3 * Copyright (c) 1991-2001, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
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)
277 if (ckWARN(WARN_DEPRECATED))
278 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
283 * Deprecate a comma-less variable list.
289 deprecate("comma-less variable list");
293 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
294 * utf16-to-utf8-reversed.
297 #ifdef PERL_CR_FILTER
301 register char *s = SvPVX(sv);
302 register char *e = s + SvCUR(sv);
303 /* outer loop optimized to do nothing if there are no CR-LFs */
305 if (*s++ == '\r' && *s == '\n') {
306 /* hit a CR-LF, need to copy the rest */
307 register char *d = s - 1;
310 if (*s == '\r' && s[1] == '\n')
321 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
323 I32 count = FILTER_READ(idx+1, sv, maxlen);
324 if (count > 0 && !maxlen)
332 * Initialize variables. Uses the Perl save_stack to save its state (for
333 * recursive calls to the parser).
337 Perl_lex_start(pTHX_ SV *line)
342 SAVEI32(PL_lex_dojoin);
343 SAVEI32(PL_lex_brackets);
344 SAVEI32(PL_lex_casemods);
345 SAVEI32(PL_lex_starts);
346 SAVEI32(PL_lex_state);
347 SAVEVPTR(PL_lex_inpat);
348 SAVEI32(PL_lex_inwhat);
349 if (PL_lex_state == LEX_KNOWNEXT) {
350 I32 toke = PL_nexttoke;
351 while (--toke >= 0) {
352 SAVEI32(PL_nexttype[toke]);
353 SAVEVPTR(PL_nextval[toke]);
355 SAVEI32(PL_nexttoke);
357 SAVECOPLINE(PL_curcop);
360 SAVEPPTR(PL_oldbufptr);
361 SAVEPPTR(PL_oldoldbufptr);
362 SAVEPPTR(PL_linestart);
363 SAVESPTR(PL_linestr);
364 SAVEPPTR(PL_lex_brackstack);
365 SAVEPPTR(PL_lex_casestack);
366 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
367 SAVESPTR(PL_lex_stuff);
368 SAVEI32(PL_lex_defer);
369 SAVEI32(PL_sublex_info.sub_inwhat);
370 SAVESPTR(PL_lex_repl);
372 SAVEINT(PL_lex_expect);
374 PL_lex_state = LEX_NORMAL;
378 New(899, PL_lex_brackstack, 120, char);
379 New(899, PL_lex_casestack, 12, char);
380 SAVEFREEPV(PL_lex_brackstack);
381 SAVEFREEPV(PL_lex_casestack);
383 *PL_lex_casestack = '\0';
386 PL_lex_stuff = Nullsv;
387 PL_lex_repl = Nullsv;
391 PL_sublex_info.sub_inwhat = 0;
393 if (SvREADONLY(PL_linestr))
394 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
395 s = SvPV(PL_linestr, len);
396 if (len && s[len-1] != ';') {
397 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
398 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
399 sv_catpvn(PL_linestr, "\n;", 2);
401 SvTEMP_off(PL_linestr);
402 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
403 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
405 PL_rs = newSVpvn("\n", 1);
411 * Finalizer for lexing operations. Must be called when the parser is
412 * done with the lexer.
418 PL_doextract = FALSE;
423 * This subroutine has nothing to do with tilting, whether at windmills
424 * or pinball tables. Its name is short for "increment line". It
425 * increments the current line number in CopLINE(PL_curcop) and checks
426 * to see whether the line starts with a comment of the form
427 * # line 500 "foo.pm"
428 * If so, it sets the current line number and file to the values in the comment.
432 S_incline(pTHX_ char *s)
439 CopLINE_inc(PL_curcop);
442 while (SPACE_OR_TAB(*s)) s++;
443 if (strnEQ(s, "line", 4))
447 if (*s == ' ' || *s == '\t')
451 while (SPACE_OR_TAB(*s)) s++;
457 while (SPACE_OR_TAB(*s))
459 if (*s == '"' && (t = strchr(s+1, '"'))) {
464 for (t = s; !isSPACE(*t); t++) ;
467 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
469 if (*e != '\n' && *e != '\0')
470 return; /* false alarm */
476 Safefree(CopFILE(PL_curcop));
478 SvREFCNT_dec(CopFILEGV(PL_curcop));
480 CopFILE_set(PL_curcop, s);
483 CopLINE_set(PL_curcop, atoi(n)-1);
488 * Called to gobble the appropriate amount and type of whitespace.
489 * Skips comments as well.
493 S_skipspace(pTHX_ register char *s)
495 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
496 while (s < PL_bufend && SPACE_OR_TAB(*s))
502 SSize_t oldprevlen, oldoldprevlen;
503 SSize_t oldloplen, oldunilen;
504 while (s < PL_bufend && isSPACE(*s)) {
505 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
510 if (s < PL_bufend && *s == '#') {
511 while (s < PL_bufend && *s != '\n')
515 if (PL_in_eval && !PL_rsfp) {
522 /* only continue to recharge the buffer if we're at the end
523 * of the buffer, we're not reading from a source filter, and
524 * we're in normal lexing mode
526 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
527 PL_lex_state == LEX_FORMLINE)
530 /* try to recharge the buffer */
531 if ((s = filter_gets(PL_linestr, PL_rsfp,
532 (prevlen = SvCUR(PL_linestr)))) == Nullch)
534 /* end of file. Add on the -p or -n magic */
535 if (PL_minus_n || PL_minus_p) {
536 sv_setpv(PL_linestr,PL_minus_p ?
537 ";}continue{print or die qq(-p destination: $!\\n)" :
539 sv_catpv(PL_linestr,";}");
540 PL_minus_n = PL_minus_p = 0;
543 sv_setpv(PL_linestr,";");
545 /* reset variables for next time we lex */
546 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
548 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
550 /* Close the filehandle. Could be from -P preprocessor,
551 * STDIN, or a regular file. If we were reading code from
552 * STDIN (because the commandline held no -e or filename)
553 * then we don't close it, we reset it so the code can
554 * read from STDIN too.
557 if (PL_preprocess && !PL_in_eval)
558 (void)PerlProc_pclose(PL_rsfp);
559 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
560 PerlIO_clearerr(PL_rsfp);
562 (void)PerlIO_close(PL_rsfp);
567 /* not at end of file, so we only read another line */
568 /* make corresponding updates to old pointers, for yyerror() */
569 oldprevlen = PL_oldbufptr - PL_bufend;
570 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
572 oldunilen = PL_last_uni - PL_bufend;
574 oldloplen = PL_last_lop - PL_bufend;
575 PL_linestart = PL_bufptr = s + prevlen;
576 PL_bufend = s + SvCUR(PL_linestr);
578 PL_oldbufptr = s + oldprevlen;
579 PL_oldoldbufptr = s + oldoldprevlen;
581 PL_last_uni = s + oldunilen;
583 PL_last_lop = s + oldloplen;
586 /* debugger active and we're not compiling the debugger code,
587 * so store the line into the debugger's array of lines
589 if (PERLDB_LINE && PL_curstash != PL_debstash) {
590 SV *sv = NEWSV(85,0);
592 sv_upgrade(sv, SVt_PVMG);
593 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
594 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
601 * Check the unary operators to ensure there's no ambiguity in how they're
602 * used. An ambiguous piece of code would be:
604 * This doesn't mean rand() + 5. Because rand() is a unary operator,
605 * the +5 is its argument.
614 if (PL_oldoldbufptr != PL_last_uni)
616 while (isSPACE(*PL_last_uni))
618 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
619 if ((t = strchr(s, '(')) && t < PL_bufptr)
621 if (ckWARN_d(WARN_AMBIGUOUS)){
624 Perl_warner(aTHX_ WARN_AMBIGUOUS,
625 "Warning: Use of \"%s\" without parens is ambiguous",
631 /* workaround to replace the UNI() macro with a function. Only the
632 * hints/uts.sh file mentions this. Other comments elsewhere in the
633 * source indicate Microport Unix might need it too.
639 #define UNI(f) return uni(f,s)
642 S_uni(pTHX_ I32 f, char *s)
647 PL_last_uni = PL_oldbufptr;
658 #endif /* CRIPPLED_CC */
661 * LOP : macro to build a list operator. Its behaviour has been replaced
662 * with a subroutine, S_lop() for which LOP is just another name.
665 #define LOP(f,x) return lop(f,x,s)
669 * Build a list operator (or something that might be one). The rules:
670 * - if we have a next token, then it's a list operator [why?]
671 * - if the next thing is an opening paren, then it's a function
672 * - else it's a list operator
676 S_lop(pTHX_ I32 f, int x, char *s)
682 PL_last_lop = PL_oldbufptr;
697 * When the lexer realizes it knows the next token (for instance,
698 * it is reordering tokens for the parser) then it can call S_force_next
699 * to know what token to return the next time the lexer is called. Caller
700 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
701 * handles the token correctly.
705 S_force_next(pTHX_ I32 type)
707 PL_nexttype[PL_nexttoke] = type;
709 if (PL_lex_state != LEX_KNOWNEXT) {
710 PL_lex_defer = PL_lex_state;
711 PL_lex_expect = PL_expect;
712 PL_lex_state = LEX_KNOWNEXT;
718 * When the lexer knows the next thing is a word (for instance, it has
719 * just seen -> and it knows that the next char is a word char, then
720 * it calls S_force_word to stick the next word into the PL_next lookahead.
723 * char *start : buffer position (must be within PL_linestr)
724 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
725 * int check_keyword : if true, Perl checks to make sure the word isn't
726 * a keyword (do this if the word is a label, e.g. goto FOO)
727 * int allow_pack : if true, : characters will also be allowed (require,
729 * int allow_initial_tick : used by the "sub" lexer only.
733 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
738 start = skipspace(start);
740 if (isIDFIRST_lazy_if(s,UTF) ||
741 (allow_pack && *s == ':') ||
742 (allow_initial_tick && *s == '\'') )
744 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
745 if (check_keyword && keyword(PL_tokenbuf, len))
747 if (token == METHOD) {
752 PL_expect = XOPERATOR;
755 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
756 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
764 * Called when the lexer wants $foo *foo &foo etc, but the program
765 * text only contains the "foo" portion. The first argument is a pointer
766 * to the "foo", and the second argument is the type symbol to prefix.
767 * Forces the next token to be a "WORD".
768 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
772 S_force_ident(pTHX_ register char *s, int kind)
775 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
776 PL_nextval[PL_nexttoke].opval = o;
779 o->op_private = OPpCONST_ENTERED;
780 /* XXX see note in pp_entereval() for why we forgo typo
781 warnings if the symbol must be introduced in an eval.
783 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
784 kind == '$' ? SVt_PV :
785 kind == '@' ? SVt_PVAV :
786 kind == '%' ? SVt_PVHV :
794 Perl_str_to_version(pTHX_ SV *sv)
799 char *start = SvPVx(sv,len);
800 bool utf = SvUTF8(sv) ? TRUE : FALSE;
801 char *end = start + len;
802 while (start < end) {
806 n = utf8_to_uv((U8*)start, len, &skip, 0);
811 retval += ((NV)n)/nshift;
820 * Forces the next token to be a version number.
824 S_force_version(pTHX_ char *s)
826 OP *version = Nullop;
835 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
836 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
838 s = scan_num(s, &yylval);
839 version = yylval.opval;
840 ver = cSVOPx(version)->op_sv;
841 if (SvPOK(ver) && !SvNIOK(ver)) {
842 (void)SvUPGRADE(ver, SVt_PVNV);
843 SvNVX(ver) = str_to_version(ver);
844 SvNOK_on(ver); /* hint that it is a version */
849 /* NOTE: The parser sees the package name and the VERSION swapped */
850 PL_nextval[PL_nexttoke].opval = version;
858 * Tokenize a quoted string passed in as an SV. It finds the next
859 * chunk, up to end of string or a backslash. It may make a new
860 * SV containing that chunk (if HINT_NEW_STRING is on). It also
865 S_tokeq(pTHX_ SV *sv)
876 s = SvPV_force(sv, len);
877 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
880 while (s < send && *s != '\\')
885 if ( PL_hints & HINT_NEW_STRING )
886 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
889 if (s + 1 < send && (s[1] == '\\'))
890 s++; /* all that, just for this */
895 SvCUR_set(sv, d - SvPVX(sv));
897 if ( PL_hints & HINT_NEW_STRING )
898 return new_constant(NULL, 0, "q", sv, pv, "q");
903 * Now come three functions related to double-quote context,
904 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
905 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
906 * interact with PL_lex_state, and create fake ( ... ) argument lists
907 * to handle functions and concatenation.
908 * They assume that whoever calls them will be setting up a fake
909 * join call, because each subthing puts a ',' after it. This lets
912 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
914 * (I'm not sure whether the spurious commas at the end of lcfirst's
915 * arguments and join's arguments are created or not).
920 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
922 * Pattern matching will set PL_lex_op to the pattern-matching op to
923 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
925 * OP_CONST and OP_READLINE are easy--just make the new op and return.
927 * Everything else becomes a FUNC.
929 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
930 * had an OP_CONST or OP_READLINE). This just sets us up for a
931 * call to S_sublex_push().
937 register I32 op_type = yylval.ival;
939 if (op_type == OP_NULL) {
940 yylval.opval = PL_lex_op;
944 if (op_type == OP_CONST || op_type == OP_READLINE) {
945 SV *sv = tokeq(PL_lex_stuff);
947 if (SvTYPE(sv) == SVt_PVIV) {
948 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
954 nsv = newSVpvn(p, len);
960 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
961 PL_lex_stuff = Nullsv;
965 PL_sublex_info.super_state = PL_lex_state;
966 PL_sublex_info.sub_inwhat = op_type;
967 PL_sublex_info.sub_op = PL_lex_op;
968 PL_lex_state = LEX_INTERPPUSH;
972 yylval.opval = PL_lex_op;
982 * Create a new scope to save the lexing state. The scope will be
983 * ended in S_sublex_done. Returns a '(', starting the function arguments
984 * to the uc, lc, etc. found before.
985 * Sets PL_lex_state to LEX_INTERPCONCAT.
993 PL_lex_state = PL_sublex_info.super_state;
994 SAVEI32(PL_lex_dojoin);
995 SAVEI32(PL_lex_brackets);
996 SAVEI32(PL_lex_casemods);
997 SAVEI32(PL_lex_starts);
998 SAVEI32(PL_lex_state);
999 SAVEVPTR(PL_lex_inpat);
1000 SAVEI32(PL_lex_inwhat);
1001 SAVECOPLINE(PL_curcop);
1002 SAVEPPTR(PL_bufptr);
1003 SAVEPPTR(PL_oldbufptr);
1004 SAVEPPTR(PL_oldoldbufptr);
1005 SAVEPPTR(PL_linestart);
1006 SAVESPTR(PL_linestr);
1007 SAVEPPTR(PL_lex_brackstack);
1008 SAVEPPTR(PL_lex_casestack);
1010 PL_linestr = PL_lex_stuff;
1011 PL_lex_stuff = Nullsv;
1013 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1014 = SvPVX(PL_linestr);
1015 PL_bufend += SvCUR(PL_linestr);
1016 SAVEFREESV(PL_linestr);
1018 PL_lex_dojoin = FALSE;
1019 PL_lex_brackets = 0;
1020 New(899, PL_lex_brackstack, 120, char);
1021 New(899, PL_lex_casestack, 12, char);
1022 SAVEFREEPV(PL_lex_brackstack);
1023 SAVEFREEPV(PL_lex_casestack);
1024 PL_lex_casemods = 0;
1025 *PL_lex_casestack = '\0';
1027 PL_lex_state = LEX_INTERPCONCAT;
1028 CopLINE_set(PL_curcop, PL_multi_start);
1030 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1031 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1032 PL_lex_inpat = PL_sublex_info.sub_op;
1034 PL_lex_inpat = Nullop;
1041 * Restores lexer state after a S_sublex_push.
1047 if (!PL_lex_starts++) {
1048 PL_expect = XOPERATOR;
1049 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1053 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1054 PL_lex_state = LEX_INTERPCASEMOD;
1058 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1059 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1060 PL_linestr = PL_lex_repl;
1062 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1063 PL_bufend += SvCUR(PL_linestr);
1064 SAVEFREESV(PL_linestr);
1065 PL_lex_dojoin = FALSE;
1066 PL_lex_brackets = 0;
1067 PL_lex_casemods = 0;
1068 *PL_lex_casestack = '\0';
1070 if (SvEVALED(PL_lex_repl)) {
1071 PL_lex_state = LEX_INTERPNORMAL;
1073 /* we don't clear PL_lex_repl here, so that we can check later
1074 whether this is an evalled subst; that means we rely on the
1075 logic to ensure sublex_done() is called again only via the
1076 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1079 PL_lex_state = LEX_INTERPCONCAT;
1080 PL_lex_repl = Nullsv;
1086 PL_bufend = SvPVX(PL_linestr);
1087 PL_bufend += SvCUR(PL_linestr);
1088 PL_expect = XOPERATOR;
1089 PL_sublex_info.sub_inwhat = 0;
1097 Extracts a pattern, double-quoted string, or transliteration. This
1100 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1101 processing a pattern (PL_lex_inpat is true), a transliteration
1102 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1104 Returns a pointer to the character scanned up to. Iff this is
1105 advanced from the start pointer supplied (ie if anything was
1106 successfully parsed), will leave an OP for the substring scanned
1107 in yylval. Caller must intuit reason for not parsing further
1108 by looking at the next characters herself.
1112 double-quoted style: \r and \n
1113 regexp special ones: \D \s
1115 backrefs: \1 (deprecated in substitution replacements)
1116 case and quoting: \U \Q \E
1117 stops on @ and $, but not for $ as tail anchor
1119 In transliterations:
1120 characters are VERY literal, except for - not at the start or end
1121 of the string, which indicates a range. scan_const expands the
1122 range to the full set of intermediate characters.
1124 In double-quoted strings:
1126 double-quoted style: \r and \n
1128 backrefs: \1 (deprecated)
1129 case and quoting: \U \Q \E
1132 scan_const does *not* construct ops to handle interpolated strings.
1133 It stops processing as soon as it finds an embedded $ or @ variable
1134 and leaves it to the caller to work out what's going on.
1136 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1138 $ in pattern could be $foo or could be tail anchor. Assumption:
1139 it's a tail anchor if $ is the last thing in the string, or if it's
1140 followed by one of ")| \n\t"
1142 \1 (backreferences) are turned into $1
1144 The structure of the code is
1145 while (there's a character to process) {
1146 handle transliteration ranges
1147 skip regexp comments
1148 skip # initiated comments in //x patterns
1149 check for embedded @foo
1150 check for embedded scalars
1152 leave intact backslashes from leave (below)
1153 deprecate \1 in strings and sub replacements
1154 handle string-changing backslashes \l \U \Q \E, etc.
1155 switch (what was escaped) {
1156 handle - in a transliteration (becomes a literal -)
1157 handle \132 octal characters
1158 handle 0x15 hex characters
1159 handle \cV (control V)
1160 handle printf backslashes (\f, \r, \n, etc)
1162 } (end if backslash)
1163 } (end while character to read)
1168 S_scan_const(pTHX_ char *start)
1170 register char *send = PL_bufend; /* end of the constant */
1171 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1172 register char *s = start; /* start of the constant */
1173 register char *d = SvPVX(sv); /* destination for copies */
1174 bool dorange = FALSE; /* are we in a translit range? */
1175 bool didrange = FALSE; /* did we just finish a range? */
1176 bool has_utf8 = FALSE; /* embedded \x{} */
1179 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1180 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1182 I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1183 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1184 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1186 const char *leaveit = /* set of acceptably-backslashed characters */
1188 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1191 while (s < send || dorange) {
1192 /* get transliterations out of the way (they're most literal) */
1193 if (PL_lex_inwhat == OP_TRANS) {
1194 /* expand a range A-Z to the full set of characters. AIE! */
1196 I32 i; /* current expanded character */
1197 I32 min; /* first character in range */
1198 I32 max; /* last character in range */
1200 i = d - SvPVX(sv); /* remember current offset */
1201 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1202 d = SvPVX(sv) + i; /* refresh d after realloc */
1203 d -= 2; /* eat the first char and the - */
1205 min = (U8)*d; /* first char in range */
1206 max = (U8)d[1]; /* last char in range */
1210 "Invalid [] range \"%c-%c\" in transliteration operator",
1211 (char)min, (char)max);
1215 if ((isLOWER(min) && isLOWER(max)) ||
1216 (isUPPER(min) && isUPPER(max))) {
1218 for (i = min; i <= max; i++)
1222 for (i = min; i <= max; i++)
1229 for (i = min; i <= max; i++)
1232 /* mark the range as done, and continue */
1238 /* range begins (ignore - as first or last char) */
1239 else if (*s == '-' && s+1 < send && s != start) {
1241 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1244 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1256 /* if we get here, we're not doing a transliteration */
1258 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1259 except for the last char, which will be done separately. */
1260 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1262 while (s < send && *s != ')')
1265 else if (s[2] == '{' /* This should match regcomp.c */
1266 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1269 char *regparse = s + (s[2] == '{' ? 3 : 4);
1272 while (count && (c = *regparse)) {
1273 if (c == '\\' && regparse[1])
1281 if (*regparse != ')') {
1282 regparse--; /* Leave one char for continuation. */
1283 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1285 while (s < regparse)
1290 /* likewise skip #-initiated comments in //x patterns */
1291 else if (*s == '#' && PL_lex_inpat &&
1292 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1293 while (s+1 < send && *s != '\n')
1297 /* check for embedded arrays
1298 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1300 else if (*s == '@' && s[1]
1301 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1304 /* check for embedded scalars. only stop if we're sure it's a
1307 else if (*s == '$') {
1308 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1310 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1311 break; /* in regexp, $ might be tail anchor */
1315 if (*s == '\\' && s+1 < send) {
1316 bool to_be_utf8 = FALSE;
1320 /* some backslashes we leave behind */
1321 if (*leaveit && *s && strchr(leaveit, *s)) {
1327 /* deprecate \1 in strings and substitution replacements */
1328 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1329 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1331 if (ckWARN(WARN_SYNTAX))
1332 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1337 /* string-change backslash escapes */
1338 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1343 /* if we get here, it's either a quoted -, or a digit */
1346 /* quoted - in transliterations */
1348 if (PL_lex_inwhat == OP_TRANS) {
1355 if (ckWARN(WARN_MISC) && isALNUM(*s))
1356 Perl_warner(aTHX_ WARN_MISC,
1357 "Unrecognized escape \\%c passed through",
1359 /* default action is to copy the quoted character */
1364 /* \132 indicates an octal constant */
1365 case '0': case '1': case '2': case '3':
1366 case '4': case '5': case '6': case '7':
1368 STRLEN len = 0; /* disallow underscores */
1369 uv = (UV)scan_oct(s, 3, &len);
1372 goto NUM_ESCAPE_INSERT;
1374 /* \x24 indicates a hex constant */
1378 char* e = strchr(s, '}');
1380 yyerror("Missing right brace on \\x{}");
1384 STRLEN len = 1; /* allow underscores */
1385 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1392 STRLEN len = 0; /* disallow underscores */
1393 uv = (UV)scan_hex(s, 2, &len);
1399 /* Insert oct or hex escaped character.
1400 * There will always enough room in sv since such escapes will
1401 * be longer than any utf8 sequence they can end up as
1404 if (!has_utf8 && (to_be_utf8 || uv > 255)) {
1405 /* might need to recode whatever we have accumulated so far
1406 * if it contains any hibit chars
1410 for (c = SvPVX(sv); c < d; c++) {
1415 char *old_pvx = SvPVX(sv);
1417 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1426 uv_to_utf8((U8*)dst, (U8)*src--);
1436 if (to_be_utf8 || uv > 255) {
1437 d = (char*)uv_to_utf8((U8*)d, uv);
1449 /* \N{latin small letter a} is a named character */
1453 char* e = strchr(s, '}');
1459 yyerror("Missing right brace on \\N{}");
1463 res = newSVpvn(s + 1, e - s - 1);
1464 res = new_constant( Nullch, 0, "charnames",
1465 res, Nullsv, "\\N{...}" );
1466 str = SvPV(res,len);
1467 if (!has_utf8 && SvUTF8(res)) {
1468 char *ostart = SvPVX(sv);
1469 SvCUR_set(sv, d - ostart);
1472 sv_utf8_upgrade(sv);
1473 /* this just broke our allocation above... */
1474 SvGROW(sv, send - start);
1475 d = SvPVX(sv) + SvCUR(sv);
1478 if (len > e - s + 4) {
1479 char *odest = SvPVX(sv);
1481 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1482 d = SvPVX(sv) + (d - odest);
1484 Copy(str, d, len, char);
1491 yyerror("Missing braces on \\N{}");
1494 /* \c is a control character */
1511 /* printf-style backslashes, formfeeds, newlines, etc */
1529 *d++ = '\047'; /* CP 1047 */
1532 *d++ = '\057'; /* CP 1047 */
1546 } /* end if (backslash) */
1548 /* (now in tr/// code again) */
1550 if (*s & 0x80 && (this_utf8 || has_utf8)) {
1551 STRLEN len = (STRLEN) -1;
1554 uv = utf8_to_uv((U8*)s, send - s, &len, 0);
1556 if (len == (STRLEN)-1) {
1557 /* Illegal UTF8 (a high-bit byte), make it valid. */
1558 char *old_pvx = SvPVX(sv);
1559 /* need space for one extra char (NOTE: SvCUR() not set here) */
1560 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1561 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1572 } /* while loop to process each character */
1574 /* terminate the string and set up the sv */
1576 SvCUR_set(sv, d - SvPVX(sv));
1581 /* shrink the sv if we allocated more than we used */
1582 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1583 SvLEN_set(sv, SvCUR(sv) + 1);
1584 Renew(SvPVX(sv), SvLEN(sv), char);
1587 /* return the substring (via yylval) only if we parsed anything */
1588 if (s > PL_bufptr) {
1589 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1590 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1592 ( PL_lex_inwhat == OP_TRANS
1594 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1597 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1604 * Returns TRUE if there's more to the expression (e.g., a subscript),
1607 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1609 * ->[ and ->{ return TRUE
1610 * { and [ outside a pattern are always subscripts, so return TRUE
1611 * if we're outside a pattern and it's not { or [, then return FALSE
1612 * if we're in a pattern and the first char is a {
1613 * {4,5} (any digits around the comma) returns FALSE
1614 * if we're in a pattern and the first char is a [
1616 * [SOMETHING] has a funky algorithm to decide whether it's a
1617 * character class or not. It has to deal with things like
1618 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1619 * anything else returns TRUE
1622 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1625 S_intuit_more(pTHX_ register char *s)
1627 if (PL_lex_brackets)
1629 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1631 if (*s != '{' && *s != '[')
1636 /* In a pattern, so maybe we have {n,m}. */
1653 /* On the other hand, maybe we have a character class */
1656 if (*s == ']' || *s == '^')
1659 /* this is terrifying, and it works */
1660 int weight = 2; /* let's weigh the evidence */
1662 unsigned char un_char = 255, last_un_char;
1663 char *send = strchr(s,']');
1664 char tmpbuf[sizeof PL_tokenbuf * 4];
1666 if (!send) /* has to be an expression */
1669 Zero(seen,256,char);
1672 else if (isDIGIT(*s)) {
1674 if (isDIGIT(s[1]) && s[2] == ']')
1680 for (; s < send; s++) {
1681 last_un_char = un_char;
1682 un_char = (unsigned char)*s;
1687 weight -= seen[un_char] * 10;
1688 if (isALNUM_lazy_if(s+1,UTF)) {
1689 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1690 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1695 else if (*s == '$' && s[1] &&
1696 strchr("[#!%*<>()-=",s[1])) {
1697 if (/*{*/ strchr("])} =",s[2]))
1706 if (strchr("wds]",s[1]))
1708 else if (seen['\''] || seen['"'])
1710 else if (strchr("rnftbxcav",s[1]))
1712 else if (isDIGIT(s[1])) {
1714 while (s[1] && isDIGIT(s[1]))
1724 if (strchr("aA01! ",last_un_char))
1726 if (strchr("zZ79~",s[1]))
1728 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1729 weight -= 5; /* cope with negative subscript */
1732 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1733 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1738 if (keyword(tmpbuf, d - tmpbuf))
1741 if (un_char == last_un_char + 1)
1743 weight -= seen[un_char];
1748 if (weight >= 0) /* probably a character class */
1758 * Does all the checking to disambiguate
1760 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1761 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1763 * First argument is the stuff after the first token, e.g. "bar".
1765 * Not a method if bar is a filehandle.
1766 * Not a method if foo is a subroutine prototyped to take a filehandle.
1767 * Not a method if it's really "Foo $bar"
1768 * Method if it's "foo $bar"
1769 * Not a method if it's really "print foo $bar"
1770 * Method if it's really "foo package::" (interpreted as package->foo)
1771 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1772 * Not a method if bar is a filehandle or package, but is quoted with
1777 S_intuit_method(pTHX_ char *start, GV *gv)
1779 char *s = start + (*start == '$');
1780 char tmpbuf[sizeof PL_tokenbuf];
1788 if ((cv = GvCVu(gv))) {
1789 char *proto = SvPVX(cv);
1799 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1800 /* start is the beginning of the possible filehandle/object,
1801 * and s is the end of it
1802 * tmpbuf is a copy of it
1805 if (*start == '$') {
1806 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1811 return *s == '(' ? FUNCMETH : METHOD;
1813 if (!keyword(tmpbuf, len)) {
1814 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1819 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1820 if (indirgv && GvCVu(indirgv))
1822 /* filehandle or package name makes it a method */
1823 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1825 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1826 return 0; /* no assumptions -- "=>" quotes bearword */
1828 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1829 newSVpvn(tmpbuf,len));
1830 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1834 return *s == '(' ? FUNCMETH : METHOD;
1842 * Return a string of Perl code to load the debugger. If PERL5DB
1843 * is set, it will return the contents of that, otherwise a
1844 * compile-time require of perl5db.pl.
1851 char *pdb = PerlEnv_getenv("PERL5DB");
1855 SETERRNO(0,SS$_NORMAL);
1856 return "BEGIN { require 'perl5db.pl' }";
1862 /* Encoded script support. filter_add() effectively inserts a
1863 * 'pre-processing' function into the current source input stream.
1864 * Note that the filter function only applies to the current source file
1865 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1867 * The datasv parameter (which may be NULL) can be used to pass
1868 * private data to this instance of the filter. The filter function
1869 * can recover the SV using the FILTER_DATA macro and use it to
1870 * store private buffers and state information.
1872 * The supplied datasv parameter is upgraded to a PVIO type
1873 * and the IoDIRP/IoANY field is used to store the function pointer,
1874 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1875 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1876 * private use must be set using malloc'd pointers.
1880 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1885 if (!PL_rsfp_filters)
1886 PL_rsfp_filters = newAV();
1888 datasv = NEWSV(255,0);
1889 if (!SvUPGRADE(datasv, SVt_PVIO))
1890 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1891 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1892 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1893 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1894 funcp, SvPV_nolen(datasv)));
1895 av_unshift(PL_rsfp_filters, 1);
1896 av_store(PL_rsfp_filters, 0, datasv) ;
1901 /* Delete most recently added instance of this filter function. */
1903 Perl_filter_del(pTHX_ filter_t funcp)
1906 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1907 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1909 /* if filter is on top of stack (usual case) just pop it off */
1910 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1911 if (IoANY(datasv) == (void *)funcp) {
1912 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1913 IoANY(datasv) = (void *)NULL;
1914 sv_free(av_pop(PL_rsfp_filters));
1918 /* we need to search for the correct entry and clear it */
1919 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1923 /* Invoke the n'th filter function for the current rsfp. */
1925 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1928 /* 0 = read one text line */
1933 if (!PL_rsfp_filters)
1935 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1936 /* Provide a default input filter to make life easy. */
1937 /* Note that we append to the line. This is handy. */
1938 DEBUG_P(PerlIO_printf(Perl_debug_log,
1939 "filter_read %d: from rsfp\n", idx));
1943 int old_len = SvCUR(buf_sv) ;
1945 /* ensure buf_sv is large enough */
1946 SvGROW(buf_sv, old_len + maxlen) ;
1947 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1948 if (PerlIO_error(PL_rsfp))
1949 return -1; /* error */
1951 return 0 ; /* end of file */
1953 SvCUR_set(buf_sv, old_len + len) ;
1956 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1957 if (PerlIO_error(PL_rsfp))
1958 return -1; /* error */
1960 return 0 ; /* end of file */
1963 return SvCUR(buf_sv);
1965 /* Skip this filter slot if filter has been deleted */
1966 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1967 DEBUG_P(PerlIO_printf(Perl_debug_log,
1968 "filter_read %d: skipped (filter deleted)\n",
1970 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1972 /* Get function pointer hidden within datasv */
1973 funcp = (filter_t)IoANY(datasv);
1974 DEBUG_P(PerlIO_printf(Perl_debug_log,
1975 "filter_read %d: via function %p (%s)\n",
1976 idx, funcp, SvPV_nolen(datasv)));
1977 /* Call function. The function is expected to */
1978 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1979 /* Return: <0:error, =0:eof, >0:not eof */
1980 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1984 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1986 #ifdef PERL_CR_FILTER
1987 if (!PL_rsfp_filters) {
1988 filter_add(S_cr_textfilter,NULL);
1991 if (PL_rsfp_filters) {
1994 SvCUR_set(sv, 0); /* start with empty line */
1995 if (FILTER_READ(0, sv, 0) > 0)
1996 return ( SvPVX(sv) ) ;
2001 return (sv_gets(sv, fp, append));
2005 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2009 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2013 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2014 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2016 return GvHV(gv); /* Foo:: */
2019 /* use constant CLASS => 'MyClass' */
2020 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2022 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2023 pkgname = SvPV_nolen(sv);
2027 return gv_stashpv(pkgname, FALSE);
2031 static char* exp_name[] =
2032 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2033 "ATTRTERM", "TERMBLOCK"
2040 Works out what to call the token just pulled out of the input
2041 stream. The yacc parser takes care of taking the ops we return and
2042 stitching them into a tree.
2048 if read an identifier
2049 if we're in a my declaration
2050 croak if they tried to say my($foo::bar)
2051 build the ops for a my() declaration
2052 if it's an access to a my() variable
2053 are we in a sort block?
2054 croak if my($a); $a <=> $b
2055 build ops for access to a my() variable
2056 if in a dq string, and they've said @foo and we can't find @foo
2058 build ops for a bareword
2059 if we already built the token before, use it.
2062 #ifdef USE_PURE_BISON
2064 #pragma segment Perl_yylex_r
2067 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2072 yylval_pointer[yyactlevel] = lvalp;
2073 yychar_pointer[yyactlevel] = lcharp;
2074 if (yyactlevel >= YYMAXLEVEL)
2075 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2077 r = Perl_yylex(aTHX);
2086 #pragma segment Perl_yylex
2098 /* check if there's an identifier for us to look at */
2099 if (PL_pending_ident) {
2100 /* pit holds the identifier we read and pending_ident is reset */
2101 char pit = PL_pending_ident;
2102 PL_pending_ident = 0;
2104 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2105 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2107 /* if we're in a my(), we can't allow dynamics here.
2108 $foo'bar has already been turned into $foo::bar, so
2109 just check for colons.
2111 if it's a legal name, the OP is a PADANY.
2114 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2115 if (strchr(PL_tokenbuf,':'))
2116 yyerror(Perl_form(aTHX_ "No package name allowed for "
2117 "variable %s in \"our\"",
2119 tmp = pad_allocmy(PL_tokenbuf);
2122 if (strchr(PL_tokenbuf,':'))
2123 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2125 yylval.opval = newOP(OP_PADANY, 0);
2126 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2132 build the ops for accesses to a my() variable.
2134 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2135 then used in a comparison. This catches most, but not
2136 all cases. For instance, it catches
2137 sort { my($a); $a <=> $b }
2139 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2140 (although why you'd do that is anyone's guess).
2143 if (!strchr(PL_tokenbuf,':')) {
2145 /* Check for single character per-thread SVs */
2146 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2147 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2148 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2150 yylval.opval = newOP(OP_THREADSV, 0);
2151 yylval.opval->op_targ = tmp;
2154 #endif /* USE_THREADS */
2155 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2156 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2157 /* might be an "our" variable" */
2158 if (SvFLAGS(namesv) & SVpad_OUR) {
2159 /* build ops for a bareword */
2160 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2161 sv_catpvn(sym, "::", 2);
2162 sv_catpv(sym, PL_tokenbuf+1);
2163 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2164 yylval.opval->op_private = OPpCONST_ENTERED;
2165 gv_fetchpv(SvPVX(sym),
2167 ? (GV_ADDMULTI | GV_ADDINEVAL)
2170 ((PL_tokenbuf[0] == '$') ? SVt_PV
2171 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2176 /* if it's a sort block and they're naming $a or $b */
2177 if (PL_last_lop_op == OP_SORT &&
2178 PL_tokenbuf[0] == '$' &&
2179 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2182 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2183 d < PL_bufend && *d != '\n';
2186 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2187 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2193 yylval.opval = newOP(OP_PADANY, 0);
2194 yylval.opval->op_targ = tmp;
2200 Whine if they've said @foo in a doublequoted string,
2201 and @foo isn't a variable we can find in the symbol
2204 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2205 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2206 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2207 && ckWARN(WARN_AMBIGUOUS))
2209 /* Downgraded from fatal to warning 20000522 mjd */
2210 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2211 "Possible unintended interpolation of %s in string",
2216 /* build ops for a bareword */
2217 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2218 yylval.opval->op_private = OPpCONST_ENTERED;
2219 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2220 ((PL_tokenbuf[0] == '$') ? SVt_PV
2221 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2226 /* no identifier pending identification */
2228 switch (PL_lex_state) {
2230 case LEX_NORMAL: /* Some compilers will produce faster */
2231 case LEX_INTERPNORMAL: /* code if we comment these out. */
2235 /* when we've already built the next token, just pull it out of the queue */
2238 yylval = PL_nextval[PL_nexttoke];
2240 PL_lex_state = PL_lex_defer;
2241 PL_expect = PL_lex_expect;
2242 PL_lex_defer = LEX_NORMAL;
2244 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2245 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2246 (IV)PL_nexttype[PL_nexttoke]); })
2248 return(PL_nexttype[PL_nexttoke]);
2250 /* interpolated case modifiers like \L \U, including \Q and \E.
2251 when we get here, PL_bufptr is at the \
2253 case LEX_INTERPCASEMOD:
2255 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2256 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2258 /* handle \E or end of string */
2259 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2263 if (PL_lex_casemods) {
2264 oldmod = PL_lex_casestack[--PL_lex_casemods];
2265 PL_lex_casestack[PL_lex_casemods] = '\0';
2267 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2269 PL_lex_state = LEX_INTERPCONCAT;
2273 if (PL_bufptr != PL_bufend)
2275 PL_lex_state = LEX_INTERPCONCAT;
2279 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2280 "### Saw case modifier at '%s'\n", PL_bufptr); })
2282 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2283 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2284 if (strchr("LU", *s) &&
2285 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2287 PL_lex_casestack[--PL_lex_casemods] = '\0';
2290 if (PL_lex_casemods > 10) {
2291 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2292 if (newlb != PL_lex_casestack) {
2294 PL_lex_casestack = newlb;
2297 PL_lex_casestack[PL_lex_casemods++] = *s;
2298 PL_lex_casestack[PL_lex_casemods] = '\0';
2299 PL_lex_state = LEX_INTERPCONCAT;
2300 PL_nextval[PL_nexttoke].ival = 0;
2303 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2305 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2307 PL_nextval[PL_nexttoke].ival = OP_LC;
2309 PL_nextval[PL_nexttoke].ival = OP_UC;
2311 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2313 Perl_croak(aTHX_ "panic: yylex");
2316 if (PL_lex_starts) {
2325 case LEX_INTERPPUSH:
2326 return sublex_push();
2328 case LEX_INTERPSTART:
2329 if (PL_bufptr == PL_bufend)
2330 return sublex_done();
2331 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2332 "### Interpolated variable at '%s'\n", PL_bufptr); })
2334 PL_lex_dojoin = (*PL_bufptr == '@');
2335 PL_lex_state = LEX_INTERPNORMAL;
2336 if (PL_lex_dojoin) {
2337 PL_nextval[PL_nexttoke].ival = 0;
2340 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2341 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2342 force_next(PRIVATEREF);
2344 force_ident("\"", '$');
2345 #endif /* USE_THREADS */
2346 PL_nextval[PL_nexttoke].ival = 0;
2348 PL_nextval[PL_nexttoke].ival = 0;
2350 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2353 if (PL_lex_starts++) {
2359 case LEX_INTERPENDMAYBE:
2360 if (intuit_more(PL_bufptr)) {
2361 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2367 if (PL_lex_dojoin) {
2368 PL_lex_dojoin = FALSE;
2369 PL_lex_state = LEX_INTERPCONCAT;
2372 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2373 && SvEVALED(PL_lex_repl))
2375 if (PL_bufptr != PL_bufend)
2376 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2377 PL_lex_repl = Nullsv;
2380 case LEX_INTERPCONCAT:
2382 if (PL_lex_brackets)
2383 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2385 if (PL_bufptr == PL_bufend)
2386 return sublex_done();
2388 if (SvIVX(PL_linestr) == '\'') {
2389 SV *sv = newSVsv(PL_linestr);
2392 else if ( PL_hints & HINT_NEW_RE )
2393 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2394 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2398 s = scan_const(PL_bufptr);
2400 PL_lex_state = LEX_INTERPCASEMOD;
2402 PL_lex_state = LEX_INTERPSTART;
2405 if (s != PL_bufptr) {
2406 PL_nextval[PL_nexttoke] = yylval;
2409 if (PL_lex_starts++)
2419 PL_lex_state = LEX_NORMAL;
2420 s = scan_formline(PL_bufptr);
2421 if (!PL_lex_formbrack)
2427 PL_oldoldbufptr = PL_oldbufptr;
2430 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2431 exp_name[PL_expect], s);
2437 if (isIDFIRST_lazy_if(s,UTF))
2439 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2442 goto fake_eof; /* emulate EOF on ^D or ^Z */
2447 if (PL_lex_brackets)
2448 yyerror("Missing right curly or square bracket");
2449 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2450 "### Tokener got EOF\n");
2454 if (s++ < PL_bufend)
2455 goto retry; /* ignore stray nulls */
2458 if (!PL_in_eval && !PL_preambled) {
2459 PL_preambled = TRUE;
2460 sv_setpv(PL_linestr,incl_perldb());
2461 if (SvCUR(PL_linestr))
2462 sv_catpv(PL_linestr,";");
2464 while(AvFILLp(PL_preambleav) >= 0) {
2465 SV *tmpsv = av_shift(PL_preambleav);
2466 sv_catsv(PL_linestr, tmpsv);
2467 sv_catpv(PL_linestr, ";");
2470 sv_free((SV*)PL_preambleav);
2471 PL_preambleav = NULL;
2473 if (PL_minus_n || PL_minus_p) {
2474 sv_catpv(PL_linestr, "LINE: while (<>) {");
2476 sv_catpv(PL_linestr,"chomp;");
2478 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2480 GvIMPORTED_AV_on(gv);
2482 if (strchr("/'\"", *PL_splitstr)
2483 && strchr(PL_splitstr + 1, *PL_splitstr))
2484 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2487 s = "'~#\200\1'"; /* surely one char is unused...*/
2488 while (s[1] && strchr(PL_splitstr, *s)) s++;
2490 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2491 "q" + (delim == '\''), delim);
2492 for (s = PL_splitstr; *s; s++) {
2494 sv_catpvn(PL_linestr, "\\", 1);
2495 sv_catpvn(PL_linestr, s, 1);
2497 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2501 sv_catpv(PL_linestr,"@F=split(' ');");
2504 sv_catpv(PL_linestr, "\n");
2505 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2506 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2507 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2508 SV *sv = NEWSV(85,0);
2510 sv_upgrade(sv, SVt_PVMG);
2511 sv_setsv(sv,PL_linestr);
2512 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2517 bool bof = PL_rsfp ? TRUE : FALSE;
2519 #ifdef PERLIO_IS_STDIO
2520 # ifdef __GNU_LIBRARY__
2521 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2522 # define FTELL_FOR_PIPE_IS_BROKEN
2526 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2527 # define FTELL_FOR_PIPE_IS_BROKEN
2532 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2533 /* This loses the possibility to detect the bof
2534 * situation on perl -P when the libc5 is being used.
2535 * Workaround? Maybe attach some extra state to PL_rsfp?
2538 bof = PerlIO_tell(PL_rsfp) == 0;
2540 bof = PerlIO_tell(PL_rsfp) == 0;
2543 s = filter_gets(PL_linestr, PL_rsfp, 0);
2547 if (PL_preprocess && !PL_in_eval)
2548 (void)PerlProc_pclose(PL_rsfp);
2549 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2550 PerlIO_clearerr(PL_rsfp);
2552 (void)PerlIO_close(PL_rsfp);
2554 PL_doextract = FALSE;
2556 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2557 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2558 sv_catpv(PL_linestr,";}");
2559 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2560 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2561 PL_minus_n = PL_minus_p = 0;
2564 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2565 sv_setpv(PL_linestr,"");
2566 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2568 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2569 s = swallow_bom((U8*)s);
2572 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2573 PL_doextract = FALSE;
2575 /* Incest with pod. */
2576 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2577 sv_setpv(PL_linestr, "");
2578 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2579 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2580 PL_doextract = FALSE;
2584 } while (PL_doextract);
2585 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2586 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2587 SV *sv = NEWSV(85,0);
2589 sv_upgrade(sv, SVt_PVMG);
2590 sv_setsv(sv,PL_linestr);
2591 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2593 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2594 if (CopLINE(PL_curcop) == 1) {
2595 while (s < PL_bufend && isSPACE(*s))
2597 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2601 if (*s == '#' && *(s+1) == '!')
2603 #ifdef ALTERNATE_SHEBANG
2605 static char as[] = ALTERNATE_SHEBANG;
2606 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2607 d = s + (sizeof(as) - 1);
2609 #endif /* ALTERNATE_SHEBANG */
2618 while (*d && !isSPACE(*d))
2622 #ifdef ARG_ZERO_IS_SCRIPT
2623 if (ipathend > ipath) {
2625 * HP-UX (at least) sets argv[0] to the script name,
2626 * which makes $^X incorrect. And Digital UNIX and Linux,
2627 * at least, set argv[0] to the basename of the Perl
2628 * interpreter. So, having found "#!", we'll set it right.
2630 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2631 assert(SvPOK(x) || SvGMAGICAL(x));
2632 if (sv_eq(x, CopFILESV(PL_curcop))) {
2633 sv_setpvn(x, ipath, ipathend - ipath);
2636 TAINT_NOT; /* $^X is always tainted, but that's OK */
2638 #endif /* ARG_ZERO_IS_SCRIPT */
2643 d = instr(s,"perl -");
2645 d = instr(s,"perl");
2647 /* avoid getting into infinite loops when shebang
2648 * line contains "Perl" rather than "perl" */
2650 for (d = ipathend-4; d >= ipath; --d) {
2651 if ((*d == 'p' || *d == 'P')
2652 && !ibcmp(d, "perl", 4))
2662 #ifdef ALTERNATE_SHEBANG
2664 * If the ALTERNATE_SHEBANG on this system starts with a
2665 * character that can be part of a Perl expression, then if
2666 * we see it but not "perl", we're probably looking at the
2667 * start of Perl code, not a request to hand off to some
2668 * other interpreter. Similarly, if "perl" is there, but
2669 * not in the first 'word' of the line, we assume the line
2670 * contains the start of the Perl program.
2672 if (d && *s != '#') {
2674 while (*c && !strchr("; \t\r\n\f\v#", *c))
2677 d = Nullch; /* "perl" not in first word; ignore */
2679 *s = '#'; /* Don't try to parse shebang line */
2681 #endif /* ALTERNATE_SHEBANG */
2682 #ifndef MACOS_TRADITIONAL
2687 !instr(s,"indir") &&
2688 instr(PL_origargv[0],"perl"))
2694 while (s < PL_bufend && isSPACE(*s))
2696 if (s < PL_bufend) {
2697 Newz(899,newargv,PL_origargc+3,char*);
2699 while (s < PL_bufend && !isSPACE(*s))
2702 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2705 newargv = PL_origargv;
2707 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2708 Perl_croak(aTHX_ "Can't exec %s", ipath);
2712 U32 oldpdb = PL_perldb;
2713 bool oldn = PL_minus_n;
2714 bool oldp = PL_minus_p;
2716 while (*d && !isSPACE(*d)) d++;
2717 while (SPACE_OR_TAB(*d)) d++;
2721 if (*d == 'M' || *d == 'm') {
2723 while (*d && !isSPACE(*d)) d++;
2724 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2727 d = moreswitches(d);
2729 if ((PERLDB_LINE && !oldpdb) ||
2730 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2731 /* if we have already added "LINE: while (<>) {",
2732 we must not do it again */
2734 sv_setpv(PL_linestr, "");
2735 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2736 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2737 PL_preambled = FALSE;
2739 (void)gv_fetchfile(PL_origfilename);
2746 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2748 PL_lex_state = LEX_FORMLINE;
2753 #ifdef PERL_STRICT_CR
2754 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2756 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2758 case ' ': case '\t': case '\f': case 013:
2759 #ifdef MACOS_TRADITIONAL
2766 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2767 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2768 /* handle eval qq[#line 1 "foo"\n ...] */
2769 CopLINE_dec(PL_curcop);
2773 while (s < d && *s != '\n')
2778 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2780 PL_lex_state = LEX_FORMLINE;
2790 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2797 while (s < PL_bufend && SPACE_OR_TAB(*s))
2800 if (strnEQ(s,"=>",2)) {
2801 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2802 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2803 "### Saw unary minus before =>, forcing word '%s'\n", s);
2805 OPERATOR('-'); /* unary minus */
2807 PL_last_uni = PL_oldbufptr;
2809 case 'r': ftst = OP_FTEREAD; break;
2810 case 'w': ftst = OP_FTEWRITE; break;
2811 case 'x': ftst = OP_FTEEXEC; break;
2812 case 'o': ftst = OP_FTEOWNED; break;
2813 case 'R': ftst = OP_FTRREAD; break;
2814 case 'W': ftst = OP_FTRWRITE; break;
2815 case 'X': ftst = OP_FTREXEC; break;
2816 case 'O': ftst = OP_FTROWNED; break;
2817 case 'e': ftst = OP_FTIS; break;
2818 case 'z': ftst = OP_FTZERO; break;
2819 case 's': ftst = OP_FTSIZE; break;
2820 case 'f': ftst = OP_FTFILE; break;
2821 case 'd': ftst = OP_FTDIR; break;
2822 case 'l': ftst = OP_FTLINK; break;
2823 case 'p': ftst = OP_FTPIPE; break;
2824 case 'S': ftst = OP_FTSOCK; break;
2825 case 'u': ftst = OP_FTSUID; break;
2826 case 'g': ftst = OP_FTSGID; break;
2827 case 'k': ftst = OP_FTSVTX; break;
2828 case 'b': ftst = OP_FTBLK; break;
2829 case 'c': ftst = OP_FTCHR; break;
2830 case 't': ftst = OP_FTTTY; break;
2831 case 'T': ftst = OP_FTTEXT; break;
2832 case 'B': ftst = OP_FTBINARY; break;
2833 case 'M': case 'A': case 'C':
2834 gv_fetchpv("\024",TRUE, SVt_PV);
2836 case 'M': ftst = OP_FTMTIME; break;
2837 case 'A': ftst = OP_FTATIME; break;
2838 case 'C': ftst = OP_FTCTIME; break;
2846 PL_last_lop_op = ftst;
2847 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2848 "### Saw file test %c\n", ftst);
2853 /* Assume it was a minus followed by a one-letter named
2854 * subroutine call (or a -bareword), then. */
2855 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2856 "### %c looked like a file test but was not\n", ftst);
2864 if (PL_expect == XOPERATOR)
2869 else if (*s == '>') {
2872 if (isIDFIRST_lazy_if(s,UTF)) {
2873 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2881 if (PL_expect == XOPERATOR)
2884 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2886 OPERATOR('-'); /* unary minus */
2893 if (PL_expect == XOPERATOR)
2898 if (PL_expect == XOPERATOR)
2901 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2907 if (PL_expect != XOPERATOR) {
2908 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2909 PL_expect = XOPERATOR;
2910 force_ident(PL_tokenbuf, '*');
2923 if (PL_expect == XOPERATOR) {
2927 PL_tokenbuf[0] = '%';
2928 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2929 if (!PL_tokenbuf[1]) {
2931 yyerror("Final % should be \\% or %name");
2934 PL_pending_ident = '%';
2953 switch (PL_expect) {
2956 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2958 PL_bufptr = s; /* update in case we back off */
2964 PL_expect = XTERMBLOCK;
2968 while (isIDFIRST_lazy_if(s,UTF)) {
2969 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2970 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2971 if (tmp < 0) tmp = -tmp;
2986 d = scan_str(d,TRUE,TRUE);
2989 SvREFCNT_dec(PL_lex_stuff);
2990 PL_lex_stuff = Nullsv;
2992 /* MUST advance bufptr here to avoid bogus
2993 "at end of line" context messages from yyerror().
2995 PL_bufptr = s + len;
2996 yyerror("Unterminated attribute parameter in attribute list");
2999 return 0; /* EOF indicator */
3003 SV *sv = newSVpvn(s, len);
3004 sv_catsv(sv, PL_lex_stuff);
3005 attrs = append_elem(OP_LIST, attrs,
3006 newSVOP(OP_CONST, 0, sv));
3007 SvREFCNT_dec(PL_lex_stuff);
3008 PL_lex_stuff = Nullsv;
3011 attrs = append_elem(OP_LIST, attrs,
3012 newSVOP(OP_CONST, 0,
3016 if (*s == ':' && s[1] != ':')
3019 break; /* require real whitespace or :'s */
3021 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3022 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3023 char q = ((*s == '\'') ? '"' : '\'');
3024 /* If here for an expression, and parsed no attrs, back off. */
3025 if (tmp == '=' && !attrs) {
3029 /* MUST advance bufptr here to avoid bogus "at end of line"
3030 context messages from yyerror().
3034 yyerror("Unterminated attribute list");
3036 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3044 PL_nextval[PL_nexttoke].opval = attrs;
3052 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3053 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3069 if (PL_lex_brackets <= 0)
3070 yyerror("Unmatched right square bracket");
3073 if (PL_lex_state == LEX_INTERPNORMAL) {
3074 if (PL_lex_brackets == 0) {
3075 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3076 PL_lex_state = LEX_INTERPEND;
3083 if (PL_lex_brackets > 100) {
3084 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3085 if (newlb != PL_lex_brackstack) {
3087 PL_lex_brackstack = newlb;
3090 switch (PL_expect) {
3092 if (PL_lex_formbrack) {
3096 if (PL_oldoldbufptr == PL_last_lop)
3097 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3099 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3100 OPERATOR(HASHBRACK);
3102 while (s < PL_bufend && SPACE_OR_TAB(*s))
3105 PL_tokenbuf[0] = '\0';
3106 if (d < PL_bufend && *d == '-') {
3107 PL_tokenbuf[0] = '-';
3109 while (d < PL_bufend && SPACE_OR_TAB(*d))
3112 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3113 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3115 while (d < PL_bufend && SPACE_OR_TAB(*d))
3118 char minus = (PL_tokenbuf[0] == '-');
3119 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3120 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, 0) &&
3121 PL_nextval[PL_nexttoke-1].opval)
3122 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke-1].opval)->op_sv);
3130 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3135 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3140 if (PL_oldoldbufptr == PL_last_lop)
3141 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3143 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3146 OPERATOR(HASHBRACK);
3147 /* This hack serves to disambiguate a pair of curlies
3148 * as being a block or an anon hash. Normally, expectation
3149 * determines that, but in cases where we're not in a
3150 * position to expect anything in particular (like inside
3151 * eval"") we have to resolve the ambiguity. This code
3152 * covers the case where the first term in the curlies is a
3153 * quoted string. Most other cases need to be explicitly
3154 * disambiguated by prepending a `+' before the opening
3155 * curly in order to force resolution as an anon hash.
3157 * XXX should probably propagate the outer expectation
3158 * into eval"" to rely less on this hack, but that could
3159 * potentially break current behavior of eval"".
3163 if (*s == '\'' || *s == '"' || *s == '`') {
3164 /* common case: get past first string, handling escapes */
3165 for (t++; t < PL_bufend && *t != *s;)
3166 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3170 else if (*s == 'q') {
3173 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3177 char open, close, term;
3180 while (t < PL_bufend && isSPACE(*t))
3184 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3188 for (t++; t < PL_bufend; t++) {
3189 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3191 else if (*t == open)
3195 for (t++; t < PL_bufend; t++) {
3196 if (*t == '\\' && t+1 < PL_bufend)
3198 else if (*t == close && --brackets <= 0)
3200 else if (*t == open)
3206 else if (isALNUM_lazy_if(t,UTF)) {
3208 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3211 while (t < PL_bufend && isSPACE(*t))
3213 /* if comma follows first term, call it an anon hash */
3214 /* XXX it could be a comma expression with loop modifiers */
3215 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3216 || (*t == '=' && t[1] == '>')))
3217 OPERATOR(HASHBRACK);
3218 if (PL_expect == XREF)
3221 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3227 yylval.ival = CopLINE(PL_curcop);
3228 if (isSPACE(*s) || *s == '#')
3229 PL_copline = NOLINE; /* invalidate current command line number */
3234 if (PL_lex_brackets <= 0)
3235 yyerror("Unmatched right curly bracket");
3237 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3238 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3239 PL_lex_formbrack = 0;
3240 if (PL_lex_state == LEX_INTERPNORMAL) {
3241 if (PL_lex_brackets == 0) {
3242 if (PL_expect & XFAKEBRACK) {
3243 PL_expect &= XENUMMASK;
3244 PL_lex_state = LEX_INTERPEND;
3246 return yylex(); /* ignore fake brackets */
3248 if (*s == '-' && s[1] == '>')
3249 PL_lex_state = LEX_INTERPENDMAYBE;
3250 else if (*s != '[' && *s != '{')
3251 PL_lex_state = LEX_INTERPEND;
3254 if (PL_expect & XFAKEBRACK) {
3255 PL_expect &= XENUMMASK;
3257 return yylex(); /* ignore fake brackets */
3267 if (PL_expect == XOPERATOR) {
3268 if (ckWARN(WARN_SEMICOLON)
3269 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3271 CopLINE_dec(PL_curcop);
3272 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3273 CopLINE_inc(PL_curcop);
3278 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3280 PL_expect = XOPERATOR;
3281 force_ident(PL_tokenbuf, '&');
3285 yylval.ival = (OPpENTERSUB_AMPER<<8);
3304 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3305 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3307 if (PL_expect == XSTATE && isALPHA(tmp) &&
3308 (s == PL_linestart+1 || s[-2] == '\n') )
3310 if (PL_in_eval && !PL_rsfp) {
3315 if (strnEQ(s,"=cut",4)) {
3329 PL_doextract = TRUE;
3332 if (PL_lex_brackets < PL_lex_formbrack) {
3334 #ifdef PERL_STRICT_CR
3335 for (t = s; SPACE_OR_TAB(*t); t++) ;
3337 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3339 if (*t == '\n' || *t == '#') {
3357 if (PL_expect != XOPERATOR) {
3358 if (s[1] != '<' && !strchr(s,'>'))
3361 s = scan_heredoc(s);
3363 s = scan_inputsymbol(s);
3364 TERM(sublex_start());
3369 SHop(OP_LEFT_SHIFT);
3383 SHop(OP_RIGHT_SHIFT);
3392 if (PL_expect == XOPERATOR) {
3393 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3396 return ','; /* grandfather non-comma-format format */
3400 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3401 PL_tokenbuf[0] = '@';
3402 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3403 sizeof PL_tokenbuf - 1, FALSE);
3404 if (PL_expect == XOPERATOR)
3405 no_op("Array length", s);
3406 if (!PL_tokenbuf[1])
3408 PL_expect = XOPERATOR;
3409 PL_pending_ident = '#';
3413 PL_tokenbuf[0] = '$';
3414 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3415 sizeof PL_tokenbuf - 1, FALSE);
3416 if (PL_expect == XOPERATOR)
3418 if (!PL_tokenbuf[1]) {
3420 yyerror("Final $ should be \\$ or $name");
3424 /* This kludge not intended to be bulletproof. */
3425 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3426 yylval.opval = newSVOP(OP_CONST, 0,
3427 newSViv(PL_compiling.cop_arybase));
3428 yylval.opval->op_private = OPpCONST_ARYBASE;
3434 if (PL_lex_state == LEX_NORMAL)
3437 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3440 PL_tokenbuf[0] = '@';
3441 if (ckWARN(WARN_SYNTAX)) {
3443 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3446 PL_bufptr = skipspace(PL_bufptr);
3447 while (t < PL_bufend && *t != ']')
3449 Perl_warner(aTHX_ WARN_SYNTAX,
3450 "Multidimensional syntax %.*s not supported",
3451 (t - PL_bufptr) + 1, PL_bufptr);
3455 else if (*s == '{') {
3456 PL_tokenbuf[0] = '%';
3457 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3458 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3460 char tmpbuf[sizeof PL_tokenbuf];
3462 for (t++; isSPACE(*t); t++) ;
3463 if (isIDFIRST_lazy_if(t,UTF)) {
3464 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3465 for (; isSPACE(*t); t++) ;
3466 if (*t == ';' && get_cv(tmpbuf, FALSE))
3467 Perl_warner(aTHX_ WARN_SYNTAX,
3468 "You need to quote \"%s\"", tmpbuf);
3474 PL_expect = XOPERATOR;
3475 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3476 bool islop = (PL_last_lop == PL_oldoldbufptr);
3477 if (!islop || PL_last_lop_op == OP_GREPSTART)
3478 PL_expect = XOPERATOR;
3479 else if (strchr("$@\"'`q", *s))
3480 PL_expect = XTERM; /* e.g. print $fh "foo" */
3481 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3482 PL_expect = XTERM; /* e.g. print $fh &sub */
3483 else if (isIDFIRST_lazy_if(s,UTF)) {
3484 char tmpbuf[sizeof PL_tokenbuf];
3485 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3486 if ((tmp = keyword(tmpbuf, len))) {
3487 /* binary operators exclude handle interpretations */
3499 PL_expect = XTERM; /* e.g. print $fh length() */
3504 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3505 if (gv && GvCVu(gv))
3506 PL_expect = XTERM; /* e.g. print $fh subr() */
3509 else if (isDIGIT(*s))
3510 PL_expect = XTERM; /* e.g. print $fh 3 */
3511 else if (*s == '.' && isDIGIT(s[1]))
3512 PL_expect = XTERM; /* e.g. print $fh .3 */
3513 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3514 PL_expect = XTERM; /* e.g. print $fh -1 */
3515 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3516 PL_expect = XTERM; /* print $fh <<"EOF" */
3518 PL_pending_ident = '$';
3522 if (PL_expect == XOPERATOR)
3524 PL_tokenbuf[0] = '@';
3525 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3526 if (!PL_tokenbuf[1]) {
3528 yyerror("Final @ should be \\@ or @name");
3531 if (PL_lex_state == LEX_NORMAL)
3533 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3535 PL_tokenbuf[0] = '%';
3537 /* Warn about @ where they meant $. */
3538 if (ckWARN(WARN_SYNTAX)) {
3539 if (*s == '[' || *s == '{') {
3541 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3543 if (*t == '}' || *t == ']') {
3545 PL_bufptr = skipspace(PL_bufptr);
3546 Perl_warner(aTHX_ WARN_SYNTAX,
3547 "Scalar value %.*s better written as $%.*s",
3548 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3553 PL_pending_ident = '@';
3556 case '/': /* may either be division or pattern */
3557 case '?': /* may either be conditional or pattern */
3558 if (PL_expect != XOPERATOR) {
3559 /* Disable warning on "study /blah/" */
3560 if (PL_oldoldbufptr == PL_last_uni
3561 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3562 || memNE(PL_last_uni, "study", 5)
3563 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3565 s = scan_pat(s,OP_MATCH);
3566 TERM(sublex_start());
3574 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3575 #ifdef PERL_STRICT_CR
3578 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3580 && (s == PL_linestart || s[-1] == '\n') )
3582 PL_lex_formbrack = 0;
3586 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3592 yylval.ival = OPf_SPECIAL;
3598 if (PL_expect != XOPERATOR)
3603 case '0': case '1': case '2': case '3': case '4':
3604 case '5': case '6': case '7': case '8': case '9':
3605 s = scan_num(s, &yylval);
3606 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3607 "### Saw number in '%s'\n", s);
3609 if (PL_expect == XOPERATOR)
3614 s = scan_str(s,FALSE,FALSE);
3615 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3616 "### Saw string in '%s'\n", s);
3618 if (PL_expect == XOPERATOR) {
3619 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3622 return ','; /* grandfather non-comma-format format */
3628 missingterm((char*)0);
3629 yylval.ival = OP_CONST;
3630 TERM(sublex_start());
3633 s = scan_str(s,FALSE,FALSE);
3634 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3635 "### Saw string in '%s'\n", s);
3637 if (PL_expect == XOPERATOR) {
3638 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3641 return ','; /* grandfather non-comma-format format */
3647 missingterm((char*)0);
3648 yylval.ival = OP_CONST;
3649 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3650 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3651 yylval.ival = OP_STRINGIFY;
3655 TERM(sublex_start());
3658 s = scan_str(s,FALSE,FALSE);
3659 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3660 "### Saw backtick string in '%s'\n", s);
3662 if (PL_expect == XOPERATOR)
3663 no_op("Backticks",s);
3665 missingterm((char*)0);
3666 yylval.ival = OP_BACKTICK;
3668 TERM(sublex_start());
3672 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3673 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3675 if (PL_expect == XOPERATOR)
3676 no_op("Backslash",s);
3680 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3684 while (isDIGIT(*start) || *start == '_')
3686 if (*start == '.' && isDIGIT(start[1])) {
3687 s = scan_num(s, &yylval);
3690 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3691 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3695 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3698 s = scan_num(s, &yylval);
3705 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3744 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3746 /* Some keywords can be followed by any delimiter, including ':' */
3747 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3748 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3749 (PL_tokenbuf[0] == 'q' &&
3750 strchr("qwxr", PL_tokenbuf[1])))));
3752 /* x::* is just a word, unless x is "CORE" */
3753 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3757 while (d < PL_bufend && isSPACE(*d))
3758 d++; /* no comments skipped here, or s### is misparsed */
3760 /* Is this a label? */
3761 if (!tmp && PL_expect == XSTATE
3762 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3764 yylval.pval = savepv(PL_tokenbuf);
3769 /* Check for keywords */
3770 tmp = keyword(PL_tokenbuf, len);
3772 /* Is this a word before a => operator? */
3773 if (*d == '=' && d[1] == '>') {
3775 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3776 yylval.opval->op_private = OPpCONST_BARE;
3777 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3778 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3782 if (tmp < 0) { /* second-class keyword? */
3783 GV *ogv = Nullgv; /* override (winner) */
3784 GV *hgv = Nullgv; /* hidden (loser) */
3785 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3787 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3790 if (GvIMPORTED_CV(gv))
3792 else if (! CvMETHOD(cv))
3796 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3797 (gv = *gvp) != (GV*)&PL_sv_undef &&
3798 GvCVu(gv) && GvIMPORTED_CV(gv))
3804 tmp = 0; /* overridden by import or by GLOBAL */
3807 && -tmp==KEY_lock /* XXX generalizable kludge */
3809 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3811 tmp = 0; /* any sub overrides "weak" keyword */
3813 else { /* no override */
3817 if (ckWARN(WARN_AMBIGUOUS) && hgv
3818 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3819 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3820 "Ambiguous call resolved as CORE::%s(), %s",
3821 GvENAME(hgv), "qualify as such or use &");
3828 default: /* not a keyword */
3831 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3833 /* Get the rest if it looks like a package qualifier */
3835 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3837 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3840 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3841 *s == '\'' ? "'" : "::");
3845 if (PL_expect == XOPERATOR) {
3846 if (PL_bufptr == PL_linestart) {
3847 CopLINE_dec(PL_curcop);
3848 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3849 CopLINE_inc(PL_curcop);
3852 no_op("Bareword",s);
3855 /* Look for a subroutine with this name in current package,
3856 unless name is "Foo::", in which case Foo is a bearword
3857 (and a package name). */
3860 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3862 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3863 Perl_warner(aTHX_ WARN_BAREWORD,
3864 "Bareword \"%s\" refers to nonexistent package",
3867 PL_tokenbuf[len] = '\0';
3874 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3877 /* if we saw a global override before, get the right name */
3880 sv = newSVpvn("CORE::GLOBAL::",14);
3881 sv_catpv(sv,PL_tokenbuf);
3884 sv = newSVpv(PL_tokenbuf,0);
3886 /* Presume this is going to be a bareword of some sort. */
3889 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3890 yylval.opval->op_private = OPpCONST_BARE;
3892 /* And if "Foo::", then that's what it certainly is. */
3897 /* See if it's the indirect object for a list operator. */
3899 if (PL_oldoldbufptr &&
3900 PL_oldoldbufptr < PL_bufptr &&
3901 (PL_oldoldbufptr == PL_last_lop
3902 || PL_oldoldbufptr == PL_last_uni) &&
3903 /* NO SKIPSPACE BEFORE HERE! */
3904 (PL_expect == XREF ||
3905 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3907 bool immediate_paren = *s == '(';
3909 /* (Now we can afford to cross potential line boundary.) */
3912 /* Two barewords in a row may indicate method call. */
3914 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3917 /* If not a declared subroutine, it's an indirect object. */
3918 /* (But it's an indir obj regardless for sort.) */
3920 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3921 ((!gv || !GvCVu(gv)) &&
3922 (PL_last_lop_op != OP_MAPSTART &&
3923 PL_last_lop_op != OP_GREPSTART))))
3925 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3931 PL_expect = XOPERATOR;
3934 /* Is this a word before a => operator? */
3935 if (*s == '=' && s[1] == '>') {
3937 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3938 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3939 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3943 /* If followed by a paren, it's certainly a subroutine. */
3946 if (gv && GvCVu(gv)) {
3947 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3948 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3953 PL_nextval[PL_nexttoke].opval = yylval.opval;
3954 PL_expect = XOPERATOR;
3960 /* If followed by var or block, call it a method (unless sub) */
3962 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3963 PL_last_lop = PL_oldbufptr;
3964 PL_last_lop_op = OP_METHOD;
3968 /* If followed by a bareword, see if it looks like indir obj. */
3970 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3973 /* Not a method, so call it a subroutine (if defined) */
3975 if (gv && GvCVu(gv)) {
3977 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3978 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3979 "Ambiguous use of -%s resolved as -&%s()",
3980 PL_tokenbuf, PL_tokenbuf);
3981 /* Check for a constant sub */
3983 if ((sv = cv_const_sv(cv))) {
3985 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3986 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3987 yylval.opval->op_private = 0;
3991 /* Resolve to GV now. */
3992 op_free(yylval.opval);
3993 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3994 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
3995 PL_last_lop = PL_oldbufptr;
3996 PL_last_lop_op = OP_ENTERSUB;
3997 /* Is there a prototype? */
4000 char *proto = SvPV((SV*)cv, len);
4003 if (strEQ(proto, "$"))
4005 if (*proto == '&' && *s == '{') {
4006 sv_setpv(PL_subname,"__ANON__");
4010 PL_nextval[PL_nexttoke].opval = yylval.opval;
4016 /* Call it a bare word */
4018 if (PL_hints & HINT_STRICT_SUBS)
4019 yylval.opval->op_private |= OPpCONST_STRICT;
4022 if (ckWARN(WARN_RESERVED)) {
4023 if (lastchar != '-') {
4024 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4026 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4033 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4034 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4035 "Operator or semicolon missing before %c%s",
4036 lastchar, PL_tokenbuf);
4037 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4038 "Ambiguous use of %c resolved as operator %c",
4039 lastchar, lastchar);
4045 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4046 newSVpv(CopFILE(PL_curcop),0));
4050 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4051 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4054 case KEY___PACKAGE__:
4055 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4057 ? newSVsv(PL_curstname)
4066 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4067 char *pname = "main";
4068 if (PL_tokenbuf[2] == 'D')
4069 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4070 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4073 GvIOp(gv) = newIO();
4074 IoIFP(GvIOp(gv)) = PL_rsfp;
4075 #if defined(HAS_FCNTL) && defined(F_SETFD)
4077 int fd = PerlIO_fileno(PL_rsfp);
4078 fcntl(fd,F_SETFD,fd >= 3);
4081 /* Mark this internal pseudo-handle as clean */
4082 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4084 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4085 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4086 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4088 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4089 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4090 /* if the script was opened in binmode, we need to revert
4091 * it to text mode for compatibility; but only iff it has CRs
4092 * XXX this is a questionable hack at best. */
4093 if (PL_bufend-PL_bufptr > 2
4094 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4097 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4098 loc = PerlIO_tell(PL_rsfp);
4099 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4101 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4102 #if defined(__BORLANDC__)
4103 /* XXX see note in do_binmode() */
4104 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4107 PerlIO_seek(PL_rsfp, loc, 0);
4111 #ifdef PERLIO_LAYERS
4112 if (UTF && !IN_BYTE)
4113 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4126 if (PL_expect == XSTATE) {
4133 if (*s == ':' && s[1] == ':') {
4136 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4137 if (!(tmp = keyword(PL_tokenbuf, len)))
4138 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4152 LOP(OP_ACCEPT,XTERM);
4158 LOP(OP_ATAN2,XTERM);
4164 LOP(OP_BINMODE,XTERM);
4167 LOP(OP_BLESS,XTERM);
4176 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4193 if (!PL_cryptseen) {
4194 PL_cryptseen = TRUE;
4198 LOP(OP_CRYPT,XTERM);
4201 if (ckWARN(WARN_CHMOD)) {
4202 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4203 if (*d != '0' && isDIGIT(*d))
4204 Perl_warner(aTHX_ WARN_CHMOD,
4205 "chmod() mode argument is missing initial 0");
4207 LOP(OP_CHMOD,XTERM);
4210 LOP(OP_CHOWN,XTERM);
4213 LOP(OP_CONNECT,XTERM);
4229 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4233 PL_hints |= HINT_BLOCK_SCOPE;
4243 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4244 LOP(OP_DBMOPEN,XTERM);
4250 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4257 yylval.ival = CopLINE(PL_curcop);
4271 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4272 UNIBRACK(OP_ENTEREVAL);
4287 case KEY_endhostent:
4293 case KEY_endservent:
4296 case KEY_endprotoent:
4307 yylval.ival = CopLINE(PL_curcop);
4309 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4311 if ((PL_bufend - p) >= 3 &&
4312 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4314 else if ((PL_bufend - p) >= 4 &&
4315 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4318 if (isIDFIRST_lazy_if(p,UTF)) {
4319 p = scan_ident(p, PL_bufend,
4320 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4324 Perl_croak(aTHX_ "Missing $ on loop variable");
4329 LOP(OP_FORMLINE,XTERM);
4335 LOP(OP_FCNTL,XTERM);
4341 LOP(OP_FLOCK,XTERM);
4350 LOP(OP_GREPSTART, XREF);
4353 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4368 case KEY_getpriority:
4369 LOP(OP_GETPRIORITY,XTERM);
4371 case KEY_getprotobyname:
4374 case KEY_getprotobynumber:
4375 LOP(OP_GPBYNUMBER,XTERM);
4377 case KEY_getprotoent:
4389 case KEY_getpeername:
4390 UNI(OP_GETPEERNAME);
4392 case KEY_gethostbyname:
4395 case KEY_gethostbyaddr:
4396 LOP(OP_GHBYADDR,XTERM);
4398 case KEY_gethostent:
4401 case KEY_getnetbyname:
4404 case KEY_getnetbyaddr:
4405 LOP(OP_GNBYADDR,XTERM);
4410 case KEY_getservbyname:
4411 LOP(OP_GSBYNAME,XTERM);
4413 case KEY_getservbyport:
4414 LOP(OP_GSBYPORT,XTERM);
4416 case KEY_getservent:
4419 case KEY_getsockname:
4420 UNI(OP_GETSOCKNAME);
4422 case KEY_getsockopt:
4423 LOP(OP_GSOCKOPT,XTERM);
4445 yylval.ival = CopLINE(PL_curcop);
4449 LOP(OP_INDEX,XTERM);
4455 LOP(OP_IOCTL,XTERM);
4467 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4499 LOP(OP_LISTEN,XTERM);
4508 s = scan_pat(s,OP_MATCH);
4509 TERM(sublex_start());
4512 LOP(OP_MAPSTART, XREF);
4515 LOP(OP_MKDIR,XTERM);
4518 LOP(OP_MSGCTL,XTERM);
4521 LOP(OP_MSGGET,XTERM);
4524 LOP(OP_MSGRCV,XTERM);
4527 LOP(OP_MSGSND,XTERM);
4533 if (isIDFIRST_lazy_if(s,UTF)) {
4534 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4535 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4537 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4538 if (!PL_in_my_stash) {
4541 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4549 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4556 if (PL_expect != XSTATE)
4557 yyerror("\"no\" not allowed in expression");
4558 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4559 s = force_version(s);
4564 if (*s == '(' || (s = skipspace(s), *s == '('))
4571 if (isIDFIRST_lazy_if(s,UTF)) {
4573 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4575 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4576 Perl_warner(aTHX_ WARN_PRECEDENCE,
4577 "Precedence problem: open %.*s should be open(%.*s)",
4583 yylval.ival = OP_OR;
4593 LOP(OP_OPEN_DIR,XTERM);
4596 checkcomma(s,PL_tokenbuf,"filehandle");
4600 checkcomma(s,PL_tokenbuf,"filehandle");
4619 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4623 LOP(OP_PIPE_OP,XTERM);
4626 s = scan_str(s,FALSE,FALSE);
4628 missingterm((char*)0);
4629 yylval.ival = OP_CONST;
4630 TERM(sublex_start());
4636 s = scan_str(s,FALSE,FALSE);
4638 missingterm((char*)0);
4640 if (SvCUR(PL_lex_stuff)) {
4643 d = SvPV_force(PL_lex_stuff, len);
4646 for (; isSPACE(*d) && len; --len, ++d) ;
4649 if (!warned && ckWARN(WARN_QW)) {
4650 for (; !isSPACE(*d) && len; --len, ++d) {
4652 Perl_warner(aTHX_ WARN_QW,
4653 "Possible attempt to separate words with commas");
4656 else if (*d == '#') {
4657 Perl_warner(aTHX_ WARN_QW,
4658 "Possible attempt to put comments in qw() list");
4664 for (; !isSPACE(*d) && len; --len, ++d) ;
4666 sv = newSVpvn(b, d-b);
4667 if (DO_UTF8(PL_lex_stuff))
4669 words = append_elem(OP_LIST, words,
4670 newSVOP(OP_CONST, 0, tokeq(sv)));
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 if (ckWARN(WARN_SYNTAX)) {
5764 for (w = s+2; *w && level; w++) {
5771 for (; *w && isSPACE(*w); w++) ;
5772 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5773 Perl_warner(aTHX_ WARN_SYNTAX,
5774 "%s (...) interpreted as function",name);
5777 while (s < PL_bufend && isSPACE(*s))
5781 while (s < PL_bufend && isSPACE(*s))
5783 if (isIDFIRST_lazy_if(s,UTF)) {
5785 while (isALNUM_lazy_if(s,UTF))
5787 while (s < PL_bufend && isSPACE(*s))
5792 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5796 Perl_croak(aTHX_ "No comma allowed after %s", what);
5801 /* Either returns sv, or mortalizes sv and returns a new SV*.
5802 Best used as sv=new_constant(..., sv, ...).
5803 If s, pv are NULL, calls subroutine with one argument,
5804 and type is used with error messages only. */
5807 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5811 HV *table = GvHV(PL_hintgv); /* ^H */
5815 const char *why1, *why2, *why3;
5817 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5820 why2 = strEQ(key,"charnames")
5821 ? "(possibly a missing \"use charnames ...\")"
5823 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5824 (type ? type: "undef"), why2);
5826 /* This is convoluted and evil ("goto considered harmful")
5827 * but I do not understand the intricacies of all the different
5828 * failure modes of %^H in here. The goal here is to make
5829 * the most probable error message user-friendly. --jhi */
5834 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5835 (type ? type: "undef"), why1, why2, why3);
5837 yyerror(SvPVX(msg));
5841 cvp = hv_fetch(table, key, strlen(key), FALSE);
5842 if (!cvp || !SvOK(*cvp)) {
5845 why3 = "} is not defined";
5848 sv_2mortal(sv); /* Parent created it permanently */
5851 pv = sv_2mortal(newSVpvn(s, len));
5853 typesv = sv_2mortal(newSVpv(type, 0));
5855 typesv = &PL_sv_undef;
5857 PUSHSTACKi(PERLSI_OVERLOAD);
5869 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5873 /* Check the eval first */
5874 if (!PL_in_eval && SvTRUE(ERRSV)) {
5876 sv_catpv(ERRSV, "Propagated");
5877 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5879 res = SvREFCNT_inc(sv);
5883 (void)SvREFCNT_inc(res);
5892 why1 = "Call to &{$^H{";
5894 why3 = "}} did not return a defined value";
5903 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5905 register char *d = dest;
5906 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5909 Perl_croak(aTHX_ ident_too_long);
5910 if (isALNUM(*s)) /* UTF handled below */
5912 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5917 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5921 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5922 char *t = s + UTF8SKIP(s);
5923 while (*t & 0x80 && is_utf8_mark((U8*)t))
5925 if (d + (t - s) > e)
5926 Perl_croak(aTHX_ ident_too_long);
5927 Copy(s, d, t - s, char);
5940 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5950 e = d + destlen - 3; /* two-character token, ending NUL */
5952 while (isDIGIT(*s)) {
5954 Perl_croak(aTHX_ ident_too_long);
5961 Perl_croak(aTHX_ ident_too_long);
5962 if (isALNUM(*s)) /* UTF handled below */
5964 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5969 else if (*s == ':' && s[1] == ':') {
5973 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5974 char *t = s + UTF8SKIP(s);
5975 while (*t & 0x80 && is_utf8_mark((U8*)t))
5977 if (d + (t - s) > e)
5978 Perl_croak(aTHX_ ident_too_long);
5979 Copy(s, d, t - s, char);
5990 if (PL_lex_state != LEX_NORMAL)
5991 PL_lex_state = LEX_INTERPENDMAYBE;
5994 if (*s == '$' && s[1] &&
5995 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6008 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6013 if (isSPACE(s[-1])) {
6016 if (!SPACE_OR_TAB(ch)) {
6022 if (isIDFIRST_lazy_if(d,UTF)) {
6026 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6028 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
6031 Copy(s, d, e - s, char);
6036 while ((isALNUM(*s) || *s == ':') && d < e)
6039 Perl_croak(aTHX_ ident_too_long);
6042 while (s < send && SPACE_OR_TAB(*s)) s++;
6043 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6044 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6045 const char *brack = *s == '[' ? "[...]" : "{...}";
6046 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6047 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6048 funny, dest, brack, funny, dest, brack);
6051 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6055 /* Handle extended ${^Foo} variables
6056 * 1999-02-27 mjd-perl-patch@plover.com */
6057 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6061 while (isALNUM(*s) && d < e) {
6065 Perl_croak(aTHX_ ident_too_long);
6070 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6071 PL_lex_state = LEX_INTERPEND;
6074 if (PL_lex_state == LEX_NORMAL) {
6075 if (ckWARN(WARN_AMBIGUOUS) &&
6076 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6078 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6079 "Ambiguous use of %c{%s} resolved to %c%s",
6080 funny, dest, funny, dest);
6085 s = bracket; /* let the parser handle it */
6089 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6090 PL_lex_state = LEX_INTERPEND;
6095 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6100 *pmfl |= PMf_GLOBAL;
6102 *pmfl |= PMf_CONTINUE;
6106 *pmfl |= PMf_MULTILINE;
6108 *pmfl |= PMf_SINGLELINE;
6110 *pmfl |= PMf_EXTENDED;
6114 S_scan_pat(pTHX_ char *start, I32 type)
6119 s = scan_str(start,FALSE,FALSE);
6122 SvREFCNT_dec(PL_lex_stuff);
6123 PL_lex_stuff = Nullsv;
6124 Perl_croak(aTHX_ "Search pattern not terminated");
6127 pm = (PMOP*)newPMOP(type, 0);
6128 if (PL_multi_open == '?')
6129 pm->op_pmflags |= PMf_ONCE;
6131 while (*s && strchr("iomsx", *s))
6132 pmflag(&pm->op_pmflags,*s++);
6135 while (*s && strchr("iogcmsx", *s))
6136 pmflag(&pm->op_pmflags,*s++);
6138 pm->op_pmpermflags = pm->op_pmflags;
6140 PL_lex_op = (OP*)pm;
6141 yylval.ival = OP_MATCH;
6146 S_scan_subst(pTHX_ char *start)
6153 yylval.ival = OP_NULL;
6155 s = scan_str(start,FALSE,FALSE);
6159 SvREFCNT_dec(PL_lex_stuff);
6160 PL_lex_stuff = Nullsv;
6161 Perl_croak(aTHX_ "Substitution pattern not terminated");
6164 if (s[-1] == PL_multi_open)
6167 first_start = PL_multi_start;
6168 s = scan_str(s,FALSE,FALSE);
6171 SvREFCNT_dec(PL_lex_stuff);
6172 PL_lex_stuff = Nullsv;
6174 SvREFCNT_dec(PL_lex_repl);
6175 PL_lex_repl = Nullsv;
6176 Perl_croak(aTHX_ "Substitution replacement not terminated");
6178 PL_multi_start = first_start; /* so whole substitution is taken together */
6180 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6186 else if (strchr("iogcmsx", *s))
6187 pmflag(&pm->op_pmflags,*s++);
6194 PL_sublex_info.super_bufptr = s;
6195 PL_sublex_info.super_bufend = PL_bufend;
6197 pm->op_pmflags |= PMf_EVAL;
6198 repl = newSVpvn("",0);
6200 sv_catpv(repl, es ? "eval " : "do ");
6201 sv_catpvn(repl, "{ ", 2);
6202 sv_catsv(repl, PL_lex_repl);
6203 sv_catpvn(repl, " };", 2);
6205 SvREFCNT_dec(PL_lex_repl);
6209 pm->op_pmpermflags = pm->op_pmflags;
6210 PL_lex_op = (OP*)pm;
6211 yylval.ival = OP_SUBST;
6216 S_scan_trans(pTHX_ char *start)
6227 yylval.ival = OP_NULL;
6229 s = scan_str(start,FALSE,FALSE);
6232 SvREFCNT_dec(PL_lex_stuff);
6233 PL_lex_stuff = Nullsv;
6234 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6236 if (s[-1] == PL_multi_open)
6239 s = scan_str(s,FALSE,FALSE);
6242 SvREFCNT_dec(PL_lex_stuff);
6243 PL_lex_stuff = Nullsv;
6245 SvREFCNT_dec(PL_lex_repl);
6246 PL_lex_repl = Nullsv;
6247 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6250 New(803,tbl,256,short);
6251 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6253 complement = del = squash = 0;
6254 while (strchr("cds", *s)) {
6256 complement = OPpTRANS_COMPLEMENT;
6258 del = OPpTRANS_DELETE;
6260 squash = OPpTRANS_SQUASH;
6263 o->op_private = del|squash|complement|
6264 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6265 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6268 yylval.ival = OP_TRANS;
6273 S_scan_heredoc(pTHX_ register char *s)
6276 I32 op_type = OP_SCALAR;
6283 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6287 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6290 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6291 if (*peek && strchr("`'\"",*peek)) {
6294 s = delimcpy(d, e, s, PL_bufend, term, &len);
6304 if (!isALNUM_lazy_if(s,UTF))
6305 deprecate("bare << to mean <<\"\"");
6306 for (; isALNUM_lazy_if(s,UTF); s++) {
6311 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6312 Perl_croak(aTHX_ "Delimiter for here document is too long");
6315 len = d - PL_tokenbuf;
6316 #ifndef PERL_STRICT_CR
6317 d = strchr(s, '\r');
6321 while (s < PL_bufend) {
6327 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6336 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6341 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6342 herewas = newSVpvn(s,PL_bufend-s);
6344 s--, herewas = newSVpvn(s,d-s);
6345 s += SvCUR(herewas);
6347 tmpstr = NEWSV(87,79);
6348 sv_upgrade(tmpstr, SVt_PVIV);
6353 else if (term == '`') {
6354 op_type = OP_BACKTICK;
6355 SvIVX(tmpstr) = '\\';
6359 PL_multi_start = CopLINE(PL_curcop);
6360 PL_multi_open = PL_multi_close = '<';
6361 term = *PL_tokenbuf;
6362 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6363 char *bufptr = PL_sublex_info.super_bufptr;
6364 char *bufend = PL_sublex_info.super_bufend;
6365 char *olds = s - SvCUR(herewas);
6366 s = strchr(bufptr, '\n');
6370 while (s < bufend &&
6371 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6373 CopLINE_inc(PL_curcop);
6376 CopLINE_set(PL_curcop, PL_multi_start);
6377 missingterm(PL_tokenbuf);
6379 sv_setpvn(herewas,bufptr,d-bufptr+1);
6380 sv_setpvn(tmpstr,d+1,s-d);
6382 sv_catpvn(herewas,s,bufend-s);
6383 (void)strcpy(bufptr,SvPVX(herewas));
6390 while (s < PL_bufend &&
6391 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6393 CopLINE_inc(PL_curcop);
6395 if (s >= PL_bufend) {
6396 CopLINE_set(PL_curcop, PL_multi_start);
6397 missingterm(PL_tokenbuf);
6399 sv_setpvn(tmpstr,d+1,s-d);
6401 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6403 sv_catpvn(herewas,s,PL_bufend-s);
6404 sv_setsv(PL_linestr,herewas);
6405 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6406 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6409 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6410 while (s >= PL_bufend) { /* multiple line string? */
6412 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6413 CopLINE_set(PL_curcop, PL_multi_start);
6414 missingterm(PL_tokenbuf);
6416 CopLINE_inc(PL_curcop);
6417 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6418 #ifndef PERL_STRICT_CR
6419 if (PL_bufend - PL_linestart >= 2) {
6420 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6421 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6423 PL_bufend[-2] = '\n';
6425 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6427 else if (PL_bufend[-1] == '\r')
6428 PL_bufend[-1] = '\n';
6430 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6431 PL_bufend[-1] = '\n';
6433 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6434 SV *sv = NEWSV(88,0);
6436 sv_upgrade(sv, SVt_PVMG);
6437 sv_setsv(sv,PL_linestr);
6438 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6440 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6443 sv_catsv(PL_linestr,herewas);
6444 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6448 sv_catsv(tmpstr,PL_linestr);
6453 PL_multi_end = CopLINE(PL_curcop);
6454 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6455 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6456 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6458 SvREFCNT_dec(herewas);
6459 if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
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)
6628 SV *sv; /* scalar value: string */
6629 char *tmps; /* temp string, used for delimiter matching */
6630 register char *s = start; /* current position in the buffer */
6631 register char term; /* terminating character */
6632 register char *to; /* current position in the sv's data */
6633 I32 brackets = 1; /* bracket nesting level */
6634 bool has_utf8 = FALSE; /* is there any utf8 content? */
6636 /* skip space before the delimiter */
6640 /* mark where we are, in case we need to report errors */
6643 /* after skipping whitespace, the next character is the terminator */
6645 if ((term & 0x80) && UTF)
6648 /* mark where we are */
6649 PL_multi_start = CopLINE(PL_curcop);
6650 PL_multi_open = term;
6652 /* find corresponding closing delimiter */
6653 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6655 PL_multi_close = term;
6657 /* create a new SV to hold the contents. 87 is leak category, I'm
6658 assuming. 79 is the SV's initial length. What a random number. */
6660 sv_upgrade(sv, SVt_PVIV);
6662 (void)SvPOK_only(sv); /* validate pointer */
6664 /* move past delimiter and try to read a complete string */
6666 sv_catpvn(sv, s, 1);
6669 /* extend sv if need be */
6670 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6671 /* set 'to' to the next character in the sv's string */
6672 to = SvPVX(sv)+SvCUR(sv);
6674 /* if open delimiter is the close delimiter read unbridle */
6675 if (PL_multi_open == PL_multi_close) {
6676 for (; s < PL_bufend; s++,to++) {
6677 /* embedded newlines increment the current line number */
6678 if (*s == '\n' && !PL_rsfp)
6679 CopLINE_inc(PL_curcop);
6680 /* handle quoted delimiters */
6681 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6682 if (!keep_quoted && s[1] == term)
6684 /* any other quotes are simply copied straight through */
6688 /* terminate when run out of buffer (the for() condition), or
6689 have found the terminator */
6690 else if (*s == term)
6692 else if (!has_utf8 && (*s & 0x80) && UTF)
6698 /* if the terminator isn't the same as the start character (e.g.,
6699 matched brackets), we have to allow more in the quoting, and
6700 be prepared for nested brackets.
6703 /* read until we run out of string, or we find the terminator */
6704 for (; s < PL_bufend; s++,to++) {
6705 /* embedded newlines increment the line count */
6706 if (*s == '\n' && !PL_rsfp)
6707 CopLINE_inc(PL_curcop);
6708 /* backslashes can escape the open or closing characters */
6709 if (*s == '\\' && s+1 < PL_bufend) {
6711 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6716 /* allow nested opens and closes */
6717 else if (*s == PL_multi_close && --brackets <= 0)
6719 else if (*s == PL_multi_open)
6721 else if (!has_utf8 && (*s & 0x80) && UTF)
6726 /* terminate the copied string and update the sv's end-of-string */
6728 SvCUR_set(sv, to - SvPVX(sv));
6731 * this next chunk reads more into the buffer if we're not done yet
6735 break; /* handle case where we are done yet :-) */
6737 #ifndef PERL_STRICT_CR
6738 if (to - SvPVX(sv) >= 2) {
6739 if ((to[-2] == '\r' && to[-1] == '\n') ||
6740 (to[-2] == '\n' && to[-1] == '\r'))
6744 SvCUR_set(sv, to - SvPVX(sv));
6746 else if (to[-1] == '\r')
6749 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6753 /* if we're out of file, or a read fails, bail and reset the current
6754 line marker so we can report where the unterminated string began
6757 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6759 CopLINE_set(PL_curcop, PL_multi_start);
6762 /* we read a line, so increment our line counter */
6763 CopLINE_inc(PL_curcop);
6765 /* update debugger info */
6766 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6767 SV *sv = NEWSV(88,0);
6769 sv_upgrade(sv, SVt_PVMG);
6770 sv_setsv(sv,PL_linestr);
6771 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6774 /* having changed the buffer, we must update PL_bufend */
6775 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6778 /* at this point, we have successfully read the delimited string */
6781 sv_catpvn(sv, s, 1);
6784 PL_multi_end = CopLINE(PL_curcop);
6787 /* if we allocated too much space, give some back */
6788 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6789 SvLEN_set(sv, SvCUR(sv) + 1);
6790 Renew(SvPVX(sv), SvLEN(sv), char);
6793 /* decide whether this is the first or second quoted string we've read
6806 takes: pointer to position in buffer
6807 returns: pointer to new position in buffer
6808 side-effects: builds ops for the constant in yylval.op
6810 Read a number in any of the formats that Perl accepts:
6812 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6813 [\d_]+(\.[\d_]*)?[Ee](\d+)
6815 Underbars (_) are allowed in decimal numbers. If -w is on,
6816 underbars before a decimal point must be at three digit intervals.
6818 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6821 If it reads a number without a decimal point or an exponent, it will
6822 try converting the number to an integer and see if it can do so
6823 without loss of precision.
6827 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6829 register char *s = start; /* current position in buffer */
6830 register char *d; /* destination in temp buffer */
6831 register char *e; /* end of temp buffer */
6832 NV nv; /* number read, as a double */
6833 SV *sv = Nullsv; /* place to put the converted number */
6834 bool floatit; /* boolean: int or float? */
6835 char *lastub = 0; /* position of last underbar */
6836 static char number_too_long[] = "Number too long";
6838 /* We use the first character to decide what type of number this is */
6842 Perl_croak(aTHX_ "panic: scan_num");
6844 /* if it starts with a 0, it could be an octal number, a decimal in
6845 0.13 disguise, or a hexadecimal number, or a binary number. */
6849 u holds the "number so far"
6850 shift the power of 2 of the base
6851 (hex == 4, octal == 3, binary == 1)
6852 overflowed was the number more than we can hold?
6854 Shift is used when we add a digit. It also serves as an "are
6855 we in octal/hex/binary?" indicator to disallow hex characters
6861 bool overflowed = FALSE;
6862 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6863 static char* bases[5] = { "", "binary", "", "octal",
6865 static char* Bases[5] = { "", "Binary", "", "Octal",
6867 static char *maxima[5] = { "",
6868 "0b11111111111111111111111111111111",
6872 char *base, *Base, *max;
6878 } else if (s[1] == 'b') {
6882 /* check for a decimal in disguise */
6883 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6885 /* so it must be octal */
6889 base = bases[shift];
6890 Base = Bases[shift];
6891 max = maxima[shift];
6893 /* read the rest of the number */
6895 /* x is used in the overflow test,
6896 b is the digit we're adding on. */
6901 /* if we don't mention it, we're done */
6910 /* 8 and 9 are not octal */
6913 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6917 case '2': case '3': case '4':
6918 case '5': case '6': case '7':
6920 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6924 b = *s++ & 15; /* ASCII digit -> value of digit */
6928 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6929 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6930 /* make sure they said 0x */
6935 /* Prepare to put the digit we have onto the end
6936 of the number so far. We check for overflows.
6941 x = u << shift; /* make room for the digit */
6943 if ((x >> shift) != u
6944 && !(PL_hints & HINT_NEW_BINARY)) {
6947 if (ckWARN_d(WARN_OVERFLOW))
6948 Perl_warner(aTHX_ WARN_OVERFLOW,
6949 "Integer overflow in %s number",
6952 u = x | b; /* add the digit to the end */
6955 n *= nvshift[shift];
6956 /* If an NV has not enough bits in its
6957 * mantissa to represent an UV this summing of
6958 * small low-order numbers is a waste of time
6959 * (because the NV cannot preserve the
6960 * low-order bits anyway): we could just
6961 * remember when did we overflow and in the
6962 * end just multiply n by the right
6970 /* if we get here, we had success: make a scalar value from
6976 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6977 Perl_warner(aTHX_ WARN_PORTABLE,
6978 "%s number > %s non-portable",
6984 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6985 Perl_warner(aTHX_ WARN_PORTABLE,
6986 "%s number > %s non-portable",
6991 if (PL_hints & HINT_NEW_BINARY)
6992 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
6997 handle decimal numbers.
6998 we're also sent here when we read a 0 as the first digit
7000 case '1': case '2': case '3': case '4': case '5':
7001 case '6': case '7': case '8': case '9': case '.':
7004 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7007 /* read next group of digits and _ and copy into d */
7008 while (isDIGIT(*s) || *s == '_') {
7009 /* skip underscores, checking for misplaced ones
7013 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7014 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7018 /* check for end of fixed-length buffer */
7020 Perl_croak(aTHX_ number_too_long);
7021 /* if we're ok, copy the character */
7026 /* final misplaced underbar check */
7027 if (lastub && s - lastub != 3) {
7028 if (ckWARN(WARN_SYNTAX))
7029 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7032 /* read a decimal portion if there is one. avoid
7033 3..5 being interpreted as the number 3. followed
7036 if (*s == '.' && s[1] != '.') {
7040 /* copy, ignoring underbars, until we run out of
7041 digits. Note: no misplaced underbar checks!
7043 for (; isDIGIT(*s) || *s == '_'; s++) {
7044 /* fixed length buffer check */
7046 Perl_croak(aTHX_ number_too_long);
7050 if (*s == '.' && isDIGIT(s[1])) {
7051 /* oops, it's really a v-string, but without the "v" */
7057 /* read exponent part, if present */
7058 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7062 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7063 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7065 /* allow positive or negative exponent */
7066 if (*s == '+' || *s == '-')
7069 /* read digits of exponent (no underbars :-) */
7070 while (isDIGIT(*s)) {
7072 Perl_croak(aTHX_ number_too_long);
7077 /* terminate the string */
7080 /* make an sv from the string */
7083 #if defined(Strtol) && defined(Strtoul)
7086 strtol/strtoll sets errno to ERANGE if the number is too big
7087 for an integer. We try to do an integer conversion first
7088 if no characters indicating "float" have been found.
7095 if (*PL_tokenbuf == '-')
7096 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7098 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7100 floatit = TRUE; /* Probably just too large. */
7101 else if (*PL_tokenbuf == '-')
7103 else if (uv <= IV_MAX)
7104 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7109 nv = Atof(PL_tokenbuf);
7114 No working strtou?ll?.
7116 Unfortunately atol() doesn't do range checks (returning
7117 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7118 everywhere [1], so we cannot use use atol() (or atoll()).
7119 If we could, they would be used, as Atol(), very much like
7120 Strtol() and Strtoul() are used above.
7122 [1] XXX Configure test needed to check for atol()
7123 (and atoll()) overflow behaviour XXX
7127 We need to do this the hard way. */
7129 nv = Atof(PL_tokenbuf);
7131 /* See if we can make do with an integer value without loss of
7132 precision. We use U_V to cast to a UV, because some
7133 compilers have issues. Then we try casting it back and see
7134 if it was the same [1]. We only do this if we know we
7135 specifically read an integer. If floatit is true, then we
7136 don't need to do the conversion at all.
7138 [1] Note that this is lossy if our NVs cannot preserve our
7139 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7140 and NV_PRESERVES_UV_BITS (a number), but in general we really
7141 do hope all such potentially lossy platforms have strtou?ll?
7142 to do a lossless IV/UV conversion.
7144 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7145 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7146 as NV_DIG and NV_MANT_DIG)?
7152 if (!floatit && (NV)uv == nv) {
7154 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7162 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7163 (PL_hints & HINT_NEW_INTEGER) )
7164 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7165 (floatit ? "float" : "integer"),
7169 /* if it starts with a v, it could be a v-string */
7175 while (isDIGIT(*pos) || *pos == '_')
7177 if (!isALPHA(*pos)) {
7179 U8 tmpbuf[UTF8_MAXLEN+1];
7182 s++; /* get past 'v' */
7185 sv_setpvn(sv, "", 0);
7188 if (*s == '0' && isDIGIT(s[1]))
7189 yyerror("Octal number in vector unsupported");
7192 /* this is atoi() that tolerates underscores */
7195 while (--end >= s) {
7200 rev += (*end - '0') * mult;
7202 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7203 Perl_warner(aTHX_ WARN_OVERFLOW,
7204 "Integer overflow in decimal number");
7207 tmpend = uv_to_utf8(tmpbuf, rev);
7208 utf8 = utf8 || rev > 127;
7209 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7210 if (*pos == '.' && isDIGIT(pos[1]))
7216 while (isDIGIT(*pos) || *pos == '_')
7225 sv_utf8_downgrade(sv, TRUE);
7232 /* make the op for the constant and return */
7235 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7237 lvalp->opval = Nullop;
7243 S_scan_formline(pTHX_ register char *s)
7247 SV *stuff = newSVpvn("",0);
7248 bool needargs = FALSE;
7251 if (*s == '.' || *s == /*{*/'}') {
7253 #ifdef PERL_STRICT_CR
7254 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7256 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7258 if (*t == '\n' || t == PL_bufend)
7261 if (PL_in_eval && !PL_rsfp) {
7262 eol = strchr(s,'\n');
7267 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7269 for (t = s; t < eol; t++) {
7270 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7272 goto enough; /* ~~ must be first line in formline */
7274 if (*t == '@' || *t == '^')
7277 sv_catpvn(stuff, s, eol-s);
7278 #ifndef PERL_STRICT_CR
7279 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7280 char *end = SvPVX(stuff) + SvCUR(stuff);
7289 s = filter_gets(PL_linestr, PL_rsfp, 0);
7290 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7291 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7294 yyerror("Format not terminated");
7304 PL_lex_state = LEX_NORMAL;
7305 PL_nextval[PL_nexttoke].ival = 0;
7309 PL_lex_state = LEX_FORMLINE;
7310 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7312 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7316 SvREFCNT_dec(stuff);
7317 PL_lex_formbrack = 0;
7328 PL_cshlen = strlen(PL_cshname);
7333 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7335 I32 oldsavestack_ix = PL_savestack_ix;
7336 CV* outsidecv = PL_compcv;
7340 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7342 SAVEI32(PL_subline);
7343 save_item(PL_subname);
7346 SAVESPTR(PL_comppad_name);
7347 SAVESPTR(PL_compcv);
7348 SAVEI32(PL_comppad_name_fill);
7349 SAVEI32(PL_min_intro_pending);
7350 SAVEI32(PL_max_intro_pending);
7351 SAVEI32(PL_pad_reset_pending);
7353 PL_compcv = (CV*)NEWSV(1104,0);
7354 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7355 CvFLAGS(PL_compcv) |= flags;
7357 PL_comppad = newAV();
7358 av_push(PL_comppad, Nullsv);
7359 PL_curpad = AvARRAY(PL_comppad);
7360 PL_comppad_name = newAV();
7361 PL_comppad_name_fill = 0;
7362 PL_min_intro_pending = 0;
7364 PL_subline = CopLINE(PL_curcop);
7366 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7367 PL_curpad[0] = (SV*)newAV();
7368 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7369 #endif /* USE_THREADS */
7371 comppadlist = newAV();
7372 AvREAL_off(comppadlist);
7373 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7374 av_store(comppadlist, 1, (SV*)PL_comppad);
7376 CvPADLIST(PL_compcv) = comppadlist;
7377 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7379 CvOWNER(PL_compcv) = 0;
7380 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7381 MUTEX_INIT(CvMUTEXP(PL_compcv));
7382 #endif /* USE_THREADS */
7384 return oldsavestack_ix;
7388 Perl_yywarn(pTHX_ char *s)
7390 PL_in_eval |= EVAL_WARNONLY;
7392 PL_in_eval &= ~EVAL_WARNONLY;
7397 Perl_yyerror(pTHX_ char *s)
7400 char *context = NULL;
7404 if (!yychar || (yychar == ';' && !PL_rsfp))
7406 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7407 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7408 while (isSPACE(*PL_oldoldbufptr))
7410 context = PL_oldoldbufptr;
7411 contlen = PL_bufptr - PL_oldoldbufptr;
7413 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7414 PL_oldbufptr != PL_bufptr) {
7415 while (isSPACE(*PL_oldbufptr))
7417 context = PL_oldbufptr;
7418 contlen = PL_bufptr - PL_oldbufptr;
7420 else if (yychar > 255)
7421 where = "next token ???";
7422 #ifdef USE_PURE_BISON
7423 /* GNU Bison sets the value -2 */
7424 else if (yychar == -2) {
7426 else if ((yychar & 127) == 127) {
7428 if (PL_lex_state == LEX_NORMAL ||
7429 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7430 where = "at end of line";
7431 else if (PL_lex_inpat)
7432 where = "within pattern";
7434 where = "within string";
7437 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7439 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7440 else if (isPRINT_LC(yychar))
7441 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7443 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7444 where = SvPVX(where_sv);
7446 msg = sv_2mortal(newSVpv(s, 0));
7447 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7448 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7450 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7452 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7453 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7454 Perl_sv_catpvf(aTHX_ msg,
7455 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7456 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7459 if (PL_in_eval & EVAL_WARNONLY)
7460 Perl_warn(aTHX_ "%"SVf, msg);
7463 if (PL_error_count >= 10) {
7464 if (PL_in_eval && SvCUR(ERRSV))
7465 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7466 ERRSV, CopFILE(PL_curcop));
7468 Perl_croak(aTHX_ "%s has too many errors.\n",
7469 CopFILE(PL_curcop));
7472 PL_in_my_stash = Nullhv;
7477 S_swallow_bom(pTHX_ U8 *s)
7480 slen = SvCUR(PL_linestr);
7484 /* UTF-16 little-endian */
7485 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7486 Perl_croak(aTHX_ "Unsupported script encoding");
7487 #ifndef PERL_NO_UTF16_FILTER
7488 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7490 if (PL_bufend > (char*)s) {
7494 filter_add(utf16rev_textfilter, NULL);
7495 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7496 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7497 PL_bufend - (char*)s - 1,
7499 Copy(news, s, newlen, U8);
7500 SvCUR_set(PL_linestr, newlen);
7501 PL_bufend = SvPVX(PL_linestr) + newlen;
7502 news[newlen++] = '\0';
7506 Perl_croak(aTHX_ "Unsupported script encoding");
7511 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7512 #ifndef PERL_NO_UTF16_FILTER
7513 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7515 if (PL_bufend > (char *)s) {
7519 filter_add(utf16_textfilter, NULL);
7520 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7521 PL_bufend = (char*)utf16_to_utf8(s, news,
7522 PL_bufend - (char*)s,
7524 Copy(news, s, newlen, U8);
7525 SvCUR_set(PL_linestr, newlen);
7526 PL_bufend = SvPVX(PL_linestr) + newlen;
7527 news[newlen++] = '\0';
7531 Perl_croak(aTHX_ "Unsupported script encoding");
7536 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7537 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7542 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7543 s[2] == 0xFE && s[3] == 0xFF)
7545 Perl_croak(aTHX_ "Unsupported script encoding");
7557 * Restore a source filter.
7561 restore_rsfp(pTHXo_ void *f)
7563 PerlIO *fp = (PerlIO*)f;
7565 if (PL_rsfp == PerlIO_stdin())
7566 PerlIO_clearerr(PL_rsfp);
7567 else if (PL_rsfp && (PL_rsfp != fp))
7568 PerlIO_close(PL_rsfp);
7572 #ifndef PERL_NO_UTF16_FILTER
7574 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7576 I32 count = FILTER_READ(idx+1, sv, maxlen);
7581 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7582 if (!*SvPV_nolen(sv))
7583 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7586 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7587 sv_usepvn(sv, (char*)tmps, tend - tmps);
7593 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7595 I32 count = FILTER_READ(idx+1, sv, maxlen);
7600 if (!*SvPV_nolen(sv))
7601 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7604 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7605 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7606 sv_usepvn(sv, (char*)tmps, tend - tmps);