST_lazy_if(s,UTF)) {
3042 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3043 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3044 if (tmp < 0) tmp = -tmp;
3059 d = scan_str(d,TRUE,TRUE);
3061 /* MUST advance bufptr here to avoid bogus
3062 "at end of line" context messages from yyerror().
3064 PL_bufptr = s + len;
3065 yyerror("Unterminated attribute parameter in attribute list");
3068 return 0; /* EOF indicator */
3072 SV *sv = newSVpvn(s, len);
3073 sv_catsv(sv, PL_lex_stuff);
3074 attrs = append_elem(OP_LIST, attrs,
3075 newSVOP(OP_CONST, 0, sv));
3076 SvREFCNT_dec(PL_lex_stuff);
3077 PL_lex_stuff = Nullsv;
3080 if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3081 CvLVALUE_on(PL_compcv);
3082 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3083 CvLOCKED_on(PL_compcv);
3084 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3085 CvMETHOD_on(PL_compcv);
3087 else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "shared", len))
3088 GvSHARED_on(cGVOPx_gv(yylval.opval));
3090 /* After we've set the flags, it could be argued that
3091 we don't need to do the attributes.pm-based setting
3092 process, and shouldn't bother appending recognized
3093 flags. To experiment with that, uncomment the
3094 following "else": */
3096 attrs = append_elem(OP_LIST, attrs,
3097 newSVOP(OP_CONST, 0,
3101 if (*s == ':' && s[1] != ':')
3104 break; /* require real whitespace or :'s */
3106 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3107 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3108 char q = ((*s == '\'') ? '"' : '\'');
3109 /* If here for an expression, and parsed no attrs, back off. */
3110 if (tmp == '=' && !attrs) {
3114 /* MUST advance bufptr here to avoid bogus "at end of line"
3115 context messages from yyerror().
3119 yyerror("Unterminated attribute list");
3121 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3129 PL_nextval[PL_nexttoke].opval = attrs;
3137 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3138 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3154 if (PL_lex_brackets <= 0)
3155 yyerror("Unmatched right square bracket");
3158 if (PL_lex_state == LEX_INTERPNORMAL) {
3159 if (PL_lex_brackets == 0) {
3160 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3161 PL_lex_state = LEX_INTERPEND;
3168 if (PL_lex_brackets > 100) {
3169 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3170 if (newlb != PL_lex_brackstack) {
3172 PL_lex_brackstack = newlb;
3175 switch (PL_expect) {
3177 if (PL_lex_formbrack) {
3181 if (PL_oldoldbufptr == PL_last_lop)
3182 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3184 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3185 OPERATOR(HASHBRACK);
3187 while (s < PL_bufend && SPACE_OR_TAB(*s))
3190 PL_tokenbuf[0] = '\0';
3191 if (d < PL_bufend && *d == '-') {
3192 PL_tokenbuf[0] = '-';
3194 while (d < PL_bufend && SPACE_OR_TAB(*d))
3197 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3198 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3200 while (d < PL_bufend && SPACE_OR_TAB(*d))
3203 char minus = (PL_tokenbuf[0] == '-');
3204 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3212 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3217 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3222 if (PL_oldoldbufptr == PL_last_lop)
3223 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3225 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3228 OPERATOR(HASHBRACK);
3229 /* This hack serves to disambiguate a pair of curlies
3230 * as being a block or an anon hash. Normally, expectation
3231 * determines that, but in cases where we're not in a
3232 * position to expect anything in particular (like inside
3233 * eval"") we have to resolve the ambiguity. This code
3234 * covers the case where the first term in the curlies is a
3235 * quoted string. Most other cases need to be explicitly
3236 * disambiguated by prepending a `+' before the opening
3237 * curly in order to force resolution as an anon hash.
3239 * XXX should probably propagate the outer expectation
3240 * into eval"" to rely less on this hack, but that could
3241 * potentially break current behavior of eval"".
3245 if (*s == '\'' || *s == '"' || *s == '`') {
3246 /* common case: get past first string, handling escapes */
3247 for (t++; t < PL_bufend && *t != *s;)
3248 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3252 else if (*s == 'q') {
3255 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3259 char open, close, term;
3262 while (t < PL_bufend && isSPACE(*t))
3266 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3270 for (t++; t < PL_bufend; t++) {
3271 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3273 else if (*t == open)
3277 for (t++; t < PL_bufend; t++) {
3278 if (*t == '\\' && t+1 < PL_bufend)
3280 else if (*t == close && --brackets <= 0)
3282 else if (*t == open)
3288 else if (isALNUM_lazy_if(t,UTF)) {
3290 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3293 while (t < PL_bufend && isSPACE(*t))
3295 /* if comma follows first term, call it an anon hash */
3296 /* XXX it could be a comma expression with loop modifiers */
3297 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3298 || (*t == '=' && t[1] == '>')))
3299 OPERATOR(HASHBRACK);
3300 if (PL_expect == XREF)
3303 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3309 yylval.ival = CopLINE(PL_curcop);
3310 if (isSPACE(*s) || *s == '#')
3311 PL_copline = NOLINE; /* invalidate current command line number */
3316 if (PL_lex_brackets <= 0)
3317 yyerror("Unmatched right curly bracket");
3319 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3320 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3321 PL_lex_formbrack = 0;
3322 if (PL_lex_state == LEX_INTERPNORMAL) {
3323 if (PL_lex_brackets == 0) {
3324 if (PL_expect & XFAKEBRACK) {
3325 PL_expect &= XENUMMASK;
3326 PL_lex_state = LEX_INTERPEND;
3328 return yylex(); /* ignore fake brackets */
3330 if (*s == '-' && s[1] == '>')
3331 PL_lex_state = LEX_INTERPENDMAYBE;
3332 else if (*s != '[' && *s != '{')
3333 PL_lex_state = LEX_INTERPEND;
3336 if (PL_expect & XFAKEBRACK) {
3337 PL_expect &= XENUMMASK;
3339 return yylex(); /* ignore fake brackets */
3349 if (PL_expect == XOPERATOR) {
3350 if (ckWARN(WARN_SEMICOLON)
3351 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3353 CopLINE_dec(PL_curcop);
3354 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3355 CopLINE_inc(PL_curcop);
3360 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3362 PL_expect = XOPERATOR;
3363 force_ident(PL_tokenbuf, '&');
3367 yylval.ival = (OPpENTERSUB_AMPER<<8);
3386 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3387 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3389 if (PL_expect == XSTATE && isALPHA(tmp) &&
3390 (s == PL_linestart+1 || s[-2] == '\n') )
3392 if (PL_in_eval && !PL_rsfp) {
3397 if (strnEQ(s,"=cut",4)) {
3411 PL_doextract = TRUE;
3414 if (PL_lex_brackets < PL_lex_formbrack) {
3416 #ifdef PERL_STRICT_CR
3417 for (t = s; SPACE_OR_TAB(*t); t++) ;
3419 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3421 if (*t == '\n' || *t == '#') {
3439 if (PL_expect != XOPERATOR) {
3440 if (s[1] != '<' && !strchr(s,'>'))
3443 s = scan_heredoc(s);
3445 s = scan_inputsymbol(s);
3446 TERM(sublex_start());
3451 SHop(OP_LEFT_SHIFT);
3465 SHop(OP_RIGHT_SHIFT);
3474 if (PL_expect == XOPERATOR) {
3475 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3478 return ','; /* grandfather non-comma-format format */
3482 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3483 PL_tokenbuf[0] = '@';
3484 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3485 sizeof PL_tokenbuf - 1, FALSE);
3486 if (PL_expect == XOPERATOR)
3487 no_op("Array length", s);
3488 if (!PL_tokenbuf[1])
3490 PL_expect = XOPERATOR;
3491 PL_pending_ident = '#';
3495 PL_tokenbuf[0] = '$';
3496 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3497 sizeof PL_tokenbuf - 1, FALSE);
3498 if (PL_expect == XOPERATOR)
3500 if (!PL_tokenbuf[1]) {
3502 yyerror("Final $ should be \\$ or $name");
3506 /* This kludge not intended to be bulletproof. */
3507 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3508 yylval.opval = newSVOP(OP_CONST, 0,
3509 newSViv(PL_compiling.cop_arybase));
3510 yylval.opval->op_private = OPpCONST_ARYBASE;
3516 if (PL_lex_state == LEX_NORMAL)
3519 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3522 PL_tokenbuf[0] = '@';
3523 if (ckWARN(WARN_SYNTAX)) {
3525 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3528 PL_bufptr = skipspace(PL_bufptr);
3529 while (t < PL_bufend && *t != ']')
3531 Perl_warner(aTHX_ WARN_SYNTAX,
3532 "Multidimensional syntax %.*s not supported",
3533 (t - PL_bufptr) + 1, PL_bufptr);
3537 else if (*s == '{') {
3538 PL_tokenbuf[0] = '%';
3539 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3540 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3542 char tmpbuf[sizeof PL_tokenbuf];
3544 for (t++; isSPACE(*t); t++) ;
3545 if (isIDFIRST_lazy_if(t,UTF)) {
3546 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3547 for (; isSPACE(*t); t++) ;
3548 if (*t == ';' && get_cv(tmpbuf, FALSE))
3549 Perl_warner(aTHX_ WARN_SYNTAX,
3550 "You need to quote \"%s\"", tmpbuf);
3556 PL_expect = XOPERATOR;
3557 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3558 bool islop = (PL_last_lop == PL_oldoldbufptr);
3559 if (!islop || PL_last_lop_op == OP_GREPSTART)
3560 PL_expect = XOPERATOR;
3561 else if (strchr("$@\"'`q", *s))
3562 PL_expect = XTERM; /* e.g. print $fh "foo" */
3563 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3564 PL_expect = XTERM; /* e.g. print $fh &sub */
3565 else if (isIDFIRST_lazy_if(s,UTF)) {
3566 char tmpbuf[sizeof PL_tokenbuf];
3567 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3568 if ((tmp = keyword(tmpbuf, len))) {
3569 /* binary operators exclude handle interpretations */
3581 PL_expect = XTERM; /* e.g. print $fh length() */
3586 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3587 if (gv && GvCVu(gv))
3588 PL_expect = XTERM; /* e.g. print $fh subr() */
3591 else if (isDIGIT(*s))
3592 PL_expect = XTERM; /* e.g. print $fh 3 */
3593 else if (*s == '.' && isDIGIT(s[1]))
3594 PL_expect = XTERM; /* e.g. print $fh .3 */
3595 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3596 PL_expect = XTERM; /* e.g. print $fh -1 */
3597 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3598 PL_expect = XTERM; /* print $fh <<"EOF" */
3600 PL_pending_ident = '$';
3604 if (PL_expect == XOPERATOR)
3606 PL_tokenbuf[0] = '@';
3607 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3608 if (!PL_tokenbuf[1]) {
3610 yyerror("Final @ should be \\@ or @name");
3613 if (PL_lex_state == LEX_NORMAL)
3615 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3617 PL_tokenbuf[0] = '%';
3619 /* Warn about @ where they meant $. */
3620 if (ckWARN(WARN_SYNTAX)) {
3621 if (*s == '[' || *s == '{') {
3623 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3625 if (*t == '}' || *t == ']') {
3627 PL_bufptr = skipspace(PL_bufptr);
3628 Perl_warner(aTHX_ WARN_SYNTAX,
3629 "Scalar value %.*s better written as $%.*s",
3630 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3635 PL_pending_ident = '@';
3638 case '/': /* may either be division or pattern */
3639 case '?': /* may either be conditional or pattern */
3640 if (PL_expect != XOPERATOR) {
3641 /* Disable warning on "study /blah/" */
3642 if (PL_oldoldbufptr == PL_last_uni
3643 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3644 || memNE(PL_last_uni, "study", 5)
3645 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3647 s = scan_pat(s,OP_MATCH);
3648 TERM(sublex_start());
3656 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3657 #ifdef PERL_STRICT_CR
3660 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3662 && (s == PL_linestart || s[-1] == '\n') )
3664 PL_lex_formbrack = 0;
3668 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3674 yylval.ival = OPf_SPECIAL;
3680 if (PL_expect != XOPERATOR)
3685 case '0': case '1': case '2': case '3': case '4':
3686 case '5': case '6': case '7': case '8': case '9':
3687 s = scan_num(s, &yylval);
3688 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3689 "### Saw number in '%s'\n", s);
3691 if (PL_expect == XOPERATOR)
3696 s = scan_str(s,FALSE,FALSE);
3697 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3698 "### Saw string before '%s'\n", s);
3700 if (PL_expect == XOPERATOR) {
3701 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3704 return ','; /* grandfather non-comma-format format */
3710 missingterm((char*)0);
3711 yylval.ival = OP_CONST;
3712 TERM(sublex_start());
3715 s = scan_str(s,FALSE,FALSE);
3716 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3717 "### Saw string before '%s'\n", s);
3719 if (PL_expect == XOPERATOR) {
3720 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3723 return ','; /* grandfather non-comma-format format */
3729 missingterm((char*)0);
3730 yylval.ival = OP_CONST;
3731 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3732 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3733 yylval.ival = OP_STRINGIFY;
3737 TERM(sublex_start());
3740 s = scan_str(s,FALSE,FALSE);
3741 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3742 "### Saw backtick string before '%s'\n", s);
3744 if (PL_expect == XOPERATOR)
3745 no_op("Backticks",s);
3747 missingterm((char*)0);
3748 yylval.ival = OP_BACKTICK;
3750 TERM(sublex_start());
3754 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3755 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3757 if (PL_expect == XOPERATOR)
3758 no_op("Backslash",s);
3762 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3766 while (isDIGIT(*start) || *start == '_')
3768 if (*start == '.' && isDIGIT(start[1])) {
3769 s = scan_num(s, &yylval);
3772 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3773 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3777 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3780 s = scan_num(s, &yylval);
3787 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3826 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3828 /* Some keywords can be followed by any delimiter, including ':' */
3829 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3830 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3831 (PL_tokenbuf[0] == 'q' &&
3832 strchr("qwxr", PL_tokenbuf[1])))));
3834 /* x::* is just a word, unless x is "CORE" */
3835 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3839 while (d < PL_bufend && isSPACE(*d))
3840 d++; /* no comments skipped here, or s### is misparsed */
3842 /* Is this a label? */
3843 if (!tmp && PL_expect == XSTATE
3844 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3846 yylval.pval = savepv(PL_tokenbuf);
3851 /* Check for keywords */
3852 tmp = keyword(PL_tokenbuf, len);
3854 /* Is this a word before a => operator? */
3855 if (*d == '=' && d[1] == '>') {
3857 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3858 yylval.opval->op_private = OPpCONST_BARE;
3859 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3860 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3864 if (tmp < 0) { /* second-class keyword? */
3865 GV *ogv = Nullgv; /* override (winner) */
3866 GV *hgv = Nullgv; /* hidden (loser) */
3867 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3869 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3872 if (GvIMPORTED_CV(gv))
3874 else if (! CvMETHOD(cv))
3878 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3879 (gv = *gvp) != (GV*)&PL_sv_undef &&
3880 GvCVu(gv) && GvIMPORTED_CV(gv))
3886 tmp = 0; /* overridden by import or by GLOBAL */
3889 && -tmp==KEY_lock /* XXX generalizable kludge */
3891 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3893 tmp = 0; /* any sub overrides "weak" keyword */
3895 else { /* no override */
3899 if (ckWARN(WARN_AMBIGUOUS) && hgv
3900 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3901 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3902 "Ambiguous call resolved as CORE::%s(), %s",
3903 GvENAME(hgv), "qualify as such or use &");
3910 default: /* not a keyword */
3913 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3915 /* Get the rest if it looks like a package qualifier */
3917 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3919 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3922 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3923 *s == '\'' ? "'" : "::");
3927 if (PL_expect == XOPERATOR) {
3928 if (PL_bufptr == PL_linestart) {
3929 CopLINE_dec(PL_curcop);
3930 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3931 CopLINE_inc(PL_curcop);
3934 no_op("Bareword",s);
3937 /* Look for a subroutine with this name in current package,
3938 unless name is "Foo::", in which case Foo is a bearword
3939 (and a package name). */
3942 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3944 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3945 Perl_warner(aTHX_ WARN_BAREWORD,
3946 "Bareword \"%s\" refers to nonexistent package",
3949 PL_tokenbuf[len] = '\0';
3956 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3959 /* if we saw a global override before, get the right name */
3962 sv = newSVpvn("CORE::GLOBAL::",14);
3963 sv_catpv(sv,PL_tokenbuf);
3966 sv = newSVpv(PL_tokenbuf,0);
3968 /* Presume this is going to be a bareword of some sort. */
3971 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3972 yylval.opval->op_private = OPpCONST_BARE;
3974 /* And if "Foo::", then that's what it certainly is. */
3979 /* See if it's the indirect object for a list operator. */
3981 if (PL_oldoldbufptr &&
3982 PL_oldoldbufptr < PL_bufptr &&
3983 (PL_oldoldbufptr == PL_last_lop
3984 || PL_oldoldbufptr == PL_last_uni) &&
3985 /* NO SKIPSPACE BEFORE HERE! */
3986 (PL_expect == XREF ||
3987 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3989 bool immediate_paren = *s == '(';
3991 /* (Now we can afford to cross potential line boundary.) */
3994 /* Two barewords in a row may indicate method call. */
3996 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3999 /* If not a declared subroutine, it's an indirect object. */
4000 /* (But it's an indir obj regardless for sort.) */
4002 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4003 ((!gv || !GvCVu(gv)) &&
4004 (PL_last_lop_op != OP_MAPSTART &&
4005 PL_last_lop_op != OP_GREPSTART))))
4007 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4013 PL_expect = XOPERATOR;
4016 /* Is this a word before a => operator? */
4017 if (*s == '=' && s[1] == '>') {
4019 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4020 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
4021 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4025 /* If followed by a paren, it's certainly a subroutine. */
4028 if (gv && GvCVu(gv)) {
4029 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4030 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4035 PL_nextval[PL_nexttoke].opval = yylval.opval;
4036 PL_expect = XOPERATOR;
4042 /* If followed by var or block, call it a method (unless sub) */
4044 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4045 PL_last_lop = PL_oldbufptr;
4046 PL_last_lop_op = OP_METHOD;
4050 /* If followed by a bareword, see if it looks like indir obj. */
4052 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
4055 /* Not a method, so call it a subroutine (if defined) */
4057 if (gv && GvCVu(gv)) {
4059 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4060 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4061 "Ambiguous use of -%s resolved as -&%s()",
4062 PL_tokenbuf, PL_tokenbuf);
4063 /* Check for a constant sub */
4065 if ((sv = cv_const_sv(cv))) {
4067 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4068 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4069 yylval.opval->op_private = 0;
4073 /* Resolve to GV now. */
4074 op_free(yylval.opval);
4075 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4076 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4077 PL_last_lop = PL_oldbufptr;
4078 PL_last_lop_op = OP_ENTERSUB;
4079 /* Is there a prototype? */
4082 char *proto = SvPV((SV*)cv, len);
4085 if (strEQ(proto, "$"))
4087 if (*proto == '&' && *s == '{') {
4088 sv_setpv(PL_subname,"__ANON__");
4092 PL_nextval[PL_nexttoke].opval = yylval.opval;
4098 /* Call it a bare word */
4100 if (PL_hints & HINT_STRICT_SUBS)
4101 yylval.opval->op_private |= OPpCONST_STRICT;
4104 if (ckWARN(WARN_RESERVED)) {
4105 if (lastchar != '-') {
4106 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4108 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4115 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4116 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4117 "Operator or semicolon missing before %c%s",
4118 lastchar, PL_tokenbuf);
4119 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4120 "Ambiguous use of %c resolved as operator %c",
4121 lastchar, lastchar);
4127 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4128 newSVpv(CopFILE(PL_curcop),0));
4132 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4133 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4136 case KEY___PACKAGE__:
4137 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4139 ? newSVsv(PL_curstname)
4148 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4149 char *pname = "main";
4150 if (PL_tokenbuf[2] == 'D')
4151 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4152 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4155 GvIOp(gv) = newIO();
4156 IoIFP(GvIOp(gv)) = PL_rsfp;
4157 #if defined(HAS_FCNTL) && defined(F_SETFD)
4159 int fd = PerlIO_fileno(PL_rsfp);
4160 fcntl(fd,F_SETFD,fd >= 3);
4163 /* Mark this internal pseudo-handle as clean */
4164 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4166 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4167 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4168 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4170 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4171 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4172 /* if the script was opened in binmode, we need to revert
4173 * it to text mode for compatibility; but only iff it has CRs
4174 * XXX this is a questionable hack at best. */
4175 if (PL_bufend-PL_bufptr > 2
4176 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4179 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4180 loc = PerlIO_tell(PL_rsfp);
4181 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4183 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4185 PerlIO_seek(PL_rsfp, loc, 0);
4189 #ifdef PERLIO_LAYERS
4190 if (UTF && !IN_BYTE)
4191 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4204 if (PL_expect == XSTATE) {
4211 if (*s == ':' && s[1] == ':') {
4214 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4215 if (!(tmp = keyword(PL_tokenbuf, len)))
4216 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4230 LOP(OP_ACCEPT,XTERM);
4236 LOP(OP_ATAN2,XTERM);
4242 LOP(OP_BINMODE,XTERM);
4245 LOP(OP_BLESS,XTERM);
4254 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4271 if (!PL_cryptseen) {
4272 PL_cryptseen = TRUE;
4276 LOP(OP_CRYPT,XTERM);
4279 if (ckWARN(WARN_CHMOD)) {
4280 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4281 if (*d != '0' && isDIGIT(*d))
4282 Perl_warner(aTHX_ WARN_CHMOD,
4283 "chmod() mode argument is missing initial 0");
4285 LOP(OP_CHMOD,XTERM);
4288 LOP(OP_CHOWN,XTERM);
4291 LOP(OP_CONNECT,XTERM);
4307 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4311 PL_hints |= HINT_BLOCK_SCOPE;
4321 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4322 LOP(OP_DBMOPEN,XTERM);
4328 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4335 yylval.ival = CopLINE(PL_curcop);
4349 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4350 UNIBRACK(OP_ENTEREVAL);
4365 case KEY_endhostent:
4371 case KEY_endservent:
4374 case KEY_endprotoent:
4385 yylval.ival = CopLINE(PL_curcop);
4387 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4389 if ((PL_bufend - p) >= 3 &&
4390 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4392 else if ((PL_bufend - p) >= 4 &&
4393 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4396 if (isIDFIRST_lazy_if(p,UTF)) {
4397 p = scan_ident(p, PL_bufend,
4398 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4402 Perl_croak(aTHX_ "Missing $ on loop variable");
4407 LOP(OP_FORMLINE,XTERM);
4413 LOP(OP_FCNTL,XTERM);
4419 LOP(OP_FLOCK,XTERM);
4428 LOP(OP_GREPSTART, XREF);
4431 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4446 case KEY_getpriority:
4447 LOP(OP_GETPRIORITY,XTERM);
4449 case KEY_getprotobyname:
4452 case KEY_getprotobynumber:
4453 LOP(OP_GPBYNUMBER,XTERM);
4455 case KEY_getprotoent:
4467 case KEY_getpeername:
4468 UNI(OP_GETPEERNAME);
4470 case KEY_gethostbyname:
4473 case KEY_gethostbyaddr:
4474 LOP(OP_GHBYADDR,XTERM);
4476 case KEY_gethostent:
4479 case KEY_getnetbyname:
4482 case KEY_getnetbyaddr:
4483 LOP(OP_GNBYADDR,XTERM);
4488 case KEY_getservbyname:
4489 LOP(OP_GSBYNAME,XTERM);
4491 case KEY_getservbyport:
4492 LOP(OP_GSBYPORT,XTERM);
4494 case KEY_getservent:
4497 case KEY_getsockname:
4498 UNI(OP_GETSOCKNAME);
4500 case KEY_getsockopt:
4501 LOP(OP_GSOCKOPT,XTERM);
4523 yylval.ival = CopLINE(PL_curcop);
4527 LOP(OP_INDEX,XTERM);
4533 LOP(OP_IOCTL,XTERM);
4545 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4577 LOP(OP_LISTEN,XTERM);
4586 s = scan_pat(s,OP_MATCH);
4587 TERM(sublex_start());
4590 LOP(OP_MAPSTART, XREF);
4593 LOP(OP_MKDIR,XTERM);
4596 LOP(OP_MSGCTL,XTERM);
4599 LOP(OP_MSGGET,XTERM);
4602 LOP(OP_MSGRCV,XTERM);
4605 LOP(OP_MSGSND,XTERM);
4611 if (isIDFIRST_lazy_if(s,UTF)) {
4612 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4613 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4615 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4616 if (!PL_in_my_stash) {
4619 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4627 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4634 if (PL_expect != XSTATE)
4635 yyerror("\"no\" not allowed in expression");
4636 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4637 s = force_version(s);
4642 if (*s == '(' || (s = skipspace(s), *s == '('))
4649 if (isIDFIRST_lazy_if(s,UTF)) {
4651 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4653 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4654 Perl_warner(aTHX_ WARN_PRECEDENCE,
4655 "Precedence problem: open %.*s should be open(%.*s)",
4661 yylval.ival = OP_OR;
4671 LOP(OP_OPEN_DIR,XTERM);
4674 checkcomma(s,PL_tokenbuf,"filehandle");
4678 checkcomma(s,PL_tokenbuf,"filehandle");
4697 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4701 LOP(OP_PIPE_OP,XTERM);
4704 s = scan_str(s,FALSE,FALSE);
4706 missingterm((char*)0);
4707 yylval.ival = OP_CONST;
4708 TERM(sublex_start());
4714 s = scan_str(s,FALSE,FALSE);
4716 missingterm((char*)0);
4718 if (SvCUR(PL_lex_stuff)) {
4721 d = SvPV_force(PL_lex_stuff, len);
4724 for (; isSPACE(*d) && len; --len, ++d) ;
4727 if (!warned && ckWARN(WARN_QW)) {
4728 for (; !isSPACE(*d) && len; --len, ++d) {
4730 Perl_warner(aTHX_ WARN_QW,
4731 "Possible attempt to separate words with commas");
4734 else if (*d == '#') {
4735 Perl_warner(aTHX_ WARN_QW,
4736 "Possible attempt to put comments in qw() list");
4742 for (; !isSPACE(*d) && len; --len, ++d) ;
4744 sv = newSVpvn(b, d-b);
4745 if (DO_UTF8(PL_lex_stuff))
4747 words = append_elem(OP_LIST, words,
4748 newSVOP(OP_CONST, 0, tokeq(sv)));
4752 PL_nextval[PL_nexttoke].opval = words;
4757 SvREFCNT_dec(PL_lex_stuff);
4758 PL_lex_stuff = Nullsv;
4764 s = scan_str(s,FALSE,FALSE);
4766 missingterm((char*)0);
4767 yylval.ival = OP_STRINGIFY;
4768 if (SvIVX(PL_lex_stuff) == '\'')
4769 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4770 TERM(sublex_start());
4773 s = scan_pat(s,OP_QR);
4774 TERM(sublex_start());
4777 s = scan_str(s,FALSE,FALSE);
4779 missingterm((char*)0);
4780 yylval.ival = OP_BACKTICK;
4782 TERM(sublex_start());
4789 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4790 s = force_version(s);
4793 *PL_tokenbuf = '\0';
4794 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4795 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4796 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4798 yyerror("<> should be quotes");
4806 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4810 LOP(OP_RENAME,XTERM);
4819 LOP(OP_RINDEX,XTERM);
4842 LOP(OP_REVERSE,XTERM);
4853 TERM(sublex_start());
4855 TOKEN(1); /* force error */
4864 LOP(OP_SELECT,XTERM);
4870 LOP(OP_SEMCTL,XTERM);
4873 LOP(OP_SEMGET,XTERM);
4876 LOP(OP_SEMOP,XTERM);
4882 LOP(OP_SETPGRP,XTERM);
4884 case KEY_setpriority:
4885 LOP(OP_SETPRIORITY,XTERM);
4887 case KEY_sethostent:
4893 case KEY_setservent:
4896 case KEY_setprotoent:
4906 LOP(OP_SEEKDIR,XTERM);
4908 case KEY_setsockopt:
4909 LOP(OP_SSOCKOPT,XTERM);
4915 LOP(OP_SHMCTL,XTERM);
4918 LOP(OP_SHMGET,XTERM);
4921 LOP(OP_SHMREAD,XTERM);
4924 LOP(OP_SHMWRITE,XTERM);
4927 LOP(OP_SHUTDOWN,XTERM);
4936 LOP(OP_SOCKET,XTERM);
4938 case KEY_socketpair:
4939 LOP(OP_SOCKPAIR,XTERM);
4942 checkcomma(s,PL_tokenbuf,"subroutine name");
4944 if (*s == ';' || *s == ')') /* probably a close */
4945 Perl_croak(aTHX_ "sort is now a reserved word");
4947 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4951 LOP(OP_SPLIT,XTERM);
4954 LOP(OP_SPRINTF,XTERM);
4957 LOP(OP_SPLICE,XTERM);
4972 LOP(OP_SUBSTR,XTERM);
4978 char tmpbuf[sizeof PL_tokenbuf];
4980 expectation attrful;
4981 bool have_name, have_proto;
4986 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4987 (*s == ':' && s[1] == ':'))
4990 attrful = XATTRBLOCK;
4991 /* remember buffer pos'n for later force_word */
4992 tboffset = s - PL_oldbufptr;
4993 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4994 if (strchr(tmpbuf, ':'))
4995 sv_setpv(PL_subname, tmpbuf);
4997 sv_setsv(PL_subname,PL_curstname);
4998 sv_catpvn(PL_subname,"::",2);
4999 sv_catpvn(PL_subname,tmpbuf,len);
5006 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5007 PL_expect = XTERMBLOCK;
5008 attrful = XATTRTERM;
5009 sv_setpv(PL_subname,"?");
5013 if (key == KEY_format) {
5015 PL_lex_formbrack = PL_lex_brackets + 1;
5017 (void) force_word(PL_oldbufptr + tboffset, WORD,
5022 /* Look for a prototype */
5026 s = scan_str(s,FALSE,FALSE);
5028 Perl_croak(aTHX_ "Prototype not terminated");
5030 d = SvPVX(PL_lex_stuff);
5032 for (p = d; *p; ++p) {
5037 SvCUR(PL_lex_stuff) = tmp;
5045 if (*s == ':' && s[1] != ':')
5046 PL_expect = attrful;
5049 PL_nextval[PL_nexttoke].opval =
5050 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5051 PL_lex_stuff = Nullsv;
5055 sv_setpv(PL_subname,"__ANON__");
5058 (void) force_word(PL_oldbufptr + tboffset, WORD,
5067 LOP(OP_SYSTEM,XREF);
5070 LOP(OP_SYMLINK,XTERM);
5073 LOP(OP_SYSCALL,XTERM);
5076 LOP(OP_SYSOPEN,XTERM);
5079 LOP(OP_SYSSEEK,XTERM);
5082 LOP(OP_SYSREAD,XTERM);
5085 LOP(OP_SYSWRITE,XTERM);
5089 TERM(sublex_start());
5110 LOP(OP_TRUNCATE,XTERM);
5122 yylval.ival = CopLINE(PL_curcop);
5126 yylval.ival = CopLINE(PL_curcop);
5130 LOP(OP_UNLINK,XTERM);
5136 LOP(OP_UNPACK,XTERM);
5139 LOP(OP_UTIME,XTERM);
5142 if (ckWARN(WARN_UMASK)) {
5143 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5144 if (*d != '0' && isDIGIT(*d))
5145 Perl_warner(aTHX_ WARN_UMASK,
5146 "umask: argument is missing initial 0");
5151 LOP(OP_UNSHIFT,XTERM);
5154 if (PL_expect != XSTATE)
5155 yyerror("\"use\" not allowed in expression");
5157 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5158 s = force_version(s);
5159 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5160 PL_nextval[PL_nexttoke].opval = Nullop;
5165 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5166 s = force_version(s);
5178 yylval.ival = CopLINE(PL_curcop);
5182 PL_hints |= HINT_BLOCK_SCOPE;
5189 LOP(OP_WAITPID,XTERM);
5197 static char ctl_l[2];
5199 if (ctl_l[0] == '\0')
5200 ctl_l[0] = toCTRL('L');
5201 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5204 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5209 if (PL_expect == XOPERATOR)
5215 yylval.ival = OP_XOR;
5220 TERM(sublex_start());
5225 #pragma segment Main
5229 Perl_keyword(pTHX_ register char *d, I32 len)
5234 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5235 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5236 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5237 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5238 if (strEQ(d,"__END__")) return KEY___END__;
5242 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5247 if (strEQ(d,"and")) return -KEY_and;
5248 if (strEQ(d,"abs")) return -KEY_abs;
5251 if (strEQ(d,"alarm")) return -KEY_alarm;
5252 if (strEQ(d,"atan2")) return -KEY_atan2;
5255 if (strEQ(d,"accept")) return -KEY_accept;
5260 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5263 if (strEQ(d,"bless")) return -KEY_bless;
5264 if (strEQ(d,"bind")) return -KEY_bind;
5265 if (strEQ(d,"binmode")) return -KEY_binmode;
5268 if (strEQ(d,"CORE")) return -KEY_CORE;
5269 if (strEQ(d,"CHECK")) return KEY_CHECK;
5274 if (strEQ(d,"cmp")) return -KEY_cmp;
5275 if (strEQ(d,"chr")) return -KEY_chr;
5276 if (strEQ(d,"cos")) return -KEY_cos;
5279 if (strEQ(d,"chop")) return -KEY_chop;
5282 if (strEQ(d,"close")) return -KEY_close;
5283 if (strEQ(d,"chdir")) return -KEY_chdir;
5284 if (strEQ(d,"chomp")) return -KEY_chomp;
5285 if (strEQ(d,"chmod")) return -KEY_chmod;
5286 if (strEQ(d,"chown")) return -KEY_chown;
5287 if (strEQ(d,"crypt")) return -KEY_crypt;
5290 if (strEQ(d,"chroot")) return -KEY_chroot;
5291 if (strEQ(d,"caller")) return -KEY_caller;
5294 if (strEQ(d,"connect")) return -KEY_connect;
5297 if (strEQ(d,"closedir")) return -KEY_closedir;
5298 if (strEQ(d,"continue")) return -KEY_continue;
5303 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5308 if (strEQ(d,"do")) return KEY_do;
5311 if (strEQ(d,"die")) return -KEY_die;
5314 if (strEQ(d,"dump")) return -KEY_dump;
5317 if (strEQ(d,"delete")) return KEY_delete;
5320 if (strEQ(d,"defined")) return KEY_defined;
5321 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5324 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5329 if (strEQ(d,"END")) return KEY_END;
5334 if (strEQ(d,"eq")) return -KEY_eq;
5337 if (strEQ(d,"eof")) return -KEY_eof;
5338 if (strEQ(d,"exp")) return -KEY_exp;
5341 if (strEQ(d,"else")) return KEY_else;
5342 if (strEQ(d,"exit")) return -KEY_exit;
5343 if (strEQ(d,"eval")) return KEY_eval;
5344 if (strEQ(d,"exec")) return -KEY_exec;
5345 if (strEQ(d,"each")) return -KEY_each;
5348 if (strEQ(d,"elsif")) return KEY_elsif;
5351 if (strEQ(d,"exists")) return KEY_exists;
5352 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5355 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5356 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5359 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5362 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5363 if (strEQ(d,"endservent")) return -KEY_endservent;
5366 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5373 if (strEQ(d,"for")) return KEY_for;
5376 if (strEQ(d,"fork")) return -KEY_fork;
5379 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5380 if (strEQ(d,"flock")) return -KEY_flock;
5383 if (strEQ(d,"format")) return KEY_format;
5384 if (strEQ(d,"fileno")) return -KEY_fileno;
5387 if (strEQ(d,"foreach")) return KEY_foreach;
5390 if (strEQ(d,"formline")) return -KEY_formline;
5395 if (strnEQ(d,"get",3)) {
5400 if (strEQ(d,"ppid")) return -KEY_getppid;
5401 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5404 if (strEQ(d,"pwent")) return -KEY_getpwent;
5405 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5406 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5409 if (strEQ(d,"peername")) return -KEY_getpeername;
5410 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5411 if (strEQ(d,"priority")) return -KEY_getpriority;
5414 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5417 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5421 else if (*d == 'h') {
5422 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5423 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5424 if (strEQ(d,"hostent")) return -KEY_gethostent;
5426 else if (*d == 'n') {
5427 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5428 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5429 if (strEQ(d,"netent")) return -KEY_getnetent;
5431 else if (*d == 's') {
5432 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5433 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5434 if (strEQ(d,"servent")) return -KEY_getservent;
5435 if (strEQ(d,"sockname")) return -KEY_getsockname;
5436 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5438 else if (*d == 'g') {
5439 if (strEQ(d,"grent")) return -KEY_getgrent;
5440 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5441 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5443 else if (*d == 'l') {
5444 if (strEQ(d,"login")) return -KEY_getlogin;
5446 else if (strEQ(d,"c")) return -KEY_getc;
5451 if (strEQ(d,"gt")) return -KEY_gt;
5452 if (strEQ(d,"ge")) return -KEY_ge;
5455 if (strEQ(d,"grep")) return KEY_grep;
5456 if (strEQ(d,"goto")) return KEY_goto;
5457 if (strEQ(d,"glob")) return KEY_glob;
5460 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5465 if (strEQ(d,"hex")) return -KEY_hex;
5468 if (strEQ(d,"INIT")) return KEY_INIT;
5473 if (strEQ(d,"if")) return KEY_if;
5476 if (strEQ(d,"int")) return -KEY_int;
5479 if (strEQ(d,"index")) return -KEY_index;
5480 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5485 if (strEQ(d,"join")) return -KEY_join;
5489 if (strEQ(d,"keys")) return -KEY_keys;
5490 if (strEQ(d,"kill")) return -KEY_kill;
5496 if (strEQ(d,"lt")) return -KEY_lt;
5497 if (strEQ(d,"le")) return -KEY_le;
5498 if (strEQ(d,"lc")) return -KEY_lc;
5501 if (strEQ(d,"log")) return -KEY_log;
5504 if (strEQ(d,"last")) return KEY_last;
5505 if (strEQ(d,"link")) return -KEY_link;
5506 if (strEQ(d,"lock")) return -KEY_lock;
5509 if (strEQ(d,"local")) return KEY_local;
5510 if (strEQ(d,"lstat")) return -KEY_lstat;
5513 if (strEQ(d,"length")) return -KEY_length;
5514 if (strEQ(d,"listen")) return -KEY_listen;
5517 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5520 if (strEQ(d,"localtime")) return -KEY_localtime;
5526 case 1: return KEY_m;
5528 if (strEQ(d,"my")) return KEY_my;
5531 if (strEQ(d,"map")) return KEY_map;
5534 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5537 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5538 if (strEQ(d,"msgget")) return -KEY_msgget;
5539 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5540 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5545 if (strEQ(d,"next")) return KEY_next;
5546 if (strEQ(d,"ne")) return -KEY_ne;
5547 if (strEQ(d,"not")) return -KEY_not;
5548 if (strEQ(d,"no")) return KEY_no;
5553 if (strEQ(d,"or")) return -KEY_or;
5556 if (strEQ(d,"ord")) return -KEY_ord;
5557 if (strEQ(d,"oct")) return -KEY_oct;
5558 if (strEQ(d,"our")) return KEY_our;
5561 if (strEQ(d,"open")) return -KEY_open;
5564 if (strEQ(d,"opendir")) return -KEY_opendir;
5571 if (strEQ(d,"pop")) return -KEY_pop;
5572 if (strEQ(d,"pos")) return KEY_pos;
5575 if (strEQ(d,"push")) return -KEY_push;
5576 if (strEQ(d,"pack")) return -KEY_pack;
5577 if (strEQ(d,"pipe")) return -KEY_pipe;
5580 if (strEQ(d,"print")) return KEY_print;
5583 if (strEQ(d,"printf")) return KEY_printf;
5586 if (strEQ(d,"package")) return KEY_package;
5589 if (strEQ(d,"prototype")) return KEY_prototype;
5594 if (strEQ(d,"q")) return KEY_q;
5595 if (strEQ(d,"qr")) return KEY_qr;
5596 if (strEQ(d,"qq")) return KEY_qq;
5597 if (strEQ(d,"qw")) return KEY_qw;
5598 if (strEQ(d,"qx")) return KEY_qx;
5600 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5605 if (strEQ(d,"ref")) return -KEY_ref;
5608 if (strEQ(d,"read")) return -KEY_read;
5609 if (strEQ(d,"rand")) return -KEY_rand;
5610 if (strEQ(d,"recv")) return -KEY_recv;
5611 if (strEQ(d,"redo")) return KEY_redo;
5614 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5615 if (strEQ(d,"reset")) return -KEY_reset;
5618 if (strEQ(d,"return")) return KEY_return;
5619 if (strEQ(d,"rename")) return -KEY_rename;
5620 if (strEQ(d,"rindex")) return -KEY_rindex;
5623 if (strEQ(d,"require")) return -KEY_require;
5624 if (strEQ(d,"reverse")) return -KEY_reverse;
5625 if (strEQ(d,"readdir")) return -KEY_readdir;
5628 if (strEQ(d,"readlink")) return -KEY_readlink;
5629 if (strEQ(d,"readline")) return -KEY_readline;
5630 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5633 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5639 case 0: return KEY_s;
5641 if (strEQ(d,"scalar")) return KEY_scalar;
5646 if (strEQ(d,"seek")) return -KEY_seek;
5647 if (strEQ(d,"send")) return -KEY_send;
5650 if (strEQ(d,"semop")) return -KEY_semop;
5653 if (strEQ(d,"select")) return -KEY_select;
5654 if (strEQ(d,"semctl")) return -KEY_semctl;
5655 if (strEQ(d,"semget")) return -KEY_semget;
5658 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5659 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5662 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5663 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5666 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5669 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5670 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5671 if (strEQ(d,"setservent")) return -KEY_setservent;
5674 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5675 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5682 if (strEQ(d,"shift")) return -KEY_shift;
5685 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5686 if (strEQ(d,"shmget")) return -KEY_shmget;
5689 if (strEQ(d,"shmread")) return -KEY_shmread;
5692 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5693 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5698 if (strEQ(d,"sin")) return -KEY_sin;
5701 if (strEQ(d,"sleep")) return -KEY_sleep;
5704 if (strEQ(d,"sort")) return KEY_sort;
5705 if (strEQ(d,"socket")) return -KEY_socket;
5706 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5709 if (strEQ(d,"split")) return KEY_split;
5710 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5711 if (strEQ(d,"splice")) return -KEY_splice;
5714 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5717 if (strEQ(d,"srand")) return -KEY_srand;
5720 if (strEQ(d,"stat")) return -KEY_stat;
5721 if (strEQ(d,"study")) return KEY_study;
5724 if (strEQ(d,"substr")) return -KEY_substr;
5725 if (strEQ(d,"sub")) return KEY_sub;
5730 if (strEQ(d,"system")) return -KEY_system;
5733 if (strEQ(d,"symlink")) return -KEY_symlink;
5734 if (strEQ(d,"syscall")) return -KEY_syscall;
5735 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5736 if (strEQ(d,"sysread")) return -KEY_sysread;
5737 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5740 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5749 if (strEQ(d,"tr")) return KEY_tr;
5752 if (strEQ(d,"tie")) return KEY_tie;
5755 if (strEQ(d,"tell")) return -KEY_tell;
5756 if (strEQ(d,"tied")) return KEY_tied;
5757 if (strEQ(d,"time")) return -KEY_time;
5760 if (strEQ(d,"times")) return -KEY_times;
5763 if (strEQ(d,"telldir")) return -KEY_telldir;
5766 if (strEQ(d,"truncate")) return -KEY_truncate;
5773 if (strEQ(d,"uc")) return -KEY_uc;
5776 if (strEQ(d,"use")) return KEY_use;
5779 if (strEQ(d,"undef")) return KEY_undef;
5780 if (strEQ(d,"until")) return KEY_until;
5781 if (strEQ(d,"untie")) return KEY_untie;
5782 if (strEQ(d,"utime")) return -KEY_utime;
5783 if (strEQ(d,"umask")) return -KEY_umask;
5786 if (strEQ(d,"unless")) return KEY_unless;
5787 if (strEQ(d,"unpack")) return -KEY_unpack;
5788 if (strEQ(d,"unlink")) return -KEY_unlink;
5791 if (strEQ(d,"unshift")) return -KEY_unshift;
5792 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5797 if (strEQ(d,"values")) return -KEY_values;
5798 if (strEQ(d,"vec")) return -KEY_vec;
5803 if (strEQ(d,"warn")) return -KEY_warn;
5804 if (strEQ(d,"wait")) return -KEY_wait;
5807 if (strEQ(d,"while")) return KEY_while;
5808 if (strEQ(d,"write")) return -KEY_write;
5811 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5814 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5819 if (len == 1) return -KEY_x;
5820 if (strEQ(d,"xor")) return -KEY_xor;
5823 if (len == 1) return KEY_y;
5832 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5836 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5837 if (ckWARN(WARN_SYNTAX)) {
5839 for (w = s+2; *w && level; w++) {
5846 for (; *w && isSPACE(*w); w++) ;
5847 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5848 Perl_warner(aTHX_ WARN_SYNTAX,
5849 "%s (...) interpreted as function",name);
5852 while (s < PL_bufend && isSPACE(*s))
5856 while (s < PL_bufend && isSPACE(*s))
5858 if (isIDFIRST_lazy_if(s,UTF)) {
5860 while (isALNUM_lazy_if(s,UTF))
5862 while (s < PL_bufend && isSPACE(*s))
5867 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5871 Perl_croak(aTHX_ "No comma allowed after %s", what);
5876 /* Either returns sv, or mortalizes sv and returns a new SV*.
5877 Best used as sv=new_constant(..., sv, ...).
5878 If s, pv are NULL, calls subroutine with one argument,
5879 and type is used with error messages only. */
5882 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5886 HV *table = GvHV(PL_hintgv); /* ^H */
5890 const char *why1, *why2, *why3;
5892 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5895 why2 = strEQ(key,"charnames")
5896 ? "(possibly a missing \"use charnames ...\")"
5898 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5899 (type ? type: "undef"), why2);
5901 /* This is convoluted and evil ("goto considered harmful")
5902 * but I do not understand the intricacies of all the different
5903 * failure modes of %^H in here. The goal here is to make
5904 * the most probable error message user-friendly. --jhi */
5909 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5910 (type ? type: "undef"), why1, why2, why3);
5912 yyerror(SvPVX(msg));
5916 cvp = hv_fetch(table, key, strlen(key), FALSE);
5917 if (!cvp || !SvOK(*cvp)) {
5920 why3 = "} is not defined";
5923 sv_2mortal(sv); /* Parent created it permanently */
5926 pv = sv_2mortal(newSVpvn(s, len));
5928 typesv = sv_2mortal(newSVpv(type, 0));
5930 typesv = &PL_sv_undef;
5932 PUSHSTACKi(PERLSI_OVERLOAD);
5944 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5948 /* Check the eval first */
5949 if (!PL_in_eval && SvTRUE(ERRSV)) {
5951 sv_catpv(ERRSV, "Propagated");
5952 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5954 res = SvREFCNT_inc(sv);
5958 (void)SvREFCNT_inc(res);
5967 why1 = "Call to &{$^H{";
5969 why3 = "}} did not return a defined value";
5978 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5980 register char *d = dest;
5981 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5984 Perl_croak(aTHX_ ident_too_long);
5985 if (isALNUM(*s)) /* UTF handled below */
5987 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5992 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5996 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
5997 char *t = s + UTF8SKIP(s);
5998 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6000 if (d + (t - s) > e)
6001 Perl_croak(aTHX_ ident_too_long);
6002 Copy(s, d, t - s, char);
6015 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6025 e = d + destlen - 3; /* two-character token, ending NUL */
6027 while (isDIGIT(*s)) {
6029 Perl_croak(aTHX_ ident_too_long);
6036 Perl_croak(aTHX_ ident_too_long);
6037 if (isALNUM(*s)) /* UTF handled below */
6039 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6044 else if (*s == ':' && s[1] == ':') {
6048 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6049 char *t = s + UTF8SKIP(s);
6050 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6052 if (d + (t - s) > e)
6053 Perl_croak(aTHX_ ident_too_long);
6054 Copy(s, d, t - s, char);
6065 if (PL_lex_state != LEX_NORMAL)
6066 PL_lex_state = LEX_INTERPENDMAYBE;
6069 if (*s == '$' && s[1] &&
6070 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6083 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6088 if (isSPACE(s[-1])) {
6091 if (!SPACE_OR_TAB(ch)) {
6097 if (isIDFIRST_lazy_if(d,UTF)) {
6101 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6103 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6106 Copy(s, d, e - s, char);
6111 while ((isALNUM(*s) || *s == ':') && d < e)
6114 Perl_croak(aTHX_ ident_too_long);
6117 while (s < send && SPACE_OR_TAB(*s)) s++;
6118 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6119 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6120 const char *brack = *s == '[' ? "[...]" : "{...}";
6121 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6122 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6123 funny, dest, brack, funny, dest, brack);
6126 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6130 /* Handle extended ${^Foo} variables
6131 * 1999-02-27 mjd-perl-patch@plover.com */
6132 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6136 while (isALNUM(*s) && d < e) {
6140 Perl_croak(aTHX_ ident_too_long);
6145 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6146 PL_lex_state = LEX_INTERPEND;
6149 if (PL_lex_state == LEX_NORMAL) {
6150 if (ckWARN(WARN_AMBIGUOUS) &&
6151 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6153 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6154 "Ambiguous use of %c{%s} resolved to %c%s",
6155 funny, dest, funny, dest);
6160 s = bracket; /* let the parser handle it */
6164 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6165 PL_lex_state = LEX_INTERPEND;
6170 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6175 *pmfl |= PMf_GLOBAL;
6177 *pmfl |= PMf_CONTINUE;
6181 *pmfl |= PMf_MULTILINE;
6183 *pmfl |= PMf_SINGLELINE;
6185 *pmfl |= PMf_EXTENDED;
6189 S_scan_pat(pTHX_ char *start, I32 type)
6194 s = scan_str(start,FALSE,FALSE);
6196 Perl_croak(aTHX_ "Search pattern not terminated");
6198 pm = (PMOP*)newPMOP(type, 0);
6199 if (PL_multi_open == '?')
6200 pm->op_pmflags |= PMf_ONCE;
6202 while (*s && strchr("iomsx", *s))
6203 pmflag(&pm->op_pmflags,*s++);
6206 while (*s && strchr("iogcmsx", *s))
6207 pmflag(&pm->op_pmflags,*s++);
6209 pm->op_pmpermflags = pm->op_pmflags;
6211 PL_lex_op = (OP*)pm;
6212 yylval.ival = OP_MATCH;
6217 S_scan_subst(pTHX_ char *start)
6224 yylval.ival = OP_NULL;
6226 s = scan_str(start,FALSE,FALSE);
6229 Perl_croak(aTHX_ "Substitution pattern not terminated");
6231 if (s[-1] == PL_multi_open)
6234 first_start = PL_multi_start;
6235 s = scan_str(s,FALSE,FALSE);
6238 SvREFCNT_dec(PL_lex_stuff);
6239 PL_lex_stuff = Nullsv;
6241 Perl_croak(aTHX_ "Substitution replacement not terminated");
6243 PL_multi_start = first_start; /* so whole substitution is taken together */
6245 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6251 else if (strchr("iogcmsx", *s))
6252 pmflag(&pm->op_pmflags,*s++);
6259 PL_sublex_info.super_bufptr = s;
6260 PL_sublex_info.super_bufend = PL_bufend;
6262 pm->op_pmflags |= PMf_EVAL;
6263 repl = newSVpvn("",0);
6265 sv_catpv(repl, es ? "eval " : "do ");
6266 sv_catpvn(repl, "{ ", 2);
6267 sv_catsv(repl, PL_lex_repl);
6268 sv_catpvn(repl, " };", 2);
6270 SvREFCNT_dec(PL_lex_repl);
6274 pm->op_pmpermflags = pm->op_pmflags;
6275 PL_lex_op = (OP*)pm;
6276 yylval.ival = OP_SUBST;
6281 S_scan_trans(pTHX_ char *start)
6290 yylval.ival = OP_NULL;
6292 s = scan_str(start,FALSE,FALSE);
6294 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6295 if (s[-1] == PL_multi_open)
6298 s = scan_str(s,FALSE,FALSE);
6301 SvREFCNT_dec(PL_lex_stuff);
6302 PL_lex_stuff = Nullsv;
6304 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6307 complement = del = squash = 0;
6308 while (strchr("cds", *s)) {
6310 complement = OPpTRANS_COMPLEMENT;
6312 del = OPpTRANS_DELETE;
6314 squash = OPpTRANS_SQUASH;
6318 New(803, tbl, complement&&!del?258:256, short);
6319 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6320 o->op_private = del|squash|complement|
6321 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6322 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6325 yylval.ival = OP_TRANS;
6330 S_scan_heredoc(pTHX_ register char *s)
6333 I32 op_type = OP_SCALAR;
6340 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6344 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6347 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6348 if (*peek && strchr("`'\"",*peek)) {
6351 s = delimcpy(d, e, s, PL_bufend, term, &len);
6361 if (!isALNUM_lazy_if(s,UTF))
6362 deprecate("bare << to mean <<\"\"");
6363 for (; isALNUM_lazy_if(s,UTF); s++) {
6368 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6369 Perl_croak(aTHX_ "Delimiter for here document is too long");
6372 len = d - PL_tokenbuf;
6373 #ifndef PERL_STRICT_CR
6374 d = strchr(s, '\r');
6378 while (s < PL_bufend) {
6384 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6393 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6398 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6399 herewas = newSVpvn(s,PL_bufend-s);
6401 s--, herewas = newSVpvn(s,d-s);
6402 s += SvCUR(herewas);
6404 tmpstr = NEWSV(87,79);
6405 sv_upgrade(tmpstr, SVt_PVIV);
6410 else if (term == '`') {
6411 op_type = OP_BACKTICK;
6412 SvIVX(tmpstr) = '\\';
6416 PL_multi_start = CopLINE(PL_curcop);
6417 PL_multi_open = PL_multi_close = '<';
6418 term = *PL_tokenbuf;
6419 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6420 char *bufptr = PL_sublex_info.super_bufptr;
6421 char *bufend = PL_sublex_info.super_bufend;
6422 char *olds = s - SvCUR(herewas);
6423 s = strchr(bufptr, '\n');
6427 while (s < bufend &&
6428 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6430 CopLINE_inc(PL_curcop);
6433 CopLINE_set(PL_curcop, PL_multi_start);
6434 missingterm(PL_tokenbuf);
6436 sv_setpvn(herewas,bufptr,d-bufptr+1);
6437 sv_setpvn(tmpstr,d+1,s-d);
6439 sv_catpvn(herewas,s,bufend-s);
6440 (void)strcpy(bufptr,SvPVX(herewas));
6447 while (s < PL_bufend &&
6448 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6450 CopLINE_inc(PL_curcop);
6452 if (s >= PL_bufend) {
6453 CopLINE_set(PL_curcop, PL_multi_start);
6454 missingterm(PL_tokenbuf);
6456 sv_setpvn(tmpstr,d+1,s-d);
6458 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6460 sv_catpvn(herewas,s,PL_bufend-s);
6461 sv_setsv(PL_linestr,herewas);
6462 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6463 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6464 PL_last_lop = PL_last_uni = Nullch;
6467 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6468 while (s >= PL_bufend) { /* multiple line string? */
6470 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6471 CopLINE_set(PL_curcop, PL_multi_start);
6472 missingterm(PL_tokenbuf);
6474 CopLINE_inc(PL_curcop);
6475 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6476 PL_last_lop = PL_last_uni = Nullch;
6477 #ifndef PERL_STRICT_CR
6478 if (PL_bufend - PL_linestart >= 2) {
6479 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6480 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6482 PL_bufend[-2] = '\n';
6484 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6486 else if (PL_bufend[-1] == '\r')
6487 PL_bufend[-1] = '\n';
6489 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6490 PL_bufend[-1] = '\n';
6492 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6493 SV *sv = NEWSV(88,0);
6495 sv_upgrade(sv, SVt_PVMG);
6496 sv_setsv(sv,PL_linestr);
6497 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6499 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6502 sv_catsv(PL_linestr,herewas);
6503 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6507 sv_catsv(tmpstr,PL_linestr);
6512 PL_multi_end = CopLINE(PL_curcop);
6513 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6514 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6515 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6517 SvREFCNT_dec(herewas);
6518 if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6520 PL_lex_stuff = tmpstr;
6521 yylval.ival = op_type;
6526 takes: current position in input buffer
6527 returns: new position in input buffer
6528 side-effects: yylval and lex_op are set.
6533 <FH> read from filehandle
6534 <pkg::FH> read from package qualified filehandle
6535 <pkg'FH> read from package qualified filehandle
6536 <$fh> read from filehandle in $fh
6542 S_scan_inputsymbol(pTHX_ char *start)
6544 register char *s = start; /* current position in buffer */
6550 d = PL_tokenbuf; /* start of temp holding space */
6551 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6552 end = strchr(s, '\n');
6555 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6557 /* die if we didn't have space for the contents of the <>,
6558 or if it didn't end, or if we see a newline
6561 if (len >= sizeof PL_tokenbuf)
6562 Perl_croak(aTHX_ "Excessively long <> operator");
6564 Perl_croak(aTHX_ "Unterminated <> operator");
6569 Remember, only scalar variables are interpreted as filehandles by
6570 this code. Anything more complex (e.g., <$fh{$num}>) will be
6571 treated as a glob() call.
6572 This code makes use of the fact that except for the $ at the front,
6573 a scalar variable and a filehandle look the same.
6575 if (*d == '$' && d[1]) d++;
6577 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6578 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6581 /* If we've tried to read what we allow filehandles to look like, and
6582 there's still text left, then it must be a glob() and not a getline.
6583 Use scan_str to pull out the stuff between the <> and treat it
6584 as nothing more than a string.
6587 if (d - PL_tokenbuf != len) {
6588 yylval.ival = OP_GLOB;
6590 s = scan_str(start,FALSE,FALSE);
6592 Perl_croak(aTHX_ "Glob not terminated");
6596 /* we're in a filehandle read situation */
6599 /* turn <> into <ARGV> */
6601 (void)strcpy(d,"ARGV");
6603 /* if <$fh>, create the ops to turn the variable into a
6609 /* try to find it in the pad for this block, otherwise find
6610 add symbol table ops
6612 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6613 OP *o = newOP(OP_PADSV, 0);
6615 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6618 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6619 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6620 newUNOP(OP_RV2SV, 0,
6621 newGVOP(OP_GV, 0, gv)));
6623 PL_lex_op->op_flags |= OPf_SPECIAL;
6624 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6625 yylval.ival = OP_NULL;
6628 /* If it's none of the above, it must be a literal filehandle
6629 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6631 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6632 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6633 yylval.ival = OP_NULL;
6642 takes: start position in buffer
6643 keep_quoted preserve \ on the embedded delimiter(s)
6644 keep_delims preserve the delimiters around the string
6645 returns: position to continue reading from buffer
6646 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6647 updates the read buffer.
6649 This subroutine pulls a string out of the input. It is called for:
6650 q single quotes q(literal text)
6651 ' single quotes 'literal text'
6652 qq double quotes qq(interpolate $here please)
6653 " double quotes "interpolate $here please"
6654 qx backticks qx(/bin/ls -l)
6655 ` backticks `/bin/ls -l`
6656 qw quote words @EXPORT_OK = qw( func() $spam )
6657 m// regexp match m/this/
6658 s/// regexp substitute s/this/that/
6659 tr/// string transliterate tr/this/that/
6660 y/// string transliterate y/this/that/
6661 ($*@) sub prototypes sub foo ($)
6662 (stuff) sub attr parameters sub foo : attr(stuff)
6663 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6665 In most of these cases (all but <>, patterns and transliterate)
6666 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6667 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6668 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6671 It skips whitespace before the string starts, and treats the first
6672 character as the delimiter. If the delimiter is one of ([{< then
6673 the corresponding "close" character )]}> is used as the closing
6674 delimiter. It allows quoting of delimiters, and if the string has
6675 balanced delimiters ([{<>}]) it allows nesting.
6677 On success, the SV with the resulting string is put into lex_stuff or,
6678 if that is already non-NULL, into lex_repl. The second case occurs only
6679 when parsing the RHS of the special constructs s/// and tr/// (y///).
6680 For convenience, the terminating delimiter character is stuffed into
6685 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6687 SV *sv; /* scalar value: string */
6688 char *tmps; /* temp string, used for delimiter matching */
6689 register char *s = start; /* current position in the buffer */
6690 register char term; /* terminating character */
6691 register char *to; /* current position in the sv's data */
6692 I32 brackets = 1; /* bracket nesting level */
6693 bool has_utf8 = FALSE; /* is there any utf8 content? */
6695 /* skip space before the delimiter */
6699 /* mark where we are, in case we need to report errors */
6702 /* after skipping whitespace, the next character is the terminator */
6704 if (!UTF8_IS_INVARIANT((U8)term) && UTF)
6707 /* mark where we are */
6708 PL_multi_start = CopLINE(PL_curcop);
6709 PL_multi_open = term;
6711 /* find corresponding closing delimiter */
6712 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6714 PL_multi_close = term;
6716 /* create a new SV to hold the contents. 87 is leak category, I'm
6717 assuming. 79 is the SV's initial length. What a random number. */
6719 sv_upgrade(sv, SVt_PVIV);
6721 (void)SvPOK_only(sv); /* validate pointer */
6723 /* move past delimiter and try to read a complete string */
6725 sv_catpvn(sv, s, 1);
6728 /* extend sv if need be */
6729 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6730 /* set 'to' to the next character in the sv's string */
6731 to = SvPVX(sv)+SvCUR(sv);
6733 /* if open delimiter is the close delimiter read unbridle */
6734 if (PL_multi_open == PL_multi_close) {
6735 for (; s < PL_bufend; s++,to++) {
6736 /* embedded newlines increment the current line number */
6737 if (*s == '\n' && !PL_rsfp)
6738 CopLINE_inc(PL_curcop);
6739 /* handle quoted delimiters */
6740 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6741 if (!keep_quoted && s[1] == term)
6743 /* any other quotes are simply copied straight through */
6747 /* terminate when run out of buffer (the for() condition), or
6748 have found the terminator */
6749 else if (*s == term)
6751 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6757 /* if the terminator isn't the same as the start character (e.g.,
6758 matched brackets), we have to allow more in the quoting, and
6759 be prepared for nested brackets.
6762 /* read until we run out of string, or we find the terminator */
6763 for (; s < PL_bufend; s++,to++) {
6764 /* embedded newlines increment the line count */
6765 if (*s == '\n' && !PL_rsfp)
6766 CopLINE_inc(PL_curcop);
6767 /* backslashes can escape the open or closing characters */
6768 if (*s == '\\' && s+1 < PL_bufend) {
6770 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6775 /* allow nested opens and closes */
6776 else if (*s == PL_multi_close && --brackets <= 0)
6778 else if (*s == PL_multi_open)
6780 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
6785 /* terminate the copied string and update the sv's end-of-string */
6787 SvCUR_set(sv, to - SvPVX(sv));
6790 * this next chunk reads more into the buffer if we're not done yet
6794 break; /* handle case where we are done yet :-) */
6796 #ifndef PERL_STRICT_CR
6797 if (to - SvPVX(sv) >= 2) {
6798 if ((to[-2] == '\r' && to[-1] == '\n') ||
6799 (to[-2] == '\n' && to[-1] == '\r'))
6803 SvCUR_set(sv, to - SvPVX(sv));
6805 else if (to[-1] == '\r')
6808 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6812 /* if we're out of file, or a read fails, bail and reset the current
6813 line marker so we can report where the unterminated string began
6816 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6818 CopLINE_set(PL_curcop, PL_multi_start);
6821 /* we read a line, so increment our line counter */
6822 CopLINE_inc(PL_curcop);
6824 /* update debugger info */
6825 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6826 SV *sv = NEWSV(88,0);
6828 sv_upgrade(sv, SVt_PVMG);
6829 sv_setsv(sv,PL_linestr);
6830 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6833 /* having changed the buffer, we must update PL_bufend */
6834 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6835 PL_last_lop = PL_last_uni = Nullch;
6838 /* at this point, we have successfully read the delimited string */
6841 sv_catpvn(sv, s, 1);
6844 PL_multi_end = CopLINE(PL_curcop);
6847 /* if we allocated too much space, give some back */
6848 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6849 SvLEN_set(sv, SvCUR(sv) + 1);
6850 Renew(SvPVX(sv), SvLEN(sv), char);
6853 /* decide whether this is the first or second quoted string we've read
6866 takes: pointer to position in buffer
6867 returns: pointer to new position in buffer
6868 side-effects: builds ops for the constant in yylval.op
6870 Read a number in any of the formats that Perl accepts:
6872 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6873 [\d_]+(\.[\d_]*)?[Ee](\d+)
6875 Underbars (_) are allowed in decimal numbers. If -w is on,
6876 underbars before a decimal point must be at three digit intervals.
6878 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6881 If it reads a number without a decimal point or an exponent, it will
6882 try converting the number to an integer and see if it can do so
6883 without loss of precision.
6887 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6889 register char *s = start; /* current position in buffer */
6890 register char *d; /* destination in temp buffer */
6891 register char *e; /* end of temp buffer */
6892 NV nv; /* number read, as a double */
6893 SV *sv = Nullsv; /* place to put the converted number */
6894 bool floatit; /* boolean: int or float? */
6895 char *lastub = 0; /* position of last underbar */
6896 static char number_too_long[] = "Number too long";
6898 /* We use the first character to decide what type of number this is */
6902 Perl_croak(aTHX_ "panic: scan_num");
6904 /* if it starts with a 0, it could be an octal number, a decimal in
6905 0.13 disguise, or a hexadecimal number, or a binary number. */
6909 u holds the "number so far"
6910 shift the power of 2 of the base
6911 (hex == 4, octal == 3, binary == 1)
6912 overflowed was the number more than we can hold?
6914 Shift is used when we add a digit. It also serves as an "are
6915 we in octal/hex/binary?" indicator to disallow hex characters
6921 bool overflowed = FALSE;
6922 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6923 static char* bases[5] = { "", "binary", "", "octal",
6925 static char* Bases[5] = { "", "Binary", "", "Octal",
6927 static char *maxima[5] = { "",
6928 "0b11111111111111111111111111111111",
6932 char *base, *Base, *max;
6938 } else if (s[1] == 'b') {
6942 /* check for a decimal in disguise */
6943 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6945 /* so it must be octal */
6949 base = bases[shift];
6950 Base = Bases[shift];
6951 max = maxima[shift];
6953 /* read the rest of the number */
6955 /* x is used in the overflow test,
6956 b is the digit we're adding on. */
6961 /* if we don't mention it, we're done */
6970 /* 8 and 9 are not octal */
6973 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6977 case '2': case '3': case '4':
6978 case '5': case '6': case '7':
6980 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6984 b = *s++ & 15; /* ASCII digit -> value of digit */
6988 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6989 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6990 /* make sure they said 0x */
6995 /* Prepare to put the digit we have onto the end
6996 of the number so far. We check for overflows.
7001 x = u << shift; /* make room for the digit */
7003 if ((x >> shift) != u
7004 && !(PL_hints & HINT_NEW_BINARY)) {
7007 if (ckWARN_d(WARN_OVERFLOW))
7008 Perl_warner(aTHX_ WARN_OVERFLOW,
7009 "Integer overflow in %s number",
7012 u = x | b; /* add the digit to the end */
7015 n *= nvshift[shift];
7016 /* If an NV has not enough bits in its
7017 * mantissa to represent an UV this summing of
7018 * small low-order numbers is a waste of time
7019 * (because the NV cannot preserve the
7020 * low-order bits anyway): we could just
7021 * remember when did we overflow and in the
7022 * end just multiply n by the right
7030 /* if we get here, we had success: make a scalar value from
7036 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7037 Perl_warner(aTHX_ WARN_PORTABLE,
7038 "%s number > %s non-portable",
7044 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7045 Perl_warner(aTHX_ WARN_PORTABLE,
7046 "%s number > %s non-portable",
7051 if (PL_hints & HINT_NEW_BINARY)
7052 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7057 handle decimal numbers.
7058 we're also sent here when we read a 0 as the first digit
7060 case '1': case '2': case '3': case '4': case '5':
7061 case '6': case '7': case '8': case '9': case '.':
7064 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7067 /* read next group of digits and _ and copy into d */
7068 while (isDIGIT(*s) || *s == '_') {
7069 /* skip underscores, checking for misplaced ones
7073 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7074 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7078 /* check for end of fixed-length buffer */
7080 Perl_croak(aTHX_ number_too_long);
7081 /* if we're ok, copy the character */
7086 /* final misplaced underbar check */
7087 if (lastub && s - lastub != 3) {
7088 if (ckWARN(WARN_SYNTAX))
7089 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7092 /* read a decimal portion if there is one. avoid
7093 3..5 being interpreted as the number 3. followed
7096 if (*s == '.' && s[1] != '.') {
7100 /* copy, ignoring underbars, until we run out of
7101 digits. Note: no misplaced underbar checks!
7103 for (; isDIGIT(*s) || *s == '_'; s++) {
7104 /* fixed length buffer check */
7106 Perl_croak(aTHX_ number_too_long);
7110 if (*s == '.' && isDIGIT(s[1])) {
7111 /* oops, it's really a v-string, but without the "v" */
7117 /* read exponent part, if present */
7118 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7122 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7123 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7125 /* allow positive or negative exponent */
7126 if (*s == '+' || *s == '-')
7129 /* read digits of exponent (no underbars :-) */
7130 while (isDIGIT(*s)) {
7132 Perl_croak(aTHX_ number_too_long);
7137 /* terminate the string */
7140 /* make an sv from the string */
7143 #if defined(Strtol) && defined(Strtoul)
7146 strtol/strtoll sets errno to ERANGE if the number is too big
7147 for an integer. We try to do an integer conversion first
7148 if no characters indicating "float" have been found.
7155 if (*PL_tokenbuf == '-')
7156 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7158 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7160 floatit = TRUE; /* Probably just too large. */
7161 else if (*PL_tokenbuf == '-')
7163 else if (uv <= IV_MAX)
7164 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7169 nv = Atof(PL_tokenbuf);
7174 No working strtou?ll?.
7176 Unfortunately atol() doesn't do range checks (returning
7177 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7178 everywhere [1], so we cannot use use atol() (or atoll()).
7179 If we could, they would be used, as Atol(), very much like
7180 Strtol() and Strtoul() are used above.
7182 [1] XXX Configure test needed to check for atol()
7183 (and atoll()) overflow behaviour XXX
7187 We need to do this the hard way. */
7189 nv = Atof(PL_tokenbuf);
7191 /* See if we can make do with an integer value without loss of
7192 precision. We use U_V to cast to a UV, because some
7193 compilers have issues. Then we try casting it back and see
7194 if it was the same [1]. We only do this if we know we
7195 specifically read an integer. If floatit is true, then we
7196 don't need to do the conversion at all.
7198 [1] Note that this is lossy if our NVs cannot preserve our
7199 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7200 and NV_PRESERVES_UV_BITS (a number), but in general we really
7201 do hope all such potentially lossy platforms have strtou?ll?
7202 to do a lossless IV/UV conversion.
7204 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7205 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7206 as NV_DIG and NV_MANT_DIG)?
7212 if (!floatit && (NV)uv == nv) {
7214 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7222 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7223 (PL_hints & HINT_NEW_INTEGER) )
7224 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7225 (floatit ? "float" : "integer"),
7229 /* if it starts with a v, it could be a v-string */
7235 while (isDIGIT(*pos) || *pos == '_')
7237 if (!isALPHA(*pos)) {
7239 U8 tmpbuf[UTF8_MAXLEN+1];
7241 s++; /* get past 'v' */
7244 sv_setpvn(sv, "", 0);
7247 if (*s == '0' && isDIGIT(s[1]))
7248 yyerror("Octal number in vector unsupported");
7251 /* this is atoi() that tolerates underscores */
7254 while (--end >= s) {
7259 rev += (*end - '0') * mult;
7261 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7262 Perl_warner(aTHX_ WARN_OVERFLOW,
7263 "Integer overflow in decimal number");
7266 /* Append native character for the rev point */
7267 tmpend = uvchr_to_utf8(tmpbuf, rev);
7268 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7269 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
7271 if (*pos == '.' && isDIGIT(pos[1]))
7277 while (isDIGIT(*pos) || *pos == '_')
7287 /* make the op for the constant and return */
7290 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7292 lvalp->opval = Nullop;
7298 S_scan_formline(pTHX_ register char *s)
7302 SV *stuff = newSVpvn("",0);
7303 bool needargs = FALSE;
7306 if (*s == '.' || *s == /*{*/'}') {
7308 #ifdef PERL_STRICT_CR
7309 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7311 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7313 if (*t == '\n' || t == PL_bufend)
7316 if (PL_in_eval && !PL_rsfp) {
7317 eol = strchr(s,'\n');
7322 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7324 for (t = s; t < eol; t++) {
7325 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7327 goto enough; /* ~~ must be first line in formline */
7329 if (*t == '@' || *t == '^')
7332 sv_catpvn(stuff, s, eol-s);
7333 #ifndef PERL_STRICT_CR
7334 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7335 char *end = SvPVX(stuff) + SvCUR(stuff);
7344 s = filter_gets(PL_linestr, PL_rsfp, 0);
7345 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7346 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7347 PL_last_lop = PL_last_uni = Nullch;
7350 yyerror("Format not terminated");
7360 PL_lex_state = LEX_NORMAL;
7361 PL_nextval[PL_nexttoke].ival = 0;
7365 PL_lex_state = LEX_FORMLINE;
7366 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7368 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7372 SvREFCNT_dec(stuff);
7373 PL_lex_formbrack = 0;
7384 PL_cshlen = strlen(PL_cshname);
7389 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7391 I32 oldsavestack_ix = PL_savestack_ix;
7392 CV* outsidecv = PL_compcv;
7396 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7398 SAVEI32(PL_subline);
7399 save_item(PL_subname);
7402 SAVESPTR(PL_comppad_name);
7403 SAVESPTR(PL_compcv);
7404 SAVEI32(PL_comppad_name_fill);
7405 SAVEI32(PL_min_intro_pending);
7406 SAVEI32(PL_max_intro_pending);
7407 SAVEI32(PL_pad_reset_pending);
7409 PL_compcv = (CV*)NEWSV(1104,0);
7410 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7411 CvFLAGS(PL_compcv) |= flags;
7413 PL_comppad = newAV();
7414 av_push(PL_comppad, Nullsv);
7415 PL_curpad = AvARRAY(PL_comppad);
7416 PL_comppad_name = newAV();
7417 PL_comppad_name_fill = 0;
7418 PL_min_intro_pending = 0;
7420 PL_subline = CopLINE(PL_curcop);
7422 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7423 PL_curpad[0] = (SV*)newAV();
7424 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7425 #endif /* USE_THREADS */
7427 comppadlist = newAV();
7428 AvREAL_off(comppadlist);
7429 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7430 av_store(comppadlist, 1, (SV*)PL_comppad);
7432 CvPADLIST(PL_compcv) = comppadlist;
7433 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7435 CvOWNER(PL_compcv) = 0;
7436 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7437 MUTEX_INIT(CvMUTEXP(PL_compcv));
7438 #endif /* USE_THREADS */
7440 return oldsavestack_ix;
7444 #pragma segment Perl_yylex
7447 Perl_yywarn(pTHX_ char *s)
7449 PL_in_eval |= EVAL_WARNONLY;
7451 PL_in_eval &= ~EVAL_WARNONLY;
7456 Perl_yyerror(pTHX_ char *s)
7459 char *context = NULL;
7463 if (!yychar || (yychar == ';' && !PL_rsfp))
7465 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7466 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7467 while (isSPACE(*PL_oldoldbufptr))
7469 context = PL_oldoldbufptr;
7470 contlen = PL_bufptr - PL_oldoldbufptr;
7472 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7473 PL_oldbufptr != PL_bufptr) {
7474 while (isSPACE(*PL_oldbufptr))
7476 context = PL_oldbufptr;
7477 contlen = PL_bufptr - PL_oldbufptr;
7479 else if (yychar > 255)
7480 where = "next token ???";
7481 #ifdef USE_PURE_BISON
7482 /* GNU Bison sets the value -2 */
7483 else if (yychar == -2) {
7485 else if ((yychar & 127) == 127) {
7487 if (PL_lex_state == LEX_NORMAL ||
7488 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7489 where = "at end of line";
7490 else if (PL_lex_inpat)
7491 where = "within pattern";
7493 where = "within string";
7496 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7498 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7499 else if (isPRINT_LC(yychar))
7500 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7502 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7503 where = SvPVX(where_sv);
7505 msg = sv_2mortal(newSVpv(s, 0));
7506 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7507 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7509 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7511 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7512 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7513 Perl_sv_catpvf(aTHX_ msg,
7514 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7515 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7518 if (PL_in_eval & EVAL_WARNONLY)
7519 Perl_warn(aTHX_ "%"SVf, msg);
7522 if (PL_error_count >= 10) {
7523 if (PL_in_eval && SvCUR(ERRSV))
7524 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7525 ERRSV, CopFILE(PL_curcop));
7527 Perl_croak(aTHX_ "%s has too many errors.\n",
7528 CopFILE(PL_curcop));
7531 PL_in_my_stash = Nullhv;
7535 #pragma segment Main
7539 S_swallow_bom(pTHX_ U8 *s)
7542 slen = SvCUR(PL_linestr);
7546 /* UTF-16 little-endian */
7547 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7548 Perl_croak(aTHX_ "Unsupported script encoding");
7549 #ifndef PERL_NO_UTF16_FILTER
7550 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7552 if (PL_bufend > (char*)s) {
7556 filter_add(utf16rev_textfilter, NULL);
7557 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7558 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7559 PL_bufend - (char*)s - 1,
7561 Copy(news, s, newlen, U8);
7562 SvCUR_set(PL_linestr, newlen);
7563 PL_bufend = SvPVX(PL_linestr) + newlen;
7564 news[newlen++] = '\0';
7568 Perl_croak(aTHX_ "Unsupported script encoding");
7573 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7574 #ifndef PERL_NO_UTF16_FILTER
7575 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7577 if (PL_bufend > (char *)s) {
7581 filter_add(utf16_textfilter, NULL);
7582 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7583 PL_bufend = (char*)utf16_to_utf8(s, news,
7584 PL_bufend - (char*)s,
7586 Copy(news, s, newlen, U8);
7587 SvCUR_set(PL_linestr, newlen);
7588 PL_bufend = SvPVX(PL_linestr) + newlen;
7589 news[newlen++] = '\0';
7593 Perl_croak(aTHX_ "Unsupported script encoding");
7598 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7599 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7604 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7605 s[2] == 0xFE && s[3] == 0xFF)
7607 Perl_croak(aTHX_ "Unsupported script encoding");
7619 * Restore a source filter.
7623 restore_rsfp(pTHXo_ void *f)
7625 PerlIO *fp = (PerlIO*)f;
7627 if (PL_rsfp == PerlIO_stdin())
7628 PerlIO_clearerr(PL_rsfp);
7629 else if (PL_rsfp && (PL_rsfp != fp))
7630 PerlIO_close(PL_rsfp);
7634 #ifndef PERL_NO_UTF16_FILTER
7636 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7638 I32 count = FILTER_READ(idx+1, sv, maxlen);
7643 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7644 if (!*SvPV_nolen(sv))
7645 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7648 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7649 sv_usepvn(sv, (char*)tmps, tend - tmps);
7655 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7657 I32 count = FILTER_READ(idx+1, sv, maxlen);
7662 if (!*SvPV_nolen(sv))
7663 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7666 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7667 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7668 sv_usepvn(sv, (char*)tmps, tend - tmps);