3 * Copyright (c) 1991-2001, Larry Wall
5 * You may distribute under the terms of either the GNU General Public
6 * License or the Artistic License, as specified in the README file.
11 * "It all comes from here, the stench and the peril." --Frodo
15 * This file is the lexer for Perl. It's closely linked to the
18 * The main routine is yylex(), which returns the next token.
22 #define PERL_IN_TOKE_C
25 #define yychar PL_yychar
26 #define yylval PL_yylval
28 static char ident_too_long[] = "Identifier too long";
30 static void restore_rsfp(pTHXo_ void *f);
31 #ifndef PERL_NO_UTF16_FILTER
32 static I32 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
33 static I32 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen);
36 #define XFAKEBRACK 128
39 /*#define UTF (SvUTF8(PL_linestr) && !(PL_hints & HINT_BYTE))*/
40 #define UTF (PL_hints & HINT_UTF8)
42 /* In variables name $^X, these are the legal values for X.
43 * 1999-02-27 mjd-perl-patch@plover.com */
44 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
46 /* On MacOS, respect nonbreaking spaces */
47 #ifdef MACOS_TRADITIONAL
48 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
50 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
53 /* LEX_* are values for PL_lex_state, the state of the lexer.
54 * They are arranged oddly so that the guard on the switch statement
55 * can get by with a single comparison (if the compiler is smart enough).
58 /* #define LEX_NOTPARSING 11 is done in perl.h. */
61 #define LEX_INTERPNORMAL 9
62 #define LEX_INTERPCASEMOD 8
63 #define LEX_INTERPPUSH 7
64 #define LEX_INTERPSTART 6
65 #define LEX_INTERPEND 5
66 #define LEX_INTERPENDMAYBE 4
67 #define LEX_INTERPCONCAT 3
68 #define LEX_INTERPCONST 2
69 #define LEX_FORMLINE 1
70 #define LEX_KNOWNEXT 0
78 # define YYMAXLEVEL 100
80 YYSTYPE* yylval_pointer[YYMAXLEVEL];
81 int* yychar_pointer[YYMAXLEVEL];
85 # define yylval (*yylval_pointer[yyactlevel])
86 # define yychar (*yychar_pointer[yyactlevel])
87 # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]
89 # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel])
94 /* CLINE is a macro that ensures PL_copline has a sane value */
99 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
102 * Convenience functions to return different tokens and prime the
103 * lexer for the next token. They all take an argument.
105 * TOKEN : generic token (used for '(', DOLSHARP, etc)
106 * OPERATOR : generic operator
107 * AOPERATOR : assignment operator
108 * PREBLOCK : beginning the block after an if, while, foreach, ...
109 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
110 * PREREF : *EXPR where EXPR is not a simple identifier
111 * TERM : expression term
112 * LOOPX : loop exiting command (goto, last, dump, etc)
113 * FTST : file test operator
114 * FUN0 : zero-argument function
115 * FUN1 : not used, except for not, which isn't a UNIOP
116 * BOop : bitwise or or xor
118 * SHop : shift operator
119 * PWop : power operator
120 * PMop : pattern-matching operator
121 * Aop : addition-level operator
122 * Mop : multiplication-level operator
123 * Eop : equality-testing operator
124 * Rop : relational operator <= != gt
126 * Also see LOP and lop() below.
129 #define TOKEN(retval) return (PL_bufptr = s,(int)retval)
130 #define OPERATOR(retval) return (PL_expect = XTERM,PL_bufptr = s,(int)retval)
131 #define AOPERATOR(retval) return ao((PL_expect = XTERM,PL_bufptr = s,(int)retval))
132 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
133 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
134 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s,(int)retval)
135 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR,PL_bufptr = s,(int)retval)
136 #define LOOPX(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
137 #define FTST(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)UNIOP)
138 #define FUN0(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
139 #define FUN1(f) return(yylval.ival = f,PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
140 #define BOop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
141 #define BAop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
142 #define SHop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
143 #define PWop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
144 #define PMop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
145 #define Aop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
146 #define Mop(f) return ao((yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
147 #define Eop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
148 #define Rop(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)RELOP)
150 /* This bit of chicanery makes a unary function followed by
151 * a parenthesis into a function with one argument, highest precedence.
153 #define UNI(f) return(yylval.ival = f, \
156 PL_last_uni = PL_oldbufptr, \
157 PL_last_lop_op = f, \
158 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
160 #define UNIBRACK(f) return(yylval.ival = f, \
162 PL_last_uni = PL_oldbufptr, \
163 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
165 /* grandfather return to old style */
166 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
171 * This subroutine detects &&= and ||= and turns an ANDAND or OROR
172 * into an OP_ANDASSIGN or OP_ORASSIGN
176 S_ao(pTHX_ int toketype)
178 if (*PL_bufptr == '=') {
180 if (toketype == ANDAND)
181 yylval.ival = OP_ANDASSIGN;
182 else if (toketype == OROR)
183 yylval.ival = OP_ORASSIGN;
191 * When Perl expects an operator and finds something else, no_op
192 * prints the warning. It always prints "<something> found where
193 * operator expected. It prints "Missing semicolon on previous line?"
194 * if the surprise occurs at the start of the line. "do you need to
195 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
196 * where the compiler doesn't know if foo is a method call or a function.
197 * It prints "Missing operator before end of line" if there's nothing
198 * after the missing operator, or "... before <...>" if there is something
199 * after the missing operator.
203 S_no_op(pTHX_ char *what, char *s)
205 char *oldbp = PL_bufptr;
206 bool is_first = (PL_oldbufptr == PL_linestart);
212 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
214 Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n");
215 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
217 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
218 if (t < PL_bufptr && isSPACE(*t))
219 Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n",
220 t - PL_oldoldbufptr, PL_oldoldbufptr);
224 Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
231 * Complain about missing quote/regexp/heredoc terminator.
232 * If it's called with (char *)NULL then it cauterizes the line buffer.
233 * If we're in a delimited string and the delimiter is a control
234 * character, it's reformatted into a two-char sequence like ^C.
239 S_missingterm(pTHX_ char *s)
244 char *nl = strrchr(s,'\n');
250 iscntrl(PL_multi_close)
252 PL_multi_close < 32 || PL_multi_close == 127
256 tmpbuf[1] = toCTRL(PL_multi_close);
262 *tmpbuf = PL_multi_close;
266 q = strchr(s,'"') ? '\'' : '"';
267 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
275 Perl_deprecate(pTHX_ char *s)
277 if (ckWARN(WARN_DEPRECATED))
278 Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s);
283 * Deprecate a comma-less variable list.
289 deprecate("comma-less variable list");
293 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
294 * utf16-to-utf8-reversed.
297 #ifdef PERL_CR_FILTER
301 register char *s = SvPVX(sv);
302 register char *e = s + SvCUR(sv);
303 /* outer loop optimized to do nothing if there are no CR-LFs */
305 if (*s++ == '\r' && *s == '\n') {
306 /* hit a CR-LF, need to copy the rest */
307 register char *d = s - 1;
310 if (*s == '\r' && s[1] == '\n')
321 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
323 I32 count = FILTER_READ(idx+1, sv, maxlen);
324 if (count > 0 && !maxlen)
332 * Initialize variables. Uses the Perl save_stack to save its state (for
333 * recursive calls to the parser).
337 Perl_lex_start(pTHX_ SV *line)
342 SAVEI32(PL_lex_dojoin);
343 SAVEI32(PL_lex_brackets);
344 SAVEI32(PL_lex_casemods);
345 SAVEI32(PL_lex_starts);
346 SAVEI32(PL_lex_state);
347 SAVEVPTR(PL_lex_inpat);
348 SAVEI32(PL_lex_inwhat);
349 if (PL_lex_state == LEX_KNOWNEXT) {
350 I32 toke = PL_nexttoke;
351 while (--toke >= 0) {
352 SAVEI32(PL_nexttype[toke]);
353 SAVEVPTR(PL_nextval[toke]);
355 SAVEI32(PL_nexttoke);
357 SAVECOPLINE(PL_curcop);
360 SAVEPPTR(PL_oldbufptr);
361 SAVEPPTR(PL_oldoldbufptr);
362 SAVEPPTR(PL_linestart);
363 SAVESPTR(PL_linestr);
364 SAVEPPTR(PL_lex_brackstack);
365 SAVEPPTR(PL_lex_casestack);
366 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
367 SAVESPTR(PL_lex_stuff);
368 SAVEI32(PL_lex_defer);
369 SAVEI32(PL_sublex_info.sub_inwhat);
370 SAVESPTR(PL_lex_repl);
372 SAVEINT(PL_lex_expect);
374 PL_lex_state = LEX_NORMAL;
378 New(899, PL_lex_brackstack, 120, char);
379 New(899, PL_lex_casestack, 12, char);
380 SAVEFREEPV(PL_lex_brackstack);
381 SAVEFREEPV(PL_lex_casestack);
383 *PL_lex_casestack = '\0';
386 PL_lex_stuff = Nullsv;
387 PL_lex_repl = Nullsv;
391 PL_sublex_info.sub_inwhat = 0;
393 if (SvREADONLY(PL_linestr))
394 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
395 s = SvPV(PL_linestr, len);
396 if (len && s[len-1] != ';') {
397 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
398 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
399 sv_catpvn(PL_linestr, "\n;", 2);
401 SvTEMP_off(PL_linestr);
402 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
403 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
405 PL_rs = newSVpvn("\n", 1);
411 * Finalizer for lexing operations. Must be called when the parser is
412 * done with the lexer.
418 PL_doextract = FALSE;
423 * This subroutine has nothing to do with tilting, whether at windmills
424 * or pinball tables. Its name is short for "increment line". It
425 * increments the current line number in CopLINE(PL_curcop) and checks
426 * to see whether the line starts with a comment of the form
427 * # line 500 "foo.pm"
428 * If so, it sets the current line number and file to the values in the comment.
432 S_incline(pTHX_ char *s)
439 CopLINE_inc(PL_curcop);
442 while (SPACE_OR_TAB(*s)) s++;
443 if (strnEQ(s, "line", 4))
447 if (*s == ' ' || *s == '\t')
451 while (SPACE_OR_TAB(*s)) s++;
457 while (SPACE_OR_TAB(*s))
459 if (*s == '"' && (t = strchr(s+1, '"'))) {
464 for (t = s; !isSPACE(*t); t++) ;
467 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
469 if (*e != '\n' && *e != '\0')
470 return; /* false alarm */
476 Safefree(CopFILE(PL_curcop));
478 SvREFCNT_dec(CopFILEGV(PL_curcop));
480 CopFILE_set(PL_curcop, s);
483 CopLINE_set(PL_curcop, atoi(n)-1);
488 * Called to gobble the appropriate amount and type of whitespace.
489 * Skips comments as well.
493 S_skipspace(pTHX_ register char *s)
495 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
496 while (s < PL_bufend && SPACE_OR_TAB(*s))
502 SSize_t oldprevlen, oldoldprevlen;
503 SSize_t oldloplen, oldunilen;
504 while (s < PL_bufend && isSPACE(*s)) {
505 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
510 if (s < PL_bufend && *s == '#') {
511 while (s < PL_bufend && *s != '\n')
515 if (PL_in_eval && !PL_rsfp) {
522 /* only continue to recharge the buffer if we're at the end
523 * of the buffer, we're not reading from a source filter, and
524 * we're in normal lexing mode
526 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
527 PL_lex_state == LEX_FORMLINE)
530 /* try to recharge the buffer */
531 if ((s = filter_gets(PL_linestr, PL_rsfp,
532 (prevlen = SvCUR(PL_linestr)))) == Nullch)
534 /* end of file. Add on the -p or -n magic */
535 if (PL_minus_n || PL_minus_p) {
536 sv_setpv(PL_linestr,PL_minus_p ?
537 ";}continue{print or die qq(-p destination: $!\\n)" :
539 sv_catpv(PL_linestr,";}");
540 PL_minus_n = PL_minus_p = 0;
543 sv_setpv(PL_linestr,";");
545 /* reset variables for next time we lex */
546 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
548 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
550 /* Close the filehandle. Could be from -P preprocessor,
551 * STDIN, or a regular file. If we were reading code from
552 * STDIN (because the commandline held no -e or filename)
553 * then we don't close it, we reset it so the code can
554 * read from STDIN too.
557 if (PL_preprocess && !PL_in_eval)
558 (void)PerlProc_pclose(PL_rsfp);
559 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
560 PerlIO_clearerr(PL_rsfp);
562 (void)PerlIO_close(PL_rsfp);
567 /* not at end of file, so we only read another line */
568 /* make corresponding updates to old pointers, for yyerror() */
569 oldprevlen = PL_oldbufptr - PL_bufend;
570 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
572 oldunilen = PL_last_uni - PL_bufend;
574 oldloplen = PL_last_lop - PL_bufend;
575 PL_linestart = PL_bufptr = s + prevlen;
576 PL_bufend = s + SvCUR(PL_linestr);
578 PL_oldbufptr = s + oldprevlen;
579 PL_oldoldbufptr = s + oldoldprevlen;
581 PL_last_uni = s + oldunilen;
583 PL_last_lop = s + oldloplen;
586 /* debugger active and we're not compiling the debugger code,
587 * so store the line into the debugger's array of lines
589 if (PERLDB_LINE && PL_curstash != PL_debstash) {
590 SV *sv = NEWSV(85,0);
592 sv_upgrade(sv, SVt_PVMG);
593 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
594 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
601 * Check the unary operators to ensure there's no ambiguity in how they're
602 * used. An ambiguous piece of code would be:
604 * This doesn't mean rand() + 5. Because rand() is a unary operator,
605 * the +5 is its argument.
614 if (PL_oldoldbufptr != PL_last_uni)
616 while (isSPACE(*PL_last_uni))
618 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
619 if ((t = strchr(s, '(')) && t < PL_bufptr)
621 if (ckWARN_d(WARN_AMBIGUOUS)){
624 Perl_warner(aTHX_ WARN_AMBIGUOUS,
625 "Warning: Use of \"%s\" without parens is ambiguous",
631 /* workaround to replace the UNI() macro with a function. Only the
632 * hints/uts.sh file mentions this. Other comments elsewhere in the
633 * source indicate Microport Unix might need it too.
639 #define UNI(f) return uni(f,s)
642 S_uni(pTHX_ I32 f, char *s)
647 PL_last_uni = PL_oldbufptr;
658 #endif /* CRIPPLED_CC */
661 * LOP : macro to build a list operator. Its behaviour has been replaced
662 * with a subroutine, S_lop() for which LOP is just another name.
665 #define LOP(f,x) return lop(f,x,s)
669 * Build a list operator (or something that might be one). The rules:
670 * - if we have a next token, then it's a list operator [why?]
671 * - if the next thing is an opening paren, then it's a function
672 * - else it's a list operator
676 S_lop(pTHX_ I32 f, int x, char *s)
682 PL_last_lop = PL_oldbufptr;
697 * When the lexer realizes it knows the next token (for instance,
698 * it is reordering tokens for the parser) then it can call S_force_next
699 * to know what token to return the next time the lexer is called. Caller
700 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
701 * handles the token correctly.
705 S_force_next(pTHX_ I32 type)
707 PL_nexttype[PL_nexttoke] = type;
709 if (PL_lex_state != LEX_KNOWNEXT) {
710 PL_lex_defer = PL_lex_state;
711 PL_lex_expect = PL_expect;
712 PL_lex_state = LEX_KNOWNEXT;
718 * When the lexer knows the next thing is a word (for instance, it has
719 * just seen -> and it knows that the next char is a word char, then
720 * it calls S_force_word to stick the next word into the PL_next lookahead.
723 * char *start : buffer position (must be within PL_linestr)
724 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
725 * int check_keyword : if true, Perl checks to make sure the word isn't
726 * a keyword (do this if the word is a label, e.g. goto FOO)
727 * int allow_pack : if true, : characters will also be allowed (require,
729 * int allow_initial_tick : used by the "sub" lexer only.
733 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
738 start = skipspace(start);
740 if (isIDFIRST_lazy_if(s,UTF) ||
741 (allow_pack && *s == ':') ||
742 (allow_initial_tick && *s == '\'') )
744 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
745 if (check_keyword && keyword(PL_tokenbuf, len))
747 if (token == METHOD) {
752 PL_expect = XOPERATOR;
755 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
756 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
764 * Called when the lexer wants $foo *foo &foo etc, but the program
765 * text only contains the "foo" portion. The first argument is a pointer
766 * to the "foo", and the second argument is the type symbol to prefix.
767 * Forces the next token to be a "WORD".
768 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
772 S_force_ident(pTHX_ register char *s, int kind)
775 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
776 PL_nextval[PL_nexttoke].opval = o;
779 o->op_private = OPpCONST_ENTERED;
780 /* XXX see note in pp_entereval() for why we forgo typo
781 warnings if the symbol must be introduced in an eval.
783 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
784 kind == '$' ? SVt_PV :
785 kind == '@' ? SVt_PVAV :
786 kind == '%' ? SVt_PVHV :
794 Perl_str_to_version(pTHX_ SV *sv)
799 char *start = SvPVx(sv,len);
800 bool utf = SvUTF8(sv) ? TRUE : FALSE;
801 char *end = start + len;
802 while (start < end) {
806 n = utf8_to_uv((U8*)start, len, &skip, 0);
811 retval += ((NV)n)/nshift;
820 * Forces the next token to be a version number.
824 S_force_version(pTHX_ char *s)
826 OP *version = Nullop;
835 for (; isDIGIT(*d) || *d == '_' || *d == '.'; d++);
836 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
838 s = scan_num(s, &yylval);
839 version = yylval.opval;
840 ver = cSVOPx(version)->op_sv;
841 if (SvPOK(ver) && !SvNIOK(ver)) {
842 (void)SvUPGRADE(ver, SVt_PVNV);
843 SvNVX(ver) = str_to_version(ver);
844 SvNOK_on(ver); /* hint that it is a version */
849 /* NOTE: The parser sees the package name and the VERSION swapped */
850 PL_nextval[PL_nexttoke].opval = version;
858 * Tokenize a quoted string passed in as an SV. It finds the next
859 * chunk, up to end of string or a backslash. It may make a new
860 * SV containing that chunk (if HINT_NEW_STRING is on). It also
865 S_tokeq(pTHX_ SV *sv)
876 s = SvPV_force(sv, len);
877 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
880 while (s < send && *s != '\\')
885 if ( PL_hints & HINT_NEW_STRING )
886 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
889 if (s + 1 < send && (s[1] == '\\'))
890 s++; /* all that, just for this */
895 SvCUR_set(sv, d - SvPVX(sv));
897 if ( PL_hints & HINT_NEW_STRING )
898 return new_constant(NULL, 0, "q", sv, pv, "q");
903 * Now come three functions related to double-quote context,
904 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
905 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
906 * interact with PL_lex_state, and create fake ( ... ) argument lists
907 * to handle functions and concatenation.
908 * They assume that whoever calls them will be setting up a fake
909 * join call, because each subthing puts a ',' after it. This lets
912 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
914 * (I'm not sure whether the spurious commas at the end of lcfirst's
915 * arguments and join's arguments are created or not).
920 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
922 * Pattern matching will set PL_lex_op to the pattern-matching op to
923 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
925 * OP_CONST and OP_READLINE are easy--just make the new op and return.
927 * Everything else becomes a FUNC.
929 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
930 * had an OP_CONST or OP_READLINE). This just sets us up for a
931 * call to S_sublex_push().
937 register I32 op_type = yylval.ival;
939 if (op_type == OP_NULL) {
940 yylval.opval = PL_lex_op;
944 if (op_type == OP_CONST || op_type == OP_READLINE) {
945 SV *sv = tokeq(PL_lex_stuff);
947 if (SvTYPE(sv) == SVt_PVIV) {
948 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
954 nsv = newSVpvn(p, len);
960 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
961 PL_lex_stuff = Nullsv;
965 PL_sublex_info.super_state = PL_lex_state;
966 PL_sublex_info.sub_inwhat = op_type;
967 PL_sublex_info.sub_op = PL_lex_op;
968 PL_lex_state = LEX_INTERPPUSH;
972 yylval.opval = PL_lex_op;
982 * Create a new scope to save the lexing state. The scope will be
983 * ended in S_sublex_done. Returns a '(', starting the function arguments
984 * to the uc, lc, etc. found before.
985 * Sets PL_lex_state to LEX_INTERPCONCAT.
993 PL_lex_state = PL_sublex_info.super_state;
994 SAVEI32(PL_lex_dojoin);
995 SAVEI32(PL_lex_brackets);
996 SAVEI32(PL_lex_casemods);
997 SAVEI32(PL_lex_starts);
998 SAVEI32(PL_lex_state);
999 SAVEVPTR(PL_lex_inpat);
1000 SAVEI32(PL_lex_inwhat);
1001 SAVECOPLINE(PL_curcop);
1002 SAVEPPTR(PL_bufptr);
1003 SAVEPPTR(PL_oldbufptr);
1004 SAVEPPTR(PL_oldoldbufptr);
1005 SAVEPPTR(PL_linestart);
1006 SAVESPTR(PL_linestr);
1007 SAVEPPTR(PL_lex_brackstack);
1008 SAVEPPTR(PL_lex_casestack);
1010 PL_linestr = PL_lex_stuff;
1011 PL_lex_stuff = Nullsv;
1013 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1014 = SvPVX(PL_linestr);
1015 PL_bufend += SvCUR(PL_linestr);
1016 SAVEFREESV(PL_linestr);
1018 PL_lex_dojoin = FALSE;
1019 PL_lex_brackets = 0;
1020 New(899, PL_lex_brackstack, 120, char);
1021 New(899, PL_lex_casestack, 12, char);
1022 SAVEFREEPV(PL_lex_brackstack);
1023 SAVEFREEPV(PL_lex_casestack);
1024 PL_lex_casemods = 0;
1025 *PL_lex_casestack = '\0';
1027 PL_lex_state = LEX_INTERPCONCAT;
1028 CopLINE_set(PL_curcop, PL_multi_start);
1030 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1031 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1032 PL_lex_inpat = PL_sublex_info.sub_op;
1034 PL_lex_inpat = Nullop;
1041 * Restores lexer state after a S_sublex_push.
1047 if (!PL_lex_starts++) {
1048 PL_expect = XOPERATOR;
1049 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn("",0));
1053 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1054 PL_lex_state = LEX_INTERPCASEMOD;
1058 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1059 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1060 PL_linestr = PL_lex_repl;
1062 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1063 PL_bufend += SvCUR(PL_linestr);
1064 SAVEFREESV(PL_linestr);
1065 PL_lex_dojoin = FALSE;
1066 PL_lex_brackets = 0;
1067 PL_lex_casemods = 0;
1068 *PL_lex_casestack = '\0';
1070 if (SvEVALED(PL_lex_repl)) {
1071 PL_lex_state = LEX_INTERPNORMAL;
1073 /* we don't clear PL_lex_repl here, so that we can check later
1074 whether this is an evalled subst; that means we rely on the
1075 logic to ensure sublex_done() is called again only via the
1076 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1079 PL_lex_state = LEX_INTERPCONCAT;
1080 PL_lex_repl = Nullsv;
1086 PL_bufend = SvPVX(PL_linestr);
1087 PL_bufend += SvCUR(PL_linestr);
1088 PL_expect = XOPERATOR;
1089 PL_sublex_info.sub_inwhat = 0;
1097 Extracts a pattern, double-quoted string, or transliteration. This
1100 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1101 processing a pattern (PL_lex_inpat is true), a transliteration
1102 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1104 Returns a pointer to the character scanned up to. Iff this is
1105 advanced from the start pointer supplied (ie if anything was
1106 successfully parsed), will leave an OP for the substring scanned
1107 in yylval. Caller must intuit reason for not parsing further
1108 by looking at the next characters herself.
1112 double-quoted style: \r and \n
1113 regexp special ones: \D \s
1115 backrefs: \1 (deprecated in substitution replacements)
1116 case and quoting: \U \Q \E
1117 stops on @ and $, but not for $ as tail anchor
1119 In transliterations:
1120 characters are VERY literal, except for - not at the start or end
1121 of the string, which indicates a range. scan_const expands the
1122 range to the full set of intermediate characters.
1124 In double-quoted strings:
1126 double-quoted style: \r and \n
1128 backrefs: \1 (deprecated)
1129 case and quoting: \U \Q \E
1132 scan_const does *not* construct ops to handle interpolated strings.
1133 It stops processing as soon as it finds an embedded $ or @ variable
1134 and leaves it to the caller to work out what's going on.
1136 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo.
1138 $ in pattern could be $foo or could be tail anchor. Assumption:
1139 it's a tail anchor if $ is the last thing in the string, or if it's
1140 followed by one of ")| \n\t"
1142 \1 (backreferences) are turned into $1
1144 The structure of the code is
1145 while (there's a character to process) {
1146 handle transliteration ranges
1147 skip regexp comments
1148 skip # initiated comments in //x patterns
1149 check for embedded @foo
1150 check for embedded scalars
1152 leave intact backslashes from leave (below)
1153 deprecate \1 in strings and sub replacements
1154 handle string-changing backslashes \l \U \Q \E, etc.
1155 switch (what was escaped) {
1156 handle - in a transliteration (becomes a literal -)
1157 handle \132 octal characters
1158 handle 0x15 hex characters
1159 handle \cV (control V)
1160 handle printf backslashes (\f, \r, \n, etc)
1162 } (end if backslash)
1163 } (end while character to read)
1168 S_scan_const(pTHX_ char *start)
1170 register char *send = PL_bufend; /* end of the constant */
1171 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1172 register char *s = start; /* start of the constant */
1173 register char *d = SvPVX(sv); /* destination for copies */
1174 bool dorange = FALSE; /* are we in a translit range? */
1175 bool didrange = FALSE; /* did we just finish a range? */
1176 bool has_utf8 = FALSE; /* embedded \x{} */
1179 I32 utf = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1180 ? (PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF))
1182 I32 this_utf8 = (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op)
1183 ? (PL_sublex_info.sub_op->op_private & (PL_lex_repl ?
1184 OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF))
1186 const char *leaveit = /* set of acceptably-backslashed characters */
1188 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#"
1191 while (s < send || dorange) {
1192 /* get transliterations out of the way (they're most literal) */
1193 if (PL_lex_inwhat == OP_TRANS) {
1194 /* expand a range A-Z to the full set of characters. AIE! */
1196 I32 i; /* current expanded character */
1197 I32 min; /* first character in range */
1198 I32 max; /* last character in range */
1200 i = d - SvPVX(sv); /* remember current offset */
1201 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1202 d = SvPVX(sv) + i; /* refresh d after realloc */
1203 d -= 2; /* eat the first char and the - */
1205 min = (U8)*d; /* first char in range */
1206 max = (U8)d[1]; /* last char in range */
1210 "Invalid [] range \"%c-%c\" in transliteration operator",
1211 (char)min, (char)max);
1215 if ((isLOWER(min) && isLOWER(max)) ||
1216 (isUPPER(min) && isUPPER(max))) {
1218 for (i = min; i <= max; i++)
1222 for (i = min; i <= max; i++)
1229 for (i = min; i <= max; i++)
1232 /* mark the range as done, and continue */
1238 /* range begins (ignore - as first or last char) */
1239 else if (*s == '-' && s+1 < send && s != start) {
1241 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1244 *d++ = (char)0xff; /* use illegal utf8 byte--see pmtrans */
1256 /* if we get here, we're not doing a transliteration */
1258 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1259 except for the last char, which will be done separately. */
1260 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1262 while (s < send && *s != ')')
1265 else if (s[2] == '{' /* This should match regcomp.c */
1266 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1269 char *regparse = s + (s[2] == '{' ? 3 : 4);
1272 while (count && (c = *regparse)) {
1273 if (c == '\\' && regparse[1])
1281 if (*regparse != ')') {
1282 regparse--; /* Leave one char for continuation. */
1283 yyerror("Sequence (?{...}) not terminated or not {}-balanced");
1285 while (s < regparse)
1290 /* likewise skip #-initiated comments in //x patterns */
1291 else if (*s == '#' && PL_lex_inpat &&
1292 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1293 while (s+1 < send && *s != '\n')
1297 /* check for embedded arrays
1298 (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-)
1300 else if (*s == '@' && s[1]
1301 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1304 /* check for embedded scalars. only stop if we're sure it's a
1307 else if (*s == '$') {
1308 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1310 if (s + 1 < send && !strchr("()| \n\t", s[1]))
1311 break; /* in regexp, $ might be tail anchor */
1315 if (*s == '\\' && s+1 < send) {
1316 bool to_be_utf8 = FALSE;
1320 /* some backslashes we leave behind */
1321 if (*leaveit && *s && strchr(leaveit, *s)) {
1327 /* deprecate \1 in strings and substitution replacements */
1328 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1329 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1331 if (ckWARN(WARN_SYNTAX))
1332 Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s);
1337 /* string-change backslash escapes */
1338 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1343 /* if we get here, it's either a quoted -, or a digit */
1346 /* quoted - in transliterations */
1348 if (PL_lex_inwhat == OP_TRANS) {
1355 if (ckWARN(WARN_MISC) && isALNUM(*s))
1356 Perl_warner(aTHX_ WARN_MISC,
1357 "Unrecognized escape \\%c passed through",
1359 /* default action is to copy the quoted character */
1364 /* \132 indicates an octal constant */
1365 case '0': case '1': case '2': case '3':
1366 case '4': case '5': case '6': case '7':
1368 STRLEN len = 0; /* disallow underscores */
1369 uv = (UV)scan_oct(s, 3, &len);
1372 goto NUM_ESCAPE_INSERT;
1374 /* \x24 indicates a hex constant */
1378 char* e = strchr(s, '}');
1380 yyerror("Missing right brace on \\x{}");
1384 STRLEN len = 1; /* allow underscores */
1385 uv = (UV)scan_hex(s + 1, e - s - 1, &len);
1386 if (PL_hints & HINT_UTF8)
1393 STRLEN len = 0; /* disallow underscores */
1394 uv = (UV)scan_hex(s, 2, &len);
1400 /* Insert oct or hex escaped character.
1401 * There will always enough room in sv since such
1402 * escapes will be longer than any UT-F8 sequence
1403 * they can end up as. */
1405 /* This spot is wrong for EBCDIC. Characters like
1406 * the lowercase letters and digits are >127 in EBCDIC,
1407 * so here they would need to be mapped to the Unicode
1408 * repertoire. --jhi */
1411 if (!has_utf8 && (to_be_utf8 || uv > 255)) {
1412 /* Might need to recode whatever we have
1413 * accumulated so far if it contains any
1416 * (Can't we keep track of that and avoid
1417 * this rescan? --jhi)
1422 for (c = SvPVX(sv); c < d; c++) {
1423 if (UTF8_IS_CONTINUED(*c))
1427 char *old_pvx = SvPVX(sv);
1431 SvCUR(sv) + hicount + 1) +
1439 if (UTF8_IS_CONTINUED(*src)) {
1440 *dst-- = UTF8_EIGHT_BIT_LO(*src);
1441 *dst-- = UTF8_EIGHT_BIT_HI(*src--);
1450 if (to_be_utf8 || has_utf8 || uv > 255) {
1451 d = (char*)uv_to_utf8((U8*)d, uv);
1463 /* \N{latin small letter a} is a named character */
1467 char* e = strchr(s, '}');
1473 yyerror("Missing right brace on \\N{}");
1477 res = newSVpvn(s + 1, e - s - 1);
1478 res = new_constant( Nullch, 0, "charnames",
1479 res, Nullsv, "\\N{...}" );
1480 str = SvPV(res,len);
1481 if (!has_utf8 && SvUTF8(res)) {
1482 char *ostart = SvPVX(sv);
1483 SvCUR_set(sv, d - ostart);
1486 sv_utf8_upgrade(sv);
1487 /* this just broke our allocation above... */
1488 SvGROW(sv, send - start);
1489 d = SvPVX(sv) + SvCUR(sv);
1492 if (len > e - s + 4) {
1493 char *odest = SvPVX(sv);
1495 SvGROW(sv, (SvCUR(sv) + len - (e - s + 4)));
1496 d = SvPVX(sv) + (d - odest);
1498 Copy(str, d, len, char);
1505 yyerror("Missing braces on \\N{}");
1508 /* \c is a control character */
1525 /* printf-style backslashes, formfeeds, newlines, etc */
1543 *d++ = '\047'; /* CP 1047 */
1546 *d++ = '\057'; /* CP 1047 */
1560 } /* end if (backslash) */
1562 /* (now in tr/// code again) */
1564 if (UTF8_IS_CONTINUED(*s) && (this_utf8 || has_utf8)) {
1565 STRLEN len = (STRLEN) -1;
1568 uv = utf8_to_uv((U8*)s, send - s, &len, 0);
1570 if (len == (STRLEN)-1) {
1571 /* Illegal UTF8 (a high-bit byte), make it valid. */
1572 char *old_pvx = SvPVX(sv);
1573 /* need space for one extra char (NOTE: SvCUR() not set here) */
1574 d = SvGROW(sv, SvLEN(sv) + 1) + (d - old_pvx);
1575 d = (char*)uv_to_utf8((U8*)d, (U8)*s++);
1586 } /* while loop to process each character */
1588 /* terminate the string and set up the sv */
1590 SvCUR_set(sv, d - SvPVX(sv));
1595 /* shrink the sv if we allocated more than we used */
1596 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1597 SvLEN_set(sv, SvCUR(sv) + 1);
1598 Renew(SvPVX(sv), SvLEN(sv), char);
1601 /* return the substring (via yylval) only if we parsed anything */
1602 if (s > PL_bufptr) {
1603 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1604 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1606 ( PL_lex_inwhat == OP_TRANS
1608 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1611 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1618 * Returns TRUE if there's more to the expression (e.g., a subscript),
1621 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1623 * ->[ and ->{ return TRUE
1624 * { and [ outside a pattern are always subscripts, so return TRUE
1625 * if we're outside a pattern and it's not { or [, then return FALSE
1626 * if we're in a pattern and the first char is a {
1627 * {4,5} (any digits around the comma) returns FALSE
1628 * if we're in a pattern and the first char is a [
1630 * [SOMETHING] has a funky algorithm to decide whether it's a
1631 * character class or not. It has to deal with things like
1632 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1633 * anything else returns TRUE
1636 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1639 S_intuit_more(pTHX_ register char *s)
1641 if (PL_lex_brackets)
1643 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1645 if (*s != '{' && *s != '[')
1650 /* In a pattern, so maybe we have {n,m}. */
1667 /* On the other hand, maybe we have a character class */
1670 if (*s == ']' || *s == '^')
1673 /* this is terrifying, and it works */
1674 int weight = 2; /* let's weigh the evidence */
1676 unsigned char un_char = 255, last_un_char;
1677 char *send = strchr(s,']');
1678 char tmpbuf[sizeof PL_tokenbuf * 4];
1680 if (!send) /* has to be an expression */
1683 Zero(seen,256,char);
1686 else if (isDIGIT(*s)) {
1688 if (isDIGIT(s[1]) && s[2] == ']')
1694 for (; s < send; s++) {
1695 last_un_char = un_char;
1696 un_char = (unsigned char)*s;
1701 weight -= seen[un_char] * 10;
1702 if (isALNUM_lazy_if(s+1,UTF)) {
1703 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1704 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1709 else if (*s == '$' && s[1] &&
1710 strchr("[#!%*<>()-=",s[1])) {
1711 if (/*{*/ strchr("])} =",s[2]))
1720 if (strchr("wds]",s[1]))
1722 else if (seen['\''] || seen['"'])
1724 else if (strchr("rnftbxcav",s[1]))
1726 else if (isDIGIT(s[1])) {
1728 while (s[1] && isDIGIT(s[1]))
1738 if (strchr("aA01! ",last_un_char))
1740 if (strchr("zZ79~",s[1]))
1742 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1743 weight -= 5; /* cope with negative subscript */
1746 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1747 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1752 if (keyword(tmpbuf, d - tmpbuf))
1755 if (un_char == last_un_char + 1)
1757 weight -= seen[un_char];
1762 if (weight >= 0) /* probably a character class */
1772 * Does all the checking to disambiguate
1774 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1775 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1777 * First argument is the stuff after the first token, e.g. "bar".
1779 * Not a method if bar is a filehandle.
1780 * Not a method if foo is a subroutine prototyped to take a filehandle.
1781 * Not a method if it's really "Foo $bar"
1782 * Method if it's "foo $bar"
1783 * Not a method if it's really "print foo $bar"
1784 * Method if it's really "foo package::" (interpreted as package->foo)
1785 * Not a method if bar is known to be a subroutne ("sub bar; foo bar")
1786 * Not a method if bar is a filehandle or package, but is quoted with
1791 S_intuit_method(pTHX_ char *start, GV *gv)
1793 char *s = start + (*start == '$');
1794 char tmpbuf[sizeof PL_tokenbuf];
1802 if ((cv = GvCVu(gv))) {
1803 char *proto = SvPVX(cv);
1813 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1814 /* start is the beginning of the possible filehandle/object,
1815 * and s is the end of it
1816 * tmpbuf is a copy of it
1819 if (*start == '$') {
1820 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1825 return *s == '(' ? FUNCMETH : METHOD;
1827 if (!keyword(tmpbuf, len)) {
1828 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1833 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1834 if (indirgv && GvCVu(indirgv))
1836 /* filehandle or package name makes it a method */
1837 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1839 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1840 return 0; /* no assumptions -- "=>" quotes bearword */
1842 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1843 newSVpvn(tmpbuf,len));
1844 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1848 return *s == '(' ? FUNCMETH : METHOD;
1856 * Return a string of Perl code to load the debugger. If PERL5DB
1857 * is set, it will return the contents of that, otherwise a
1858 * compile-time require of perl5db.pl.
1865 char *pdb = PerlEnv_getenv("PERL5DB");
1869 SETERRNO(0,SS$_NORMAL);
1870 return "BEGIN { require 'perl5db.pl' }";
1876 /* Encoded script support. filter_add() effectively inserts a
1877 * 'pre-processing' function into the current source input stream.
1878 * Note that the filter function only applies to the current source file
1879 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1881 * The datasv parameter (which may be NULL) can be used to pass
1882 * private data to this instance of the filter. The filter function
1883 * can recover the SV using the FILTER_DATA macro and use it to
1884 * store private buffers and state information.
1886 * The supplied datasv parameter is upgraded to a PVIO type
1887 * and the IoDIRP/IoANY field is used to store the function pointer,
1888 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1889 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1890 * private use must be set using malloc'd pointers.
1894 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1899 if (!PL_rsfp_filters)
1900 PL_rsfp_filters = newAV();
1902 datasv = NEWSV(255,0);
1903 if (!SvUPGRADE(datasv, SVt_PVIO))
1904 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1905 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1906 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1907 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1908 funcp, SvPV_nolen(datasv)));
1909 av_unshift(PL_rsfp_filters, 1);
1910 av_store(PL_rsfp_filters, 0, datasv) ;
1915 /* Delete most recently added instance of this filter function. */
1917 Perl_filter_del(pTHX_ filter_t funcp)
1920 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp));
1921 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
1923 /* if filter is on top of stack (usual case) just pop it off */
1924 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
1925 if (IoANY(datasv) == (void *)funcp) {
1926 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
1927 IoANY(datasv) = (void *)NULL;
1928 sv_free(av_pop(PL_rsfp_filters));
1932 /* we need to search for the correct entry and clear it */
1933 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
1937 /* Invoke the n'th filter function for the current rsfp. */
1939 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
1942 /* 0 = read one text line */
1947 if (!PL_rsfp_filters)
1949 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
1950 /* Provide a default input filter to make life easy. */
1951 /* Note that we append to the line. This is handy. */
1952 DEBUG_P(PerlIO_printf(Perl_debug_log,
1953 "filter_read %d: from rsfp\n", idx));
1957 int old_len = SvCUR(buf_sv) ;
1959 /* ensure buf_sv is large enough */
1960 SvGROW(buf_sv, old_len + maxlen) ;
1961 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
1962 if (PerlIO_error(PL_rsfp))
1963 return -1; /* error */
1965 return 0 ; /* end of file */
1967 SvCUR_set(buf_sv, old_len + len) ;
1970 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
1971 if (PerlIO_error(PL_rsfp))
1972 return -1; /* error */
1974 return 0 ; /* end of file */
1977 return SvCUR(buf_sv);
1979 /* Skip this filter slot if filter has been deleted */
1980 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
1981 DEBUG_P(PerlIO_printf(Perl_debug_log,
1982 "filter_read %d: skipped (filter deleted)\n",
1984 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
1986 /* Get function pointer hidden within datasv */
1987 funcp = (filter_t)IoANY(datasv);
1988 DEBUG_P(PerlIO_printf(Perl_debug_log,
1989 "filter_read %d: via function %p (%s)\n",
1990 idx, funcp, SvPV_nolen(datasv)));
1991 /* Call function. The function is expected to */
1992 /* call "FILTER_READ(idx+1, buf_sv)" first. */
1993 /* Return: <0:error, =0:eof, >0:not eof */
1994 return (*funcp)(aTHXo_ idx, buf_sv, maxlen);
1998 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2000 #ifdef PERL_CR_FILTER
2001 if (!PL_rsfp_filters) {
2002 filter_add(S_cr_textfilter,NULL);
2005 if (PL_rsfp_filters) {
2008 SvCUR_set(sv, 0); /* start with empty line */
2009 if (FILTER_READ(0, sv, 0) > 0)
2010 return ( SvPVX(sv) ) ;
2015 return (sv_gets(sv, fp, append));
2019 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2023 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2027 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2028 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2030 return GvHV(gv); /* Foo:: */
2033 /* use constant CLASS => 'MyClass' */
2034 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2036 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2037 pkgname = SvPV_nolen(sv);
2041 return gv_stashpv(pkgname, FALSE);
2045 static char* exp_name[] =
2046 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2047 "ATTRTERM", "TERMBLOCK"
2054 Works out what to call the token just pulled out of the input
2055 stream. The yacc parser takes care of taking the ops we return and
2056 stitching them into a tree.
2062 if read an identifier
2063 if we're in a my declaration
2064 croak if they tried to say my($foo::bar)
2065 build the ops for a my() declaration
2066 if it's an access to a my() variable
2067 are we in a sort block?
2068 croak if my($a); $a <=> $b
2069 build ops for access to a my() variable
2070 if in a dq string, and they've said @foo and we can't find @foo
2072 build ops for a bareword
2073 if we already built the token before, use it.
2076 #ifdef USE_PURE_BISON
2078 #pragma segment Perl_yylex_r
2081 Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp)
2086 yylval_pointer[yyactlevel] = lvalp;
2087 yychar_pointer[yyactlevel] = lcharp;
2088 if (yyactlevel >= YYMAXLEVEL)
2089 Perl_croak(aTHX_ "panic: YYMAXLEVEL");
2091 r = Perl_yylex(aTHX);
2100 #pragma segment Perl_yylex
2113 /* check if there's an identifier for us to look at */
2114 if (PL_pending_ident) {
2115 /* pit holds the identifier we read and pending_ident is reset */
2116 char pit = PL_pending_ident;
2117 PL_pending_ident = 0;
2119 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2120 "### Tokener saw identifier '%s'\n", PL_tokenbuf); })
2122 /* if we're in a my(), we can't allow dynamics here.
2123 $foo'bar has already been turned into $foo::bar, so
2124 just check for colons.
2126 if it's a legal name, the OP is a PADANY.
2129 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
2130 if (strchr(PL_tokenbuf,':'))
2131 yyerror(Perl_form(aTHX_ "No package name allowed for "
2132 "variable %s in \"our\"",
2134 tmp = pad_allocmy(PL_tokenbuf);
2137 if (strchr(PL_tokenbuf,':'))
2138 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
2140 yylval.opval = newOP(OP_PADANY, 0);
2141 yylval.opval->op_targ = pad_allocmy(PL_tokenbuf);
2147 build the ops for accesses to a my() variable.
2149 Deny my($a) or my($b) in a sort block, *if* $a or $b is
2150 then used in a comparison. This catches most, but not
2151 all cases. For instance, it catches
2152 sort { my($a); $a <=> $b }
2154 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
2155 (although why you'd do that is anyone's guess).
2158 if (!strchr(PL_tokenbuf,':')) {
2160 /* Check for single character per-thread SVs */
2161 if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0'
2162 && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */
2163 && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD)
2165 yylval.opval = newOP(OP_THREADSV, 0);
2166 yylval.opval->op_targ = tmp;
2169 #endif /* USE_THREADS */
2170 if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) {
2171 SV *namesv = AvARRAY(PL_comppad_name)[tmp];
2172 /* might be an "our" variable" */
2173 if (SvFLAGS(namesv) & SVpad_OUR) {
2174 /* build ops for a bareword */
2175 SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0);
2176 sv_catpvn(sym, "::", 2);
2177 sv_catpv(sym, PL_tokenbuf+1);
2178 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
2179 yylval.opval->op_private = OPpCONST_ENTERED;
2180 gv_fetchpv(SvPVX(sym),
2182 ? (GV_ADDMULTI | GV_ADDINEVAL)
2185 ((PL_tokenbuf[0] == '$') ? SVt_PV
2186 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2191 /* if it's a sort block and they're naming $a or $b */
2192 if (PL_last_lop_op == OP_SORT &&
2193 PL_tokenbuf[0] == '$' &&
2194 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
2197 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
2198 d < PL_bufend && *d != '\n';
2201 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
2202 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
2208 yylval.opval = newOP(OP_PADANY, 0);
2209 yylval.opval->op_targ = tmp;
2215 Whine if they've said @foo in a doublequoted string,
2216 and @foo isn't a variable we can find in the symbol
2219 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
2220 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
2221 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
2222 && ckWARN(WARN_AMBIGUOUS))
2224 /* Downgraded from fatal to warning 20000522 mjd */
2225 Perl_warner(aTHX_ WARN_AMBIGUOUS,
2226 "Possible unintended interpolation of %s in string",
2231 /* build ops for a bareword */
2232 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
2233 yylval.opval->op_private = OPpCONST_ENTERED;
2234 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
2235 ((PL_tokenbuf[0] == '$') ? SVt_PV
2236 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
2241 /* no identifier pending identification */
2243 switch (PL_lex_state) {
2245 case LEX_NORMAL: /* Some compilers will produce faster */
2246 case LEX_INTERPNORMAL: /* code if we comment these out. */
2250 /* when we've already built the next token, just pull it out of the queue */
2253 yylval = PL_nextval[PL_nexttoke];
2255 PL_lex_state = PL_lex_defer;
2256 PL_expect = PL_lex_expect;
2257 PL_lex_defer = LEX_NORMAL;
2259 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2260 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2261 (IV)PL_nexttype[PL_nexttoke]); })
2263 return(PL_nexttype[PL_nexttoke]);
2265 /* interpolated case modifiers like \L \U, including \Q and \E.
2266 when we get here, PL_bufptr is at the \
2268 case LEX_INTERPCASEMOD:
2270 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2271 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2273 /* handle \E or end of string */
2274 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2278 if (PL_lex_casemods) {
2279 oldmod = PL_lex_casestack[--PL_lex_casemods];
2280 PL_lex_casestack[PL_lex_casemods] = '\0';
2282 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2284 PL_lex_state = LEX_INTERPCONCAT;
2288 if (PL_bufptr != PL_bufend)
2290 PL_lex_state = LEX_INTERPCONCAT;
2294 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2295 "### Saw case modifier at '%s'\n", PL_bufptr); })
2297 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2298 tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */
2299 if (strchr("LU", *s) &&
2300 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U')))
2302 PL_lex_casestack[--PL_lex_casemods] = '\0';
2305 if (PL_lex_casemods > 10) {
2306 char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2307 if (newlb != PL_lex_casestack) {
2309 PL_lex_casestack = newlb;
2312 PL_lex_casestack[PL_lex_casemods++] = *s;
2313 PL_lex_casestack[PL_lex_casemods] = '\0';
2314 PL_lex_state = LEX_INTERPCONCAT;
2315 PL_nextval[PL_nexttoke].ival = 0;
2318 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2320 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2322 PL_nextval[PL_nexttoke].ival = OP_LC;
2324 PL_nextval[PL_nexttoke].ival = OP_UC;
2326 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2328 Perl_croak(aTHX_ "panic: yylex");
2331 if (PL_lex_starts) {
2340 case LEX_INTERPPUSH:
2341 return sublex_push();
2343 case LEX_INTERPSTART:
2344 if (PL_bufptr == PL_bufend)
2345 return sublex_done();
2346 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2347 "### Interpolated variable at '%s'\n", PL_bufptr); })
2349 PL_lex_dojoin = (*PL_bufptr == '@');
2350 PL_lex_state = LEX_INTERPNORMAL;
2351 if (PL_lex_dojoin) {
2352 PL_nextval[PL_nexttoke].ival = 0;
2355 PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0);
2356 PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\"");
2357 force_next(PRIVATEREF);
2359 force_ident("\"", '$');
2360 #endif /* USE_THREADS */
2361 PL_nextval[PL_nexttoke].ival = 0;
2363 PL_nextval[PL_nexttoke].ival = 0;
2365 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2368 if (PL_lex_starts++) {
2374 case LEX_INTERPENDMAYBE:
2375 if (intuit_more(PL_bufptr)) {
2376 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2382 if (PL_lex_dojoin) {
2383 PL_lex_dojoin = FALSE;
2384 PL_lex_state = LEX_INTERPCONCAT;
2387 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2388 && SvEVALED(PL_lex_repl))
2390 if (PL_bufptr != PL_bufend)
2391 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2392 PL_lex_repl = Nullsv;
2395 case LEX_INTERPCONCAT:
2397 if (PL_lex_brackets)
2398 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2400 if (PL_bufptr == PL_bufend)
2401 return sublex_done();
2403 if (SvIVX(PL_linestr) == '\'') {
2404 SV *sv = newSVsv(PL_linestr);
2407 else if ( PL_hints & HINT_NEW_RE )
2408 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2409 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2413 s = scan_const(PL_bufptr);
2415 PL_lex_state = LEX_INTERPCASEMOD;
2417 PL_lex_state = LEX_INTERPSTART;
2420 if (s != PL_bufptr) {
2421 PL_nextval[PL_nexttoke] = yylval;
2424 if (PL_lex_starts++)
2434 PL_lex_state = LEX_NORMAL;
2435 s = scan_formline(PL_bufptr);
2436 if (!PL_lex_formbrack)
2442 PL_oldoldbufptr = PL_oldbufptr;
2445 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2446 exp_name[PL_expect], s);
2452 if (isIDFIRST_lazy_if(s,UTF))
2454 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2457 goto fake_eof; /* emulate EOF on ^D or ^Z */
2462 if (PL_lex_brackets)
2463 yyerror("Missing right curly or square bracket");
2464 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2465 "### Tokener got EOF\n");
2469 if (s++ < PL_bufend)
2470 goto retry; /* ignore stray nulls */
2473 if (!PL_in_eval && !PL_preambled) {
2474 PL_preambled = TRUE;
2475 sv_setpv(PL_linestr,incl_perldb());
2476 if (SvCUR(PL_linestr))
2477 sv_catpv(PL_linestr,";");
2479 while(AvFILLp(PL_preambleav) >= 0) {
2480 SV *tmpsv = av_shift(PL_preambleav);
2481 sv_catsv(PL_linestr, tmpsv);
2482 sv_catpv(PL_linestr, ";");
2485 sv_free((SV*)PL_preambleav);
2486 PL_preambleav = NULL;
2488 if (PL_minus_n || PL_minus_p) {
2489 sv_catpv(PL_linestr, "LINE: while (<>) {");
2491 sv_catpv(PL_linestr,"chomp;");
2493 GV* gv = gv_fetchpv("::F", TRUE, SVt_PVAV);
2495 GvIMPORTED_AV_on(gv);
2497 if (strchr("/'\"", *PL_splitstr)
2498 && strchr(PL_splitstr + 1, *PL_splitstr))
2499 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr);
2502 s = "'~#\200\1'"; /* surely one char is unused...*/
2503 while (s[1] && strchr(PL_splitstr, *s)) s++;
2505 Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s%c",
2506 "q" + (delim == '\''), delim);
2507 for (s = PL_splitstr; *s; s++) {
2509 sv_catpvn(PL_linestr, "\\", 1);
2510 sv_catpvn(PL_linestr, s, 1);
2512 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2516 sv_catpv(PL_linestr,"@F=split(' ');");
2519 sv_catpv(PL_linestr, "\n");
2520 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2521 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2522 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2523 SV *sv = NEWSV(85,0);
2525 sv_upgrade(sv, SVt_PVMG);
2526 sv_setsv(sv,PL_linestr);
2527 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2532 bof = PL_rsfp ? TRUE : FALSE;
2534 #ifdef PERLIO_IS_STDIO
2535 # ifdef __GNU_LIBRARY__
2536 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2537 # define FTELL_FOR_PIPE_IS_BROKEN
2541 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2542 # define FTELL_FOR_PIPE_IS_BROKEN
2547 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2548 /* This loses the possibility to detect the bof
2549 * situation on perl -P when the libc5 is being used.
2550 * Workaround? Maybe attach some extra state to PL_rsfp?
2553 bof = PerlIO_tell(PL_rsfp) == 0;
2555 bof = PerlIO_tell(PL_rsfp) == 0;
2558 s = filter_gets(PL_linestr, PL_rsfp, 0);
2562 if (PL_preprocess && !PL_in_eval)
2563 (void)PerlProc_pclose(PL_rsfp);
2564 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2565 PerlIO_clearerr(PL_rsfp);
2567 (void)PerlIO_close(PL_rsfp);
2569 PL_doextract = FALSE;
2571 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2572 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2573 sv_catpv(PL_linestr,";}");
2574 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2575 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2576 PL_minus_n = PL_minus_p = 0;
2579 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2580 sv_setpv(PL_linestr,"");
2581 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2583 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2584 s = swallow_bom((U8*)s);
2587 if (*s == '#' && s[1] == '!' && instr(s,"perl"))
2588 PL_doextract = FALSE;
2590 /* Incest with pod. */
2591 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2592 sv_setpv(PL_linestr, "");
2593 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2594 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2595 PL_doextract = FALSE;
2599 } while (PL_doextract);
2600 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2601 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2602 SV *sv = NEWSV(85,0);
2604 sv_upgrade(sv, SVt_PVMG);
2605 sv_setsv(sv,PL_linestr);
2606 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2608 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2609 if (CopLINE(PL_curcop) == 1) {
2610 while (s < PL_bufend && isSPACE(*s))
2612 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2616 if (*s == '#' && *(s+1) == '!')
2618 #ifdef ALTERNATE_SHEBANG
2620 static char as[] = ALTERNATE_SHEBANG;
2621 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2622 d = s + (sizeof(as) - 1);
2624 #endif /* ALTERNATE_SHEBANG */
2633 while (*d && !isSPACE(*d))
2637 #ifdef ARG_ZERO_IS_SCRIPT
2638 if (ipathend > ipath) {
2640 * HP-UX (at least) sets argv[0] to the script name,
2641 * which makes $^X incorrect. And Digital UNIX and Linux,
2642 * at least, set argv[0] to the basename of the Perl
2643 * interpreter. So, having found "#!", we'll set it right.
2645 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV));
2646 assert(SvPOK(x) || SvGMAGICAL(x));
2647 if (sv_eq(x, CopFILESV(PL_curcop))) {
2648 sv_setpvn(x, ipath, ipathend - ipath);
2651 TAINT_NOT; /* $^X is always tainted, but that's OK */
2653 #endif /* ARG_ZERO_IS_SCRIPT */
2658 d = instr(s,"perl -");
2660 d = instr(s,"perl");
2662 /* avoid getting into infinite loops when shebang
2663 * line contains "Perl" rather than "perl" */
2665 for (d = ipathend-4; d >= ipath; --d) {
2666 if ((*d == 'p' || *d == 'P')
2667 && !ibcmp(d, "perl", 4))
2677 #ifdef ALTERNATE_SHEBANG
2679 * If the ALTERNATE_SHEBANG on this system starts with a
2680 * character that can be part of a Perl expression, then if
2681 * we see it but not "perl", we're probably looking at the
2682 * start of Perl code, not a request to hand off to some
2683 * other interpreter. Similarly, if "perl" is there, but
2684 * not in the first 'word' of the line, we assume the line
2685 * contains the start of the Perl program.
2687 if (d && *s != '#') {
2689 while (*c && !strchr("; \t\r\n\f\v#", *c))
2692 d = Nullch; /* "perl" not in first word; ignore */
2694 *s = '#'; /* Don't try to parse shebang line */
2696 #endif /* ALTERNATE_SHEBANG */
2697 #ifndef MACOS_TRADITIONAL
2702 !instr(s,"indir") &&
2703 instr(PL_origargv[0],"perl"))
2709 while (s < PL_bufend && isSPACE(*s))
2711 if (s < PL_bufend) {
2712 Newz(899,newargv,PL_origargc+3,char*);
2714 while (s < PL_bufend && !isSPACE(*s))
2717 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2720 newargv = PL_origargv;
2722 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2723 Perl_croak(aTHX_ "Can't exec %s", ipath);
2727 U32 oldpdb = PL_perldb;
2728 bool oldn = PL_minus_n;
2729 bool oldp = PL_minus_p;
2731 while (*d && !isSPACE(*d)) d++;
2732 while (SPACE_OR_TAB(*d)) d++;
2736 if (*d == 'M' || *d == 'm') {
2738 while (*d && !isSPACE(*d)) d++;
2739 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2742 d = moreswitches(d);
2744 if ((PERLDB_LINE && !oldpdb) ||
2745 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2746 /* if we have already added "LINE: while (<>) {",
2747 we must not do it again */
2749 sv_setpv(PL_linestr, "");
2750 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2751 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2752 PL_preambled = FALSE;
2754 (void)gv_fetchfile(PL_origfilename);
2761 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2763 PL_lex_state = LEX_FORMLINE;
2768 #ifdef PERL_STRICT_CR
2769 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2771 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2773 case ' ': case '\t': case '\f': case 013:
2774 #ifdef MACOS_TRADITIONAL
2781 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2782 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2783 /* handle eval qq[#line 1 "foo"\n ...] */
2784 CopLINE_dec(PL_curcop);
2788 while (s < d && *s != '\n')
2793 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2795 PL_lex_state = LEX_FORMLINE;
2805 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2812 while (s < PL_bufend && SPACE_OR_TAB(*s))
2815 if (strnEQ(s,"=>",2)) {
2816 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2817 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2818 "### Saw unary minus before =>, forcing word '%s'\n", s);
2820 OPERATOR('-'); /* unary minus */
2822 PL_last_uni = PL_oldbufptr;
2824 case 'r': ftst = OP_FTEREAD; break;
2825 case 'w': ftst = OP_FTEWRITE; break;
2826 case 'x': ftst = OP_FTEEXEC; break;
2827 case 'o': ftst = OP_FTEOWNED; break;
2828 case 'R': ftst = OP_FTRREAD; break;
2829 case 'W': ftst = OP_FTRWRITE; break;
2830 case 'X': ftst = OP_FTREXEC; break;
2831 case 'O': ftst = OP_FTROWNED; break;
2832 case 'e': ftst = OP_FTIS; break;
2833 case 'z': ftst = OP_FTZERO; break;
2834 case 's': ftst = OP_FTSIZE; break;
2835 case 'f': ftst = OP_FTFILE; break;
2836 case 'd': ftst = OP_FTDIR; break;
2837 case 'l': ftst = OP_FTLINK; break;
2838 case 'p': ftst = OP_FTPIPE; break;
2839 case 'S': ftst = OP_FTSOCK; break;
2840 case 'u': ftst = OP_FTSUID; break;
2841 case 'g': ftst = OP_FTSGID; break;
2842 case 'k': ftst = OP_FTSVTX; break;
2843 case 'b': ftst = OP_FTBLK; break;
2844 case 'c': ftst = OP_FTCHR; break;
2845 case 't': ftst = OP_FTTTY; break;
2846 case 'T': ftst = OP_FTTEXT; break;
2847 case 'B': ftst = OP_FTBINARY; break;
2848 case 'M': case 'A': case 'C':
2849 gv_fetchpv("\024",TRUE, SVt_PV);
2851 case 'M': ftst = OP_FTMTIME; break;
2852 case 'A': ftst = OP_FTATIME; break;
2853 case 'C': ftst = OP_FTCTIME; break;
2861 PL_last_lop_op = ftst;
2862 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2863 "### Saw file test %c\n", ftst);
2868 /* Assume it was a minus followed by a one-letter named
2869 * subroutine call (or a -bareword), then. */
2870 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2871 "### %c looked like a file test but was not\n", ftst);
2879 if (PL_expect == XOPERATOR)
2884 else if (*s == '>') {
2887 if (isIDFIRST_lazy_if(s,UTF)) {
2888 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2896 if (PL_expect == XOPERATOR)
2899 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2901 OPERATOR('-'); /* unary minus */
2908 if (PL_expect == XOPERATOR)
2913 if (PL_expect == XOPERATOR)
2916 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2922 if (PL_expect != XOPERATOR) {
2923 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2924 PL_expect = XOPERATOR;
2925 force_ident(PL_tokenbuf, '*');
2938 if (PL_expect == XOPERATOR) {
2942 PL_tokenbuf[0] = '%';
2943 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2944 if (!PL_tokenbuf[1]) {
2946 yyerror("Final % should be \\% or %name");
2949 PL_pending_ident = '%';
2968 switch (PL_expect) {
2971 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2973 PL_bufptr = s; /* update in case we back off */
2979 PL_expect = XTERMBLOCK;
2983 while (isIDFIRST_lazy_if(s,UTF)) {
2984 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2985 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2986 if (tmp < 0) tmp = -tmp;
3001 d = scan_str(d,TRUE,TRUE);
3004 SvREFCNT_dec(PL_lex_stuff);
3005 PL_lex_stuff = Nullsv;
3007 /* MUST advance bufptr here to avoid bogus
3008 "at end of line" context messages from yyerror().
3010 PL_bufptr = s + len;
3011 yyerror("Unterminated attribute parameter in attribute list");
3014 return 0; /* EOF indicator */
3018 SV *sv = newSVpvn(s, len);
3019 sv_catsv(sv, PL_lex_stuff);
3020 attrs = append_elem(OP_LIST, attrs,
3021 newSVOP(OP_CONST, 0, sv));
3022 SvREFCNT_dec(PL_lex_stuff);
3023 PL_lex_stuff = Nullsv;
3026 attrs = append_elem(OP_LIST, attrs,
3027 newSVOP(OP_CONST, 0,
3031 if (*s == ':' && s[1] != ':')
3034 break; /* require real whitespace or :'s */
3036 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3037 if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) {
3038 char q = ((*s == '\'') ? '"' : '\'');
3039 /* If here for an expression, and parsed no attrs, back off. */
3040 if (tmp == '=' && !attrs) {
3044 /* MUST advance bufptr here to avoid bogus "at end of line"
3045 context messages from yyerror().
3049 yyerror("Unterminated attribute list");
3051 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3059 PL_nextval[PL_nexttoke].opval = attrs;
3067 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3068 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3084 if (PL_lex_brackets <= 0)
3085 yyerror("Unmatched right square bracket");
3088 if (PL_lex_state == LEX_INTERPNORMAL) {
3089 if (PL_lex_brackets == 0) {
3090 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3091 PL_lex_state = LEX_INTERPEND;
3098 if (PL_lex_brackets > 100) {
3099 char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char);
3100 if (newlb != PL_lex_brackstack) {
3102 PL_lex_brackstack = newlb;
3105 switch (PL_expect) {
3107 if (PL_lex_formbrack) {
3111 if (PL_oldoldbufptr == PL_last_lop)
3112 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3114 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3115 OPERATOR(HASHBRACK);
3117 while (s < PL_bufend && SPACE_OR_TAB(*s))
3120 PL_tokenbuf[0] = '\0';
3121 if (d < PL_bufend && *d == '-') {
3122 PL_tokenbuf[0] = '-';
3124 while (d < PL_bufend && SPACE_OR_TAB(*d))
3127 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3128 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3130 while (d < PL_bufend && SPACE_OR_TAB(*d))
3133 char minus = (PL_tokenbuf[0] == '-');
3134 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3135 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, 0) &&
3136 PL_nextval[PL_nexttoke-1].opval)
3137 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke-1].opval)->op_sv);
3145 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3150 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3155 if (PL_oldoldbufptr == PL_last_lop)
3156 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3158 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3161 OPERATOR(HASHBRACK);
3162 /* This hack serves to disambiguate a pair of curlies
3163 * as being a block or an anon hash. Normally, expectation
3164 * determines that, but in cases where we're not in a
3165 * position to expect anything in particular (like inside
3166 * eval"") we have to resolve the ambiguity. This code
3167 * covers the case where the first term in the curlies is a
3168 * quoted string. Most other cases need to be explicitly
3169 * disambiguated by prepending a `+' before the opening
3170 * curly in order to force resolution as an anon hash.
3172 * XXX should probably propagate the outer expectation
3173 * into eval"" to rely less on this hack, but that could
3174 * potentially break current behavior of eval"".
3178 if (*s == '\'' || *s == '"' || *s == '`') {
3179 /* common case: get past first string, handling escapes */
3180 for (t++; t < PL_bufend && *t != *s;)
3181 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3185 else if (*s == 'q') {
3188 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3192 char open, close, term;
3195 while (t < PL_bufend && isSPACE(*t))
3199 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3203 for (t++; t < PL_bufend; t++) {
3204 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3206 else if (*t == open)
3210 for (t++; t < PL_bufend; t++) {
3211 if (*t == '\\' && t+1 < PL_bufend)
3213 else if (*t == close && --brackets <= 0)
3215 else if (*t == open)
3221 else if (isALNUM_lazy_if(t,UTF)) {
3223 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3226 while (t < PL_bufend && isSPACE(*t))
3228 /* if comma follows first term, call it an anon hash */
3229 /* XXX it could be a comma expression with loop modifiers */
3230 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3231 || (*t == '=' && t[1] == '>')))
3232 OPERATOR(HASHBRACK);
3233 if (PL_expect == XREF)
3236 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3242 yylval.ival = CopLINE(PL_curcop);
3243 if (isSPACE(*s) || *s == '#')
3244 PL_copline = NOLINE; /* invalidate current command line number */
3249 if (PL_lex_brackets <= 0)
3250 yyerror("Unmatched right curly bracket");
3252 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3253 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3254 PL_lex_formbrack = 0;
3255 if (PL_lex_state == LEX_INTERPNORMAL) {
3256 if (PL_lex_brackets == 0) {
3257 if (PL_expect & XFAKEBRACK) {
3258 PL_expect &= XENUMMASK;
3259 PL_lex_state = LEX_INTERPEND;
3261 return yylex(); /* ignore fake brackets */
3263 if (*s == '-' && s[1] == '>')
3264 PL_lex_state = LEX_INTERPENDMAYBE;
3265 else if (*s != '[' && *s != '{')
3266 PL_lex_state = LEX_INTERPEND;
3269 if (PL_expect & XFAKEBRACK) {
3270 PL_expect &= XENUMMASK;
3272 return yylex(); /* ignore fake brackets */
3282 if (PL_expect == XOPERATOR) {
3283 if (ckWARN(WARN_SEMICOLON)
3284 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3286 CopLINE_dec(PL_curcop);
3287 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3288 CopLINE_inc(PL_curcop);
3293 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3295 PL_expect = XOPERATOR;
3296 force_ident(PL_tokenbuf, '&');
3300 yylval.ival = (OPpENTERSUB_AMPER<<8);
3319 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3320 Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp);
3322 if (PL_expect == XSTATE && isALPHA(tmp) &&
3323 (s == PL_linestart+1 || s[-2] == '\n') )
3325 if (PL_in_eval && !PL_rsfp) {
3330 if (strnEQ(s,"=cut",4)) {
3344 PL_doextract = TRUE;
3347 if (PL_lex_brackets < PL_lex_formbrack) {
3349 #ifdef PERL_STRICT_CR
3350 for (t = s; SPACE_OR_TAB(*t); t++) ;
3352 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3354 if (*t == '\n' || *t == '#') {
3372 if (PL_expect != XOPERATOR) {
3373 if (s[1] != '<' && !strchr(s,'>'))
3376 s = scan_heredoc(s);
3378 s = scan_inputsymbol(s);
3379 TERM(sublex_start());
3384 SHop(OP_LEFT_SHIFT);
3398 SHop(OP_RIGHT_SHIFT);
3407 if (PL_expect == XOPERATOR) {
3408 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3411 return ','; /* grandfather non-comma-format format */
3415 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3416 PL_tokenbuf[0] = '@';
3417 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3418 sizeof PL_tokenbuf - 1, FALSE);
3419 if (PL_expect == XOPERATOR)
3420 no_op("Array length", s);
3421 if (!PL_tokenbuf[1])
3423 PL_expect = XOPERATOR;
3424 PL_pending_ident = '#';
3428 PL_tokenbuf[0] = '$';
3429 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3430 sizeof PL_tokenbuf - 1, FALSE);
3431 if (PL_expect == XOPERATOR)
3433 if (!PL_tokenbuf[1]) {
3435 yyerror("Final $ should be \\$ or $name");
3439 /* This kludge not intended to be bulletproof. */
3440 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3441 yylval.opval = newSVOP(OP_CONST, 0,
3442 newSViv(PL_compiling.cop_arybase));
3443 yylval.opval->op_private = OPpCONST_ARYBASE;
3449 if (PL_lex_state == LEX_NORMAL)
3452 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3455 PL_tokenbuf[0] = '@';
3456 if (ckWARN(WARN_SYNTAX)) {
3458 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3461 PL_bufptr = skipspace(PL_bufptr);
3462 while (t < PL_bufend && *t != ']')
3464 Perl_warner(aTHX_ WARN_SYNTAX,
3465 "Multidimensional syntax %.*s not supported",
3466 (t - PL_bufptr) + 1, PL_bufptr);
3470 else if (*s == '{') {
3471 PL_tokenbuf[0] = '%';
3472 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3473 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3475 char tmpbuf[sizeof PL_tokenbuf];
3477 for (t++; isSPACE(*t); t++) ;
3478 if (isIDFIRST_lazy_if(t,UTF)) {
3479 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3480 for (; isSPACE(*t); t++) ;
3481 if (*t == ';' && get_cv(tmpbuf, FALSE))
3482 Perl_warner(aTHX_ WARN_SYNTAX,
3483 "You need to quote \"%s\"", tmpbuf);
3489 PL_expect = XOPERATOR;
3490 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3491 bool islop = (PL_last_lop == PL_oldoldbufptr);
3492 if (!islop || PL_last_lop_op == OP_GREPSTART)
3493 PL_expect = XOPERATOR;
3494 else if (strchr("$@\"'`q", *s))
3495 PL_expect = XTERM; /* e.g. print $fh "foo" */
3496 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3497 PL_expect = XTERM; /* e.g. print $fh &sub */
3498 else if (isIDFIRST_lazy_if(s,UTF)) {
3499 char tmpbuf[sizeof PL_tokenbuf];
3500 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3501 if ((tmp = keyword(tmpbuf, len))) {
3502 /* binary operators exclude handle interpretations */
3514 PL_expect = XTERM; /* e.g. print $fh length() */
3519 GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
3520 if (gv && GvCVu(gv))
3521 PL_expect = XTERM; /* e.g. print $fh subr() */
3524 else if (isDIGIT(*s))
3525 PL_expect = XTERM; /* e.g. print $fh 3 */
3526 else if (*s == '.' && isDIGIT(s[1]))
3527 PL_expect = XTERM; /* e.g. print $fh .3 */
3528 else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3529 PL_expect = XTERM; /* e.g. print $fh -1 */
3530 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3531 PL_expect = XTERM; /* print $fh <<"EOF" */
3533 PL_pending_ident = '$';
3537 if (PL_expect == XOPERATOR)
3539 PL_tokenbuf[0] = '@';
3540 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3541 if (!PL_tokenbuf[1]) {
3543 yyerror("Final @ should be \\@ or @name");
3546 if (PL_lex_state == LEX_NORMAL)
3548 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3550 PL_tokenbuf[0] = '%';
3552 /* Warn about @ where they meant $. */
3553 if (ckWARN(WARN_SYNTAX)) {
3554 if (*s == '[' || *s == '{') {
3556 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3558 if (*t == '}' || *t == ']') {
3560 PL_bufptr = skipspace(PL_bufptr);
3561 Perl_warner(aTHX_ WARN_SYNTAX,
3562 "Scalar value %.*s better written as $%.*s",
3563 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3568 PL_pending_ident = '@';
3571 case '/': /* may either be division or pattern */
3572 case '?': /* may either be conditional or pattern */
3573 if (PL_expect != XOPERATOR) {
3574 /* Disable warning on "study /blah/" */
3575 if (PL_oldoldbufptr == PL_last_uni
3576 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3577 || memNE(PL_last_uni, "study", 5)
3578 || isALNUM_lazy_if(PL_last_uni+5,UTF)))
3580 s = scan_pat(s,OP_MATCH);
3581 TERM(sublex_start());
3589 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3590 #ifdef PERL_STRICT_CR
3593 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3595 && (s == PL_linestart || s[-1] == '\n') )
3597 PL_lex_formbrack = 0;
3601 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3607 yylval.ival = OPf_SPECIAL;
3613 if (PL_expect != XOPERATOR)
3618 case '0': case '1': case '2': case '3': case '4':
3619 case '5': case '6': case '7': case '8': case '9':
3620 s = scan_num(s, &yylval);
3621 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3622 "### Saw number in '%s'\n", s);
3624 if (PL_expect == XOPERATOR)
3629 s = scan_str(s,FALSE,FALSE);
3630 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3631 "### Saw string in '%s'\n", s);
3633 if (PL_expect == XOPERATOR) {
3634 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3637 return ','; /* grandfather non-comma-format format */
3643 missingterm((char*)0);
3644 yylval.ival = OP_CONST;
3645 TERM(sublex_start());
3648 s = scan_str(s,FALSE,FALSE);
3649 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3650 "### Saw string in '%s'\n", s);
3652 if (PL_expect == XOPERATOR) {
3653 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3656 return ','; /* grandfather non-comma-format format */
3662 missingterm((char*)0);
3663 yylval.ival = OP_CONST;
3664 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3665 if (*d == '$' || *d == '@' || *d == '\\' || UTF8_IS_CONTINUED(*d)) {
3666 yylval.ival = OP_STRINGIFY;
3670 TERM(sublex_start());
3673 s = scan_str(s,FALSE,FALSE);
3674 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3675 "### Saw backtick string in '%s'\n", s);
3677 if (PL_expect == XOPERATOR)
3678 no_op("Backticks",s);
3680 missingterm((char*)0);
3681 yylval.ival = OP_BACKTICK;
3683 TERM(sublex_start());
3687 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3688 Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression",
3690 if (PL_expect == XOPERATOR)
3691 no_op("Backslash",s);
3695 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3699 while (isDIGIT(*start) || *start == '_')
3701 if (*start == '.' && isDIGIT(start[1])) {
3702 s = scan_num(s, &yylval);
3705 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3706 else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF)) {
3710 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3713 s = scan_num(s, &yylval);
3720 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3759 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3761 /* Some keywords can be followed by any delimiter, including ':' */
3762 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3763 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3764 (PL_tokenbuf[0] == 'q' &&
3765 strchr("qwxr", PL_tokenbuf[1])))));
3767 /* x::* is just a word, unless x is "CORE" */
3768 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3772 while (d < PL_bufend && isSPACE(*d))
3773 d++; /* no comments skipped here, or s### is misparsed */
3775 /* Is this a label? */
3776 if (!tmp && PL_expect == XSTATE
3777 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3779 yylval.pval = savepv(PL_tokenbuf);
3784 /* Check for keywords */
3785 tmp = keyword(PL_tokenbuf, len);
3787 /* Is this a word before a => operator? */
3788 if (*d == '=' && d[1] == '>') {
3790 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3791 yylval.opval->op_private = OPpCONST_BARE;
3792 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3793 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3797 if (tmp < 0) { /* second-class keyword? */
3798 GV *ogv = Nullgv; /* override (winner) */
3799 GV *hgv = Nullgv; /* hidden (loser) */
3800 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3802 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3805 if (GvIMPORTED_CV(gv))
3807 else if (! CvMETHOD(cv))
3811 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3812 (gv = *gvp) != (GV*)&PL_sv_undef &&
3813 GvCVu(gv) && GvIMPORTED_CV(gv))
3819 tmp = 0; /* overridden by import or by GLOBAL */
3822 && -tmp==KEY_lock /* XXX generalizable kludge */
3824 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3826 tmp = 0; /* any sub overrides "weak" keyword */
3828 else { /* no override */
3832 if (ckWARN(WARN_AMBIGUOUS) && hgv
3833 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3834 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3835 "Ambiguous call resolved as CORE::%s(), %s",
3836 GvENAME(hgv), "qualify as such or use &");
3843 default: /* not a keyword */
3846 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3848 /* Get the rest if it looks like a package qualifier */
3850 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3852 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3855 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3856 *s == '\'' ? "'" : "::");
3860 if (PL_expect == XOPERATOR) {
3861 if (PL_bufptr == PL_linestart) {
3862 CopLINE_dec(PL_curcop);
3863 Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi);
3864 CopLINE_inc(PL_curcop);
3867 no_op("Bareword",s);
3870 /* Look for a subroutine with this name in current package,
3871 unless name is "Foo::", in which case Foo is a bearword
3872 (and a package name). */
3875 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3877 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3878 Perl_warner(aTHX_ WARN_BAREWORD,
3879 "Bareword \"%s\" refers to nonexistent package",
3882 PL_tokenbuf[len] = '\0';
3889 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3892 /* if we saw a global override before, get the right name */
3895 sv = newSVpvn("CORE::GLOBAL::",14);
3896 sv_catpv(sv,PL_tokenbuf);
3899 sv = newSVpv(PL_tokenbuf,0);
3901 /* Presume this is going to be a bareword of some sort. */
3904 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3905 yylval.opval->op_private = OPpCONST_BARE;
3907 /* And if "Foo::", then that's what it certainly is. */
3912 /* See if it's the indirect object for a list operator. */
3914 if (PL_oldoldbufptr &&
3915 PL_oldoldbufptr < PL_bufptr &&
3916 (PL_oldoldbufptr == PL_last_lop
3917 || PL_oldoldbufptr == PL_last_uni) &&
3918 /* NO SKIPSPACE BEFORE HERE! */
3919 (PL_expect == XREF ||
3920 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3922 bool immediate_paren = *s == '(';
3924 /* (Now we can afford to cross potential line boundary.) */
3927 /* Two barewords in a row may indicate method call. */
3929 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
3932 /* If not a declared subroutine, it's an indirect object. */
3933 /* (But it's an indir obj regardless for sort.) */
3935 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
3936 ((!gv || !GvCVu(gv)) &&
3937 (PL_last_lop_op != OP_MAPSTART &&
3938 PL_last_lop_op != OP_GREPSTART))))
3940 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
3946 PL_expect = XOPERATOR;
3949 /* Is this a word before a => operator? */
3950 if (*s == '=' && s[1] == '>') {
3952 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
3953 if (UTF && !IN_BYTE && is_utf8_string((U8*)PL_tokenbuf, len))
3954 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3958 /* If followed by a paren, it's certainly a subroutine. */
3961 if (gv && GvCVu(gv)) {
3962 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
3963 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
3968 PL_nextval[PL_nexttoke].opval = yylval.opval;
3969 PL_expect = XOPERATOR;
3975 /* If followed by var or block, call it a method (unless sub) */
3977 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
3978 PL_last_lop = PL_oldbufptr;
3979 PL_last_lop_op = OP_METHOD;
3983 /* If followed by a bareword, see if it looks like indir obj. */
3985 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv)))
3988 /* Not a method, so call it a subroutine (if defined) */
3990 if (gv && GvCVu(gv)) {
3992 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
3993 Perl_warner(aTHX_ WARN_AMBIGUOUS,
3994 "Ambiguous use of -%s resolved as -&%s()",
3995 PL_tokenbuf, PL_tokenbuf);
3996 /* Check for a constant sub */
3998 if ((sv = cv_const_sv(cv))) {
4000 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4001 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4002 yylval.opval->op_private = 0;
4006 /* Resolve to GV now. */
4007 op_free(yylval.opval);
4008 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4009 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4010 PL_last_lop = PL_oldbufptr;
4011 PL_last_lop_op = OP_ENTERSUB;
4012 /* Is there a prototype? */
4015 char *proto = SvPV((SV*)cv, len);
4018 if (strEQ(proto, "$"))
4020 if (*proto == '&' && *s == '{') {
4021 sv_setpv(PL_subname,"__ANON__");
4025 PL_nextval[PL_nexttoke].opval = yylval.opval;
4031 /* Call it a bare word */
4033 if (PL_hints & HINT_STRICT_SUBS)
4034 yylval.opval->op_private |= OPpCONST_STRICT;
4037 if (ckWARN(WARN_RESERVED)) {
4038 if (lastchar != '-') {
4039 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4041 Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved,
4048 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4049 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4050 "Operator or semicolon missing before %c%s",
4051 lastchar, PL_tokenbuf);
4052 Perl_warner(aTHX_ WARN_AMBIGUOUS,
4053 "Ambiguous use of %c resolved as operator %c",
4054 lastchar, lastchar);
4060 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4061 newSVpv(CopFILE(PL_curcop),0));
4065 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4066 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4069 case KEY___PACKAGE__:
4070 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4072 ? newSVsv(PL_curstname)
4081 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4082 char *pname = "main";
4083 if (PL_tokenbuf[2] == 'D')
4084 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4085 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4088 GvIOp(gv) = newIO();
4089 IoIFP(GvIOp(gv)) = PL_rsfp;
4090 #if defined(HAS_FCNTL) && defined(F_SETFD)
4092 int fd = PerlIO_fileno(PL_rsfp);
4093 fcntl(fd,F_SETFD,fd >= 3);
4096 /* Mark this internal pseudo-handle as clean */
4097 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4099 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4100 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4101 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4103 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4104 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4105 /* if the script was opened in binmode, we need to revert
4106 * it to text mode for compatibility; but only iff it has CRs
4107 * XXX this is a questionable hack at best. */
4108 if (PL_bufend-PL_bufptr > 2
4109 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4112 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4113 loc = PerlIO_tell(PL_rsfp);
4114 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4116 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4117 #if defined(__BORLANDC__)
4118 /* XXX see note in do_binmode() */
4119 ((FILE*)PL_rsfp)->flags |= _F_BIN;
4122 PerlIO_seek(PL_rsfp, loc, 0);
4126 #ifdef PERLIO_LAYERS
4127 if (UTF && !IN_BYTE)
4128 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4141 if (PL_expect == XSTATE) {
4148 if (*s == ':' && s[1] == ':') {
4151 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4152 if (!(tmp = keyword(PL_tokenbuf, len)))
4153 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4167 LOP(OP_ACCEPT,XTERM);
4173 LOP(OP_ATAN2,XTERM);
4179 LOP(OP_BINMODE,XTERM);
4182 LOP(OP_BLESS,XTERM);
4191 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4208 if (!PL_cryptseen) {
4209 PL_cryptseen = TRUE;
4213 LOP(OP_CRYPT,XTERM);
4216 if (ckWARN(WARN_CHMOD)) {
4217 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
4218 if (*d != '0' && isDIGIT(*d))
4219 Perl_warner(aTHX_ WARN_CHMOD,
4220 "chmod() mode argument is missing initial 0");
4222 LOP(OP_CHMOD,XTERM);
4225 LOP(OP_CHOWN,XTERM);
4228 LOP(OP_CONNECT,XTERM);
4244 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4248 PL_hints |= HINT_BLOCK_SCOPE;
4258 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4259 LOP(OP_DBMOPEN,XTERM);
4265 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4272 yylval.ival = CopLINE(PL_curcop);
4286 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4287 UNIBRACK(OP_ENTEREVAL);
4302 case KEY_endhostent:
4308 case KEY_endservent:
4311 case KEY_endprotoent:
4322 yylval.ival = CopLINE(PL_curcop);
4324 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4326 if ((PL_bufend - p) >= 3 &&
4327 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4329 else if ((PL_bufend - p) >= 4 &&
4330 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4333 if (isIDFIRST_lazy_if(p,UTF)) {
4334 p = scan_ident(p, PL_bufend,
4335 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4339 Perl_croak(aTHX_ "Missing $ on loop variable");
4344 LOP(OP_FORMLINE,XTERM);
4350 LOP(OP_FCNTL,XTERM);
4356 LOP(OP_FLOCK,XTERM);
4365 LOP(OP_GREPSTART, XREF);
4368 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4383 case KEY_getpriority:
4384 LOP(OP_GETPRIORITY,XTERM);
4386 case KEY_getprotobyname:
4389 case KEY_getprotobynumber:
4390 LOP(OP_GPBYNUMBER,XTERM);
4392 case KEY_getprotoent:
4404 case KEY_getpeername:
4405 UNI(OP_GETPEERNAME);
4407 case KEY_gethostbyname:
4410 case KEY_gethostbyaddr:
4411 LOP(OP_GHBYADDR,XTERM);
4413 case KEY_gethostent:
4416 case KEY_getnetbyname:
4419 case KEY_getnetbyaddr:
4420 LOP(OP_GNBYADDR,XTERM);
4425 case KEY_getservbyname:
4426 LOP(OP_GSBYNAME,XTERM);
4428 case KEY_getservbyport:
4429 LOP(OP_GSBYPORT,XTERM);
4431 case KEY_getservent:
4434 case KEY_getsockname:
4435 UNI(OP_GETSOCKNAME);
4437 case KEY_getsockopt:
4438 LOP(OP_GSOCKOPT,XTERM);
4460 yylval.ival = CopLINE(PL_curcop);
4464 LOP(OP_INDEX,XTERM);
4470 LOP(OP_IOCTL,XTERM);
4482 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4514 LOP(OP_LISTEN,XTERM);
4523 s = scan_pat(s,OP_MATCH);
4524 TERM(sublex_start());
4527 LOP(OP_MAPSTART, XREF);
4530 LOP(OP_MKDIR,XTERM);
4533 LOP(OP_MSGCTL,XTERM);
4536 LOP(OP_MSGGET,XTERM);
4539 LOP(OP_MSGRCV,XTERM);
4542 LOP(OP_MSGSND,XTERM);
4548 if (isIDFIRST_lazy_if(s,UTF)) {
4549 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4550 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4552 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4553 if (!PL_in_my_stash) {
4556 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4564 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4571 if (PL_expect != XSTATE)
4572 yyerror("\"no\" not allowed in expression");
4573 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4574 s = force_version(s);
4579 if (*s == '(' || (s = skipspace(s), *s == '('))
4586 if (isIDFIRST_lazy_if(s,UTF)) {
4588 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4590 if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE))
4591 Perl_warner(aTHX_ WARN_PRECEDENCE,
4592 "Precedence problem: open %.*s should be open(%.*s)",
4598 yylval.ival = OP_OR;
4608 LOP(OP_OPEN_DIR,XTERM);
4611 checkcomma(s,PL_tokenbuf,"filehandle");
4615 checkcomma(s,PL_tokenbuf,"filehandle");
4634 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4638 LOP(OP_PIPE_OP,XTERM);
4641 s = scan_str(s,FALSE,FALSE);
4643 missingterm((char*)0);
4644 yylval.ival = OP_CONST;
4645 TERM(sublex_start());
4651 s = scan_str(s,FALSE,FALSE);
4653 missingterm((char*)0);
4655 if (SvCUR(PL_lex_stuff)) {
4658 d = SvPV_force(PL_lex_stuff, len);
4661 for (; isSPACE(*d) && len; --len, ++d) ;
4664 if (!warned && ckWARN(WARN_QW)) {
4665 for (; !isSPACE(*d) && len; --len, ++d) {
4667 Perl_warner(aTHX_ WARN_QW,
4668 "Possible attempt to separate words with commas");
4671 else if (*d == '#') {
4672 Perl_warner(aTHX_ WARN_QW,
4673 "Possible attempt to put comments in qw() list");
4679 for (; !isSPACE(*d) && len; --len, ++d) ;
4681 sv = newSVpvn(b, d-b);
4682 if (DO_UTF8(PL_lex_stuff))
4684 words = append_elem(OP_LIST, words,
4685 newSVOP(OP_CONST, 0, tokeq(sv)));
4689 PL_nextval[PL_nexttoke].opval = words;
4694 SvREFCNT_dec(PL_lex_stuff);
4695 PL_lex_stuff = Nullsv;
4700 s = scan_str(s,FALSE,FALSE);
4702 missingterm((char*)0);
4703 yylval.ival = OP_STRINGIFY;
4704 if (SvIVX(PL_lex_stuff) == '\'')
4705 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4706 TERM(sublex_start());
4709 s = scan_pat(s,OP_QR);
4710 TERM(sublex_start());
4713 s = scan_str(s,FALSE,FALSE);
4715 missingterm((char*)0);
4716 yylval.ival = OP_BACKTICK;
4718 TERM(sublex_start());
4725 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
4726 s = force_version(s);
4729 *PL_tokenbuf = '\0';
4730 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4731 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4732 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4734 yyerror("<> should be quotes");
4742 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4746 LOP(OP_RENAME,XTERM);
4755 LOP(OP_RINDEX,XTERM);
4778 LOP(OP_REVERSE,XTERM);
4789 TERM(sublex_start());
4791 TOKEN(1); /* force error */
4800 LOP(OP_SELECT,XTERM);
4806 LOP(OP_SEMCTL,XTERM);
4809 LOP(OP_SEMGET,XTERM);
4812 LOP(OP_SEMOP,XTERM);
4818 LOP(OP_SETPGRP,XTERM);
4820 case KEY_setpriority:
4821 LOP(OP_SETPRIORITY,XTERM);
4823 case KEY_sethostent:
4829 case KEY_setservent:
4832 case KEY_setprotoent:
4842 LOP(OP_SEEKDIR,XTERM);
4844 case KEY_setsockopt:
4845 LOP(OP_SSOCKOPT,XTERM);
4851 LOP(OP_SHMCTL,XTERM);
4854 LOP(OP_SHMGET,XTERM);
4857 LOP(OP_SHMREAD,XTERM);
4860 LOP(OP_SHMWRITE,XTERM);
4863 LOP(OP_SHUTDOWN,XTERM);
4872 LOP(OP_SOCKET,XTERM);
4874 case KEY_socketpair:
4875 LOP(OP_SOCKPAIR,XTERM);
4878 checkcomma(s,PL_tokenbuf,"subroutine name");
4880 if (*s == ';' || *s == ')') /* probably a close */
4881 Perl_croak(aTHX_ "sort is now a reserved word");
4883 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4887 LOP(OP_SPLIT,XTERM);
4890 LOP(OP_SPRINTF,XTERM);
4893 LOP(OP_SPLICE,XTERM);
4908 LOP(OP_SUBSTR,XTERM);
4914 char tmpbuf[sizeof PL_tokenbuf];
4916 expectation attrful;
4917 bool have_name, have_proto;
4922 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
4923 (*s == ':' && s[1] == ':'))
4926 attrful = XATTRBLOCK;
4927 /* remember buffer pos'n for later force_word */
4928 tboffset = s - PL_oldbufptr;
4929 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4930 if (strchr(tmpbuf, ':'))
4931 sv_setpv(PL_subname, tmpbuf);
4933 sv_setsv(PL_subname,PL_curstname);
4934 sv_catpvn(PL_subname,"::",2);
4935 sv_catpvn(PL_subname,tmpbuf,len);
4942 Perl_croak(aTHX_ "Missing name in \"my sub\"");
4943 PL_expect = XTERMBLOCK;
4944 attrful = XATTRTERM;
4945 sv_setpv(PL_subname,"?");
4949 if (key == KEY_format) {
4951 PL_lex_formbrack = PL_lex_brackets + 1;
4953 (void) force_word(PL_oldbufptr + tboffset, WORD,
4958 /* Look for a prototype */
4962 s = scan_str(s,FALSE,FALSE);
4965 SvREFCNT_dec(PL_lex_stuff);
4966 PL_lex_stuff = Nullsv;
4967 Perl_croak(aTHX_ "Prototype not terminated");
4970 d = SvPVX(PL_lex_stuff);
4972 for (p = d; *p; ++p) {
4977 SvCUR(PL_lex_stuff) = tmp;
4985 if (*s == ':' && s[1] != ':')
4986 PL_expect = attrful;
4989 PL_nextval[PL_nexttoke].opval =
4990 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
4991 PL_lex_stuff = Nullsv;
4995 sv_setpv(PL_subname,"__ANON__");
4998 (void) force_word(PL_oldbufptr + tboffset, WORD,
5007 LOP(OP_SYSTEM,XREF);
5010 LOP(OP_SYMLINK,XTERM);
5013 LOP(OP_SYSCALL,XTERM);
5016 LOP(OP_SYSOPEN,XTERM);
5019 LOP(OP_SYSSEEK,XTERM);
5022 LOP(OP_SYSREAD,XTERM);
5025 LOP(OP_SYSWRITE,XTERM);
5029 TERM(sublex_start());
5050 LOP(OP_TRUNCATE,XTERM);
5062 yylval.ival = CopLINE(PL_curcop);
5066 yylval.ival = CopLINE(PL_curcop);
5070 LOP(OP_UNLINK,XTERM);
5076 LOP(OP_UNPACK,XTERM);
5079 LOP(OP_UTIME,XTERM);
5082 if (ckWARN(WARN_UMASK)) {
5083 for (d = s; d < PL_bufend && (isSPACE(*d) || *d == '('); d++) ;
5084 if (*d != '0' && isDIGIT(*d))
5085 Perl_warner(aTHX_ WARN_UMASK,
5086 "umask: argument is missing initial 0");
5091 LOP(OP_UNSHIFT,XTERM);
5094 if (PL_expect != XSTATE)
5095 yyerror("\"use\" not allowed in expression");
5097 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5098 s = force_version(s);
5099 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5100 PL_nextval[PL_nexttoke].opval = Nullop;
5105 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5106 s = force_version(s);
5118 yylval.ival = CopLINE(PL_curcop);
5122 PL_hints |= HINT_BLOCK_SCOPE;
5129 LOP(OP_WAITPID,XTERM);
5137 static char ctl_l[2];
5139 if (ctl_l[0] == '\0')
5140 ctl_l[0] = toCTRL('L');
5141 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5144 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5149 if (PL_expect == XOPERATOR)
5155 yylval.ival = OP_XOR;
5160 TERM(sublex_start());
5165 #pragma segment Main
5169 Perl_keyword(pTHX_ register char *d, I32 len)
5174 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5175 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5176 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5177 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5178 if (strEQ(d,"__END__")) return KEY___END__;
5182 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5187 if (strEQ(d,"and")) return -KEY_and;
5188 if (strEQ(d,"abs")) return -KEY_abs;
5191 if (strEQ(d,"alarm")) return -KEY_alarm;
5192 if (strEQ(d,"atan2")) return -KEY_atan2;
5195 if (strEQ(d,"accept")) return -KEY_accept;
5200 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5203 if (strEQ(d,"bless")) return -KEY_bless;
5204 if (strEQ(d,"bind")) return -KEY_bind;
5205 if (strEQ(d,"binmode")) return -KEY_binmode;
5208 if (strEQ(d,"CORE")) return -KEY_CORE;
5209 if (strEQ(d,"CHECK")) return KEY_CHECK;
5214 if (strEQ(d,"cmp")) return -KEY_cmp;
5215 if (strEQ(d,"chr")) return -KEY_chr;
5216 if (strEQ(d,"cos")) return -KEY_cos;
5219 if (strEQ(d,"chop")) return -KEY_chop;
5222 if (strEQ(d,"close")) return -KEY_close;
5223 if (strEQ(d,"chdir")) return -KEY_chdir;
5224 if (strEQ(d,"chomp")) return -KEY_chomp;
5225 if (strEQ(d,"chmod")) return -KEY_chmod;
5226 if (strEQ(d,"chown")) return -KEY_chown;
5227 if (strEQ(d,"crypt")) return -KEY_crypt;
5230 if (strEQ(d,"chroot")) return -KEY_chroot;
5231 if (strEQ(d,"caller")) return -KEY_caller;
5234 if (strEQ(d,"connect")) return -KEY_connect;
5237 if (strEQ(d,"closedir")) return -KEY_closedir;
5238 if (strEQ(d,"continue")) return -KEY_continue;
5243 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5248 if (strEQ(d,"do")) return KEY_do;
5251 if (strEQ(d,"die")) return -KEY_die;
5254 if (strEQ(d,"dump")) return -KEY_dump;
5257 if (strEQ(d,"delete")) return KEY_delete;
5260 if (strEQ(d,"defined")) return KEY_defined;
5261 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5264 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5269 if (strEQ(d,"END")) return KEY_END;
5274 if (strEQ(d,"eq")) return -KEY_eq;
5277 if (strEQ(d,"eof")) return -KEY_eof;
5278 if (strEQ(d,"exp")) return -KEY_exp;
5281 if (strEQ(d,"else")) return KEY_else;
5282 if (strEQ(d,"exit")) return -KEY_exit;
5283 if (strEQ(d,"eval")) return KEY_eval;
5284 if (strEQ(d,"exec")) return -KEY_exec;
5285 if (strEQ(d,"each")) return -KEY_each;
5288 if (strEQ(d,"elsif")) return KEY_elsif;
5291 if (strEQ(d,"exists")) return KEY_exists;
5292 if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif");
5295 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5296 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5299 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5302 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5303 if (strEQ(d,"endservent")) return -KEY_endservent;
5306 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5313 if (strEQ(d,"for")) return KEY_for;
5316 if (strEQ(d,"fork")) return -KEY_fork;
5319 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5320 if (strEQ(d,"flock")) return -KEY_flock;
5323 if (strEQ(d,"format")) return KEY_format;
5324 if (strEQ(d,"fileno")) return -KEY_fileno;
5327 if (strEQ(d,"foreach")) return KEY_foreach;
5330 if (strEQ(d,"formline")) return -KEY_formline;
5335 if (strnEQ(d,"get",3)) {
5340 if (strEQ(d,"ppid")) return -KEY_getppid;
5341 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5344 if (strEQ(d,"pwent")) return -KEY_getpwent;
5345 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5346 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5349 if (strEQ(d,"peername")) return -KEY_getpeername;
5350 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5351 if (strEQ(d,"priority")) return -KEY_getpriority;
5354 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5357 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5361 else if (*d == 'h') {
5362 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5363 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5364 if (strEQ(d,"hostent")) return -KEY_gethostent;
5366 else if (*d == 'n') {
5367 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5368 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5369 if (strEQ(d,"netent")) return -KEY_getnetent;
5371 else if (*d == 's') {
5372 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5373 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5374 if (strEQ(d,"servent")) return -KEY_getservent;
5375 if (strEQ(d,"sockname")) return -KEY_getsockname;
5376 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5378 else if (*d == 'g') {
5379 if (strEQ(d,"grent")) return -KEY_getgrent;
5380 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5381 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5383 else if (*d == 'l') {
5384 if (strEQ(d,"login")) return -KEY_getlogin;
5386 else if (strEQ(d,"c")) return -KEY_getc;
5391 if (strEQ(d,"gt")) return -KEY_gt;
5392 if (strEQ(d,"ge")) return -KEY_ge;
5395 if (strEQ(d,"grep")) return KEY_grep;
5396 if (strEQ(d,"goto")) return KEY_goto;
5397 if (strEQ(d,"glob")) return KEY_glob;
5400 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5405 if (strEQ(d,"hex")) return -KEY_hex;
5408 if (strEQ(d,"INIT")) return KEY_INIT;
5413 if (strEQ(d,"if")) return KEY_if;
5416 if (strEQ(d,"int")) return -KEY_int;
5419 if (strEQ(d,"index")) return -KEY_index;
5420 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5425 if (strEQ(d,"join")) return -KEY_join;
5429 if (strEQ(d,"keys")) return -KEY_keys;
5430 if (strEQ(d,"kill")) return -KEY_kill;
5436 if (strEQ(d,"lt")) return -KEY_lt;
5437 if (strEQ(d,"le")) return -KEY_le;
5438 if (strEQ(d,"lc")) return -KEY_lc;
5441 if (strEQ(d,"log")) return -KEY_log;
5444 if (strEQ(d,"last")) return KEY_last;
5445 if (strEQ(d,"link")) return -KEY_link;
5446 if (strEQ(d,"lock")) return -KEY_lock;
5449 if (strEQ(d,"local")) return KEY_local;
5450 if (strEQ(d,"lstat")) return -KEY_lstat;
5453 if (strEQ(d,"length")) return -KEY_length;
5454 if (strEQ(d,"listen")) return -KEY_listen;
5457 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5460 if (strEQ(d,"localtime")) return -KEY_localtime;
5466 case 1: return KEY_m;
5468 if (strEQ(d,"my")) return KEY_my;
5471 if (strEQ(d,"map")) return KEY_map;
5474 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5477 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5478 if (strEQ(d,"msgget")) return -KEY_msgget;
5479 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5480 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5485 if (strEQ(d,"next")) return KEY_next;
5486 if (strEQ(d,"ne")) return -KEY_ne;
5487 if (strEQ(d,"not")) return -KEY_not;
5488 if (strEQ(d,"no")) return KEY_no;
5493 if (strEQ(d,"or")) return -KEY_or;
5496 if (strEQ(d,"ord")) return -KEY_ord;
5497 if (strEQ(d,"oct")) return -KEY_oct;
5498 if (strEQ(d,"our")) return KEY_our;
5501 if (strEQ(d,"open")) return -KEY_open;
5504 if (strEQ(d,"opendir")) return -KEY_opendir;
5511 if (strEQ(d,"pop")) return -KEY_pop;
5512 if (strEQ(d,"pos")) return KEY_pos;
5515 if (strEQ(d,"push")) return -KEY_push;
5516 if (strEQ(d,"pack")) return -KEY_pack;
5517 if (strEQ(d,"pipe")) return -KEY_pipe;
5520 if (strEQ(d,"print")) return KEY_print;
5523 if (strEQ(d,"printf")) return KEY_printf;
5526 if (strEQ(d,"package")) return KEY_package;
5529 if (strEQ(d,"prototype")) return KEY_prototype;
5534 if (strEQ(d,"q")) return KEY_q;
5535 if (strEQ(d,"qr")) return KEY_qr;
5536 if (strEQ(d,"qq")) return KEY_qq;
5537 if (strEQ(d,"qw")) return KEY_qw;
5538 if (strEQ(d,"qx")) return KEY_qx;
5540 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5545 if (strEQ(d,"ref")) return -KEY_ref;
5548 if (strEQ(d,"read")) return -KEY_read;
5549 if (strEQ(d,"rand")) return -KEY_rand;
5550 if (strEQ(d,"recv")) return -KEY_recv;
5551 if (strEQ(d,"redo")) return KEY_redo;
5554 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5555 if (strEQ(d,"reset")) return -KEY_reset;
5558 if (strEQ(d,"return")) return KEY_return;
5559 if (strEQ(d,"rename")) return -KEY_rename;
5560 if (strEQ(d,"rindex")) return -KEY_rindex;
5563 if (strEQ(d,"require")) return -KEY_require;
5564 if (strEQ(d,"reverse")) return -KEY_reverse;
5565 if (strEQ(d,"readdir")) return -KEY_readdir;
5568 if (strEQ(d,"readlink")) return -KEY_readlink;
5569 if (strEQ(d,"readline")) return -KEY_readline;
5570 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5573 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5579 case 0: return KEY_s;
5581 if (strEQ(d,"scalar")) return KEY_scalar;
5586 if (strEQ(d,"seek")) return -KEY_seek;
5587 if (strEQ(d,"send")) return -KEY_send;
5590 if (strEQ(d,"semop")) return -KEY_semop;
5593 if (strEQ(d,"select")) return -KEY_select;
5594 if (strEQ(d,"semctl")) return -KEY_semctl;
5595 if (strEQ(d,"semget")) return -KEY_semget;
5598 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5599 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5602 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5603 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5606 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5609 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5610 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5611 if (strEQ(d,"setservent")) return -KEY_setservent;
5614 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5615 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5622 if (strEQ(d,"shift")) return -KEY_shift;
5625 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5626 if (strEQ(d,"shmget")) return -KEY_shmget;
5629 if (strEQ(d,"shmread")) return -KEY_shmread;
5632 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5633 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5638 if (strEQ(d,"sin")) return -KEY_sin;
5641 if (strEQ(d,"sleep")) return -KEY_sleep;
5644 if (strEQ(d,"sort")) return KEY_sort;
5645 if (strEQ(d,"socket")) return -KEY_socket;
5646 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5649 if (strEQ(d,"split")) return KEY_split;
5650 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5651 if (strEQ(d,"splice")) return -KEY_splice;
5654 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5657 if (strEQ(d,"srand")) return -KEY_srand;
5660 if (strEQ(d,"stat")) return -KEY_stat;
5661 if (strEQ(d,"study")) return KEY_study;
5664 if (strEQ(d,"substr")) return -KEY_substr;
5665 if (strEQ(d,"sub")) return KEY_sub;
5670 if (strEQ(d,"system")) return -KEY_system;
5673 if (strEQ(d,"symlink")) return -KEY_symlink;
5674 if (strEQ(d,"syscall")) return -KEY_syscall;
5675 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5676 if (strEQ(d,"sysread")) return -KEY_sysread;
5677 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5680 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5689 if (strEQ(d,"tr")) return KEY_tr;
5692 if (strEQ(d,"tie")) return KEY_tie;
5695 if (strEQ(d,"tell")) return -KEY_tell;
5696 if (strEQ(d,"tied")) return KEY_tied;
5697 if (strEQ(d,"time")) return -KEY_time;
5700 if (strEQ(d,"times")) return -KEY_times;
5703 if (strEQ(d,"telldir")) return -KEY_telldir;
5706 if (strEQ(d,"truncate")) return -KEY_truncate;
5713 if (strEQ(d,"uc")) return -KEY_uc;
5716 if (strEQ(d,"use")) return KEY_use;
5719 if (strEQ(d,"undef")) return KEY_undef;
5720 if (strEQ(d,"until")) return KEY_until;
5721 if (strEQ(d,"untie")) return KEY_untie;
5722 if (strEQ(d,"utime")) return -KEY_utime;
5723 if (strEQ(d,"umask")) return -KEY_umask;
5726 if (strEQ(d,"unless")) return KEY_unless;
5727 if (strEQ(d,"unpack")) return -KEY_unpack;
5728 if (strEQ(d,"unlink")) return -KEY_unlink;
5731 if (strEQ(d,"unshift")) return -KEY_unshift;
5732 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5737 if (strEQ(d,"values")) return -KEY_values;
5738 if (strEQ(d,"vec")) return -KEY_vec;
5743 if (strEQ(d,"warn")) return -KEY_warn;
5744 if (strEQ(d,"wait")) return -KEY_wait;
5747 if (strEQ(d,"while")) return KEY_while;
5748 if (strEQ(d,"write")) return -KEY_write;
5751 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5754 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5759 if (len == 1) return -KEY_x;
5760 if (strEQ(d,"xor")) return -KEY_xor;
5763 if (len == 1) return KEY_y;
5772 S_checkcomma(pTHX_ register char *s, char *name, char *what)
5776 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
5777 if (ckWARN(WARN_SYNTAX)) {
5779 for (w = s+2; *w && level; w++) {
5786 for (; *w && isSPACE(*w); w++) ;
5787 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
5788 Perl_warner(aTHX_ WARN_SYNTAX,
5789 "%s (...) interpreted as function",name);
5792 while (s < PL_bufend && isSPACE(*s))
5796 while (s < PL_bufend && isSPACE(*s))
5798 if (isIDFIRST_lazy_if(s,UTF)) {
5800 while (isALNUM_lazy_if(s,UTF))
5802 while (s < PL_bufend && isSPACE(*s))
5807 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
5811 Perl_croak(aTHX_ "No comma allowed after %s", what);
5816 /* Either returns sv, or mortalizes sv and returns a new SV*.
5817 Best used as sv=new_constant(..., sv, ...).
5818 If s, pv are NULL, calls subroutine with one argument,
5819 and type is used with error messages only. */
5822 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
5826 HV *table = GvHV(PL_hintgv); /* ^H */
5830 const char *why1, *why2, *why3;
5832 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
5835 why2 = strEQ(key,"charnames")
5836 ? "(possibly a missing \"use charnames ...\")"
5838 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
5839 (type ? type: "undef"), why2);
5841 /* This is convoluted and evil ("goto considered harmful")
5842 * but I do not understand the intricacies of all the different
5843 * failure modes of %^H in here. The goal here is to make
5844 * the most probable error message user-friendly. --jhi */
5849 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
5850 (type ? type: "undef"), why1, why2, why3);
5852 yyerror(SvPVX(msg));
5856 cvp = hv_fetch(table, key, strlen(key), FALSE);
5857 if (!cvp || !SvOK(*cvp)) {
5860 why3 = "} is not defined";
5863 sv_2mortal(sv); /* Parent created it permanently */
5866 pv = sv_2mortal(newSVpvn(s, len));
5868 typesv = sv_2mortal(newSVpv(type, 0));
5870 typesv = &PL_sv_undef;
5872 PUSHSTACKi(PERLSI_OVERLOAD);
5884 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
5888 /* Check the eval first */
5889 if (!PL_in_eval && SvTRUE(ERRSV)) {
5891 sv_catpv(ERRSV, "Propagated");
5892 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
5894 res = SvREFCNT_inc(sv);
5898 (void)SvREFCNT_inc(res);
5907 why1 = "Call to &{$^H{";
5909 why3 = "}} did not return a defined value";
5918 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
5920 register char *d = dest;
5921 register char *e = d + destlen - 3; /* two-character token, ending NUL */
5924 Perl_croak(aTHX_ ident_too_long);
5925 if (isALNUM(*s)) /* UTF handled below */
5927 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
5932 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
5936 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
5937 char *t = s + UTF8SKIP(s);
5938 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
5940 if (d + (t - s) > e)
5941 Perl_croak(aTHX_ ident_too_long);
5942 Copy(s, d, t - s, char);
5955 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
5965 e = d + destlen - 3; /* two-character token, ending NUL */
5967 while (isDIGIT(*s)) {
5969 Perl_croak(aTHX_ ident_too_long);
5976 Perl_croak(aTHX_ ident_too_long);
5977 if (isALNUM(*s)) /* UTF handled below */
5979 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
5984 else if (*s == ':' && s[1] == ':') {
5988 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
5989 char *t = s + UTF8SKIP(s);
5990 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
5992 if (d + (t - s) > e)
5993 Perl_croak(aTHX_ ident_too_long);
5994 Copy(s, d, t - s, char);
6005 if (PL_lex_state != LEX_NORMAL)
6006 PL_lex_state = LEX_INTERPENDMAYBE;
6009 if (*s == '$' && s[1] &&
6010 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6023 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6028 if (isSPACE(s[-1])) {
6031 if (!SPACE_OR_TAB(ch)) {
6037 if (isIDFIRST_lazy_if(d,UTF)) {
6041 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6043 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6046 Copy(s, d, e - s, char);
6051 while ((isALNUM(*s) || *s == ':') && d < e)
6054 Perl_croak(aTHX_ ident_too_long);
6057 while (s < send && SPACE_OR_TAB(*s)) s++;
6058 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6059 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6060 const char *brack = *s == '[' ? "[...]" : "{...}";
6061 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6062 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6063 funny, dest, brack, funny, dest, brack);
6066 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6070 /* Handle extended ${^Foo} variables
6071 * 1999-02-27 mjd-perl-patch@plover.com */
6072 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6076 while (isALNUM(*s) && d < e) {
6080 Perl_croak(aTHX_ ident_too_long);
6085 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets)
6086 PL_lex_state = LEX_INTERPEND;
6089 if (PL_lex_state == LEX_NORMAL) {
6090 if (ckWARN(WARN_AMBIGUOUS) &&
6091 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6093 Perl_warner(aTHX_ WARN_AMBIGUOUS,
6094 "Ambiguous use of %c{%s} resolved to %c%s",
6095 funny, dest, funny, dest);
6100 s = bracket; /* let the parser handle it */
6104 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6105 PL_lex_state = LEX_INTERPEND;
6110 Perl_pmflag(pTHX_ U16 *pmfl, int ch)
6115 *pmfl |= PMf_GLOBAL;
6117 *pmfl |= PMf_CONTINUE;
6121 *pmfl |= PMf_MULTILINE;
6123 *pmfl |= PMf_SINGLELINE;
6125 *pmfl |= PMf_EXTENDED;
6129 S_scan_pat(pTHX_ char *start, I32 type)
6134 s = scan_str(start,FALSE,FALSE);
6137 SvREFCNT_dec(PL_lex_stuff);
6138 PL_lex_stuff = Nullsv;
6139 Perl_croak(aTHX_ "Search pattern not terminated");
6142 pm = (PMOP*)newPMOP(type, 0);
6143 if (PL_multi_open == '?')
6144 pm->op_pmflags |= PMf_ONCE;
6146 while (*s && strchr("iomsx", *s))
6147 pmflag(&pm->op_pmflags,*s++);
6150 while (*s && strchr("iogcmsx", *s))
6151 pmflag(&pm->op_pmflags,*s++);
6153 pm->op_pmpermflags = pm->op_pmflags;
6155 PL_lex_op = (OP*)pm;
6156 yylval.ival = OP_MATCH;
6161 S_scan_subst(pTHX_ char *start)
6168 yylval.ival = OP_NULL;
6170 s = scan_str(start,FALSE,FALSE);
6174 SvREFCNT_dec(PL_lex_stuff);
6175 PL_lex_stuff = Nullsv;
6176 Perl_croak(aTHX_ "Substitution pattern not terminated");
6179 if (s[-1] == PL_multi_open)
6182 first_start = PL_multi_start;
6183 s = scan_str(s,FALSE,FALSE);
6186 SvREFCNT_dec(PL_lex_stuff);
6187 PL_lex_stuff = Nullsv;
6189 SvREFCNT_dec(PL_lex_repl);
6190 PL_lex_repl = Nullsv;
6191 Perl_croak(aTHX_ "Substitution replacement not terminated");
6193 PL_multi_start = first_start; /* so whole substitution is taken together */
6195 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6201 else if (strchr("iogcmsx", *s))
6202 pmflag(&pm->op_pmflags,*s++);
6209 PL_sublex_info.super_bufptr = s;
6210 PL_sublex_info.super_bufend = PL_bufend;
6212 pm->op_pmflags |= PMf_EVAL;
6213 repl = newSVpvn("",0);
6215 sv_catpv(repl, es ? "eval " : "do ");
6216 sv_catpvn(repl, "{ ", 2);
6217 sv_catsv(repl, PL_lex_repl);
6218 sv_catpvn(repl, " };", 2);
6220 SvREFCNT_dec(PL_lex_repl);
6224 pm->op_pmpermflags = pm->op_pmflags;
6225 PL_lex_op = (OP*)pm;
6226 yylval.ival = OP_SUBST;
6231 S_scan_trans(pTHX_ char *start)
6242 yylval.ival = OP_NULL;
6244 s = scan_str(start,FALSE,FALSE);
6247 SvREFCNT_dec(PL_lex_stuff);
6248 PL_lex_stuff = Nullsv;
6249 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6251 if (s[-1] == PL_multi_open)
6254 s = scan_str(s,FALSE,FALSE);
6257 SvREFCNT_dec(PL_lex_stuff);
6258 PL_lex_stuff = Nullsv;
6260 SvREFCNT_dec(PL_lex_repl);
6261 PL_lex_repl = Nullsv;
6262 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6265 New(803,tbl,256,short);
6266 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6268 complement = del = squash = 0;
6269 while (strchr("cds", *s)) {
6271 complement = OPpTRANS_COMPLEMENT;
6273 del = OPpTRANS_DELETE;
6275 squash = OPpTRANS_SQUASH;
6278 o->op_private = del|squash|complement|
6279 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6280 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6283 yylval.ival = OP_TRANS;
6288 S_scan_heredoc(pTHX_ register char *s)
6291 I32 op_type = OP_SCALAR;
6298 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6302 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6305 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6306 if (*peek && strchr("`'\"",*peek)) {
6309 s = delimcpy(d, e, s, PL_bufend, term, &len);
6319 if (!isALNUM_lazy_if(s,UTF))
6320 deprecate("bare << to mean <<\"\"");
6321 for (; isALNUM_lazy_if(s,UTF); s++) {
6326 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6327 Perl_croak(aTHX_ "Delimiter for here document is too long");
6330 len = d - PL_tokenbuf;
6331 #ifndef PERL_STRICT_CR
6332 d = strchr(s, '\r');
6336 while (s < PL_bufend) {
6342 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6351 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6356 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6357 herewas = newSVpvn(s,PL_bufend-s);
6359 s--, herewas = newSVpvn(s,d-s);
6360 s += SvCUR(herewas);
6362 tmpstr = NEWSV(87,79);
6363 sv_upgrade(tmpstr, SVt_PVIV);
6368 else if (term == '`') {
6369 op_type = OP_BACKTICK;
6370 SvIVX(tmpstr) = '\\';
6374 PL_multi_start = CopLINE(PL_curcop);
6375 PL_multi_open = PL_multi_close = '<';
6376 term = *PL_tokenbuf;
6377 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6378 char *bufptr = PL_sublex_info.super_bufptr;
6379 char *bufend = PL_sublex_info.super_bufend;
6380 char *olds = s - SvCUR(herewas);
6381 s = strchr(bufptr, '\n');
6385 while (s < bufend &&
6386 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6388 CopLINE_inc(PL_curcop);
6391 CopLINE_set(PL_curcop, PL_multi_start);
6392 missingterm(PL_tokenbuf);
6394 sv_setpvn(herewas,bufptr,d-bufptr+1);
6395 sv_setpvn(tmpstr,d+1,s-d);
6397 sv_catpvn(herewas,s,bufend-s);
6398 (void)strcpy(bufptr,SvPVX(herewas));
6405 while (s < PL_bufend &&
6406 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6408 CopLINE_inc(PL_curcop);
6410 if (s >= PL_bufend) {
6411 CopLINE_set(PL_curcop, PL_multi_start);
6412 missingterm(PL_tokenbuf);
6414 sv_setpvn(tmpstr,d+1,s-d);
6416 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6418 sv_catpvn(herewas,s,PL_bufend-s);
6419 sv_setsv(PL_linestr,herewas);
6420 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6421 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6424 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6425 while (s >= PL_bufend) { /* multiple line string? */
6427 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6428 CopLINE_set(PL_curcop, PL_multi_start);
6429 missingterm(PL_tokenbuf);
6431 CopLINE_inc(PL_curcop);
6432 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6433 #ifndef PERL_STRICT_CR
6434 if (PL_bufend - PL_linestart >= 2) {
6435 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6436 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6438 PL_bufend[-2] = '\n';
6440 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6442 else if (PL_bufend[-1] == '\r')
6443 PL_bufend[-1] = '\n';
6445 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6446 PL_bufend[-1] = '\n';
6448 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6449 SV *sv = NEWSV(88,0);
6451 sv_upgrade(sv, SVt_PVMG);
6452 sv_setsv(sv,PL_linestr);
6453 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6455 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6458 sv_catsv(PL_linestr,herewas);
6459 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6463 sv_catsv(tmpstr,PL_linestr);
6468 PL_multi_end = CopLINE(PL_curcop);
6469 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6470 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6471 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6473 SvREFCNT_dec(herewas);
6474 if (UTF && !IN_BYTE && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6476 PL_lex_stuff = tmpstr;
6477 yylval.ival = op_type;
6482 takes: current position in input buffer
6483 returns: new position in input buffer
6484 side-effects: yylval and lex_op are set.
6489 <FH> read from filehandle
6490 <pkg::FH> read from package qualified filehandle
6491 <pkg'FH> read from package qualified filehandle
6492 <$fh> read from filehandle in $fh
6498 S_scan_inputsymbol(pTHX_ char *start)
6500 register char *s = start; /* current position in buffer */
6506 d = PL_tokenbuf; /* start of temp holding space */
6507 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6508 end = strchr(s, '\n');
6511 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6513 /* die if we didn't have space for the contents of the <>,
6514 or if it didn't end, or if we see a newline
6517 if (len >= sizeof PL_tokenbuf)
6518 Perl_croak(aTHX_ "Excessively long <> operator");
6520 Perl_croak(aTHX_ "Unterminated <> operator");
6525 Remember, only scalar variables are interpreted as filehandles by
6526 this code. Anything more complex (e.g., <$fh{$num}>) will be
6527 treated as a glob() call.
6528 This code makes use of the fact that except for the $ at the front,
6529 a scalar variable and a filehandle look the same.
6531 if (*d == '$' && d[1]) d++;
6533 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6534 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6537 /* If we've tried to read what we allow filehandles to look like, and
6538 there's still text left, then it must be a glob() and not a getline.
6539 Use scan_str to pull out the stuff between the <> and treat it
6540 as nothing more than a string.
6543 if (d - PL_tokenbuf != len) {
6544 yylval.ival = OP_GLOB;
6546 s = scan_str(start,FALSE,FALSE);
6548 Perl_croak(aTHX_ "Glob not terminated");
6552 /* we're in a filehandle read situation */
6555 /* turn <> into <ARGV> */
6557 (void)strcpy(d,"ARGV");
6559 /* if <$fh>, create the ops to turn the variable into a
6565 /* try to find it in the pad for this block, otherwise find
6566 add symbol table ops
6568 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6569 OP *o = newOP(OP_PADSV, 0);
6571 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o);
6574 GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV);
6575 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0,
6576 newUNOP(OP_RV2SV, 0,
6577 newGVOP(OP_GV, 0, gv)));
6579 PL_lex_op->op_flags |= OPf_SPECIAL;
6580 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6581 yylval.ival = OP_NULL;
6584 /* If it's none of the above, it must be a literal filehandle
6585 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6587 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6588 PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6589 yylval.ival = OP_NULL;
6598 takes: start position in buffer
6599 keep_quoted preserve \ on the embedded delimiter(s)
6600 keep_delims preserve the delimiters around the string
6601 returns: position to continue reading from buffer
6602 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6603 updates the read buffer.
6605 This subroutine pulls a string out of the input. It is called for:
6606 q single quotes q(literal text)
6607 ' single quotes 'literal text'
6608 qq double quotes qq(interpolate $here please)
6609 " double quotes "interpolate $here please"
6610 qx backticks qx(/bin/ls -l)
6611 ` backticks `/bin/ls -l`
6612 qw quote words @EXPORT_OK = qw( func() $spam )
6613 m// regexp match m/this/
6614 s/// regexp substitute s/this/that/
6615 tr/// string transliterate tr/this/that/
6616 y/// string transliterate y/this/that/
6617 ($*@) sub prototypes sub foo ($)
6618 (stuff) sub attr parameters sub foo : attr(stuff)
6619 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6621 In most of these cases (all but <>, patterns and transliterate)
6622 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6623 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6624 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6627 It skips whitespace before the string starts, and treats the first
6628 character as the delimiter. If the delimiter is one of ([{< then
6629 the corresponding "close" character )]}> is used as the closing
6630 delimiter. It allows quoting of delimiters, and if the string has
6631 balanced delimiters ([{<>}]) it allows nesting.
6633 The lexer always reads these strings into lex_stuff, except in the
6634 case of the operators which take *two* arguments (s/// and tr///)
6635 when it checks to see if lex_stuff is full (presumably with the 1st
6636 arg to s or tr) and if so puts the string into lex_repl.
6641 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6643 SV *sv; /* scalar value: string */
6644 char *tmps; /* temp string, used for delimiter matching */
6645 register char *s = start; /* current position in the buffer */
6646 register char term; /* terminating character */
6647 register char *to; /* current position in the sv's data */
6648 I32 brackets = 1; /* bracket nesting level */
6649 bool has_utf8 = FALSE; /* is there any utf8 content? */
6651 /* skip space before the delimiter */
6655 /* mark where we are, in case we need to report errors */
6658 /* after skipping whitespace, the next character is the terminator */
6660 if (UTF8_IS_CONTINUED(term) && UTF)
6663 /* mark where we are */
6664 PL_multi_start = CopLINE(PL_curcop);
6665 PL_multi_open = term;
6667 /* find corresponding closing delimiter */
6668 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6670 PL_multi_close = term;
6672 /* create a new SV to hold the contents. 87 is leak category, I'm
6673 assuming. 79 is the SV's initial length. What a random number. */
6675 sv_upgrade(sv, SVt_PVIV);
6677 (void)SvPOK_only(sv); /* validate pointer */
6679 /* move past delimiter and try to read a complete string */
6681 sv_catpvn(sv, s, 1);
6684 /* extend sv if need be */
6685 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
6686 /* set 'to' to the next character in the sv's string */
6687 to = SvPVX(sv)+SvCUR(sv);
6689 /* if open delimiter is the close delimiter read unbridle */
6690 if (PL_multi_open == PL_multi_close) {
6691 for (; s < PL_bufend; s++,to++) {
6692 /* embedded newlines increment the current line number */
6693 if (*s == '\n' && !PL_rsfp)
6694 CopLINE_inc(PL_curcop);
6695 /* handle quoted delimiters */
6696 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
6697 if (!keep_quoted && s[1] == term)
6699 /* any other quotes are simply copied straight through */
6703 /* terminate when run out of buffer (the for() condition), or
6704 have found the terminator */
6705 else if (*s == term)
6707 else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF)
6713 /* if the terminator isn't the same as the start character (e.g.,
6714 matched brackets), we have to allow more in the quoting, and
6715 be prepared for nested brackets.
6718 /* read until we run out of string, or we find the terminator */
6719 for (; s < PL_bufend; s++,to++) {
6720 /* embedded newlines increment the line count */
6721 if (*s == '\n' && !PL_rsfp)
6722 CopLINE_inc(PL_curcop);
6723 /* backslashes can escape the open or closing characters */
6724 if (*s == '\\' && s+1 < PL_bufend) {
6726 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
6731 /* allow nested opens and closes */
6732 else if (*s == PL_multi_close && --brackets <= 0)
6734 else if (*s == PL_multi_open)
6736 else if (!has_utf8 && UTF8_IS_CONTINUED(*s) && UTF)
6741 /* terminate the copied string and update the sv's end-of-string */
6743 SvCUR_set(sv, to - SvPVX(sv));
6746 * this next chunk reads more into the buffer if we're not done yet
6750 break; /* handle case where we are done yet :-) */
6752 #ifndef PERL_STRICT_CR
6753 if (to - SvPVX(sv) >= 2) {
6754 if ((to[-2] == '\r' && to[-1] == '\n') ||
6755 (to[-2] == '\n' && to[-1] == '\r'))
6759 SvCUR_set(sv, to - SvPVX(sv));
6761 else if (to[-1] == '\r')
6764 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
6768 /* if we're out of file, or a read fails, bail and reset the current
6769 line marker so we can report where the unterminated string began
6772 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6774 CopLINE_set(PL_curcop, PL_multi_start);
6777 /* we read a line, so increment our line counter */
6778 CopLINE_inc(PL_curcop);
6780 /* update debugger info */
6781 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6782 SV *sv = NEWSV(88,0);
6784 sv_upgrade(sv, SVt_PVMG);
6785 sv_setsv(sv,PL_linestr);
6786 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
6789 /* having changed the buffer, we must update PL_bufend */
6790 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6793 /* at this point, we have successfully read the delimited string */
6796 sv_catpvn(sv, s, 1);
6799 PL_multi_end = CopLINE(PL_curcop);
6802 /* if we allocated too much space, give some back */
6803 if (SvCUR(sv) + 5 < SvLEN(sv)) {
6804 SvLEN_set(sv, SvCUR(sv) + 1);
6805 Renew(SvPVX(sv), SvLEN(sv), char);
6808 /* decide whether this is the first or second quoted string we've read
6821 takes: pointer to position in buffer
6822 returns: pointer to new position in buffer
6823 side-effects: builds ops for the constant in yylval.op
6825 Read a number in any of the formats that Perl accepts:
6827 0(x[0-7A-F]+)|([0-7]+)|(b[01])
6828 [\d_]+(\.[\d_]*)?[Ee](\d+)
6830 Underbars (_) are allowed in decimal numbers. If -w is on,
6831 underbars before a decimal point must be at three digit intervals.
6833 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
6836 If it reads a number without a decimal point or an exponent, it will
6837 try converting the number to an integer and see if it can do so
6838 without loss of precision.
6842 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
6844 register char *s = start; /* current position in buffer */
6845 register char *d; /* destination in temp buffer */
6846 register char *e; /* end of temp buffer */
6847 NV nv; /* number read, as a double */
6848 SV *sv = Nullsv; /* place to put the converted number */
6849 bool floatit; /* boolean: int or float? */
6850 char *lastub = 0; /* position of last underbar */
6851 static char number_too_long[] = "Number too long";
6853 /* We use the first character to decide what type of number this is */
6857 Perl_croak(aTHX_ "panic: scan_num");
6859 /* if it starts with a 0, it could be an octal number, a decimal in
6860 0.13 disguise, or a hexadecimal number, or a binary number. */
6864 u holds the "number so far"
6865 shift the power of 2 of the base
6866 (hex == 4, octal == 3, binary == 1)
6867 overflowed was the number more than we can hold?
6869 Shift is used when we add a digit. It also serves as an "are
6870 we in octal/hex/binary?" indicator to disallow hex characters
6876 bool overflowed = FALSE;
6877 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
6878 static char* bases[5] = { "", "binary", "", "octal",
6880 static char* Bases[5] = { "", "Binary", "", "Octal",
6882 static char *maxima[5] = { "",
6883 "0b11111111111111111111111111111111",
6887 char *base, *Base, *max;
6893 } else if (s[1] == 'b') {
6897 /* check for a decimal in disguise */
6898 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
6900 /* so it must be octal */
6904 base = bases[shift];
6905 Base = Bases[shift];
6906 max = maxima[shift];
6908 /* read the rest of the number */
6910 /* x is used in the overflow test,
6911 b is the digit we're adding on. */
6916 /* if we don't mention it, we're done */
6925 /* 8 and 9 are not octal */
6928 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
6932 case '2': case '3': case '4':
6933 case '5': case '6': case '7':
6935 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
6939 b = *s++ & 15; /* ASCII digit -> value of digit */
6943 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
6944 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
6945 /* make sure they said 0x */
6950 /* Prepare to put the digit we have onto the end
6951 of the number so far. We check for overflows.
6956 x = u << shift; /* make room for the digit */
6958 if ((x >> shift) != u
6959 && !(PL_hints & HINT_NEW_BINARY)) {
6962 if (ckWARN_d(WARN_OVERFLOW))
6963 Perl_warner(aTHX_ WARN_OVERFLOW,
6964 "Integer overflow in %s number",
6967 u = x | b; /* add the digit to the end */
6970 n *= nvshift[shift];
6971 /* If an NV has not enough bits in its
6972 * mantissa to represent an UV this summing of
6973 * small low-order numbers is a waste of time
6974 * (because the NV cannot preserve the
6975 * low-order bits anyway): we could just
6976 * remember when did we overflow and in the
6977 * end just multiply n by the right
6985 /* if we get here, we had success: make a scalar value from
6991 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
6992 Perl_warner(aTHX_ WARN_PORTABLE,
6993 "%s number > %s non-portable",
6999 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7000 Perl_warner(aTHX_ WARN_PORTABLE,
7001 "%s number > %s non-portable",
7006 if (PL_hints & HINT_NEW_BINARY)
7007 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7012 handle decimal numbers.
7013 we're also sent here when we read a 0 as the first digit
7015 case '1': case '2': case '3': case '4': case '5':
7016 case '6': case '7': case '8': case '9': case '.':
7019 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7022 /* read next group of digits and _ and copy into d */
7023 while (isDIGIT(*s) || *s == '_') {
7024 /* skip underscores, checking for misplaced ones
7028 if (ckWARN(WARN_SYNTAX) && lastub && s - lastub != 3)
7029 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7033 /* check for end of fixed-length buffer */
7035 Perl_croak(aTHX_ number_too_long);
7036 /* if we're ok, copy the character */
7041 /* final misplaced underbar check */
7042 if (lastub && s - lastub != 3) {
7043 if (ckWARN(WARN_SYNTAX))
7044 Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number");
7047 /* read a decimal portion if there is one. avoid
7048 3..5 being interpreted as the number 3. followed
7051 if (*s == '.' && s[1] != '.') {
7055 /* copy, ignoring underbars, until we run out of
7056 digits. Note: no misplaced underbar checks!
7058 for (; isDIGIT(*s) || *s == '_'; s++) {
7059 /* fixed length buffer check */
7061 Perl_croak(aTHX_ number_too_long);
7065 if (*s == '.' && isDIGIT(s[1])) {
7066 /* oops, it's really a v-string, but without the "v" */
7072 /* read exponent part, if present */
7073 if (*s && strchr("eE",*s) && strchr("+-0123456789",s[1])) {
7077 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7078 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7080 /* allow positive or negative exponent */
7081 if (*s == '+' || *s == '-')
7084 /* read digits of exponent (no underbars :-) */
7085 while (isDIGIT(*s)) {
7087 Perl_croak(aTHX_ number_too_long);
7092 /* terminate the string */
7095 /* make an sv from the string */
7098 #if defined(Strtol) && defined(Strtoul)
7101 strtol/strtoll sets errno to ERANGE if the number is too big
7102 for an integer. We try to do an integer conversion first
7103 if no characters indicating "float" have been found.
7110 if (*PL_tokenbuf == '-')
7111 iv = Strtol(PL_tokenbuf, (char**)NULL, 10);
7113 uv = Strtoul(PL_tokenbuf, (char**)NULL, 10);
7115 floatit = TRUE; /* Probably just too large. */
7116 else if (*PL_tokenbuf == '-')
7118 else if (uv <= IV_MAX)
7119 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7124 nv = Atof(PL_tokenbuf);
7129 No working strtou?ll?.
7131 Unfortunately atol() doesn't do range checks (returning
7132 LONG_MIN/LONG_MAX, and setting errno to ERANGE on overflows)
7133 everywhere [1], so we cannot use use atol() (or atoll()).
7134 If we could, they would be used, as Atol(), very much like
7135 Strtol() and Strtoul() are used above.
7137 [1] XXX Configure test needed to check for atol()
7138 (and atoll()) overflow behaviour XXX
7142 We need to do this the hard way. */
7144 nv = Atof(PL_tokenbuf);
7146 /* See if we can make do with an integer value without loss of
7147 precision. We use U_V to cast to a UV, because some
7148 compilers have issues. Then we try casting it back and see
7149 if it was the same [1]. We only do this if we know we
7150 specifically read an integer. If floatit is true, then we
7151 don't need to do the conversion at all.
7153 [1] Note that this is lossy if our NVs cannot preserve our
7154 UVs. There are metaconfig defines NV_PRESERVES_UV (a boolean)
7155 and NV_PRESERVES_UV_BITS (a number), but in general we really
7156 do hope all such potentially lossy platforms have strtou?ll?
7157 to do a lossless IV/UV conversion.
7159 Maybe could do some tricks with DBL_DIG, LDBL_DIG and
7160 DBL_MANT_DIG and LDBL_MANT_DIG (these are already available
7161 as NV_DIG and NV_MANT_DIG)?
7167 if (!floatit && (NV)uv == nv) {
7169 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7177 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7178 (PL_hints & HINT_NEW_INTEGER) )
7179 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7180 (floatit ? "float" : "integer"),
7184 /* if it starts with a v, it could be a v-string */
7190 while (isDIGIT(*pos) || *pos == '_')
7192 if (!isALPHA(*pos)) {
7194 U8 tmpbuf[UTF8_MAXLEN+1];
7197 s++; /* get past 'v' */
7200 sv_setpvn(sv, "", 0);
7203 if (*s == '0' && isDIGIT(s[1]))
7204 yyerror("Octal number in vector unsupported");
7207 /* this is atoi() that tolerates underscores */
7210 while (--end >= s) {
7215 rev += (*end - '0') * mult;
7217 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
7218 Perl_warner(aTHX_ WARN_OVERFLOW,
7219 "Integer overflow in decimal number");
7222 tmpend = uv_to_utf8(tmpbuf, rev);
7223 utf8 = utf8 || rev > 127;
7224 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
7225 if (*pos == '.' && isDIGIT(pos[1]))
7231 while (isDIGIT(*pos) || *pos == '_')
7240 sv_utf8_downgrade(sv, TRUE);
7247 /* make the op for the constant and return */
7250 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7252 lvalp->opval = Nullop;
7258 S_scan_formline(pTHX_ register char *s)
7262 SV *stuff = newSVpvn("",0);
7263 bool needargs = FALSE;
7266 if (*s == '.' || *s == /*{*/'}') {
7268 #ifdef PERL_STRICT_CR
7269 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7271 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7273 if (*t == '\n' || t == PL_bufend)
7276 if (PL_in_eval && !PL_rsfp) {
7277 eol = strchr(s,'\n');
7282 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7284 for (t = s; t < eol; t++) {
7285 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7287 goto enough; /* ~~ must be first line in formline */
7289 if (*t == '@' || *t == '^')
7292 sv_catpvn(stuff, s, eol-s);
7293 #ifndef PERL_STRICT_CR
7294 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7295 char *end = SvPVX(stuff) + SvCUR(stuff);
7304 s = filter_gets(PL_linestr, PL_rsfp, 0);
7305 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7306 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7309 yyerror("Format not terminated");
7319 PL_lex_state = LEX_NORMAL;
7320 PL_nextval[PL_nexttoke].ival = 0;
7324 PL_lex_state = LEX_FORMLINE;
7325 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7327 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7331 SvREFCNT_dec(stuff);
7332 PL_lex_formbrack = 0;
7343 PL_cshlen = strlen(PL_cshname);
7348 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7350 I32 oldsavestack_ix = PL_savestack_ix;
7351 CV* outsidecv = PL_compcv;
7355 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7357 SAVEI32(PL_subline);
7358 save_item(PL_subname);
7361 SAVESPTR(PL_comppad_name);
7362 SAVESPTR(PL_compcv);
7363 SAVEI32(PL_comppad_name_fill);
7364 SAVEI32(PL_min_intro_pending);
7365 SAVEI32(PL_max_intro_pending);
7366 SAVEI32(PL_pad_reset_pending);
7368 PL_compcv = (CV*)NEWSV(1104,0);
7369 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7370 CvFLAGS(PL_compcv) |= flags;
7372 PL_comppad = newAV();
7373 av_push(PL_comppad, Nullsv);
7374 PL_curpad = AvARRAY(PL_comppad);
7375 PL_comppad_name = newAV();
7376 PL_comppad_name_fill = 0;
7377 PL_min_intro_pending = 0;
7379 PL_subline = CopLINE(PL_curcop);
7381 av_store(PL_comppad_name, 0, newSVpvn("@_", 2));
7382 PL_curpad[0] = (SV*)newAV();
7383 SvPADMY_on(PL_curpad[0]); /* XXX Needed? */
7384 #endif /* USE_THREADS */
7386 comppadlist = newAV();
7387 AvREAL_off(comppadlist);
7388 av_store(comppadlist, 0, (SV*)PL_comppad_name);
7389 av_store(comppadlist, 1, (SV*)PL_comppad);
7391 CvPADLIST(PL_compcv) = comppadlist;
7392 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7394 CvOWNER(PL_compcv) = 0;
7395 New(666, CvMUTEXP(PL_compcv), 1, perl_mutex);
7396 MUTEX_INIT(CvMUTEXP(PL_compcv));
7397 #endif /* USE_THREADS */
7399 return oldsavestack_ix;
7403 Perl_yywarn(pTHX_ char *s)
7405 PL_in_eval |= EVAL_WARNONLY;
7407 PL_in_eval &= ~EVAL_WARNONLY;
7412 Perl_yyerror(pTHX_ char *s)
7415 char *context = NULL;
7419 if (!yychar || (yychar == ';' && !PL_rsfp))
7421 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7422 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7423 while (isSPACE(*PL_oldoldbufptr))
7425 context = PL_oldoldbufptr;
7426 contlen = PL_bufptr - PL_oldoldbufptr;
7428 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7429 PL_oldbufptr != PL_bufptr) {
7430 while (isSPACE(*PL_oldbufptr))
7432 context = PL_oldbufptr;
7433 contlen = PL_bufptr - PL_oldbufptr;
7435 else if (yychar > 255)
7436 where = "next token ???";
7437 #ifdef USE_PURE_BISON
7438 /* GNU Bison sets the value -2 */
7439 else if (yychar == -2) {
7441 else if ((yychar & 127) == 127) {
7443 if (PL_lex_state == LEX_NORMAL ||
7444 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7445 where = "at end of line";
7446 else if (PL_lex_inpat)
7447 where = "within pattern";
7449 where = "within string";
7452 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7454 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7455 else if (isPRINT_LC(yychar))
7456 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7458 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7459 where = SvPVX(where_sv);
7461 msg = sv_2mortal(newSVpv(s, 0));
7462 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7463 CopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7465 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7467 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7468 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7469 Perl_sv_catpvf(aTHX_ msg,
7470 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7471 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7474 if (PL_in_eval & EVAL_WARNONLY)
7475 Perl_warn(aTHX_ "%"SVf, msg);
7478 if (PL_error_count >= 10) {
7479 if (PL_in_eval && SvCUR(ERRSV))
7480 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7481 ERRSV, CopFILE(PL_curcop));
7483 Perl_croak(aTHX_ "%s has too many errors.\n",
7484 CopFILE(PL_curcop));
7487 PL_in_my_stash = Nullhv;
7492 S_swallow_bom(pTHX_ U8 *s)
7495 slen = SvCUR(PL_linestr);
7499 /* UTF-16 little-endian */
7500 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7501 Perl_croak(aTHX_ "Unsupported script encoding");
7502 #ifndef PERL_NO_UTF16_FILTER
7503 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n"));
7505 if (PL_bufend > (char*)s) {
7509 filter_add(utf16rev_textfilter, NULL);
7510 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7511 PL_bufend = (char*)utf16_to_utf8_reversed(s, news,
7512 PL_bufend - (char*)s - 1,
7514 Copy(news, s, newlen, U8);
7515 SvCUR_set(PL_linestr, newlen);
7516 PL_bufend = SvPVX(PL_linestr) + newlen;
7517 news[newlen++] = '\0';
7521 Perl_croak(aTHX_ "Unsupported script encoding");
7526 if (s[1] == 0xFF) { /* UTF-16 big-endian */
7527 #ifndef PERL_NO_UTF16_FILTER
7528 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n"));
7530 if (PL_bufend > (char *)s) {
7534 filter_add(utf16_textfilter, NULL);
7535 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7536 PL_bufend = (char*)utf16_to_utf8(s, news,
7537 PL_bufend - (char*)s,
7539 Copy(news, s, newlen, U8);
7540 SvCUR_set(PL_linestr, newlen);
7541 PL_bufend = SvPVX(PL_linestr) + newlen;
7542 news[newlen++] = '\0';
7546 Perl_croak(aTHX_ "Unsupported script encoding");
7551 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7552 DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n"));
7557 if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */
7558 s[2] == 0xFE && s[3] == 0xFF)
7560 Perl_croak(aTHX_ "Unsupported script encoding");
7572 * Restore a source filter.
7576 restore_rsfp(pTHXo_ void *f)
7578 PerlIO *fp = (PerlIO*)f;
7580 if (PL_rsfp == PerlIO_stdin())
7581 PerlIO_clearerr(PL_rsfp);
7582 else if (PL_rsfp && (PL_rsfp != fp))
7583 PerlIO_close(PL_rsfp);
7587 #ifndef PERL_NO_UTF16_FILTER
7589 utf16_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7591 I32 count = FILTER_READ(idx+1, sv, maxlen);
7596 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7597 if (!*SvPV_nolen(sv))
7598 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7601 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7602 sv_usepvn(sv, (char*)tmps, tend - tmps);
7608 utf16rev_textfilter(pTHXo_ int idx, SV *sv, int maxlen)
7610 I32 count = FILTER_READ(idx+1, sv, maxlen);
7615 if (!*SvPV_nolen(sv))
7616 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7619 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7620 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7621 sv_usepvn(sv, (char*)tmps, tend - tmps);