3 * Copyright (c) 1991-2000, 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
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
78 # define YYMAXLEVEL 100
80 YYSTYPE* yylval_pointer[YYMAXLEVEL];
81 int* yychar_pointer[YYMAXLEVEL];
85 # define yylval (*yylval_pointer[yyactlevel])
86 # define yychar (*yychar_pointer[yyactlevel])
87 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
89 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
94 /* CLINE is a macro that ensures PL_copline has a sane value */
99 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
102 * Convenience functions to return different tokens and prime the
103 * lexer for the next token. They all take an argument.
105 * TOKEN : generic token (used for '(', DOLSHARP, etc)
106 * OPERATOR : generic operator
107 * AOPERATOR : assignment operator
108 * PREBLOCK : beginning the block after an if, while, foreach, ...
109 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
110 * PREREF : *EXPR where EXPR is not a simple identifier
111 * TERM : expression term
112 * LOOPX : loop exiting command (goto, last, dump, etc)
113 * FTST : file test operator
114 * FUN0 : zero-argument function
115 * FUN1 : not used, except for not, which isn't a UNIOP
116 * BOop : bitwise or or xor
118 * SHop : shift operator
119 * PWop : power operator
120 * PMop : pattern-matching operator
121 * Aop : addition-level operator
122 * Mop : multiplication-level operator
123 * Eop : equality-testing operator
124 * Rop : relational operator <= != gt
126 * Also see LOP and lop() below.
129 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
130 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
131 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
132 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
133 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
134 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
135 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
136 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
137 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
138 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
139 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
140 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
141 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
142 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
143 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
144 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
145 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
146 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
147 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
148 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
150 /* This bit of chicanery makes a unary function followed by
151 * a parenthesis into a function with one argument, highest precedence.
153 #define UNI(f) return(yylval.ival = f, \
156 PL_last_uni = PL_oldbufptr, \
157 PL_last_lop_op = f, \
158 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
160 #define UNIBRACK(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
165 /* grandfather return to old style */
166 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
171 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
172 * into an OP_ANDASSIGN or OP_ORASSIGN
176 S_ao(pTHX_ int toketype)
178 if (*PL_bufptr == '=') {
180 if (toketype == ANDAND)
181 yylval.ival = OP_ANDASSIGN;
182 else if (toketype == OROR)
183 yylval.ival = OP_ORASSIGN;
191 * When Perl expects an operator and finds something else, no_op
192 * prints the warning. It always prints "<something> found where
193 * operator expected. It prints "Missing semicolon on previous line?"
194 * if the surprise occurs at the start of the line. "do you need to
195 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
196 * where the compiler doesn't know if foo is a method call or a function.
197 * It prints "Missing operator before end of line" if there's nothing
198 * after the missing operator, or "... before <...>" if there is something
199 * after the missing operator.
203 S_no_op(pTHX_ char *what, char *s)
205 char *oldbp = PL_bufptr;
206 bool is_first = (PL_oldbufptr == PL_linestart);
212 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
214 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
215 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
217 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
218 if (t < PL_bufptr && isSPACE(*t))
219 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
220 t - PL_oldoldbufptr, PL_oldoldbufptr);
224 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
231 * Complain about missing quote/regexp/heredoc terminator.
232 * If it's called with (char *)NULL then it cauterizes the line buffer.
233 * If we're in a delimited string and the delimiter is a control
234 * character, it's reformatted into a two-char sequence like ^C.
239 S_missingterm(pTHX_ char *s)
244 char *nl = strrchr(s,'\n');
250 iscntrl(PL_multi_close)
252 PL_multi_close < 32 || PL_multi_close == 127
256 tmpbuf[1] = toCTRL(PL_multi_close);
262 *tmpbuf = PL_multi_close;
266 q = strchr(s,'"') ? '\'' : '"';
267 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
275 Perl_deprecate(pTHX_ char *s)
278 if (ckWARN(WARN_DEPRECATED))
279 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
284 * Deprecate a comma-less variable list.
290 deprecate("comma-less variable list");
294 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
295 * utf16-to-utf8-reversed.
298 #ifdef PERL_CR_FILTER
302 register char *s = SvPVX(sv);
303 register char *e = s + SvCUR(sv);
304 /* outer loop optimized to do nothing if there are no CR-LFs */
306 if (*s++ == '\r' && *s == '\n') {
307 /* hit a CR-LF, need to copy the rest */
308 register char *d = s - 1;
311 if (*s == '\r' && s[1] == '\n')
322 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
324 I32 count = FILTER_READ(idx+1, sv, maxlen);
325 if (count > 0 && !maxlen)
333 * Initialize variables. Uses the Perl save_stack to save its state (for
334 * recursive calls to the parser).
338 Perl_lex_start(pTHX_ SV *line)
344 SAVEI32(PL_lex_dojoin);
345 SAVEI32(PL_lex_brackets);
346 SAVEI32(PL_lex_casemods);
347 SAVEI32(PL_lex_starts);
348 SAVEI32(PL_lex_state);
349 SAVEVPTR(PL_lex_inpat);
350 SAVEI32(PL_lex_inwhat);
351 if (PL_lex_state == LEX_KNOWNEXT) {
352 I32 toke = PL_nexttoke;
353 while (--toke >= 0) {
354 SAVEI32(PL_nexttype[toke]);
355 SAVEVPTR(PL_nextval[toke]);
357 SAVEI32(PL_nexttoke);
359 SAVECOPLINE(PL_curcop);
362 SAVEPPTR(PL_oldbufptr);
363 SAVEPPTR(PL_oldoldbufptr);
364 SAVEPPTR(PL_linestart);
365 SAVESPTR(PL_linestr);
366 SAVEPPTR(PL_lex_brackstack);
367 SAVEPPTR(PL_lex_casestack);
368 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
369 SAVESPTR(PL_lex_stuff);
370 SAVEI32(PL_lex_defer);
371 SAVEI32(PL_sublex_info.sub_inwhat);
372 SAVESPTR(PL_lex_repl);
374 SAVEINT(PL_lex_expect);
376 PL_lex_state = LEX_NORMAL;
380 New(899, PL_lex_brackstack, 120, char);
381 New(899, PL_lex_casestack, 12, char);
382 SAVEFREEPV(PL_lex_brackstack);
383 SAVEFREEPV(PL_lex_casestack);
385 *PL_lex_casestack = '\0';
388 PL_lex_stuff = Nullsv;
389 PL_lex_repl = Nullsv;
393 PL_sublex_info.sub_inwhat = 0;
395 if (SvREADONLY(PL_linestr))
396 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
397 s = SvPV(PL_linestr, len);
398 if (len && s[len-1] != ';') {
399 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
400 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
401 sv_catpvn(PL_linestr, "\n;", 2);
403 SvTEMP_off(PL_linestr);
404 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
405 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
407 PL_rs = newSVpvn("\n", 1);
413 * Finalizer for lexing operations. Must be called when the parser is
414 * done with the lexer.
420 PL_doextract = FALSE;
425 * This subroutine has nothing to do with tilting, whether at windmills
426 * or pinball tables. Its name is short for "increment line". It
427 * increments the current line number in CopLINE(PL_curcop) and checks
428 * to see whether the line starts with a comment of the form
429 * # line 500 "foo.pm"
430 * If so, it sets the current line number and file to the values in the comment.
434 S_incline(pTHX_ char *s)
442 CopLINE_inc(PL_curcop);
445 while (SPACE_OR_TAB(*s)) s++;
446 if (strnEQ(s, "line", 4))
450 if (*s == ' ' || *s == '\t')
454 while (SPACE_OR_TAB(*s)) s++;
460 while (SPACE_OR_TAB(*s))
462 if (*s == '"' && (t = strchr(s+1, '"'))) {
467 for (t = s; !isSPACE(*t); t++) ;
470 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
472 if (*e != '\n' && *e != '\0')
473 return; /* false alarm */
479 Safefree(CopFILE(PL_curcop));
481 SvREFCNT_dec(CopFILEGV(PL_curcop));
483 CopFILE_set(PL_curcop, s);
486 CopLINE_set(PL_curcop, atoi(n)-1);
491 * Called to gobble the appropriate amount and type of whitespace.
492 * Skips comments as well.
496 S_skipspace(pTHX_ register char *s)
499 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
500 while (s < PL_bufend && SPACE_OR_TAB(*s))
506 SSize_t oldprevlen, oldoldprevlen;
507 SSize_t oldloplen, oldunilen;
508 while (s < PL_bufend && isSPACE(*s)) {
509 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
514 if (s < PL_bufend && *s == '#') {
515 while (s < PL_bufend && *s != '\n')
519 if (PL_in_eval && !PL_rsfp) {
526 /* only continue to recharge the buffer if we're at the end
527 * of the buffer, we're not reading from a source filter, and
528 * we're in normal lexing mode
530 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
531 PL_lex_state == LEX_FORMLINE)
534 /* try to recharge the buffer */
535 if ((s = filter_gets(PL_linestr, PL_rsfp,
536 (prevlen = SvCUR(PL_linestr)))) == Nullch)
538 /* end of file. Add on the -p or -n magic */
539 if (PL_minus_n || PL_minus_p) {
540 sv_setpv(PL_linestr,PL_minus_p ?
541 ";}continue{print or die qq(-p destination: $!\\n)" :
543 sv_catpv(PL_linestr,";}");
544 PL_minus_n = PL_minus_p = 0;
547 sv_setpv(PL_linestr,";");
549 /* reset variables for next time we lex */
550 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
552 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
554 /* Close the filehandle. Could be from -P preprocessor,
555 * STDIN, or a regular file. If we were reading code from
556 * STDIN (because the commandline held no -e or filename)
557 * then we don't close it, we reset it so the code can
558 * read from STDIN too.
561 if (PL_preprocess && !PL_in_eval)
562 (void)PerlProc_pclose(PL_rsfp);
563 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
564 PerlIO_clearerr(PL_rsfp);
566 (void)PerlIO_close(PL_rsfp);
571 /* not at end of file, so we only read another line */
572 /* make corresponding updates to old pointers, for yyerror() */
573 oldprevlen = PL_oldbufptr - PL_bufend;
574 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
576 oldunilen = PL_last_uni - PL_bufend;
578 oldloplen = PL_last_lop - PL_bufend;
579 PL_linestart = PL_bufptr = s + prevlen;
580 PL_bufend = s + SvCUR(PL_linestr);
582 PL_oldbufptr = s + oldprevlen;
583 PL_oldoldbufptr = s + oldoldprevlen;
585 PL_last_uni = s + oldunilen;
587 PL_last_lop = s + oldloplen;
590 /* debugger active and we're not compiling the debugger code,
591 * so store the line into the debugger's array of lines
593 if (PERLDB_LINE && PL_curstash != PL_debstash) {
594 SV *sv = NEWSV(85,0);
596 sv_upgrade(sv, SVt_PVMG);
597 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
598 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
605 * Check the unary operators to ensure there's no ambiguity in how they're
606 * used. An ambiguous piece of code would be:
608 * This doesn't mean rand() + 5. Because rand() is a unary operator,
609 * the +5 is its argument.
619 if (PL_oldoldbufptr != PL_last_uni)
621 while (isSPACE(*PL_last_uni))
623 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
624 if ((t = strchr(s, '(')) && t < PL_bufptr)
626 if (ckWARN_d(WARN_AMBIGUOUS)){
629 Perl_warner(aTHX_ WARN_AMBIGUOUS,
630 "Warning: Use of \"%s\" without parens is ambiguous",
636 /* workaround to replace the UNI() macro with a function. Only the
637 * hints/uts.sh file mentions this. Other comments elsewhere in the
638 * source indicate Microport Unix might need it too.
644 #define UNI(f) return uni(f,s)
647 S_uni(pTHX_ I32 f, char *s)
652 PL_last_uni = PL_oldbufptr;
663 #endif /* CRIPPLED_CC */
666 * LOP : macro to build a list operator. Its behaviour has been replaced
667 * with a subroutine, S_lop() for which LOP is just another name.
670 #define LOP(f,x) return lop(f,x,s)
674 * Build a list operator (or something that might be one). The rules:
675 * - if we have a next token, then it's a list operator [why?]
676 * - if the next thing is an opening paren, then it's a function
677 * - else it's a list operator
681 S_lop(pTHX_ I32 f, int x, char *s)
688 PL_last_lop = PL_oldbufptr;
703 * When the lexer realizes it knows the next token (for instance,
704 * it is reordering tokens for the parser) then it can call S_force_next
705 * to know what token to return the next time the lexer is called. Caller
706 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
707 * handles the token correctly.
711 S_force_next(pTHX_ I32 type)
713 PL_nexttype[PL_nexttoke] = type;
715 if (PL_lex_state != LEX_KNOWNEXT) {
716 PL_lex_defer = PL_lex_state;
717 PL_lex_expect = PL_expect;
718 PL_lex_state = LEX_KNOWNEXT;
724 * When the lexer knows the next thing is a word (for instance, it has
725 * just seen -> and it knows that the next char is a word char, then
726 * it calls S_force_word to stick the next word into the PL_next lookahead.
729 * char *start : buffer position (must be within PL_linestr)
730 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
731 * int check_keyword : if true, Perl checks to make sure the word isn't
732 * a keyword (do this if the word is a label, e.g. goto FOO)
733 * int allow_pack : if true, : characters will also be allowed (require,
735 * int allow_initial_tick : used by the "sub" lexer only.
739 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
744 start = skipspace(start);
746 if (isIDFIRST_lazy_if(s,UTF) ||
747 (allow_pack && *s == ':') ||
748 (allow_initial_tick && *s == '\'') )
750 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
751 if (check_keyword && keyword(PL_tokenbuf, len))
753 if (token == METHOD) {
758 PL_expect = XOPERATOR;
761 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
762 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
770 * Called when the lexer wants $foo *foo &foo etc, but the program
771 * text only contains the "foo" portion. The first argument is a pointer
772 * to the "foo", and the second argument is the type symbol to prefix.
773 * Forces the next token to be a "WORD".
774 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
778 S_force_ident(pTHX_ register char *s, int kind)
781 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
782 PL_nextval[PL_nexttoke].opval = o;
785 dTHR; /* just for in_eval */
786 o->op_private = OPpCONST_ENTERED;
787 /* XXX see note in pp_entereval() for why we forgo typo
788 warnings if the symbol must be introduced in an eval.
790 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
791 kind == '$' ? SVt_PV :
792 kind == '@' ? SVt_PVAV :
793 kind == '%' ? SVt_PVHV :
801 Perl_str_to_version(pTHX_ SV *sv)
806 char *start = SvPVx(sv,len);
807 bool utf = SvUTF8(sv) ? TRUE : FALSE;
808 char *end = start + len;
809 while (start < end) {
813 n = utf8_to_uv((U8*)start, len, &skip, 0);
818 retval += ((NV)n)/nshift;
827 * Forces the next token to be a version number.
831 S_force_version(pTHX_ char *s)
833 OP *version = Nullop;
842 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
843 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
845 s = scan_num(s, &yylval);
846 version = yylval.opval;
847 ver = cSVOPx(version)->op_sv;
848 if (SvPOK(ver) && !SvNIOK(ver)) {
849 (void)SvUPGRADE(ver, SVt_PVNV);
850 SvNVX(ver) = str_to_version(ver);
851 SvNOK_on(ver); /* hint that it is a version */
856 /* NOTE: The parser sees the package name and the VERSION swapped */
857 PL_nextval[PL_nexttoke].opval = version;
865 * Tokenize a quoted string passed in as an SV. It finds the next
866 * chunk, up to end of string or a backslash. It may make a new
867 * SV containing that chunk (if HINT_NEW_STRING is on). It also
872 S_tokeq(pTHX_ SV *sv)
883 s = SvPV_force(sv, len);
884 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
887 while (s < send && *s != '\\')
892 if ( PL_hints & HINT_NEW_STRING )
893 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
896 if (s + 1 < send && (s[1] == '\\'))
897 s++; /* all that, just for this */
902 SvCUR_set(sv, d - SvPVX(sv));
904 if ( PL_hints & HINT_NEW_STRING )
905 return new_constant(NULL, 0, "q", sv, pv, "q");
910 * Now come three functions related to double-quote context,
911 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
912 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
913 * interact with PL_lex_state, and create fake ( ... ) argument lists
914 * to handle functions and concatenation.
915 * They assume that whoever calls them will be setting up a fake
916 * join call, because each subthing puts a ',' after it. This lets
919 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
921 * (I'm not sure whether the spurious commas at the end of lcfirst's
922 * arguments and join's arguments are created or not).
927 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
929 * Pattern matching will set PL_lex_op to the pattern-matching op to
930 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
932 * OP_CONST and OP_READLINE are easy--just make the new op and return.
934 * Everything else becomes a FUNC.
936 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
937 * had an OP_CONST or OP_READLINE). This just sets us up for a
938 * call to S_sublex_push().
944 register I32 op_type = yylval.ival;
946 if (op_type == OP_NULL) {
947 yylval.opval = PL_lex_op;
951 if (op_type == OP_CONST || op_type == OP_READLINE) {
952 SV *sv = tokeq(PL_lex_stuff);
954 if (SvTYPE(sv) == SVt_PVIV) {
955 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
961 nsv = newSVpvn(p, len);
967 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
968 PL_lex_stuff = Nullsv;
972 PL_sublex_info.super_state = PL_lex_state;
973 PL_sublex_info.sub_inwhat = op_type;
974 PL_sublex_info.sub_op = PL_lex_op;
975 PL_lex_state = LEX_INTERPPUSH;
979 yylval.opval = PL_lex_op;
989 * Create a new scope to save the lexing state. The scope will be
990 * ended in S_sublex_done. Returns a '(', starting the function arguments
991 * to the uc, lc, etc. found before.
992 * Sets PL_lex_state to LEX_INTERPCONCAT.
1001 PL_lex_state = PL_sublex_info.super_state;
1002 SAVEI32(PL_lex_dojoin);
1003 SAVEI32(PL_lex_brackets);
1004 SAVEI32(PL_lex_casemods);
1005 SAVEI32(PL_lex_starts);
1006 SAVEI32(PL_lex_state);
1007 SAVEVPTR(PL_lex_inpat);
1008 SAVEI32(PL_lex_inwhat);
1009 SAVECOPLINE(PL_curcop);
1010 SAVEPPTR(PL_bufptr);
1011 SAVEPPTR(PL_oldbufptr);
1012 SAVEPPTR(PL_oldoldbufptr);
1013 SAVEPPTR(PL_linestart);
1014 SAVESPTR(PL_linestr);
1015 SAVEPPTR(PL_lex_brackstack);
1016 SAVEPPTR(PL_lex_casestack);
1018 PL_linestr = PL_lex_stuff;
1019 PL_lex_stuff = Nullsv;
1021 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1022 = SvPVX(PL_linestr);
1023 PL_bufend += SvCUR(PL_linestr);
1024 SAVEFREESV(PL_linestr);
1026 PL_lex_dojoin = FALSE;
1027 PL_lex_brackets = 0;
1028 New(899, PL_lex_brackstack, 120, char);
1029 New(899, PL_lex_casestack, 12, char);
1030 SAVEFREEPV(PL_lex_brackstack);
1031 SAVEFREEPV(PL_lex_casestack);
1032 PL_lex_casemods = 0;
1033 *PL_lex_casestack = '\0';
1035 PL_lex_state = LEX_INTERPCONCAT;
1036 CopLINE_set(PL_curcop, PL_multi_start);
1038 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1039 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1040 PL_lex_inpat = PL_sublex_info.sub_op;
1042 PL_lex_inpat = Nullop;
1049 * Restores lexer state after a S_sublex_push.
1055 if (!PL_lex_starts++) {
1056 PL_expect = XOPERATOR;
1057 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1061 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1062 PL_lex_state = LEX_INTERPCASEMOD;
1066 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1067 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1068 PL_linestr = PL_lex_repl;
1070 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1071 PL_bufend += SvCUR(PL_linestr);
1072 SAVEFREESV(PL_linestr);
1073 PL_lex_dojoin = FALSE;
1074 PL_lex_brackets = 0;
1075 PL_lex_casemods = 0;
1076 *PL_lex_casestack = '\0';
1078 if (SvEVALED(PL_lex_repl)) {
1079 PL_lex_state = LEX_INTERPNORMAL;
1081 /* we don't clear PL_lex_repl here, so that we can check later
1082 whether this is an evalled subst; that means we rely on the
1083 logic to ensure sublex_done() is called again only via the
1084 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1087 PL_lex_state = LEX_INTERPCONCAT;
1088 PL_lex_repl = Nullsv;
1094 PL_bufend = SvPVX(PL_linestr);
1095 PL_bufend += SvCUR(PL_linestr);
1096 PL_expect = XOPERATOR;
1097 PL_sublex_info.sub_inwhat = 0;
1105 Extracts a pattern, double-quoted string, or transliteration. This
1108 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1109 processing a pattern (PL_lex_inpat is true), a transliteration
1110 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1112 Returns a pointer to the character scanned up to. Iff this is
1113 advanced from the start pointer supplied (ie if anything was
1114 successfully parsed), will leave an OP for the substring scanned
1115 in yylval. Caller must intuit reason for not parsing further
1116 by looking at the next characters herself.
1120 double-quoted style: \r and \n
1121 regexp special ones: \D \s
1123 backrefs: \1 (deprecated in substitution replacements)
1124 case and quoting: \U \Q \E
1125 stops on @ and $, but not for $ as tail anchor
1127 In transliterations:
1128 characters are VERY literal, except for - not at the start or end
1129 of the string, which indicates a range. scan_const expands the
1130 range to the full set of intermediate characters.
1132 In double-quoted strings:
1134 double-quoted style: \r and \n
1136 backrefs: \1 (deprecated)
1137 case and quoting: \U \Q \E
1140 scan_const does *not* construct ops to handle interpolated strings.
1141 It stops processing as soon as it finds an embedded $ or @ variable
1142 and leaves it to the caller to work out what's going on.
1144 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1146 $ in pattern could be $foo or could be tail anchor. Assumption:
1147 it's a tail anchor if $ is the last thing in the string, or if it's
1148 followed by one of ")| \n\t"
1150 \1 (backreferences) are turned into $1
1152 The structure of the code is
1153 while (there's a character to process) {
1154 handle transliteration ranges
1155 skip regexp comments
1156 skip # initiated comments in //x patterns
1157 check for embedded @foo
1158 check for embedded scalars
1160 leave intact backslashes from leave (below)
1161 deprecate \1 in strings and sub replacements
1162 handle string-changing backslashes \l \U \Q \E, etc.
1163 switch (what was escaped) {
1164 handle - in a transliteration (becomes a literal -)
1165 handle \132 octal characters
1166 handle 0x15 hex characters
1167 handle \cV (control V)
1168 handle printf backslashes (\f, \r, \n, etc)
1170 } (end if backslash)
1171 } (end while character to read)
1176 S_scan_const(pTHX_ char *start)
1178 register char *send = PL_bufend; /* end of the constant */
1179 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1180 register char *s = start; /* start of the constant */
1181 register char *d = SvPVX(sv); /* destination for copies */
1182 bool dorange = FALSE; /* are we in a translit range? */
1183 bool didrange = FALSE; /* did we just finish a range? */
1184 bool has_utf8 = FALSE; /* embedded \x{} */
1187 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1188 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1190 I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1191 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1192 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1194 const char *leaveit = /* set of acceptably-backslashed characters */
1196 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1199 while (s < send || dorange) {
1200 /* get transliterations out of the way (they're most literal) */
1201 if (PL_lex_inwhat == OP_TRANS) {
1202 /* expand a range A-Z to the full set of characters. AIE! */
1204 I32 i; /* current expanded character */
1205 I32 min; /* first character in range */
1206 I32 max; /* last character in range */
1208 i = d - SvPVX(sv); /* remember current offset */
1209 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1210 d = SvPVX(sv) + i; /* refresh d after realloc */
1211 d -= 2; /* eat the first char and the - */
1213 min = (U8)*d; /* first char in range */
1214 max = (U8)d[1]; /* last char in range */
1218 "Invalid [] range \"%c-%c\" in transliteration operator",
1219 (char)min, (char)max);
1223 if ((isLOWER(min) && isLOWER(max)) ||
1224 (isUPPER(min) && isUPPER(max))) {
1226 for (i = min; i <= max; i++)
1230 for (i = min; i <= max; i++)
1237 for (i = min; i <= max; i++)
1240 /* mark the range as done, and continue */
1246 /* range begins (ignore - as first or last char) */
1247 else if (*s == '-' && s+1 < send && s != start) {
1249 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1252 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1264 /* if we get here, we're not doing a transliteration */
1266 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1267 except for the last char, which will be done separately. */
1268 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1270 while (s < send && *s != ')')
1273 else if (s[2] == '{' /* This should match regcomp.c */
1274 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1277 char *regparse = s + (s[2] == '{' ? 3 : 4);
1280 while (count && (c = *regparse)) {
1281 if (c == '\\' && regparse[1])
1289 if (*regparse != ')') {
1290 regparse--; /* Leave one char for continuation. */
1291 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1293 while (s < regparse)
1298 /* likewise skip #-initiated comments in //x patterns */
1299 else if (*s == '#' && PL_lex_inpat &&
1300 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1301 while (s+1 < send && *s != '\n')
1305 /* check for embedded arrays
1306 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1308 else if (*s == '@' && s[1]
1309 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1312 /* check for embedded scalars. only stop if we're sure it's a
1315 else if (*s == '$') {
1316 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1318 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1319 break; /* in regexp, $ might be tail anchor */
1322 /* (now in tr/// code again) */
1324 if (*s & 0x80 && this_utf8) {
1328 uv = utf8_to_uv((U8*)s, send - s, &len, UTF8_CHECK_ONLY);
1329 if (len == (STRLEN)-1) {
1330 /* Illegal UTF8 (a high-bit byte), make it valid. */
1331 char *old_pvx = SvPVX(sv);
1332 /* need space for one extra char (NOTE: SvCUR() not set here) */
1333 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1334 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1345 if (*s == '\\' && s+1 < send) {
1348 /* some backslashes we leave behind */
1349 if (*leaveit && *s && strchr(leaveit, *s)) {
1355 /* deprecate \1 in strings and substitution replacements */
1356 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1357 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1359 dTHR; /* only for ckWARN */
1360 if (ckWARN(WARN_SYNTAX))
1361 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1366 /* string-change backslash escapes */
1367 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1372 /* if we get here, it's either a quoted -, or a digit */
1375 /* quoted - in transliterations */
1377 if (PL_lex_inwhat == OP_TRANS) {
1385 if (ckWARN(WARN_MISC) && isALNUM(*s))
1386 Perl_warner(aTHX_ WARN_MISC,
1387 "Unrecognized escape \\%c passed through",
1389 /* default action is to copy the quoted character */
1394 /* \132 indicates an octal constant */
1395 case '0': case '1': case '2': case '3':
1396 case '4': case '5': case '6': case '7':
1398 STRLEN len = 0; /* disallow underscores */
1399 uv = (UV)scan_oct(s, 3, &len);
1402 goto NUM_ESCAPE_INSERT;
1404 /* \x24 indicates a hex constant */
1408 char* e = strchr(s, '}');
1410 yyerror("Missing right brace on \\x{}");
1414 STRLEN len = 1; /* allow underscores */
1415 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1422 STRLEN len = 0; /* disallow underscores */
1423 uv = (UV)scan_hex(s, 2, &len);
1429 /* Insert oct or hex escaped character.
1430 * There will always enough room in sv since such escapes will
1431 * be longer than any utf8 sequence they can end up as
1433 if (uv > 127 || has_utf8) {
1434 if (!this_utf8 && !has_utf8 && uv > 255) {
1435 /* might need to recode whatever we have accumulated so far
1436 * if it contains any hibit chars
1440 for (c = SvPVX(sv); c < d; c++) {
1445 char *old_pvx = SvPVX(sv);
1447 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1456 uv_to_utf8((U8*)dst, (U8)*src--);
1466 if (has_utf8 || uv > 255) {
1467 d = (char*)uv_to_utf8((U8*)d, uv);
1479 /* \N{latin small letter a} is a named character */
1483 char* e = strchr(s, '}');
1489 yyerror("Missing right brace on \\N{}");
1493 res = newSVpvn(s + 1, e - s - 1);
1494 res = new_constant( Nullch, 0, "charnames",
1495 res, Nullsv, "\\N{...}" );
1496 str = SvPV(res,len);
1497 if (!has_utf8 && SvUTF8(res)) {
1498 char *ostart = SvPVX(sv);
1499 SvCUR_set(sv, d - ostart);
1502 sv_utf8_upgrade(sv);
1503 /* this just broke our allocation above... */
1504 SvGROW(sv, send - start);
1505 d = SvPVX(sv) + SvCUR(sv);
1508 if (len > e - s + 4) {
1509 char *odest = SvPVX(sv);
1511 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1512 d = SvPVX(sv) + (d - odest);
1514 Copy(str, d, len, char);
1521 yyerror("Missing braces on \\N{}");
1524 /* \c is a control character */
1541 /* printf-style backslashes, formfeeds, newlines, etc */
1559 *d++ = '\047'; /* CP 1047 */
1562 *d++ = '\057'; /* CP 1047 */
1576 } /* end if (backslash) */
1579 } /* while loop to process each character */
1581 /* terminate the string and set up the sv */
1583 SvCUR_set(sv, d - SvPVX(sv));
1588 /* shrink the sv if we allocated more than we used */
1589 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1590 SvLEN_set(sv, SvCUR(sv) + 1);
1591 Renew(SvPVX(sv), SvLEN(sv), char);
1594 /* return the substring (via yylval) only if we parsed anything */
1595 if (s > PL_bufptr) {
1596 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1597 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1599 ( PL_lex_inwhat == OP_TRANS
1601 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1604 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1611 * Returns TRUE if there's more to the expression (e.g., a subscript),
1614 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1616 * ->[ and ->{ return TRUE
1617 * { and [ outside a pattern are always subscripts, so return TRUE
1618 * if we're outside a pattern and it's not { or [, then return FALSE
1619 * if we're in a pattern and the first char is a {
1620 * {4,5} (any digits around the comma) returns FALSE
1621 * if we're in a pattern and the first char is a [
1623 * [SOMETHING] has a funky algorithm to decide whether it's a
1624 * character class or not. It has to deal with things like
1625 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1626 * anything else returns TRUE
1629 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1632 S_intuit_more(pTHX_ register char *s)
1634 if (PL_lex_brackets)
1636 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1638 if (*s != '{' && *s != '[')
1643 /* In a pattern, so maybe we have {n,m}. */
1660 /* On the other hand, maybe we have a character class */
1663 if (*s == ']' || *s == '^')
1666 /* this is terrifying, and it works */
1667 int weight = 2; /* let's weigh the evidence */
1669 unsigned char un_char = 255, last_un_char;
1670 char *send = strchr(s,']');
1671 char tmpbuf[sizeof PL_tokenbuf * 4];
1673 if (!send) /* has to be an expression */
1676 Zero(seen,256,char);
1679 else if (isDIGIT(*s)) {
1681 if (isDIGIT(s[1]) && s[2] == ']')
1687 for (; s < send; s++) {
1688 last_un_char = un_char;
1689 un_char = (unsigned char)*s;
1694 weight -= seen[un_char] * 10;
1695 if (isALNUM_lazy_if(s+1,UTF)) {
1696 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1697 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1702 else if (*s == '$' && s[1] &&
1703 strchr("[#!%*<>()-=",s[1])) {
1704 if (/*{*/ strchr("])} =",s[2]))
1713 if (strchr("wds]",s[1]))
1715 else if (seen['\''] || seen['"'])
1717 else if (strchr("rnftbxcav",s[1]))
1719 else if (isDIGIT(s[1])) {
1721 while (s[1] && isDIGIT(s[1]))
1731 if (strchr("aA01! ",last_un_char))
1733 if (strchr("zZ79~",s[1]))
1735 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1736 weight -= 5; /* cope with negative subscript */
1739 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1740 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1745 if (keyword(tmpbuf, d - tmpbuf))
1748 if (un_char == last_un_char + 1)
1750 weight -= seen[un_char];
1755 if (weight >= 0) /* probably a character class */
1765 * Does all the checking to disambiguate
1767 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1768 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1770 * First argument is the stuff after the first token, e.g. "bar".
1772 * Not a method if bar is a filehandle.
1773 * Not a method if foo is a subroutine prototyped to take a filehandle.
1774 * Not a method if it's really "Foo $bar"
1775 * Method if it's "foo $bar"
1776 * Not a method if it's really "print foo $bar"
1777 * Method if it's really "foo package::" (interpreted as package->foo)
1778 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1779 * Not a method if bar is a filehandle or package, but is quoted with
1784 S_intuit_method(pTHX_ char *start, GV *gv)
1786 char *s = start + (*start == '$');
1787 char tmpbuf[sizeof PL_tokenbuf];
1795 if ((cv = GvCVu(gv))) {
1796 char *proto = SvPVX(cv);
1806 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1807 /* start is the beginning of the possible filehandle/object,
1808 * and s is the end of it
1809 * tmpbuf is a copy of it
1812 if (*start == '$') {
1813 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1818 return *s == '(' ? FUNCMETH : METHOD;
1820 if (!keyword(tmpbuf, len)) {
1821 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1826 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1827 if (indirgv && GvCVu(indirgv))
1829 /* filehandle or package name makes it a method */
1830 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1832 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1833 return 0; /* no assumptions -- "=>" quotes bearword */
1835 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1836 newSVpvn(tmpbuf,len));
1837 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1841 return *s == '(' ? FUNCMETH : METHOD;
1849 * Return a string of Perl code to load the debugger. If PERL5DB
1850 * is set, it will return the contents of that, otherwise a
1851 * compile-time require of perl5db.pl.
1858 char *pdb = PerlEnv_getenv("PERL5DB");
1862 SETERRNO(0,SS$_NORMAL);
1863 return "BEGIN { require 'perl5db.pl' }";
1869 /* Encoded script support. filter_add() effectively inserts a
1870 * 'pre-processing' function into the current source input stream.
1871 * Note that the filter function only applies to the current source file
1872 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1874 * The datasv parameter (which may be NULL) can be used to pass
1875 * private data to this instance of the filter. The filter function
1876 * can recover the SV using the FILTER_DATA macro and use it to
1877 * store private buffers and state information.
1879 * The supplied datasv parameter is upgraded to a PVIO type
1880 * and the IoDIRP/IoANY field is used to store the function pointer,
1881 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1882 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1883 * private use must be set using malloc'd pointers.
1887 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1892 if (!PL_rsfp_filters)
1893 PL_rsfp_filters = newAV();
1895 datasv = NEWSV(255,0);
1896 if (!SvUPGRADE(datasv, SVt_PVIO))
1897 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1898 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1899 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1900 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1901 funcp, SvPV_nolen(datasv)));
1902 av_unshift(PL_rsfp_filters, 1);
1903 av_store(PL_rsfp_filters, 0, datasv) ;
1908 /* Delete most recently added instance of this filter function. */
1910 Perl_filter_del(pTHX_ filter_t funcp)
1913 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1914 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1916 /* if filter is on top of stack (usual case) just pop it off */
1917 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1918 if (IoANY(datasv) == (void *)funcp) {
1919 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1920 IoANY(datasv) = (void *)NULL;
1921 sv_free(av_pop(PL_rsfp_filters));
1925 /* we need to search for the correct entry and clear it */
1926 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1930 /* Invoke the n'th filter function for the current rsfp. */
1932 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1935 /* 0 = read one text line */
1940 if (!PL_rsfp_filters)
1942 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1943 /* Provide a default input filter to make life easy. */
1944 /* Note that we append to the line. This is handy. */
1945 DEBUG_P(PerlIO_printf(Perl_debug_log,
1946 "filter_read %d: from rsfp\n", idx));
1950 int old_len = SvCUR(buf_sv) ;
1952 /* ensure buf_sv is large enough */
1953 SvGROW(buf_sv, old_len + maxlen) ;
1954 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1955 if (PerlIO_error(PL_rsfp))
1956 return -1; /* error */
1958 return 0 ; /* end of file */
1960 SvCUR_set(buf_sv, old_len + len) ;
1963 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1964 if (PerlIO_error(PL_rsfp))
1965 return -1; /* error */
1967 return 0 ; /* end of file */
1970 return SvCUR(buf_sv);
1972 /* Skip this filter slot if filter has been deleted */
1973 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1974 DEBUG_P(PerlIO_printf(Perl_debug_log,
1975 "filter_read %d: skipped (filter deleted)\n",
1977 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1979 /* Get function pointer hidden within datasv */
1980 funcp = (filter_t)IoANY(datasv);
1981 DEBUG_P(PerlIO_printf(Perl_debug_log,
1982 "filter_read %d: via function %p (%s)\n",
1983 idx, funcp, SvPV_nolen(datasv)));
1984 /* Call function. The function is expected to */
1985 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1986 /* Return: <0:error, =0:eof, >0:not eof */
1987 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1991 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1993 #ifdef PERL_CR_FILTER
1994 if (!PL_rsfp_filters) {
1995 filter_add(S_cr_textfilter,NULL);
1998 if (PL_rsfp_filters) {
2001 SvCUR_set(sv, 0); /* start with empty line */
2002 if (FILTER_READ(0, sv, 0) > 0)
2003 return ( SvPVX(sv) ) ;
2008 return (sv_gets(sv, fp, append));
2012 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2016 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2020 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2021 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2023 return GvHV(gv); /* Foo:: */
2026 /* use constant CLASS => 'MyClass' */
2027 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2029 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2030 pkgname = SvPV_nolen(sv);
2034 return gv_stashpv(pkgname, FALSE);
2038 static char* exp_name[] =
2039 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2040 "ATTRTERM", "TERMBLOCK"
2047 Works out what to call the token just pulled out of the input
2048 stream. The yacc parser takes care of taking the ops we return and
2049 stitching them into a tree.
2055 if read an identifier
2056 if we're in a my declaration
2057 croak if they tried to say my($foo::bar)
2058 build the ops for a my() declaration
2059 if it's an access to a my() variable
2060 are we in a sort block?
2061 croak if my($a); $a <=> $b
2062 build ops for access to a my() variable
2063 if in a dq string, and they've said @foo and we can't find @foo
2065 build ops for a bareword
2066 if we already built the token before, use it.
2069 #ifdef USE_PURE_BISON
2071 #pragma segment Perl_yylex_r
2074 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2079 yylval_pointer[yyactlevel] = lvalp;
2080 yychar_pointer[yyactlevel] = lcharp;
2082 if (yyactlevel >= YYMAXLEVEL)
2083 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2085 r = Perl_yylex(aTHX);
2094 #pragma segment Perl_yylex
2098 #ifdef USE_PURE_BISON
2099 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2112 /* check if there's an identifier for us to look at */
2113 if (PL_pending_ident) {
2114 /* pit holds the identifier we read and pending_ident is reset */
2115 char pit = PL_pending_ident;
2116 PL_pending_ident = 0;
2118 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2119 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2121 /* if we're in a my(), we can't allow dynamics here.
2122 $foo'bar has already been turned into $foo::bar, so
2123 just check for colons.
2125 if it's a legal name, the OP is a PADANY.
2128 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2129 if (strchr(PL_tokenbuf,':'))
2130 yyerror(Perl_form(aTHX_ "No package name allowed for "
2131 "variable %s in \"our\"",
2133 tmp = pad_allocmy(PL_tokenbuf);
2136 if (strchr(PL_tokenbuf,':'))
2137 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2139 yylval.opval = newOP(OP_PADANY, 0);
2140 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2146 build the ops for accesses to a my() variable.
2148 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2149 then used in a comparison. This catches most, but not
2150 all cases. For instance, it catches
2151 sort { my($a); $a <=> $b }
2153 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2154 (although why you'd do that is anyone's guess).
2157 if (!strchr(PL_tokenbuf,':')) {
2159 /* Check for single character per-thread SVs */
2160 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2161 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2162 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2164 yylval.opval = newOP(OP_THREADSV, 0);
2165 yylval.opval->op_targ = tmp;
2168 #endif /* USE_THREADS */
2169 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2170 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2171 /* might be an "our" variable" */
2172 if (SvFLAGS(namesv) & SVpad_OUR) {
2173 /* build ops for a bareword */
2174 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2175 sv_catpvn(sym, "::", 2);
2176 sv_catpv(sym, PL_tokenbuf+1);
2177 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2178 yylval.opval->op_private = OPpCONST_ENTERED;
2179 gv_fetchpv(SvPVX(sym),
2181 ? (GV_ADDMULTI | GV_ADDINEVAL)
2184 ((PL_tokenbuf[0] == '$') ? SVt_PV
2185 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2190 /* if it's a sort block and they're naming $a or $b */
2191 if (PL_last_lop_op == OP_SORT &&
2192 PL_tokenbuf[0] == '$' &&
2193 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2196 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2197 d < PL_bufend && *d != '\n';
2200 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2201 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2207 yylval.opval = newOP(OP_PADANY, 0);
2208 yylval.opval->op_targ = tmp;
2214 Whine if they've said @foo in a doublequoted string,
2215 and @foo isn't a variable we can find in the symbol
2218 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2219 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2220 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2221 && ckWARN(WARN_AMBIGUOUS))
2223 /* Downgraded from fatal to warning 20000522 mjd */
2224 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2225 "Possible unintended interpolation of %s in string",
2230 /* build ops for a bareword */
2231 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2232 yylval.opval->op_private = OPpCONST_ENTERED;
2233 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2234 ((PL_tokenbuf[0] == '$') ? SVt_PV
2235 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2240 /* no identifier pending identification */
2242 switch (PL_lex_state) {
2244 case LEX_NORMAL: /* Some compilers will produce faster */
2245 case LEX_INTERPNORMAL: /* code if we comment these out. */
2249 /* when we've already built the next token, just pull it out of the queue */
2252 yylval = PL_nextval[PL_nexttoke];
2254 PL_lex_state = PL_lex_defer;
2255 PL_expect = PL_lex_expect;
2256 PL_lex_defer = LEX_NORMAL;
2258 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2259 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2260 (IV)PL_nexttype[PL_nexttoke]); })
2262 return(PL_nexttype[PL_nexttoke]);
2264 /* interpolated case modifiers like \L \U, including \Q and \E.
2265 when we get here, PL_bufptr is at the \
2267 case LEX_INTERPCASEMOD:
2269 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2270 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2272 /* handle \E or end of string */
2273 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2277 if (PL_lex_casemods) {
2278 oldmod = PL_lex_casestack[--PL_lex_casemods];
2279 PL_lex_casestack[PL_lex_casemods] = '\0';
2281 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2283 PL_lex_state = LEX_INTERPCONCAT;
2287 if (PL_bufptr != PL_bufend)
2289 PL_lex_state = LEX_INTERPCONCAT;
2293 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2294 "### Saw case modifier at '%s'\n", PL_bufptr); })
2296 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2297 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2298 if (strchr("LU", *s) &&
2299 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2301 PL_lex_casestack[--PL_lex_casemods] = '\0';
2304 if (PL_lex_casemods > 10) {
2305 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2306 if (newlb != PL_lex_casestack) {
2308 PL_lex_casestack = newlb;
2311 PL_lex_casestack[PL_lex_casemods++] = *s;
2312 PL_lex_casestack[PL_lex_casemods] = '\0';
2313 PL_lex_state = LEX_INTERPCONCAT;
2314 PL_nextval[PL_nexttoke].ival = 0;
2317 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2319 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2321 PL_nextval[PL_nexttoke].ival = OP_LC;
2323 PL_nextval[PL_nexttoke].ival = OP_UC;
2325 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2327 Perl_croak(aTHX_ "panic: yylex");
2330 if (PL_lex_starts) {
2339 case LEX_INTERPPUSH:
2340 return sublex_push();
2342 case LEX_INTERPSTART:
2343 if (PL_bufptr == PL_bufend)
2344 return sublex_done();
2345 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2346 "### Interpolated variable at '%s'\n", PL_bufptr); })
2348 PL_lex_dojoin = (*PL_bufptr == '@');
2349 PL_lex_state = LEX_INTERPNORMAL;
2350 if (PL_lex_dojoin) {
2351 PL_nextval[PL_nexttoke].ival = 0;
2354 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2355 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2356 force_next(PRIVATEREF);
2358 force_ident("\"", '$');
2359 #endif /* USE_THREADS */
2360 PL_nextval[PL_nexttoke].ival = 0;
2362 PL_nextval[PL_nexttoke].ival = 0;
2364 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2367 if (PL_lex_starts++) {
2373 case LEX_INTERPENDMAYBE:
2374 if (intuit_more(PL_bufptr)) {
2375 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2381 if (PL_lex_dojoin) {
2382 PL_lex_dojoin = FALSE;
2383 PL_lex_state = LEX_INTERPCONCAT;
2386 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2387 && SvEVALED(PL_lex_repl))
2389 if (PL_bufptr != PL_bufend)
2390 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2391 PL_lex_repl = Nullsv;
2394 case LEX_INTERPCONCAT:
2396 if (PL_lex_brackets)
2397 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2399 if (PL_bufptr == PL_bufend)
2400 return sublex_done();
2402 if (SvIVX(PL_linestr) == '\'') {
2403 SV *sv = newSVsv(PL_linestr);
2406 else if ( PL_hints & HINT_NEW_RE )
2407 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2408 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2412 s = scan_const(PL_bufptr);
2414 PL_lex_state = LEX_INTERPCASEMOD;
2416 PL_lex_state = LEX_INTERPSTART;
2419 if (s != PL_bufptr) {
2420 PL_nextval[PL_nexttoke] = yylval;
2423 if (PL_lex_starts++)
2433 PL_lex_state = LEX_NORMAL;
2434 s = scan_formline(PL_bufptr);
2435 if (!PL_lex_formbrack)
2441 PL_oldoldbufptr = PL_oldbufptr;
2444 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2445 exp_name[PL_expect], s);
2451 if (isIDFIRST_lazy_if(s,UTF))
2453 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2456 goto fake_eof; /* emulate EOF on ^D or ^Z */
2461 if (PL_lex_brackets)
2462 yyerror("Missing right curly or square bracket");
2463 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2464 "### Tokener got EOF\n");
2468 if (s++ < PL_bufend)
2469 goto retry; /* ignore stray nulls */
2472 if (!PL_in_eval && !PL_preambled) {
2473 PL_preambled = TRUE;
2474 sv_setpv(PL_linestr,incl_perldb());
2475 if (SvCUR(PL_linestr))
2476 sv_catpv(PL_linestr,";");
2478 while(AvFILLp(PL_preambleav) >= 0) {
2479 SV *tmpsv = av_shift(PL_preambleav);
2480 sv_catsv(PL_linestr, tmpsv);
2481 sv_catpv(PL_linestr, ";");
2484 sv_free((SV*)PL_preambleav);
2485 PL_preambleav = NULL;
2487 if (PL_minus_n || PL_minus_p) {
2488 sv_catpv(PL_linestr, "LINE: while (<>) {");
2490 sv_catpv(PL_linestr,"chomp;");
2492 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2494 GvIMPORTED_AV_on(gv);
2496 if (strchr("/'\"", *PL_splitstr)
2497 && strchr(PL_splitstr + 1, *PL_splitstr))
2498 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2501 s = "'~#\200\1'"; /* surely one char is unused...*/
2502 while (s[1] && strchr(PL_splitstr, *s)) s++;
2504 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2505 "q" + (delim == '\''), delim);
2506 for (s = PL_splitstr; *s; s++) {
2508 sv_catpvn(PL_linestr, "\\", 1);
2509 sv_catpvn(PL_linestr, s, 1);
2511 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2515 sv_catpv(PL_linestr,"@F=split(' ');");
2518 sv_catpv(PL_linestr, "\n");
2519 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2520 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2521 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2522 SV *sv = NEWSV(85,0);
2524 sv_upgrade(sv, SVt_PVMG);
2525 sv_setsv(sv,PL_linestr);
2526 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2531 bool bof = PL_rsfp ? TRUE : FALSE;
2533 #ifdef PERLIO_IS_STDIO
2534 # ifdef __GNU_LIBRARY__
2535 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2536 # define FTELL_FOR_PIPE_IS_BROKEN
2540 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2541 # define FTELL_FOR_PIPE_IS_BROKEN
2546 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2547 /* This loses the possibility to detect the bof
2548 * situation on perl -P when the libc5 is being used.
2549 * Workaround? Maybe attach some extra state to PL_rsfp?
2552 bof = PerlIO_tell(PL_rsfp) == 0;
2554 bof = PerlIO_tell(PL_rsfp) == 0;
2557 s = filter_gets(PL_linestr, PL_rsfp, 0);
2561 if (PL_preprocess && !PL_in_eval)
2562 (void)PerlProc_pclose(PL_rsfp);
2563 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2564 PerlIO_clearerr(PL_rsfp);
2566 (void)PerlIO_close(PL_rsfp);
2568 PL_doextract = FALSE;
2570 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2571 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2572 sv_catpv(PL_linestr,";}");
2573 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2574 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2575 PL_minus_n = PL_minus_p = 0;
2578 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2579 sv_setpv(PL_linestr,"");
2580 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2582 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2583 s = swallow_bom((U8*)s);
2586 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2587 PL_doextract = FALSE;
2589 /* Incest with pod. */
2590 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2591 sv_setpv(PL_linestr, "");
2592 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2593 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2594 PL_doextract = FALSE;
2598 } while (PL_doextract);
2599 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2600 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2601 SV *sv = NEWSV(85,0);
2603 sv_upgrade(sv, SVt_PVMG);
2604 sv_setsv(sv,PL_linestr);
2605 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2607 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2608 if (CopLINE(PL_curcop) == 1) {
2609 while (s < PL_bufend && isSPACE(*s))
2611 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2615 if (*s == '#' && *(s+1) == '!')
2617 #ifdef ALTERNATE_SHEBANG
2619 static char as[] = ALTERNATE_SHEBANG;
2620 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2621 d = s + (sizeof(as) - 1);
2623 #endif /* ALTERNATE_SHEBANG */
2632 while (*d && !isSPACE(*d))
2636 #ifdef ARG_ZERO_IS_SCRIPT
2637 if (ipathend > ipath) {
2639 * HP-UX (at least) sets argv[0] to the script name,
2640 * which makes $^X incorrect. And Digital UNIX and Linux,
2641 * at least, set argv[0] to the basename of the Perl
2642 * interpreter. So, having found "#!", we'll set it right.
2644 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2645 assert(SvPOK(x) || SvGMAGICAL(x));
2646 if (sv_eq(x, CopFILESV(PL_curcop))) {
2647 sv_setpvn(x, ipath, ipathend - ipath);
2650 TAINT_NOT; /* $^X is always tainted, but that's OK */
2652 #endif /* ARG_ZERO_IS_SCRIPT */
2657 d = instr(s,"perl -");
2659 d = instr(s,"perl");
2661 /* avoid getting into infinite loops when shebang
2662 * line contains "Perl" rather than "perl" */
2664 for (d = ipathend-4; d >= ipath; --d) {
2665 if ((*d == 'p' || *d == 'P')
2666 && !ibcmp(d, "perl", 4))
2676 #ifdef ALTERNATE_SHEBANG
2678 * If the ALTERNATE_SHEBANG on this system starts with a
2679 * character that can be part of a Perl expression, then if
2680 * we see it but not "perl", we're probably looking at the
2681 * start of Perl code, not a request to hand off to some
2682 * other interpreter. Similarly, if "perl" is there, but
2683 * not in the first 'word' of the line, we assume the line
2684 * contains the start of the Perl program.
2686 if (d && *s != '#') {
2688 while (*c && !strchr("; \t\r\n\f\v#", *c))
2691 d = Nullch; /* "perl" not in first word; ignore */
2693 *s = '#'; /* Don't try to parse shebang line */
2695 #endif /* ALTERNATE_SHEBANG */
2696 #ifndef MACOS_TRADITIONAL
2701 !instr(s,"indir") &&
2702 instr(PL_origargv[0],"perl"))
2708 while (s < PL_bufend && isSPACE(*s))
2710 if (s < PL_bufend) {
2711 Newz(899,newargv,PL_origargc+3,char*);
2713 while (s < PL_bufend && !isSPACE(*s))
2716 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2719 newargv = PL_origargv;
2721 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2722 Perl_croak(aTHX_ "Can't exec %s", ipath);
2726 U32 oldpdb = PL_perldb;
2727 bool oldn = PL_minus_n;
2728 bool oldp = PL_minus_p;
2730 while (*d && !isSPACE(*d)) d++;
2731 while (SPACE_OR_TAB(*d)) d++;
2735 if (*d == 'M' || *d == 'm') {
2737 while (*d && !isSPACE(*d)) d++;
2738 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2741 d = moreswitches(d);
2743 if ((PERLDB_LINE && !oldpdb) ||
2744 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2745 /* if we have already added "LINE: while (<>) {",
2746 we must not do it again */
2748 sv_setpv(PL_linestr, "");
2749 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2750 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2751 PL_preambled = FALSE;
2753 (void)gv_fetchfile(PL_origfilename);
2760 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2762 PL_lex_state = LEX_FORMLINE;
2767 #ifdef PERL_STRICT_CR
2768 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2770 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2772 case ' ': case '\t': case '\f': case 013:
2773 #ifdef MACOS_TRADITIONAL
2780 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2781 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2782 /* handle eval qq[#line 1 "foo"\n ...] */
2783 CopLINE_dec(PL_curcop);
2787 while (s < d && *s != '\n')
2792 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2794 PL_lex_state = LEX_FORMLINE;
2804 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2811 while (s < PL_bufend && SPACE_OR_TAB(*s))
2814 if (strnEQ(s,"=>",2)) {
2815 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2816 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2817 "### Saw unary minus before =>, forcing word '%s'\n", s);
2819 OPERATOR('-'); /* unary minus */
2821 PL_last_uni = PL_oldbufptr;
2823 case 'r': ftst = OP_FTEREAD; break;
2824 case 'w': ftst = OP_FTEWRITE; break;
2825 case 'x': ftst = OP_FTEEXEC; break;
2826 case 'o': ftst = OP_FTEOWNED; break;
2827 case 'R': ftst = OP_FTRREAD; break;
2828 case 'W': ftst = OP_FTRWRITE; break;
2829 case 'X': ftst = OP_FTREXEC; break;
2830 case 'O': ftst = OP_FTROWNED; break;
2831 case 'e': ftst = OP_FTIS; break;
2832 case 'z': ftst = OP_FTZERO; break;
2833 case 's': ftst = OP_FTSIZE; break;
2834 case 'f': ftst = OP_FTFILE; break;
2835 case 'd': ftst = OP_FTDIR; break;
2836 case 'l': ftst = OP_FTLINK; break;
2837 case 'p': ftst = OP_FTPIPE; break;
2838 case 'S': ftst = OP_FTSOCK; break;
2839 case 'u': ftst = OP_FTSUID; break;
2840 case 'g': ftst = OP_FTSGID; break;
2841 case 'k': ftst = OP_FTSVTX; break;
2842 case 'b': ftst = OP_FTBLK; break;
2843 case 'c': ftst = OP_FTCHR; break;
2844 case 't': ftst = OP_FTTTY; break;
2845 case 'T': ftst = OP_FTTEXT; break;
2846 case 'B': ftst = OP_FTBINARY; break;
2847 case 'M': case 'A': case 'C':
2848 gv_fetchpv("\024",TRUE, SVt_PV);
2850 case 'M': ftst = OP_FTMTIME; break;
2851 case 'A': ftst = OP_FTATIME; break;
2852 case 'C': ftst = OP_FTCTIME; break;
2860 PL_last_lop_op = ftst;
2861 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2862 "### Saw file test %c\n", ftst);
2864 if (*s == '(' && ckWARN(WARN_AMBIGUOUS))
2865 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2866 "Ambiguous -%c() resolved as a file test",
2871 /* Assume it was a minus followed by a one-letter named
2872 * subroutine call (or a -bareword), then. */
2879 if (PL_expect == XOPERATOR)
2884 else if (*s == '>') {
2887 if (isIDFIRST_lazy_if(s,UTF)) {
2888 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2896 if (PL_expect == XOPERATOR)
2899 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2901 OPERATOR('-'); /* unary minus */
2908 if (PL_expect == XOPERATOR)
2913 if (PL_expect == XOPERATOR)
2916 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2922 if (PL_expect != XOPERATOR) {
2923 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2924 PL_expect = XOPERATOR;
2925 force_ident(PL_tokenbuf, '*');
2938 if (PL_expect == XOPERATOR) {
2942 PL_tokenbuf[0] = '%';
2943 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2944 if (!PL_tokenbuf[1]) {
2946 yyerror("Final % should be \\% or %name");
2949 PL_pending_ident = '%';
2968 switch (PL_expect) {
2971 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2973 PL_bufptr = s; /* update in case we back off */
2979 PL_expect = XTERMBLOCK;
2983 while (isIDFIRST_lazy_if(s,UTF)) {
2984 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2985 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2986 if (tmp < 0) tmp = -tmp;
3001 d = scan_str(d,TRUE,TRUE);
3004 SvREFCNT_dec(PL_lex_stuff);
3005 PL_lex_stuff = Nullsv;
3007 /* MUST advance bufptr here to avoid bogus
3008 "at end of line" context messages from yyerror().
3010 PL_bufptr = s + len;
3011 yyerror("Unterminated attribute parameter in attribute list");
3014 return 0; /* EOF indicator */
3018 SV *sv = newSVpvn(s, len);
3019 sv_catsv(sv, PL_lex_stuff);
3020 attrs = append_elem(OP_LIST, attrs,
3021 newSVOP(OP_CONST, 0, sv));
3022 SvREFCNT_dec(PL_lex_stuff);
3023 PL_lex_stuff = Nullsv;
3026 attrs = append_elem(OP_LIST, attrs,
3027 newSVOP(OP_CONST, 0,
3031 if (*s == ':' && s[1] != ':')
3034 break; /* require real whitespace or :'s */
3036 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3037 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3038 char q = ((*s == '\'') ? '"' : '\'');
3039 /* If here for an expression, and parsed no attrs, back off. */
3040 if (tmp == '=' && !attrs) {
3044 /* MUST advance bufptr here to avoid bogus "at end of line"
3045 context messages from yyerror().
3049 yyerror("Unterminated attribute list");
3051 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3059 PL_nextval[PL_nexttoke].opval = attrs;
3067 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3068 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3084 if (PL_lex_brackets <= 0)
3085 yyerror("Unmatched right square bracket");
3088 if (PL_lex_state == LEX_INTERPNORMAL) {
3089 if (PL_lex_brackets == 0) {
3090 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3091 PL_lex_state = LEX_INTERPEND;
3098 if (PL_lex_brackets > 100) {
3099 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3100 if (newlb != PL_lex_brackstack) {
3102 PL_lex_brackstack = newlb;
3105 switch (PL_expect) {
3107 if (PL_lex_formbrack) {
3111 if (PL_oldoldbufptr == PL_last_lop)
3112 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3114 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3115 OPERATOR(HASHBRACK);
3117 while (s < PL_bufend && SPACE_OR_TAB(*s))
3120 PL_tokenbuf[0] = '\0';
3121 if (d < PL_bufend && *d == '-') {
3122 PL_tokenbuf[0] = '-';
3124 while (d < PL_bufend && SPACE_OR_TAB(*d))
3127 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3128 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3130 while (d < PL_bufend && SPACE_OR_TAB(*d))
3133 char minus = (PL_tokenbuf[0] == '-');
3134 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3142 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3147 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3152 if (PL_oldoldbufptr == PL_last_lop)
3153 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3155 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3158 OPERATOR(HASHBRACK);
3159 /* This hack serves to disambiguate a pair of curlies
3160 * as being a block or an anon hash. Normally, expectation
3161 * determines that, but in cases where we're not in a
3162 * position to expect anything in particular (like inside
3163 * eval"") we have to resolve the ambiguity. This code
3164 * covers the case where the first term in the curlies is a
3165 * quoted string. Most other cases need to be explicitly
3166 * disambiguated by prepending a `+' before the opening
3167 * curly in order to force resolution as an anon hash.
3169 * XXX should probably propagate the outer expectation
3170 * into eval"" to rely less on this hack, but that could
3171 * potentially break current behavior of eval"".
3175 if (*s == '\'' || *s == '"' || *s == '`') {
3176 /* common case: get past first string, handling escapes */
3177 for (t++; t < PL_bufend && *t != *s;)
3178 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3182 else if (*s == 'q') {
3185 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3189 char open, close, term;
3192 while (t < PL_bufend && isSPACE(*t))
3196 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3200 for (t++; t < PL_bufend; t++) {
3201 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3203 else if (*t == open)
3207 for (t++; t < PL_bufend; t++) {
3208 if (*t == '\\' && t+1 < PL_bufend)
3210 else if (*t == close && --brackets <= 0)
3212 else if (*t == open)
3218 else if (isALNUM_lazy_if(t,UTF)) {
3220 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3223 while (t < PL_bufend && isSPACE(*t))
3225 /* if comma follows first term, call it an anon hash */
3226 /* XXX it could be a comma expression with loop modifiers */
3227 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3228 || (*t == '=' && t[1] == '>')))
3229 OPERATOR(HASHBRACK);
3230 if (PL_expect == XREF)
3233 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3239 yylval.ival = CopLINE(PL_curcop);
3240 if (isSPACE(*s) || *s == '#')
3241 PL_copline = NOLINE; /* invalidate current command line number */
3246 if (PL_lex_brackets <= 0)
3247 yyerror("Unmatched right curly bracket");
3249 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3250 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3251 PL_lex_formbrack = 0;
3252 if (PL_lex_state == LEX_INTERPNORMAL) {
3253 if (PL_lex_brackets == 0) {
3254 if (PL_expect & XFAKEBRACK) {
3255 PL_expect &= XENUMMASK;
3256 PL_lex_state = LEX_INTERPEND;
3258 return yylex(); /* ignore fake brackets */
3260 if (*s == '-' && s[1] == '>')
3261 PL_lex_state = LEX_INTERPENDMAYBE;
3262 else if (*s != '[' && *s != '{')
3263 PL_lex_state = LEX_INTERPEND;
3266 if (PL_expect & XFAKEBRACK) {
3267 PL_expect &= XENUMMASK;
3269 return yylex(); /* ignore fake brackets */
3279 if (PL_expect == XOPERATOR) {
3280 if (ckWARN(WARN_SEMICOLON)
3281 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3283 CopLINE_dec(PL_curcop);
3284 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3285 CopLINE_inc(PL_curcop);
3290 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3292 PL_expect = XOPERATOR;
3293 force_ident(PL_tokenbuf, '&');
3297 yylval.ival = (OPpENTERSUB_AMPER<<8);
3316 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3317 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3319 if (PL_expect == XSTATE && isALPHA(tmp) &&
3320 (s == PL_linestart+1 || s[-2] == '\n') )
3322 if (PL_in_eval && !PL_rsfp) {
3327 if (strnEQ(s,"=cut",4)) {
3341 PL_doextract = TRUE;
3344 if (PL_lex_brackets < PL_lex_formbrack) {
3346 #ifdef PERL_STRICT_CR
3347 for (t = s; SPACE_OR_TAB(*t); t++) ;
3349 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3351 if (*t == '\n' || *t == '#') {
3369 if (PL_expect != XOPERATOR) {
3370 if (s[1] != '<' && !strchr(s,'>'))
3373 s = scan_heredoc(s);
3375 s = scan_inputsymbol(s);
3376 TERM(sublex_start());
3381 SHop(OP_LEFT_SHIFT);
3395 SHop(OP_RIGHT_SHIFT);
3404 if (PL_expect == XOPERATOR) {
3405 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3408 return ','; /* grandfather non-comma-format format */
3412 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3413 PL_tokenbuf[0] = '@';
3414 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3415 sizeof PL_tokenbuf - 1, FALSE);
3416 if (PL_expect == XOPERATOR)
3417 no_op("Array length", s);
3418 if (!PL_tokenbuf[1])
3420 PL_expect = XOPERATOR;
3421 PL_pending_ident = '#';
3425 PL_tokenbuf[0] = '$';
3426 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3427 sizeof PL_tokenbuf - 1, FALSE);
3428 if (PL_expect == XOPERATOR)
3430 if (!PL_tokenbuf[1]) {
3432 yyerror("Final $ should be \\$ or $name");
3436 /* This kludge not intended to be bulletproof. */
3437 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3438 yylval.opval = newSVOP(OP_CONST, 0,
3439 newSViv(PL_compiling.cop_arybase));
3440 yylval.opval->op_private = OPpCONST_ARYBASE;
3446 if (PL_lex_state == LEX_NORMAL)
3449 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3452 PL_tokenbuf[0] = '@';
3453 if (ckWARN(WARN_SYNTAX)) {
3455 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3458 PL_bufptr = skipspace(PL_bufptr);
3459 while (t < PL_bufend && *t != ']')
3461 Perl_warner(aTHX_ WARN_SYNTAX,
3462 "Multidimensional syntax %.*s not supported",
3463 (t - PL_bufptr) + 1, PL_bufptr);
3467 else if (*s == '{') {
3468 PL_tokenbuf[0] = '%';
3469 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3470 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3472 char tmpbuf[sizeof PL_tokenbuf];
3474 for (t++; isSPACE(*t); t++) ;
3475 if (isIDFIRST_lazy_if(t,UTF)) {
3476 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3477 for (; isSPACE(*t); t++) ;
3478 if (*t == ';' && get_cv(tmpbuf, FALSE))
3479 Perl_warner(aTHX_ WARN_SYNTAX,
3480 "You need to quote \"%s\"", tmpbuf);
3486 PL_expect = XOPERATOR;
3487 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3488 bool islop = (PL_last_lop == PL_oldoldbufptr);
3489 if (!islop || PL_last_lop_op == OP_GREPSTART)
3490 PL_expect = XOPERATOR;
3491 else if (strchr("$@\"'`q", *s))
3492 PL_expect = XTERM; /* e.g. print $fh "foo" */
3493 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3494 PL_expect = XTERM; /* e.g. print $fh &sub */
3495 else if (isIDFIRST_lazy_if(s,UTF)) {
3496 char tmpbuf[sizeof PL_tokenbuf];
3497 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3498 if ((tmp = keyword(tmpbuf, len))) {
3499 /* binary operators exclude handle interpretations */
3511 PL_expect = XTERM; /* e.g. print $fh length() */
3516 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3517 if (gv && GvCVu(gv))
3518 PL_expect = XTERM; /* e.g. print $fh subr() */
3521 else if (isDIGIT(*s))
3522 PL_expect = XTERM; /* e.g. print $fh 3 */
3523 else if (*s == '.' && isDIGIT(s[1]))
3524 PL_expect = XTERM; /* e.g. print $fh .3 */
3525 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3526 PL_expect = XTERM; /* e.g. print $fh -1 */
3527 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3528 PL_expect = XTERM; /* print $fh <<"EOF" */
3530 PL_pending_ident = '$';
3534 if (PL_expect == XOPERATOR)
3536 PL_tokenbuf[0] = '@';
3537 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3538 if (!PL_tokenbuf[1]) {
3540 yyerror("Final @ should be \\@ or @name");
3543 if (PL_lex_state == LEX_NORMAL)
3545 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3547 PL_tokenbuf[0] = '%';
3549 /* Warn about @ where they meant $. */
3550 if (ckWARN(WARN_SYNTAX)) {
3551 if (*s == '[' || *s == '{') {
3553 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3555 if (*t == '}' || *t == ']') {
3557 PL_bufptr = skipspace(PL_bufptr);
3558 Perl_warner(aTHX_ WARN_SYNTAX,
3559 "Scalar value %.*s better written as $%.*s",
3560 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3565 PL_pending_ident = '@';
3568 case '/': /* may either be division or pattern */
3569 case '?': /* may either be conditional or pattern */
3570 if (PL_expect != XOPERATOR) {
3571 /* Disable warning on "study /blah/" */
3572 if (PL_oldoldbufptr == PL_last_uni
3573 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3574 || memNE(PL_last_uni, "study", 5)
3575 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3577 s = scan_pat(s,OP_MATCH);
3578 TERM(sublex_start());
3586 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3587 #ifdef PERL_STRICT_CR
3590 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3592 && (s == PL_linestart || s[-1] == '\n') )
3594 PL_lex_formbrack = 0;
3598 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3604 yylval.ival = OPf_SPECIAL;
3610 if (PL_expect != XOPERATOR)
3615 case '0': case '1': case '2': case '3': case '4':
3616 case '5': case '6': case '7': case '8': case '9':
3617 s = scan_num(s, &yylval);
3618 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3619 "### Saw number in '%s'\n", s);
3621 if (PL_expect == XOPERATOR)
3626 s = scan_str(s,FALSE,FALSE);
3627 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3628 "### Saw string in '%s'\n", s);
3630 if (PL_expect == XOPERATOR) {
3631 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3634 return ','; /* grandfather non-comma-format format */
3640 missingterm((char*)0);
3641 yylval.ival = OP_CONST;
3642 TERM(sublex_start());
3645 s = scan_str(s,FALSE,FALSE);
3646 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3647 "### Saw string in '%s'\n", s);
3649 if (PL_expect == XOPERATOR) {
3650 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3653 return ','; /* grandfather non-comma-format format */
3659 missingterm((char*)0);
3660 yylval.ival = OP_CONST;
3661 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3662 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3663 yylval.ival = OP_STRINGIFY;
3667 TERM(sublex_start());
3670 s = scan_str(s,FALSE,FALSE);
3671 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3672 "### Saw backtick string in '%s'\n", s);
3674 if (PL_expect == XOPERATOR)
3675 no_op("Backticks",s);
3677 missingterm((char*)0);
3678 yylval.ival = OP_BACKTICK;
3680 TERM(sublex_start());
3684 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3685 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3687 if (PL_expect == XOPERATOR)
3688 no_op("Backslash",s);
3692 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3696 while (isDIGIT(*start) || *start == '_')
3698 if (*start == '.' && isDIGIT(start[1])) {
3699 s = scan_num(s, &yylval);
3702 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3703 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3707 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3710 s = scan_num(s, &yylval);
3717 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3756 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3758 /* Some keywords can be followed by any delimiter, including ':' */
3759 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3760 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3761 (PL_tokenbuf[0] == 'q' &&
3762 strchr("qwxr", PL_tokenbuf[1])))));
3764 /* x::* is just a word, unless x is "CORE" */
3765 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3769 while (d < PL_bufend && isSPACE(*d))
3770 d++; /* no comments skipped here, or s### is misparsed */
3772 /* Is this a label? */
3773 if (!tmp && PL_expect == XSTATE
3774 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3776 yylval.pval = savepv(PL_tokenbuf);
3781 /* Check for keywords */
3782 tmp = keyword(PL_tokenbuf, len);
3784 /* Is this a word before a => operator? */
3785 if (*d == '=' && d[1] == '>') {
3787 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3788 yylval.opval->op_private = OPpCONST_BARE;
3792 if (tmp < 0) { /* second-class keyword? */
3793 GV *ogv = Nullgv; /* override (winner) */
3794 GV *hgv = Nullgv; /* hidden (loser) */
3795 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3797 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3800 if (GvIMPORTED_CV(gv))
3802 else if (! CvMETHOD(cv))
3806 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3807 (gv = *gvp) != (GV*)&PL_sv_undef &&
3808 GvCVu(gv) && GvIMPORTED_CV(gv))
3814 tmp = 0; /* overridden by import or by GLOBAL */
3817 && -tmp==KEY_lock /* XXX generalizable kludge */
3819 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3821 tmp = 0; /* any sub overrides "weak" keyword */
3823 else { /* no override */
3827 if (ckWARN(WARN_AMBIGUOUS) && hgv
3828 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3829 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3830 "Ambiguous call resolved as CORE::%s(), %s",
3831 GvENAME(hgv), "qualify as such or use &");
3838 default: /* not a keyword */
3841 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3843 /* Get the rest if it looks like a package qualifier */
3845 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3847 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3850 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3851 *s == '\'' ? "'" : "::");
3855 if (PL_expect == XOPERATOR) {
3856 if (PL_bufptr == PL_linestart) {
3857 CopLINE_dec(PL_curcop);
3858 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3859 CopLINE_inc(PL_curcop);
3862 no_op("Bareword",s);
3865 /* Look for a subroutine with this name in current package,
3866 unless name is "Foo::", in which case Foo is a bearword
3867 (and a package name). */
3870 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3872 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3873 Perl_warner(aTHX_ WARN_BAREWORD,
3874 "Bareword \"%s\" refers to nonexistent package",
3877 PL_tokenbuf[len] = '\0';
3884 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3887 /* if we saw a global override before, get the right name */
3890 sv = newSVpvn("CORE::GLOBAL::",14);
3891 sv_catpv(sv,PL_tokenbuf);
3894 sv = newSVpv(PL_tokenbuf,0);
3896 /* Presume this is going to be a bareword of some sort. */
3899 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3900 yylval.opval->op_private = OPpCONST_BARE;
3902 /* And if "Foo::", then that's what it certainly is. */
3907 /* See if it's the indirect object for a list operator. */
3909 if (PL_oldoldbufptr &&
3910 PL_oldoldbufptr < PL_bufptr &&
3911 (PL_oldoldbufptr == PL_last_lop
3912 || PL_oldoldbufptr == PL_last_uni) &&
3913 /* NO SKIPSPACE BEFORE HERE! */
3914 (PL_expect == XREF ||
3915 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3917 bool immediate_paren = *s == '(';
3919 /* (Now we can afford to cross potential line boundary.) */
3922 /* Two barewords in a row may indicate method call. */
3924 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3927 /* If not a declared subroutine, it's an indirect object. */
3928 /* (But it's an indir obj regardless for sort.) */
3930 if ((PL_last_lop_op == OP_SORT ||
3931 (!immediate_paren && (!gv || !GvCVu(gv)))) &&
3932 (PL_last_lop_op != OP_MAPSTART &&
3933 PL_last_lop_op != OP_GREPSTART))
3935 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3941 PL_expect = XOPERATOR;
3944 /* Is this a word before a => operator? */
3945 if (*s == '=' && s[1] == '>') {
3947 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3951 /* If followed by a paren, it's certainly a subroutine. */
3954 if (gv && GvCVu(gv)) {
3955 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3956 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3961 PL_nextval[PL_nexttoke].opval = yylval.opval;
3962 PL_expect = XOPERATOR;
3968 /* If followed by var or block, call it a method (unless sub) */
3970 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3971 PL_last_lop = PL_oldbufptr;
3972 PL_last_lop_op = OP_METHOD;
3976 /* If followed by a bareword, see if it looks like indir obj. */
3978 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3981 /* Not a method, so call it a subroutine (if defined) */
3983 if (gv && GvCVu(gv)) {
3985 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3986 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3987 "Ambiguous use of -%s resolved as -&%s()",
3988 PL_tokenbuf, PL_tokenbuf);
3989 /* Check for a constant sub */
3991 if ((sv = cv_const_sv(cv))) {
3993 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3994 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3995 yylval.opval->op_private = 0;
3999 /* Resolve to GV now. */
4000 op_free(yylval.opval);
4001 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4002 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4003 PL_last_lop = PL_oldbufptr;
4004 PL_last_lop_op = OP_ENTERSUB;
4005 /* Is there a prototype? */
4008 char *proto = SvPV((SV*)cv, len);
4011 if (strEQ(proto, "$"))
4013 if (*proto == '&' && *s == '{') {
4014 sv_setpv(PL_subname,"__ANON__");
4018 PL_nextval[PL_nexttoke].opval = yylval.opval;
4024 /* Call it a bare word */
4026 if (PL_hints & HINT_STRICT_SUBS)
4027 yylval.opval->op_private |= OPpCONST_STRICT;
4030 if (ckWARN(WARN_RESERVED)) {
4031 if (lastchar != '-') {
4032 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4034 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4041 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4042 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4043 "Operator or semicolon missing before %c%s",
4044 lastchar, PL_tokenbuf);
4045 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4046 "Ambiguous use of %c resolved as operator %c",
4047 lastchar, lastchar);
4053 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4054 newSVpv(CopFILE(PL_curcop),0));
4058 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4059 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4062 case KEY___PACKAGE__:
4063 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4065 ? newSVsv(PL_curstname)
4074 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4075 char *pname = "main";
4076 if (PL_tokenbuf[2] == 'D')
4077 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4078 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4081 GvIOp(gv) = newIO();
4082 IoIFP(GvIOp(gv)) = PL_rsfp;
4083 #if defined(HAS_FCNTL) && defined(F_SETFD)
4085 int fd = PerlIO_fileno(PL_rsfp);
4086 fcntl(fd,F_SETFD,fd >= 3);
4089 /* Mark this internal pseudo-handle as clean */
4090 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4092 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4093 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4094 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4096 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4097 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4098 /* if the script was opened in binmode, we need to revert
4099 * it to text mode for compatibility; but only iff it has CRs
4100 * XXX this is a questionable hack at best. */
4101 if (PL_bufend-PL_bufptr > 2
4102 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4105 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4106 loc = PerlIO_tell(PL_rsfp);
4107 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4109 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4110 #if defined(__BORLANDC__)
4111 /* XXX see note in do_binmode() */
4112 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4115 PerlIO_seek(PL_rsfp, loc, 0);
4130 if (PL_expect == XSTATE) {
4137 if (*s == ':' && s[1] == ':') {
4140 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4141 if (!(tmp = keyword(PL_tokenbuf, len)))
4142 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4156 LOP(OP_ACCEPT,XTERM);
4162 LOP(OP_ATAN2,XTERM);
4168 LOP(OP_BINMODE,XTERM);
4171 LOP(OP_BLESS,XTERM);
4180 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4197 if (!PL_cryptseen) {
4198 PL_cryptseen = TRUE;
4202 LOP(OP_CRYPT,XTERM);
4205 if (ckWARN(WARN_CHMOD)) {
4206 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4207 if (*d != '0' && isDIGIT(*d))
4208 Perl_warner(aTHX_ WARN_CHMOD,
4209 "chmod() mode argument is missing initial 0");
4211 LOP(OP_CHMOD,XTERM);
4214 LOP(OP_CHOWN,XTERM);
4217 LOP(OP_CONNECT,XTERM);
4233 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4237 PL_hints |= HINT_BLOCK_SCOPE;
4247 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4248 LOP(OP_DBMOPEN,XTERM);
4254 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4261 yylval.ival = CopLINE(PL_curcop);
4275 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4276 UNIBRACK(OP_ENTEREVAL);
4291 case KEY_endhostent:
4297 case KEY_endservent:
4300 case KEY_endprotoent:
4311 yylval.ival = CopLINE(PL_curcop);
4313 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4315 if ((PL_bufend - p) >= 3 &&
4316 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4318 else if ((PL_bufend - p) >= 4 &&
4319 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4322 if (isIDFIRST_lazy_if(p,UTF)) {
4323 p = scan_ident(p, PL_bufend,
4324 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4328 Perl_croak(aTHX_ "Missing $ on loop variable");
4333 LOP(OP_FORMLINE,XTERM);
4339 LOP(OP_FCNTL,XTERM);
4345 LOP(OP_FLOCK,XTERM);
4354 LOP(OP_GREPSTART, XREF);
4357 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4372 case KEY_getpriority:
4373 LOP(OP_GETPRIORITY,XTERM);
4375 case KEY_getprotobyname:
4378 case KEY_getprotobynumber:
4379 LOP(OP_GPBYNUMBER,XTERM);
4381 case KEY_getprotoent:
4393 case KEY_getpeername:
4394 UNI(OP_GETPEERNAME);
4396 case KEY_gethostbyname:
4399 case KEY_gethostbyaddr:
4400 LOP(OP_GHBYADDR,XTERM);
4402 case KEY_gethostent:
4405 case KEY_getnetbyname:
4408 case KEY_getnetbyaddr:
4409 LOP(OP_GNBYADDR,XTERM);
4414 case KEY_getservbyname:
4415 LOP(OP_GSBYNAME,XTERM);
4417 case KEY_getservbyport:
4418 LOP(OP_GSBYPORT,XTERM);
4420 case KEY_getservent:
4423 case KEY_getsockname:
4424 UNI(OP_GETSOCKNAME);
4426 case KEY_getsockopt:
4427 LOP(OP_GSOCKOPT,XTERM);
4449 yylval.ival = CopLINE(PL_curcop);
4453 LOP(OP_INDEX,XTERM);
4459 LOP(OP_IOCTL,XTERM);
4471 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4503 LOP(OP_LISTEN,XTERM);
4512 s = scan_pat(s,OP_MATCH);
4513 TERM(sublex_start());
4516 LOP(OP_MAPSTART, XREF);
4519 LOP(OP_MKDIR,XTERM);
4522 LOP(OP_MSGCTL,XTERM);
4525 LOP(OP_MSGGET,XTERM);
4528 LOP(OP_MSGRCV,XTERM);
4531 LOP(OP_MSGSND,XTERM);
4537 if (isIDFIRST_lazy_if(s,UTF)) {
4538 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4539 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4541 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4542 if (!PL_in_my_stash) {
4545 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4553 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4560 if (PL_expect != XSTATE)
4561 yyerror("\"no\" not allowed in expression");
4562 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4563 s = force_version(s);
4568 if (*s == '(' || (s = skipspace(s), *s == '('))
4575 if (isIDFIRST_lazy_if(s,UTF)) {
4577 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4579 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4580 Perl_warner(aTHX_ WARN_PRECEDENCE,
4581 "Precedence problem: open %.*s should be open(%.*s)",
4587 yylval.ival = OP_OR;
4597 LOP(OP_OPEN_DIR,XTERM);
4600 checkcomma(s,PL_tokenbuf,"filehandle");
4604 checkcomma(s,PL_tokenbuf,"filehandle");
4623 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4627 LOP(OP_PIPE_OP,XTERM);
4630 s = scan_str(s,FALSE,FALSE);
4632 missingterm((char*)0);
4633 yylval.ival = OP_CONST;
4634 TERM(sublex_start());
4640 s = scan_str(s,FALSE,FALSE);
4642 missingterm((char*)0);
4644 if (SvCUR(PL_lex_stuff)) {
4647 d = SvPV_force(PL_lex_stuff, len);
4649 for (; isSPACE(*d) && len; --len, ++d) ;
4652 if (!warned && ckWARN(WARN_QW)) {
4653 for (; !isSPACE(*d) && len; --len, ++d) {
4655 Perl_warner(aTHX_ WARN_QW,
4656 "Possible attempt to separate words with commas");
4659 else if (*d == '#') {
4660 Perl_warner(aTHX_ WARN_QW,
4661 "Possible attempt to put comments in qw() list");
4667 for (; !isSPACE(*d) && len; --len, ++d) ;
4669 words = append_elem(OP_LIST, words,
4670 newSVOP(OP_CONST, 0, tokeq(newSVpvn(b, d-b))));
4674 PL_nextval[PL_nexttoke].opval = words;
4679 SvREFCNT_dec(PL_lex_stuff);
4680 PL_lex_stuff = Nullsv;
4685 s = scan_str(s,FALSE,FALSE);
4687 missingterm((char*)0);
4688 yylval.ival = OP_STRINGIFY;
4689 if (SvIVX(PL_lex_stuff) == '\'')
4690 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4691 TERM(sublex_start());
4694 s = scan_pat(s,OP_QR);
4695 TERM(sublex_start());
4698 s = scan_str(s,FALSE,FALSE);
4700 missingterm((char*)0);
4701 yylval.ival = OP_BACKTICK;
4703 TERM(sublex_start());
4710 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4711 s = force_version(s);
4714 *PL_tokenbuf = '\0';
4715 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4716 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4717 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4719 yyerror("<> should be quotes");
4727 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4731 LOP(OP_RENAME,XTERM);
4740 LOP(OP_RINDEX,XTERM);
4763 LOP(OP_REVERSE,XTERM);
4774 TERM(sublex_start());
4776 TOKEN(1); /* force error */
4785 LOP(OP_SELECT,XTERM);
4791 LOP(OP_SEMCTL,XTERM);
4794 LOP(OP_SEMGET,XTERM);
4797 LOP(OP_SEMOP,XTERM);
4803 LOP(OP_SETPGRP,XTERM);
4805 case KEY_setpriority:
4806 LOP(OP_SETPRIORITY,XTERM);
4808 case KEY_sethostent:
4814 case KEY_setservent:
4817 case KEY_setprotoent:
4827 LOP(OP_SEEKDIR,XTERM);
4829 case KEY_setsockopt:
4830 LOP(OP_SSOCKOPT,XTERM);
4836 LOP(OP_SHMCTL,XTERM);
4839 LOP(OP_SHMGET,XTERM);
4842 LOP(OP_SHMREAD,XTERM);
4845 LOP(OP_SHMWRITE,XTERM);
4848 LOP(OP_SHUTDOWN,XTERM);
4857 LOP(OP_SOCKET,XTERM);
4859 case KEY_socketpair:
4860 LOP(OP_SOCKPAIR,XTERM);
4863 checkcomma(s,PL_tokenbuf,"subroutine name");
4865 if (*s == ';' || *s == ')') /* probably a close */
4866 Perl_croak(aTHX_ "sort is now a reserved word");
4868 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4872 LOP(OP_SPLIT,XTERM);
4875 LOP(OP_SPRINTF,XTERM);
4878 LOP(OP_SPLICE,XTERM);
4893 LOP(OP_SUBSTR,XTERM);
4899 char tmpbuf[sizeof PL_tokenbuf];
4901 expectation attrful;
4902 bool have_name, have_proto;
4907 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4908 (*s == ':' && s[1] == ':'))
4911 attrful = XATTRBLOCK;
4912 /* remember buffer pos'n for later force_word */
4913 tboffset = s - PL_oldbufptr;
4914 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4915 if (strchr(tmpbuf, ':'))
4916 sv_setpv(PL_subname, tmpbuf);
4918 sv_setsv(PL_subname,PL_curstname);
4919 sv_catpvn(PL_subname,"::",2);
4920 sv_catpvn(PL_subname,tmpbuf,len);
4927 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4928 PL_expect = XTERMBLOCK;
4929 attrful = XATTRTERM;
4930 sv_setpv(PL_subname,"?");
4934 if (key == KEY_format) {
4936 PL_lex_formbrack = PL_lex_brackets + 1;
4938 (void) force_word(PL_oldbufptr + tboffset, WORD,
4943 /* Look for a prototype */
4947 s = scan_str(s,FALSE,FALSE);
4950 SvREFCNT_dec(PL_lex_stuff);
4951 PL_lex_stuff = Nullsv;
4952 Perl_croak(aTHX_ "Prototype not terminated");
4955 d = SvPVX(PL_lex_stuff);
4957 for (p = d; *p; ++p) {
4962 SvCUR(PL_lex_stuff) = tmp;
4970 if (*s == ':' && s[1] != ':')
4971 PL_expect = attrful;
4974 PL_nextval[PL_nexttoke].opval =
4975 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4976 PL_lex_stuff = Nullsv;
4980 sv_setpv(PL_subname,"__ANON__");
4983 (void) force_word(PL_oldbufptr + tboffset, WORD,
4992 LOP(OP_SYSTEM,XREF);
4995 LOP(OP_SYMLINK,XTERM);
4998 LOP(OP_SYSCALL,XTERM);
5001 LOP(OP_SYSOPEN,XTERM);
5004 LOP(OP_SYSSEEK,XTERM);
5007 LOP(OP_SYSREAD,XTERM);
5010 LOP(OP_SYSWRITE,XTERM);
5014 TERM(sublex_start());
5035 LOP(OP_TRUNCATE,XTERM);
5047 yylval.ival = CopLINE(PL_curcop);
5051 yylval.ival = CopLINE(PL_curcop);
5055 LOP(OP_UNLINK,XTERM);
5061 LOP(OP_UNPACK,XTERM);
5064 LOP(OP_UTIME,XTERM);
5067 if (ckWARN(WARN_UMASK)) {
5068 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5069 if (*d != '0' && isDIGIT(*d))
5070 Perl_warner(aTHX_ WARN_UMASK,
5071 "umask: argument is missing initial 0");
5076 LOP(OP_UNSHIFT,XTERM);
5079 if (PL_expect != XSTATE)
5080 yyerror("\"use\" not allowed in expression");
5082 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5083 s = force_version(s);
5084 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5085 PL_nextval[PL_nexttoke].opval = Nullop;
5090 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5091 s = force_version(s);
5103 yylval.ival = CopLINE(PL_curcop);
5107 PL_hints |= HINT_BLOCK_SCOPE;
5114 LOP(OP_WAITPID,XTERM);
5122 static char ctl_l[2];
5124 if (ctl_l[0] == '\0')
5125 ctl_l[0] = toCTRL('L');
5126 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5129 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5134 if (PL_expect == XOPERATOR)
5140 yylval.ival = OP_XOR;
5145 TERM(sublex_start());
5150 #pragma segment Main
5154 Perl_keyword(pTHX_ register char *d, I32 len)
5159 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5160 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5161 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5162 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5163 if (strEQ(d,"__END__")) return KEY___END__;
5167 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5172 if (strEQ(d,"and")) return -KEY_and;
5173 if (strEQ(d,"abs")) return -KEY_abs;
5176 if (strEQ(d,"alarm")) return -KEY_alarm;
5177 if (strEQ(d,"atan2")) return -KEY_atan2;
5180 if (strEQ(d,"accept")) return -KEY_accept;
5185 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5188 if (strEQ(d,"bless")) return -KEY_bless;
5189 if (strEQ(d,"bind")) return -KEY_bind;
5190 if (strEQ(d,"binmode")) return -KEY_binmode;
5193 if (strEQ(d,"CORE")) return -KEY_CORE;
5194 if (strEQ(d,"CHECK")) return KEY_CHECK;
5199 if (strEQ(d,"cmp")) return -KEY_cmp;
5200 if (strEQ(d,"chr")) return -KEY_chr;
5201 if (strEQ(d,"cos")) return -KEY_cos;
5204 if (strEQ(d,"chop")) return -KEY_chop;
5207 if (strEQ(d,"close")) return -KEY_close;
5208 if (strEQ(d,"chdir")) return -KEY_chdir;
5209 if (strEQ(d,"chomp")) return -KEY_chomp;
5210 if (strEQ(d,"chmod")) return -KEY_chmod;
5211 if (strEQ(d,"chown")) return -KEY_chown;
5212 if (strEQ(d,"crypt")) return -KEY_crypt;
5215 if (strEQ(d,"chroot")) return -KEY_chroot;
5216 if (strEQ(d,"caller")) return -KEY_caller;
5219 if (strEQ(d,"connect")) return -KEY_connect;
5222 if (strEQ(d,"closedir")) return -KEY_closedir;
5223 if (strEQ(d,"continue")) return -KEY_continue;
5228 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5233 if (strEQ(d,"do")) return KEY_do;
5236 if (strEQ(d,"die")) return -KEY_die;
5239 if (strEQ(d,"dump")) return -KEY_dump;
5242 if (strEQ(d,"delete")) return KEY_delete;
5245 if (strEQ(d,"defined")) return KEY_defined;
5246 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5249 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5254 if (strEQ(d,"END")) return KEY_END;
5259 if (strEQ(d,"eq")) return -KEY_eq;
5262 if (strEQ(d,"eof")) return -KEY_eof;
5263 if (strEQ(d,"exp")) return -KEY_exp;
5266 if (strEQ(d,"else")) return KEY_else;
5267 if (strEQ(d,"exit")) return -KEY_exit;
5268 if (strEQ(d,"eval")) return KEY_eval;
5269 if (strEQ(d,"exec")) return -KEY_exec;
5270 if (strEQ(d,"each")) return -KEY_each;
5273 if (strEQ(d,"elsif")) return KEY_elsif;
5276 if (strEQ(d,"exists")) return KEY_exists;
5277 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5280 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5281 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5284 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5287 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5288 if (strEQ(d,"endservent")) return -KEY_endservent;
5291 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5298 if (strEQ(d,"for")) return KEY_for;
5301 if (strEQ(d,"fork")) return -KEY_fork;
5304 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5305 if (strEQ(d,"flock")) return -KEY_flock;
5308 if (strEQ(d,"format")) return KEY_format;
5309 if (strEQ(d,"fileno")) return -KEY_fileno;
5312 if (strEQ(d,"foreach")) return KEY_foreach;
5315 if (strEQ(d,"formline")) return -KEY_formline;
5320 if (strnEQ(d,"get",3)) {
5325 if (strEQ(d,"ppid")) return -KEY_getppid;
5326 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5329 if (strEQ(d,"pwent")) return -KEY_getpwent;
5330 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5331 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5334 if (strEQ(d,"peername")) return -KEY_getpeername;
5335 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5336 if (strEQ(d,"priority")) return -KEY_getpriority;
5339 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5342 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5346 else if (*d == 'h') {
5347 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5348 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5349 if (strEQ(d,"hostent")) return -KEY_gethostent;
5351 else if (*d == 'n') {
5352 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5353 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5354 if (strEQ(d,"netent")) return -KEY_getnetent;
5356 else if (*d == 's') {
5357 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5358 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5359 if (strEQ(d,"servent")) return -KEY_getservent;
5360 if (strEQ(d,"sockname")) return -KEY_getsockname;
5361 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5363 else if (*d == 'g') {
5364 if (strEQ(d,"grent")) return -KEY_getgrent;
5365 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5366 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5368 else if (*d == 'l') {
5369 if (strEQ(d,"login")) return -KEY_getlogin;
5371 else if (strEQ(d,"c")) return -KEY_getc;
5376 if (strEQ(d,"gt")) return -KEY_gt;
5377 if (strEQ(d,"ge")) return -KEY_ge;
5380 if (strEQ(d,"grep")) return KEY_grep;
5381 if (strEQ(d,"goto")) return KEY_goto;
5382 if (strEQ(d,"glob")) return KEY_glob;
5385 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5390 if (strEQ(d,"hex")) return -KEY_hex;
5393 if (strEQ(d,"INIT")) return KEY_INIT;
5398 if (strEQ(d,"if")) return KEY_if;
5401 if (strEQ(d,"int")) return -KEY_int;
5404 if (strEQ(d,"index")) return -KEY_index;
5405 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5410 if (strEQ(d,"join")) return -KEY_join;
5414 if (strEQ(d,"keys")) return -KEY_keys;
5415 if (strEQ(d,"kill")) return -KEY_kill;
5421 if (strEQ(d,"lt")) return -KEY_lt;
5422 if (strEQ(d,"le")) return -KEY_le;
5423 if (strEQ(d,"lc")) return -KEY_lc;
5426 if (strEQ(d,"log")) return -KEY_log;
5429 if (strEQ(d,"last")) return KEY_last;
5430 if (strEQ(d,"link")) return -KEY_link;
5431 if (strEQ(d,"lock")) return -KEY_lock;
5434 if (strEQ(d,"local")) return KEY_local;
5435 if (strEQ(d,"lstat")) return -KEY_lstat;
5438 if (strEQ(d,"length")) return -KEY_length;
5439 if (strEQ(d,"listen")) return -KEY_listen;
5442 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5445 if (strEQ(d,"localtime")) return -KEY_localtime;
5451 case 1: return KEY_m;
5453 if (strEQ(d,"my")) return KEY_my;
5456 if (strEQ(d,"map")) return KEY_map;
5459 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5462 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5463 if (strEQ(d,"msgget")) return -KEY_msgget;
5464 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5465 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5470 if (strEQ(d,"next")) return KEY_next;
5471 if (strEQ(d,"ne")) return -KEY_ne;
5472 if (strEQ(d,"not")) return -KEY_not;
5473 if (strEQ(d,"no")) return KEY_no;
5478 if (strEQ(d,"or")) return -KEY_or;
5481 if (strEQ(d,"ord")) return -KEY_ord;
5482 if (strEQ(d,"oct")) return -KEY_oct;
5483 if (strEQ(d,"our")) return KEY_our;
5486 if (strEQ(d,"open")) return -KEY_open;
5489 if (strEQ(d,"opendir")) return -KEY_opendir;
5496 if (strEQ(d,"pop")) return -KEY_pop;
5497 if (strEQ(d,"pos")) return KEY_pos;
5500 if (strEQ(d,"push")) return -KEY_push;
5501 if (strEQ(d,"pack")) return -KEY_pack;
5502 if (strEQ(d,"pipe")) return -KEY_pipe;
5505 if (strEQ(d,"print")) return KEY_print;
5508 if (strEQ(d,"printf")) return KEY_printf;
5511 if (strEQ(d,"package")) return KEY_package;
5514 if (strEQ(d,"prototype")) return KEY_prototype;
5519 if (strEQ(d,"q")) return KEY_q;
5520 if (strEQ(d,"qr")) return KEY_qr;
5521 if (strEQ(d,"qq")) return KEY_qq;
5522 if (strEQ(d,"qw")) return KEY_qw;
5523 if (strEQ(d,"qx")) return KEY_qx;
5525 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5530 if (strEQ(d,"ref")) return -KEY_ref;
5533 if (strEQ(d,"read")) return -KEY_read;
5534 if (strEQ(d,"rand")) return -KEY_rand;
5535 if (strEQ(d,"recv")) return -KEY_recv;
5536 if (strEQ(d,"redo")) return KEY_redo;
5539 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5540 if (strEQ(d,"reset")) return -KEY_reset;
5543 if (strEQ(d,"return")) return KEY_return;
5544 if (strEQ(d,"rename")) return -KEY_rename;
5545 if (strEQ(d,"rindex")) return -KEY_rindex;
5548 if (strEQ(d,"require")) return -KEY_require;
5549 if (strEQ(d,"reverse")) return -KEY_reverse;
5550 if (strEQ(d,"readdir")) return -KEY_readdir;
5553 if (strEQ(d,"readlink")) return -KEY_readlink;
5554 if (strEQ(d,"readline")) return -KEY_readline;
5555 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5558 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5564 case 0: return KEY_s;
5566 if (strEQ(d,"scalar")) return KEY_scalar;
5571 if (strEQ(d,"seek")) return -KEY_seek;
5572 if (strEQ(d,"send")) return -KEY_send;
5575 if (strEQ(d,"semop")) return -KEY_semop;
5578 if (strEQ(d,"select")) return -KEY_select;
5579 if (strEQ(d,"semctl")) return -KEY_semctl;
5580 if (strEQ(d,"semget")) return -KEY_semget;
5583 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5584 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5587 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5588 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5591 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5594 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5595 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5596 if (strEQ(d,"setservent")) return -KEY_setservent;
5599 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5600 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5607 if (strEQ(d,"shift")) return -KEY_shift;
5610 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5611 if (strEQ(d,"shmget")) return -KEY_shmget;
5614 if (strEQ(d,"shmread")) return -KEY_shmread;
5617 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5618 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5623 if (strEQ(d,"sin")) return -KEY_sin;
5626 if (strEQ(d,"sleep")) return -KEY_sleep;
5629 if (strEQ(d,"sort")) return KEY_sort;
5630 if (strEQ(d,"socket")) return -KEY_socket;
5631 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5634 if (strEQ(d,"split")) return KEY_split;
5635 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5636 if (strEQ(d,"splice")) return -KEY_splice;
5639 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5642 if (strEQ(d,"srand")) return -KEY_srand;
5645 if (strEQ(d,"stat")) return -KEY_stat;
5646 if (strEQ(d,"study")) return KEY_study;
5649 if (strEQ(d,"substr")) return -KEY_substr;
5650 if (strEQ(d,"sub")) return KEY_sub;
5655 if (strEQ(d,"system")) return -KEY_system;
5658 if (strEQ(d,"symlink")) return -KEY_symlink;
5659 if (strEQ(d,"syscall")) return -KEY_syscall;
5660 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5661 if (strEQ(d,"sysread")) return -KEY_sysread;
5662 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5665 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5674 if (strEQ(d,"tr")) return KEY_tr;
5677 if (strEQ(d,"tie")) return KEY_tie;
5680 if (strEQ(d,"tell")) return -KEY_tell;
5681 if (strEQ(d,"tied")) return KEY_tied;
5682 if (strEQ(d,"time")) return -KEY_time;
5685 if (strEQ(d,"times")) return -KEY_times;
5688 if (strEQ(d,"telldir")) return -KEY_telldir;
5691 if (strEQ(d,"truncate")) return -KEY_truncate;
5698 if (strEQ(d,"uc")) return -KEY_uc;
5701 if (strEQ(d,"use")) return KEY_use;
5704 if (strEQ(d,"undef")) return KEY_undef;
5705 if (strEQ(d,"until")) return KEY_until;
5706 if (strEQ(d,"untie")) return KEY_untie;
5707 if (strEQ(d,"utime")) return -KEY_utime;
5708 if (strEQ(d,"umask")) return -KEY_umask;
5711 if (strEQ(d,"unless")) return KEY_unless;
5712 if (strEQ(d,"unpack")) return -KEY_unpack;
5713 if (strEQ(d,"unlink")) return -KEY_unlink;
5716 if (strEQ(d,"unshift")) return -KEY_unshift;
5717 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5722 if (strEQ(d,"values")) return -KEY_values;
5723 if (strEQ(d,"vec")) return -KEY_vec;
5728 if (strEQ(d,"warn")) return -KEY_warn;
5729 if (strEQ(d,"wait")) return -KEY_wait;
5732 if (strEQ(d,"while")) return KEY_while;
5733 if (strEQ(d,"write")) return -KEY_write;
5736 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5739 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5744 if (len == 1) return -KEY_x;
5745 if (strEQ(d,"xor")) return -KEY_xor;
5748 if (len == 1) return KEY_y;
5757 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5761 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5762 dTHR; /* only for ckWARN */
5763 if (ckWARN(WARN_SYNTAX)) {
5765 for (w = s+2; *w && level; w++) {
5772 for (; *w && isSPACE(*w); w++) ;
5773 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5774 Perl_warner(aTHX_ WARN_SYNTAX,
5775 "%s (...) interpreted as function",name);
5778 while (s < PL_bufend && isSPACE(*s))
5782 while (s < PL_bufend && isSPACE(*s))
5784 if (isIDFIRST_lazy_if(s,UTF)) {
5786 while (isALNUM_lazy_if(s,UTF))
5788 while (s < PL_bufend && isSPACE(*s))
5793 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5797 Perl_croak(aTHX_ "No comma allowed after %s", what);
5802 /* Either returns sv, or mortalizes sv and returns a new SV*.
5803 Best used as sv=new_constant(..., sv, ...).
5804 If s, pv are NULL, calls subroutine with one argument,
5805 and type is used with error messages only. */
5808 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5812 HV *table = GvHV(PL_hintgv); /* ^H */
5816 const char *why1, *why2, *why3;
5818 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5821 why2 = strEQ(key,"charnames")
5822 ? "(possibly a missing \"use charnames ...\")"
5824 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5825 (type ? type: "undef"), why2);
5827 /* This is convoluted and evil ("goto considered harmful")
5828 * but I do not understand the intricacies of all the different
5829 * failure modes of %^H in here. The goal here is to make
5830 * the most probable error message user-friendly. --jhi */
5835 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5836 (type ? type: "undef"), why1, why2, why3);
5838 yyerror(SvPVX(msg));
5842 cvp = hv_fetch(table, key, strlen(key), FALSE);
5843 if (!cvp || !SvOK(*cvp)) {
5846 why3 = "} is not defined";
5849 sv_2mortal(sv); /* Parent created it permanently */
5852 pv = sv_2mortal(newSVpvn(s, len));
5854 typesv = sv_2mortal(newSVpv(type, 0));
5856 typesv = &PL_sv_undef;
5858 PUSHSTACKi(PERLSI_OVERLOAD);
5870 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5874 /* Check the eval first */
5875 if (!PL_in_eval && SvTRUE(ERRSV)) {
5877 sv_catpv(ERRSV, "Propagated");
5878 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5880 res = SvREFCNT_inc(sv);
5884 (void)SvREFCNT_inc(res);
5893 why1 = "Call to &{$^H{";
5895 why3 = "}} did not return a defined value";
5904 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5906 register char *d = dest;
5907 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5910 Perl_croak(aTHX_ ident_too_long);
5911 if (isALNUM(*s)) /* UTF handled below */
5913 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5918 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5922 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5923 char *t = s + UTF8SKIP(s);
5924 while (*t & 0x80 && is_utf8_mark((U8*)t))
5926 if (d + (t - s) > e)
5927 Perl_croak(aTHX_ ident_too_long);
5928 Copy(s, d, t - s, char);
5941 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5951 e = d + destlen - 3; /* two-character token, ending NUL */
5953 while (isDIGIT(*s)) {
5955 Perl_croak(aTHX_ ident_too_long);
5962 Perl_croak(aTHX_ ident_too_long);
5963 if (isALNUM(*s)) /* UTF handled below */
5965 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5970 else if (*s == ':' && s[1] == ':') {
5974 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5975 char *t = s + UTF8SKIP(s);
5976 while (*t & 0x80 && is_utf8_mark((U8*)t))
5978 if (d + (t - s) > e)
5979 Perl_croak(aTHX_ ident_too_long);
5980 Copy(s, d, t - s, char);
5991 if (PL_lex_state != LEX_NORMAL)
5992 PL_lex_state = LEX_INTERPENDMAYBE;
5995 if (*s == '$' && s[1] &&
5996 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6009 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6014 if (isSPACE(s[-1])) {
6017 if (!SPACE_OR_TAB(ch)) {
6023 if (isIDFIRST_lazy_if(d,UTF)) {
6027 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6029 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
6032 Copy(s, d, e - s, char);
6037 while ((isALNUM(*s) || *s == ':') && d < e)
6040 Perl_croak(aTHX_ ident_too_long);
6043 while (s < send && SPACE_OR_TAB(*s)) s++;
6044 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6045 dTHR; /* only for ckWARN */
6046 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6047 const char *brack = *s == '[' ? "[...]" : "{...}";
6048 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6049 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6050 funny, dest, brack, funny, dest, brack);
6053 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6057 /* Handle extended ${^Foo} variables
6058 * 1999-02-27 mjd-perl-patch@plover.com */
6059 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6063 while (isALNUM(*s) && d < e) {
6067 Perl_croak(aTHX_ ident_too_long);
6072 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6073 PL_lex_state = LEX_INTERPEND;
6076 if (PL_lex_state == LEX_NORMAL) {
6077 dTHR; /* only for ckWARN */
6078 if (ckWARN(WARN_AMBIGUOUS) &&
6079 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6081 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6082 "Ambiguous use of %c{%s} resolved to %c%s",
6083 funny, dest, funny, dest);
6088 s = bracket; /* let the parser handle it */
6092 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6093 PL_lex_state = LEX_INTERPEND;
6098 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6103 *pmfl |= PMf_GLOBAL;
6105 *pmfl |= PMf_CONTINUE;
6109 *pmfl |= PMf_MULTILINE;
6111 *pmfl |= PMf_SINGLELINE;
6113 *pmfl |= PMf_EXTENDED;
6117 S_scan_pat(pTHX_ char *start, I32 type)
6122 s = scan_str(start,FALSE,FALSE);
6125 SvREFCNT_dec(PL_lex_stuff);
6126 PL_lex_stuff = Nullsv;
6127 Perl_croak(aTHX_ "Search pattern not terminated");
6130 pm = (PMOP*)newPMOP(type, 0);
6131 if (PL_multi_open == '?')
6132 pm->op_pmflags |= PMf_ONCE;
6134 while (*s && strchr("iomsx", *s))
6135 pmflag(&pm->op_pmflags,*s++);
6138 while (*s && strchr("iogcmsx", *s))
6139 pmflag(&pm->op_pmflags,*s++);
6141 pm->op_pmpermflags = pm->op_pmflags;
6143 PL_lex_op = (OP*)pm;
6144 yylval.ival = OP_MATCH;
6149 S_scan_subst(pTHX_ char *start)
6156 yylval.ival = OP_NULL;
6158 s = scan_str(start,FALSE,FALSE);
6162 SvREFCNT_dec(PL_lex_stuff);
6163 PL_lex_stuff = Nullsv;
6164 Perl_croak(aTHX_ "Substitution pattern not terminated");
6167 if (s[-1] == PL_multi_open)
6170 first_start = PL_multi_start;
6171 s = scan_str(s,FALSE,FALSE);
6174 SvREFCNT_dec(PL_lex_stuff);
6175 PL_lex_stuff = Nullsv;
6177 SvREFCNT_dec(PL_lex_repl);
6178 PL_lex_repl = Nullsv;
6179 Perl_croak(aTHX_ "Substitution replacement not terminated");
6181 PL_multi_start = first_start; /* so whole substitution is taken together */
6183 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6189 else if (strchr("iogcmsx", *s))
6190 pmflag(&pm->op_pmflags,*s++);
6197 PL_sublex_info.super_bufptr = s;
6198 PL_sublex_info.super_bufend = PL_bufend;
6200 pm->op_pmflags |= PMf_EVAL;
6201 repl = newSVpvn("",0);
6203 sv_catpv(repl, es ? "eval " : "do ");
6204 sv_catpvn(repl, "{ ", 2);
6205 sv_catsv(repl, PL_lex_repl);
6206 sv_catpvn(repl, " };", 2);
6208 SvREFCNT_dec(PL_lex_repl);
6212 pm->op_pmpermflags = pm->op_pmflags;
6213 PL_lex_op = (OP*)pm;
6214 yylval.ival = OP_SUBST;
6219 S_scan_trans(pTHX_ char *start)
6230 yylval.ival = OP_NULL;
6232 s = scan_str(start,FALSE,FALSE);
6235 SvREFCNT_dec(PL_lex_stuff);
6236 PL_lex_stuff = Nullsv;
6237 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6239 if (s[-1] == PL_multi_open)
6242 s = scan_str(s,FALSE,FALSE);
6245 SvREFCNT_dec(PL_lex_stuff);
6246 PL_lex_stuff = Nullsv;
6248 SvREFCNT_dec(PL_lex_repl);
6249 PL_lex_repl = Nullsv;
6250 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6253 New(803,tbl,256,short);
6254 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6256 complement = del = squash = 0;
6257 while (strchr("cds", *s)) {
6259 complement = OPpTRANS_COMPLEMENT;
6261 del = OPpTRANS_DELETE;
6263 squash = OPpTRANS_SQUASH;
6266 o->op_private = del|squash|complement;
6269 yylval.ival = OP_TRANS;
6274 S_scan_heredoc(pTHX_ register char *s)
6278 I32 op_type = OP_SCALAR;
6285 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6289 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6292 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6293 if (*peek && strchr("`'\"",*peek)) {
6296 s = delimcpy(d, e, s, PL_bufend, term, &len);
6306 if (!isALNUM_lazy_if(s,UTF))
6307 deprecate("bare << to mean <<\"\"");
6308 for (; isALNUM_lazy_if(s,UTF); s++) {
6313 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6314 Perl_croak(aTHX_ "Delimiter for here document is too long");
6317 len = d - PL_tokenbuf;
6318 #ifndef PERL_STRICT_CR
6319 d = strchr(s, '\r');
6323 while (s < PL_bufend) {
6329 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6338 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6343 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6344 herewas = newSVpvn(s,PL_bufend-s);
6346 s--, herewas = newSVpvn(s,d-s);
6347 s += SvCUR(herewas);
6349 tmpstr = NEWSV(87,79);
6350 sv_upgrade(tmpstr, SVt_PVIV);
6355 else if (term == '`') {
6356 op_type = OP_BACKTICK;
6357 SvIVX(tmpstr) = '\\';
6361 PL_multi_start = CopLINE(PL_curcop);
6362 PL_multi_open = PL_multi_close = '<';
6363 term = *PL_tokenbuf;
6364 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6365 char *bufptr = PL_sublex_info.super_bufptr;
6366 char *bufend = PL_sublex_info.super_bufend;
6367 char *olds = s - SvCUR(herewas);
6368 s = strchr(bufptr, '\n');
6372 while (s < bufend &&
6373 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6375 CopLINE_inc(PL_curcop);
6378 CopLINE_set(PL_curcop, PL_multi_start);
6379 missingterm(PL_tokenbuf);
6381 sv_setpvn(herewas,bufptr,d-bufptr+1);
6382 sv_setpvn(tmpstr,d+1,s-d);
6384 sv_catpvn(herewas,s,bufend-s);
6385 (void)strcpy(bufptr,SvPVX(herewas));
6392 while (s < PL_bufend &&
6393 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6395 CopLINE_inc(PL_curcop);
6397 if (s >= PL_bufend) {
6398 CopLINE_set(PL_curcop, PL_multi_start);
6399 missingterm(PL_tokenbuf);
6401 sv_setpvn(tmpstr,d+1,s-d);
6403 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6405 sv_catpvn(herewas,s,PL_bufend-s);
6406 sv_setsv(PL_linestr,herewas);
6407 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6408 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6411 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6412 while (s >= PL_bufend) { /* multiple line string? */
6414 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6415 CopLINE_set(PL_curcop, PL_multi_start);
6416 missingterm(PL_tokenbuf);
6418 CopLINE_inc(PL_curcop);
6419 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6420 #ifndef PERL_STRICT_CR
6421 if (PL_bufend - PL_linestart >= 2) {
6422 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6423 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6425 PL_bufend[-2] = '\n';
6427 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6429 else if (PL_bufend[-1] == '\r')
6430 PL_bufend[-1] = '\n';
6432 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6433 PL_bufend[-1] = '\n';
6435 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6436 SV *sv = NEWSV(88,0);
6438 sv_upgrade(sv, SVt_PVMG);
6439 sv_setsv(sv,PL_linestr);
6440 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6442 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6445 sv_catsv(PL_linestr,herewas);
6446 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6450 sv_catsv(tmpstr,PL_linestr);
6455 PL_multi_end = CopLINE(PL_curcop);
6456 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6457 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6458 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6460 SvREFCNT_dec(herewas);
6461 PL_lex_stuff = tmpstr;
6462 yylval.ival = op_type;
6467 takes: current position in input buffer
6468 returns: new position in input buffer
6469 side-effects: yylval and lex_op are set.
6474 <FH> read from filehandle
6475 <pkg::FH> read from package qualified filehandle
6476 <pkg'FH> read from package qualified filehandle
6477 <$fh> read from filehandle in $fh
6483 S_scan_inputsymbol(pTHX_ char *start)
6485 register char *s = start; /* current position in buffer */
6491 d = PL_tokenbuf; /* start of temp holding space */
6492 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6493 end = strchr(s, '\n');
6496 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6498 /* die if we didn't have space for the contents of the <>,
6499 or if it didn't end, or if we see a newline
6502 if (len >= sizeof PL_tokenbuf)
6503 Perl_croak(aTHX_ "Excessively long <> operator");
6505 Perl_croak(aTHX_ "Unterminated <> operator");
6510 Remember, only scalar variables are interpreted as filehandles by
6511 this code. Anything more complex (e.g., <$fh{$num}>) will be
6512 treated as a glob() call.
6513 This code makes use of the fact that except for the $ at the front,
6514 a scalar variable and a filehandle look the same.
6516 if (*d == '$' && d[1]) d++;
6518 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6519 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6522 /* If we've tried to read what we allow filehandles to look like, and
6523 there's still text left, then it must be a glob() and not a getline.
6524 Use scan_str to pull out the stuff between the <> and treat it
6525 as nothing more than a string.
6528 if (d - PL_tokenbuf != len) {
6529 yylval.ival = OP_GLOB;
6531 s = scan_str(start,FALSE,FALSE);
6533 Perl_croak(aTHX_ "Glob not terminated");
6537 /* we're in a filehandle read situation */
6540 /* turn <> into <ARGV> */
6542 (void)strcpy(d,"ARGV");
6544 /* if <$fh>, create the ops to turn the variable into a
6550 /* try to find it in the pad for this block, otherwise find
6551 add symbol table ops
6553 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6554 OP *o = newOP(OP_PADSV, 0);
6556 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6559 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6560 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6561 newUNOP(OP_RV2SV, 0,
6562 newGVOP(OP_GV, 0, gv)));
6564 PL_lex_op->op_flags |= OPf_SPECIAL;
6565 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6566 yylval.ival = OP_NULL;
6569 /* If it's none of the above, it must be a literal filehandle
6570 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6572 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6573 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6574 yylval.ival = OP_NULL;
6583 takes: start position in buffer
6584 keep_quoted preserve \ on the embedded delimiter(s)
6585 keep_delims preserve the delimiters around the string
6586 returns: position to continue reading from buffer
6587 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6588 updates the read buffer.
6590 This subroutine pulls a string out of the input. It is called for:
6591 q single quotes q(literal text)
6592 ' single quotes 'literal text'
6593 qq double quotes qq(interpolate $here please)
6594 " double quotes "interpolate $here please"
6595 qx backticks qx(/bin/ls -l)
6596 ` backticks `/bin/ls -l`
6597 qw quote words @EXPORT_OK = qw( func() $spam )
6598 m// regexp match m/this/
6599 s/// regexp substitute s/this/that/
6600 tr/// string transliterate tr/this/that/
6601 y/// string transliterate y/this/that/
6602 ($*@) sub prototypes sub foo ($)
6603 (stuff) sub attr parameters sub foo : attr(stuff)
6604 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6606 In most of these cases (all but <>, patterns and transliterate)
6607 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6608 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6609 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6612 It skips whitespace before the string starts, and treats the first
6613 character as the delimiter. If the delimiter is one of ([{< then
6614 the corresponding "close" character )]}> is used as the closing
6615 delimiter. It allows quoting of delimiters, and if the string has
6616 balanced delimiters ([{<>}]) it allows nesting.
6618 The lexer always reads these strings into lex_stuff, except in the
6619 case of the operators which take *two* arguments (s/// and tr///)
6620 when it checks to see if lex_stuff is full (presumably with the 1st
6621 arg to s or tr) and if so puts the string into lex_repl.
6626 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6629 SV *sv; /* scalar value: string */
6630 char *tmps; /* temp string, used for delimiter matching */
6631 register char *s = start; /* current position in the buffer */
6632 register char term; /* terminating character */
6633 register char *to; /* current position in the sv's data */
6634 I32 brackets = 1; /* bracket nesting level */
6635 bool has_utf8 = FALSE; /* is there any utf8 content? */
6637 /* skip space before the delimiter */
6641 /* mark where we are, in case we need to report errors */
6644 /* after skipping whitespace, the next character is the terminator */
6646 if ((term & 0x80) && UTF)
6649 /* mark where we are */
6650 PL_multi_start = CopLINE(PL_curcop);
6651 PL_multi_open = term;
6653 /* find corresponding closing delimiter */
6654 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6656 PL_multi_close = term;
6658 /* create a new SV to hold the contents. 87 is leak category, I'm
6659 assuming. 79 is the SV's initial length. What a random number. */
6661 sv_upgrade(sv, SVt_PVIV);
6663 (void)SvPOK_only(sv); /* validate pointer */
6665 /* move past delimiter and try to read a complete string */
6667 sv_catpvn(sv, s, 1);
6670 /* extend sv if need be */
6671 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6672 /* set 'to' to the next character in the sv's string */
6673 to = SvPVX(sv)+SvCUR(sv);
6675 /* if open delimiter is the close delimiter read unbridle */
6676 if (PL_multi_open == PL_multi_close) {
6677 for (; s < PL_bufend; s++,to++) {
6678 /* embedded newlines increment the current line number */
6679 if (*s == '\n' && !PL_rsfp)
6680 CopLINE_inc(PL_curcop);
6681 /* handle quoted delimiters */
6682 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6683 if (!keep_quoted && s[1] == term)
6685 /* any other quotes are simply copied straight through */
6689 /* terminate when run out of buffer (the for() condition), or
6690 have found the terminator */
6691 else if (*s == term)
6693 else if (!has_utf8 && (*s & 0x80) && UTF)
6699 /* if the terminator isn't the same as the start character (e.g.,
6700 matched brackets), we have to allow more in the quoting, and
6701 be prepared for nested brackets.
6704 /* read until we run out of string, or we find the terminator */
6705 for (; s < PL_bufend; s++,to++) {
6706 /* embedded newlines increment the line count */
6707 if (*s == '\n' && !PL_rsfp)
6708 CopLINE_inc(PL_curcop);
6709 /* backslashes can escape the open or closing characters */
6710 if (*s == '\\' && s+1 < PL_bufend) {
6712 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6717 /* allow nested opens and closes */
6718 else if (*s == PL_multi_close && --brackets <= 0)
6720 else if (*s == PL_multi_open)
6722 else if (!has_utf8 && (*s & 0x80) && UTF)
6727 /* terminate the copied string and update the sv's end-of-string */
6729 SvCUR_set(sv, to - SvPVX(sv));
6732 * this next chunk reads more into the buffer if we're not done yet
6736 break; /* handle case where we are done yet :-) */
6738 #ifndef PERL_STRICT_CR
6739 if (to - SvPVX(sv) >= 2) {
6740 if ((to[-2] == '\r' && to[-1] == '\n') ||
6741 (to[-2] == '\n' && to[-1] == '\r'))
6745 SvCUR_set(sv, to - SvPVX(sv));
6747 else if (to[-1] == '\r')
6750 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6754 /* if we're out of file, or a read fails, bail and reset the current
6755 line marker so we can report where the unterminated string began
6758 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6760 CopLINE_set(PL_curcop, PL_multi_start);
6763 /* we read a line, so increment our line counter */
6764 CopLINE_inc(PL_curcop);
6766 /* update debugger info */
6767 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6768 SV *sv = NEWSV(88,0);
6770 sv_upgrade(sv, SVt_PVMG);
6771 sv_setsv(sv,PL_linestr);
6772 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6775 /* having changed the buffer, we must update PL_bufend */
6776 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6779 /* at this point, we have successfully read the delimited string */
6782 sv_catpvn(sv, s, 1);
6785 PL_multi_end = CopLINE(PL_curcop);
6788 /* if we allocated too much space, give some back */
6789 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6790 SvLEN_set(sv, SvCUR(sv) + 1);
6791 Renew(SvPVX(sv), SvLEN(sv), char);
6794 /* decide whether this is the first or second quoted string we've read
6807 takes: pointer to position in buffer
6808 returns: pointer to new position in buffer
6809 side-effects: builds ops for the constant in yylval.op
6811 Read a number in any of the formats that Perl accepts:
6813 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6814 [\d_]+(\.[\d_]*)?[Ee](\d+)
6816 Underbars (_) are allowed in decimal numbers. If -w is on,
6817 underbars before a decimal point must be at three digit intervals.
6819 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6822 If it reads a number without a decimal point or an exponent, it will
6823 try converting the number to an integer and see if it can do so
6824 without loss of precision.
6828 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6830 register char *s = start; /* current position in buffer */
6831 register char *d; /* destination in temp buffer */
6832 register char *e; /* end of temp buffer */
6833 NV nv; /* number read, as a double */
6834 SV *sv = Nullsv; /* place to put the converted number */
6835 bool floatit; /* boolean: int or float? */
6836 char *lastub = 0; /* position of last underbar */
6837 static char number_too_long[] = "Number too long";
6839 /* We use the first character to decide what type of number this is */
6843 Perl_croak(aTHX_ "panic: scan_num");
6845 /* if it starts with a 0, it could be an octal number, a decimal in
6846 0.13 disguise, or a hexadecimal number, or a binary number. */
6850 u holds the "number so far"
6851 shift the power of 2 of the base
6852 (hex == 4, octal == 3, binary == 1)
6853 overflowed was the number more than we can hold?
6855 Shift is used when we add a digit. It also serves as an "are
6856 we in octal/hex/binary?" indicator to disallow hex characters
6863 bool overflowed = FALSE;
6864 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6865 static char* bases[5] = { "", "binary", "", "octal",
6867 static char* Bases[5] = { "", "Binary", "", "Octal",
6869 static char *maxima[5] = { "",
6870 "0b11111111111111111111111111111111",
6874 char *base, *Base, *max;
6880 } else if (s[1] == 'b') {
6884 /* check for a decimal in disguise */
6885 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6887 /* so it must be octal */
6891 base = bases[shift];
6892 Base = Bases[shift];
6893 max = maxima[shift];
6895 /* read the rest of the number */
6897 /* x is used in the overflow test,
6898 b is the digit we're adding on. */
6903 /* if we don't mention it, we're done */
6912 /* 8 and 9 are not octal */
6915 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6919 case '2': case '3': case '4':
6920 case '5': case '6': case '7':
6922 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6926 b = *s++ & 15; /* ASCII digit -> value of digit */
6930 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6931 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6932 /* make sure they said 0x */
6937 /* Prepare to put the digit we have onto the end
6938 of the number so far. We check for overflows.
6943 x = u << shift; /* make room for the digit */
6945 if ((x >> shift) != u
6946 && !(PL_hints & HINT_NEW_BINARY)) {
6950 if (ckWARN_d(WARN_OVERFLOW))
6951 Perl_warner(aTHX_ WARN_OVERFLOW,
6952 "Integer overflow in %s number",
6955 u = x | b; /* add the digit to the end */
6958 n *= nvshift[shift];
6959 /* If an NV has not enough bits in its
6960 * mantissa to represent an UV this summing of
6961 * small low-order numbers is a waste of time
6962 * (because the NV cannot preserve the
6963 * low-order bits anyway): we could just
6964 * remember when did we overflow and in the
6965 * end just multiply n by the right
6973 /* if we get here, we had success: make a scalar value from
6980 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6981 Perl_warner(aTHX_ WARN_PORTABLE,
6982 "%s number > %s non-portable",
6989 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6990 Perl_warner(aTHX_ WARN_PORTABLE,
6991 "%s number > %s non-portable",
6996 if (PL_hints & HINT_NEW_BINARY)
6997 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7002 handle decimal numbers.
7003 we're also sent here when we read a 0 as the first digit
7005 case '1': case '2': case '3': case '4': case '5':
7006 case '6': case '7': case '8': case '9': case '.':
7009 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7012 /* read next group of digits and _ and copy into d */
7013 while (isDIGIT(*s) || *s == '_') {
7014 /* skip underscores, checking for misplaced ones
7018 dTHR; /* only for ckWARN */
7019 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7020 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7024 /* check for end of fixed-length buffer */
7026 Perl_croak(aTHX_ number_too_long);
7027 /* if we're ok, copy the character */
7032 /* final misplaced underbar check */
7033 if (lastub && s - lastub != 3) {
7035 if (ckWARN(WARN_SYNTAX))
7036 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7039 /* read a decimal portion if there is one. avoid
7040 3..5 being interpreted as the number 3. followed
7043 if (*s == '.' && s[1] != '.') {
7047 /* copy, ignoring underbars, until we run out of
7048 digits. Note: no misplaced underbar checks!
7050 for (; isDIGIT(*s) || *s == '_'; s++) {
7051 /* fixed length buffer check */
7053 Perl_croak(aTHX_ number_too_long);
7057 if (*s == '.' && isDIGIT(s[1])) {
7058 /* oops, it's really a v-string, but without the "v" */
7064 /* read exponent part, if present */
7065 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7069 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7070 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7072 /* allow positive or negative exponent */
7073 if (*s == '+' || *s == '-')
7076 /* read digits of exponent (no underbars :-) */
7077 while (isDIGIT(*s)) {
7079 Perl_croak(aTHX_ number_too_long);
7084 /* terminate the string */
7087 /* make an sv from the string */
7090 #if defined(Strtol) && defined(Strtoul)
7093 strtol/strtoll sets errno to ERANGE if the number is too big
7094 for an integer. We try to do an integer conversion first
7095 if no characters indicating "float" have been found.
7102 if (*PL_tokenbuf == '-')
7103 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7105 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7107 floatit = TRUE; /* Probably just too large. */
7108 else if (*PL_tokenbuf == '-')
7110 else if (uv <= IV_MAX)
7111 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7116 nv = Atof(PL_tokenbuf);
7121 No working strtou?ll?.
7123 Unfortunately atol() doesn't do range checks (returning
7124 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7125 everywhere [1], so we cannot use use atol() (or atoll()).
7126 If we could, they would be used, as Atol(), very much like
7127 Strtol() and Strtoul() are used above.
7129 [1] XXX Configure test needed to check for atol()
7130 (and atoll()) overflow behaviour XXX
7134 We need to do this the hard way. */
7136 nv = Atof(PL_tokenbuf);
7138 /* See if we can make do with an integer value without loss of
7139 precision. We use U_V to cast to a UV, because some
7140 compilers have issues. Then we try casting it back and see
7141 if it was the same [1]. We only do this if we know we
7142 specifically read an integer. If floatit is true, then we
7143 don't need to do the conversion at all.
7145 [1] Note that this is lossy if our NVs cannot preserve our
7146 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7147 and NV_PRESERVES_UV_BITS (a number), but in general we really
7148 do hope all such potentially lossy platforms have strtou?ll?
7149 to do a lossless IV/UV conversion.
7151 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7152 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7153 as NV_DIG and NV_MANT_DIG)?
7159 if (!floatit && (NV)uv == nv) {
7161 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7169 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7170 (PL_hints & HINT_NEW_INTEGER) )
7171 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7172 (floatit ? "float" : "integer"),
7176 /* if it starts with a v, it could be a v-string */
7182 while (isDIGIT(*pos) || *pos == '_')
7184 if (!isALPHA(*pos)) {
7186 U8 tmpbuf[UTF8_MAXLEN+1];
7189 s++; /* get past 'v' */
7192 sv_setpvn(sv, "", 0);
7195 if (*s == '0' && isDIGIT(s[1]))
7196 yyerror("Octal number in vector unsupported");
7199 /* this is atoi() that tolerates underscores */
7202 while (--end >= s) {
7207 rev += (*end - '0') * mult;
7209 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7210 Perl_warner(aTHX_ WARN_OVERFLOW,
7211 "Integer overflow in decimal number");
7214 tmpend = uv_to_utf8(tmpbuf, rev);
7215 utf8 = utf8 || rev > 127;
7216 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7217 if (*pos == '.' && isDIGIT(pos[1]))
7223 while (isDIGIT(*pos) || *pos == '_')
7231 sv_utf8_downgrade(sv, TRUE);
7238 /* make the op for the constant and return */
7241 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7243 lvalp->opval = Nullop;
7249 S_scan_formline(pTHX_ register char *s)
7254 SV *stuff = newSVpvn("",0);
7255 bool needargs = FALSE;
7258 if (*s == '.' || *s == /*{*/'}') {
7260 #ifdef PERL_STRICT_CR
7261 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7263 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7265 if (*t == '\n' || t == PL_bufend)
7268 if (PL_in_eval && !PL_rsfp) {
7269 eol = strchr(s,'\n');
7274 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7276 for (t = s; t < eol; t++) {
7277 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7279 goto enough; /* ~~ must be first line in formline */
7281 if (*t == '@' || *t == '^')
7284 sv_catpvn(stuff, s, eol-s);
7285 #ifndef PERL_STRICT_CR
7286 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7287 char *end = SvPVX(stuff) + SvCUR(stuff);
7296 s = filter_gets(PL_linestr, PL_rsfp, 0);
7297 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7298 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7301 yyerror("Format not terminated");
7311 PL_lex_state = LEX_NORMAL;
7312 PL_nextval[PL_nexttoke].ival = 0;
7316 PL_lex_state = LEX_FORMLINE;
7317 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7319 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7323 SvREFCNT_dec(stuff);
7324 PL_lex_formbrack = 0;
7335 PL_cshlen = strlen(PL_cshname);
7340 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7343 I32 oldsavestack_ix = PL_savestack_ix;
7344 CV* outsidecv = PL_compcv;
7348 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7350 SAVEI32(PL_subline);
7351 save_item(PL_subname);
7354 SAVESPTR(PL_comppad_name);
7355 SAVESPTR(PL_compcv);
7356 SAVEI32(PL_comppad_name_fill);
7357 SAVEI32(PL_min_intro_pending);
7358 SAVEI32(PL_max_intro_pending);
7359 SAVEI32(PL_pad_reset_pending);
7361 PL_compcv = (CV*)NEWSV(1104,0);
7362 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7363 CvFLAGS(PL_compcv) |= flags;
7365 PL_comppad = newAV();
7366 av_push(PL_comppad, Nullsv);
7367 PL_curpad = AvARRAY(PL_comppad);
7368 PL_comppad_name = newAV();
7369 PL_comppad_name_fill = 0;
7370 PL_min_intro_pending = 0;
7372 PL_subline = CopLINE(PL_curcop);
7374 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7375 PL_curpad[0] = (SV*)newAV();
7376 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7377 #endif /* USE_THREADS */
7379 comppadlist = newAV();
7380 AvREAL_off(comppadlist);
7381 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7382 av_store(comppadlist, 1, (SV*)PL_comppad);
7384 CvPADLIST(PL_compcv) = comppadlist;
7385 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7387 CvOWNER(PL_compcv) = 0;
7388 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7389 MUTEX_INIT(CvMUTEXP(PL_compcv));
7390 #endif /* USE_THREADS */
7392 return oldsavestack_ix;
7396 Perl_yywarn(pTHX_ char *s)
7399 PL_in_eval |= EVAL_WARNONLY;
7401 PL_in_eval &= ~EVAL_WARNONLY;
7406 Perl_yyerror(pTHX_ char *s)
7410 char *context = NULL;
7414 if (!yychar || (yychar == ';' && !PL_rsfp))
7416 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7417 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7418 while (isSPACE(*PL_oldoldbufptr))
7420 context = PL_oldoldbufptr;
7421 contlen = PL_bufptr - PL_oldoldbufptr;
7423 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7424 PL_oldbufptr != PL_bufptr) {
7425 while (isSPACE(*PL_oldbufptr))
7427 context = PL_oldbufptr;
7428 contlen = PL_bufptr - PL_oldbufptr;
7430 else if (yychar > 255)
7431 where = "next token ???";
7432 #ifdef USE_PURE_BISON
7433 /* GNU Bison sets the value -2 */
7434 else if (yychar == -2) {
7436 else if ((yychar & 127) == 127) {
7438 if (PL_lex_state == LEX_NORMAL ||
7439 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7440 where = "at end of line";
7441 else if (PL_lex_inpat)
7442 where = "within pattern";
7444 where = "within string";
7447 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7449 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7450 else if (isPRINT_LC(yychar))
7451 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7453 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7454 where = SvPVX(where_sv);
7456 msg = sv_2mortal(newSVpv(s, 0));
7457 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7458 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7460 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7462 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7463 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7464 Perl_sv_catpvf(aTHX_ msg,
7465 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7466 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7469 if (PL_in_eval & EVAL_WARNONLY)
7470 Perl_warn(aTHX_ "%"SVf, msg);
7473 if (PL_error_count >= 10) {
7474 if (PL_in_eval && SvCUR(ERRSV))
7475 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7476 ERRSV, CopFILE(PL_curcop));
7478 Perl_croak(aTHX_ "%s has too many errors.\n",
7479 CopFILE(PL_curcop));
7482 PL_in_my_stash = Nullhv;
7487 S_swallow_bom(pTHX_ U8 *s)
7490 slen = SvCUR(PL_linestr);
7494 /* UTF-16 little-endian */
7495 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7496 Perl_croak(aTHX_ "Unsupported script encoding");
7497 #ifndef PERL_NO_UTF16_FILTER
7498 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7500 if (PL_bufend > (char*)s) {
7504 filter_add(utf16rev_textfilter, NULL);
7505 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7506 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7507 PL_bufend - (char*)s - 1,
7509 Copy(news, s, newlen, U8);
7510 SvCUR_set(PL_linestr, newlen);
7511 PL_bufend = SvPVX(PL_linestr) + newlen;
7512 news[newlen++] = '\0';
7516 Perl_croak(aTHX_ "Unsupported script encoding");
7521 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7522 #ifndef PERL_NO_UTF16_FILTER
7523 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7525 if (PL_bufend > (char *)s) {
7529 filter_add(utf16_textfilter, NULL);
7530 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7531 PL_bufend = (char*)utf16_to_utf8(s, news,
7532 PL_bufend - (char*)s,
7534 Copy(news, s, newlen, U8);
7535 SvCUR_set(PL_linestr, newlen);
7536 PL_bufend = SvPVX(PL_linestr) + newlen;
7537 news[newlen++] = '\0';
7541 Perl_croak(aTHX_ "Unsupported script encoding");
7546 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7547 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7552 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7553 s[2] == 0xFE && s[3] == 0xFF)
7555 Perl_croak(aTHX_ "Unsupported script encoding");
7567 * Restore a source filter.
7571 restore_rsfp(pTHXo_ void *f)
7573 PerlIO *fp = (PerlIO*)f;
7575 if (PL_rsfp == PerlIO_stdin())
7576 PerlIO_clearerr(PL_rsfp);
7577 else if (PL_rsfp && (PL_rsfp != fp))
7578 PerlIO_close(PL_rsfp);
7582 #ifndef PERL_NO_UTF16_FILTER
7584 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7586 I32 count = FILTER_READ(idx+1, sv, maxlen);
7591 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7592 if (!*SvPV_nolen(sv))
7593 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7596 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7597 sv_usepvn(sv, (char*)tmps, tend - tmps);
7603 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7605 I32 count = FILTER_READ(idx+1, sv, maxlen);
7610 if (!*SvPV_nolen(sv))
7611 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7614 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7615 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7616 sv_usepvn(sv, (char*)tmps, tend - tmps);