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
1401 * escapes will be longer than any UT-F8 sequence
1402 * they can end up as. */
1404 if (!has_utf8 && (to_be_utf8 || uv > 255)) {
1405 /* Might need to recode whatever we have
1406 * accumulated so far if it contains any
1409 * (Can't we keep track of that and avoid
1410 * this rescan? --jhi)
1415 for (c = SvPVX(sv); c < d; c++) {
1416 if (UTF8_IS_CONTINUED(*c))
1420 char *old_pvx = SvPVX(sv);
1422 U8 tmpbuf[UTF8_MAXLEN+1];
1426 SvCUR(sv) + hicount + 1) +
1434 if (UTF8_IS_CONTINUED(*src)) {
1435 tmpend = uv_to_utf8(tmpbuf, (U8)*src--);
1436 dst -= tmpend - tmpbuf;
1437 Copy((char *)tmpbuf, dst+1,
1438 tmpend - tmpbuf, char);
1447 if (to_be_utf8 || (has_utf8 && uv > 127) || uv > 255) {
1448 d = (char*)uv_to_utf8((U8*)d, uv);
1460 /* \N{latin small letter a} is a named character */
1464 char* e = strchr(s, '}');
1470 yyerror("Missing right brace on \\N{}");
1474 res = newSVpvn(s + 1, e - s - 1);
1475 res = new_constant( Nullch, 0, "charnames",
1476 res, Nullsv, "\\N{...}" );
1477 str = SvPV(res,len);
1478 if (!has_utf8 && SvUTF8(res)) {
1479 char *ostart = SvPVX(sv);
1480 SvCUR_set(sv, d - ostart);
1483 sv_utf8_upgrade(sv);
1484 /* this just broke our allocation above... */
1485 SvGROW(sv, send - start);
1486 d = SvPVX(sv) + SvCUR(sv);
1489 if (len > e - s + 4) {
1490 char *odest = SvPVX(sv);
1492 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1493 d = SvPVX(sv) + (d - odest);
1495 Copy(str, d, len, char);
1502 yyerror("Missing braces on \\N{}");
1505 /* \c is a control character */
1522 /* printf-style backslashes, formfeeds, newlines, etc */
1540 *d++ = '\047'; /* CP 1047 */
1543 *d++ = '\057'; /* CP 1047 */
1557 } /* end if (backslash) */
1559 /* (now in tr/// code again) */
1561 if (UTF8_IS_CONTINUED(*s) && (this_utf8 || has_utf8)) {
1562 STRLEN len = (STRLEN) -1;
1565 uv = utf8_to_uv((U8*)s, send - s, &len, 0);
1567 if (len == (STRLEN)-1) {
1568 /* Illegal UTF8 (a high-bit byte), make it valid. */
1569 char *old_pvx = SvPVX(sv);
1570 /* need space for one extra char (NOTE: SvCUR() not set here) */
1571 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1572 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1583 } /* while loop to process each character */
1585 /* terminate the string and set up the sv */
1587 SvCUR_set(sv, d - SvPVX(sv));
1592 /* shrink the sv if we allocated more than we used */
1593 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1594 SvLEN_set(sv, SvCUR(sv) + 1);
1595 Renew(SvPVX(sv), SvLEN(sv), char);
1598 /* return the substring (via yylval) only if we parsed anything */
1599 if (s > PL_bufptr) {
1600 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1601 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1603 ( PL_lex_inwhat == OP_TRANS
1605 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1608 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1615 * Returns TRUE if there's more to the expression (e.g., a subscript),
1618 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1620 * ->[ and ->{ return TRUE
1621 * { and [ outside a pattern are always subscripts, so return TRUE
1622 * if we're outside a pattern and it's not { or [, then return FALSE
1623 * if we're in a pattern and the first char is a {
1624 * {4,5} (any digits around the comma) returns FALSE
1625 * if we're in a pattern and the first char is a [
1627 * [SOMETHING] has a funky algorithm to decide whether it's a
1628 * character class or not. It has to deal with things like
1629 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1630 * anything else returns TRUE
1633 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1636 S_intuit_more(pTHX_ register char *s)
1638 if (PL_lex_brackets)
1640 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1642 if (*s != '{' && *s != '[')
1647 /* In a pattern, so maybe we have {n,m}. */
1664 /* On the other hand, maybe we have a character class */
1667 if (*s == ']' || *s == '^')
1670 /* this is terrifying, and it works */
1671 int weight = 2; /* let's weigh the evidence */
1673 unsigned char un_char = 255, last_un_char;
1674 char *send = strchr(s,']');
1675 char tmpbuf[sizeof PL_tokenbuf * 4];
1677 if (!send) /* has to be an expression */
1680 Zero(seen,256,char);
1683 else if (isDIGIT(*s)) {
1685 if (isDIGIT(s[1]) && s[2] == ']')
1691 for (; s < send; s++) {
1692 last_un_char = un_char;
1693 un_char = (unsigned char)*s;
1698 weight -= seen[un_char] * 10;
1699 if (isALNUM_lazy_if(s+1,UTF)) {
1700 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1701 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1706 else if (*s == '$' && s[1] &&
1707 strchr("[#!%*<>()-=",s[1])) {
1708 if (/*{*/ strchr("])} =",s[2]))
1717 if (strchr("wds]",s[1]))
1719 else if (seen['\''] || seen['"'])
1721 else if (strchr("rnftbxcav",s[1]))
1723 else if (isDIGIT(s[1])) {
1725 while (s[1] && isDIGIT(s[1]))
1735 if (strchr("aA01! ",last_un_char))
1737 if (strchr("zZ79~",s[1]))
1739 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1740 weight -= 5; /* cope with negative subscript */
1743 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1744 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1749 if (keyword(tmpbuf, d - tmpbuf))
1752 if (un_char == last_un_char + 1)
1754 weight -= seen[un_char];
1759 if (weight >= 0) /* probably a character class */
1769 * Does all the checking to disambiguate
1771 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1772 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1774 * First argument is the stuff after the first token, e.g. "bar".
1776 * Not a method if bar is a filehandle.
1777 * Not a method if foo is a subroutine prototyped to take a filehandle.
1778 * Not a method if it's really "Foo $bar"
1779 * Method if it's "foo $bar"
1780 * Not a method if it's really "print foo $bar"
1781 * Method if it's really "foo package::" (interpreted as package->foo)
1782 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1783 * Not a method if bar is a filehandle or package, but is quoted with
1788 S_intuit_method(pTHX_ char *start, GV *gv)
1790 char *s = start + (*start == '$');
1791 char tmpbuf[sizeof PL_tokenbuf];
1799 if ((cv = GvCVu(gv))) {
1800 char *proto = SvPVX(cv);
1810 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1811 /* start is the beginning of the possible filehandle/object,
1812 * and s is the end of it
1813 * tmpbuf is a copy of it
1816 if (*start == '$') {
1817 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1822 return *s == '(' ? FUNCMETH : METHOD;
1824 if (!keyword(tmpbuf, len)) {
1825 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1830 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1831 if (indirgv && GvCVu(indirgv))
1833 /* filehandle or package name makes it a method */
1834 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1836 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1837 return 0; /* no assumptions -- "=>" quotes bearword */
1839 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1840 newSVpvn(tmpbuf,len));
1841 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1845 return *s == '(' ? FUNCMETH : METHOD;
1853 * Return a string of Perl code to load the debugger. If PERL5DB
1854 * is set, it will return the contents of that, otherwise a
1855 * compile-time require of perl5db.pl.
1862 char *pdb = PerlEnv_getenv("PERL5DB");
1866 SETERRNO(0,SS$_NORMAL);
1867 return "BEGIN { require 'perl5db.pl' }";
1873 /* Encoded script support. filter_add() effectively inserts a
1874 * 'pre-processing' function into the current source input stream.
1875 * Note that the filter function only applies to the current source file
1876 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1878 * The datasv parameter (which may be NULL) can be used to pass
1879 * private data to this instance of the filter. The filter function
1880 * can recover the SV using the FILTER_DATA macro and use it to
1881 * store private buffers and state information.
1883 * The supplied datasv parameter is upgraded to a PVIO type
1884 * and the IoDIRP/IoANY field is used to store the function pointer,
1885 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1886 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1887 * private use must be set using malloc'd pointers.
1891 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1896 if (!PL_rsfp_filters)
1897 PL_rsfp_filters = newAV();
1899 datasv = NEWSV(255,0);
1900 if (!SvUPGRADE(datasv, SVt_PVIO))
1901 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1902 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1903 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1904 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1905 funcp, SvPV_nolen(datasv)));
1906 av_unshift(PL_rsfp_filters, 1);
1907 av_store(PL_rsfp_filters, 0, datasv) ;
1912 /* Delete most recently added instance of this filter function. */
1914 Perl_filter_del(pTHX_ filter_t funcp)
1917 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1918 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1920 /* if filter is on top of stack (usual case) just pop it off */
1921 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1922 if (IoANY(datasv) == (void *)funcp) {
1923 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1924 IoANY(datasv) = (void *)NULL;
1925 sv_free(av_pop(PL_rsfp_filters));
1929 /* we need to search for the correct entry and clear it */
1930 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1934 /* Invoke the n'th filter function for the current rsfp. */
1936 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1939 /* 0 = read one text line */
1944 if (!PL_rsfp_filters)
1946 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1947 /* Provide a default input filter to make life easy. */
1948 /* Note that we append to the line. This is handy. */
1949 DEBUG_P(PerlIO_printf(Perl_debug_log,
1950 "filter_read %d: from rsfp\n", idx));
1954 int old_len = SvCUR(buf_sv) ;
1956 /* ensure buf_sv is large enough */
1957 SvGROW(buf_sv, old_len + maxlen) ;
1958 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1959 if (PerlIO_error(PL_rsfp))
1960 return -1; /* error */
1962 return 0 ; /* end of file */
1964 SvCUR_set(buf_sv, old_len + len) ;
1967 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1968 if (PerlIO_error(PL_rsfp))
1969 return -1; /* error */
1971 return 0 ; /* end of file */
1974 return SvCUR(buf_sv);
1976 /* Skip this filter slot if filter has been deleted */
1977 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1978 DEBUG_P(PerlIO_printf(Perl_debug_log,
1979 "filter_read %d: skipped (filter deleted)\n",
1981 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1983 /* Get function pointer hidden within datasv */
1984 funcp = (filter_t)IoANY(datasv);
1985 DEBUG_P(PerlIO_printf(Perl_debug_log,
1986 "filter_read %d: via function %p (%s)\n",
1987 idx, funcp, SvPV_nolen(datasv)));
1988 /* Call function. The function is expected to */
1989 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1990 /* Return: <0:error, =0:eof, >0:not eof */
1991 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1995 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1997 #ifdef PERL_CR_FILTER
1998 if (!PL_rsfp_filters) {
1999 filter_add(S_cr_textfilter,NULL);
2002 if (PL_rsfp_filters) {
2005 SvCUR_set(sv, 0); /* start with empty line */
2006 if (FILTER_READ(0, sv, 0) > 0)
2007 return ( SvPVX(sv) ) ;
2012 return (sv_gets(sv, fp, append));
2016 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2020 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2024 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2025 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2027 return GvHV(gv); /* Foo:: */
2030 /* use constant CLASS => 'MyClass' */
2031 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2033 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2034 pkgname = SvPV_nolen(sv);
2038 return gv_stashpv(pkgname, FALSE);
2042 static char* exp_name[] =
2043 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2044 "ATTRTERM", "TERMBLOCK"
2051 Works out what to call the token just pulled out of the input
2052 stream. The yacc parser takes care of taking the ops we return and
2053 stitching them into a tree.
2059 if read an identifier
2060 if we're in a my declaration
2061 croak if they tried to say my($foo::bar)
2062 build the ops for a my() declaration
2063 if it's an access to a my() variable
2064 are we in a sort block?
2065 croak if my($a); $a <=> $b
2066 build ops for access to a my() variable
2067 if in a dq string, and they've said @foo and we can't find @foo
2069 build ops for a bareword
2070 if we already built the token before, use it.
2073 #ifdef USE_PURE_BISON
2075 #pragma segment Perl_yylex_r
2078 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2083 yylval_pointer[yyactlevel] = lvalp;
2084 yychar_pointer[yyactlevel] = lcharp;
2085 if (yyactlevel >= YYMAXLEVEL)
2086 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2088 r = Perl_yylex(aTHX);
2097 #pragma segment Perl_yylex
2109 /* check if there's an identifier for us to look at */
2110 if (PL_pending_ident) {
2111 /* pit holds the identifier we read and pending_ident is reset */
2112 char pit = PL_pending_ident;
2113 PL_pending_ident = 0;
2115 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2116 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2118 /* if we're in a my(), we can't allow dynamics here.
2119 $foo'bar has already been turned into $foo::bar, so
2120 just check for colons.
2122 if it's a legal name, the OP is a PADANY.
2125 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2126 if (strchr(PL_tokenbuf,':'))
2127 yyerror(Perl_form(aTHX_ "No package name allowed for "
2128 "variable %s in \"our\"",
2130 tmp = pad_allocmy(PL_tokenbuf);
2133 if (strchr(PL_tokenbuf,':'))
2134 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2136 yylval.opval = newOP(OP_PADANY, 0);
2137 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2143 build the ops for accesses to a my() variable.
2145 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2146 then used in a comparison. This catches most, but not
2147 all cases. For instance, it catches
2148 sort { my($a); $a <=> $b }
2150 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2151 (although why you'd do that is anyone's guess).
2154 if (!strchr(PL_tokenbuf,':')) {
2156 /* Check for single character per-thread SVs */
2157 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2158 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2159 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2161 yylval.opval = newOP(OP_THREADSV, 0);
2162 yylval.opval->op_targ = tmp;
2165 #endif /* USE_THREADS */
2166 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2167 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2168 /* might be an "our" variable" */
2169 if (SvFLAGS(namesv) & SVpad_OUR) {
2170 /* build ops for a bareword */
2171 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2172 sv_catpvn(sym, "::", 2);
2173 sv_catpv(sym, PL_tokenbuf+1);
2174 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2175 yylval.opval->op_private = OPpCONST_ENTERED;
2176 gv_fetchpv(SvPVX(sym),
2178 ? (GV_ADDMULTI | GV_ADDINEVAL)
2181 ((PL_tokenbuf[0] == '$') ? SVt_PV
2182 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2187 /* if it's a sort block and they're naming $a or $b */
2188 if (PL_last_lop_op == OP_SORT &&
2189 PL_tokenbuf[0] == '$' &&
2190 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2193 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2194 d < PL_bufend && *d != '\n';
2197 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2198 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2204 yylval.opval = newOP(OP_PADANY, 0);
2205 yylval.opval->op_targ = tmp;
2211 Whine if they've said @foo in a doublequoted string,
2212 and @foo isn't a variable we can find in the symbol
2215 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2216 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2217 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2218 && ckWARN(WARN_AMBIGUOUS))
2220 /* Downgraded from fatal to warning 20000522 mjd */
2221 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2222 "Possible unintended interpolation of %s in string",
2227 /* build ops for a bareword */
2228 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2229 yylval.opval->op_private = OPpCONST_ENTERED;
2230 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2231 ((PL_tokenbuf[0] == '$') ? SVt_PV
2232 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2237 /* no identifier pending identification */
2239 switch (PL_lex_state) {
2241 case LEX_NORMAL: /* Some compilers will produce faster */
2242 case LEX_INTERPNORMAL: /* code if we comment these out. */
2246 /* when we've already built the next token, just pull it out of the queue */
2249 yylval = PL_nextval[PL_nexttoke];
2251 PL_lex_state = PL_lex_defer;
2252 PL_expect = PL_lex_expect;
2253 PL_lex_defer = LEX_NORMAL;
2255 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2256 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2257 (IV)PL_nexttype[PL_nexttoke]); })
2259 return(PL_nexttype[PL_nexttoke]);
2261 /* interpolated case modifiers like \L \U, including \Q and \E.
2262 when we get here, PL_bufptr is at the \
2264 case LEX_INTERPCASEMOD:
2266 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2267 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2269 /* handle \E or end of string */
2270 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2274 if (PL_lex_casemods) {
2275 oldmod = PL_lex_casestack[--PL_lex_casemods];
2276 PL_lex_casestack[PL_lex_casemods] = '\0';
2278 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2280 PL_lex_state = LEX_INTERPCONCAT;
2284 if (PL_bufptr != PL_bufend)
2286 PL_lex_state = LEX_INTERPCONCAT;
2290 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2291 "### Saw case modifier at '%s'\n", PL_bufptr); })
2293 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2294 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2295 if (strchr("LU", *s) &&
2296 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2298 PL_lex_casestack[--PL_lex_casemods] = '\0';
2301 if (PL_lex_casemods > 10) {
2302 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2303 if (newlb != PL_lex_casestack) {
2305 PL_lex_casestack = newlb;
2308 PL_lex_casestack[PL_lex_casemods++] = *s;
2309 PL_lex_casestack[PL_lex_casemods] = '\0';
2310 PL_lex_state = LEX_INTERPCONCAT;
2311 PL_nextval[PL_nexttoke].ival = 0;
2314 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2316 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2318 PL_nextval[PL_nexttoke].ival = OP_LC;
2320 PL_nextval[PL_nexttoke].ival = OP_UC;
2322 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2324 Perl_croak(aTHX_ "panic: yylex");
2327 if (PL_lex_starts) {
2336 case LEX_INTERPPUSH:
2337 return sublex_push();
2339 case LEX_INTERPSTART:
2340 if (PL_bufptr == PL_bufend)
2341 return sublex_done();
2342 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2343 "### Interpolated variable at '%s'\n", PL_bufptr); })
2345 PL_lex_dojoin = (*PL_bufptr == '@');
2346 PL_lex_state = LEX_INTERPNORMAL;
2347 if (PL_lex_dojoin) {
2348 PL_nextval[PL_nexttoke].ival = 0;
2351 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2352 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2353 force_next(PRIVATEREF);
2355 force_ident("\"", '$');
2356 #endif /* USE_THREADS */
2357 PL_nextval[PL_nexttoke].ival = 0;
2359 PL_nextval[PL_nexttoke].ival = 0;
2361 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2364 if (PL_lex_starts++) {
2370 case LEX_INTERPENDMAYBE:
2371 if (intuit_more(PL_bufptr)) {
2372 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2378 if (PL_lex_dojoin) {
2379 PL_lex_dojoin = FALSE;
2380 PL_lex_state = LEX_INTERPCONCAT;
2383 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2384 && SvEVALED(PL_lex_repl))
2386 if (PL_bufptr != PL_bufend)
2387 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2388 PL_lex_repl = Nullsv;
2391 case LEX_INTERPCONCAT:
2393 if (PL_lex_brackets)
2394 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2396 if (PL_bufptr == PL_bufend)
2397 return sublex_done();
2399 if (SvIVX(PL_linestr) == '\'') {
2400 SV *sv = newSVsv(PL_linestr);
2403 else if ( PL_hints & HINT_NEW_RE )
2404 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2405 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2409 s = scan_const(PL_bufptr);
2411 PL_lex_state = LEX_INTERPCASEMOD;
2413 PL_lex_state = LEX_INTERPSTART;
2416 if (s != PL_bufptr) {
2417 PL_nextval[PL_nexttoke] = yylval;
2420 if (PL_lex_starts++)
2430 PL_lex_state = LEX_NORMAL;
2431 s = scan_formline(PL_bufptr);
2432 if (!PL_lex_formbrack)
2438 PL_oldoldbufptr = PL_oldbufptr;
2441 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2442 exp_name[PL_expect], s);
2448 if (isIDFIRST_lazy_if(s,UTF))
2450 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2453 goto fake_eof; /* emulate EOF on ^D or ^Z */
2458 if (PL_lex_brackets)
2459 yyerror("Missing right curly or square bracket");
2460 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2461 "### Tokener got EOF\n");
2465 if (s++ < PL_bufend)
2466 goto retry; /* ignore stray nulls */
2469 if (!PL_in_eval && !PL_preambled) {
2470 PL_preambled = TRUE;
2471 sv_setpv(PL_linestr,incl_perldb());
2472 if (SvCUR(PL_linestr))
2473 sv_catpv(PL_linestr,";");
2475 while(AvFILLp(PL_preambleav) >= 0) {
2476 SV *tmpsv = av_shift(PL_preambleav);
2477 sv_catsv(PL_linestr, tmpsv);
2478 sv_catpv(PL_linestr, ";");
2481 sv_free((SV*)PL_preambleav);
2482 PL_preambleav = NULL;
2484 if (PL_minus_n || PL_minus_p) {
2485 sv_catpv(PL_linestr, "LINE: while (<>) {");
2487 sv_catpv(PL_linestr,"chomp;");
2489 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2491 GvIMPORTED_AV_on(gv);
2493 if (strchr("/'\"", *PL_splitstr)
2494 && strchr(PL_splitstr + 1, *PL_splitstr))
2495 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2498 s = "'~#\200\1'"; /* surely one char is unused...*/
2499 while (s[1] && strchr(PL_splitstr, *s)) s++;
2501 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2502 "q" + (delim == '\''), delim);
2503 for (s = PL_splitstr; *s; s++) {
2505 sv_catpvn(PL_linestr, "\\", 1);
2506 sv_catpvn(PL_linestr, s, 1);
2508 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2512 sv_catpv(PL_linestr,"@F=split(' ');");
2515 sv_catpv(PL_linestr, "\n");
2516 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2517 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2518 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2519 SV *sv = NEWSV(85,0);
2521 sv_upgrade(sv, SVt_PVMG);
2522 sv_setsv(sv,PL_linestr);
2523 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2528 bool bof = PL_rsfp ? TRUE : FALSE;
2530 #ifdef PERLIO_IS_STDIO
2531 # ifdef __GNU_LIBRARY__
2532 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2533 # define FTELL_FOR_PIPE_IS_BROKEN
2537 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2538 # define FTELL_FOR_PIPE_IS_BROKEN
2543 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2544 /* This loses the possibility to detect the bof
2545 * situation on perl -P when the libc5 is being used.
2546 * Workaround? Maybe attach some extra state to PL_rsfp?
2549 bof = PerlIO_tell(PL_rsfp) == 0;
2551 bof = PerlIO_tell(PL_rsfp) == 0;
2554 s = filter_gets(PL_linestr, PL_rsfp, 0);
2558 if (PL_preprocess && !PL_in_eval)
2559 (void)PerlProc_pclose(PL_rsfp);
2560 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2561 PerlIO_clearerr(PL_rsfp);
2563 (void)PerlIO_close(PL_rsfp);
2565 PL_doextract = FALSE;
2567 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2568 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2569 sv_catpv(PL_linestr,";}");
2570 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2571 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2572 PL_minus_n = PL_minus_p = 0;
2575 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2576 sv_setpv(PL_linestr,"");
2577 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2579 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2580 s = swallow_bom((U8*)s);
2583 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2584 PL_doextract = FALSE;
2586 /* Incest with pod. */
2587 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2588 sv_setpv(PL_linestr, "");
2589 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2590 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2591 PL_doextract = FALSE;
2595 } while (PL_doextract);
2596 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2597 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2598 SV *sv = NEWSV(85,0);
2600 sv_upgrade(sv, SVt_PVMG);
2601 sv_setsv(sv,PL_linestr);
2602 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2604 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2605 if (CopLINE(PL_curcop) == 1) {
2606 while (s < PL_bufend && isSPACE(*s))
2608 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2612 if (*s == '#' && *(s+1) == '!')
2614 #ifdef ALTERNATE_SHEBANG
2616 static char as[] = ALTERNATE_SHEBANG;
2617 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2618 d = s + (sizeof(as) - 1);
2620 #endif /* ALTERNATE_SHEBANG */
2629 while (*d && !isSPACE(*d))
2633 #ifdef ARG_ZERO_IS_SCRIPT
2634 if (ipathend > ipath) {
2636 * HP-UX (at least) sets argv[0] to the script name,
2637 * which makes $^X incorrect. And Digital UNIX and Linux,
2638 * at least, set argv[0] to the basename of the Perl
2639 * interpreter. So, having found "#!", we'll set it right.
2641 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2642 assert(SvPOK(x) || SvGMAGICAL(x));
2643 if (sv_eq(x, CopFILESV(PL_curcop))) {
2644 sv_setpvn(x, ipath, ipathend - ipath);
2647 TAINT_NOT; /* $^X is always tainted, but that's OK */
2649 #endif /* ARG_ZERO_IS_SCRIPT */
2654 d = instr(s,"perl -");
2656 d = instr(s,"perl");
2658 /* avoid getting into infinite loops when shebang
2659 * line contains "Perl" rather than "perl" */
2661 for (d = ipathend-4; d >= ipath; --d) {
2662 if ((*d == 'p' || *d == 'P')
2663 && !ibcmp(d, "perl", 4))
2673 #ifdef ALTERNATE_SHEBANG
2675 * If the ALTERNATE_SHEBANG on this system starts with a
2676 * character that can be part of a Perl expression, then if
2677 * we see it but not "perl", we're probably looking at the
2678 * start of Perl code, not a request to hand off to some
2679 * other interpreter. Similarly, if "perl" is there, but
2680 * not in the first 'word' of the line, we assume the line
2681 * contains the start of the Perl program.
2683 if (d && *s != '#') {
2685 while (*c && !strchr("; \t\r\n\f\v#", *c))
2688 d = Nullch; /* "perl" not in first word; ignore */
2690 *s = '#'; /* Don't try to parse shebang line */
2692 #endif /* ALTERNATE_SHEBANG */
2693 #ifndef MACOS_TRADITIONAL
2698 !instr(s,"indir") &&
2699 instr(PL_origargv[0],"perl"))
2705 while (s < PL_bufend && isSPACE(*s))
2707 if (s < PL_bufend) {
2708 Newz(899,newargv,PL_origargc+3,char*);
2710 while (s < PL_bufend && !isSPACE(*s))
2713 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2716 newargv = PL_origargv;
2718 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2719 Perl_croak(aTHX_ "Can't exec %s", ipath);
2723 U32 oldpdb = PL_perldb;
2724 bool oldn = PL_minus_n;
2725 bool oldp = PL_minus_p;
2727 while (*d && !isSPACE(*d)) d++;
2728 while (SPACE_OR_TAB(*d)) d++;
2732 if (*d == 'M' || *d == 'm') {
2734 while (*d && !isSPACE(*d)) d++;
2735 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2738 d = moreswitches(d);
2740 if ((PERLDB_LINE && !oldpdb) ||
2741 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2742 /* if we have already added "LINE: while (<>) {",
2743 we must not do it again */
2745 sv_setpv(PL_linestr, "");
2746 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2747 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2748 PL_preambled = FALSE;
2750 (void)gv_fetchfile(PL_origfilename);
2757 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2759 PL_lex_state = LEX_FORMLINE;
2764 #ifdef PERL_STRICT_CR
2765 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2767 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2769 case ' ': case '\t': case '\f': case 013:
2770 #ifdef MACOS_TRADITIONAL
2777 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2778 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2779 /* handle eval qq[#line 1 "foo"\n ...] */
2780 CopLINE_dec(PL_curcop);
2784 while (s < d && *s != '\n')
2789 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2791 PL_lex_state = LEX_FORMLINE;
2801 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2808 while (s < PL_bufend && SPACE_OR_TAB(*s))
2811 if (strnEQ(s,"=>",2)) {
2812 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2813 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2814 "### Saw unary minus before =>, forcing word '%s'\n", s);
2816 OPERATOR('-'); /* unary minus */
2818 PL_last_uni = PL_oldbufptr;
2820 case 'r': ftst = OP_FTEREAD; break;
2821 case 'w': ftst = OP_FTEWRITE; break;
2822 case 'x': ftst = OP_FTEEXEC; break;
2823 case 'o': ftst = OP_FTEOWNED; break;
2824 case 'R': ftst = OP_FTRREAD; break;
2825 case 'W': ftst = OP_FTRWRITE; break;
2826 case 'X': ftst = OP_FTREXEC; break;
2827 case 'O': ftst = OP_FTROWNED; break;
2828 case 'e': ftst = OP_FTIS; break;
2829 case 'z': ftst = OP_FTZERO; break;
2830 case 's': ftst = OP_FTSIZE; break;
2831 case 'f': ftst = OP_FTFILE; break;
2832 case 'd': ftst = OP_FTDIR; break;
2833 case 'l': ftst = OP_FTLINK; break;
2834 case 'p': ftst = OP_FTPIPE; break;
2835 case 'S': ftst = OP_FTSOCK; break;
2836 case 'u': ftst = OP_FTSUID; break;
2837 case 'g': ftst = OP_FTSGID; break;
2838 case 'k': ftst = OP_FTSVTX; break;
2839 case 'b': ftst = OP_FTBLK; break;
2840 case 'c': ftst = OP_FTCHR; break;
2841 case 't': ftst = OP_FTTTY; break;
2842 case 'T': ftst = OP_FTTEXT; break;
2843 case 'B': ftst = OP_FTBINARY; break;
2844 case 'M': case 'A': case 'C':
2845 gv_fetchpv("\024",TRUE, SVt_PV);
2847 case 'M': ftst = OP_FTMTIME; break;
2848 case 'A': ftst = OP_FTATIME; break;
2849 case 'C': ftst = OP_FTCTIME; break;
2857 PL_last_lop_op = ftst;
2858 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2859 "### Saw file test %c\n", ftst);
2864 /* Assume it was a minus followed by a one-letter named
2865 * subroutine call (or a -bareword), then. */
2866 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2867 "### %c looked like a file test but was not\n", ftst);
2875 if (PL_expect == XOPERATOR)
2880 else if (*s == '>') {
2883 if (isIDFIRST_lazy_if(s,UTF)) {
2884 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2892 if (PL_expect == XOPERATOR)
2895 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2897 OPERATOR('-'); /* unary minus */
2904 if (PL_expect == XOPERATOR)
2909 if (PL_expect == XOPERATOR)
2912 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2918 if (PL_expect != XOPERATOR) {
2919 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2920 PL_expect = XOPERATOR;
2921 force_ident(PL_tokenbuf, '*');
2934 if (PL_expect == XOPERATOR) {
2938 PL_tokenbuf[0] = '%';
2939 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2940 if (!PL_tokenbuf[1]) {
2942 yyerror("Final % should be \\% or %name");
2945 PL_pending_ident = '%';
2964 switch (PL_expect) {
2967 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2969 PL_bufptr = s; /* update in case we back off */
2975 PL_expect = XTERMBLOCK;
2979 while (isIDFIRST_lazy_if(s,UTF)) {
2980 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2981 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2982 if (tmp < 0) tmp = -tmp;
2997 d = scan_str(d,TRUE,TRUE);
3000 SvREFCNT_dec(PL_lex_stuff);
3001 PL_lex_stuff = Nullsv;
3003 /* MUST advance bufptr here to avoid bogus
3004 "at end of line" context messages from yyerror().
3006 PL_bufptr = s + len;
3007 yyerror("Unterminated attribute parameter in attribute list");
3010 return 0; /* EOF indicator */
3014 SV *sv = newSVpvn(s, len);
3015 sv_catsv(sv, PL_lex_stuff);
3016 attrs = append_elem(OP_LIST, attrs,
3017 newSVOP(OP_CONST, 0, sv));
3018 SvREFCNT_dec(PL_lex_stuff);
3019 PL_lex_stuff = Nullsv;
3022 attrs = append_elem(OP_LIST, attrs,
3023 newSVOP(OP_CONST, 0,
3027 if (*s == ':' && s[1] != ':')
3030 break; /* require real whitespace or :'s */
3032 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3033 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3034 char q = ((*s == '\'') ? '"' : '\'');
3035 /* If here for an expression, and parsed no attrs, back off. */
3036 if (tmp == '=' && !attrs) {
3040 /* MUST advance bufptr here to avoid bogus "at end of line"
3041 context messages from yyerror().
3045 yyerror("Unterminated attribute list");
3047 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3055 PL_nextval[PL_nexttoke].opval = attrs;
3063 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3064 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3080 if (PL_lex_brackets <= 0)
3081 yyerror("Unmatched right square bracket");
3084 if (PL_lex_state == LEX_INTERPNORMAL) {
3085 if (PL_lex_brackets == 0) {
3086 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3087 PL_lex_state = LEX_INTERPEND;
3094 if (PL_lex_brackets > 100) {
3095 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3096 if (newlb != PL_lex_brackstack) {
3098 PL_lex_brackstack = newlb;
3101 switch (PL_expect) {
3103 if (PL_lex_formbrack) {
3107 if (PL_oldoldbufptr == PL_last_lop)
3108 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3110 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3111 OPERATOR(HASHBRACK);
3113 while (s < PL_bufend && SPACE_OR_TAB(*s))
3116 PL_tokenbuf[0] = '\0';
3117 if (d < PL_bufend && *d == '-') {
3118 PL_tokenbuf[0] = '-';
3120 while (d < PL_bufend && SPACE_OR_TAB(*d))
3123 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3124 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3126 while (d < PL_bufend && SPACE_OR_TAB(*d))
3129 char minus = (PL_tokenbuf[0] == '-');
3130 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3131 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, 0) &&
3132 PL_nextval[PL_nexttoke-1].opval)
3133 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke-1].opval)->op_sv);
3141 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3146 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3151 if (PL_oldoldbufptr == PL_last_lop)
3152 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3154 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3157 OPERATOR(HASHBRACK);
3158 /* This hack serves to disambiguate a pair of curlies
3159 * as being a block or an anon hash. Normally, expectation
3160 * determines that, but in cases where we're not in a
3161 * position to expect anything in particular (like inside
3162 * eval"") we have to resolve the ambiguity. This code
3163 * covers the case where the first term in the curlies is a
3164 * quoted string. Most other cases need to be explicitly
3165 * disambiguated by prepending a `+' before the opening
3166 * curly in order to force resolution as an anon hash.
3168 * XXX should probably propagate the outer expectation
3169 * into eval"" to rely less on this hack, but that could
3170 * potentially break current behavior of eval"".
3174 if (*s == '\'' || *s == '"' || *s == '`') {
3175 /* common case: get past first string, handling escapes */
3176 for (t++; t < PL_bufend && *t != *s;)
3177 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3181 else if (*s == 'q') {
3184 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3188 char open, close, term;
3191 while (t < PL_bufend && isSPACE(*t))
3195 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3199 for (t++; t < PL_bufend; t++) {
3200 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3202 else if (*t == open)
3206 for (t++; t < PL_bufend; t++) {
3207 if (*t == '\\' && t+1 < PL_bufend)
3209 else if (*t == close && --brackets <= 0)
3211 else if (*t == open)
3217 else if (isALNUM_lazy_if(t,UTF)) {
3219 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3222 while (t < PL_bufend && isSPACE(*t))
3224 /* if comma follows first term, call it an anon hash */
3225 /* XXX it could be a comma expression with loop modifiers */
3226 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3227 || (*t == '=' && t[1] == '>')))
3228 OPERATOR(HASHBRACK);
3229 if (PL_expect == XREF)
3232 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3238 yylval.ival = CopLINE(PL_curcop);
3239 if (isSPACE(*s) || *s == '#')
3240 PL_copline = NOLINE; /* invalidate current command line number */
3245 if (PL_lex_brackets <= 0)
3246 yyerror("Unmatched right curly bracket");
3248 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3249 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3250 PL_lex_formbrack = 0;
3251 if (PL_lex_state == LEX_INTERPNORMAL) {
3252 if (PL_lex_brackets == 0) {
3253 if (PL_expect & XFAKEBRACK) {
3254 PL_expect &= XENUMMASK;
3255 PL_lex_state = LEX_INTERPEND;
3257 return yylex(); /* ignore fake brackets */
3259 if (*s == '-' && s[1] == '>')
3260 PL_lex_state = LEX_INTERPENDMAYBE;
3261 else if (*s != '[' && *s != '{')
3262 PL_lex_state = LEX_INTERPEND;
3265 if (PL_expect & XFAKEBRACK) {
3266 PL_expect &= XENUMMASK;
3268 return yylex(); /* ignore fake brackets */
3278 if (PL_expect == XOPERATOR) {
3279 if (ckWARN(WARN_SEMICOLON)
3280 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3282 CopLINE_dec(PL_curcop);
3283 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3284 CopLINE_inc(PL_curcop);
3289 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3291 PL_expect = XOPERATOR;
3292 force_ident(PL_tokenbuf, '&');
3296 yylval.ival = (OPpENTERSUB_AMPER<<8);
3315 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3316 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3318 if (PL_expect == XSTATE && isALPHA(tmp) &&
3319 (s == PL_linestart+1 || s[-2] == '\n') )
3321 if (PL_in_eval && !PL_rsfp) {
3326 if (strnEQ(s,"=cut",4)) {
3340 PL_doextract = TRUE;
3343 if (PL_lex_brackets < PL_lex_formbrack) {
3345 #ifdef PERL_STRICT_CR
3346 for (t = s; SPACE_OR_TAB(*t); t++) ;
3348 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3350 if (*t == '\n' || *t == '#') {
3368 if (PL_expect != XOPERATOR) {
3369 if (s[1] != '<' && !strchr(s,'>'))
3372 s = scan_heredoc(s);
3374 s = scan_inputsymbol(s);
3375 TERM(sublex_start());
3380 SHop(OP_LEFT_SHIFT);
3394 SHop(OP_RIGHT_SHIFT);
3403 if (PL_expect == XOPERATOR) {
3404 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3407 return ','; /* grandfather non-comma-format format */
3411 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3412 PL_tokenbuf[0] = '@';
3413 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3414 sizeof PL_tokenbuf - 1, FALSE);
3415 if (PL_expect == XOPERATOR)
3416 no_op("Array length", s);
3417 if (!PL_tokenbuf[1])
3419 PL_expect = XOPERATOR;
3420 PL_pending_ident = '#';
3424 PL_tokenbuf[0] = '$';
3425 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3426 sizeof PL_tokenbuf - 1, FALSE);
3427 if (PL_expect == XOPERATOR)
3429 if (!PL_tokenbuf[1]) {
3431 yyerror("Final $ should be \\$ or $name");
3435 /* This kludge not intended to be bulletproof. */
3436 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3437 yylval.opval = newSVOP(OP_CONST, 0,
3438 newSViv(PL_compiling.cop_arybase));
3439 yylval.opval->op_private = OPpCONST_ARYBASE;
3445 if (PL_lex_state == LEX_NORMAL)
3448 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3451 PL_tokenbuf[0] = '@';
3452 if (ckWARN(WARN_SYNTAX)) {
3454 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3457 PL_bufptr = skipspace(PL_bufptr);
3458 while (t < PL_bufend && *t != ']')
3460 Perl_warner(aTHX_ WARN_SYNTAX,
3461 "Multidimensional syntax %.*s not supported",
3462 (t - PL_bufptr) + 1, PL_bufptr);
3466 else if (*s == '{') {
3467 PL_tokenbuf[0] = '%';
3468 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3469 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3471 char tmpbuf[sizeof PL_tokenbuf];
3473 for (t++; isSPACE(*t); t++) ;
3474 if (isIDFIRST_lazy_if(t,UTF)) {
3475 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3476 for (; isSPACE(*t); t++) ;
3477 if (*t == ';' && get_cv(tmpbuf, FALSE))
3478 Perl_warner(aTHX_ WARN_SYNTAX,
3479 "You need to quote \"%s\"", tmpbuf);
3485 PL_expect = XOPERATOR;
3486 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3487 bool islop = (PL_last_lop == PL_oldoldbufptr);
3488 if (!islop || PL_last_lop_op == OP_GREPSTART)
3489 PL_expect = XOPERATOR;
3490 else if (strchr("$@\"'`q", *s))
3491 PL_expect = XTERM; /* e.g. print $fh "foo" */
3492 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3493 PL_expect = XTERM; /* e.g. print $fh &sub */
3494 else if (isIDFIRST_lazy_if(s,UTF)) {
3495 char tmpbuf[sizeof PL_tokenbuf];
3496 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3497 if ((tmp = keyword(tmpbuf, len))) {
3498 /* binary operators exclude handle interpretations */
3510 PL_expect = XTERM; /* e.g. print $fh length() */
3515 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3516 if (gv && GvCVu(gv))
3517 PL_expect = XTERM; /* e.g. print $fh subr() */
3520 else if (isDIGIT(*s))
3521 PL_expect = XTERM; /* e.g. print $fh 3 */
3522 else if (*s == '.' && isDIGIT(s[1]))
3523 PL_expect = XTERM; /* e.g. print $fh .3 */
3524 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3525 PL_expect = XTERM; /* e.g. print $fh -1 */
3526 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3527 PL_expect = XTERM; /* print $fh <<"EOF" */
3529 PL_pending_ident = '$';
3533 if (PL_expect == XOPERATOR)
3535 PL_tokenbuf[0] = '@';
3536 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3537 if (!PL_tokenbuf[1]) {
3539 yyerror("Final @ should be \\@ or @name");
3542 if (PL_lex_state == LEX_NORMAL)
3544 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3546 PL_tokenbuf[0] = '%';
3548 /* Warn about @ where they meant $. */
3549 if (ckWARN(WARN_SYNTAX)) {
3550 if (*s == '[' || *s == '{') {
3552 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3554 if (*t == '}' || *t == ']') {
3556 PL_bufptr = skipspace(PL_bufptr);
3557 Perl_warner(aTHX_ WARN_SYNTAX,
3558 "Scalar value %.*s better written as $%.*s",
3559 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3564 PL_pending_ident = '@';
3567 case '/': /* may either be division or pattern */
3568 case '?': /* may either be conditional or pattern */
3569 if (PL_expect != XOPERATOR) {
3570 /* Disable warning on "study /blah/" */
3571 if (PL_oldoldbufptr == PL_last_uni
3572 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3573 || memNE(PL_last_uni, "study", 5)
3574 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3576 s = scan_pat(s,OP_MATCH);
3577 TERM(sublex_start());
3585 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3586 #ifdef PERL_STRICT_CR
3589 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3591 && (s == PL_linestart || s[-1] == '\n') )
3593 PL_lex_formbrack = 0;
3597 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3603 yylval.ival = OPf_SPECIAL;
3609 if (PL_expect != XOPERATOR)
3614 case '0': case '1': case '2': case '3': case '4':
3615 case '5': case '6': case '7': case '8': case '9':
3616 s = scan_num(s, &yylval);
3617 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3618 "### Saw number in '%s'\n", s);
3620 if (PL_expect == XOPERATOR)
3625 s = scan_str(s,FALSE,FALSE);
3626 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3627 "### Saw string in '%s'\n", s);
3629 if (PL_expect == XOPERATOR) {
3630 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3633 return ','; /* grandfather non-comma-format format */
3639 missingterm((char*)0);
3640 yylval.ival = OP_CONST;
3641 TERM(sublex_start());
3644 s = scan_str(s,FALSE,FALSE);
3645 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3646 "### Saw string in '%s'\n", s);
3648 if (PL_expect == XOPERATOR) {
3649 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3652 return ','; /* grandfather non-comma-format format */
3658 missingterm((char*)0);
3659 yylval.ival = OP_CONST;
3660 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3661 if (*d == '$' || *d == '@' || *d == '\\' || UTF8_IS_CONTINUED(*d)) {
3662 yylval.ival = OP_STRINGIFY;
3666 TERM(sublex_start());
3669 s = scan_str(s,FALSE,FALSE);
3670 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3671 "### Saw backtick string in '%s'\n", s);
3673 if (PL_expect == XOPERATOR)
3674 no_op("Backticks",s);
3676 missingterm((char*)0);
3677 yylval.ival = OP_BACKTICK;
3679 TERM(sublex_start());
3683 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3684 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3686 if (PL_expect == XOPERATOR)
3687 no_op("Backslash",s);
3691 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3695 while (isDIGIT(*start) || *start == '_')
3697 if (*start == '.' && isDIGIT(start[1])) {
3698 s = scan_num(s, &yylval);
3701 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3702 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3706 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3709 s = scan_num(s, &yylval);
3716 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3755 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3757 /* Some keywords can be followed by any delimiter, including ':' */
3758 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3759 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3760 (PL_tokenbuf[0] == 'q' &&
3761 strchr("qwxr", PL_tokenbuf[1])))));
3763 /* x::* is just a word, unless x is "CORE" */
3764 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3768 while (d < PL_bufend && isSPACE(*d))
3769 d++; /* no comments skipped here, or s### is misparsed */
3771 /* Is this a label? */
3772 if (!tmp && PL_expect == XSTATE
3773 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3775 yylval.pval = savepv(PL_tokenbuf);
3780 /* Check for keywords */
3781 tmp = keyword(PL_tokenbuf, len);
3783 /* Is this a word before a => operator? */
3784 if (*d == '=' && d[1] == '>') {
3786 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3787 yylval.opval->op_private = OPpCONST_BARE;
3788 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3789 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3793 if (tmp < 0) { /* second-class keyword? */
3794 GV *ogv = Nullgv; /* override (winner) */
3795 GV *hgv = Nullgv; /* hidden (loser) */
3796 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3798 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3801 if (GvIMPORTED_CV(gv))
3803 else if (! CvMETHOD(cv))
3807 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3808 (gv = *gvp) != (GV*)&PL_sv_undef &&
3809 GvCVu(gv) && GvIMPORTED_CV(gv))
3815 tmp = 0; /* overridden by import or by GLOBAL */
3818 && -tmp==KEY_lock /* XXX generalizable kludge */
3820 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3822 tmp = 0; /* any sub overrides "weak" keyword */
3824 else { /* no override */
3828 if (ckWARN(WARN_AMBIGUOUS) && hgv
3829 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3830 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3831 "Ambiguous call resolved as CORE::%s(), %s",
3832 GvENAME(hgv), "qualify as such or use &");
3839 default: /* not a keyword */
3842 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3844 /* Get the rest if it looks like a package qualifier */
3846 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3848 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3851 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3852 *s == '\'' ? "'" : "::");
3856 if (PL_expect == XOPERATOR) {
3857 if (PL_bufptr == PL_linestart) {
3858 CopLINE_dec(PL_curcop);
3859 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3860 CopLINE_inc(PL_curcop);
3863 no_op("Bareword",s);
3866 /* Look for a subroutine with this name in current package,
3867 unless name is "Foo::", in which case Foo is a bearword
3868 (and a package name). */
3871 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3873 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3874 Perl_warner(aTHX_ WARN_BAREWORD,
3875 "Bareword \"%s\" refers to nonexistent package",
3878 PL_tokenbuf[len] = '\0';
3885 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3888 /* if we saw a global override before, get the right name */
3891 sv = newSVpvn("CORE::GLOBAL::",14);
3892 sv_catpv(sv,PL_tokenbuf);
3895 sv = newSVpv(PL_tokenbuf,0);
3897 /* Presume this is going to be a bareword of some sort. */
3900 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3901 yylval.opval->op_private = OPpCONST_BARE;
3903 /* And if "Foo::", then that's what it certainly is. */
3908 /* See if it's the indirect object for a list operator. */
3910 if (PL_oldoldbufptr &&
3911 PL_oldoldbufptr < PL_bufptr &&
3912 (PL_oldoldbufptr == PL_last_lop
3913 || PL_oldoldbufptr == PL_last_uni) &&
3914 /* NO SKIPSPACE BEFORE HERE! */
3915 (PL_expect == XREF ||
3916 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3918 bool immediate_paren = *s == '(';
3920 /* (Now we can afford to cross potential line boundary.) */
3923 /* Two barewords in a row may indicate method call. */
3925 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3928 /* If not a declared subroutine, it's an indirect object. */
3929 /* (But it's an indir obj regardless for sort.) */
3931 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3932 ((!gv || !GvCVu(gv)) &&
3933 (PL_last_lop_op != OP_MAPSTART &&
3934 PL_last_lop_op != OP_GREPSTART))))
3936 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3942 PL_expect = XOPERATOR;
3945 /* Is this a word before a => operator? */
3946 if (*s == '=' && s[1] == '>') {
3948 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3949 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3950 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3954 /* If followed by a paren, it's certainly a subroutine. */
3957 if (gv && GvCVu(gv)) {
3958 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3959 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3964 PL_nextval[PL_nexttoke].opval = yylval.opval;
3965 PL_expect = XOPERATOR;
3971 /* If followed by var or block, call it a method (unless sub) */
3973 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3974 PL_last_lop = PL_oldbufptr;
3975 PL_last_lop_op = OP_METHOD;
3979 /* If followed by a bareword, see if it looks like indir obj. */
3981 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3984 /* Not a method, so call it a subroutine (if defined) */
3986 if (gv && GvCVu(gv)) {
3988 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3989 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3990 "Ambiguous use of -%s resolved as -&%s()",
3991 PL_tokenbuf, PL_tokenbuf);
3992 /* Check for a constant sub */
3994 if ((sv = cv_const_sv(cv))) {
3996 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3997 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3998 yylval.opval->op_private = 0;
4002 /* Resolve to GV now. */
4003 op_free(yylval.opval);
4004 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4005 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4006 PL_last_lop = PL_oldbufptr;
4007 PL_last_lop_op = OP_ENTERSUB;
4008 /* Is there a prototype? */
4011 char *proto = SvPV((SV*)cv, len);
4014 if (strEQ(proto, "$"))
4016 if (*proto == '&' && *s == '{') {
4017 sv_setpv(PL_subname,"__ANON__");
4021 PL_nextval[PL_nexttoke].opval = yylval.opval;
4027 /* Call it a bare word */
4029 if (PL_hints & HINT_STRICT_SUBS)
4030 yylval.opval->op_private |= OPpCONST_STRICT;
4033 if (ckWARN(WARN_RESERVED)) {
4034 if (lastchar != '-') {
4035 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4037 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4044 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4045 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4046 "Operator or semicolon missing before %c%s",
4047 lastchar, PL_tokenbuf);
4048 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4049 "Ambiguous use of %c resolved as operator %c",
4050 lastchar, lastchar);
4056 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4057 newSVpv(CopFILE(PL_curcop),0));
4061 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4062 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4065 case KEY___PACKAGE__:
4066 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4068 ? newSVsv(PL_curstname)
4077 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4078 char *pname = "main";
4079 if (PL_tokenbuf[2] == 'D')
4080 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4081 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4084 GvIOp(gv) = newIO();
4085 IoIFP(GvIOp(gv)) = PL_rsfp;
4086 #if defined(HAS_FCNTL) && defined(F_SETFD)
4088 int fd = PerlIO_fileno(PL_rsfp);
4089 fcntl(fd,F_SETFD,fd >= 3);
4092 /* Mark this internal pseudo-handle as clean */
4093 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4095 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4096 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4097 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4099 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4100 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4101 /* if the script was opened in binmode, we need to revert
4102 * it to text mode for compatibility; but only iff it has CRs
4103 * XXX this is a questionable hack at best. */
4104 if (PL_bufend-PL_bufptr > 2
4105 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4108 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4109 loc = PerlIO_tell(PL_rsfp);
4110 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4112 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4113 #if defined(__BORLANDC__)
4114 /* XXX see note in do_binmode() */
4115 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4118 PerlIO_seek(PL_rsfp, loc, 0);
4122 #ifdef PERLIO_LAYERS
4123 if (UTF && !IN_BYTE)
4124 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4137 if (PL_expect == XSTATE) {
4144 if (*s == ':' && s[1] == ':') {
4147 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4148 if (!(tmp = keyword(PL_tokenbuf, len)))
4149 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4163 LOP(OP_ACCEPT,XTERM);
4169 LOP(OP_ATAN2,XTERM);
4175 LOP(OP_BINMODE,XTERM);
4178 LOP(OP_BLESS,XTERM);
4187 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4204 if (!PL_cryptseen) {
4205 PL_cryptseen = TRUE;
4209 LOP(OP_CRYPT,XTERM);
4212 if (ckWARN(WARN_CHMOD)) {
4213 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4214 if (*d != '0' && isDIGIT(*d))
4215 Perl_warner(aTHX_ WARN_CHMOD,
4216 "chmod() mode argument is missing initial 0");
4218 LOP(OP_CHMOD,XTERM);
4221 LOP(OP_CHOWN,XTERM);
4224 LOP(OP_CONNECT,XTERM);
4240 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4244 PL_hints |= HINT_BLOCK_SCOPE;
4254 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4255 LOP(OP_DBMOPEN,XTERM);
4261 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4268 yylval.ival = CopLINE(PL_curcop);
4282 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4283 UNIBRACK(OP_ENTEREVAL);
4298 case KEY_endhostent:
4304 case KEY_endservent:
4307 case KEY_endprotoent:
4318 yylval.ival = CopLINE(PL_curcop);
4320 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4322 if ((PL_bufend - p) >= 3 &&
4323 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4325 else if ((PL_bufend - p) >= 4 &&
4326 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4329 if (isIDFIRST_lazy_if(p,UTF)) {
4330 p = scan_ident(p, PL_bufend,
4331 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4335 Perl_croak(aTHX_ "Missing $ on loop variable");
4340 LOP(OP_FORMLINE,XTERM);
4346 LOP(OP_FCNTL,XTERM);
4352 LOP(OP_FLOCK,XTERM);
4361 LOP(OP_GREPSTART, XREF);
4364 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4379 case KEY_getpriority:
4380 LOP(OP_GETPRIORITY,XTERM);
4382 case KEY_getprotobyname:
4385 case KEY_getprotobynumber:
4386 LOP(OP_GPBYNUMBER,XTERM);
4388 case KEY_getprotoent:
4400 case KEY_getpeername:
4401 UNI(OP_GETPEERNAME);
4403 case KEY_gethostbyname:
4406 case KEY_gethostbyaddr:
4407 LOP(OP_GHBYADDR,XTERM);
4409 case KEY_gethostent:
4412 case KEY_getnetbyname:
4415 case KEY_getnetbyaddr:
4416 LOP(OP_GNBYADDR,XTERM);
4421 case KEY_getservbyname:
4422 LOP(OP_GSBYNAME,XTERM);
4424 case KEY_getservbyport:
4425 LOP(OP_GSBYPORT,XTERM);
4427 case KEY_getservent:
4430 case KEY_getsockname:
4431 UNI(OP_GETSOCKNAME);
4433 case KEY_getsockopt:
4434 LOP(OP_GSOCKOPT,XTERM);
4456 yylval.ival = CopLINE(PL_curcop);
4460 LOP(OP_INDEX,XTERM);
4466 LOP(OP_IOCTL,XTERM);
4478 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4510 LOP(OP_LISTEN,XTERM);
4519 s = scan_pat(s,OP_MATCH);
4520 TERM(sublex_start());
4523 LOP(OP_MAPSTART, XREF);
4526 LOP(OP_MKDIR,XTERM);
4529 LOP(OP_MSGCTL,XTERM);
4532 LOP(OP_MSGGET,XTERM);
4535 LOP(OP_MSGRCV,XTERM);
4538 LOP(OP_MSGSND,XTERM);
4544 if (isIDFIRST_lazy_if(s,UTF)) {
4545 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4546 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4548 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4549 if (!PL_in_my_stash) {
4552 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4560 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4567 if (PL_expect != XSTATE)
4568 yyerror("\"no\" not allowed in expression");
4569 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4570 s = force_version(s);
4575 if (*s == '(' || (s = skipspace(s), *s == '('))
4582 if (isIDFIRST_lazy_if(s,UTF)) {
4584 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4586 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4587 Perl_warner(aTHX_ WARN_PRECEDENCE,
4588 "Precedence problem: open %.*s should be open(%.*s)",
4594 yylval.ival = OP_OR;
4604 LOP(OP_OPEN_DIR,XTERM);
4607 checkcomma(s,PL_tokenbuf,"filehandle");
4611 checkcomma(s,PL_tokenbuf,"filehandle");
4630 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4634 LOP(OP_PIPE_OP,XTERM);
4637 s = scan_str(s,FALSE,FALSE);
4639 missingterm((char*)0);
4640 yylval.ival = OP_CONST;
4641 TERM(sublex_start());
4647 s = scan_str(s,FALSE,FALSE);
4649 missingterm((char*)0);
4651 if (SvCUR(PL_lex_stuff)) {
4654 d = SvPV_force(PL_lex_stuff, len);
4657 for (; isSPACE(*d) && len; --len, ++d) ;
4660 if (!warned && ckWARN(WARN_QW)) {
4661 for (; !isSPACE(*d) && len; --len, ++d) {
4663 Perl_warner(aTHX_ WARN_QW,
4664 "Possible attempt to separate words with commas");
4667 else if (*d == '#') {
4668 Perl_warner(aTHX_ WARN_QW,
4669 "Possible attempt to put comments in qw() list");
4675 for (; !isSPACE(*d) && len; --len, ++d) ;
4677 sv = newSVpvn(b, d-b);
4678 if (DO_UTF8(PL_lex_stuff))
4680 words = append_elem(OP_LIST, words,
4681 newSVOP(OP_CONST, 0, tokeq(sv)));
4685 PL_nextval[PL_nexttoke].opval = words;
4690 SvREFCNT_dec(PL_lex_stuff);
4691 PL_lex_stuff = Nullsv;
4696 s = scan_str(s,FALSE,FALSE);
4698 missingterm((char*)0);
4699 yylval.ival = OP_STRINGIFY;
4700 if (SvIVX(PL_lex_stuff) == '\'')
4701 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4702 TERM(sublex_start());
4705 s = scan_pat(s,OP_QR);
4706 TERM(sublex_start());
4709 s = scan_str(s,FALSE,FALSE);
4711 missingterm((char*)0);
4712 yylval.ival = OP_BACKTICK;
4714 TERM(sublex_start());
4721 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4722 s = force_version(s);
4725 *PL_tokenbuf = '\0';
4726 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4727 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4728 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4730 yyerror("<> should be quotes");
4738 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4742 LOP(OP_RENAME,XTERM);
4751 LOP(OP_RINDEX,XTERM);
4774 LOP(OP_REVERSE,XTERM);
4785 TERM(sublex_start());
4787 TOKEN(1); /* force error */
4796 LOP(OP_SELECT,XTERM);
4802 LOP(OP_SEMCTL,XTERM);
4805 LOP(OP_SEMGET,XTERM);
4808 LOP(OP_SEMOP,XTERM);
4814 LOP(OP_SETPGRP,XTERM);
4816 case KEY_setpriority:
4817 LOP(OP_SETPRIORITY,XTERM);
4819 case KEY_sethostent:
4825 case KEY_setservent:
4828 case KEY_setprotoent:
4838 LOP(OP_SEEKDIR,XTERM);
4840 case KEY_setsockopt:
4841 LOP(OP_SSOCKOPT,XTERM);
4847 LOP(OP_SHMCTL,XTERM);
4850 LOP(OP_SHMGET,XTERM);
4853 LOP(OP_SHMREAD,XTERM);
4856 LOP(OP_SHMWRITE,XTERM);
4859 LOP(OP_SHUTDOWN,XTERM);
4868 LOP(OP_SOCKET,XTERM);
4870 case KEY_socketpair:
4871 LOP(OP_SOCKPAIR,XTERM);
4874 checkcomma(s,PL_tokenbuf,"subroutine name");
4876 if (*s == ';' || *s == ')') /* probably a close */
4877 Perl_croak(aTHX_ "sort is now a reserved word");
4879 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4883 LOP(OP_SPLIT,XTERM);
4886 LOP(OP_SPRINTF,XTERM);
4889 LOP(OP_SPLICE,XTERM);
4904 LOP(OP_SUBSTR,XTERM);
4910 char tmpbuf[sizeof PL_tokenbuf];
4912 expectation attrful;
4913 bool have_name, have_proto;
4918 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4919 (*s == ':' && s[1] == ':'))
4922 attrful = XATTRBLOCK;
4923 /* remember buffer pos'n for later force_word */
4924 tboffset = s - PL_oldbufptr;
4925 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4926 if (strchr(tmpbuf, ':'))
4927 sv_setpv(PL_subname, tmpbuf);
4929 sv_setsv(PL_subname,PL_curstname);
4930 sv_catpvn(PL_subname,"::",2);
4931 sv_catpvn(PL_subname,tmpbuf,len);
4938 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4939 PL_expect = XTERMBLOCK;
4940 attrful = XATTRTERM;
4941 sv_setpv(PL_subname,"?");
4945 if (key == KEY_format) {
4947 PL_lex_formbrack = PL_lex_brackets + 1;
4949 (void) force_word(PL_oldbufptr + tboffset, WORD,
4954 /* Look for a prototype */
4958 s = scan_str(s,FALSE,FALSE);
4961 SvREFCNT_dec(PL_lex_stuff);
4962 PL_lex_stuff = Nullsv;
4963 Perl_croak(aTHX_ "Prototype not terminated");
4966 d = SvPVX(PL_lex_stuff);
4968 for (p = d; *p; ++p) {
4973 SvCUR(PL_lex_stuff) = tmp;
4981 if (*s == ':' && s[1] != ':')
4982 PL_expect = attrful;
4985 PL_nextval[PL_nexttoke].opval =
4986 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4987 PL_lex_stuff = Nullsv;
4991 sv_setpv(PL_subname,"__ANON__");
4994 (void) force_word(PL_oldbufptr + tboffset, WORD,
5003 LOP(OP_SYSTEM,XREF);
5006 LOP(OP_SYMLINK,XTERM);
5009 LOP(OP_SYSCALL,XTERM);
5012 LOP(OP_SYSOPEN,XTERM);
5015 LOP(OP_SYSSEEK,XTERM);
5018 LOP(OP_SYSREAD,XTERM);
5021 LOP(OP_SYSWRITE,XTERM);
5025 TERM(sublex_start());
5046 LOP(OP_TRUNCATE,XTERM);
5058 yylval.ival = CopLINE(PL_curcop);
5062 yylval.ival = CopLINE(PL_curcop);
5066 LOP(OP_UNLINK,XTERM);
5072 LOP(OP_UNPACK,XTERM);
5075 LOP(OP_UTIME,XTERM);
5078 if (ckWARN(WARN_UMASK)) {
5079 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5080 if (*d != '0' && isDIGIT(*d))
5081 Perl_warner(aTHX_ WARN_UMASK,
5082 "umask: argument is missing initial 0");
5087 LOP(OP_UNSHIFT,XTERM);
5090 if (PL_expect != XSTATE)
5091 yyerror("\"use\" not allowed in expression");
5093 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5094 s = force_version(s);
5095 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5096 PL_nextval[PL_nexttoke].opval = Nullop;
5101 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5102 s = force_version(s);
5114 yylval.ival = CopLINE(PL_curcop);
5118 PL_hints |= HINT_BLOCK_SCOPE;
5125 LOP(OP_WAITPID,XTERM);
5133 static char ctl_l[2];
5135 if (ctl_l[0] == '\0')
5136 ctl_l[0] = toCTRL('L');
5137 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5140 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5145 if (PL_expect == XOPERATOR)
5151 yylval.ival = OP_XOR;
5156 TERM(sublex_start());
5161 #pragma segment Main
5165 Perl_keyword(pTHX_ register char *d, I32 len)
5170 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5171 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5172 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5173 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5174 if (strEQ(d,"__END__")) return KEY___END__;
5178 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5183 if (strEQ(d,"and")) return -KEY_and;
5184 if (strEQ(d,"abs")) return -KEY_abs;
5187 if (strEQ(d,"alarm")) return -KEY_alarm;
5188 if (strEQ(d,"atan2")) return -KEY_atan2;
5191 if (strEQ(d,"accept")) return -KEY_accept;
5196 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5199 if (strEQ(d,"bless")) return -KEY_bless;
5200 if (strEQ(d,"bind")) return -KEY_bind;
5201 if (strEQ(d,"binmode")) return -KEY_binmode;
5204 if (strEQ(d,"CORE")) return -KEY_CORE;
5205 if (strEQ(d,"CHECK")) return KEY_CHECK;
5210 if (strEQ(d,"cmp")) return -KEY_cmp;
5211 if (strEQ(d,"chr")) return -KEY_chr;
5212 if (strEQ(d,"cos")) return -KEY_cos;
5215 if (strEQ(d,"chop")) return -KEY_chop;
5218 if (strEQ(d,"close")) return -KEY_close;
5219 if (strEQ(d,"chdir")) return -KEY_chdir;
5220 if (strEQ(d,"chomp")) return -KEY_chomp;
5221 if (strEQ(d,"chmod")) return -KEY_chmod;
5222 if (strEQ(d,"chown")) return -KEY_chown;
5223 if (strEQ(d,"crypt")) return -KEY_crypt;
5226 if (strEQ(d,"chroot")) return -KEY_chroot;
5227 if (strEQ(d,"caller")) return -KEY_caller;
5230 if (strEQ(d,"connect")) return -KEY_connect;
5233 if (strEQ(d,"closedir")) return -KEY_closedir;
5234 if (strEQ(d,"continue")) return -KEY_continue;
5239 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5244 if (strEQ(d,"do")) return KEY_do;
5247 if (strEQ(d,"die")) return -KEY_die;
5250 if (strEQ(d,"dump")) return -KEY_dump;
5253 if (strEQ(d,"delete")) return KEY_delete;
5256 if (strEQ(d,"defined")) return KEY_defined;
5257 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5260 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5265 if (strEQ(d,"END")) return KEY_END;
5270 if (strEQ(d,"eq")) return -KEY_eq;
5273 if (strEQ(d,"eof")) return -KEY_eof;
5274 if (strEQ(d,"exp")) return -KEY_exp;
5277 if (strEQ(d,"else")) return KEY_else;
5278 if (strEQ(d,"exit")) return -KEY_exit;
5279 if (strEQ(d,"eval")) return KEY_eval;
5280 if (strEQ(d,"exec")) return -KEY_exec;
5281 if (strEQ(d,"each")) return -KEY_each;
5284 if (strEQ(d,"elsif")) return KEY_elsif;
5287 if (strEQ(d,"exists")) return KEY_exists;
5288 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5291 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5292 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5295 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5298 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5299 if (strEQ(d,"endservent")) return -KEY_endservent;
5302 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5309 if (strEQ(d,"for")) return KEY_for;
5312 if (strEQ(d,"fork")) return -KEY_fork;
5315 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5316 if (strEQ(d,"flock")) return -KEY_flock;
5319 if (strEQ(d,"format")) return KEY_format;
5320 if (strEQ(d,"fileno")) return -KEY_fileno;
5323 if (strEQ(d,"foreach")) return KEY_foreach;
5326 if (strEQ(d,"formline")) return -KEY_formline;
5331 if (strnEQ(d,"get",3)) {
5336 if (strEQ(d,"ppid")) return -KEY_getppid;
5337 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5340 if (strEQ(d,"pwent")) return -KEY_getpwent;
5341 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5342 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5345 if (strEQ(d,"peername")) return -KEY_getpeername;
5346 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5347 if (strEQ(d,"priority")) return -KEY_getpriority;
5350 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5353 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5357 else if (*d == 'h') {
5358 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5359 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5360 if (strEQ(d,"hostent")) return -KEY_gethostent;
5362 else if (*d == 'n') {
5363 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5364 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5365 if (strEQ(d,"netent")) return -KEY_getnetent;
5367 else if (*d == 's') {
5368 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5369 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5370 if (strEQ(d,"servent")) return -KEY_getservent;
5371 if (strEQ(d,"sockname")) return -KEY_getsockname;
5372 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5374 else if (*d == 'g') {
5375 if (strEQ(d,"grent")) return -KEY_getgrent;
5376 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5377 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5379 else if (*d == 'l') {
5380 if (strEQ(d,"login")) return -KEY_getlogin;
5382 else if (strEQ(d,"c")) return -KEY_getc;
5387 if (strEQ(d,"gt")) return -KEY_gt;
5388 if (strEQ(d,"ge")) return -KEY_ge;
5391 if (strEQ(d,"grep")) return KEY_grep;
5392 if (strEQ(d,"goto")) return KEY_goto;
5393 if (strEQ(d,"glob")) return KEY_glob;
5396 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5401 if (strEQ(d,"hex")) return -KEY_hex;
5404 if (strEQ(d,"INIT")) return KEY_INIT;
5409 if (strEQ(d,"if")) return KEY_if;
5412 if (strEQ(d,"int")) return -KEY_int;
5415 if (strEQ(d,"index")) return -KEY_index;
5416 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5421 if (strEQ(d,"join")) return -KEY_join;
5425 if (strEQ(d,"keys")) return -KEY_keys;
5426 if (strEQ(d,"kill")) return -KEY_kill;
5432 if (strEQ(d,"lt")) return -KEY_lt;
5433 if (strEQ(d,"le")) return -KEY_le;
5434 if (strEQ(d,"lc")) return -KEY_lc;
5437 if (strEQ(d,"log")) return -KEY_log;
5440 if (strEQ(d,"last")) return KEY_last;
5441 if (strEQ(d,"link")) return -KEY_link;
5442 if (strEQ(d,"lock")) return -KEY_lock;
5445 if (strEQ(d,"local")) return KEY_local;
5446 if (strEQ(d,"lstat")) return -KEY_lstat;
5449 if (strEQ(d,"length")) return -KEY_length;
5450 if (strEQ(d,"listen")) return -KEY_listen;
5453 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5456 if (strEQ(d,"localtime")) return -KEY_localtime;
5462 case 1: return KEY_m;
5464 if (strEQ(d,"my")) return KEY_my;
5467 if (strEQ(d,"map")) return KEY_map;
5470 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5473 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5474 if (strEQ(d,"msgget")) return -KEY_msgget;
5475 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5476 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5481 if (strEQ(d,"next")) return KEY_next;
5482 if (strEQ(d,"ne")) return -KEY_ne;
5483 if (strEQ(d,"not")) return -KEY_not;
5484 if (strEQ(d,"no")) return KEY_no;
5489 if (strEQ(d,"or")) return -KEY_or;
5492 if (strEQ(d,"ord")) return -KEY_ord;
5493 if (strEQ(d,"oct")) return -KEY_oct;
5494 if (strEQ(d,"our")) return KEY_our;
5497 if (strEQ(d,"open")) return -KEY_open;
5500 if (strEQ(d,"opendir")) return -KEY_opendir;
5507 if (strEQ(d,"pop")) return -KEY_pop;
5508 if (strEQ(d,"pos")) return KEY_pos;
5511 if (strEQ(d,"push")) return -KEY_push;
5512 if (strEQ(d,"pack")) return -KEY_pack;
5513 if (strEQ(d,"pipe")) return -KEY_pipe;
5516 if (strEQ(d,"print")) return KEY_print;
5519 if (strEQ(d,"printf")) return KEY_printf;
5522 if (strEQ(d,"package")) return KEY_package;
5525 if (strEQ(d,"prototype")) return KEY_prototype;
5530 if (strEQ(d,"q")) return KEY_q;
5531 if (strEQ(d,"qr")) return KEY_qr;
5532 if (strEQ(d,"qq")) return KEY_qq;
5533 if (strEQ(d,"qw")) return KEY_qw;
5534 if (strEQ(d,"qx")) return KEY_qx;
5536 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5541 if (strEQ(d,"ref")) return -KEY_ref;
5544 if (strEQ(d,"read")) return -KEY_read;
5545 if (strEQ(d,"rand")) return -KEY_rand;
5546 if (strEQ(d,"recv")) return -KEY_recv;
5547 if (strEQ(d,"redo")) return KEY_redo;
5550 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5551 if (strEQ(d,"reset")) return -KEY_reset;
5554 if (strEQ(d,"return")) return KEY_return;
5555 if (strEQ(d,"rename")) return -KEY_rename;
5556 if (strEQ(d,"rindex")) return -KEY_rindex;
5559 if (strEQ(d,"require")) return -KEY_require;
5560 if (strEQ(d,"reverse")) return -KEY_reverse;
5561 if (strEQ(d,"readdir")) return -KEY_readdir;
5564 if (strEQ(d,"readlink")) return -KEY_readlink;
5565 if (strEQ(d,"readline")) return -KEY_readline;
5566 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5569 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5575 case 0: return KEY_s;
5577 if (strEQ(d,"scalar")) return KEY_scalar;
5582 if (strEQ(d,"seek")) return -KEY_seek;
5583 if (strEQ(d,"send")) return -KEY_send;
5586 if (strEQ(d,"semop")) return -KEY_semop;
5589 if (strEQ(d,"select")) return -KEY_select;
5590 if (strEQ(d,"semctl")) return -KEY_semctl;
5591 if (strEQ(d,"semget")) return -KEY_semget;
5594 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5595 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5598 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5599 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5602 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5605 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5606 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5607 if (strEQ(d,"setservent")) return -KEY_setservent;
5610 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5611 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5618 if (strEQ(d,"shift")) return -KEY_shift;
5621 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5622 if (strEQ(d,"shmget")) return -KEY_shmget;
5625 if (strEQ(d,"shmread")) return -KEY_shmread;
5628 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5629 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5634 if (strEQ(d,"sin")) return -KEY_sin;
5637 if (strEQ(d,"sleep")) return -KEY_sleep;
5640 if (strEQ(d,"sort")) return KEY_sort;
5641 if (strEQ(d,"socket")) return -KEY_socket;
5642 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5645 if (strEQ(d,"split")) return KEY_split;
5646 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5647 if (strEQ(d,"splice")) return -KEY_splice;
5650 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5653 if (strEQ(d,"srand")) return -KEY_srand;
5656 if (strEQ(d,"stat")) return -KEY_stat;
5657 if (strEQ(d,"study")) return KEY_study;
5660 if (strEQ(d,"substr")) return -KEY_substr;
5661 if (strEQ(d,"sub")) return KEY_sub;
5666 if (strEQ(d,"system")) return -KEY_system;
5669 if (strEQ(d,"symlink")) return -KEY_symlink;
5670 if (strEQ(d,"syscall")) return -KEY_syscall;
5671 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5672 if (strEQ(d,"sysread")) return -KEY_sysread;
5673 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5676 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5685 if (strEQ(d,"tr")) return KEY_tr;
5688 if (strEQ(d,"tie")) return KEY_tie;
5691 if (strEQ(d,"tell")) return -KEY_tell;
5692 if (strEQ(d,"tied")) return KEY_tied;
5693 if (strEQ(d,"time")) return -KEY_time;
5696 if (strEQ(d,"times")) return -KEY_times;
5699 if (strEQ(d,"telldir")) return -KEY_telldir;
5702 if (strEQ(d,"truncate")) return -KEY_truncate;
5709 if (strEQ(d,"uc")) return -KEY_uc;
5712 if (strEQ(d,"use")) return KEY_use;
5715 if (strEQ(d,"undef")) return KEY_undef;
5716 if (strEQ(d,"until")) return KEY_until;
5717 if (strEQ(d,"untie")) return KEY_untie;
5718 if (strEQ(d,"utime")) return -KEY_utime;
5719 if (strEQ(d,"umask")) return -KEY_umask;
5722 if (strEQ(d,"unless")) return KEY_unless;
5723 if (strEQ(d,"unpack")) return -KEY_unpack;
5724 if (strEQ(d,"unlink")) return -KEY_unlink;
5727 if (strEQ(d,"unshift")) return -KEY_unshift;
5728 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5733 if (strEQ(d,"values")) return -KEY_values;
5734 if (strEQ(d,"vec")) return -KEY_vec;
5739 if (strEQ(d,"warn")) return -KEY_warn;
5740 if (strEQ(d,"wait")) return -KEY_wait;
5743 if (strEQ(d,"while")) return KEY_while;
5744 if (strEQ(d,"write")) return -KEY_write;
5747 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5750 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5755 if (len == 1) return -KEY_x;
5756 if (strEQ(d,"xor")) return -KEY_xor;
5759 if (len == 1) return KEY_y;
5768 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5772 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5773 if (ckWARN(WARN_SYNTAX)) {
5775 for (w = s+2; *w && level; w++) {
5782 for (; *w && isSPACE(*w); w++) ;
5783 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5784 Perl_warner(aTHX_ WARN_SYNTAX,
5785 "%s (...) interpreted as function",name);
5788 while (s < PL_bufend && isSPACE(*s))
5792 while (s < PL_bufend && isSPACE(*s))
5794 if (isIDFIRST_lazy_if(s,UTF)) {
5796 while (isALNUM_lazy_if(s,UTF))
5798 while (s < PL_bufend && isSPACE(*s))
5803 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5807 Perl_croak(aTHX_ "No comma allowed after %s", what);
5812 /* Either returns sv, or mortalizes sv and returns a new SV*.
5813 Best used as sv=new_constant(..., sv, ...).
5814 If s, pv are NULL, calls subroutine with one argument,
5815 and type is used with error messages only. */
5818 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5822 HV *table = GvHV(PL_hintgv); /* ^H */
5826 const char *why1, *why2, *why3;
5828 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5831 why2 = strEQ(key,"charnames")
5832 ? "(possibly a missing \"use charnames ...\")"
5834 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5835 (type ? type: "undef"), why2);
5837 /* This is convoluted and evil ("goto considered harmful")
5838 * but I do not understand the intricacies of all the different
5839 * failure modes of %^H in here. The goal here is to make
5840 * the most probable error message user-friendly. --jhi */
5845 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5846 (type ? type: "undef"), why1, why2, why3);
5848 yyerror(SvPVX(msg));
5852 cvp = hv_fetch(table, key, strlen(key), FALSE);
5853 if (!cvp || !SvOK(*cvp)) {
5856 why3 = "} is not defined";
5859 sv_2mortal(sv); /* Parent created it permanently */
5862 pv = sv_2mortal(newSVpvn(s, len));
5864 typesv = sv_2mortal(newSVpv(type, 0));
5866 typesv = &PL_sv_undef;
5868 PUSHSTACKi(PERLSI_OVERLOAD);
5880 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5884 /* Check the eval first */
5885 if (!PL_in_eval && SvTRUE(ERRSV)) {
5887 sv_catpv(ERRSV, "Propagated");
5888 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5890 res = SvREFCNT_inc(sv);
5894 (void)SvREFCNT_inc(res);
5903 why1 = "Call to &{$^H{";
5905 why3 = "}} did not return a defined value";
5914 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5916 register char *d = dest;
5917 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5920 Perl_croak(aTHX_ ident_too_long);
5921 if (isALNUM(*s)) /* UTF handled below */
5923 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5928 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5932 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
5933 char *t = s + UTF8SKIP(s);
5934 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
5936 if (d + (t - s) > e)
5937 Perl_croak(aTHX_ ident_too_long);
5938 Copy(s, d, t - s, char);
5951 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5961 e = d + destlen - 3; /* two-character token, ending NUL */
5963 while (isDIGIT(*s)) {
5965 Perl_croak(aTHX_ ident_too_long);
5972 Perl_croak(aTHX_ ident_too_long);
5973 if (isALNUM(*s)) /* UTF handled below */
5975 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5980 else if (*s == ':' && s[1] == ':') {
5984 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
5985 char *t = s + UTF8SKIP(s);
5986 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
5988 if (d + (t - s) > e)
5989 Perl_croak(aTHX_ ident_too_long);
5990 Copy(s, d, t - s, char);
6001 if (PL_lex_state != LEX_NORMAL)
6002 PL_lex_state = LEX_INTERPENDMAYBE;
6005 if (*s == '$' && s[1] &&
6006 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6019 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6024 if (isSPACE(s[-1])) {
6027 if (!SPACE_OR_TAB(ch)) {
6033 if (isIDFIRST_lazy_if(d,UTF)) {
6037 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6039 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6042 Copy(s, d, e - s, char);
6047 while ((isALNUM(*s) || *s == ':') && d < e)
6050 Perl_croak(aTHX_ ident_too_long);
6053 while (s < send && SPACE_OR_TAB(*s)) s++;
6054 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6055 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6056 const char *brack = *s == '[' ? "[...]" : "{...}";
6057 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6058 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6059 funny, dest, brack, funny, dest, brack);
6062 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6066 /* Handle extended ${^Foo} variables
6067 * 1999-02-27 mjd-perl-patch@plover.com */
6068 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6072 while (isALNUM(*s) && d < e) {
6076 Perl_croak(aTHX_ ident_too_long);
6081 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6082 PL_lex_state = LEX_INTERPEND;
6085 if (PL_lex_state == LEX_NORMAL) {
6086 if (ckWARN(WARN_AMBIGUOUS) &&
6087 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6089 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6090 "Ambiguous use of %c{%s} resolved to %c%s",
6091 funny, dest, funny, dest);
6096 s = bracket; /* let the parser handle it */
6100 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6101 PL_lex_state = LEX_INTERPEND;
6106 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6111 *pmfl |= PMf_GLOBAL;
6113 *pmfl |= PMf_CONTINUE;
6117 *pmfl |= PMf_MULTILINE;
6119 *pmfl |= PMf_SINGLELINE;
6121 *pmfl |= PMf_EXTENDED;
6125 S_scan_pat(pTHX_ char *start, I32 type)
6130 s = scan_str(start,FALSE,FALSE);
6133 SvREFCNT_dec(PL_lex_stuff);
6134 PL_lex_stuff = Nullsv;
6135 Perl_croak(aTHX_ "Search pattern not terminated");
6138 pm = (PMOP*)newPMOP(type, 0);
6139 if (PL_multi_open == '?')
6140 pm->op_pmflags |= PMf_ONCE;
6142 while (*s && strchr("iomsx", *s))
6143 pmflag(&pm->op_pmflags,*s++);
6146 while (*s && strchr("iogcmsx", *s))
6147 pmflag(&pm->op_pmflags,*s++);
6149 pm->op_pmpermflags = pm->op_pmflags;
6151 PL_lex_op = (OP*)pm;
6152 yylval.ival = OP_MATCH;
6157 S_scan_subst(pTHX_ char *start)
6164 yylval.ival = OP_NULL;
6166 s = scan_str(start,FALSE,FALSE);
6170 SvREFCNT_dec(PL_lex_stuff);
6171 PL_lex_stuff = Nullsv;
6172 Perl_croak(aTHX_ "Substitution pattern not terminated");
6175 if (s[-1] == PL_multi_open)
6178 first_start = PL_multi_start;
6179 s = scan_str(s,FALSE,FALSE);
6182 SvREFCNT_dec(PL_lex_stuff);
6183 PL_lex_stuff = Nullsv;
6185 SvREFCNT_dec(PL_lex_repl);
6186 PL_lex_repl = Nullsv;
6187 Perl_croak(aTHX_ "Substitution replacement not terminated");
6189 PL_multi_start = first_start; /* so whole substitution is taken together */
6191 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6197 else if (strchr("iogcmsx", *s))
6198 pmflag(&pm->op_pmflags,*s++);
6205 PL_sublex_info.super_bufptr = s;
6206 PL_sublex_info.super_bufend = PL_bufend;
6208 pm->op_pmflags |= PMf_EVAL;
6209 repl = newSVpvn("",0);
6211 sv_catpv(repl, es ? "eval " : "do ");
6212 sv_catpvn(repl, "{ ", 2);
6213 sv_catsv(repl, PL_lex_repl);
6214 sv_catpvn(repl, " };", 2);
6216 SvREFCNT_dec(PL_lex_repl);
6220 pm->op_pmpermflags = pm->op_pmflags;
6221 PL_lex_op = (OP*)pm;
6222 yylval.ival = OP_SUBST;
6227 S_scan_trans(pTHX_ char *start)
6238 yylval.ival = OP_NULL;
6240 s = scan_str(start,FALSE,FALSE);
6243 SvREFCNT_dec(PL_lex_stuff);
6244 PL_lex_stuff = Nullsv;
6245 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6247 if (s[-1] == PL_multi_open)
6250 s = scan_str(s,FALSE,FALSE);
6253 SvREFCNT_dec(PL_lex_stuff);
6254 PL_lex_stuff = Nullsv;
6256 SvREFCNT_dec(PL_lex_repl);
6257 PL_lex_repl = Nullsv;
6258 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6261 New(803,tbl,256,short);
6262 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6264 complement = del = squash = 0;
6265 while (strchr("cds", *s)) {
6267 complement = OPpTRANS_COMPLEMENT;
6269 del = OPpTRANS_DELETE;
6271 squash = OPpTRANS_SQUASH;
6274 o->op_private = del|squash|complement|
6275 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6276 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6279 yylval.ival = OP_TRANS;
6284 S_scan_heredoc(pTHX_ register char *s)
6287 I32 op_type = OP_SCALAR;
6294 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6298 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6301 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6302 if (*peek && strchr("`'\"",*peek)) {
6305 s = delimcpy(d, e, s, PL_bufend, term, &len);
6315 if (!isALNUM_lazy_if(s,UTF))
6316 deprecate("bare << to mean <<\"\"");
6317 for (; isALNUM_lazy_if(s,UTF); s++) {
6322 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6323 Perl_croak(aTHX_ "Delimiter for here document is too long");
6326 len = d - PL_tokenbuf;
6327 #ifndef PERL_STRICT_CR
6328 d = strchr(s, '\r');
6332 while (s < PL_bufend) {
6338 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6347 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6352 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6353 herewas = newSVpvn(s,PL_bufend-s);
6355 s--, herewas = newSVpvn(s,d-s);
6356 s += SvCUR(herewas);
6358 tmpstr = NEWSV(87,79);
6359 sv_upgrade(tmpstr, SVt_PVIV);
6364 else if (term == '`') {
6365 op_type = OP_BACKTICK;
6366 SvIVX(tmpstr) = '\\';
6370 PL_multi_start = CopLINE(PL_curcop);
6371 PL_multi_open = PL_multi_close = '<';
6372 term = *PL_tokenbuf;
6373 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6374 char *bufptr = PL_sublex_info.super_bufptr;
6375 char *bufend = PL_sublex_info.super_bufend;
6376 char *olds = s - SvCUR(herewas);
6377 s = strchr(bufptr, '\n');
6381 while (s < bufend &&
6382 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6384 CopLINE_inc(PL_curcop);
6387 CopLINE_set(PL_curcop, PL_multi_start);
6388 missingterm(PL_tokenbuf);
6390 sv_setpvn(herewas,bufptr,d-bufptr+1);
6391 sv_setpvn(tmpstr,d+1,s-d);
6393 sv_catpvn(herewas,s,bufend-s);
6394 (void)strcpy(bufptr,SvPVX(herewas));
6401 while (s < PL_bufend &&
6402 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6404 CopLINE_inc(PL_curcop);
6406 if (s >= PL_bufend) {
6407 CopLINE_set(PL_curcop, PL_multi_start);
6408 missingterm(PL_tokenbuf);
6410 sv_setpvn(tmpstr,d+1,s-d);
6412 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6414 sv_catpvn(herewas,s,PL_bufend-s);
6415 sv_setsv(PL_linestr,herewas);
6416 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6417 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6420 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6421 while (s >= PL_bufend) { /* multiple line string? */
6423 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6424 CopLINE_set(PL_curcop, PL_multi_start);
6425 missingterm(PL_tokenbuf);
6427 CopLINE_inc(PL_curcop);
6428 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6429 #ifndef PERL_STRICT_CR
6430 if (PL_bufend - PL_linestart >= 2) {
6431 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6432 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6434 PL_bufend[-2] = '\n';
6436 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6438 else if (PL_bufend[-1] == '\r')
6439 PL_bufend[-1] = '\n';
6441 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6442 PL_bufend[-1] = '\n';
6444 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6445 SV *sv = NEWSV(88,0);
6447 sv_upgrade(sv, SVt_PVMG);
6448 sv_setsv(sv,PL_linestr);
6449 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6451 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6454 sv_catsv(PL_linestr,herewas);
6455 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6459 sv_catsv(tmpstr,PL_linestr);
6464 PL_multi_end = CopLINE(PL_curcop);
6465 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6466 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6467 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6469 SvREFCNT_dec(herewas);
6470 if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6472 PL_lex_stuff = tmpstr;
6473 yylval.ival = op_type;
6478 takes: current position in input buffer
6479 returns: new position in input buffer
6480 side-effects: yylval and lex_op are set.
6485 <FH> read from filehandle
6486 <pkg::FH> read from package qualified filehandle
6487 <pkg'FH> read from package qualified filehandle
6488 <$fh> read from filehandle in $fh
6494 S_scan_inputsymbol(pTHX_ char *start)
6496 register char *s = start; /* current position in buffer */
6502 d = PL_tokenbuf; /* start of temp holding space */
6503 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6504 end = strchr(s, '\n');
6507 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6509 /* die if we didn't have space for the contents of the <>,
6510 or if it didn't end, or if we see a newline
6513 if (len >= sizeof PL_tokenbuf)
6514 Perl_croak(aTHX_ "Excessively long <> operator");
6516 Perl_croak(aTHX_ "Unterminated <> operator");
6521 Remember, only scalar variables are interpreted as filehandles by
6522 this code. Anything more complex (e.g., <$fh{$num}>) will be
6523 treated as a glob() call.
6524 This code makes use of the fact that except for the $ at the front,
6525 a scalar variable and a filehandle look the same.
6527 if (*d == '$' && d[1]) d++;
6529 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6530 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6533 /* If we've tried to read what we allow filehandles to look like, and
6534 there's still text left, then it must be a glob() and not a getline.
6535 Use scan_str to pull out the stuff between the <> and treat it
6536 as nothing more than a string.
6539 if (d - PL_tokenbuf != len) {
6540 yylval.ival = OP_GLOB;
6542 s = scan_str(start,FALSE,FALSE);
6544 Perl_croak(aTHX_ "Glob not terminated");
6548 /* we're in a filehandle read situation */
6551 /* turn <> into <ARGV> */
6553 (void)strcpy(d,"ARGV");
6555 /* if <$fh>, create the ops to turn the variable into a
6561 /* try to find it in the pad for this block, otherwise find
6562 add symbol table ops
6564 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6565 OP *o = newOP(OP_PADSV, 0);
6567 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6570 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6571 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6572 newUNOP(OP_RV2SV, 0,
6573 newGVOP(OP_GV, 0, gv)));
6575 PL_lex_op->op_flags |= OPf_SPECIAL;
6576 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6577 yylval.ival = OP_NULL;
6580 /* If it's none of the above, it must be a literal filehandle
6581 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6583 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6584 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6585 yylval.ival = OP_NULL;
6594 takes: start position in buffer
6595 keep_quoted preserve \ on the embedded delimiter(s)
6596 keep_delims preserve the delimiters around the string
6597 returns: position to continue reading from buffer
6598 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6599 updates the read buffer.
6601 This subroutine pulls a string out of the input. It is called for:
6602 q single quotes q(literal text)
6603 ' single quotes 'literal text'
6604 qq double quotes qq(interpolate $here please)
6605 " double quotes "interpolate $here please"
6606 qx backticks qx(/bin/ls -l)
6607 ` backticks `/bin/ls -l`
6608 qw quote words @EXPORT_OK = qw( func() $spam )
6609 m// regexp match m/this/
6610 s/// regexp substitute s/this/that/
6611 tr/// string transliterate tr/this/that/
6612 y/// string transliterate y/this/that/
6613 ($*@) sub prototypes sub foo ($)
6614 (stuff) sub attr parameters sub foo : attr(stuff)
6615 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6617 In most of these cases (all but <>, patterns and transliterate)
6618 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6619 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6620 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6623 It skips whitespace before the string starts, and treats the first
6624 character as the delimiter. If the delimiter is one of ([{< then
6625 the corresponding "close" character )]}> is used as the closing
6626 delimiter. It allows quoting of delimiters, and if the string has
6627 balanced delimiters ([{<>}]) it allows nesting.
6629 The lexer always reads these strings into lex_stuff, except in the
6630 case of the operators which take *two* arguments (s/// and tr///)
6631 when it checks to see if lex_stuff is full (presumably with the 1st
6632 arg to s or tr) and if so puts the string into lex_repl.
6637 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6639 SV *sv; /* scalar value: string */
6640 char *tmps; /* temp string, used for delimiter matching */
6641 register char *s = start; /* current position in the buffer */
6642 register char term; /* terminating character */
6643 register char *to; /* current position in the sv's data */
6644 I32 brackets = 1; /* bracket nesting level */
6645 bool has_utf8 = FALSE; /* is there any utf8 content? */
6647 /* skip space before the delimiter */
6651 /* mark where we are, in case we need to report errors */
6654 /* after skipping whitespace, the next character is the terminator */
6656 if (UTF8_IS_CONTINUED(term) && UTF)
6659 /* mark where we are */
6660 PL_multi_start = CopLINE(PL_curcop);
6661 PL_multi_open = term;
6663 /* find corresponding closing delimiter */
6664 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6666 PL_multi_close = term;
6668 /* create a new SV to hold the contents. 87 is leak category, I'm
6669 assuming. 79 is the SV's initial length. What a random number. */
6671 sv_upgrade(sv, SVt_PVIV);
6673 (void)SvPOK_only(sv); /* validate pointer */
6675 /* move past delimiter and try to read a complete string */
6677 sv_catpvn(sv, s, 1);
6680 /* extend sv if need be */
6681 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6682 /* set 'to' to the next character in the sv's string */
6683 to = SvPVX(sv)+SvCUR(sv);
6685 /* if open delimiter is the close delimiter read unbridle */
6686 if (PL_multi_open == PL_multi_close) {
6687 for (; s < PL_bufend; s++,to++) {
6688 /* embedded newlines increment the current line number */
6689 if (*s == '\n' && !PL_rsfp)
6690 CopLINE_inc(PL_curcop);
6691 /* handle quoted delimiters */
6692 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6693 if (!keep_quoted && s[1] == term)
6695 /* any other quotes are simply copied straight through */
6699 /* terminate when run out of buffer (the for() condition), or
6700 have found the terminator */
6701 else if (*s == term)
6703 else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF)
6709 /* if the terminator isn't the same as the start character (e.g.,
6710 matched brackets), we have to allow more in the quoting, and
6711 be prepared for nested brackets.
6714 /* read until we run out of string, or we find the terminator */
6715 for (; s < PL_bufend; s++,to++) {
6716 /* embedded newlines increment the line count */
6717 if (*s == '\n' && !PL_rsfp)
6718 CopLINE_inc(PL_curcop);
6719 /* backslashes can escape the open or closing characters */
6720 if (*s == '\\' && s+1 < PL_bufend) {
6722 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6727 /* allow nested opens and closes */
6728 else if (*s == PL_multi_close && --brackets <= 0)
6730 else if (*s == PL_multi_open)
6732 else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF)
6737 /* terminate the copied string and update the sv's end-of-string */
6739 SvCUR_set(sv, to - SvPVX(sv));
6742 * this next chunk reads more into the buffer if we're not done yet
6746 break; /* handle case where we are done yet :-) */
6748 #ifndef PERL_STRICT_CR
6749 if (to - SvPVX(sv) >= 2) {
6750 if ((to[-2] == '\r' && to[-1] == '\n') ||
6751 (to[-2] == '\n' && to[-1] == '\r'))
6755 SvCUR_set(sv, to - SvPVX(sv));
6757 else if (to[-1] == '\r')
6760 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6764 /* if we're out of file, or a read fails, bail and reset the current
6765 line marker so we can report where the unterminated string began
6768 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6770 CopLINE_set(PL_curcop, PL_multi_start);
6773 /* we read a line, so increment our line counter */
6774 CopLINE_inc(PL_curcop);
6776 /* update debugger info */
6777 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6778 SV *sv = NEWSV(88,0);
6780 sv_upgrade(sv, SVt_PVMG);
6781 sv_setsv(sv,PL_linestr);
6782 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6785 /* having changed the buffer, we must update PL_bufend */
6786 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6789 /* at this point, we have successfully read the delimited string */
6792 sv_catpvn(sv, s, 1);
6795 PL_multi_end = CopLINE(PL_curcop);
6798 /* if we allocated too much space, give some back */
6799 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6800 SvLEN_set(sv, SvCUR(sv) + 1);
6801 Renew(SvPVX(sv), SvLEN(sv), char);
6804 /* decide whether this is the first or second quoted string we've read
6817 takes: pointer to position in buffer
6818 returns: pointer to new position in buffer
6819 side-effects: builds ops for the constant in yylval.op
6821 Read a number in any of the formats that Perl accepts:
6823 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6824 [\d_]+(\.[\d_]*)?[Ee](\d+)
6826 Underbars (_) are allowed in decimal numbers. If -w is on,
6827 underbars before a decimal point must be at three digit intervals.
6829 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6832 If it reads a number without a decimal point or an exponent, it will
6833 try converting the number to an integer and see if it can do so
6834 without loss of precision.
6838 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6840 register char *s = start; /* current position in buffer */
6841 register char *d; /* destination in temp buffer */
6842 register char *e; /* end of temp buffer */
6843 NV nv; /* number read, as a double */
6844 SV *sv = Nullsv; /* place to put the converted number */
6845 bool floatit; /* boolean: int or float? */
6846 char *lastub = 0; /* position of last underbar */
6847 static char number_too_long[] = "Number too long";
6849 /* We use the first character to decide what type of number this is */
6853 Perl_croak(aTHX_ "panic: scan_num");
6855 /* if it starts with a 0, it could be an octal number, a decimal in
6856 0.13 disguise, or a hexadecimal number, or a binary number. */
6860 u holds the "number so far"
6861 shift the power of 2 of the base
6862 (hex == 4, octal == 3, binary == 1)
6863 overflowed was the number more than we can hold?
6865 Shift is used when we add a digit. It also serves as an "are
6866 we in octal/hex/binary?" indicator to disallow hex characters
6872 bool overflowed = FALSE;
6873 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6874 static char* bases[5] = { "", "binary", "", "octal",
6876 static char* Bases[5] = { "", "Binary", "", "Octal",
6878 static char *maxima[5] = { "",
6879 "0b11111111111111111111111111111111",
6883 char *base, *Base, *max;
6889 } else if (s[1] == 'b') {
6893 /* check for a decimal in disguise */
6894 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6896 /* so it must be octal */
6900 base = bases[shift];
6901 Base = Bases[shift];
6902 max = maxima[shift];
6904 /* read the rest of the number */
6906 /* x is used in the overflow test,
6907 b is the digit we're adding on. */
6912 /* if we don't mention it, we're done */
6921 /* 8 and 9 are not octal */
6924 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6928 case '2': case '3': case '4':
6929 case '5': case '6': case '7':
6931 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6935 b = *s++ & 15; /* ASCII digit -> value of digit */
6939 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6940 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6941 /* make sure they said 0x */
6946 /* Prepare to put the digit we have onto the end
6947 of the number so far. We check for overflows.
6952 x = u << shift; /* make room for the digit */
6954 if ((x >> shift) != u
6955 && !(PL_hints & HINT_NEW_BINARY)) {
6958 if (ckWARN_d(WARN_OVERFLOW))
6959 Perl_warner(aTHX_ WARN_OVERFLOW,
6960 "Integer overflow in %s number",
6963 u = x | b; /* add the digit to the end */
6966 n *= nvshift[shift];
6967 /* If an NV has not enough bits in its
6968 * mantissa to represent an UV this summing of
6969 * small low-order numbers is a waste of time
6970 * (because the NV cannot preserve the
6971 * low-order bits anyway): we could just
6972 * remember when did we overflow and in the
6973 * end just multiply n by the right
6981 /* if we get here, we had success: make a scalar value from
6987 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6988 Perl_warner(aTHX_ WARN_PORTABLE,
6989 "%s number > %s non-portable",
6995 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6996 Perl_warner(aTHX_ WARN_PORTABLE,
6997 "%s number > %s non-portable",
7002 if (PL_hints & HINT_NEW_BINARY)
7003 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7008 handle decimal numbers.
7009 we're also sent here when we read a 0 as the first digit
7011 case '1': case '2': case '3': case '4': case '5':
7012 case '6': case '7': case '8': case '9': case '.':
7015 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7018 /* read next group of digits and _ and copy into d */
7019 while (isDIGIT(*s) || *s == '_') {
7020 /* skip underscores, checking for misplaced ones
7024 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7025 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7029 /* check for end of fixed-length buffer */
7031 Perl_croak(aTHX_ number_too_long);
7032 /* if we're ok, copy the character */
7037 /* final misplaced underbar check */
7038 if (lastub && s - lastub != 3) {
7039 if (ckWARN(WARN_SYNTAX))
7040 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7043 /* read a decimal portion if there is one. avoid
7044 3..5 being interpreted as the number 3. followed
7047 if (*s == '.' && s[1] != '.') {
7051 /* copy, ignoring underbars, until we run out of
7052 digits. Note: no misplaced underbar checks!
7054 for (; isDIGIT(*s) || *s == '_'; s++) {
7055 /* fixed length buffer check */
7057 Perl_croak(aTHX_ number_too_long);
7061 if (*s == '.' && isDIGIT(s[1])) {
7062 /* oops, it's really a v-string, but without the "v" */
7068 /* read exponent part, if present */
7069 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7073 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7074 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7076 /* allow positive or negative exponent */
7077 if (*s == '+' || *s == '-')
7080 /* read digits of exponent (no underbars :-) */
7081 while (isDIGIT(*s)) {
7083 Perl_croak(aTHX_ number_too_long);
7088 /* terminate the string */
7091 /* make an sv from the string */
7094 #if defined(Strtol) && defined(Strtoul)
7097 strtol/strtoll sets errno to ERANGE if the number is too big
7098 for an integer. We try to do an integer conversion first
7099 if no characters indicating "float" have been found.
7106 if (*PL_tokenbuf == '-')
7107 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7109 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7111 floatit = TRUE; /* Probably just too large. */
7112 else if (*PL_tokenbuf == '-')
7114 else if (uv <= IV_MAX)
7115 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7120 nv = Atof(PL_tokenbuf);
7125 No working strtou?ll?.
7127 Unfortunately atol() doesn't do range checks (returning
7128 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7129 everywhere [1], so we cannot use use atol() (or atoll()).
7130 If we could, they would be used, as Atol(), very much like
7131 Strtol() and Strtoul() are used above.
7133 [1] XXX Configure test needed to check for atol()
7134 (and atoll()) overflow behaviour XXX
7138 We need to do this the hard way. */
7140 nv = Atof(PL_tokenbuf);
7142 /* See if we can make do with an integer value without loss of
7143 precision. We use U_V to cast to a UV, because some
7144 compilers have issues. Then we try casting it back and see
7145 if it was the same [1]. We only do this if we know we
7146 specifically read an integer. If floatit is true, then we
7147 don't need to do the conversion at all.
7149 [1] Note that this is lossy if our NVs cannot preserve our
7150 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7151 and NV_PRESERVES_UV_BITS (a number), but in general we really
7152 do hope all such potentially lossy platforms have strtou?ll?
7153 to do a lossless IV/UV conversion.
7155 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7156 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7157 as NV_DIG and NV_MANT_DIG)?
7163 if (!floatit && (NV)uv == nv) {
7165 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7173 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7174 (PL_hints & HINT_NEW_INTEGER) )
7175 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7176 (floatit ? "float" : "integer"),
7180 /* if it starts with a v, it could be a v-string */
7186 while (isDIGIT(*pos) || *pos == '_')
7188 if (!isALPHA(*pos)) {
7190 U8 tmpbuf[UTF8_MAXLEN+1];
7193 s++; /* get past 'v' */
7196 sv_setpvn(sv, "", 0);
7199 if (*s == '0' && isDIGIT(s[1]))
7200 yyerror("Octal number in vector unsupported");
7203 /* this is atoi() that tolerates underscores */
7206 while (--end >= s) {
7211 rev += (*end - '0') * mult;
7213 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7214 Perl_warner(aTHX_ WARN_OVERFLOW,
7215 "Integer overflow in decimal number");
7218 tmpend = uv_to_utf8(tmpbuf, rev);
7219 utf8 = utf8 || rev > 127;
7220 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7221 if (*pos == '.' && isDIGIT(pos[1]))
7227 while (isDIGIT(*pos) || *pos == '_')
7236 sv_utf8_downgrade(sv, TRUE);
7243 /* make the op for the constant and return */
7246 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7248 lvalp->opval = Nullop;
7254 S_scan_formline(pTHX_ register char *s)
7258 SV *stuff = newSVpvn("",0);
7259 bool needargs = FALSE;
7262 if (*s == '.' || *s == /*{*/'}') {
7264 #ifdef PERL_STRICT_CR
7265 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7267 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7269 if (*t == '\n' || t == PL_bufend)
7272 if (PL_in_eval && !PL_rsfp) {
7273 eol = strchr(s,'\n');
7278 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7280 for (t = s; t < eol; t++) {
7281 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7283 goto enough; /* ~~ must be first line in formline */
7285 if (*t == '@' || *t == '^')
7288 sv_catpvn(stuff, s, eol-s);
7289 #ifndef PERL_STRICT_CR
7290 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7291 char *end = SvPVX(stuff) + SvCUR(stuff);
7300 s = filter_gets(PL_linestr, PL_rsfp, 0);
7301 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7302 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7305 yyerror("Format not terminated");
7315 PL_lex_state = LEX_NORMAL;
7316 PL_nextval[PL_nexttoke].ival = 0;
7320 PL_lex_state = LEX_FORMLINE;
7321 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7323 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7327 SvREFCNT_dec(stuff);
7328 PL_lex_formbrack = 0;
7339 PL_cshlen = strlen(PL_cshname);
7344 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7346 I32 oldsavestack_ix = PL_savestack_ix;
7347 CV* outsidecv = PL_compcv;
7351 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7353 SAVEI32(PL_subline);
7354 save_item(PL_subname);
7357 SAVESPTR(PL_comppad_name);
7358 SAVESPTR(PL_compcv);
7359 SAVEI32(PL_comppad_name_fill);
7360 SAVEI32(PL_min_intro_pending);
7361 SAVEI32(PL_max_intro_pending);
7362 SAVEI32(PL_pad_reset_pending);
7364 PL_compcv = (CV*)NEWSV(1104,0);
7365 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7366 CvFLAGS(PL_compcv) |= flags;
7368 PL_comppad = newAV();
7369 av_push(PL_comppad, Nullsv);
7370 PL_curpad = AvARRAY(PL_comppad);
7371 PL_comppad_name = newAV();
7372 PL_comppad_name_fill = 0;
7373 PL_min_intro_pending = 0;
7375 PL_subline = CopLINE(PL_curcop);
7377 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7378 PL_curpad[0] = (SV*)newAV();
7379 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7380 #endif /* USE_THREADS */
7382 comppadlist = newAV();
7383 AvREAL_off(comppadlist);
7384 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7385 av_store(comppadlist, 1, (SV*)PL_comppad);
7387 CvPADLIST(PL_compcv) = comppadlist;
7388 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7390 CvOWNER(PL_compcv) = 0;
7391 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7392 MUTEX_INIT(CvMUTEXP(PL_compcv));
7393 #endif /* USE_THREADS */
7395 return oldsavestack_ix;
7399 Perl_yywarn(pTHX_ char *s)
7401 PL_in_eval |= EVAL_WARNONLY;
7403 PL_in_eval &= ~EVAL_WARNONLY;
7408 Perl_yyerror(pTHX_ char *s)
7411 char *context = NULL;
7415 if (!yychar || (yychar == ';' && !PL_rsfp))
7417 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7418 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7419 while (isSPACE(*PL_oldoldbufptr))
7421 context = PL_oldoldbufptr;
7422 contlen = PL_bufptr - PL_oldoldbufptr;
7424 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7425 PL_oldbufptr != PL_bufptr) {
7426 while (isSPACE(*PL_oldbufptr))
7428 context = PL_oldbufptr;
7429 contlen = PL_bufptr - PL_oldbufptr;
7431 else if (yychar > 255)
7432 where = "next token ???";
7433 #ifdef USE_PURE_BISON
7434 /* GNU Bison sets the value -2 */
7435 else if (yychar == -2) {
7437 else if ((yychar & 127) == 127) {
7439 if (PL_lex_state == LEX_NORMAL ||
7440 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7441 where = "at end of line";
7442 else if (PL_lex_inpat)
7443 where = "within pattern";
7445 where = "within string";
7448 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7450 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7451 else if (isPRINT_LC(yychar))
7452 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7454 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7455 where = SvPVX(where_sv);
7457 msg = sv_2mortal(newSVpv(s, 0));
7458 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7459 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7461 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7463 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7464 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7465 Perl_sv_catpvf(aTHX_ msg,
7466 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7467 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7470 if (PL_in_eval & EVAL_WARNONLY)
7471 Perl_warn(aTHX_ "%"SVf, msg);
7474 if (PL_error_count >= 10) {
7475 if (PL_in_eval && SvCUR(ERRSV))
7476 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7477 ERRSV, CopFILE(PL_curcop));
7479 Perl_croak(aTHX_ "%s has too many errors.\n",
7480 CopFILE(PL_curcop));
7483 PL_in_my_stash = Nullhv;
7488 S_swallow_bom(pTHX_ U8 *s)
7491 slen = SvCUR(PL_linestr);
7495 /* UTF-16 little-endian */
7496 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7497 Perl_croak(aTHX_ "Unsupported script encoding");
7498 #ifndef PERL_NO_UTF16_FILTER
7499 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7501 if (PL_bufend > (char*)s) {
7505 filter_add(utf16rev_textfilter, NULL);
7506 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7507 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7508 PL_bufend - (char*)s - 1,
7510 Copy(news, s, newlen, U8);
7511 SvCUR_set(PL_linestr, newlen);
7512 PL_bufend = SvPVX(PL_linestr) + newlen;
7513 news[newlen++] = '\0';
7517 Perl_croak(aTHX_ "Unsupported script encoding");
7522 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7523 #ifndef PERL_NO_UTF16_FILTER
7524 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7526 if (PL_bufend > (char *)s) {
7530 filter_add(utf16_textfilter, NULL);
7531 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7532 PL_bufend = (char*)utf16_to_utf8(s, news,
7533 PL_bufend - (char*)s,
7535 Copy(news, s, newlen, U8);
7536 SvCUR_set(PL_linestr, newlen);
7537 PL_bufend = SvPVX(PL_linestr) + newlen;
7538 news[newlen++] = '\0';
7542 Perl_croak(aTHX_ "Unsupported script encoding");
7547 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7548 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7553 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7554 s[2] == 0xFE && s[3] == 0xFF)
7556 Perl_croak(aTHX_ "Unsupported script encoding");
7568 * Restore a source filter.
7572 restore_rsfp(pTHXo_ void *f)
7574 PerlIO *fp = (PerlIO*)f;
7576 if (PL_rsfp == PerlIO_stdin())
7577 PerlIO_clearerr(PL_rsfp);
7578 else if (PL_rsfp && (PL_rsfp != fp))
7579 PerlIO_close(PL_rsfp);
7583 #ifndef PERL_NO_UTF16_FILTER
7585 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7587 I32 count = FILTER_READ(idx+1, sv, maxlen);
7592 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7593 if (!*SvPV_nolen(sv))
7594 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7597 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7598 sv_usepvn(sv, (char*)tmps, tend - tmps);
7604 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7606 I32 count = FILTER_READ(idx+1, sv, maxlen);
7611 if (!*SvPV_nolen(sv))
7612 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7615 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7616 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7617 sv_usepvn(sv, (char*)tmps, tend - tmps);