3 * Copyright (c) 1991-2000, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
78 # define YYMAXLEVEL 100
80 YYSTYPE* yylval_pointer[YYMAXLEVEL];
81 int* yychar_pointer[YYMAXLEVEL];
85 # define yylval (*yylval_pointer[yyactlevel])
86 # define yychar (*yychar_pointer[yyactlevel])
87 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
89 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
94 /* CLINE is a macro that ensures PL_copline has a sane value */
99 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
102 * Convenience functions to return different tokens and prime the
103 * lexer for the next token. They all take an argument.
105 * TOKEN : generic token (used for '(', DOLSHARP, etc)
106 * OPERATOR : generic operator
107 * AOPERATOR : assignment operator
108 * PREBLOCK : beginning the block after an if, while, foreach, ...
109 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
110 * PREREF : *EXPR where EXPR is not a simple identifier
111 * TERM : expression term
112 * LOOPX : loop exiting command (goto, last, dump, etc)
113 * FTST : file test operator
114 * FUN0 : zero-argument function
115 * FUN1 : not used, except for not, which isn't a UNIOP
116 * BOop : bitwise or or xor
118 * SHop : shift operator
119 * PWop : power operator
120 * PMop : pattern-matching operator
121 * Aop : addition-level operator
122 * Mop : multiplication-level operator
123 * Eop : equality-testing operator
124 * Rop : relational operator <= != gt
126 * Also see LOP and lop() below.
129 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
130 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
131 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
132 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
133 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
134 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
135 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
136 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
137 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
138 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
139 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
140 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
141 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
142 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
143 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
144 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
145 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
146 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
147 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
148 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
150 /* This bit of chicanery makes a unary function followed by
151 * a parenthesis into a function with one argument, highest precedence.
153 #define UNI(f) return(yylval.ival = f, \
156 PL_last_uni = PL_oldbufptr, \
157 PL_last_lop_op = f, \
158 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
160 #define UNIBRACK(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
165 /* grandfather return to old style */
166 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
171 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
172 * into an OP_ANDASSIGN or OP_ORASSIGN
176 S_ao(pTHX_ int toketype)
178 if (*PL_bufptr == '=') {
180 if (toketype == ANDAND)
181 yylval.ival = OP_ANDASSIGN;
182 else if (toketype == OROR)
183 yylval.ival = OP_ORASSIGN;
191 * When Perl expects an operator and finds something else, no_op
192 * prints the warning. It always prints "<something> found where
193 * operator expected. It prints "Missing semicolon on previous line?"
194 * if the surprise occurs at the start of the line. "do you need to
195 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
196 * where the compiler doesn't know if foo is a method call or a function.
197 * It prints "Missing operator before end of line" if there's nothing
198 * after the missing operator, or "... before <...>" if there is something
199 * after the missing operator.
203 S_no_op(pTHX_ char *what, char *s)
205 char *oldbp = PL_bufptr;
206 bool is_first = (PL_oldbufptr == PL_linestart);
212 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
214 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
215 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
217 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
218 if (t < PL_bufptr && isSPACE(*t))
219 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
220 t - PL_oldoldbufptr, PL_oldoldbufptr);
224 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
231 * Complain about missing quote/regexp/heredoc terminator.
232 * If it's called with (char *)NULL then it cauterizes the line buffer.
233 * If we're in a delimited string and the delimiter is a control
234 * character, it's reformatted into a two-char sequence like ^C.
239 S_missingterm(pTHX_ char *s)
244 char *nl = strrchr(s,'\n');
250 iscntrl(PL_multi_close)
252 PL_multi_close < 32 || PL_multi_close == 127
256 tmpbuf[1] = toCTRL(PL_multi_close);
262 *tmpbuf = PL_multi_close;
266 q = strchr(s,'"') ? '\'' : '"';
267 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
275 Perl_deprecate(pTHX_ char *s)
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 */
1314 /* (now in tr/// code again) */
1316 if (*s & 0x80 && (this_utf8 || has_utf8)) {
1317 STRLEN len = (STRLEN) -1;
1320 uv = utf8_to_uv((U8*)s, send - s, &len, UTF8_CHECK_ONLY);
1322 if (len == (STRLEN)-1) {
1323 /* Illegal UTF8 (a high-bit byte), make it valid. */
1324 char *old_pvx = SvPVX(sv);
1325 /* need space for one extra char (NOTE: SvCUR() not set here) */
1326 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1327 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1338 if (*s == '\\' && s+1 < send) {
1339 bool to_be_utf8 = FALSE;
1343 /* some backslashes we leave behind */
1344 if (*leaveit && *s && strchr(leaveit, *s)) {
1350 /* deprecate \1 in strings and substitution replacements */
1351 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1352 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1354 if (ckWARN(WARN_SYNTAX))
1355 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1360 /* string-change backslash escapes */
1361 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1366 /* if we get here, it's either a quoted -, or a digit */
1369 /* quoted - in transliterations */
1371 if (PL_lex_inwhat == OP_TRANS) {
1378 if (ckWARN(WARN_MISC) && isALNUM(*s))
1379 Perl_warner(aTHX_ WARN_MISC,
1380 "Unrecognized escape \\%c passed through",
1382 /* default action is to copy the quoted character */
1387 /* \132 indicates an octal constant */
1388 case '0': case '1': case '2': case '3':
1389 case '4': case '5': case '6': case '7':
1391 STRLEN len = 0; /* disallow underscores */
1392 uv = (UV)scan_oct(s, 3, &len);
1395 goto NUM_ESCAPE_INSERT;
1397 /* \x24 indicates a hex constant */
1401 char* e = strchr(s, '}');
1403 yyerror("Missing right brace on \\x{}");
1407 STRLEN len = 1; /* allow underscores */
1408 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1415 STRLEN len = 0; /* disallow underscores */
1416 uv = (UV)scan_hex(s, 2, &len);
1422 /* Insert oct or hex escaped character.
1423 * There will always enough room in sv since such escapes will
1424 * be longer than any utf8 sequence they can end up as
1427 if (!has_utf8 && (to_be_utf8 || uv > 255)) {
1428 /* might need to recode whatever we have accumulated so far
1429 * if it contains any hibit chars
1433 for (c = SvPVX(sv); c < d; c++) {
1438 char *old_pvx = SvPVX(sv);
1440 d = SvGROW(sv, SvCUR(sv) + hicount + 1) + (d - old_pvx);
1449 uv_to_utf8((U8*)dst, (U8)*src--);
1459 if (to_be_utf8 || uv > 255) {
1460 d = (char*)uv_to_utf8((U8*)d, uv);
1472 /* \N{latin small letter a} is a named character */
1476 char* e = strchr(s, '}');
1482 yyerror("Missing right brace on \\N{}");
1486 res = newSVpvn(s + 1, e - s - 1);
1487 res = new_constant( Nullch, 0, "charnames",
1488 res, Nullsv, "\\N{...}" );
1489 str = SvPV(res,len);
1490 if (!has_utf8 && SvUTF8(res)) {
1491 char *ostart = SvPVX(sv);
1492 SvCUR_set(sv, d - ostart);
1495 sv_utf8_upgrade(sv);
1496 /* this just broke our allocation above... */
1497 SvGROW(sv, send - start);
1498 d = SvPVX(sv) + SvCUR(sv);
1501 if (len > e - s + 4) {
1502 char *odest = SvPVX(sv);
1504 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1505 d = SvPVX(sv) + (d - odest);
1507 Copy(str, d, len, char);
1514 yyerror("Missing braces on \\N{}");
1517 /* \c is a control character */
1534 /* printf-style backslashes, formfeeds, newlines, etc */
1552 *d++ = '\047'; /* CP 1047 */
1555 *d++ = '\057'; /* CP 1047 */
1569 } /* end if (backslash) */
1572 } /* while loop to process each character */
1574 /* terminate the string and set up the sv */
1576 SvCUR_set(sv, d - SvPVX(sv));
1581 /* shrink the sv if we allocated more than we used */
1582 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1583 SvLEN_set(sv, SvCUR(sv) + 1);
1584 Renew(SvPVX(sv), SvLEN(sv), char);
1587 /* return the substring (via yylval) only if we parsed anything */
1588 if (s > PL_bufptr) {
1589 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1590 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1592 ( PL_lex_inwhat == OP_TRANS
1594 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1597 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1604 * Returns TRUE if there's more to the expression (e.g., a subscript),
1607 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1609 * ->[ and ->{ return TRUE
1610 * { and [ outside a pattern are always subscripts, so return TRUE
1611 * if we're outside a pattern and it's not { or [, then return FALSE
1612 * if we're in a pattern and the first char is a {
1613 * {4,5} (any digits around the comma) returns FALSE
1614 * if we're in a pattern and the first char is a [
1616 * [SOMETHING] has a funky algorithm to decide whether it's a
1617 * character class or not. It has to deal with things like
1618 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1619 * anything else returns TRUE
1622 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1625 S_intuit_more(pTHX_ register char *s)
1627 if (PL_lex_brackets)
1629 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1631 if (*s != '{' && *s != '[')
1636 /* In a pattern, so maybe we have {n,m}. */
1653 /* On the other hand, maybe we have a character class */
1656 if (*s == ']' || *s == '^')
1659 /* this is terrifying, and it works */
1660 int weight = 2; /* let's weigh the evidence */
1662 unsigned char un_char = 255, last_un_char;
1663 char *send = strchr(s,']');
1664 char tmpbuf[sizeof PL_tokenbuf * 4];
1666 if (!send) /* has to be an expression */
1669 Zero(seen,256,char);
1672 else if (isDIGIT(*s)) {
1674 if (isDIGIT(s[1]) && s[2] == ']')
1680 for (; s < send; s++) {
1681 last_un_char = un_char;
1682 un_char = (unsigned char)*s;
1687 weight -= seen[un_char] * 10;
1688 if (isALNUM_lazy_if(s+1,UTF)) {
1689 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1690 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1695 else if (*s == '$' && s[1] &&
1696 strchr("[#!%*<>()-=",s[1])) {
1697 if (/*{*/ strchr("])} =",s[2]))
1706 if (strchr("wds]",s[1]))
1708 else if (seen['\''] || seen['"'])
1710 else if (strchr("rnftbxcav",s[1]))
1712 else if (isDIGIT(s[1])) {
1714 while (s[1] && isDIGIT(s[1]))
1724 if (strchr("aA01! ",last_un_char))
1726 if (strchr("zZ79~",s[1]))
1728 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1729 weight -= 5; /* cope with negative subscript */
1732 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1733 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1738 if (keyword(tmpbuf, d - tmpbuf))
1741 if (un_char == last_un_char + 1)
1743 weight -= seen[un_char];
1748 if (weight >= 0) /* probably a character class */
1758 * Does all the checking to disambiguate
1760 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1761 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1763 * First argument is the stuff after the first token, e.g. "bar".
1765 * Not a method if bar is a filehandle.
1766 * Not a method if foo is a subroutine prototyped to take a filehandle.
1767 * Not a method if it's really "Foo $bar"
1768 * Method if it's "foo $bar"
1769 * Not a method if it's really "print foo $bar"
1770 * Method if it's really "foo package::" (interpreted as package->foo)
1771 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1772 * Not a method if bar is a filehandle or package, but is quoted with
1777 S_intuit_method(pTHX_ char *start, GV *gv)
1779 char *s = start + (*start == '$');
1780 char tmpbuf[sizeof PL_tokenbuf];
1788 if ((cv = GvCVu(gv))) {
1789 char *proto = SvPVX(cv);
1799 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1800 /* start is the beginning of the possible filehandle/object,
1801 * and s is the end of it
1802 * tmpbuf is a copy of it
1805 if (*start == '$') {
1806 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1811 return *s == '(' ? FUNCMETH : METHOD;
1813 if (!keyword(tmpbuf, len)) {
1814 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1819 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1820 if (indirgv && GvCVu(indirgv))
1822 /* filehandle or package name makes it a method */
1823 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1825 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1826 return 0; /* no assumptions -- "=>" quotes bearword */
1828 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1829 newSVpvn(tmpbuf,len));
1830 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1834 return *s == '(' ? FUNCMETH : METHOD;
1842 * Return a string of Perl code to load the debugger. If PERL5DB
1843 * is set, it will return the contents of that, otherwise a
1844 * compile-time require of perl5db.pl.
1851 char *pdb = PerlEnv_getenv("PERL5DB");
1855 SETERRNO(0,SS$_NORMAL);
1856 return "BEGIN { require 'perl5db.pl' }";
1862 /* Encoded script support. filter_add() effectively inserts a
1863 * 'pre-processing' function into the current source input stream.
1864 * Note that the filter function only applies to the current source file
1865 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1867 * The datasv parameter (which may be NULL) can be used to pass
1868 * private data to this instance of the filter. The filter function
1869 * can recover the SV using the FILTER_DATA macro and use it to
1870 * store private buffers and state information.
1872 * The supplied datasv parameter is upgraded to a PVIO type
1873 * and the IoDIRP/IoANY field is used to store the function pointer,
1874 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1875 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1876 * private use must be set using malloc'd pointers.
1880 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1885 if (!PL_rsfp_filters)
1886 PL_rsfp_filters = newAV();
1888 datasv = NEWSV(255,0);
1889 if (!SvUPGRADE(datasv, SVt_PVIO))
1890 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1891 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1892 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1893 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1894 funcp, SvPV_nolen(datasv)));
1895 av_unshift(PL_rsfp_filters, 1);
1896 av_store(PL_rsfp_filters, 0, datasv) ;
1901 /* Delete most recently added instance of this filter function. */
1903 Perl_filter_del(pTHX_ filter_t funcp)
1906 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1907 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1909 /* if filter is on top of stack (usual case) just pop it off */
1910 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1911 if (IoANY(datasv) == (void *)funcp) {
1912 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1913 IoANY(datasv) = (void *)NULL;
1914 sv_free(av_pop(PL_rsfp_filters));
1918 /* we need to search for the correct entry and clear it */
1919 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1923 /* Invoke the n'th filter function for the current rsfp. */
1925 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1928 /* 0 = read one text line */
1933 if (!PL_rsfp_filters)
1935 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1936 /* Provide a default input filter to make life easy. */
1937 /* Note that we append to the line. This is handy. */
1938 DEBUG_P(PerlIO_printf(Perl_debug_log,
1939 "filter_read %d: from rsfp\n", idx));
1943 int old_len = SvCUR(buf_sv) ;
1945 /* ensure buf_sv is large enough */
1946 SvGROW(buf_sv, old_len + maxlen) ;
1947 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1948 if (PerlIO_error(PL_rsfp))
1949 return -1; /* error */
1951 return 0 ; /* end of file */
1953 SvCUR_set(buf_sv, old_len + len) ;
1956 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1957 if (PerlIO_error(PL_rsfp))
1958 return -1; /* error */
1960 return 0 ; /* end of file */
1963 return SvCUR(buf_sv);
1965 /* Skip this filter slot if filter has been deleted */
1966 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1967 DEBUG_P(PerlIO_printf(Perl_debug_log,
1968 "filter_read %d: skipped (filter deleted)\n",
1970 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1972 /* Get function pointer hidden within datasv */
1973 funcp = (filter_t)IoANY(datasv);
1974 DEBUG_P(PerlIO_printf(Perl_debug_log,
1975 "filter_read %d: via function %p (%s)\n",
1976 idx, funcp, SvPV_nolen(datasv)));
1977 /* Call function. The function is expected to */
1978 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1979 /* Return: <0:error, =0:eof, >0:not eof */
1980 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1984 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
1986 #ifdef PERL_CR_FILTER
1987 if (!PL_rsfp_filters) {
1988 filter_add(S_cr_textfilter,NULL);
1991 if (PL_rsfp_filters) {
1994 SvCUR_set(sv, 0); /* start with empty line */
1995 if (FILTER_READ(0, sv, 0) > 0)
1996 return ( SvPVX(sv) ) ;
2001 return (sv_gets(sv, fp, append));
2005 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2009 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2013 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2014 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2016 return GvHV(gv); /* Foo:: */
2019 /* use constant CLASS => 'MyClass' */
2020 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2022 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2023 pkgname = SvPV_nolen(sv);
2027 return gv_stashpv(pkgname, FALSE);
2031 static char* exp_name[] =
2032 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2033 "ATTRTERM", "TERMBLOCK"
2040 Works out what to call the token just pulled out of the input
2041 stream. The yacc parser takes care of taking the ops we return and
2042 stitching them into a tree.
2048 if read an identifier
2049 if we're in a my declaration
2050 croak if they tried to say my($foo::bar)
2051 build the ops for a my() declaration
2052 if it's an access to a my() variable
2053 are we in a sort block?
2054 croak if my($a); $a <=> $b
2055 build ops for access to a my() variable
2056 if in a dq string, and they've said @foo and we can't find @foo
2058 build ops for a bareword
2059 if we already built the token before, use it.
2062 #ifdef USE_PURE_BISON
2064 #pragma segment Perl_yylex_r
2067 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2071 yylval_pointer[yyactlevel] = lvalp;
2072 yychar_pointer[yyactlevel] = lcharp;
2074 if (yyactlevel >= YYMAXLEVEL)
2075 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2077 r = Perl_yylex(aTHX);
2086 #pragma segment Perl_yylex
2090 #ifdef USE_PURE_BISON
2091 Perl_yylex(pTHX_ YYSTYPE *lvalp, int *lcharp)
2103 /* check if there's an identifier for us to look at */
2104 if (PL_pending_ident) {
2105 /* pit holds the identifier we read and pending_ident is reset */
2106 char pit = PL_pending_ident;
2107 PL_pending_ident = 0;
2109 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2110 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2112 /* if we're in a my(), we can't allow dynamics here.
2113 $foo'bar has already been turned into $foo::bar, so
2114 just check for colons.
2116 if it's a legal name, the OP is a PADANY.
2119 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2120 if (strchr(PL_tokenbuf,':'))
2121 yyerror(Perl_form(aTHX_ "No package name allowed for "
2122 "variable %s in \"our\"",
2124 tmp = pad_allocmy(PL_tokenbuf);
2127 if (strchr(PL_tokenbuf,':'))
2128 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2130 yylval.opval = newOP(OP_PADANY, 0);
2131 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2137 build the ops for accesses to a my() variable.
2139 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2140 then used in a comparison. This catches most, but not
2141 all cases. For instance, it catches
2142 sort { my($a); $a <=> $b }
2144 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2145 (although why you'd do that is anyone's guess).
2148 if (!strchr(PL_tokenbuf,':')) {
2150 /* Check for single character per-thread SVs */
2151 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2152 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2153 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2155 yylval.opval = newOP(OP_THREADSV, 0);
2156 yylval.opval->op_targ = tmp;
2159 #endif /* USE_THREADS */
2160 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2161 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2162 /* might be an "our" variable" */
2163 if (SvFLAGS(namesv) & SVpad_OUR) {
2164 /* build ops for a bareword */
2165 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2166 sv_catpvn(sym, "::", 2);
2167 sv_catpv(sym, PL_tokenbuf+1);
2168 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2169 yylval.opval->op_private = OPpCONST_ENTERED;
2170 gv_fetchpv(SvPVX(sym),
2172 ? (GV_ADDMULTI | GV_ADDINEVAL)
2175 ((PL_tokenbuf[0] == '$') ? SVt_PV
2176 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2181 /* if it's a sort block and they're naming $a or $b */
2182 if (PL_last_lop_op == OP_SORT &&
2183 PL_tokenbuf[0] == '$' &&
2184 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2187 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2188 d < PL_bufend && *d != '\n';
2191 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2192 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2198 yylval.opval = newOP(OP_PADANY, 0);
2199 yylval.opval->op_targ = tmp;
2205 Whine if they've said @foo in a doublequoted string,
2206 and @foo isn't a variable we can find in the symbol
2209 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2210 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2211 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2212 && ckWARN(WARN_AMBIGUOUS))
2214 /* Downgraded from fatal to warning 20000522 mjd */
2215 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2216 "Possible unintended interpolation of %s in string",
2221 /* build ops for a bareword */
2222 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2223 yylval.opval->op_private = OPpCONST_ENTERED;
2224 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2225 ((PL_tokenbuf[0] == '$') ? SVt_PV
2226 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2231 /* no identifier pending identification */
2233 switch (PL_lex_state) {
2235 case LEX_NORMAL: /* Some compilers will produce faster */
2236 case LEX_INTERPNORMAL: /* code if we comment these out. */
2240 /* when we've already built the next token, just pull it out of the queue */
2243 yylval = PL_nextval[PL_nexttoke];
2245 PL_lex_state = PL_lex_defer;
2246 PL_expect = PL_lex_expect;
2247 PL_lex_defer = LEX_NORMAL;
2249 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2250 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2251 (IV)PL_nexttype[PL_nexttoke]); })
2253 return(PL_nexttype[PL_nexttoke]);
2255 /* interpolated case modifiers like \L \U, including \Q and \E.
2256 when we get here, PL_bufptr is at the \
2258 case LEX_INTERPCASEMOD:
2260 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2261 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2263 /* handle \E or end of string */
2264 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2268 if (PL_lex_casemods) {
2269 oldmod = PL_lex_casestack[--PL_lex_casemods];
2270 PL_lex_casestack[PL_lex_casemods] = '\0';
2272 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2274 PL_lex_state = LEX_INTERPCONCAT;
2278 if (PL_bufptr != PL_bufend)
2280 PL_lex_state = LEX_INTERPCONCAT;
2284 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2285 "### Saw case modifier at '%s'\n", PL_bufptr); })
2287 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2288 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2289 if (strchr("LU", *s) &&
2290 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2292 PL_lex_casestack[--PL_lex_casemods] = '\0';
2295 if (PL_lex_casemods > 10) {
2296 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2297 if (newlb != PL_lex_casestack) {
2299 PL_lex_casestack = newlb;
2302 PL_lex_casestack[PL_lex_casemods++] = *s;
2303 PL_lex_casestack[PL_lex_casemods] = '\0';
2304 PL_lex_state = LEX_INTERPCONCAT;
2305 PL_nextval[PL_nexttoke].ival = 0;
2308 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2310 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2312 PL_nextval[PL_nexttoke].ival = OP_LC;
2314 PL_nextval[PL_nexttoke].ival = OP_UC;
2316 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2318 Perl_croak(aTHX_ "panic: yylex");
2321 if (PL_lex_starts) {
2330 case LEX_INTERPPUSH:
2331 return sublex_push();
2333 case LEX_INTERPSTART:
2334 if (PL_bufptr == PL_bufend)
2335 return sublex_done();
2336 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2337 "### Interpolated variable at '%s'\n", PL_bufptr); })
2339 PL_lex_dojoin = (*PL_bufptr == '@');
2340 PL_lex_state = LEX_INTERPNORMAL;
2341 if (PL_lex_dojoin) {
2342 PL_nextval[PL_nexttoke].ival = 0;
2345 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2346 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2347 force_next(PRIVATEREF);
2349 force_ident("\"", '$');
2350 #endif /* USE_THREADS */
2351 PL_nextval[PL_nexttoke].ival = 0;
2353 PL_nextval[PL_nexttoke].ival = 0;
2355 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2358 if (PL_lex_starts++) {
2364 case LEX_INTERPENDMAYBE:
2365 if (intuit_more(PL_bufptr)) {
2366 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2372 if (PL_lex_dojoin) {
2373 PL_lex_dojoin = FALSE;
2374 PL_lex_state = LEX_INTERPCONCAT;
2377 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2378 && SvEVALED(PL_lex_repl))
2380 if (PL_bufptr != PL_bufend)
2381 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2382 PL_lex_repl = Nullsv;
2385 case LEX_INTERPCONCAT:
2387 if (PL_lex_brackets)
2388 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2390 if (PL_bufptr == PL_bufend)
2391 return sublex_done();
2393 if (SvIVX(PL_linestr) == '\'') {
2394 SV *sv = newSVsv(PL_linestr);
2397 else if ( PL_hints & HINT_NEW_RE )
2398 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2399 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2403 s = scan_const(PL_bufptr);
2405 PL_lex_state = LEX_INTERPCASEMOD;
2407 PL_lex_state = LEX_INTERPSTART;
2410 if (s != PL_bufptr) {
2411 PL_nextval[PL_nexttoke] = yylval;
2414 if (PL_lex_starts++)
2424 PL_lex_state = LEX_NORMAL;
2425 s = scan_formline(PL_bufptr);
2426 if (!PL_lex_formbrack)
2432 PL_oldoldbufptr = PL_oldbufptr;
2435 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2436 exp_name[PL_expect], s);
2442 if (isIDFIRST_lazy_if(s,UTF))
2444 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2447 goto fake_eof; /* emulate EOF on ^D or ^Z */
2452 if (PL_lex_brackets)
2453 yyerror("Missing right curly or square bracket");
2454 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2455 "### Tokener got EOF\n");
2459 if (s++ < PL_bufend)
2460 goto retry; /* ignore stray nulls */
2463 if (!PL_in_eval && !PL_preambled) {
2464 PL_preambled = TRUE;
2465 sv_setpv(PL_linestr,incl_perldb());
2466 if (SvCUR(PL_linestr))
2467 sv_catpv(PL_linestr,";");
2469 while(AvFILLp(PL_preambleav) >= 0) {
2470 SV *tmpsv = av_shift(PL_preambleav);
2471 sv_catsv(PL_linestr, tmpsv);
2472 sv_catpv(PL_linestr, ";");
2475 sv_free((SV*)PL_preambleav);
2476 PL_preambleav = NULL;
2478 if (PL_minus_n || PL_minus_p) {
2479 sv_catpv(PL_linestr, "LINE: while (<>) {");
2481 sv_catpv(PL_linestr,"chomp;");
2483 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2485 GvIMPORTED_AV_on(gv);
2487 if (strchr("/'\"", *PL_splitstr)
2488 && strchr(PL_splitstr + 1, *PL_splitstr))
2489 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2492 s = "'~#\200\1'"; /* surely one char is unused...*/
2493 while (s[1] && strchr(PL_splitstr, *s)) s++;
2495 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2496 "q" + (delim == '\''), delim);
2497 for (s = PL_splitstr; *s; s++) {
2499 sv_catpvn(PL_linestr, "\\", 1);
2500 sv_catpvn(PL_linestr, s, 1);
2502 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2506 sv_catpv(PL_linestr,"@F=split(' ');");
2509 sv_catpv(PL_linestr, "\n");
2510 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2511 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2512 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2513 SV *sv = NEWSV(85,0);
2515 sv_upgrade(sv, SVt_PVMG);
2516 sv_setsv(sv,PL_linestr);
2517 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2522 bool bof = PL_rsfp ? TRUE : FALSE;
2524 #ifdef PERLIO_IS_STDIO
2525 # ifdef __GNU_LIBRARY__
2526 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2527 # define FTELL_FOR_PIPE_IS_BROKEN
2531 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2532 # define FTELL_FOR_PIPE_IS_BROKEN
2537 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2538 /* This loses the possibility to detect the bof
2539 * situation on perl -P when the libc5 is being used.
2540 * Workaround? Maybe attach some extra state to PL_rsfp?
2543 bof = PerlIO_tell(PL_rsfp) == 0;
2545 bof = PerlIO_tell(PL_rsfp) == 0;
2548 s = filter_gets(PL_linestr, PL_rsfp, 0);
2552 if (PL_preprocess && !PL_in_eval)
2553 (void)PerlProc_pclose(PL_rsfp);
2554 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2555 PerlIO_clearerr(PL_rsfp);
2557 (void)PerlIO_close(PL_rsfp);
2559 PL_doextract = FALSE;
2561 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2562 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2563 sv_catpv(PL_linestr,";}");
2564 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2565 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2566 PL_minus_n = PL_minus_p = 0;
2569 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2570 sv_setpv(PL_linestr,"");
2571 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2573 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2574 s = swallow_bom((U8*)s);
2577 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2578 PL_doextract = FALSE;
2580 /* Incest with pod. */
2581 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2582 sv_setpv(PL_linestr, "");
2583 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2584 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2585 PL_doextract = FALSE;
2589 } while (PL_doextract);
2590 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2591 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2592 SV *sv = NEWSV(85,0);
2594 sv_upgrade(sv, SVt_PVMG);
2595 sv_setsv(sv,PL_linestr);
2596 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2598 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2599 if (CopLINE(PL_curcop) == 1) {
2600 while (s < PL_bufend && isSPACE(*s))
2602 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2606 if (*s == '#' && *(s+1) == '!')
2608 #ifdef ALTERNATE_SHEBANG
2610 static char as[] = ALTERNATE_SHEBANG;
2611 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2612 d = s + (sizeof(as) - 1);
2614 #endif /* ALTERNATE_SHEBANG */
2623 while (*d && !isSPACE(*d))
2627 #ifdef ARG_ZERO_IS_SCRIPT
2628 if (ipathend > ipath) {
2630 * HP-UX (at least) sets argv[0] to the script name,
2631 * which makes $^X incorrect. And Digital UNIX and Linux,
2632 * at least, set argv[0] to the basename of the Perl
2633 * interpreter. So, having found "#!", we'll set it right.
2635 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2636 assert(SvPOK(x) || SvGMAGICAL(x));
2637 if (sv_eq(x, CopFILESV(PL_curcop))) {
2638 sv_setpvn(x, ipath, ipathend - ipath);
2641 TAINT_NOT; /* $^X is always tainted, but that's OK */
2643 #endif /* ARG_ZERO_IS_SCRIPT */
2648 d = instr(s,"perl -");
2650 d = instr(s,"perl");
2652 /* avoid getting into infinite loops when shebang
2653 * line contains "Perl" rather than "perl" */
2655 for (d = ipathend-4; d >= ipath; --d) {
2656 if ((*d == 'p' || *d == 'P')
2657 && !ibcmp(d, "perl", 4))
2667 #ifdef ALTERNATE_SHEBANG
2669 * If the ALTERNATE_SHEBANG on this system starts with a
2670 * character that can be part of a Perl expression, then if
2671 * we see it but not "perl", we're probably looking at the
2672 * start of Perl code, not a request to hand off to some
2673 * other interpreter. Similarly, if "perl" is there, but
2674 * not in the first 'word' of the line, we assume the line
2675 * contains the start of the Perl program.
2677 if (d && *s != '#') {
2679 while (*c && !strchr("; \t\r\n\f\v#", *c))
2682 d = Nullch; /* "perl" not in first word; ignore */
2684 *s = '#'; /* Don't try to parse shebang line */
2686 #endif /* ALTERNATE_SHEBANG */
2687 #ifndef MACOS_TRADITIONAL
2692 !instr(s,"indir") &&
2693 instr(PL_origargv[0],"perl"))
2699 while (s < PL_bufend && isSPACE(*s))
2701 if (s < PL_bufend) {
2702 Newz(899,newargv,PL_origargc+3,char*);
2704 while (s < PL_bufend && !isSPACE(*s))
2707 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2710 newargv = PL_origargv;
2712 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2713 Perl_croak(aTHX_ "Can't exec %s", ipath);
2717 U32 oldpdb = PL_perldb;
2718 bool oldn = PL_minus_n;
2719 bool oldp = PL_minus_p;
2721 while (*d && !isSPACE(*d)) d++;
2722 while (SPACE_OR_TAB(*d)) d++;
2726 if (*d == 'M' || *d == 'm') {
2728 while (*d && !isSPACE(*d)) d++;
2729 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2732 d = moreswitches(d);
2734 if ((PERLDB_LINE && !oldpdb) ||
2735 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2736 /* if we have already added "LINE: while (<>) {",
2737 we must not do it again */
2739 sv_setpv(PL_linestr, "");
2740 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2741 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2742 PL_preambled = FALSE;
2744 (void)gv_fetchfile(PL_origfilename);
2751 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2753 PL_lex_state = LEX_FORMLINE;
2758 #ifdef PERL_STRICT_CR
2759 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2761 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2763 case ' ': case '\t': case '\f': case 013:
2764 #ifdef MACOS_TRADITIONAL
2771 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2772 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2773 /* handle eval qq[#line 1 "foo"\n ...] */
2774 CopLINE_dec(PL_curcop);
2778 while (s < d && *s != '\n')
2783 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2785 PL_lex_state = LEX_FORMLINE;
2795 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2802 while (s < PL_bufend && SPACE_OR_TAB(*s))
2805 if (strnEQ(s,"=>",2)) {
2806 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2807 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2808 "### Saw unary minus before =>, forcing word '%s'\n", s);
2810 OPERATOR('-'); /* unary minus */
2812 PL_last_uni = PL_oldbufptr;
2814 case 'r': ftst = OP_FTEREAD; break;
2815 case 'w': ftst = OP_FTEWRITE; break;
2816 case 'x': ftst = OP_FTEEXEC; break;
2817 case 'o': ftst = OP_FTEOWNED; break;
2818 case 'R': ftst = OP_FTRREAD; break;
2819 case 'W': ftst = OP_FTRWRITE; break;
2820 case 'X': ftst = OP_FTREXEC; break;
2821 case 'O': ftst = OP_FTROWNED; break;
2822 case 'e': ftst = OP_FTIS; break;
2823 case 'z': ftst = OP_FTZERO; break;
2824 case 's': ftst = OP_FTSIZE; break;
2825 case 'f': ftst = OP_FTFILE; break;
2826 case 'd': ftst = OP_FTDIR; break;
2827 case 'l': ftst = OP_FTLINK; break;
2828 case 'p': ftst = OP_FTPIPE; break;
2829 case 'S': ftst = OP_FTSOCK; break;
2830 case 'u': ftst = OP_FTSUID; break;
2831 case 'g': ftst = OP_FTSGID; break;
2832 case 'k': ftst = OP_FTSVTX; break;
2833 case 'b': ftst = OP_FTBLK; break;
2834 case 'c': ftst = OP_FTCHR; break;
2835 case 't': ftst = OP_FTTTY; break;
2836 case 'T': ftst = OP_FTTEXT; break;
2837 case 'B': ftst = OP_FTBINARY; break;
2838 case 'M': case 'A': case 'C':
2839 gv_fetchpv("\024",TRUE, SVt_PV);
2841 case 'M': ftst = OP_FTMTIME; break;
2842 case 'A': ftst = OP_FTATIME; break;
2843 case 'C': ftst = OP_FTCTIME; break;
2851 PL_last_lop_op = ftst;
2852 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2853 "### Saw file test %c\n", ftst);
2858 /* Assume it was a minus followed by a one-letter named
2859 * subroutine call (or a -bareword), then. */
2860 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2861 "### %c looked like a file test but was not\n", ftst);
2869 if (PL_expect == XOPERATOR)
2874 else if (*s == '>') {
2877 if (isIDFIRST_lazy_if(s,UTF)) {
2878 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2886 if (PL_expect == XOPERATOR)
2889 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2891 OPERATOR('-'); /* unary minus */
2898 if (PL_expect == XOPERATOR)
2903 if (PL_expect == XOPERATOR)
2906 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2912 if (PL_expect != XOPERATOR) {
2913 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2914 PL_expect = XOPERATOR;
2915 force_ident(PL_tokenbuf, '*');
2928 if (PL_expect == XOPERATOR) {
2932 PL_tokenbuf[0] = '%';
2933 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2934 if (!PL_tokenbuf[1]) {
2936 yyerror("Final % should be \\% or %name");
2939 PL_pending_ident = '%';
2958 switch (PL_expect) {
2961 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2963 PL_bufptr = s; /* update in case we back off */
2969 PL_expect = XTERMBLOCK;
2973 while (isIDFIRST_lazy_if(s,UTF)) {
2974 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2975 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2976 if (tmp < 0) tmp = -tmp;
2991 d = scan_str(d,TRUE,TRUE);
2994 SvREFCNT_dec(PL_lex_stuff);
2995 PL_lex_stuff = Nullsv;
2997 /* MUST advance bufptr here to avoid bogus
2998 "at end of line" context messages from yyerror().
3000 PL_bufptr = s + len;
3001 yyerror("Unterminated attribute parameter in attribute list");
3004 return 0; /* EOF indicator */
3008 SV *sv = newSVpvn(s, len);
3009 sv_catsv(sv, PL_lex_stuff);
3010 attrs = append_elem(OP_LIST, attrs,
3011 newSVOP(OP_CONST, 0, sv));
3012 SvREFCNT_dec(PL_lex_stuff);
3013 PL_lex_stuff = Nullsv;
3016 attrs = append_elem(OP_LIST, attrs,
3017 newSVOP(OP_CONST, 0,
3021 if (*s == ':' && s[1] != ':')
3024 break; /* require real whitespace or :'s */
3026 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3027 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3028 char q = ((*s == '\'') ? '"' : '\'');
3029 /* If here for an expression, and parsed no attrs, back off. */
3030 if (tmp == '=' && !attrs) {
3034 /* MUST advance bufptr here to avoid bogus "at end of line"
3035 context messages from yyerror().
3039 yyerror("Unterminated attribute list");
3041 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3049 PL_nextval[PL_nexttoke].opval = attrs;
3057 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3058 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3074 if (PL_lex_brackets <= 0)
3075 yyerror("Unmatched right square bracket");
3078 if (PL_lex_state == LEX_INTERPNORMAL) {
3079 if (PL_lex_brackets == 0) {
3080 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3081 PL_lex_state = LEX_INTERPEND;
3088 if (PL_lex_brackets > 100) {
3089 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3090 if (newlb != PL_lex_brackstack) {
3092 PL_lex_brackstack = newlb;
3095 switch (PL_expect) {
3097 if (PL_lex_formbrack) {
3101 if (PL_oldoldbufptr == PL_last_lop)
3102 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3104 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3105 OPERATOR(HASHBRACK);
3107 while (s < PL_bufend && SPACE_OR_TAB(*s))
3110 PL_tokenbuf[0] = '\0';
3111 if (d < PL_bufend && *d == '-') {
3112 PL_tokenbuf[0] = '-';
3114 while (d < PL_bufend && SPACE_OR_TAB(*d))
3117 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3118 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3120 while (d < PL_bufend && SPACE_OR_TAB(*d))
3123 char minus = (PL_tokenbuf[0] == '-');
3124 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3125 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, 0) &&
3126 PL_nextval[PL_nexttoke-1].opval)
3127 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke-1].opval)->op_sv);
3135 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3140 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3145 if (PL_oldoldbufptr == PL_last_lop)
3146 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3148 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3151 OPERATOR(HASHBRACK);
3152 /* This hack serves to disambiguate a pair of curlies
3153 * as being a block or an anon hash. Normally, expectation
3154 * determines that, but in cases where we're not in a
3155 * position to expect anything in particular (like inside
3156 * eval"") we have to resolve the ambiguity. This code
3157 * covers the case where the first term in the curlies is a
3158 * quoted string. Most other cases need to be explicitly
3159 * disambiguated by prepending a `+' before the opening
3160 * curly in order to force resolution as an anon hash.
3162 * XXX should probably propagate the outer expectation
3163 * into eval"" to rely less on this hack, but that could
3164 * potentially break current behavior of eval"".
3168 if (*s == '\'' || *s == '"' || *s == '`') {
3169 /* common case: get past first string, handling escapes */
3170 for (t++; t < PL_bufend && *t != *s;)
3171 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3175 else if (*s == 'q') {
3178 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3182 char open, close, term;
3185 while (t < PL_bufend && isSPACE(*t))
3189 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3193 for (t++; t < PL_bufend; t++) {
3194 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3196 else if (*t == open)
3200 for (t++; t < PL_bufend; t++) {
3201 if (*t == '\\' && t+1 < PL_bufend)
3203 else if (*t == close && --brackets <= 0)
3205 else if (*t == open)
3211 else if (isALNUM_lazy_if(t,UTF)) {
3213 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3216 while (t < PL_bufend && isSPACE(*t))
3218 /* if comma follows first term, call it an anon hash */
3219 /* XXX it could be a comma expression with loop modifiers */
3220 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3221 || (*t == '=' && t[1] == '>')))
3222 OPERATOR(HASHBRACK);
3223 if (PL_expect == XREF)
3226 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3232 yylval.ival = CopLINE(PL_curcop);
3233 if (isSPACE(*s) || *s == '#')
3234 PL_copline = NOLINE; /* invalidate current command line number */
3239 if (PL_lex_brackets <= 0)
3240 yyerror("Unmatched right curly bracket");
3242 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3243 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3244 PL_lex_formbrack = 0;
3245 if (PL_lex_state == LEX_INTERPNORMAL) {
3246 if (PL_lex_brackets == 0) {
3247 if (PL_expect & XFAKEBRACK) {
3248 PL_expect &= XENUMMASK;
3249 PL_lex_state = LEX_INTERPEND;
3251 return yylex(); /* ignore fake brackets */
3253 if (*s == '-' && s[1] == '>')
3254 PL_lex_state = LEX_INTERPENDMAYBE;
3255 else if (*s != '[' && *s != '{')
3256 PL_lex_state = LEX_INTERPEND;
3259 if (PL_expect & XFAKEBRACK) {
3260 PL_expect &= XENUMMASK;
3262 return yylex(); /* ignore fake brackets */
3272 if (PL_expect == XOPERATOR) {
3273 if (ckWARN(WARN_SEMICOLON)
3274 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3276 CopLINE_dec(PL_curcop);
3277 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3278 CopLINE_inc(PL_curcop);
3283 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3285 PL_expect = XOPERATOR;
3286 force_ident(PL_tokenbuf, '&');
3290 yylval.ival = (OPpENTERSUB_AMPER<<8);
3309 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3310 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3312 if (PL_expect == XSTATE && isALPHA(tmp) &&
3313 (s == PL_linestart+1 || s[-2] == '\n') )
3315 if (PL_in_eval && !PL_rsfp) {
3320 if (strnEQ(s,"=cut",4)) {
3334 PL_doextract = TRUE;
3337 if (PL_lex_brackets < PL_lex_formbrack) {
3339 #ifdef PERL_STRICT_CR
3340 for (t = s; SPACE_OR_TAB(*t); t++) ;
3342 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3344 if (*t == '\n' || *t == '#') {
3362 if (PL_expect != XOPERATOR) {
3363 if (s[1] != '<' && !strchr(s,'>'))
3366 s = scan_heredoc(s);
3368 s = scan_inputsymbol(s);
3369 TERM(sublex_start());
3374 SHop(OP_LEFT_SHIFT);
3388 SHop(OP_RIGHT_SHIFT);
3397 if (PL_expect == XOPERATOR) {
3398 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3401 return ','; /* grandfather non-comma-format format */
3405 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3406 PL_tokenbuf[0] = '@';
3407 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3408 sizeof PL_tokenbuf - 1, FALSE);
3409 if (PL_expect == XOPERATOR)
3410 no_op("Array length", s);
3411 if (!PL_tokenbuf[1])
3413 PL_expect = XOPERATOR;
3414 PL_pending_ident = '#';
3418 PL_tokenbuf[0] = '$';
3419 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3420 sizeof PL_tokenbuf - 1, FALSE);
3421 if (PL_expect == XOPERATOR)
3423 if (!PL_tokenbuf[1]) {
3425 yyerror("Final $ should be \\$ or $name");
3429 /* This kludge not intended to be bulletproof. */
3430 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3431 yylval.opval = newSVOP(OP_CONST, 0,
3432 newSViv(PL_compiling.cop_arybase));
3433 yylval.opval->op_private = OPpCONST_ARYBASE;
3439 if (PL_lex_state == LEX_NORMAL)
3442 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3445 PL_tokenbuf[0] = '@';
3446 if (ckWARN(WARN_SYNTAX)) {
3448 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3451 PL_bufptr = skipspace(PL_bufptr);
3452 while (t < PL_bufend && *t != ']')
3454 Perl_warner(aTHX_ WARN_SYNTAX,
3455 "Multidimensional syntax %.*s not supported",
3456 (t - PL_bufptr) + 1, PL_bufptr);
3460 else if (*s == '{') {
3461 PL_tokenbuf[0] = '%';
3462 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3463 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3465 char tmpbuf[sizeof PL_tokenbuf];
3467 for (t++; isSPACE(*t); t++) ;
3468 if (isIDFIRST_lazy_if(t,UTF)) {
3469 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3470 for (; isSPACE(*t); t++) ;
3471 if (*t == ';' && get_cv(tmpbuf, FALSE))
3472 Perl_warner(aTHX_ WARN_SYNTAX,
3473 "You need to quote \"%s\"", tmpbuf);
3479 PL_expect = XOPERATOR;
3480 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3481 bool islop = (PL_last_lop == PL_oldoldbufptr);
3482 if (!islop || PL_last_lop_op == OP_GREPSTART)
3483 PL_expect = XOPERATOR;
3484 else if (strchr("$@\"'`q", *s))
3485 PL_expect = XTERM; /* e.g. print $fh "foo" */
3486 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3487 PL_expect = XTERM; /* e.g. print $fh &sub */
3488 else if (isIDFIRST_lazy_if(s,UTF)) {
3489 char tmpbuf[sizeof PL_tokenbuf];
3490 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3491 if ((tmp = keyword(tmpbuf, len))) {
3492 /* binary operators exclude handle interpretations */
3504 PL_expect = XTERM; /* e.g. print $fh length() */
3509 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3510 if (gv && GvCVu(gv))
3511 PL_expect = XTERM; /* e.g. print $fh subr() */
3514 else if (isDIGIT(*s))
3515 PL_expect = XTERM; /* e.g. print $fh 3 */
3516 else if (*s == '.' && isDIGIT(s[1]))
3517 PL_expect = XTERM; /* e.g. print $fh .3 */
3518 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3519 PL_expect = XTERM; /* e.g. print $fh -1 */
3520 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3521 PL_expect = XTERM; /* print $fh <<"EOF" */
3523 PL_pending_ident = '$';
3527 if (PL_expect == XOPERATOR)
3529 PL_tokenbuf[0] = '@';
3530 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3531 if (!PL_tokenbuf[1]) {
3533 yyerror("Final @ should be \\@ or @name");
3536 if (PL_lex_state == LEX_NORMAL)
3538 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3540 PL_tokenbuf[0] = '%';
3542 /* Warn about @ where they meant $. */
3543 if (ckWARN(WARN_SYNTAX)) {
3544 if (*s == '[' || *s == '{') {
3546 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3548 if (*t == '}' || *t == ']') {
3550 PL_bufptr = skipspace(PL_bufptr);
3551 Perl_warner(aTHX_ WARN_SYNTAX,
3552 "Scalar value %.*s better written as $%.*s",
3553 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3558 PL_pending_ident = '@';
3561 case '/': /* may either be division or pattern */
3562 case '?': /* may either be conditional or pattern */
3563 if (PL_expect != XOPERATOR) {
3564 /* Disable warning on "study /blah/" */
3565 if (PL_oldoldbufptr == PL_last_uni
3566 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3567 || memNE(PL_last_uni, "study", 5)
3568 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3570 s = scan_pat(s,OP_MATCH);
3571 TERM(sublex_start());
3579 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3580 #ifdef PERL_STRICT_CR
3583 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3585 && (s == PL_linestart || s[-1] == '\n') )
3587 PL_lex_formbrack = 0;
3591 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3597 yylval.ival = OPf_SPECIAL;
3603 if (PL_expect != XOPERATOR)
3608 case '0': case '1': case '2': case '3': case '4':
3609 case '5': case '6': case '7': case '8': case '9':
3610 s = scan_num(s, &yylval);
3611 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3612 "### Saw number in '%s'\n", s);
3614 if (PL_expect == XOPERATOR)
3619 s = scan_str(s,FALSE,FALSE);
3620 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3621 "### Saw string in '%s'\n", s);
3623 if (PL_expect == XOPERATOR) {
3624 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3627 return ','; /* grandfather non-comma-format format */
3633 missingterm((char*)0);
3634 yylval.ival = OP_CONST;
3635 TERM(sublex_start());
3638 s = scan_str(s,FALSE,FALSE);
3639 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3640 "### Saw string in '%s'\n", s);
3642 if (PL_expect == XOPERATOR) {
3643 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3646 return ','; /* grandfather non-comma-format format */
3652 missingterm((char*)0);
3653 yylval.ival = OP_CONST;
3654 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3655 if (*d == '$' || *d == '@' || *d == '\\' || *d & 0x80) {
3656 yylval.ival = OP_STRINGIFY;
3660 TERM(sublex_start());
3663 s = scan_str(s,FALSE,FALSE);
3664 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3665 "### Saw backtick string in '%s'\n", s);
3667 if (PL_expect == XOPERATOR)
3668 no_op("Backticks",s);
3670 missingterm((char*)0);
3671 yylval.ival = OP_BACKTICK;
3673 TERM(sublex_start());
3677 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3678 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3680 if (PL_expect == XOPERATOR)
3681 no_op("Backslash",s);
3685 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3689 while (isDIGIT(*start) || *start == '_')
3691 if (*start == '.' && isDIGIT(start[1])) {
3692 s = scan_num(s, &yylval);
3695 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3696 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3700 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3703 s = scan_num(s, &yylval);
3710 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3749 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3751 /* Some keywords can be followed by any delimiter, including ':' */
3752 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3753 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3754 (PL_tokenbuf[0] == 'q' &&
3755 strchr("qwxr", PL_tokenbuf[1])))));
3757 /* x::* is just a word, unless x is "CORE" */
3758 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3762 while (d < PL_bufend && isSPACE(*d))
3763 d++; /* no comments skipped here, or s### is misparsed */
3765 /* Is this a label? */
3766 if (!tmp && PL_expect == XSTATE
3767 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3769 yylval.pval = savepv(PL_tokenbuf);
3774 /* Check for keywords */
3775 tmp = keyword(PL_tokenbuf, len);
3777 /* Is this a word before a => operator? */
3778 if (*d == '=' && d[1] == '>') {
3780 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3781 yylval.opval->op_private = OPpCONST_BARE;
3782 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3783 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3787 if (tmp < 0) { /* second-class keyword? */
3788 GV *ogv = Nullgv; /* override (winner) */
3789 GV *hgv = Nullgv; /* hidden (loser) */
3790 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3792 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3795 if (GvIMPORTED_CV(gv))
3797 else if (! CvMETHOD(cv))
3801 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3802 (gv = *gvp) != (GV*)&PL_sv_undef &&
3803 GvCVu(gv) && GvIMPORTED_CV(gv))
3809 tmp = 0; /* overridden by import or by GLOBAL */
3812 && -tmp==KEY_lock /* XXX generalizable kludge */
3814 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3816 tmp = 0; /* any sub overrides "weak" keyword */
3818 else { /* no override */
3822 if (ckWARN(WARN_AMBIGUOUS) && hgv
3823 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3824 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3825 "Ambiguous call resolved as CORE::%s(), %s",
3826 GvENAME(hgv), "qualify as such or use &");
3833 default: /* not a keyword */
3836 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3838 /* Get the rest if it looks like a package qualifier */
3840 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3842 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3845 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3846 *s == '\'' ? "'" : "::");
3850 if (PL_expect == XOPERATOR) {
3851 if (PL_bufptr == PL_linestart) {
3852 CopLINE_dec(PL_curcop);
3853 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3854 CopLINE_inc(PL_curcop);
3857 no_op("Bareword",s);
3860 /* Look for a subroutine with this name in current package,
3861 unless name is "Foo::", in which case Foo is a bearword
3862 (and a package name). */
3865 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3867 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3868 Perl_warner(aTHX_ WARN_BAREWORD,
3869 "Bareword \"%s\" refers to nonexistent package",
3872 PL_tokenbuf[len] = '\0';
3879 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3882 /* if we saw a global override before, get the right name */
3885 sv = newSVpvn("CORE::GLOBAL::",14);
3886 sv_catpv(sv,PL_tokenbuf);
3889 sv = newSVpv(PL_tokenbuf,0);
3891 /* Presume this is going to be a bareword of some sort. */
3894 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3895 yylval.opval->op_private = OPpCONST_BARE;
3897 /* And if "Foo::", then that's what it certainly is. */
3902 /* See if it's the indirect object for a list operator. */
3904 if (PL_oldoldbufptr &&
3905 PL_oldoldbufptr < PL_bufptr &&
3906 (PL_oldoldbufptr == PL_last_lop
3907 || PL_oldoldbufptr == PL_last_uni) &&
3908 /* NO SKIPSPACE BEFORE HERE! */
3909 (PL_expect == XREF ||
3910 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3912 bool immediate_paren = *s == '(';
3914 /* (Now we can afford to cross potential line boundary.) */
3917 /* Two barewords in a row may indicate method call. */
3919 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3922 /* If not a declared subroutine, it's an indirect object. */
3923 /* (But it's an indir obj regardless for sort.) */
3925 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3926 ((!gv || !GvCVu(gv)) &&
3927 (PL_last_lop_op != OP_MAPSTART &&
3928 PL_last_lop_op != OP_GREPSTART))))
3930 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3936 PL_expect = XOPERATOR;
3939 /* Is this a word before a => operator? */
3940 if (*s == '=' && s[1] == '>') {
3942 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3943 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3944 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3948 /* If followed by a paren, it's certainly a subroutine. */
3951 if (gv && GvCVu(gv)) {
3952 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3953 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3958 PL_nextval[PL_nexttoke].opval = yylval.opval;
3959 PL_expect = XOPERATOR;
3965 /* If followed by var or block, call it a method (unless sub) */
3967 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3968 PL_last_lop = PL_oldbufptr;
3969 PL_last_lop_op = OP_METHOD;
3973 /* If followed by a bareword, see if it looks like indir obj. */
3975 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3978 /* Not a method, so call it a subroutine (if defined) */
3980 if (gv && GvCVu(gv)) {
3982 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3983 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3984 "Ambiguous use of -%s resolved as -&%s()",
3985 PL_tokenbuf, PL_tokenbuf);
3986 /* Check for a constant sub */
3988 if ((sv = cv_const_sv(cv))) {
3990 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
3991 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
3992 yylval.opval->op_private = 0;
3996 /* Resolve to GV now. */
3997 op_free(yylval.opval);
3998 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
3999 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4000 PL_last_lop = PL_oldbufptr;
4001 PL_last_lop_op = OP_ENTERSUB;
4002 /* Is there a prototype? */
4005 char *proto = SvPV((SV*)cv, len);
4008 if (strEQ(proto, "$"))
4010 if (*proto == '&' && *s == '{') {
4011 sv_setpv(PL_subname,"__ANON__");
4015 PL_nextval[PL_nexttoke].opval = yylval.opval;
4021 /* Call it a bare word */
4023 if (PL_hints & HINT_STRICT_SUBS)
4024 yylval.opval->op_private |= OPpCONST_STRICT;
4027 if (ckWARN(WARN_RESERVED)) {
4028 if (lastchar != '-') {
4029 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4031 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4038 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4039 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4040 "Operator or semicolon missing before %c%s",
4041 lastchar, PL_tokenbuf);
4042 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4043 "Ambiguous use of %c resolved as operator %c",
4044 lastchar, lastchar);
4050 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4051 newSVpv(CopFILE(PL_curcop),0));
4055 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4056 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4059 case KEY___PACKAGE__:
4060 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4062 ? newSVsv(PL_curstname)
4071 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4072 char *pname = "main";
4073 if (PL_tokenbuf[2] == 'D')
4074 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4075 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4078 GvIOp(gv) = newIO();
4079 IoIFP(GvIOp(gv)) = PL_rsfp;
4080 #if defined(HAS_FCNTL) && defined(F_SETFD)
4082 int fd = PerlIO_fileno(PL_rsfp);
4083 fcntl(fd,F_SETFD,fd >= 3);
4086 /* Mark this internal pseudo-handle as clean */
4087 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4089 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4090 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4091 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4093 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4094 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4095 /* if the script was opened in binmode, we need to revert
4096 * it to text mode for compatibility; but only iff it has CRs
4097 * XXX this is a questionable hack at best. */
4098 if (PL_bufend-PL_bufptr > 2
4099 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4102 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4103 loc = PerlIO_tell(PL_rsfp);
4104 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4106 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4107 #if defined(__BORLANDC__)
4108 /* XXX see note in do_binmode() */
4109 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4112 PerlIO_seek(PL_rsfp, loc, 0);
4116 #ifdef PERLIO_LAYERS
4117 if (UTF && !IN_BYTE)
4118 PerlIO_apply_layers(PL_rsfp, NULL, ":utf8");
4131 if (PL_expect == XSTATE) {
4138 if (*s == ':' && s[1] == ':') {
4141 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4142 if (!(tmp = keyword(PL_tokenbuf, len)))
4143 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4157 LOP(OP_ACCEPT,XTERM);
4163 LOP(OP_ATAN2,XTERM);
4169 LOP(OP_BINMODE,XTERM);
4172 LOP(OP_BLESS,XTERM);
4181 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4198 if (!PL_cryptseen) {
4199 PL_cryptseen = TRUE;
4203 LOP(OP_CRYPT,XTERM);
4206 if (ckWARN(WARN_CHMOD)) {
4207 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4208 if (*d != '0' && isDIGIT(*d))
4209 Perl_warner(aTHX_ WARN_CHMOD,
4210 "chmod() mode argument is missing initial 0");
4212 LOP(OP_CHMOD,XTERM);
4215 LOP(OP_CHOWN,XTERM);
4218 LOP(OP_CONNECT,XTERM);
4234 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4238 PL_hints |= HINT_BLOCK_SCOPE;
4248 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4249 LOP(OP_DBMOPEN,XTERM);
4255 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4262 yylval.ival = CopLINE(PL_curcop);
4276 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4277 UNIBRACK(OP_ENTEREVAL);
4292 case KEY_endhostent:
4298 case KEY_endservent:
4301 case KEY_endprotoent:
4312 yylval.ival = CopLINE(PL_curcop);
4314 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4316 if ((PL_bufend - p) >= 3 &&
4317 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4319 else if ((PL_bufend - p) >= 4 &&
4320 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4323 if (isIDFIRST_lazy_if(p,UTF)) {
4324 p = scan_ident(p, PL_bufend,
4325 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4329 Perl_croak(aTHX_ "Missing $ on loop variable");
4334 LOP(OP_FORMLINE,XTERM);
4340 LOP(OP_FCNTL,XTERM);
4346 LOP(OP_FLOCK,XTERM);
4355 LOP(OP_GREPSTART, XREF);
4358 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4373 case KEY_getpriority:
4374 LOP(OP_GETPRIORITY,XTERM);
4376 case KEY_getprotobyname:
4379 case KEY_getprotobynumber:
4380 LOP(OP_GPBYNUMBER,XTERM);
4382 case KEY_getprotoent:
4394 case KEY_getpeername:
4395 UNI(OP_GETPEERNAME);
4397 case KEY_gethostbyname:
4400 case KEY_gethostbyaddr:
4401 LOP(OP_GHBYADDR,XTERM);
4403 case KEY_gethostent:
4406 case KEY_getnetbyname:
4409 case KEY_getnetbyaddr:
4410 LOP(OP_GNBYADDR,XTERM);
4415 case KEY_getservbyname:
4416 LOP(OP_GSBYNAME,XTERM);
4418 case KEY_getservbyport:
4419 LOP(OP_GSBYPORT,XTERM);
4421 case KEY_getservent:
4424 case KEY_getsockname:
4425 UNI(OP_GETSOCKNAME);
4427 case KEY_getsockopt:
4428 LOP(OP_GSOCKOPT,XTERM);
4450 yylval.ival = CopLINE(PL_curcop);
4454 LOP(OP_INDEX,XTERM);
4460 LOP(OP_IOCTL,XTERM);
4472 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4504 LOP(OP_LISTEN,XTERM);
4513 s = scan_pat(s,OP_MATCH);
4514 TERM(sublex_start());
4517 LOP(OP_MAPSTART, XREF);
4520 LOP(OP_MKDIR,XTERM);
4523 LOP(OP_MSGCTL,XTERM);
4526 LOP(OP_MSGGET,XTERM);
4529 LOP(OP_MSGRCV,XTERM);
4532 LOP(OP_MSGSND,XTERM);
4538 if (isIDFIRST_lazy_if(s,UTF)) {
4539 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4540 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4542 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4543 if (!PL_in_my_stash) {
4546 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4554 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4561 if (PL_expect != XSTATE)
4562 yyerror("\"no\" not allowed in expression");
4563 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4564 s = force_version(s);
4569 if (*s == '(' || (s = skipspace(s), *s == '('))
4576 if (isIDFIRST_lazy_if(s,UTF)) {
4578 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4580 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4581 Perl_warner(aTHX_ WARN_PRECEDENCE,
4582 "Precedence problem: open %.*s should be open(%.*s)",
4588 yylval.ival = OP_OR;
4598 LOP(OP_OPEN_DIR,XTERM);
4601 checkcomma(s,PL_tokenbuf,"filehandle");
4605 checkcomma(s,PL_tokenbuf,"filehandle");
4624 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4628 LOP(OP_PIPE_OP,XTERM);
4631 s = scan_str(s,FALSE,FALSE);
4633 missingterm((char*)0);
4634 yylval.ival = OP_CONST;
4635 TERM(sublex_start());
4641 s = scan_str(s,FALSE,FALSE);
4643 missingterm((char*)0);
4645 if (SvCUR(PL_lex_stuff)) {
4648 d = SvPV_force(PL_lex_stuff, len);
4651 for (; isSPACE(*d) && len; --len, ++d) ;
4654 if (!warned && ckWARN(WARN_QW)) {
4655 for (; !isSPACE(*d) && len; --len, ++d) {
4657 Perl_warner(aTHX_ WARN_QW,
4658 "Possible attempt to separate words with commas");
4661 else if (*d == '#') {
4662 Perl_warner(aTHX_ WARN_QW,
4663 "Possible attempt to put comments in qw() list");
4669 for (; !isSPACE(*d) && len; --len, ++d) ;
4671 sv = newSVpvn(b, d-b);
4672 if (DO_UTF8(PL_lex_stuff))
4674 words = append_elem(OP_LIST, words,
4675 newSVOP(OP_CONST, 0, tokeq(sv)));
4679 PL_nextval[PL_nexttoke].opval = words;
4684 SvREFCNT_dec(PL_lex_stuff);
4685 PL_lex_stuff = Nullsv;
4690 s = scan_str(s,FALSE,FALSE);
4692 missingterm((char*)0);
4693 yylval.ival = OP_STRINGIFY;
4694 if (SvIVX(PL_lex_stuff) == '\'')
4695 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4696 TERM(sublex_start());
4699 s = scan_pat(s,OP_QR);
4700 TERM(sublex_start());
4703 s = scan_str(s,FALSE,FALSE);
4705 missingterm((char*)0);
4706 yylval.ival = OP_BACKTICK;
4708 TERM(sublex_start());
4715 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4716 s = force_version(s);
4719 *PL_tokenbuf = '\0';
4720 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4721 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4722 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4724 yyerror("<> should be quotes");
4732 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4736 LOP(OP_RENAME,XTERM);
4745 LOP(OP_RINDEX,XTERM);
4768 LOP(OP_REVERSE,XTERM);
4779 TERM(sublex_start());
4781 TOKEN(1); /* force error */
4790 LOP(OP_SELECT,XTERM);
4796 LOP(OP_SEMCTL,XTERM);
4799 LOP(OP_SEMGET,XTERM);
4802 LOP(OP_SEMOP,XTERM);
4808 LOP(OP_SETPGRP,XTERM);
4810 case KEY_setpriority:
4811 LOP(OP_SETPRIORITY,XTERM);
4813 case KEY_sethostent:
4819 case KEY_setservent:
4822 case KEY_setprotoent:
4832 LOP(OP_SEEKDIR,XTERM);
4834 case KEY_setsockopt:
4835 LOP(OP_SSOCKOPT,XTERM);
4841 LOP(OP_SHMCTL,XTERM);
4844 LOP(OP_SHMGET,XTERM);
4847 LOP(OP_SHMREAD,XTERM);
4850 LOP(OP_SHMWRITE,XTERM);
4853 LOP(OP_SHUTDOWN,XTERM);
4862 LOP(OP_SOCKET,XTERM);
4864 case KEY_socketpair:
4865 LOP(OP_SOCKPAIR,XTERM);
4868 checkcomma(s,PL_tokenbuf,"subroutine name");
4870 if (*s == ';' || *s == ')') /* probably a close */
4871 Perl_croak(aTHX_ "sort is now a reserved word");
4873 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4877 LOP(OP_SPLIT,XTERM);
4880 LOP(OP_SPRINTF,XTERM);
4883 LOP(OP_SPLICE,XTERM);
4898 LOP(OP_SUBSTR,XTERM);
4904 char tmpbuf[sizeof PL_tokenbuf];
4906 expectation attrful;
4907 bool have_name, have_proto;
4912 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4913 (*s == ':' && s[1] == ':'))
4916 attrful = XATTRBLOCK;
4917 /* remember buffer pos'n for later force_word */
4918 tboffset = s - PL_oldbufptr;
4919 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4920 if (strchr(tmpbuf, ':'))
4921 sv_setpv(PL_subname, tmpbuf);
4923 sv_setsv(PL_subname,PL_curstname);
4924 sv_catpvn(PL_subname,"::",2);
4925 sv_catpvn(PL_subname,tmpbuf,len);
4932 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4933 PL_expect = XTERMBLOCK;
4934 attrful = XATTRTERM;
4935 sv_setpv(PL_subname,"?");
4939 if (key == KEY_format) {
4941 PL_lex_formbrack = PL_lex_brackets + 1;
4943 (void) force_word(PL_oldbufptr + tboffset, WORD,
4948 /* Look for a prototype */
4952 s = scan_str(s,FALSE,FALSE);
4955 SvREFCNT_dec(PL_lex_stuff);
4956 PL_lex_stuff = Nullsv;
4957 Perl_croak(aTHX_ "Prototype not terminated");
4960 d = SvPVX(PL_lex_stuff);
4962 for (p = d; *p; ++p) {
4967 SvCUR(PL_lex_stuff) = tmp;
4975 if (*s == ':' && s[1] != ':')
4976 PL_expect = attrful;
4979 PL_nextval[PL_nexttoke].opval =
4980 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4981 PL_lex_stuff = Nullsv;
4985 sv_setpv(PL_subname,"__ANON__");
4988 (void) force_word(PL_oldbufptr + tboffset, WORD,
4997 LOP(OP_SYSTEM,XREF);
5000 LOP(OP_SYMLINK,XTERM);
5003 LOP(OP_SYSCALL,XTERM);
5006 LOP(OP_SYSOPEN,XTERM);
5009 LOP(OP_SYSSEEK,XTERM);
5012 LOP(OP_SYSREAD,XTERM);
5015 LOP(OP_SYSWRITE,XTERM);
5019 TERM(sublex_start());
5040 LOP(OP_TRUNCATE,XTERM);
5052 yylval.ival = CopLINE(PL_curcop);
5056 yylval.ival = CopLINE(PL_curcop);
5060 LOP(OP_UNLINK,XTERM);
5066 LOP(OP_UNPACK,XTERM);
5069 LOP(OP_UTIME,XTERM);
5072 if (ckWARN(WARN_UMASK)) {
5073 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5074 if (*d != '0' && isDIGIT(*d))
5075 Perl_warner(aTHX_ WARN_UMASK,
5076 "umask: argument is missing initial 0");
5081 LOP(OP_UNSHIFT,XTERM);
5084 if (PL_expect != XSTATE)
5085 yyerror("\"use\" not allowed in expression");
5087 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5088 s = force_version(s);
5089 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5090 PL_nextval[PL_nexttoke].opval = Nullop;
5095 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5096 s = force_version(s);
5108 yylval.ival = CopLINE(PL_curcop);
5112 PL_hints |= HINT_BLOCK_SCOPE;
5119 LOP(OP_WAITPID,XTERM);
5127 static char ctl_l[2];
5129 if (ctl_l[0] == '\0')
5130 ctl_l[0] = toCTRL('L');
5131 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5134 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5139 if (PL_expect == XOPERATOR)
5145 yylval.ival = OP_XOR;
5150 TERM(sublex_start());
5155 #pragma segment Main
5159 Perl_keyword(pTHX_ register char *d, I32 len)
5164 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5165 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5166 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5167 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5168 if (strEQ(d,"__END__")) return KEY___END__;
5172 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5177 if (strEQ(d,"and")) return -KEY_and;
5178 if (strEQ(d,"abs")) return -KEY_abs;
5181 if (strEQ(d,"alarm")) return -KEY_alarm;
5182 if (strEQ(d,"atan2")) return -KEY_atan2;
5185 if (strEQ(d,"accept")) return -KEY_accept;
5190 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5193 if (strEQ(d,"bless")) return -KEY_bless;
5194 if (strEQ(d,"bind")) return -KEY_bind;
5195 if (strEQ(d,"binmode")) return -KEY_binmode;
5198 if (strEQ(d,"CORE")) return -KEY_CORE;
5199 if (strEQ(d,"CHECK")) return KEY_CHECK;
5204 if (strEQ(d,"cmp")) return -KEY_cmp;
5205 if (strEQ(d,"chr")) return -KEY_chr;
5206 if (strEQ(d,"cos")) return -KEY_cos;
5209 if (strEQ(d,"chop")) return -KEY_chop;
5212 if (strEQ(d,"close")) return -KEY_close;
5213 if (strEQ(d,"chdir")) return -KEY_chdir;
5214 if (strEQ(d,"chomp")) return -KEY_chomp;
5215 if (strEQ(d,"chmod")) return -KEY_chmod;
5216 if (strEQ(d,"chown")) return -KEY_chown;
5217 if (strEQ(d,"crypt")) return -KEY_crypt;
5220 if (strEQ(d,"chroot")) return -KEY_chroot;
5221 if (strEQ(d,"caller")) return -KEY_caller;
5224 if (strEQ(d,"connect")) return -KEY_connect;
5227 if (strEQ(d,"closedir")) return -KEY_closedir;
5228 if (strEQ(d,"continue")) return -KEY_continue;
5233 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5238 if (strEQ(d,"do")) return KEY_do;
5241 if (strEQ(d,"die")) return -KEY_die;
5244 if (strEQ(d,"dump")) return -KEY_dump;
5247 if (strEQ(d,"delete")) return KEY_delete;
5250 if (strEQ(d,"defined")) return KEY_defined;
5251 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5254 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5259 if (strEQ(d,"END")) return KEY_END;
5264 if (strEQ(d,"eq")) return -KEY_eq;
5267 if (strEQ(d,"eof")) return -KEY_eof;
5268 if (strEQ(d,"exp")) return -KEY_exp;
5271 if (strEQ(d,"else")) return KEY_else;
5272 if (strEQ(d,"exit")) return -KEY_exit;
5273 if (strEQ(d,"eval")) return KEY_eval;
5274 if (strEQ(d,"exec")) return -KEY_exec;
5275 if (strEQ(d,"each")) return -KEY_each;
5278 if (strEQ(d,"elsif")) return KEY_elsif;
5281 if (strEQ(d,"exists")) return KEY_exists;
5282 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5285 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5286 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5289 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5292 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5293 if (strEQ(d,"endservent")) return -KEY_endservent;
5296 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5303 if (strEQ(d,"for")) return KEY_for;
5306 if (strEQ(d,"fork")) return -KEY_fork;
5309 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5310 if (strEQ(d,"flock")) return -KEY_flock;
5313 if (strEQ(d,"format")) return KEY_format;
5314 if (strEQ(d,"fileno")) return -KEY_fileno;
5317 if (strEQ(d,"foreach")) return KEY_foreach;
5320 if (strEQ(d,"formline")) return -KEY_formline;
5325 if (strnEQ(d,"get",3)) {
5330 if (strEQ(d,"ppid")) return -KEY_getppid;
5331 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5334 if (strEQ(d,"pwent")) return -KEY_getpwent;
5335 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5336 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5339 if (strEQ(d,"peername")) return -KEY_getpeername;
5340 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5341 if (strEQ(d,"priority")) return -KEY_getpriority;
5344 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5347 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5351 else if (*d == 'h') {
5352 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5353 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5354 if (strEQ(d,"hostent")) return -KEY_gethostent;
5356 else if (*d == 'n') {
5357 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5358 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5359 if (strEQ(d,"netent")) return -KEY_getnetent;
5361 else if (*d == 's') {
5362 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5363 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5364 if (strEQ(d,"servent")) return -KEY_getservent;
5365 if (strEQ(d,"sockname")) return -KEY_getsockname;
5366 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5368 else if (*d == 'g') {
5369 if (strEQ(d,"grent")) return -KEY_getgrent;
5370 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5371 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5373 else if (*d == 'l') {
5374 if (strEQ(d,"login")) return -KEY_getlogin;
5376 else if (strEQ(d,"c")) return -KEY_getc;
5381 if (strEQ(d,"gt")) return -KEY_gt;
5382 if (strEQ(d,"ge")) return -KEY_ge;
5385 if (strEQ(d,"grep")) return KEY_grep;
5386 if (strEQ(d,"goto")) return KEY_goto;
5387 if (strEQ(d,"glob")) return KEY_glob;
5390 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5395 if (strEQ(d,"hex")) return -KEY_hex;
5398 if (strEQ(d,"INIT")) return KEY_INIT;
5403 if (strEQ(d,"if")) return KEY_if;
5406 if (strEQ(d,"int")) return -KEY_int;
5409 if (strEQ(d,"index")) return -KEY_index;
5410 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5415 if (strEQ(d,"join")) return -KEY_join;
5419 if (strEQ(d,"keys")) return -KEY_keys;
5420 if (strEQ(d,"kill")) return -KEY_kill;
5426 if (strEQ(d,"lt")) return -KEY_lt;
5427 if (strEQ(d,"le")) return -KEY_le;
5428 if (strEQ(d,"lc")) return -KEY_lc;
5431 if (strEQ(d,"log")) return -KEY_log;
5434 if (strEQ(d,"last")) return KEY_last;
5435 if (strEQ(d,"link")) return -KEY_link;
5436 if (strEQ(d,"lock")) return -KEY_lock;
5439 if (strEQ(d,"local")) return KEY_local;
5440 if (strEQ(d,"lstat")) return -KEY_lstat;
5443 if (strEQ(d,"length")) return -KEY_length;
5444 if (strEQ(d,"listen")) return -KEY_listen;
5447 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5450 if (strEQ(d,"localtime")) return -KEY_localtime;
5456 case 1: return KEY_m;
5458 if (strEQ(d,"my")) return KEY_my;
5461 if (strEQ(d,"map")) return KEY_map;
5464 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5467 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5468 if (strEQ(d,"msgget")) return -KEY_msgget;
5469 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5470 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5475 if (strEQ(d,"next")) return KEY_next;
5476 if (strEQ(d,"ne")) return -KEY_ne;
5477 if (strEQ(d,"not")) return -KEY_not;
5478 if (strEQ(d,"no")) return KEY_no;
5483 if (strEQ(d,"or")) return -KEY_or;
5486 if (strEQ(d,"ord")) return -KEY_ord;
5487 if (strEQ(d,"oct")) return -KEY_oct;
5488 if (strEQ(d,"our")) return KEY_our;
5491 if (strEQ(d,"open")) return -KEY_open;
5494 if (strEQ(d,"opendir")) return -KEY_opendir;
5501 if (strEQ(d,"pop")) return -KEY_pop;
5502 if (strEQ(d,"pos")) return KEY_pos;
5505 if (strEQ(d,"push")) return -KEY_push;
5506 if (strEQ(d,"pack")) return -KEY_pack;
5507 if (strEQ(d,"pipe")) return -KEY_pipe;
5510 if (strEQ(d,"print")) return KEY_print;
5513 if (strEQ(d,"printf")) return KEY_printf;
5516 if (strEQ(d,"package")) return KEY_package;
5519 if (strEQ(d,"prototype")) return KEY_prototype;
5524 if (strEQ(d,"q")) return KEY_q;
5525 if (strEQ(d,"qr")) return KEY_qr;
5526 if (strEQ(d,"qq")) return KEY_qq;
5527 if (strEQ(d,"qw")) return KEY_qw;
5528 if (strEQ(d,"qx")) return KEY_qx;
5530 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5535 if (strEQ(d,"ref")) return -KEY_ref;
5538 if (strEQ(d,"read")) return -KEY_read;
5539 if (strEQ(d,"rand")) return -KEY_rand;
5540 if (strEQ(d,"recv")) return -KEY_recv;
5541 if (strEQ(d,"redo")) return KEY_redo;
5544 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5545 if (strEQ(d,"reset")) return -KEY_reset;
5548 if (strEQ(d,"return")) return KEY_return;
5549 if (strEQ(d,"rename")) return -KEY_rename;
5550 if (strEQ(d,"rindex")) return -KEY_rindex;
5553 if (strEQ(d,"require")) return -KEY_require;
5554 if (strEQ(d,"reverse")) return -KEY_reverse;
5555 if (strEQ(d,"readdir")) return -KEY_readdir;
5558 if (strEQ(d,"readlink")) return -KEY_readlink;
5559 if (strEQ(d,"readline")) return -KEY_readline;
5560 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5563 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5569 case 0: return KEY_s;
5571 if (strEQ(d,"scalar")) return KEY_scalar;
5576 if (strEQ(d,"seek")) return -KEY_seek;
5577 if (strEQ(d,"send")) return -KEY_send;
5580 if (strEQ(d,"semop")) return -KEY_semop;
5583 if (strEQ(d,"select")) return -KEY_select;
5584 if (strEQ(d,"semctl")) return -KEY_semctl;
5585 if (strEQ(d,"semget")) return -KEY_semget;
5588 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5589 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5592 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5593 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5596 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5599 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5600 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5601 if (strEQ(d,"setservent")) return -KEY_setservent;
5604 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5605 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5612 if (strEQ(d,"shift")) return -KEY_shift;
5615 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5616 if (strEQ(d,"shmget")) return -KEY_shmget;
5619 if (strEQ(d,"shmread")) return -KEY_shmread;
5622 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5623 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5628 if (strEQ(d,"sin")) return -KEY_sin;
5631 if (strEQ(d,"sleep")) return -KEY_sleep;
5634 if (strEQ(d,"sort")) return KEY_sort;
5635 if (strEQ(d,"socket")) return -KEY_socket;
5636 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5639 if (strEQ(d,"split")) return KEY_split;
5640 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5641 if (strEQ(d,"splice")) return -KEY_splice;
5644 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5647 if (strEQ(d,"srand")) return -KEY_srand;
5650 if (strEQ(d,"stat")) return -KEY_stat;
5651 if (strEQ(d,"study")) return KEY_study;
5654 if (strEQ(d,"substr")) return -KEY_substr;
5655 if (strEQ(d,"sub")) return KEY_sub;
5660 if (strEQ(d,"system")) return -KEY_system;
5663 if (strEQ(d,"symlink")) return -KEY_symlink;
5664 if (strEQ(d,"syscall")) return -KEY_syscall;
5665 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5666 if (strEQ(d,"sysread")) return -KEY_sysread;
5667 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5670 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5679 if (strEQ(d,"tr")) return KEY_tr;
5682 if (strEQ(d,"tie")) return KEY_tie;
5685 if (strEQ(d,"tell")) return -KEY_tell;
5686 if (strEQ(d,"tied")) return KEY_tied;
5687 if (strEQ(d,"time")) return -KEY_time;
5690 if (strEQ(d,"times")) return -KEY_times;
5693 if (strEQ(d,"telldir")) return -KEY_telldir;
5696 if (strEQ(d,"truncate")) return -KEY_truncate;
5703 if (strEQ(d,"uc")) return -KEY_uc;
5706 if (strEQ(d,"use")) return KEY_use;
5709 if (strEQ(d,"undef")) return KEY_undef;
5710 if (strEQ(d,"until")) return KEY_until;
5711 if (strEQ(d,"untie")) return KEY_untie;
5712 if (strEQ(d,"utime")) return -KEY_utime;
5713 if (strEQ(d,"umask")) return -KEY_umask;
5716 if (strEQ(d,"unless")) return KEY_unless;
5717 if (strEQ(d,"unpack")) return -KEY_unpack;
5718 if (strEQ(d,"unlink")) return -KEY_unlink;
5721 if (strEQ(d,"unshift")) return -KEY_unshift;
5722 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5727 if (strEQ(d,"values")) return -KEY_values;
5728 if (strEQ(d,"vec")) return -KEY_vec;
5733 if (strEQ(d,"warn")) return -KEY_warn;
5734 if (strEQ(d,"wait")) return -KEY_wait;
5737 if (strEQ(d,"while")) return KEY_while;
5738 if (strEQ(d,"write")) return -KEY_write;
5741 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5744 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5749 if (len == 1) return -KEY_x;
5750 if (strEQ(d,"xor")) return -KEY_xor;
5753 if (len == 1) return KEY_y;
5762 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5766 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5767 if (ckWARN(WARN_SYNTAX)) {
5769 for (w = s+2; *w && level; w++) {
5776 for (; *w && isSPACE(*w); w++) ;
5777 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5778 Perl_warner(aTHX_ WARN_SYNTAX,
5779 "%s (...) interpreted as function",name);
5782 while (s < PL_bufend && isSPACE(*s))
5786 while (s < PL_bufend && isSPACE(*s))
5788 if (isIDFIRST_lazy_if(s,UTF)) {
5790 while (isALNUM_lazy_if(s,UTF))
5792 while (s < PL_bufend && isSPACE(*s))
5797 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5801 Perl_croak(aTHX_ "No comma allowed after %s", what);
5806 /* Either returns sv, or mortalizes sv and returns a new SV*.
5807 Best used as sv=new_constant(..., sv, ...).
5808 If s, pv are NULL, calls subroutine with one argument,
5809 and type is used with error messages only. */
5812 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5816 HV *table = GvHV(PL_hintgv); /* ^H */
5820 const char *why1, *why2, *why3;
5822 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5825 why2 = strEQ(key,"charnames")
5826 ? "(possibly a missing \"use charnames ...\")"
5828 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5829 (type ? type: "undef"), why2);
5831 /* This is convoluted and evil ("goto considered harmful")
5832 * but I do not understand the intricacies of all the different
5833 * failure modes of %^H in here. The goal here is to make
5834 * the most probable error message user-friendly. --jhi */
5839 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5840 (type ? type: "undef"), why1, why2, why3);
5842 yyerror(SvPVX(msg));
5846 cvp = hv_fetch(table, key, strlen(key), FALSE);
5847 if (!cvp || !SvOK(*cvp)) {
5850 why3 = "} is not defined";
5853 sv_2mortal(sv); /* Parent created it permanently */
5856 pv = sv_2mortal(newSVpvn(s, len));
5858 typesv = sv_2mortal(newSVpv(type, 0));
5860 typesv = &PL_sv_undef;
5862 PUSHSTACKi(PERLSI_OVERLOAD);
5874 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5878 /* Check the eval first */
5879 if (!PL_in_eval && SvTRUE(ERRSV)) {
5881 sv_catpv(ERRSV, "Propagated");
5882 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5884 res = SvREFCNT_inc(sv);
5888 (void)SvREFCNT_inc(res);
5897 why1 = "Call to &{$^H{";
5899 why3 = "}} did not return a defined value";
5908 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5910 register char *d = dest;
5911 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5914 Perl_croak(aTHX_ ident_too_long);
5915 if (isALNUM(*s)) /* UTF handled below */
5917 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5922 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5926 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5927 char *t = s + UTF8SKIP(s);
5928 while (*t & 0x80 && is_utf8_mark((U8*)t))
5930 if (d + (t - s) > e)
5931 Perl_croak(aTHX_ ident_too_long);
5932 Copy(s, d, t - s, char);
5945 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5955 e = d + destlen - 3; /* two-character token, ending NUL */
5957 while (isDIGIT(*s)) {
5959 Perl_croak(aTHX_ ident_too_long);
5966 Perl_croak(aTHX_ ident_too_long);
5967 if (isALNUM(*s)) /* UTF handled below */
5969 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5974 else if (*s == ':' && s[1] == ':') {
5978 else if (UTF && *(U8*)s >= 0xc0 && isALNUM_utf8((U8*)s)) {
5979 char *t = s + UTF8SKIP(s);
5980 while (*t & 0x80 && is_utf8_mark((U8*)t))
5982 if (d + (t - s) > e)
5983 Perl_croak(aTHX_ ident_too_long);
5984 Copy(s, d, t - s, char);
5995 if (PL_lex_state != LEX_NORMAL)
5996 PL_lex_state = LEX_INTERPENDMAYBE;
5999 if (*s == '$' && s[1] &&
6000 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6013 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6018 if (isSPACE(s[-1])) {
6021 if (!SPACE_OR_TAB(ch)) {
6027 if (isIDFIRST_lazy_if(d,UTF)) {
6031 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6033 while (e < send && *e & 0x80 && is_utf8_mark((U8*)e))
6036 Copy(s, d, e - s, char);
6041 while ((isALNUM(*s) || *s == ':') && d < e)
6044 Perl_croak(aTHX_ ident_too_long);
6047 while (s < send && SPACE_OR_TAB(*s)) s++;
6048 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6049 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6050 const char *brack = *s == '[' ? "[...]" : "{...}";
6051 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6052 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6053 funny, dest, brack, funny, dest, brack);
6056 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6060 /* Handle extended ${^Foo} variables
6061 * 1999-02-27 mjd-perl-patch@plover.com */
6062 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6066 while (isALNUM(*s) && d < e) {
6070 Perl_croak(aTHX_ ident_too_long);
6075 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6076 PL_lex_state = LEX_INTERPEND;
6079 if (PL_lex_state == LEX_NORMAL) {
6080 if (ckWARN(WARN_AMBIGUOUS) &&
6081 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6083 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6084 "Ambiguous use of %c{%s} resolved to %c%s",
6085 funny, dest, funny, dest);
6090 s = bracket; /* let the parser handle it */
6094 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6095 PL_lex_state = LEX_INTERPEND;
6100 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6105 *pmfl |= PMf_GLOBAL;
6107 *pmfl |= PMf_CONTINUE;
6111 *pmfl |= PMf_MULTILINE;
6113 *pmfl |= PMf_SINGLELINE;
6115 *pmfl |= PMf_EXTENDED;
6119 S_scan_pat(pTHX_ char *start, I32 type)
6124 s = scan_str(start,FALSE,FALSE);
6127 SvREFCNT_dec(PL_lex_stuff);
6128 PL_lex_stuff = Nullsv;
6129 Perl_croak(aTHX_ "Search pattern not terminated");
6132 pm = (PMOP*)newPMOP(type, 0);
6133 if (PL_multi_open == '?')
6134 pm->op_pmflags |= PMf_ONCE;
6136 while (*s && strchr("iomsx", *s))
6137 pmflag(&pm->op_pmflags,*s++);
6140 while (*s && strchr("iogcmsx", *s))
6141 pmflag(&pm->op_pmflags,*s++);
6143 pm->op_pmpermflags = pm->op_pmflags;
6145 PL_lex_op = (OP*)pm;
6146 yylval.ival = OP_MATCH;
6151 S_scan_subst(pTHX_ char *start)
6158 yylval.ival = OP_NULL;
6160 s = scan_str(start,FALSE,FALSE);
6164 SvREFCNT_dec(PL_lex_stuff);
6165 PL_lex_stuff = Nullsv;
6166 Perl_croak(aTHX_ "Substitution pattern not terminated");
6169 if (s[-1] == PL_multi_open)
6172 first_start = PL_multi_start;
6173 s = scan_str(s,FALSE,FALSE);
6176 SvREFCNT_dec(PL_lex_stuff);
6177 PL_lex_stuff = Nullsv;
6179 SvREFCNT_dec(PL_lex_repl);
6180 PL_lex_repl = Nullsv;
6181 Perl_croak(aTHX_ "Substitution replacement not terminated");
6183 PL_multi_start = first_start; /* so whole substitution is taken together */
6185 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6191 else if (strchr("iogcmsx", *s))
6192 pmflag(&pm->op_pmflags,*s++);
6199 PL_sublex_info.super_bufptr = s;
6200 PL_sublex_info.super_bufend = PL_bufend;
6202 pm->op_pmflags |= PMf_EVAL;
6203 repl = newSVpvn("",0);
6205 sv_catpv(repl, es ? "eval " : "do ");
6206 sv_catpvn(repl, "{ ", 2);
6207 sv_catsv(repl, PL_lex_repl);
6208 sv_catpvn(repl, " };", 2);
6210 SvREFCNT_dec(PL_lex_repl);
6214 pm->op_pmpermflags = pm->op_pmflags;
6215 PL_lex_op = (OP*)pm;
6216 yylval.ival = OP_SUBST;
6221 S_scan_trans(pTHX_ char *start)
6232 yylval.ival = OP_NULL;
6234 s = scan_str(start,FALSE,FALSE);
6237 SvREFCNT_dec(PL_lex_stuff);
6238 PL_lex_stuff = Nullsv;
6239 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6241 if (s[-1] == PL_multi_open)
6244 s = scan_str(s,FALSE,FALSE);
6247 SvREFCNT_dec(PL_lex_stuff);
6248 PL_lex_stuff = Nullsv;
6250 SvREFCNT_dec(PL_lex_repl);
6251 PL_lex_repl = Nullsv;
6252 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6255 New(803,tbl,256,short);
6256 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6258 complement = del = squash = 0;
6259 while (strchr("cds", *s)) {
6261 complement = OPpTRANS_COMPLEMENT;
6263 del = OPpTRANS_DELETE;
6265 squash = OPpTRANS_SQUASH;
6268 o->op_private = del|squash|complement|
6269 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6270 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6273 yylval.ival = OP_TRANS;
6278 S_scan_heredoc(pTHX_ register char *s)
6281 I32 op_type = OP_SCALAR;
6288 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6292 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6295 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6296 if (*peek && strchr("`'\"",*peek)) {
6299 s = delimcpy(d, e, s, PL_bufend, term, &len);
6309 if (!isALNUM_lazy_if(s,UTF))
6310 deprecate("bare << to mean <<\"\"");
6311 for (; isALNUM_lazy_if(s,UTF); s++) {
6316 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6317 Perl_croak(aTHX_ "Delimiter for here document is too long");
6320 len = d - PL_tokenbuf;
6321 #ifndef PERL_STRICT_CR
6322 d = strchr(s, '\r');
6326 while (s < PL_bufend) {
6332 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6341 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6346 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6347 herewas = newSVpvn(s,PL_bufend-s);
6349 s--, herewas = newSVpvn(s,d-s);
6350 s += SvCUR(herewas);
6352 tmpstr = NEWSV(87,79);
6353 sv_upgrade(tmpstr, SVt_PVIV);
6358 else if (term == '`') {
6359 op_type = OP_BACKTICK;
6360 SvIVX(tmpstr) = '\\';
6364 PL_multi_start = CopLINE(PL_curcop);
6365 PL_multi_open = PL_multi_close = '<';
6366 term = *PL_tokenbuf;
6367 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6368 char *bufptr = PL_sublex_info.super_bufptr;
6369 char *bufend = PL_sublex_info.super_bufend;
6370 char *olds = s - SvCUR(herewas);
6371 s = strchr(bufptr, '\n');
6375 while (s < bufend &&
6376 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6378 CopLINE_inc(PL_curcop);
6381 CopLINE_set(PL_curcop, PL_multi_start);
6382 missingterm(PL_tokenbuf);
6384 sv_setpvn(herewas,bufptr,d-bufptr+1);
6385 sv_setpvn(tmpstr,d+1,s-d);
6387 sv_catpvn(herewas,s,bufend-s);
6388 (void)strcpy(bufptr,SvPVX(herewas));
6395 while (s < PL_bufend &&
6396 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6398 CopLINE_inc(PL_curcop);
6400 if (s >= PL_bufend) {
6401 CopLINE_set(PL_curcop, PL_multi_start);
6402 missingterm(PL_tokenbuf);
6404 sv_setpvn(tmpstr,d+1,s-d);
6406 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6408 sv_catpvn(herewas,s,PL_bufend-s);
6409 sv_setsv(PL_linestr,herewas);
6410 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6411 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6414 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6415 while (s >= PL_bufend) { /* multiple line string? */
6417 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6418 CopLINE_set(PL_curcop, PL_multi_start);
6419 missingterm(PL_tokenbuf);
6421 CopLINE_inc(PL_curcop);
6422 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6423 #ifndef PERL_STRICT_CR
6424 if (PL_bufend - PL_linestart >= 2) {
6425 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6426 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6428 PL_bufend[-2] = '\n';
6430 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6432 else if (PL_bufend[-1] == '\r')
6433 PL_bufend[-1] = '\n';
6435 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6436 PL_bufend[-1] = '\n';
6438 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6439 SV *sv = NEWSV(88,0);
6441 sv_upgrade(sv, SVt_PVMG);
6442 sv_setsv(sv,PL_linestr);
6443 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6445 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6448 sv_catsv(PL_linestr,herewas);
6449 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6453 sv_catsv(tmpstr,PL_linestr);
6458 PL_multi_end = CopLINE(PL_curcop);
6459 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6460 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6461 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6463 SvREFCNT_dec(herewas);
6464 if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6466 PL_lex_stuff = tmpstr;
6467 yylval.ival = op_type;
6472 takes: current position in input buffer
6473 returns: new position in input buffer
6474 side-effects: yylval and lex_op are set.
6479 <FH> read from filehandle
6480 <pkg::FH> read from package qualified filehandle
6481 <pkg'FH> read from package qualified filehandle
6482 <$fh> read from filehandle in $fh
6488 S_scan_inputsymbol(pTHX_ char *start)
6490 register char *s = start; /* current position in buffer */
6496 d = PL_tokenbuf; /* start of temp holding space */
6497 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6498 end = strchr(s, '\n');
6501 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6503 /* die if we didn't have space for the contents of the <>,
6504 or if it didn't end, or if we see a newline
6507 if (len >= sizeof PL_tokenbuf)
6508 Perl_croak(aTHX_ "Excessively long <> operator");
6510 Perl_croak(aTHX_ "Unterminated <> operator");
6515 Remember, only scalar variables are interpreted as filehandles by
6516 this code. Anything more complex (e.g., <$fh{$num}>) will be
6517 treated as a glob() call.
6518 This code makes use of the fact that except for the $ at the front,
6519 a scalar variable and a filehandle look the same.
6521 if (*d == '$' && d[1]) d++;
6523 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6524 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6527 /* If we've tried to read what we allow filehandles to look like, and
6528 there's still text left, then it must be a glob() and not a getline.
6529 Use scan_str to pull out the stuff between the <> and treat it
6530 as nothing more than a string.
6533 if (d - PL_tokenbuf != len) {
6534 yylval.ival = OP_GLOB;
6536 s = scan_str(start,FALSE,FALSE);
6538 Perl_croak(aTHX_ "Glob not terminated");
6542 /* we're in a filehandle read situation */
6545 /* turn <> into <ARGV> */
6547 (void)strcpy(d,"ARGV");
6549 /* if <$fh>, create the ops to turn the variable into a
6555 /* try to find it in the pad for this block, otherwise find
6556 add symbol table ops
6558 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6559 OP *o = newOP(OP_PADSV, 0);
6561 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6564 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6565 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6566 newUNOP(OP_RV2SV, 0,
6567 newGVOP(OP_GV, 0, gv)));
6569 PL_lex_op->op_flags |= OPf_SPECIAL;
6570 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6571 yylval.ival = OP_NULL;
6574 /* If it's none of the above, it must be a literal filehandle
6575 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6577 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6578 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6579 yylval.ival = OP_NULL;
6588 takes: start position in buffer
6589 keep_quoted preserve \ on the embedded delimiter(s)
6590 keep_delims preserve the delimiters around the string
6591 returns: position to continue reading from buffer
6592 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6593 updates the read buffer.
6595 This subroutine pulls a string out of the input. It is called for:
6596 q single quotes q(literal text)
6597 ' single quotes 'literal text'
6598 qq double quotes qq(interpolate $here please)
6599 " double quotes "interpolate $here please"
6600 qx backticks qx(/bin/ls -l)
6601 ` backticks `/bin/ls -l`
6602 qw quote words @EXPORT_OK = qw( func() $spam )
6603 m// regexp match m/this/
6604 s/// regexp substitute s/this/that/
6605 tr/// string transliterate tr/this/that/
6606 y/// string transliterate y/this/that/
6607 ($*@) sub prototypes sub foo ($)
6608 (stuff) sub attr parameters sub foo : attr(stuff)
6609 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6611 In most of these cases (all but <>, patterns and transliterate)
6612 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6613 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6614 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6617 It skips whitespace before the string starts, and treats the first
6618 character as the delimiter. If the delimiter is one of ([{< then
6619 the corresponding "close" character )]}> is used as the closing
6620 delimiter. It allows quoting of delimiters, and if the string has
6621 balanced delimiters ([{<>}]) it allows nesting.
6623 The lexer always reads these strings into lex_stuff, except in the
6624 case of the operators which take *two* arguments (s/// and tr///)
6625 when it checks to see if lex_stuff is full (presumably with the 1st
6626 arg to s or tr) and if so puts the string into lex_repl.
6631 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6633 SV *sv; /* scalar value: string */
6634 char *tmps; /* temp string, used for delimiter matching */
6635 register char *s = start; /* current position in the buffer */
6636 register char term; /* terminating character */
6637 register char *to; /* current position in the sv's data */
6638 I32 brackets = 1; /* bracket nesting level */
6639 bool has_utf8 = FALSE; /* is there any utf8 content? */
6641 /* skip space before the delimiter */
6645 /* mark where we are, in case we need to report errors */
6648 /* after skipping whitespace, the next character is the terminator */
6650 if ((term & 0x80) && UTF)
6653 /* mark where we are */
6654 PL_multi_start = CopLINE(PL_curcop);
6655 PL_multi_open = term;
6657 /* find corresponding closing delimiter */
6658 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6660 PL_multi_close = term;
6662 /* create a new SV to hold the contents. 87 is leak category, I'm
6663 assuming. 79 is the SV's initial length. What a random number. */
6665 sv_upgrade(sv, SVt_PVIV);
6667 (void)SvPOK_only(sv); /* validate pointer */
6669 /* move past delimiter and try to read a complete string */
6671 sv_catpvn(sv, s, 1);
6674 /* extend sv if need be */
6675 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6676 /* set 'to' to the next character in the sv's string */
6677 to = SvPVX(sv)+SvCUR(sv);
6679 /* if open delimiter is the close delimiter read unbridle */
6680 if (PL_multi_open == PL_multi_close) {
6681 for (; s < PL_bufend; s++,to++) {
6682 /* embedded newlines increment the current line number */
6683 if (*s == '\n' && !PL_rsfp)
6684 CopLINE_inc(PL_curcop);
6685 /* handle quoted delimiters */
6686 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6687 if (!keep_quoted && s[1] == term)
6689 /* any other quotes are simply copied straight through */
6693 /* terminate when run out of buffer (the for() condition), or
6694 have found the terminator */
6695 else if (*s == term)
6697 else if (!has_utf8 && (*s & 0x80) && UTF)
6703 /* if the terminator isn't the same as the start character (e.g.,
6704 matched brackets), we have to allow more in the quoting, and
6705 be prepared for nested brackets.
6708 /* read until we run out of string, or we find the terminator */
6709 for (; s < PL_bufend; s++,to++) {
6710 /* embedded newlines increment the line count */
6711 if (*s == '\n' && !PL_rsfp)
6712 CopLINE_inc(PL_curcop);
6713 /* backslashes can escape the open or closing characters */
6714 if (*s == '\\' && s+1 < PL_bufend) {
6716 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6721 /* allow nested opens and closes */
6722 else if (*s == PL_multi_close && --brackets <= 0)
6724 else if (*s == PL_multi_open)
6726 else if (!has_utf8 && (*s & 0x80) && UTF)
6731 /* terminate the copied string and update the sv's end-of-string */
6733 SvCUR_set(sv, to - SvPVX(sv));
6736 * this next chunk reads more into the buffer if we're not done yet
6740 break; /* handle case where we are done yet :-) */
6742 #ifndef PERL_STRICT_CR
6743 if (to - SvPVX(sv) >= 2) {
6744 if ((to[-2] == '\r' && to[-1] == '\n') ||
6745 (to[-2] == '\n' && to[-1] == '\r'))
6749 SvCUR_set(sv, to - SvPVX(sv));
6751 else if (to[-1] == '\r')
6754 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6758 /* if we're out of file, or a read fails, bail and reset the current
6759 line marker so we can report where the unterminated string began
6762 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6764 CopLINE_set(PL_curcop, PL_multi_start);
6767 /* we read a line, so increment our line counter */
6768 CopLINE_inc(PL_curcop);
6770 /* update debugger info */
6771 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6772 SV *sv = NEWSV(88,0);
6774 sv_upgrade(sv, SVt_PVMG);
6775 sv_setsv(sv,PL_linestr);
6776 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6779 /* having changed the buffer, we must update PL_bufend */
6780 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6783 /* at this point, we have successfully read the delimited string */
6786 sv_catpvn(sv, s, 1);
6789 PL_multi_end = CopLINE(PL_curcop);
6792 /* if we allocated too much space, give some back */
6793 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6794 SvLEN_set(sv, SvCUR(sv) + 1);
6795 Renew(SvPVX(sv), SvLEN(sv), char);
6798 /* decide whether this is the first or second quoted string we've read
6811 takes: pointer to position in buffer
6812 returns: pointer to new position in buffer
6813 side-effects: builds ops for the constant in yylval.op
6815 Read a number in any of the formats that Perl accepts:
6817 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6818 [\d_]+(\.[\d_]*)?[Ee](\d+)
6820 Underbars (_) are allowed in decimal numbers. If -w is on,
6821 underbars before a decimal point must be at three digit intervals.
6823 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6826 If it reads a number without a decimal point or an exponent, it will
6827 try converting the number to an integer and see if it can do so
6828 without loss of precision.
6832 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6834 register char *s = start; /* current position in buffer */
6835 register char *d; /* destination in temp buffer */
6836 register char *e; /* end of temp buffer */
6837 NV nv; /* number read, as a double */
6838 SV *sv = Nullsv; /* place to put the converted number */
6839 bool floatit; /* boolean: int or float? */
6840 char *lastub = 0; /* position of last underbar */
6841 static char number_too_long[] = "Number too long";
6843 /* We use the first character to decide what type of number this is */
6847 Perl_croak(aTHX_ "panic: scan_num");
6849 /* if it starts with a 0, it could be an octal number, a decimal in
6850 0.13 disguise, or a hexadecimal number, or a binary number. */
6854 u holds the "number so far"
6855 shift the power of 2 of the base
6856 (hex == 4, octal == 3, binary == 1)
6857 overflowed was the number more than we can hold?
6859 Shift is used when we add a digit. It also serves as an "are
6860 we in octal/hex/binary?" indicator to disallow hex characters
6866 bool overflowed = FALSE;
6867 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6868 static char* bases[5] = { "", "binary", "", "octal",
6870 static char* Bases[5] = { "", "Binary", "", "Octal",
6872 static char *maxima[5] = { "",
6873 "0b11111111111111111111111111111111",
6877 char *base, *Base, *max;
6883 } else if (s[1] == 'b') {
6887 /* check for a decimal in disguise */
6888 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6890 /* so it must be octal */
6894 base = bases[shift];
6895 Base = Bases[shift];
6896 max = maxima[shift];
6898 /* read the rest of the number */
6900 /* x is used in the overflow test,
6901 b is the digit we're adding on. */
6906 /* if we don't mention it, we're done */
6915 /* 8 and 9 are not octal */
6918 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6922 case '2': case '3': case '4':
6923 case '5': case '6': case '7':
6925 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6929 b = *s++ & 15; /* ASCII digit -> value of digit */
6933 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6934 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6935 /* make sure they said 0x */
6940 /* Prepare to put the digit we have onto the end
6941 of the number so far. We check for overflows.
6946 x = u << shift; /* make room for the digit */
6948 if ((x >> shift) != u
6949 && !(PL_hints & HINT_NEW_BINARY)) {
6952 if (ckWARN_d(WARN_OVERFLOW))
6953 Perl_warner(aTHX_ WARN_OVERFLOW,
6954 "Integer overflow in %s number",
6957 u = x | b; /* add the digit to the end */
6960 n *= nvshift[shift];
6961 /* If an NV has not enough bits in its
6962 * mantissa to represent an UV this summing of
6963 * small low-order numbers is a waste of time
6964 * (because the NV cannot preserve the
6965 * low-order bits anyway): we could just
6966 * remember when did we overflow and in the
6967 * end just multiply n by the right
6975 /* if we get here, we had success: make a scalar value from
6981 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6982 Perl_warner(aTHX_ WARN_PORTABLE,
6983 "%s number > %s non-portable",
6989 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
6990 Perl_warner(aTHX_ WARN_PORTABLE,
6991 "%s number > %s non-portable",
6996 if (PL_hints & HINT_NEW_BINARY)
6997 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7002 handle decimal numbers.
7003 we're also sent here when we read a 0 as the first digit
7005 case '1': case '2': case '3': case '4': case '5':
7006 case '6': case '7': case '8': case '9': case '.':
7009 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7012 /* read next group of digits and _ and copy into d */
7013 while (isDIGIT(*s) || *s == '_') {
7014 /* skip underscores, checking for misplaced ones
7018 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7019 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7023 /* check for end of fixed-length buffer */
7025 Perl_croak(aTHX_ number_too_long);
7026 /* if we're ok, copy the character */
7031 /* final misplaced underbar check */
7032 if (lastub && s - lastub != 3) {
7033 if (ckWARN(WARN_SYNTAX))
7034 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7037 /* read a decimal portion if there is one. avoid
7038 3..5 being interpreted as the number 3. followed
7041 if (*s == '.' && s[1] != '.') {
7045 /* copy, ignoring underbars, until we run out of
7046 digits. Note: no misplaced underbar checks!
7048 for (; isDIGIT(*s) || *s == '_'; s++) {
7049 /* fixed length buffer check */
7051 Perl_croak(aTHX_ number_too_long);
7055 if (*s == '.' && isDIGIT(s[1])) {
7056 /* oops, it's really a v-string, but without the "v" */
7062 /* read exponent part, if present */
7063 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7067 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7068 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7070 /* allow positive or negative exponent */
7071 if (*s == '+' || *s == '-')
7074 /* read digits of exponent (no underbars :-) */
7075 while (isDIGIT(*s)) {
7077 Perl_croak(aTHX_ number_too_long);
7082 /* terminate the string */
7085 /* make an sv from the string */
7088 #if defined(Strtol) && defined(Strtoul)
7091 strtol/strtoll sets errno to ERANGE if the number is too big
7092 for an integer. We try to do an integer conversion first
7093 if no characters indicating "float" have been found.
7100 if (*PL_tokenbuf == '-')
7101 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7103 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7105 floatit = TRUE; /* Probably just too large. */
7106 else if (*PL_tokenbuf == '-')
7108 else if (uv <= IV_MAX)
7109 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7114 nv = Atof(PL_tokenbuf);
7119 No working strtou?ll?.
7121 Unfortunately atol() doesn't do range checks (returning
7122 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7123 everywhere [1], so we cannot use use atol() (or atoll()).
7124 If we could, they would be used, as Atol(), very much like
7125 Strtol() and Strtoul() are used above.
7127 [1] XXX Configure test needed to check for atol()
7128 (and atoll()) overflow behaviour XXX
7132 We need to do this the hard way. */
7134 nv = Atof(PL_tokenbuf);
7136 /* See if we can make do with an integer value without loss of
7137 precision. We use U_V to cast to a UV, because some
7138 compilers have issues. Then we try casting it back and see
7139 if it was the same [1]. We only do this if we know we
7140 specifically read an integer. If floatit is true, then we
7141 don't need to do the conversion at all.
7143 [1] Note that this is lossy if our NVs cannot preserve our
7144 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7145 and NV_PRESERVES_UV_BITS (a number), but in general we really
7146 do hope all such potentially lossy platforms have strtou?ll?
7147 to do a lossless IV/UV conversion.
7149 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7150 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7151 as NV_DIG and NV_MANT_DIG)?
7157 if (!floatit && (NV)uv == nv) {
7159 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7167 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7168 (PL_hints & HINT_NEW_INTEGER) )
7169 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7170 (floatit ? "float" : "integer"),
7174 /* if it starts with a v, it could be a v-string */
7180 while (isDIGIT(*pos) || *pos == '_')
7182 if (!isALPHA(*pos)) {
7184 U8 tmpbuf[UTF8_MAXLEN+1];
7187 s++; /* get past 'v' */
7190 sv_setpvn(sv, "", 0);
7193 if (*s == '0' && isDIGIT(s[1]))
7194 yyerror("Octal number in vector unsupported");
7197 /* this is atoi() that tolerates underscores */
7200 while (--end >= s) {
7205 rev += (*end - '0') * mult;
7207 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7208 Perl_warner(aTHX_ WARN_OVERFLOW,
7209 "Integer overflow in decimal number");
7212 tmpend = uv_to_utf8(tmpbuf, rev);
7213 utf8 = utf8 || rev > 127;
7214 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7215 if (*pos == '.' && isDIGIT(pos[1]))
7221 while (isDIGIT(*pos) || *pos == '_')
7230 sv_utf8_downgrade(sv, TRUE);
7237 /* make the op for the constant and return */
7240 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7242 lvalp->opval = Nullop;
7248 S_scan_formline(pTHX_ register char *s)
7252 SV *stuff = newSVpvn("",0);
7253 bool needargs = FALSE;
7256 if (*s == '.' || *s == /*{*/'}') {
7258 #ifdef PERL_STRICT_CR
7259 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7261 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7263 if (*t == '\n' || t == PL_bufend)
7266 if (PL_in_eval && !PL_rsfp) {
7267 eol = strchr(s,'\n');
7272 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7274 for (t = s; t < eol; t++) {
7275 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7277 goto enough; /* ~~ must be first line in formline */
7279 if (*t == '@' || *t == '^')
7282 sv_catpvn(stuff, s, eol-s);
7283 #ifndef PERL_STRICT_CR
7284 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7285 char *end = SvPVX(stuff) + SvCUR(stuff);
7294 s = filter_gets(PL_linestr, PL_rsfp, 0);
7295 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7296 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7299 yyerror("Format not terminated");
7309 PL_lex_state = LEX_NORMAL;
7310 PL_nextval[PL_nexttoke].ival = 0;
7314 PL_lex_state = LEX_FORMLINE;
7315 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7317 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7321 SvREFCNT_dec(stuff);
7322 PL_lex_formbrack = 0;
7333 PL_cshlen = strlen(PL_cshname);
7338 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7340 I32 oldsavestack_ix = PL_savestack_ix;
7341 CV* outsidecv = PL_compcv;
7345 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7347 SAVEI32(PL_subline);
7348 save_item(PL_subname);
7351 SAVESPTR(PL_comppad_name);
7352 SAVESPTR(PL_compcv);
7353 SAVEI32(PL_comppad_name_fill);
7354 SAVEI32(PL_min_intro_pending);
7355 SAVEI32(PL_max_intro_pending);
7356 SAVEI32(PL_pad_reset_pending);
7358 PL_compcv = (CV*)NEWSV(1104,0);
7359 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7360 CvFLAGS(PL_compcv) |= flags;
7362 PL_comppad = newAV();
7363 av_push(PL_comppad, Nullsv);
7364 PL_curpad = AvARRAY(PL_comppad);
7365 PL_comppad_name = newAV();
7366 PL_comppad_name_fill = 0;
7367 PL_min_intro_pending = 0;
7369 PL_subline = CopLINE(PL_curcop);
7371 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7372 PL_curpad[0] = (SV*)newAV();
7373 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7374 #endif /* USE_THREADS */
7376 comppadlist = newAV();
7377 AvREAL_off(comppadlist);
7378 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7379 av_store(comppadlist, 1, (SV*)PL_comppad);
7381 CvPADLIST(PL_compcv) = comppadlist;
7382 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7384 CvOWNER(PL_compcv) = 0;
7385 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7386 MUTEX_INIT(CvMUTEXP(PL_compcv));
7387 #endif /* USE_THREADS */
7389 return oldsavestack_ix;
7393 Perl_yywarn(pTHX_ char *s)
7395 PL_in_eval |= EVAL_WARNONLY;
7397 PL_in_eval &= ~EVAL_WARNONLY;
7402 Perl_yyerror(pTHX_ char *s)
7405 char *context = NULL;
7409 if (!yychar || (yychar == ';' && !PL_rsfp))
7411 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7412 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7413 while (isSPACE(*PL_oldoldbufptr))
7415 context = PL_oldoldbufptr;
7416 contlen = PL_bufptr - PL_oldoldbufptr;
7418 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7419 PL_oldbufptr != PL_bufptr) {
7420 while (isSPACE(*PL_oldbufptr))
7422 context = PL_oldbufptr;
7423 contlen = PL_bufptr - PL_oldbufptr;
7425 else if (yychar > 255)
7426 where = "next token ???";
7427 #ifdef USE_PURE_BISON
7428 /* GNU Bison sets the value -2 */
7429 else if (yychar == -2) {
7431 else if ((yychar & 127) == 127) {
7433 if (PL_lex_state == LEX_NORMAL ||
7434 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7435 where = "at end of line";
7436 else if (PL_lex_inpat)
7437 where = "within pattern";
7439 where = "within string";
7442 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7444 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7445 else if (isPRINT_LC(yychar))
7446 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7448 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7449 where = SvPVX(where_sv);
7451 msg = sv_2mortal(newSVpv(s, 0));
7452 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7453 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7455 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7457 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7458 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7459 Perl_sv_catpvf(aTHX_ msg,
7460 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7461 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7464 if (PL_in_eval & EVAL_WARNONLY)
7465 Perl_warn(aTHX_ "%"SVf, msg);
7468 if (PL_error_count >= 10) {
7469 if (PL_in_eval && SvCUR(ERRSV))
7470 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7471 ERRSV, CopFILE(PL_curcop));
7473 Perl_croak(aTHX_ "%s has too many errors.\n",
7474 CopFILE(PL_curcop));
7477 PL_in_my_stash = Nullhv;
7482 S_swallow_bom(pTHX_ U8 *s)
7485 slen = SvCUR(PL_linestr);
7489 /* UTF-16 little-endian */
7490 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7491 Perl_croak(aTHX_ "Unsupported script encoding");
7492 #ifndef PERL_NO_UTF16_FILTER
7493 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7495 if (PL_bufend > (char*)s) {
7499 filter_add(utf16rev_textfilter, NULL);
7500 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7501 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7502 PL_bufend - (char*)s - 1,
7504 Copy(news, s, newlen, U8);
7505 SvCUR_set(PL_linestr, newlen);
7506 PL_bufend = SvPVX(PL_linestr) + newlen;
7507 news[newlen++] = '\0';
7511 Perl_croak(aTHX_ "Unsupported script encoding");
7516 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7517 #ifndef PERL_NO_UTF16_FILTER
7518 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7520 if (PL_bufend > (char *)s) {
7524 filter_add(utf16_textfilter, NULL);
7525 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7526 PL_bufend = (char*)utf16_to_utf8(s, news,
7527 PL_bufend - (char*)s,
7529 Copy(news, s, newlen, U8);
7530 SvCUR_set(PL_linestr, newlen);
7531 PL_bufend = SvPVX(PL_linestr) + newlen;
7532 news[newlen++] = '\0';
7536 Perl_croak(aTHX_ "Unsupported script encoding");
7541 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7542 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7547 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7548 s[2] == 0xFE && s[3] == 0xFF)
7550 Perl_croak(aTHX_ "Unsupported script encoding");
7562 * Restore a source filter.
7566 restore_rsfp(pTHXo_ void *f)
7568 PerlIO *fp = (PerlIO*)f;
7570 if (PL_rsfp == PerlIO_stdin())
7571 PerlIO_clearerr(PL_rsfp);
7572 else if (PL_rsfp && (PL_rsfp != fp))
7573 PerlIO_close(PL_rsfp);
7577 #ifndef PERL_NO_UTF16_FILTER
7579 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7581 I32 count = FILTER_READ(idx+1, sv, maxlen);
7586 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7587 if (!*SvPV_nolen(sv))
7588 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7591 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7592 sv_usepvn(sv, (char*)tmps, tend - tmps);
7598 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7600 I32 count = FILTER_READ(idx+1, sv, maxlen);
7605 if (!*SvPV_nolen(sv))
7606 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7609 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7610 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7611 sv_usepvn(sv, (char*)tmps, tend - tmps);