3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define yychar (*PL_yycharp)
27 #define yylval (*PL_yylvalp)
29 static char ident_too_long[] = "Identifier too long";
30 static char c_without_g[] = "Use of /c modifier is meaningless without /g";
31 static char c_in_subst[] = "Use of /c modifier is meaningless in s///";
33 static void restore_rsfp(pTHX_ void *f);
34 #ifndef PERL_NO_UTF16_FILTER
35 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
36 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
39 #define XFAKEBRACK 128
42 #ifdef USE_UTF8_SCRIPTS
43 # define UTF (!IN_BYTES)
45 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
48 /* In variables named $^X, these are the legal values for X.
49 * 1999-02-27 mjd-perl-patch@plover.com */
50 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
52 /* On MacOS, respect nonbreaking spaces */
53 #ifdef MACOS_TRADITIONAL
54 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
56 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
59 /* LEX_* are values for PL_lex_state, the state of the lexer.
60 * They are arranged oddly so that the guard on the switch statement
61 * can get by with a single comparison (if the compiler is smart enough).
64 /* #define LEX_NOTPARSING 11 is done in perl.h. */
67 #define LEX_INTERPNORMAL 9
68 #define LEX_INTERPCASEMOD 8
69 #define LEX_INTERPPUSH 7
70 #define LEX_INTERPSTART 6
71 #define LEX_INTERPEND 5
72 #define LEX_INTERPENDMAYBE 4
73 #define LEX_INTERPCONCAT 3
74 #define LEX_INTERPCONST 2
75 #define LEX_FORMLINE 1
76 #define LEX_KNOWNEXT 0
84 /* CLINE is a macro that ensures PL_copline has a sane value */
89 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
92 * Convenience functions to return different tokens and prime the
93 * lexer for the next token. They all take an argument.
95 * TOKEN : generic token (used for '(', DOLSHARP, etc)
96 * OPERATOR : generic operator
97 * AOPERATOR : assignment operator
98 * PREBLOCK : beginning the block after an if, while, foreach, ...
99 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
100 * PREREF : *EXPR where EXPR is not a simple identifier
101 * TERM : expression term
102 * LOOPX : loop exiting command (goto, last, dump, etc)
103 * FTST : file test operator
104 * FUN0 : zero-argument function
105 * FUN1 : not used, except for not, which isn't a UNIOP
106 * BOop : bitwise or or xor
108 * SHop : shift operator
109 * PWop : power operator
110 * PMop : pattern-matching operator
111 * Aop : addition-level operator
112 * Mop : multiplication-level operator
113 * Eop : equality-testing operator
114 * Rop : relational operator <= != gt
116 * Also see LOP and lop() below.
119 /* Note that REPORT() and REPORT2() will be expressions that supply
120 * their own trailing comma, not suitable for statements as such. */
121 #ifdef DEBUGGING /* Serve -DT. */
122 # define REPORT(x,retval) tokereport(x,s,(int)retval),
123 # define REPORT2(x,retval) tokereport(x,s, yylval.ival),
125 # define REPORT(x,retval)
126 # define REPORT2(x,retval)
129 #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval)
130 #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)
131 #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval))
132 #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval)
133 #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval)
134 #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval)
135 #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval)
136 #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX)
137 #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERMORDORDOR,PL_bufptr = s,(int)UNIOP)
138 #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0)
139 #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1)
140 #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP))
141 #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP))
142 #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP))
143 #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP))
144 #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP)
145 #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP))
146 #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP))
147 #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP)
148 #define Rop(f) return(yylval.ival=f, REPORT("rel",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.
152 * The UNIDOR macro is for unary functions that can be followed by the //
153 * operator (such as C<shift // 0>).
155 #define UNI2(f,x) return(yylval.ival = f, \
159 PL_last_uni = PL_oldbufptr, \
160 PL_last_lop_op = f, \
161 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
162 #define UNI(f) UNI2(f,XTERM)
163 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
165 #define UNIBRACK(f) return(yylval.ival = f, \
168 PL_last_uni = PL_oldbufptr, \
169 (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) )
171 /* grandfather return to old style */
172 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
177 S_tokereport(pTHX_ char *thing, char* s, I32 rv)
180 SV* report = newSVpv(thing, 0);
181 Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop),
184 if (s - PL_bufptr > 0)
185 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
187 if (PL_oldbufptr && *PL_oldbufptr)
188 sv_catpv(report, PL_tokenbuf);
190 PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report));
199 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
200 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
204 S_ao(pTHX_ int toketype)
206 if (*PL_bufptr == '=') {
208 if (toketype == ANDAND)
209 yylval.ival = OP_ANDASSIGN;
210 else if (toketype == OROR)
211 yylval.ival = OP_ORASSIGN;
212 else if (toketype == DORDOR)
213 yylval.ival = OP_DORASSIGN;
221 * When Perl expects an operator and finds something else, no_op
222 * prints the warning. It always prints "<something> found where
223 * operator expected. It prints "Missing semicolon on previous line?"
224 * if the surprise occurs at the start of the line. "do you need to
225 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
226 * where the compiler doesn't know if foo is a method call or a function.
227 * It prints "Missing operator before end of line" if there's nothing
228 * after the missing operator, or "... before <...>" if there is something
229 * after the missing operator.
233 S_no_op(pTHX_ char *what, char *s)
235 char *oldbp = PL_bufptr;
236 bool is_first = (PL_oldbufptr == PL_linestart);
242 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
243 if (ckWARN_d(WARN_SYNTAX)) {
245 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
246 "\t(Missing semicolon on previous line?)\n");
247 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
249 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
250 if (t < PL_bufptr && isSPACE(*t))
251 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
252 "\t(Do you need to predeclare %.*s?)\n",
253 t - PL_oldoldbufptr, PL_oldoldbufptr);
257 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
258 "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
266 * Complain about missing quote/regexp/heredoc terminator.
267 * If it's called with (char *)NULL then it cauterizes the line buffer.
268 * If we're in a delimited string and the delimiter is a control
269 * character, it's reformatted into a two-char sequence like ^C.
274 S_missingterm(pTHX_ char *s)
279 char *nl = strrchr(s,'\n');
285 iscntrl(PL_multi_close)
287 PL_multi_close < 32 || PL_multi_close == 127
291 tmpbuf[1] = toCTRL(PL_multi_close);
297 *tmpbuf = (char)PL_multi_close;
301 q = strchr(s,'"') ? '\'' : '"';
302 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
310 Perl_deprecate(pTHX_ char *s)
312 if (ckWARN(WARN_DEPRECATED))
313 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
317 Perl_deprecate_old(pTHX_ char *s)
319 /* This function should NOT be called for any new deprecated warnings */
320 /* Use Perl_deprecate instead */
322 /* It is here to maintain backward compatibility with the pre-5.8 */
323 /* warnings category hierarchy. The "deprecated" category used to */
324 /* live under the "syntax" category. It is now a top-level category */
325 /* in its own right. */
327 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
328 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
329 "Use of %s is deprecated", s);
334 * Deprecate a comma-less variable list.
340 deprecate_old("comma-less variable list");
344 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
345 * utf16-to-utf8-reversed.
348 #ifdef PERL_CR_FILTER
352 register char *s = SvPVX(sv);
353 register char *e = s + SvCUR(sv);
354 /* outer loop optimized to do nothing if there are no CR-LFs */
356 if (*s++ == '\r' && *s == '\n') {
357 /* hit a CR-LF, need to copy the rest */
358 register char *d = s - 1;
361 if (*s == '\r' && s[1] == '\n')
372 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
374 I32 count = FILTER_READ(idx+1, sv, maxlen);
375 if (count > 0 && !maxlen)
383 * Initialize variables. Uses the Perl save_stack to save its state (for
384 * recursive calls to the parser).
388 Perl_lex_start(pTHX_ SV *line)
393 SAVEI32(PL_lex_dojoin);
394 SAVEI32(PL_lex_brackets);
395 SAVEI32(PL_lex_casemods);
396 SAVEI32(PL_lex_starts);
397 SAVEI32(PL_lex_state);
398 SAVEVPTR(PL_lex_inpat);
399 SAVEI32(PL_lex_inwhat);
400 if (PL_lex_state == LEX_KNOWNEXT) {
401 I32 toke = PL_nexttoke;
402 while (--toke >= 0) {
403 SAVEI32(PL_nexttype[toke]);
404 SAVEVPTR(PL_nextval[toke]);
406 SAVEI32(PL_nexttoke);
408 SAVECOPLINE(PL_curcop);
411 SAVEPPTR(PL_oldbufptr);
412 SAVEPPTR(PL_oldoldbufptr);
413 SAVEPPTR(PL_last_lop);
414 SAVEPPTR(PL_last_uni);
415 SAVEPPTR(PL_linestart);
416 SAVESPTR(PL_linestr);
417 SAVEGENERICPV(PL_lex_brackstack);
418 SAVEGENERICPV(PL_lex_casestack);
419 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
420 SAVESPTR(PL_lex_stuff);
421 SAVEI32(PL_lex_defer);
422 SAVEI32(PL_sublex_info.sub_inwhat);
423 SAVESPTR(PL_lex_repl);
425 SAVEINT(PL_lex_expect);
427 PL_lex_state = LEX_NORMAL;
431 New(899, PL_lex_brackstack, 120, char);
432 New(899, PL_lex_casestack, 12, char);
434 *PL_lex_casestack = '\0';
437 PL_lex_stuff = Nullsv;
438 PL_lex_repl = Nullsv;
442 PL_sublex_info.sub_inwhat = 0;
444 if (SvREADONLY(PL_linestr))
445 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
446 s = SvPV(PL_linestr, len);
447 if (!len || s[len-1] != ';') {
448 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
449 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
450 sv_catpvn(PL_linestr, "\n;", 2);
452 SvTEMP_off(PL_linestr);
453 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
454 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
455 PL_last_lop = PL_last_uni = Nullch;
461 * Finalizer for lexing operations. Must be called when the parser is
462 * done with the lexer.
468 PL_doextract = FALSE;
473 * This subroutine has nothing to do with tilting, whether at windmills
474 * or pinball tables. Its name is short for "increment line". It
475 * increments the current line number in CopLINE(PL_curcop) and checks
476 * to see whether the line starts with a comment of the form
477 * # line 500 "foo.pm"
478 * If so, it sets the current line number and file to the values in the comment.
482 S_incline(pTHX_ char *s)
489 CopLINE_inc(PL_curcop);
492 while (SPACE_OR_TAB(*s)) s++;
493 if (strnEQ(s, "line", 4))
497 if (SPACE_OR_TAB(*s))
501 while (SPACE_OR_TAB(*s)) s++;
507 while (SPACE_OR_TAB(*s))
509 if (*s == '"' && (t = strchr(s+1, '"'))) {
514 for (t = s; !isSPACE(*t); t++) ;
517 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
519 if (*e != '\n' && *e != '\0')
520 return; /* false alarm */
525 CopFILE_free(PL_curcop);
526 CopFILE_set(PL_curcop, s);
529 CopLINE_set(PL_curcop, atoi(n)-1);
534 * Called to gobble the appropriate amount and type of whitespace.
535 * Skips comments as well.
539 S_skipspace(pTHX_ register char *s)
541 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
542 while (s < PL_bufend && SPACE_OR_TAB(*s))
548 SSize_t oldprevlen, oldoldprevlen;
549 SSize_t oldloplen = 0, oldunilen = 0;
550 while (s < PL_bufend && isSPACE(*s)) {
551 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
556 if (s < PL_bufend && *s == '#') {
557 while (s < PL_bufend && *s != '\n')
561 if (PL_in_eval && !PL_rsfp) {
568 /* only continue to recharge the buffer if we're at the end
569 * of the buffer, we're not reading from a source filter, and
570 * we're in normal lexing mode
572 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
573 PL_lex_state == LEX_FORMLINE)
576 /* try to recharge the buffer */
577 if ((s = filter_gets(PL_linestr, PL_rsfp,
578 (prevlen = SvCUR(PL_linestr)))) == Nullch)
580 /* end of file. Add on the -p or -n magic */
581 if (PL_minus_n || PL_minus_p) {
582 sv_setpv(PL_linestr,PL_minus_p ?
583 ";}continue{print or die qq(-p destination: $!\\n)" :
585 sv_catpv(PL_linestr,";}");
586 PL_minus_n = PL_minus_p = 0;
589 sv_setpv(PL_linestr,";");
591 /* reset variables for next time we lex */
592 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
594 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
595 PL_last_lop = PL_last_uni = Nullch;
597 /* Close the filehandle. Could be from -P preprocessor,
598 * STDIN, or a regular file. If we were reading code from
599 * STDIN (because the commandline held no -e or filename)
600 * then we don't close it, we reset it so the code can
601 * read from STDIN too.
604 if (PL_preprocess && !PL_in_eval)
605 (void)PerlProc_pclose(PL_rsfp);
606 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
607 PerlIO_clearerr(PL_rsfp);
609 (void)PerlIO_close(PL_rsfp);
614 /* not at end of file, so we only read another line */
615 /* make corresponding updates to old pointers, for yyerror() */
616 oldprevlen = PL_oldbufptr - PL_bufend;
617 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
619 oldunilen = PL_last_uni - PL_bufend;
621 oldloplen = PL_last_lop - PL_bufend;
622 PL_linestart = PL_bufptr = s + prevlen;
623 PL_bufend = s + SvCUR(PL_linestr);
625 PL_oldbufptr = s + oldprevlen;
626 PL_oldoldbufptr = s + oldoldprevlen;
628 PL_last_uni = s + oldunilen;
630 PL_last_lop = s + oldloplen;
633 /* debugger active and we're not compiling the debugger code,
634 * so store the line into the debugger's array of lines
636 if (PERLDB_LINE && PL_curstash != PL_debstash) {
637 SV *sv = NEWSV(85,0);
639 sv_upgrade(sv, SVt_PVMG);
640 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
643 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
650 * Check the unary operators to ensure there's no ambiguity in how they're
651 * used. An ambiguous piece of code would be:
653 * This doesn't mean rand() + 5. Because rand() is a unary operator,
654 * the +5 is its argument.
663 if (PL_oldoldbufptr != PL_last_uni)
665 while (isSPACE(*PL_last_uni))
667 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
668 if ((t = strchr(s, '(')) && t < PL_bufptr)
670 if (ckWARN_d(WARN_AMBIGUOUS)){
673 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
674 "Warning: Use of \"%s\" without parentheses is ambiguous",
681 * LOP : macro to build a list operator. Its behaviour has been replaced
682 * with a subroutine, S_lop() for which LOP is just another name.
685 #define LOP(f,x) return lop(f,x,s)
689 * Build a list operator (or something that might be one). The rules:
690 * - if we have a next token, then it's a list operator [why?]
691 * - if the next thing is an opening paren, then it's a function
692 * - else it's a list operator
696 S_lop(pTHX_ I32 f, int x, char *s)
703 PL_last_lop = PL_oldbufptr;
704 PL_last_lop_op = (OPCODE)f;
718 * When the lexer realizes it knows the next token (for instance,
719 * it is reordering tokens for the parser) then it can call S_force_next
720 * to know what token to return the next time the lexer is called. Caller
721 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
722 * handles the token correctly.
726 S_force_next(pTHX_ I32 type)
728 PL_nexttype[PL_nexttoke] = type;
730 if (PL_lex_state != LEX_KNOWNEXT) {
731 PL_lex_defer = PL_lex_state;
732 PL_lex_expect = PL_expect;
733 PL_lex_state = LEX_KNOWNEXT;
739 * When the lexer knows the next thing is a word (for instance, it has
740 * just seen -> and it knows that the next char is a word char, then
741 * it calls S_force_word to stick the next word into the PL_next lookahead.
744 * char *start : buffer position (must be within PL_linestr)
745 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
746 * int check_keyword : if true, Perl checks to make sure the word isn't
747 * a keyword (do this if the word is a label, e.g. goto FOO)
748 * int allow_pack : if true, : characters will also be allowed (require,
750 * int allow_initial_tick : used by the "sub" lexer only.
754 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
759 start = skipspace(start);
761 if (isIDFIRST_lazy_if(s,UTF) ||
762 (allow_pack && *s == ':') ||
763 (allow_initial_tick && *s == '\'') )
765 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
766 if (check_keyword && keyword(PL_tokenbuf, len))
768 if (token == METHOD) {
773 PL_expect = XOPERATOR;
776 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0));
777 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
778 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
779 SvUTF8_on(((SVOP*)PL_nextval[PL_nexttoke].opval)->op_sv);
787 * Called when the lexer wants $foo *foo &foo etc, but the program
788 * text only contains the "foo" portion. The first argument is a pointer
789 * to the "foo", and the second argument is the type symbol to prefix.
790 * Forces the next token to be a "WORD".
791 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
795 S_force_ident(pTHX_ register char *s, int kind)
798 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
799 PL_nextval[PL_nexttoke].opval = o;
802 o->op_private = OPpCONST_ENTERED;
803 /* XXX see note in pp_entereval() for why we forgo typo
804 warnings if the symbol must be introduced in an eval.
806 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
807 kind == '$' ? SVt_PV :
808 kind == '@' ? SVt_PVAV :
809 kind == '%' ? SVt_PVHV :
817 Perl_str_to_version(pTHX_ SV *sv)
822 char *start = SvPVx(sv,len);
823 bool utf = SvUTF8(sv) ? TRUE : FALSE;
824 char *end = start + len;
825 while (start < end) {
829 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
834 retval += ((NV)n)/nshift;
843 * Forces the next token to be a version number.
844 * If the next token appears to be an invalid version number, (e.g. "v2b"),
845 * and if "guessing" is TRUE, then no new token is created (and the caller
846 * must use an alternative parsing method).
850 S_force_version(pTHX_ char *s, int guessing)
852 OP *version = Nullop;
861 while (isDIGIT(*d) || *d == '_' || *d == '.')
863 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
865 s = scan_num(s, &yylval);
866 version = yylval.opval;
867 ver = cSVOPx(version)->op_sv;
868 if (SvPOK(ver) && !SvNIOK(ver)) {
869 (void)SvUPGRADE(ver, SVt_PVNV);
870 SvNVX(ver) = str_to_version(ver);
871 SvNOK_on(ver); /* hint that it is a version */
878 /* NOTE: The parser sees the package name and the VERSION swapped */
879 PL_nextval[PL_nexttoke].opval = version;
887 * Tokenize a quoted string passed in as an SV. It finds the next
888 * chunk, up to end of string or a backslash. It may make a new
889 * SV containing that chunk (if HINT_NEW_STRING is on). It also
894 S_tokeq(pTHX_ SV *sv)
905 s = SvPV_force(sv, len);
906 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
909 while (s < send && *s != '\\')
914 if ( PL_hints & HINT_NEW_STRING ) {
915 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
921 if (s + 1 < send && (s[1] == '\\'))
922 s++; /* all that, just for this */
927 SvCUR_set(sv, d - SvPVX(sv));
929 if ( PL_hints & HINT_NEW_STRING )
930 return new_constant(NULL, 0, "q", sv, pv, "q");
935 * Now come three functions related to double-quote context,
936 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
937 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
938 * interact with PL_lex_state, and create fake ( ... ) argument lists
939 * to handle functions and concatenation.
940 * They assume that whoever calls them will be setting up a fake
941 * join call, because each subthing puts a ',' after it. This lets
944 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
946 * (I'm not sure whether the spurious commas at the end of lcfirst's
947 * arguments and join's arguments are created or not).
952 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
954 * Pattern matching will set PL_lex_op to the pattern-matching op to
955 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
957 * OP_CONST and OP_READLINE are easy--just make the new op and return.
959 * Everything else becomes a FUNC.
961 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
962 * had an OP_CONST or OP_READLINE). This just sets us up for a
963 * call to S_sublex_push().
969 register I32 op_type = yylval.ival;
971 if (op_type == OP_NULL) {
972 yylval.opval = PL_lex_op;
976 if (op_type == OP_CONST || op_type == OP_READLINE) {
977 SV *sv = tokeq(PL_lex_stuff);
979 if (SvTYPE(sv) == SVt_PVIV) {
980 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
986 nsv = newSVpvn(p, len);
992 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
993 PL_lex_stuff = Nullsv;
994 /* Allow <FH> // "foo" */
995 if (op_type == OP_READLINE)
996 PL_expect = XTERMORDORDOR;
1000 PL_sublex_info.super_state = PL_lex_state;
1001 PL_sublex_info.sub_inwhat = op_type;
1002 PL_sublex_info.sub_op = PL_lex_op;
1003 PL_lex_state = LEX_INTERPPUSH;
1007 yylval.opval = PL_lex_op;
1017 * Create a new scope to save the lexing state. The scope will be
1018 * ended in S_sublex_done. Returns a '(', starting the function arguments
1019 * to the uc, lc, etc. found before.
1020 * Sets PL_lex_state to LEX_INTERPCONCAT.
1028 PL_lex_state = PL_sublex_info.super_state;
1029 SAVEI32(PL_lex_dojoin);
1030 SAVEI32(PL_lex_brackets);
1031 SAVEI32(PL_lex_casemods);
1032 SAVEI32(PL_lex_starts);
1033 SAVEI32(PL_lex_state);
1034 SAVEVPTR(PL_lex_inpat);
1035 SAVEI32(PL_lex_inwhat);
1036 SAVECOPLINE(PL_curcop);
1037 SAVEPPTR(PL_bufptr);
1038 SAVEPPTR(PL_bufend);
1039 SAVEPPTR(PL_oldbufptr);
1040 SAVEPPTR(PL_oldoldbufptr);
1041 SAVEPPTR(PL_last_lop);
1042 SAVEPPTR(PL_last_uni);
1043 SAVEPPTR(PL_linestart);
1044 SAVESPTR(PL_linestr);
1045 SAVEGENERICPV(PL_lex_brackstack);
1046 SAVEGENERICPV(PL_lex_casestack);
1048 PL_linestr = PL_lex_stuff;
1049 PL_lex_stuff = Nullsv;
1051 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1052 = SvPVX(PL_linestr);
1053 PL_bufend += SvCUR(PL_linestr);
1054 PL_last_lop = PL_last_uni = Nullch;
1055 SAVEFREESV(PL_linestr);
1057 PL_lex_dojoin = FALSE;
1058 PL_lex_brackets = 0;
1059 New(899, PL_lex_brackstack, 120, char);
1060 New(899, PL_lex_casestack, 12, char);
1061 PL_lex_casemods = 0;
1062 *PL_lex_casestack = '\0';
1064 PL_lex_state = LEX_INTERPCONCAT;
1065 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1067 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1068 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1069 PL_lex_inpat = PL_sublex_info.sub_op;
1071 PL_lex_inpat = Nullop;
1078 * Restores lexer state after a S_sublex_push.
1084 if (!PL_lex_starts++) {
1085 SV *sv = newSVpvn("",0);
1086 if (SvUTF8(PL_linestr))
1088 PL_expect = XOPERATOR;
1089 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1093 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1094 PL_lex_state = LEX_INTERPCASEMOD;
1098 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1099 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1100 PL_linestr = PL_lex_repl;
1102 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1103 PL_bufend += SvCUR(PL_linestr);
1104 PL_last_lop = PL_last_uni = Nullch;
1105 SAVEFREESV(PL_linestr);
1106 PL_lex_dojoin = FALSE;
1107 PL_lex_brackets = 0;
1108 PL_lex_casemods = 0;
1109 *PL_lex_casestack = '\0';
1111 if (SvEVALED(PL_lex_repl)) {
1112 PL_lex_state = LEX_INTERPNORMAL;
1114 /* we don't clear PL_lex_repl here, so that we can check later
1115 whether this is an evalled subst; that means we rely on the
1116 logic to ensure sublex_done() is called again only via the
1117 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1120 PL_lex_state = LEX_INTERPCONCAT;
1121 PL_lex_repl = Nullsv;
1127 PL_bufend = SvPVX(PL_linestr);
1128 PL_bufend += SvCUR(PL_linestr);
1129 PL_expect = XOPERATOR;
1130 PL_sublex_info.sub_inwhat = 0;
1138 Extracts a pattern, double-quoted string, or transliteration. This
1141 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1142 processing a pattern (PL_lex_inpat is true), a transliteration
1143 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1145 Returns a pointer to the character scanned up to. Iff this is
1146 advanced from the start pointer supplied (ie if anything was
1147 successfully parsed), will leave an OP for the substring scanned
1148 in yylval. Caller must intuit reason for not parsing further
1149 by looking at the next characters herself.
1153 double-quoted style: \r and \n
1154 regexp special ones: \D \s
1156 backrefs: \1 (deprecated in substitution replacements)
1157 case and quoting: \U \Q \E
1158 stops on @ and $, but not for $ as tail anchor
1160 In transliterations:
1161 characters are VERY literal, except for - not at the start or end
1162 of the string, which indicates a range. scan_const expands the
1163 range to the full set of intermediate characters.
1165 In double-quoted strings:
1167 double-quoted style: \r and \n
1169 backrefs: \1 (deprecated)
1170 case and quoting: \U \Q \E
1173 scan_const does *not* construct ops to handle interpolated strings.
1174 It stops processing as soon as it finds an embedded $ or @ variable
1175 and leaves it to the caller to work out what's going on.
1177 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1179 $ in pattern could be $foo or could be tail anchor. Assumption:
1180 it's a tail anchor if $ is the last thing in the string, or if it's
1181 followed by one of ")| \n\t"
1183 \1 (backreferences) are turned into $1
1185 The structure of the code is
1186 while (there's a character to process) {
1187 handle transliteration ranges
1188 skip regexp comments
1189 skip # initiated comments in //x patterns
1190 check for embedded @foo
1191 check for embedded scalars
1193 leave intact backslashes from leave (below)
1194 deprecate \1 in strings and sub replacements
1195 handle string-changing backslashes \l \U \Q \E, etc.
1196 switch (what was escaped) {
1197 handle - in a transliteration (becomes a literal -)
1198 handle \132 octal characters
1199 handle 0x15 hex characters
1200 handle \cV (control V)
1201 handle printf backslashes (\f, \r, \n, etc)
1203 } (end if backslash)
1204 } (end while character to read)
1209 S_scan_const(pTHX_ char *start)
1211 register char *send = PL_bufend; /* end of the constant */
1212 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1213 register char *s = start; /* start of the constant */
1214 register char *d = SvPVX(sv); /* destination for copies */
1215 bool dorange = FALSE; /* are we in a translit range? */
1216 bool didrange = FALSE; /* did we just finish a range? */
1217 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1218 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1221 const char *leaveit = /* set of acceptably-backslashed characters */
1223 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1226 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1227 /* If we are doing a trans and we know we want UTF8 set expectation */
1228 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1229 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1233 while (s < send || dorange) {
1234 /* get transliterations out of the way (they're most literal) */
1235 if (PL_lex_inwhat == OP_TRANS) {
1236 /* expand a range A-Z to the full set of characters. AIE! */
1238 I32 i; /* current expanded character */
1239 I32 min; /* first character in range */
1240 I32 max; /* last character in range */
1243 char *c = (char*)utf8_hop((U8*)d, -1);
1247 *c = (char)UTF_TO_NATIVE(0xff);
1248 /* mark the range as done, and continue */
1254 i = d - SvPVX(sv); /* remember current offset */
1255 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1256 d = SvPVX(sv) + i; /* refresh d after realloc */
1257 d -= 2; /* eat the first char and the - */
1259 min = (U8)*d; /* first char in range */
1260 max = (U8)d[1]; /* last char in range */
1264 "Invalid range \"%c-%c\" in transliteration operator",
1265 (char)min, (char)max);
1269 if ((isLOWER(min) && isLOWER(max)) ||
1270 (isUPPER(min) && isUPPER(max))) {
1272 for (i = min; i <= max; i++)
1274 *d++ = NATIVE_TO_NEED(has_utf8,i);
1276 for (i = min; i <= max; i++)
1278 *d++ = NATIVE_TO_NEED(has_utf8,i);
1283 for (i = min; i <= max; i++)
1286 /* mark the range as done, and continue */
1292 /* range begins (ignore - as first or last char) */
1293 else if (*s == '-' && s+1 < send && s != start) {
1295 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1298 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1310 /* if we get here, we're not doing a transliteration */
1312 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1313 except for the last char, which will be done separately. */
1314 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1316 while (s+1 < send && *s != ')')
1317 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1319 else if (s[2] == '{' /* This should match regcomp.c */
1320 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1323 char *regparse = s + (s[2] == '{' ? 3 : 4);
1326 while (count && (c = *regparse)) {
1327 if (c == '\\' && regparse[1])
1335 if (*regparse != ')')
1336 regparse--; /* Leave one char for continuation. */
1337 while (s < regparse)
1338 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1342 /* likewise skip #-initiated comments in //x patterns */
1343 else if (*s == '#' && PL_lex_inpat &&
1344 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1345 while (s+1 < send && *s != '\n')
1346 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1349 /* check for embedded arrays
1350 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1352 else if (*s == '@' && s[1]
1353 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1356 /* check for embedded scalars. only stop if we're sure it's a
1359 else if (*s == '$') {
1360 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1362 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1363 break; /* in regexp, $ might be tail anchor */
1366 /* End of else if chain - OP_TRANS rejoin rest */
1369 if (*s == '\\' && s+1 < send) {
1372 /* some backslashes we leave behind */
1373 if (*leaveit && *s && strchr(leaveit, *s)) {
1374 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1375 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1379 /* deprecate \1 in strings and substitution replacements */
1380 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1381 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1383 if (ckWARN(WARN_SYNTAX))
1384 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1389 /* string-change backslash escapes */
1390 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1395 /* if we get here, it's either a quoted -, or a digit */
1398 /* quoted - in transliterations */
1400 if (PL_lex_inwhat == OP_TRANS) {
1407 if (ckWARN(WARN_MISC) &&
1410 Perl_warner(aTHX_ packWARN(WARN_MISC),
1411 "Unrecognized escape \\%c passed through",
1413 /* default action is to copy the quoted character */
1414 goto default_action;
1417 /* \132 indicates an octal constant */
1418 case '0': case '1': case '2': case '3':
1419 case '4': case '5': case '6': case '7':
1423 uv = grok_oct(s, &len, &flags, NULL);
1426 goto NUM_ESCAPE_INSERT;
1428 /* \x24 indicates a hex constant */
1432 char* e = strchr(s, '}');
1433 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1434 PERL_SCAN_DISALLOW_PREFIX;
1439 yyerror("Missing right brace on \\x{}");
1443 uv = grok_hex(s, &len, &flags, NULL);
1449 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1450 uv = grok_hex(s, &len, &flags, NULL);
1456 /* Insert oct or hex escaped character.
1457 * There will always enough room in sv since such
1458 * escapes will be longer than any UTF-8 sequence
1459 * they can end up as. */
1461 /* We need to map to chars to ASCII before doing the tests
1464 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1465 if (!has_utf8 && uv > 255) {
1466 /* Might need to recode whatever we have
1467 * accumulated so far if it contains any
1470 * (Can't we keep track of that and avoid
1471 * this rescan? --jhi)
1475 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1476 if (!NATIVE_IS_INVARIANT(*c)) {
1481 STRLEN offset = d - SvPVX(sv);
1483 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1487 while (src >= (U8 *)SvPVX(sv)) {
1488 if (!NATIVE_IS_INVARIANT(*src)) {
1489 U8 ch = NATIVE_TO_ASCII(*src);
1490 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1491 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1501 if (has_utf8 || uv > 255) {
1502 d = (char*)uvchr_to_utf8((U8*)d, uv);
1504 if (PL_lex_inwhat == OP_TRANS &&
1505 PL_sublex_info.sub_op) {
1506 PL_sublex_info.sub_op->op_private |=
1507 (PL_lex_repl ? OPpTRANS_FROM_UTF
1520 /* \N{LATIN SMALL LETTER A} is a named character */
1524 char* e = strchr(s, '}');
1530 yyerror("Missing right brace on \\N{}");
1534 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1536 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1537 PERL_SCAN_DISALLOW_PREFIX;
1540 uv = grok_hex(s, &len, &flags, NULL);
1542 goto NUM_ESCAPE_INSERT;
1544 res = newSVpvn(s + 1, e - s - 1);
1545 res = new_constant( Nullch, 0, "charnames",
1546 res, Nullsv, "\\N{...}" );
1548 sv_utf8_upgrade(res);
1549 str = SvPV(res,len);
1550 #ifdef EBCDIC_NEVER_MIND
1551 /* charnames uses pack U and that has been
1552 * recently changed to do the below uni->native
1553 * mapping, so this would be redundant (and wrong,
1554 * the code point would be doubly converted).
1555 * But leave this in just in case the pack U change
1556 * gets revoked, but the semantics is still
1557 * desireable for charnames. --jhi */
1559 UV uv = utf8_to_uvchr((U8*)str, 0);
1562 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1564 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1565 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1566 str = SvPV(res, len);
1570 if (!has_utf8 && SvUTF8(res)) {
1571 char *ostart = SvPVX(sv);
1572 SvCUR_set(sv, d - ostart);
1575 sv_utf8_upgrade(sv);
1576 /* this just broke our allocation above... */
1577 SvGROW(sv, (STRLEN)(send - start));
1578 d = SvPVX(sv) + SvCUR(sv);
1581 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1582 char *odest = SvPVX(sv);
1584 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1585 d = SvPVX(sv) + (d - odest);
1587 Copy(str, d, len, char);
1594 yyerror("Missing braces on \\N{}");
1597 /* \c is a control character */
1606 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1609 yyerror("Missing control char name in \\c");
1613 /* printf-style backslashes, formfeeds, newlines, etc */
1615 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1618 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1621 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1624 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1627 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1630 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1633 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1639 } /* end if (backslash) */
1642 /* If we started with encoded form, or already know we want it
1643 and then encode the next character */
1644 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1646 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1647 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1650 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1651 STRLEN off = d - SvPVX(sv);
1652 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1654 d = (char*)uvchr_to_utf8((U8*)d, uv);
1658 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1660 } /* while loop to process each character */
1662 /* terminate the string and set up the sv */
1664 SvCUR_set(sv, d - SvPVX(sv));
1665 if (SvCUR(sv) >= SvLEN(sv))
1666 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1669 if (PL_encoding && !has_utf8) {
1670 sv_recode_to_utf8(sv, PL_encoding);
1676 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1677 PL_sublex_info.sub_op->op_private |=
1678 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1682 /* shrink the sv if we allocated more than we used */
1683 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1684 SvLEN_set(sv, SvCUR(sv) + 1);
1685 Renew(SvPVX(sv), SvLEN(sv), char);
1688 /* return the substring (via yylval) only if we parsed anything */
1689 if (s > PL_bufptr) {
1690 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1691 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1693 ( PL_lex_inwhat == OP_TRANS
1695 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1698 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1705 * Returns TRUE if there's more to the expression (e.g., a subscript),
1708 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1710 * ->[ and ->{ return TRUE
1711 * { and [ outside a pattern are always subscripts, so return TRUE
1712 * if we're outside a pattern and it's not { or [, then return FALSE
1713 * if we're in a pattern and the first char is a {
1714 * {4,5} (any digits around the comma) returns FALSE
1715 * if we're in a pattern and the first char is a [
1717 * [SOMETHING] has a funky algorithm to decide whether it's a
1718 * character class or not. It has to deal with things like
1719 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1720 * anything else returns TRUE
1723 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1726 S_intuit_more(pTHX_ register char *s)
1728 if (PL_lex_brackets)
1730 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1732 if (*s != '{' && *s != '[')
1737 /* In a pattern, so maybe we have {n,m}. */
1754 /* On the other hand, maybe we have a character class */
1757 if (*s == ']' || *s == '^')
1760 /* this is terrifying, and it works */
1761 int weight = 2; /* let's weigh the evidence */
1763 unsigned char un_char = 255, last_un_char;
1764 char *send = strchr(s,']');
1765 char tmpbuf[sizeof PL_tokenbuf * 4];
1767 if (!send) /* has to be an expression */
1770 Zero(seen,256,char);
1773 else if (isDIGIT(*s)) {
1775 if (isDIGIT(s[1]) && s[2] == ']')
1781 for (; s < send; s++) {
1782 last_un_char = un_char;
1783 un_char = (unsigned char)*s;
1788 weight -= seen[un_char] * 10;
1789 if (isALNUM_lazy_if(s+1,UTF)) {
1790 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1791 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1796 else if (*s == '$' && s[1] &&
1797 strchr("[#!%*<>()-=",s[1])) {
1798 if (/*{*/ strchr("])} =",s[2]))
1807 if (strchr("wds]",s[1]))
1809 else if (seen['\''] || seen['"'])
1811 else if (strchr("rnftbxcav",s[1]))
1813 else if (isDIGIT(s[1])) {
1815 while (s[1] && isDIGIT(s[1]))
1825 if (strchr("aA01! ",last_un_char))
1827 if (strchr("zZ79~",s[1]))
1829 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1830 weight -= 5; /* cope with negative subscript */
1833 if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) &&
1834 isALPHA(*s) && s[1] && isALPHA(s[1])) {
1839 if (keyword(tmpbuf, d - tmpbuf))
1842 if (un_char == last_un_char + 1)
1844 weight -= seen[un_char];
1849 if (weight >= 0) /* probably a character class */
1859 * Does all the checking to disambiguate
1861 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
1862 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
1864 * First argument is the stuff after the first token, e.g. "bar".
1866 * Not a method if bar is a filehandle.
1867 * Not a method if foo is a subroutine prototyped to take a filehandle.
1868 * Not a method if it's really "Foo $bar"
1869 * Method if it's "foo $bar"
1870 * Not a method if it's really "print foo $bar"
1871 * Method if it's really "foo package::" (interpreted as package->foo)
1872 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
1873 * Not a method if bar is a filehandle or package, but is quoted with
1878 S_intuit_method(pTHX_ char *start, GV *gv)
1880 char *s = start + (*start == '$');
1881 char tmpbuf[sizeof PL_tokenbuf];
1889 if ((cv = GvCVu(gv))) {
1890 char *proto = SvPVX(cv);
1900 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
1901 /* start is the beginning of the possible filehandle/object,
1902 * and s is the end of it
1903 * tmpbuf is a copy of it
1906 if (*start == '$') {
1907 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
1912 return *s == '(' ? FUNCMETH : METHOD;
1914 if (!keyword(tmpbuf, len)) {
1915 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
1920 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
1921 if (indirgv && GvCVu(indirgv))
1923 /* filehandle or package name makes it a method */
1924 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
1926 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
1927 return 0; /* no assumptions -- "=>" quotes bearword */
1929 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
1930 newSVpvn(tmpbuf,len));
1931 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
1935 return *s == '(' ? FUNCMETH : METHOD;
1943 * Return a string of Perl code to load the debugger. If PERL5DB
1944 * is set, it will return the contents of that, otherwise a
1945 * compile-time require of perl5db.pl.
1952 char *pdb = PerlEnv_getenv("PERL5DB");
1956 SETERRNO(0,SS_NORMAL);
1957 return "BEGIN { require 'perl5db.pl' }";
1963 /* Encoded script support. filter_add() effectively inserts a
1964 * 'pre-processing' function into the current source input stream.
1965 * Note that the filter function only applies to the current source file
1966 * (e.g., it will not affect files 'require'd or 'use'd by this one).
1968 * The datasv parameter (which may be NULL) can be used to pass
1969 * private data to this instance of the filter. The filter function
1970 * can recover the SV using the FILTER_DATA macro and use it to
1971 * store private buffers and state information.
1973 * The supplied datasv parameter is upgraded to a PVIO type
1974 * and the IoDIRP/IoANY field is used to store the function pointer,
1975 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
1976 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
1977 * private use must be set using malloc'd pointers.
1981 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
1986 if (!PL_rsfp_filters)
1987 PL_rsfp_filters = newAV();
1989 datasv = NEWSV(255,0);
1990 if (!SvUPGRADE(datasv, SVt_PVIO))
1991 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
1992 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
1993 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
1994 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
1995 (void*)funcp, SvPV_nolen(datasv)));
1996 av_unshift(PL_rsfp_filters, 1);
1997 av_store(PL_rsfp_filters, 0, datasv) ;
2002 /* Delete most recently added instance of this filter function. */
2004 Perl_filter_del(pTHX_ filter_t funcp)
2007 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2008 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2010 /* if filter is on top of stack (usual case) just pop it off */
2011 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2012 if (IoANY(datasv) == (void *)funcp) {
2013 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2014 IoANY(datasv) = (void *)NULL;
2015 sv_free(av_pop(PL_rsfp_filters));
2019 /* we need to search for the correct entry and clear it */
2020 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2024 /* Invoke the n'th filter function for the current rsfp. */
2026 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2029 /* 0 = read one text line */
2034 if (!PL_rsfp_filters)
2036 if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */
2037 /* Provide a default input filter to make life easy. */
2038 /* Note that we append to the line. This is handy. */
2039 DEBUG_P(PerlIO_printf(Perl_debug_log,
2040 "filter_read %d: from rsfp\n", idx));
2044 int old_len = SvCUR(buf_sv) ;
2046 /* ensure buf_sv is large enough */
2047 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2048 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2049 if (PerlIO_error(PL_rsfp))
2050 return -1; /* error */
2052 return 0 ; /* end of file */
2054 SvCUR_set(buf_sv, old_len + len) ;
2057 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2058 if (PerlIO_error(PL_rsfp))
2059 return -1; /* error */
2061 return 0 ; /* end of file */
2064 return SvCUR(buf_sv);
2066 /* Skip this filter slot if filter has been deleted */
2067 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){
2068 DEBUG_P(PerlIO_printf(Perl_debug_log,
2069 "filter_read %d: skipped (filter deleted)\n",
2071 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2073 /* Get function pointer hidden within datasv */
2074 funcp = (filter_t)IoANY(datasv);
2075 DEBUG_P(PerlIO_printf(Perl_debug_log,
2076 "filter_read %d: via function %p (%s)\n",
2077 idx, (void*)funcp, SvPV_nolen(datasv)));
2078 /* Call function. The function is expected to */
2079 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2080 /* Return: <0:error, =0:eof, >0:not eof */
2081 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2085 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2087 #ifdef PERL_CR_FILTER
2088 if (!PL_rsfp_filters) {
2089 filter_add(S_cr_textfilter,NULL);
2092 if (PL_rsfp_filters) {
2095 SvCUR_set(sv, 0); /* start with empty line */
2096 if (FILTER_READ(0, sv, 0) > 0)
2097 return ( SvPVX(sv) ) ;
2102 return (sv_gets(sv, fp, append));
2106 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2110 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2114 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2115 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2117 return GvHV(gv); /* Foo:: */
2120 /* use constant CLASS => 'MyClass' */
2121 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2123 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2124 pkgname = SvPV_nolen(sv);
2128 return gv_stashpv(pkgname, FALSE);
2132 static char* exp_name[] =
2133 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2134 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2141 Works out what to call the token just pulled out of the input
2142 stream. The yacc parser takes care of taking the ops we return and
2143 stitching them into a tree.
2149 if read an identifier
2150 if we're in a my declaration
2151 croak if they tried to say my($foo::bar)
2152 build the ops for a my() declaration
2153 if it's an access to a my() variable
2154 are we in a sort block?
2155 croak if my($a); $a <=> $b
2156 build ops for access to a my() variable
2157 if in a dq string, and they've said @foo and we can't find @foo
2159 build ops for a bareword
2160 if we already built the token before, use it.
2165 #pragma segment Perl_yylex
2177 I32 orig_keyword = 0;
2179 /* check if there's an identifier for us to look at */
2180 if (PL_pending_ident)
2181 return S_pending_ident(aTHX);
2183 /* no identifier pending identification */
2185 switch (PL_lex_state) {
2187 case LEX_NORMAL: /* Some compilers will produce faster */
2188 case LEX_INTERPNORMAL: /* code if we comment these out. */
2192 /* when we've already built the next token, just pull it out of the queue */
2195 yylval = PL_nextval[PL_nexttoke];
2197 PL_lex_state = PL_lex_defer;
2198 PL_expect = PL_lex_expect;
2199 PL_lex_defer = LEX_NORMAL;
2201 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2202 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2203 (IV)PL_nexttype[PL_nexttoke]); });
2205 return(PL_nexttype[PL_nexttoke]);
2207 /* interpolated case modifiers like \L \U, including \Q and \E.
2208 when we get here, PL_bufptr is at the \
2210 case LEX_INTERPCASEMOD:
2212 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2213 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2215 /* handle \E or end of string */
2216 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2220 if (PL_lex_casemods) {
2221 oldmod = PL_lex_casestack[--PL_lex_casemods];
2222 PL_lex_casestack[PL_lex_casemods] = '\0';
2224 if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) {
2226 PL_lex_state = LEX_INTERPCONCAT;
2230 if (PL_bufptr != PL_bufend)
2232 PL_lex_state = LEX_INTERPCONCAT;
2236 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2237 "### Saw case modifier at '%s'\n", PL_bufptr); });
2239 if (s[1] == '\\' && s[2] == 'E') {
2241 PL_lex_state = LEX_INTERPCONCAT;
2245 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2246 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2247 if (strchr("LU", *s) &&
2248 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2249 PL_lex_casestack[--PL_lex_casemods] = '\0';
2252 if (PL_lex_casemods > 10)
2253 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2254 PL_lex_casestack[PL_lex_casemods++] = *s;
2255 PL_lex_casestack[PL_lex_casemods] = '\0';
2256 PL_lex_state = LEX_INTERPCONCAT;
2257 PL_nextval[PL_nexttoke].ival = 0;
2260 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2262 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2264 PL_nextval[PL_nexttoke].ival = OP_LC;
2266 PL_nextval[PL_nexttoke].ival = OP_UC;
2268 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2270 Perl_croak(aTHX_ "panic: yylex");
2274 if (PL_lex_starts) {
2283 case LEX_INTERPPUSH:
2284 return sublex_push();
2286 case LEX_INTERPSTART:
2287 if (PL_bufptr == PL_bufend)
2288 return sublex_done();
2289 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2290 "### Interpolated variable at '%s'\n", PL_bufptr); });
2292 PL_lex_dojoin = (*PL_bufptr == '@');
2293 PL_lex_state = LEX_INTERPNORMAL;
2294 if (PL_lex_dojoin) {
2295 PL_nextval[PL_nexttoke].ival = 0;
2297 force_ident("\"", '$');
2298 PL_nextval[PL_nexttoke].ival = 0;
2300 PL_nextval[PL_nexttoke].ival = 0;
2302 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2305 if (PL_lex_starts++) {
2311 case LEX_INTERPENDMAYBE:
2312 if (intuit_more(PL_bufptr)) {
2313 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2319 if (PL_lex_dojoin) {
2320 PL_lex_dojoin = FALSE;
2321 PL_lex_state = LEX_INTERPCONCAT;
2324 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2325 && SvEVALED(PL_lex_repl))
2327 if (PL_bufptr != PL_bufend)
2328 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2329 PL_lex_repl = Nullsv;
2332 case LEX_INTERPCONCAT:
2334 if (PL_lex_brackets)
2335 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2337 if (PL_bufptr == PL_bufend)
2338 return sublex_done();
2340 if (SvIVX(PL_linestr) == '\'') {
2341 SV *sv = newSVsv(PL_linestr);
2344 else if ( PL_hints & HINT_NEW_RE )
2345 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2346 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2350 s = scan_const(PL_bufptr);
2352 PL_lex_state = LEX_INTERPCASEMOD;
2354 PL_lex_state = LEX_INTERPSTART;
2357 if (s != PL_bufptr) {
2358 PL_nextval[PL_nexttoke] = yylval;
2361 if (PL_lex_starts++)
2371 PL_lex_state = LEX_NORMAL;
2372 s = scan_formline(PL_bufptr);
2373 if (!PL_lex_formbrack)
2379 PL_oldoldbufptr = PL_oldbufptr;
2382 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n",
2383 exp_name[PL_expect], s);
2389 if (isIDFIRST_lazy_if(s,UTF))
2391 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2394 goto fake_eof; /* emulate EOF on ^D or ^Z */
2399 if (PL_lex_brackets) {
2400 if (PL_lex_formbrack)
2401 yyerror("Format not terminated");
2403 yyerror("Missing right curly or square bracket");
2405 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2406 "### Tokener got EOF\n");
2410 if (s++ < PL_bufend)
2411 goto retry; /* ignore stray nulls */
2414 if (!PL_in_eval && !PL_preambled) {
2415 PL_preambled = TRUE;
2416 sv_setpv(PL_linestr,incl_perldb());
2417 if (SvCUR(PL_linestr))
2418 sv_catpv(PL_linestr,";");
2420 while(AvFILLp(PL_preambleav) >= 0) {
2421 SV *tmpsv = av_shift(PL_preambleav);
2422 sv_catsv(PL_linestr, tmpsv);
2423 sv_catpv(PL_linestr, ";");
2426 sv_free((SV*)PL_preambleav);
2427 PL_preambleav = NULL;
2429 if (PL_minus_n || PL_minus_p) {
2430 sv_catpv(PL_linestr, "LINE: while (<>) {");
2432 sv_catpv(PL_linestr,"chomp;");
2435 if (strchr("/'\"", *PL_splitstr)
2436 && strchr(PL_splitstr + 1, *PL_splitstr))
2437 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2440 s = "'~#\200\1'"; /* surely one char is unused...*/
2441 while (s[1] && strchr(PL_splitstr, *s)) s++;
2443 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c",
2444 "q" + (delim == '\''), delim);
2445 for (s = PL_splitstr; *s; s++) {
2447 sv_catpvn(PL_linestr, "\\", 1);
2448 sv_catpvn(PL_linestr, s, 1);
2450 Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim);
2454 sv_catpv(PL_linestr,"our @F=split(' ');");
2457 sv_catpv(PL_linestr, "\n");
2458 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2459 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2460 PL_last_lop = PL_last_uni = Nullch;
2461 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2462 SV *sv = NEWSV(85,0);
2464 sv_upgrade(sv, SVt_PVMG);
2465 sv_setsv(sv,PL_linestr);
2468 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2473 bof = PL_rsfp ? TRUE : FALSE;
2474 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2477 if (PL_preprocess && !PL_in_eval)
2478 (void)PerlProc_pclose(PL_rsfp);
2479 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2480 PerlIO_clearerr(PL_rsfp);
2482 (void)PerlIO_close(PL_rsfp);
2484 PL_doextract = FALSE;
2486 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2487 sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : "");
2488 sv_catpv(PL_linestr,";}");
2489 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2490 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2491 PL_last_lop = PL_last_uni = Nullch;
2492 PL_minus_n = PL_minus_p = 0;
2495 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2496 PL_last_lop = PL_last_uni = Nullch;
2497 sv_setpv(PL_linestr,"");
2498 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2500 /* If it looks like the start of a BOM or raw UTF-16,
2501 * check if it in fact is. */
2507 #ifdef PERLIO_IS_STDIO
2508 # ifdef __GNU_LIBRARY__
2509 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2510 # define FTELL_FOR_PIPE_IS_BROKEN
2514 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2515 # define FTELL_FOR_PIPE_IS_BROKEN
2520 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2521 /* This loses the possibility to detect the bof
2522 * situation on perl -P when the libc5 is being used.
2523 * Workaround? Maybe attach some extra state to PL_rsfp?
2526 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2528 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2531 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2532 s = swallow_bom((U8*)s);
2536 /* Incest with pod. */
2537 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2538 sv_setpv(PL_linestr, "");
2539 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2540 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2541 PL_last_lop = PL_last_uni = Nullch;
2542 PL_doextract = FALSE;
2546 } while (PL_doextract);
2547 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2548 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2549 SV *sv = NEWSV(85,0);
2551 sv_upgrade(sv, SVt_PVMG);
2552 sv_setsv(sv,PL_linestr);
2555 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2557 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2558 PL_last_lop = PL_last_uni = Nullch;
2559 if (CopLINE(PL_curcop) == 1) {
2560 while (s < PL_bufend && isSPACE(*s))
2562 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2566 if (*s == '#' && *(s+1) == '!')
2568 #ifdef ALTERNATE_SHEBANG
2570 static char as[] = ALTERNATE_SHEBANG;
2571 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2572 d = s + (sizeof(as) - 1);
2574 #endif /* ALTERNATE_SHEBANG */
2583 while (*d && !isSPACE(*d))
2587 #ifdef ARG_ZERO_IS_SCRIPT
2588 if (ipathend > ipath) {
2590 * HP-UX (at least) sets argv[0] to the script name,
2591 * which makes $^X incorrect. And Digital UNIX and Linux,
2592 * at least, set argv[0] to the basename of the Perl
2593 * interpreter. So, having found "#!", we'll set it right.
2595 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2596 assert(SvPOK(x) || SvGMAGICAL(x));
2597 if (sv_eq(x, CopFILESV(PL_curcop))) {
2598 sv_setpvn(x, ipath, ipathend - ipath);
2604 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2605 char *lstart = SvPV(x,llen);
2607 bstart += blen - llen;
2608 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2609 sv_setpvn(x, ipath, ipathend - ipath);
2614 TAINT_NOT; /* $^X is always tainted, but that's OK */
2616 #endif /* ARG_ZERO_IS_SCRIPT */
2621 d = instr(s,"perl -");
2623 d = instr(s,"perl");
2625 /* avoid getting into infinite loops when shebang
2626 * line contains "Perl" rather than "perl" */
2628 for (d = ipathend-4; d >= ipath; --d) {
2629 if ((*d == 'p' || *d == 'P')
2630 && !ibcmp(d, "perl", 4))
2640 #ifdef ALTERNATE_SHEBANG
2642 * If the ALTERNATE_SHEBANG on this system starts with a
2643 * character that can be part of a Perl expression, then if
2644 * we see it but not "perl", we're probably looking at the
2645 * start of Perl code, not a request to hand off to some
2646 * other interpreter. Similarly, if "perl" is there, but
2647 * not in the first 'word' of the line, we assume the line
2648 * contains the start of the Perl program.
2650 if (d && *s != '#') {
2652 while (*c && !strchr("; \t\r\n\f\v#", *c))
2655 d = Nullch; /* "perl" not in first word; ignore */
2657 *s = '#'; /* Don't try to parse shebang line */
2659 #endif /* ALTERNATE_SHEBANG */
2660 #ifndef MACOS_TRADITIONAL
2665 !instr(s,"indir") &&
2666 instr(PL_origargv[0],"perl"))
2672 while (s < PL_bufend && isSPACE(*s))
2674 if (s < PL_bufend) {
2675 Newz(899,newargv,PL_origargc+3,char*);
2677 while (s < PL_bufend && !isSPACE(*s))
2680 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2683 newargv = PL_origargv;
2686 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2688 Perl_croak(aTHX_ "Can't exec %s", ipath);
2692 U32 oldpdb = PL_perldb;
2693 bool oldn = PL_minus_n;
2694 bool oldp = PL_minus_p;
2696 while (*d && !isSPACE(*d)) d++;
2697 while (SPACE_OR_TAB(*d)) d++;
2700 bool switches_done = PL_doswitches;
2702 if (*d == 'M' || *d == 'm') {
2704 while (*d && !isSPACE(*d)) d++;
2705 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2708 d = moreswitches(d);
2710 if (PL_doswitches && !switches_done) {
2711 int argc = PL_origargc;
2712 char **argv = PL_origargv;
2715 } while (argc && argv[0][0] == '-' && argv[0][1]);
2716 init_argv_symbols(argc,argv);
2718 if ((PERLDB_LINE && !oldpdb) ||
2719 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2720 /* if we have already added "LINE: while (<>) {",
2721 we must not do it again */
2723 sv_setpv(PL_linestr, "");
2724 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2725 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2726 PL_last_lop = PL_last_uni = Nullch;
2727 PL_preambled = FALSE;
2729 (void)gv_fetchfile(PL_origfilename);
2732 if (PL_doswitches && !switches_done) {
2733 int argc = PL_origargc;
2734 char **argv = PL_origargv;
2737 } while (argc && argv[0][0] == '-' && argv[0][1]);
2738 init_argv_symbols(argc,argv);
2744 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2746 PL_lex_state = LEX_FORMLINE;
2751 #ifdef PERL_STRICT_CR
2752 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2754 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2756 case ' ': case '\t': case '\f': case 013:
2757 #ifdef MACOS_TRADITIONAL
2764 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2765 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2766 /* handle eval qq[#line 1 "foo"\n ...] */
2767 CopLINE_dec(PL_curcop);
2771 while (s < d && *s != '\n')
2775 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2776 Perl_croak(aTHX_ "panic: input overflow");
2778 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2780 PL_lex_state = LEX_FORMLINE;
2790 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2797 while (s < PL_bufend && SPACE_OR_TAB(*s))
2800 if (strnEQ(s,"=>",2)) {
2801 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2802 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2803 "### Saw unary minus before =>, forcing word '%s'\n", s);
2805 OPERATOR('-'); /* unary minus */
2807 PL_last_uni = PL_oldbufptr;
2809 case 'r': ftst = OP_FTEREAD; break;
2810 case 'w': ftst = OP_FTEWRITE; break;
2811 case 'x': ftst = OP_FTEEXEC; break;
2812 case 'o': ftst = OP_FTEOWNED; break;
2813 case 'R': ftst = OP_FTRREAD; break;
2814 case 'W': ftst = OP_FTRWRITE; break;
2815 case 'X': ftst = OP_FTREXEC; break;
2816 case 'O': ftst = OP_FTROWNED; break;
2817 case 'e': ftst = OP_FTIS; break;
2818 case 'z': ftst = OP_FTZERO; break;
2819 case 's': ftst = OP_FTSIZE; break;
2820 case 'f': ftst = OP_FTFILE; break;
2821 case 'd': ftst = OP_FTDIR; break;
2822 case 'l': ftst = OP_FTLINK; break;
2823 case 'p': ftst = OP_FTPIPE; break;
2824 case 'S': ftst = OP_FTSOCK; break;
2825 case 'u': ftst = OP_FTSUID; break;
2826 case 'g': ftst = OP_FTSGID; break;
2827 case 'k': ftst = OP_FTSVTX; break;
2828 case 'b': ftst = OP_FTBLK; break;
2829 case 'c': ftst = OP_FTCHR; break;
2830 case 't': ftst = OP_FTTTY; break;
2831 case 'T': ftst = OP_FTTEXT; break;
2832 case 'B': ftst = OP_FTBINARY; break;
2833 case 'M': case 'A': case 'C':
2834 gv_fetchpv("\024",TRUE, SVt_PV);
2836 case 'M': ftst = OP_FTMTIME; break;
2837 case 'A': ftst = OP_FTATIME; break;
2838 case 'C': ftst = OP_FTCTIME; break;
2846 PL_last_lop_op = (OPCODE)ftst;
2847 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2848 "### Saw file test %c\n", (int)ftst);
2853 /* Assume it was a minus followed by a one-letter named
2854 * subroutine call (or a -bareword), then. */
2855 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2856 "### '-%c' looked like a file test but was not\n",
2865 if (PL_expect == XOPERATOR)
2870 else if (*s == '>') {
2873 if (isIDFIRST_lazy_if(s,UTF)) {
2874 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
2882 if (PL_expect == XOPERATOR)
2885 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2887 OPERATOR('-'); /* unary minus */
2894 if (PL_expect == XOPERATOR)
2899 if (PL_expect == XOPERATOR)
2902 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
2908 if (PL_expect != XOPERATOR) {
2909 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
2910 PL_expect = XOPERATOR;
2911 force_ident(PL_tokenbuf, '*');
2924 if (PL_expect == XOPERATOR) {
2928 PL_tokenbuf[0] = '%';
2929 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
2930 if (!PL_tokenbuf[1]) {
2933 PL_pending_ident = '%';
2952 switch (PL_expect) {
2955 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
2957 PL_bufptr = s; /* update in case we back off */
2963 PL_expect = XTERMBLOCK;
2967 while (isIDFIRST_lazy_if(s,UTF)) {
2968 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
2969 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
2970 if (tmp < 0) tmp = -tmp;
2986 d = scan_str(d,TRUE,TRUE);
2988 /* MUST advance bufptr here to avoid bogus
2989 "at end of line" context messages from yyerror().
2991 PL_bufptr = s + len;
2992 yyerror("Unterminated attribute parameter in attribute list");
2995 return 0; /* EOF indicator */
2999 SV *sv = newSVpvn(s, len);
3000 sv_catsv(sv, PL_lex_stuff);
3001 attrs = append_elem(OP_LIST, attrs,
3002 newSVOP(OP_CONST, 0, sv));
3003 SvREFCNT_dec(PL_lex_stuff);
3004 PL_lex_stuff = Nullsv;
3007 if (len == 6 && strnEQ(s, "unique", len)) {
3008 if (PL_in_my == KEY_our)
3010 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3012 ; /* skip to avoid loading attributes.pm */
3015 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3018 /* NOTE: any CV attrs applied here need to be part of
3019 the CVf_BUILTIN_ATTRS define in cv.h! */
3020 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3021 CvLVALUE_on(PL_compcv);
3022 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3023 CvLOCKED_on(PL_compcv);
3024 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3025 CvMETHOD_on(PL_compcv);
3026 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3027 CvASSERTION_on(PL_compcv);
3028 /* After we've set the flags, it could be argued that
3029 we don't need to do the attributes.pm-based setting
3030 process, and shouldn't bother appending recognized
3031 flags. To experiment with that, uncomment the
3032 following "else". (Note that's already been
3033 uncommented. That keeps the above-applied built-in
3034 attributes from being intercepted (and possibly
3035 rejected) by a package's attribute routines, but is
3036 justified by the performance win for the common case
3037 of applying only built-in attributes.) */
3039 attrs = append_elem(OP_LIST, attrs,
3040 newSVOP(OP_CONST, 0,
3044 if (*s == ':' && s[1] != ':')
3047 break; /* require real whitespace or :'s */
3049 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3050 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3051 char q = ((*s == '\'') ? '"' : '\'');
3052 /* If here for an expression, and parsed no attrs, back off. */
3053 if (tmp == '=' && !attrs) {
3057 /* MUST advance bufptr here to avoid bogus "at end of line"
3058 context messages from yyerror().
3062 yyerror("Unterminated attribute list");
3064 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3072 PL_nextval[PL_nexttoke].opval = attrs;
3080 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3081 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3098 if (PL_lex_brackets <= 0)
3099 yyerror("Unmatched right square bracket");
3102 if (PL_lex_state == LEX_INTERPNORMAL) {
3103 if (PL_lex_brackets == 0) {
3104 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3105 PL_lex_state = LEX_INTERPEND;
3112 if (PL_lex_brackets > 100) {
3113 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3115 switch (PL_expect) {
3117 if (PL_lex_formbrack) {
3121 if (PL_oldoldbufptr == PL_last_lop)
3122 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3124 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3125 OPERATOR(HASHBRACK);
3127 while (s < PL_bufend && SPACE_OR_TAB(*s))
3130 PL_tokenbuf[0] = '\0';
3131 if (d < PL_bufend && *d == '-') {
3132 PL_tokenbuf[0] = '-';
3134 while (d < PL_bufend && SPACE_OR_TAB(*d))
3137 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3138 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3140 while (d < PL_bufend && SPACE_OR_TAB(*d))
3143 char minus = (PL_tokenbuf[0] == '-');
3144 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3152 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3157 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3162 if (PL_oldoldbufptr == PL_last_lop)
3163 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3165 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3168 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3170 /* This hack is to get the ${} in the message. */
3172 yyerror("syntax error");
3175 OPERATOR(HASHBRACK);
3177 /* This hack serves to disambiguate a pair of curlies
3178 * as being a block or an anon hash. Normally, expectation
3179 * determines that, but in cases where we're not in a
3180 * position to expect anything in particular (like inside
3181 * eval"") we have to resolve the ambiguity. This code
3182 * covers the case where the first term in the curlies is a
3183 * quoted string. Most other cases need to be explicitly
3184 * disambiguated by prepending a `+' before the opening
3185 * curly in order to force resolution as an anon hash.
3187 * XXX should probably propagate the outer expectation
3188 * into eval"" to rely less on this hack, but that could
3189 * potentially break current behavior of eval"".
3193 if (*s == '\'' || *s == '"' || *s == '`') {
3194 /* common case: get past first string, handling escapes */
3195 for (t++; t < PL_bufend && *t != *s;)
3196 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3200 else if (*s == 'q') {
3203 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3206 /* skip q//-like construct */
3208 char open, close, term;
3211 while (t < PL_bufend && isSPACE(*t))
3213 /* check for q => */
3214 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3215 OPERATOR(HASHBRACK);
3219 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3223 for (t++; t < PL_bufend; t++) {
3224 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3226 else if (*t == open)
3230 for (t++; t < PL_bufend; t++) {
3231 if (*t == '\\' && t+1 < PL_bufend)
3233 else if (*t == close && --brackets <= 0)
3235 else if (*t == open)
3242 /* skip plain q word */
3243 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3246 else if (isALNUM_lazy_if(t,UTF)) {
3248 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3251 while (t < PL_bufend && isSPACE(*t))
3253 /* if comma follows first term, call it an anon hash */
3254 /* XXX it could be a comma expression with loop modifiers */
3255 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3256 || (*t == '=' && t[1] == '>')))
3257 OPERATOR(HASHBRACK);
3258 if (PL_expect == XREF)
3261 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3267 yylval.ival = CopLINE(PL_curcop);
3268 if (isSPACE(*s) || *s == '#')
3269 PL_copline = NOLINE; /* invalidate current command line number */
3274 if (PL_lex_brackets <= 0)
3275 yyerror("Unmatched right curly bracket");
3277 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3278 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3279 PL_lex_formbrack = 0;
3280 if (PL_lex_state == LEX_INTERPNORMAL) {
3281 if (PL_lex_brackets == 0) {
3282 if (PL_expect & XFAKEBRACK) {
3283 PL_expect &= XENUMMASK;
3284 PL_lex_state = LEX_INTERPEND;
3286 return yylex(); /* ignore fake brackets */
3288 if (*s == '-' && s[1] == '>')
3289 PL_lex_state = LEX_INTERPENDMAYBE;
3290 else if (*s != '[' && *s != '{')
3291 PL_lex_state = LEX_INTERPEND;
3294 if (PL_expect & XFAKEBRACK) {
3295 PL_expect &= XENUMMASK;
3297 return yylex(); /* ignore fake brackets */
3307 if (PL_expect == XOPERATOR) {
3308 if (ckWARN(WARN_SEMICOLON)
3309 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3311 CopLINE_dec(PL_curcop);
3312 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3313 CopLINE_inc(PL_curcop);
3318 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3320 PL_expect = XOPERATOR;
3321 force_ident(PL_tokenbuf, '&');
3325 yylval.ival = (OPpENTERSUB_AMPER<<8);
3344 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3345 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3347 if (PL_expect == XSTATE && isALPHA(tmp) &&
3348 (s == PL_linestart+1 || s[-2] == '\n') )
3350 if (PL_in_eval && !PL_rsfp) {
3355 if (strnEQ(s,"=cut",4)) {
3369 PL_doextract = TRUE;
3372 if (PL_lex_brackets < PL_lex_formbrack) {
3374 #ifdef PERL_STRICT_CR
3375 for (t = s; SPACE_OR_TAB(*t); t++) ;
3377 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3379 if (*t == '\n' || *t == '#') {
3391 /* was this !=~ where !~ was meant?
3392 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3394 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3397 while (t < PL_bufend && isSPACE(*t))
3400 if (*t == '/' || *t == '?' ||
3401 ((*t == 'm' || *t == 's' || *t == 'y') && !isALNUM(t[1])) ||
3402 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3403 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3404 "!=~ should be !~");
3413 if (PL_expect != XOPERATOR) {
3414 if (s[1] != '<' && !strchr(s,'>'))
3417 s = scan_heredoc(s);
3419 s = scan_inputsymbol(s);
3420 TERM(sublex_start());
3425 SHop(OP_LEFT_SHIFT);
3439 SHop(OP_RIGHT_SHIFT);
3448 if (PL_expect == XOPERATOR) {
3449 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3452 return ','; /* grandfather non-comma-format format */
3456 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3457 PL_tokenbuf[0] = '@';
3458 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3459 sizeof PL_tokenbuf - 1, FALSE);
3460 if (PL_expect == XOPERATOR)
3461 no_op("Array length", s);
3462 if (!PL_tokenbuf[1])
3464 PL_expect = XOPERATOR;
3465 PL_pending_ident = '#';
3469 PL_tokenbuf[0] = '$';
3470 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3471 sizeof PL_tokenbuf - 1, FALSE);
3472 if (PL_expect == XOPERATOR)
3474 if (!PL_tokenbuf[1]) {
3476 yyerror("Final $ should be \\$ or $name");
3480 /* This kludge not intended to be bulletproof. */
3481 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3482 yylval.opval = newSVOP(OP_CONST, 0,
3483 newSViv(PL_compiling.cop_arybase));
3484 yylval.opval->op_private = OPpCONST_ARYBASE;
3490 if (PL_lex_state == LEX_NORMAL)
3493 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3496 PL_tokenbuf[0] = '@';
3497 if (ckWARN(WARN_SYNTAX)) {
3499 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3502 PL_bufptr = skipspace(PL_bufptr);
3503 while (t < PL_bufend && *t != ']')
3505 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3506 "Multidimensional syntax %.*s not supported",
3507 (t - PL_bufptr) + 1, PL_bufptr);
3511 else if (*s == '{') {
3512 PL_tokenbuf[0] = '%';
3513 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3514 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3516 char tmpbuf[sizeof PL_tokenbuf];
3518 for (t++; isSPACE(*t); t++) ;
3519 if (isIDFIRST_lazy_if(t,UTF)) {
3520 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3521 for (; isSPACE(*t); t++) ;
3522 if (*t == ';' && get_cv(tmpbuf, FALSE))
3523 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3524 "You need to quote \"%s\"", tmpbuf);
3530 PL_expect = XOPERATOR;
3531 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3532 bool islop = (PL_last_lop == PL_oldoldbufptr);
3533 if (!islop || PL_last_lop_op == OP_GREPSTART)
3534 PL_expect = XOPERATOR;
3535 else if (strchr("$@\"'`q", *s))
3536 PL_expect = XTERM; /* e.g. print $fh "foo" */
3537 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3538 PL_expect = XTERM; /* e.g. print $fh &sub */
3539 else if (isIDFIRST_lazy_if(s,UTF)) {
3540 char tmpbuf[sizeof PL_tokenbuf];
3541 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3542 if ((tmp = keyword(tmpbuf, len))) {
3543 /* binary operators exclude handle interpretations */
3555 PL_expect = XTERM; /* e.g. print $fh length() */
3560 PL_expect = XTERM; /* e.g. print $fh subr() */
3563 else if (isDIGIT(*s))
3564 PL_expect = XTERM; /* e.g. print $fh 3 */
3565 else if (*s == '.' && isDIGIT(s[1]))
3566 PL_expect = XTERM; /* e.g. print $fh .3 */
3567 else if (strchr("?-+", *s) && !isSPACE(s[1]) && s[1] != '=')
3568 PL_expect = XTERM; /* e.g. print $fh -1 */
3569 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3570 PL_expect = XTERM; /* e.g. print $fh /.../
3571 XXX except DORDOR operator */
3572 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3573 PL_expect = XTERM; /* print $fh <<"EOF" */
3575 PL_pending_ident = '$';
3579 if (PL_expect == XOPERATOR)
3581 PL_tokenbuf[0] = '@';
3582 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3583 if (!PL_tokenbuf[1]) {
3586 if (PL_lex_state == LEX_NORMAL)
3588 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3590 PL_tokenbuf[0] = '%';
3592 /* Warn about @ where they meant $. */
3593 if (ckWARN(WARN_SYNTAX)) {
3594 if (*s == '[' || *s == '{') {
3596 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3598 if (*t == '}' || *t == ']') {
3600 PL_bufptr = skipspace(PL_bufptr);
3601 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3602 "Scalar value %.*s better written as $%.*s",
3603 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3608 PL_pending_ident = '@';
3611 case '/': /* may be division, defined-or, or pattern */
3612 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3616 case '?': /* may either be conditional or pattern */
3617 if(PL_expect == XOPERATOR) {
3625 /* A // operator. */
3635 /* Disable warning on "study /blah/" */
3636 if (PL_oldoldbufptr == PL_last_uni
3637 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3638 || memNE(PL_last_uni, "study", 5)
3639 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3642 s = scan_pat(s,OP_MATCH);
3643 TERM(sublex_start());
3647 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3648 #ifdef PERL_STRICT_CR
3651 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3653 && (s == PL_linestart || s[-1] == '\n') )
3655 PL_lex_formbrack = 0;
3659 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3665 yylval.ival = OPf_SPECIAL;
3671 if (PL_expect != XOPERATOR)
3676 case '0': case '1': case '2': case '3': case '4':
3677 case '5': case '6': case '7': case '8': case '9':
3678 s = scan_num(s, &yylval);
3679 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3680 "### Saw number in '%s'\n", s);
3682 if (PL_expect == XOPERATOR)
3687 s = scan_str(s,FALSE,FALSE);
3688 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3689 "### Saw string before '%s'\n", s);
3691 if (PL_expect == XOPERATOR) {
3692 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3695 return ','; /* grandfather non-comma-format format */
3701 missingterm((char*)0);
3702 yylval.ival = OP_CONST;
3703 TERM(sublex_start());
3706 s = scan_str(s,FALSE,FALSE);
3707 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3708 "### Saw string before '%s'\n", s);
3710 if (PL_expect == XOPERATOR) {
3711 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3714 return ','; /* grandfather non-comma-format format */
3720 missingterm((char*)0);
3721 yylval.ival = OP_CONST;
3722 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3723 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3724 yylval.ival = OP_STRINGIFY;
3728 TERM(sublex_start());
3731 s = scan_str(s,FALSE,FALSE);
3732 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3733 "### Saw backtick string before '%s'\n", s);
3735 if (PL_expect == XOPERATOR)
3736 no_op("Backticks",s);
3738 missingterm((char*)0);
3739 yylval.ival = OP_BACKTICK;
3741 TERM(sublex_start());
3745 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3746 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3748 if (PL_expect == XOPERATOR)
3749 no_op("Backslash",s);
3753 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3757 while (isDIGIT(*start) || *start == '_')
3759 if (*start == '.' && isDIGIT(start[1])) {
3760 s = scan_num(s, &yylval);
3763 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3764 else if (!isALPHA(*start) && (PL_expect == XTERM
3765 || PL_expect == XREF || PL_expect == XSTATE
3766 || PL_expect == XTERMORDORDOR)) {
3770 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3773 s = scan_num(s, &yylval);
3780 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3820 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3822 /* Some keywords can be followed by any delimiter, including ':' */
3823 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3824 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3825 (PL_tokenbuf[0] == 'q' &&
3826 strchr("qwxr", PL_tokenbuf[1])))));
3828 /* x::* is just a word, unless x is "CORE" */
3829 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
3833 while (d < PL_bufend && isSPACE(*d))
3834 d++; /* no comments skipped here, or s### is misparsed */
3836 /* Is this a label? */
3837 if (!tmp && PL_expect == XSTATE
3838 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
3840 yylval.pval = savepv(PL_tokenbuf);
3845 /* Check for keywords */
3846 tmp = keyword(PL_tokenbuf, len);
3848 /* Is this a word before a => operator? */
3849 if (*d == '=' && d[1] == '>') {
3851 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0));
3852 yylval.opval->op_private = OPpCONST_BARE;
3853 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
3854 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
3858 if (tmp < 0) { /* second-class keyword? */
3859 GV *ogv = Nullgv; /* override (winner) */
3860 GV *hgv = Nullgv; /* hidden (loser) */
3861 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
3863 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
3866 if (GvIMPORTED_CV(gv))
3868 else if (! CvMETHOD(cv))
3872 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
3873 (gv = *gvp) != (GV*)&PL_sv_undef &&
3874 GvCVu(gv) && GvIMPORTED_CV(gv))
3881 tmp = 0; /* overridden by import or by GLOBAL */
3884 && -tmp==KEY_lock /* XXX generalizable kludge */
3886 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
3888 tmp = 0; /* any sub overrides "weak" keyword */
3890 else { /* no override */
3892 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
3893 Perl_warner(aTHX_ packWARN(WARN_MISC),
3894 "dump() better written as CORE::dump()");
3898 if (ckWARN(WARN_AMBIGUOUS) && hgv
3899 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
3900 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3901 "Ambiguous call resolved as CORE::%s(), %s",
3902 GvENAME(hgv), "qualify as such or use &");
3909 default: /* not a keyword */
3913 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
3915 /* Get the rest if it looks like a package qualifier */
3917 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
3919 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
3922 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
3923 *s == '\'' ? "'" : "::");
3928 if (PL_expect == XOPERATOR) {
3929 if (PL_bufptr == PL_linestart) {
3930 CopLINE_dec(PL_curcop);
3931 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3932 CopLINE_inc(PL_curcop);
3935 no_op("Bareword",s);
3938 /* Look for a subroutine with this name in current package,
3939 unless name is "Foo::", in which case Foo is a bearword
3940 (and a package name). */
3943 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
3945 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
3946 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
3947 "Bareword \"%s\" refers to nonexistent package",
3950 PL_tokenbuf[len] = '\0';
3957 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
3960 /* if we saw a global override before, get the right name */
3963 sv = newSVpvn("CORE::GLOBAL::",14);
3964 sv_catpv(sv,PL_tokenbuf);
3967 sv = newSVpv(PL_tokenbuf,0);
3969 /* Presume this is going to be a bareword of some sort. */
3972 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3973 yylval.opval->op_private = OPpCONST_BARE;
3974 /* UTF-8 package name? */
3975 if (UTF && !IN_BYTES &&
3976 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
3979 /* And if "Foo::", then that's what it certainly is. */
3984 /* See if it's the indirect object for a list operator. */
3986 if (PL_oldoldbufptr &&
3987 PL_oldoldbufptr < PL_bufptr &&
3988 (PL_oldoldbufptr == PL_last_lop
3989 || PL_oldoldbufptr == PL_last_uni) &&
3990 /* NO SKIPSPACE BEFORE HERE! */
3991 (PL_expect == XREF ||
3992 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
3994 bool immediate_paren = *s == '(';
3996 /* (Now we can afford to cross potential line boundary.) */
3999 /* Two barewords in a row may indicate method call. */
4001 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4004 /* If not a declared subroutine, it's an indirect object. */
4005 /* (But it's an indir obj regardless for sort.) */
4007 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4008 ((!gv || !GvCVu(gv)) &&
4009 (PL_last_lop_op != OP_MAPSTART &&
4010 PL_last_lop_op != OP_GREPSTART))))
4012 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4017 PL_expect = XOPERATOR;
4020 /* Is this a word before a => operator? */
4021 if (*s == '=' && s[1] == '>' && !pkgname) {
4023 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4024 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4025 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4029 /* If followed by a paren, it's certainly a subroutine. */
4032 if (gv && GvCVu(gv)) {
4033 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4034 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4039 PL_nextval[PL_nexttoke].opval = yylval.opval;
4040 PL_expect = XOPERATOR;
4046 /* If followed by var or block, call it a method (unless sub) */
4048 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4049 PL_last_lop = PL_oldbufptr;
4050 PL_last_lop_op = OP_METHOD;
4054 /* If followed by a bareword, see if it looks like indir obj. */
4057 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4058 && (tmp = intuit_method(s,gv)))
4061 /* Not a method, so call it a subroutine (if defined) */
4063 if (gv && GvCVu(gv)) {
4065 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4066 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4067 "Ambiguous use of -%s resolved as -&%s()",
4068 PL_tokenbuf, PL_tokenbuf);
4069 /* Check for a constant sub */
4071 if ((sv = cv_const_sv(cv))) {
4073 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4074 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4075 yylval.opval->op_private = 0;
4079 /* Resolve to GV now. */
4080 op_free(yylval.opval);
4081 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4082 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4083 PL_last_lop = PL_oldbufptr;
4084 PL_last_lop_op = OP_ENTERSUB;
4085 /* Is there a prototype? */
4088 char *proto = SvPV((SV*)cv, len);
4091 if (strEQ(proto, "$"))
4093 while (*proto == ';')
4095 if (*proto == '&' && *s == '{') {
4096 sv_setpv(PL_subname, PL_curstash ?
4097 "__ANON__" : "__ANON__::__ANON__");
4101 PL_nextval[PL_nexttoke].opval = yylval.opval;
4107 /* Call it a bare word */
4109 if (PL_hints & HINT_STRICT_SUBS)
4110 yylval.opval->op_private |= OPpCONST_STRICT;
4113 if (ckWARN(WARN_RESERVED)) {
4114 if (lastchar != '-') {
4115 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4116 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4117 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4124 if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) {
4125 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4126 "Operator or semicolon missing before %c%s",
4127 lastchar, PL_tokenbuf);
4128 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4129 "Ambiguous use of %c resolved as operator %c",
4130 lastchar, lastchar);
4136 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4137 newSVpv(CopFILE(PL_curcop),0));
4141 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4142 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4145 case KEY___PACKAGE__:
4146 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4148 ? newSVsv(PL_curstname)
4157 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4158 char *pname = "main";
4159 if (PL_tokenbuf[2] == 'D')
4160 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4161 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4164 GvIOp(gv) = newIO();
4165 IoIFP(GvIOp(gv)) = PL_rsfp;
4166 #if defined(HAS_FCNTL) && defined(F_SETFD)
4168 int fd = PerlIO_fileno(PL_rsfp);
4169 fcntl(fd,F_SETFD,fd >= 3);
4172 /* Mark this internal pseudo-handle as clean */
4173 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4175 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4176 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4177 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4179 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4180 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4181 /* if the script was opened in binmode, we need to revert
4182 * it to text mode for compatibility; but only iff it has CRs
4183 * XXX this is a questionable hack at best. */
4184 if (PL_bufend-PL_bufptr > 2
4185 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4188 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4189 loc = PerlIO_tell(PL_rsfp);
4190 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4193 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4195 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4196 #endif /* NETWARE */
4197 #ifdef PERLIO_IS_STDIO /* really? */
4198 # if defined(__BORLANDC__)
4199 /* XXX see note in do_binmode() */
4200 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4204 PerlIO_seek(PL_rsfp, loc, 0);
4208 #ifdef PERLIO_LAYERS
4211 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4212 else if (PL_encoding) {
4219 XPUSHs(PL_encoding);
4221 call_method("name", G_SCALAR);
4225 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4226 Perl_form(aTHX_ ":encoding(%"SVf")",
4244 if (PL_expect == XSTATE) {
4251 if (*s == ':' && s[1] == ':') {
4254 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4255 if (!(tmp = keyword(PL_tokenbuf, len)))
4256 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4270 LOP(OP_ACCEPT,XTERM);
4276 LOP(OP_ATAN2,XTERM);
4282 LOP(OP_BINMODE,XTERM);
4285 LOP(OP_BLESS,XTERM);
4294 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4311 if (!PL_cryptseen) {
4312 PL_cryptseen = TRUE;
4316 LOP(OP_CRYPT,XTERM);
4319 LOP(OP_CHMOD,XTERM);
4322 LOP(OP_CHOWN,XTERM);
4325 LOP(OP_CONNECT,XTERM);
4341 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4345 PL_hints |= HINT_BLOCK_SCOPE;
4355 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4356 LOP(OP_DBMOPEN,XTERM);
4362 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4369 yylval.ival = CopLINE(PL_curcop);
4383 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4384 UNIBRACK(OP_ENTEREVAL);
4402 case KEY_endhostent:
4408 case KEY_endservent:
4411 case KEY_endprotoent:
4422 yylval.ival = CopLINE(PL_curcop);
4424 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4426 if ((PL_bufend - p) >= 3 &&
4427 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4429 else if ((PL_bufend - p) >= 4 &&
4430 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4433 if (isIDFIRST_lazy_if(p,UTF)) {
4434 p = scan_ident(p, PL_bufend,
4435 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4439 Perl_croak(aTHX_ "Missing $ on loop variable");
4444 LOP(OP_FORMLINE,XTERM);
4450 LOP(OP_FCNTL,XTERM);
4456 LOP(OP_FLOCK,XTERM);
4465 LOP(OP_GREPSTART, XREF);
4468 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4483 case KEY_getpriority:
4484 LOP(OP_GETPRIORITY,XTERM);
4486 case KEY_getprotobyname:
4489 case KEY_getprotobynumber:
4490 LOP(OP_GPBYNUMBER,XTERM);
4492 case KEY_getprotoent:
4504 case KEY_getpeername:
4505 UNI(OP_GETPEERNAME);
4507 case KEY_gethostbyname:
4510 case KEY_gethostbyaddr:
4511 LOP(OP_GHBYADDR,XTERM);
4513 case KEY_gethostent:
4516 case KEY_getnetbyname:
4519 case KEY_getnetbyaddr:
4520 LOP(OP_GNBYADDR,XTERM);
4525 case KEY_getservbyname:
4526 LOP(OP_GSBYNAME,XTERM);
4528 case KEY_getservbyport:
4529 LOP(OP_GSBYPORT,XTERM);
4531 case KEY_getservent:
4534 case KEY_getsockname:
4535 UNI(OP_GETSOCKNAME);
4537 case KEY_getsockopt:
4538 LOP(OP_GSOCKOPT,XTERM);
4560 yylval.ival = CopLINE(PL_curcop);
4564 LOP(OP_INDEX,XTERM);
4570 LOP(OP_IOCTL,XTERM);
4582 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4614 LOP(OP_LISTEN,XTERM);
4623 s = scan_pat(s,OP_MATCH);
4624 TERM(sublex_start());
4627 LOP(OP_MAPSTART, XREF);
4630 LOP(OP_MKDIR,XTERM);
4633 LOP(OP_MSGCTL,XTERM);
4636 LOP(OP_MSGGET,XTERM);
4639 LOP(OP_MSGRCV,XTERM);
4642 LOP(OP_MSGSND,XTERM);
4648 if (isIDFIRST_lazy_if(s,UTF)) {
4649 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4650 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4652 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4653 if (!PL_in_my_stash) {
4656 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4664 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4671 if (PL_expect != XSTATE)
4672 yyerror("\"no\" not allowed in expression");
4673 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4674 s = force_version(s, FALSE);
4679 if (*s == '(' || (s = skipspace(s), *s == '('))
4686 if (isIDFIRST_lazy_if(s,UTF)) {
4688 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4689 for (t=d; *t && isSPACE(*t); t++) ;
4690 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4692 && !(t[0] == '=' && t[1] == '>')
4694 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4695 "Precedence problem: open %.*s should be open(%.*s)",
4696 d - s, s, d - s, s);
4702 yylval.ival = OP_OR;
4712 LOP(OP_OPEN_DIR,XTERM);
4715 checkcomma(s,PL_tokenbuf,"filehandle");
4719 checkcomma(s,PL_tokenbuf,"filehandle");
4738 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4742 LOP(OP_PIPE_OP,XTERM);
4745 s = scan_str(s,FALSE,FALSE);
4747 missingterm((char*)0);
4748 yylval.ival = OP_CONST;
4749 TERM(sublex_start());
4755 s = scan_str(s,FALSE,FALSE);
4757 missingterm((char*)0);
4759 if (SvCUR(PL_lex_stuff)) {
4762 d = SvPV_force(PL_lex_stuff, len);
4765 for (; isSPACE(*d) && len; --len, ++d) ;
4768 if (!warned && ckWARN(WARN_QW)) {
4769 for (; !isSPACE(*d) && len; --len, ++d) {
4771 Perl_warner(aTHX_ packWARN(WARN_QW),
4772 "Possible attempt to separate words with commas");
4775 else if (*d == '#') {
4776 Perl_warner(aTHX_ packWARN(WARN_QW),
4777 "Possible attempt to put comments in qw() list");
4783 for (; !isSPACE(*d) && len; --len, ++d) ;
4785 sv = newSVpvn(b, d-b);
4786 if (DO_UTF8(PL_lex_stuff))
4788 words = append_elem(OP_LIST, words,
4789 newSVOP(OP_CONST, 0, tokeq(sv)));
4793 PL_nextval[PL_nexttoke].opval = words;
4798 SvREFCNT_dec(PL_lex_stuff);
4799 PL_lex_stuff = Nullsv;
4805 s = scan_str(s,FALSE,FALSE);
4807 missingterm((char*)0);
4808 yylval.ival = OP_STRINGIFY;
4809 if (SvIVX(PL_lex_stuff) == '\'')
4810 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4811 TERM(sublex_start());
4814 s = scan_pat(s,OP_QR);
4815 TERM(sublex_start());
4818 s = scan_str(s,FALSE,FALSE);
4820 missingterm((char*)0);
4821 yylval.ival = OP_BACKTICK;
4823 TERM(sublex_start());
4831 s = force_version(s, FALSE);
4833 else if (*s != 'v' || !isDIGIT(s[1])
4834 || (s = force_version(s, TRUE), *s == 'v'))
4836 *PL_tokenbuf = '\0';
4837 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4838 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
4839 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
4841 yyerror("<> should be quotes");
4849 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4853 LOP(OP_RENAME,XTERM);
4862 LOP(OP_RINDEX,XTERM);
4872 UNIDOR(OP_READLINE);
4885 LOP(OP_REVERSE,XTERM);
4888 UNIDOR(OP_READLINK);
4896 TERM(sublex_start());
4898 TOKEN(1); /* force error */
4907 LOP(OP_SELECT,XTERM);
4913 LOP(OP_SEMCTL,XTERM);
4916 LOP(OP_SEMGET,XTERM);
4919 LOP(OP_SEMOP,XTERM);
4925 LOP(OP_SETPGRP,XTERM);
4927 case KEY_setpriority:
4928 LOP(OP_SETPRIORITY,XTERM);
4930 case KEY_sethostent:
4936 case KEY_setservent:
4939 case KEY_setprotoent:
4949 LOP(OP_SEEKDIR,XTERM);
4951 case KEY_setsockopt:
4952 LOP(OP_SSOCKOPT,XTERM);
4958 LOP(OP_SHMCTL,XTERM);
4961 LOP(OP_SHMGET,XTERM);
4964 LOP(OP_SHMREAD,XTERM);
4967 LOP(OP_SHMWRITE,XTERM);
4970 LOP(OP_SHUTDOWN,XTERM);
4979 LOP(OP_SOCKET,XTERM);
4981 case KEY_socketpair:
4982 LOP(OP_SOCKPAIR,XTERM);
4985 checkcomma(s,PL_tokenbuf,"subroutine name");
4987 if (*s == ';' || *s == ')') /* probably a close */
4988 Perl_croak(aTHX_ "sort is now a reserved word");
4990 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4994 LOP(OP_SPLIT,XTERM);
4997 LOP(OP_SPRINTF,XTERM);
5000 LOP(OP_SPLICE,XTERM);
5015 LOP(OP_SUBSTR,XTERM);
5021 char tmpbuf[sizeof PL_tokenbuf];
5022 SSize_t tboffset = 0;
5023 expectation attrful;
5024 bool have_name, have_proto, bad_proto;
5029 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5030 (*s == ':' && s[1] == ':'))
5033 attrful = XATTRBLOCK;
5034 /* remember buffer pos'n for later force_word */
5035 tboffset = s - PL_oldbufptr;
5036 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5037 if (strchr(tmpbuf, ':'))
5038 sv_setpv(PL_subname, tmpbuf);
5040 sv_setsv(PL_subname,PL_curstname);
5041 sv_catpvn(PL_subname,"::",2);
5042 sv_catpvn(PL_subname,tmpbuf,len);
5049 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5050 PL_expect = XTERMBLOCK;
5051 attrful = XATTRTERM;
5052 sv_setpv(PL_subname,"?");
5056 if (key == KEY_format) {
5058 PL_lex_formbrack = PL_lex_brackets + 1;
5060 (void) force_word(PL_oldbufptr + tboffset, WORD,
5065 /* Look for a prototype */
5069 s = scan_str(s,FALSE,FALSE);
5071 Perl_croak(aTHX_ "Prototype not terminated");
5072 /* strip spaces and check for bad characters */
5073 d = SvPVX(PL_lex_stuff);
5076 for (p = d; *p; ++p) {
5079 if (!strchr("$@%*;[]&\\", *p))
5084 if (bad_proto && ckWARN(WARN_SYNTAX))
5085 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5086 "Illegal character in prototype for %"SVf" : %s",
5088 SvCUR(PL_lex_stuff) = tmp;
5096 if (*s == ':' && s[1] != ':')
5097 PL_expect = attrful;
5098 else if (!have_name && *s != '{' && key == KEY_sub)
5099 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5102 PL_nextval[PL_nexttoke].opval =
5103 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5104 PL_lex_stuff = Nullsv;
5108 sv_setpv(PL_subname,
5109 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5112 (void) force_word(PL_oldbufptr + tboffset, WORD,
5121 LOP(OP_SYSTEM,XREF);
5124 LOP(OP_SYMLINK,XTERM);
5127 LOP(OP_SYSCALL,XTERM);
5130 LOP(OP_SYSOPEN,XTERM);
5133 LOP(OP_SYSSEEK,XTERM);
5136 LOP(OP_SYSREAD,XTERM);
5139 LOP(OP_SYSWRITE,XTERM);
5143 TERM(sublex_start());
5164 LOP(OP_TRUNCATE,XTERM);
5176 yylval.ival = CopLINE(PL_curcop);
5180 yylval.ival = CopLINE(PL_curcop);
5184 LOP(OP_UNLINK,XTERM);
5190 LOP(OP_UNPACK,XTERM);
5193 LOP(OP_UTIME,XTERM);
5199 LOP(OP_UNSHIFT,XTERM);
5202 if (PL_expect != XSTATE)
5203 yyerror("\"use\" not allowed in expression");
5205 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5206 s = force_version(s, TRUE);
5207 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5208 PL_nextval[PL_nexttoke].opval = Nullop;
5211 else if (*s == 'v') {
5212 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5213 s = force_version(s, FALSE);
5217 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5218 s = force_version(s, FALSE);
5230 yylval.ival = CopLINE(PL_curcop);
5234 PL_hints |= HINT_BLOCK_SCOPE;
5241 LOP(OP_WAITPID,XTERM);
5250 ctl_l[0] = toCTRL('L');
5252 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5255 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5260 if (PL_expect == XOPERATOR)
5266 yylval.ival = OP_XOR;
5271 TERM(sublex_start());
5276 #pragma segment Main
5280 S_pending_ident(pTHX)
5283 register I32 tmp = 0;
5284 /* pit holds the identifier we read and pending_ident is reset */
5285 char pit = PL_pending_ident;
5286 PL_pending_ident = 0;
5288 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5289 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5291 /* if we're in a my(), we can't allow dynamics here.
5292 $foo'bar has already been turned into $foo::bar, so
5293 just check for colons.
5295 if it's a legal name, the OP is a PADANY.
5298 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5299 if (strchr(PL_tokenbuf,':'))
5300 yyerror(Perl_form(aTHX_ "No package name allowed for "
5301 "variable %s in \"our\"",
5303 tmp = allocmy(PL_tokenbuf);
5306 if (strchr(PL_tokenbuf,':'))
5307 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5309 yylval.opval = newOP(OP_PADANY, 0);
5310 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5316 build the ops for accesses to a my() variable.
5318 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5319 then used in a comparison. This catches most, but not
5320 all cases. For instance, it catches
5321 sort { my($a); $a <=> $b }
5323 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5324 (although why you'd do that is anyone's guess).
5327 if (!strchr(PL_tokenbuf,':')) {
5329 tmp = pad_findmy(PL_tokenbuf);
5330 if (tmp != NOT_IN_PAD) {
5331 /* might be an "our" variable" */
5332 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5333 /* build ops for a bareword */
5334 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5335 sv_catpvn(sym, "::", 2);
5336 sv_catpv(sym, PL_tokenbuf+1);
5337 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5338 yylval.opval->op_private = OPpCONST_ENTERED;
5339 gv_fetchpv(SvPVX(sym),
5341 ? (GV_ADDMULTI | GV_ADDINEVAL)
5344 ((PL_tokenbuf[0] == '$') ? SVt_PV
5345 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5350 /* if it's a sort block and they're naming $a or $b */
5351 if (PL_last_lop_op == OP_SORT &&
5352 PL_tokenbuf[0] == '$' &&
5353 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5356 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5357 d < PL_bufend && *d != '\n';
5360 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5361 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5367 yylval.opval = newOP(OP_PADANY, 0);
5368 yylval.opval->op_targ = tmp;
5374 Whine if they've said @foo in a doublequoted string,
5375 and @foo isn't a variable we can find in the symbol
5378 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5379 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5380 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5381 && ckWARN(WARN_AMBIGUOUS))
5383 /* Downgraded from fatal to warning 20000522 mjd */
5384 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5385 "Possible unintended interpolation of %s in string",
5390 /* build ops for a bareword */
5391 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5392 yylval.opval->op_private = OPpCONST_ENTERED;
5393 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5394 ((PL_tokenbuf[0] == '$') ? SVt_PV
5395 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5401 Perl_keyword(pTHX_ register char *d, I32 len)
5406 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5407 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5408 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5409 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5410 if (strEQ(d,"__END__")) return KEY___END__;
5414 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5419 if (strEQ(d,"and")) return -KEY_and;
5420 if (strEQ(d,"abs")) return -KEY_abs;
5423 if (strEQ(d,"alarm")) return -KEY_alarm;
5424 if (strEQ(d,"atan2")) return -KEY_atan2;
5427 if (strEQ(d,"accept")) return -KEY_accept;
5432 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5435 if (strEQ(d,"bless")) return -KEY_bless;
5436 if (strEQ(d,"bind")) return -KEY_bind;
5437 if (strEQ(d,"binmode")) return -KEY_binmode;
5440 if (strEQ(d,"CORE")) return -KEY_CORE;
5441 if (strEQ(d,"CHECK")) return KEY_CHECK;
5446 if (strEQ(d,"cmp")) return -KEY_cmp;
5447 if (strEQ(d,"chr")) return -KEY_chr;
5448 if (strEQ(d,"cos")) return -KEY_cos;
5451 if (strEQ(d,"chop")) return -KEY_chop;
5454 if (strEQ(d,"close")) return -KEY_close;
5455 if (strEQ(d,"chdir")) return -KEY_chdir;
5456 if (strEQ(d,"chomp")) return -KEY_chomp;
5457 if (strEQ(d,"chmod")) return -KEY_chmod;
5458 if (strEQ(d,"chown")) return -KEY_chown;
5459 if (strEQ(d,"crypt")) return -KEY_crypt;
5462 if (strEQ(d,"chroot")) return -KEY_chroot;
5463 if (strEQ(d,"caller")) return -KEY_caller;
5466 if (strEQ(d,"connect")) return -KEY_connect;
5469 if (strEQ(d,"closedir")) return -KEY_closedir;
5470 if (strEQ(d,"continue")) return -KEY_continue;
5475 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5480 if (strEQ(d,"do")) return KEY_do;
5483 if (strEQ(d,"die")) return -KEY_die;
5486 if (strEQ(d,"dump")) return -KEY_dump;
5489 if (strEQ(d,"delete")) return KEY_delete;
5492 if (strEQ(d,"defined")) return KEY_defined;
5493 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5496 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5501 if (strEQ(d,"END")) return KEY_END;
5506 if (strEQ(d,"eq")) return -KEY_eq;
5509 if (strEQ(d,"eof")) return -KEY_eof;
5510 if (strEQ(d,"err")) return -KEY_err;
5511 if (strEQ(d,"exp")) return -KEY_exp;
5514 if (strEQ(d,"else")) return KEY_else;
5515 if (strEQ(d,"exit")) return -KEY_exit;
5516 if (strEQ(d,"eval")) return KEY_eval;
5517 if (strEQ(d,"exec")) return -KEY_exec;
5518 if (strEQ(d,"each")) return -KEY_each;
5521 if (strEQ(d,"elsif")) return KEY_elsif;
5524 if (strEQ(d,"exists")) return KEY_exists;
5525 if (strEQ(d,"elseif") && ckWARN_d(WARN_SYNTAX))
5526 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5527 "elseif should be elsif");
5530 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5531 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5534 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5537 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5538 if (strEQ(d,"endservent")) return -KEY_endservent;
5541 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5548 if (strEQ(d,"for")) return KEY_for;
5551 if (strEQ(d,"fork")) return -KEY_fork;
5554 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5555 if (strEQ(d,"flock")) return -KEY_flock;
5558 if (strEQ(d,"format")) return KEY_format;
5559 if (strEQ(d,"fileno")) return -KEY_fileno;
5562 if (strEQ(d,"foreach")) return KEY_foreach;
5565 if (strEQ(d,"formline")) return -KEY_formline;
5570 if (strnEQ(d,"get",3)) {
5575 if (strEQ(d,"ppid")) return -KEY_getppid;
5576 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5579 if (strEQ(d,"pwent")) return -KEY_getpwent;
5580 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5581 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5584 if (strEQ(d,"peername")) return -KEY_getpeername;
5585 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5586 if (strEQ(d,"priority")) return -KEY_getpriority;
5589 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5592 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5596 else if (*d == 'h') {
5597 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5598 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5599 if (strEQ(d,"hostent")) return -KEY_gethostent;
5601 else if (*d == 'n') {
5602 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5603 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5604 if (strEQ(d,"netent")) return -KEY_getnetent;
5606 else if (*d == 's') {
5607 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5608 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5609 if (strEQ(d,"servent")) return -KEY_getservent;
5610 if (strEQ(d,"sockname")) return -KEY_getsockname;
5611 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5613 else if (*d == 'g') {
5614 if (strEQ(d,"grent")) return -KEY_getgrent;
5615 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5616 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5618 else if (*d == 'l') {
5619 if (strEQ(d,"login")) return -KEY_getlogin;
5621 else if (strEQ(d,"c")) return -KEY_getc;
5626 if (strEQ(d,"gt")) return -KEY_gt;
5627 if (strEQ(d,"ge")) return -KEY_ge;
5630 if (strEQ(d,"grep")) return KEY_grep;
5631 if (strEQ(d,"goto")) return KEY_goto;
5632 if (strEQ(d,"glob")) return KEY_glob;
5635 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5640 if (strEQ(d,"hex")) return -KEY_hex;
5643 if (strEQ(d,"INIT")) return KEY_INIT;
5648 if (strEQ(d,"if")) return KEY_if;
5651 if (strEQ(d,"int")) return -KEY_int;
5654 if (strEQ(d,"index")) return -KEY_index;
5655 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5660 if (strEQ(d,"join")) return -KEY_join;
5664 if (strEQ(d,"keys")) return -KEY_keys;
5665 if (strEQ(d,"kill")) return -KEY_kill;
5671 if (strEQ(d,"lt")) return -KEY_lt;
5672 if (strEQ(d,"le")) return -KEY_le;
5673 if (strEQ(d,"lc")) return -KEY_lc;
5676 if (strEQ(d,"log")) return -KEY_log;
5679 if (strEQ(d,"last")) return KEY_last;
5680 if (strEQ(d,"link")) return -KEY_link;
5681 if (strEQ(d,"lock")) return -KEY_lock;
5684 if (strEQ(d,"local")) return KEY_local;
5685 if (strEQ(d,"lstat")) return -KEY_lstat;
5688 if (strEQ(d,"length")) return -KEY_length;
5689 if (strEQ(d,"listen")) return -KEY_listen;
5692 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5695 if (strEQ(d,"localtime")) return -KEY_localtime;
5701 case 1: return KEY_m;
5703 if (strEQ(d,"my")) return KEY_my;
5706 if (strEQ(d,"map")) return KEY_map;
5709 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5712 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5713 if (strEQ(d,"msgget")) return -KEY_msgget;
5714 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5715 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5720 if (strEQ(d,"next")) return KEY_next;
5721 if (strEQ(d,"ne")) return -KEY_ne;
5722 if (strEQ(d,"not")) return -KEY_not;
5723 if (strEQ(d,"no")) return KEY_no;
5728 if (strEQ(d,"or")) return -KEY_or;
5731 if (strEQ(d,"ord")) return -KEY_ord;
5732 if (strEQ(d,"oct")) return -KEY_oct;
5733 if (strEQ(d,"our")) return KEY_our;
5736 if (strEQ(d,"open")) return -KEY_open;
5739 if (strEQ(d,"opendir")) return -KEY_opendir;
5746 if (strEQ(d,"pop")) return -KEY_pop;
5747 if (strEQ(d,"pos")) return KEY_pos;
5750 if (strEQ(d,"push")) return -KEY_push;
5751 if (strEQ(d,"pack")) return -KEY_pack;
5752 if (strEQ(d,"pipe")) return -KEY_pipe;
5755 if (strEQ(d,"print")) return KEY_print;
5758 if (strEQ(d,"printf")) return KEY_printf;
5761 if (strEQ(d,"package")) return KEY_package;
5764 if (strEQ(d,"prototype")) return KEY_prototype;
5769 if (strEQ(d,"q")) return KEY_q;
5770 if (strEQ(d,"qr")) return KEY_qr;
5771 if (strEQ(d,"qq")) return KEY_qq;
5772 if (strEQ(d,"qw")) return KEY_qw;
5773 if (strEQ(d,"qx")) return KEY_qx;
5775 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5780 if (strEQ(d,"ref")) return -KEY_ref;
5783 if (strEQ(d,"read")) return -KEY_read;
5784 if (strEQ(d,"rand")) return -KEY_rand;
5785 if (strEQ(d,"recv")) return -KEY_recv;
5786 if (strEQ(d,"redo")) return KEY_redo;
5789 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5790 if (strEQ(d,"reset")) return -KEY_reset;
5793 if (strEQ(d,"return")) return KEY_return;
5794 if (strEQ(d,"rename")) return -KEY_rename;
5795 if (strEQ(d,"rindex")) return -KEY_rindex;
5798 if (strEQ(d,"require")) return KEY_require;
5799 if (strEQ(d,"reverse")) return -KEY_reverse;
5800 if (strEQ(d,"readdir")) return -KEY_readdir;
5803 if (strEQ(d,"readlink")) return -KEY_readlink;
5804 if (strEQ(d,"readline")) return -KEY_readline;
5805 if (strEQ(d,"readpipe")) return -KEY_readpipe;
5808 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
5814 case 0: return KEY_s;
5816 if (strEQ(d,"scalar")) return KEY_scalar;
5821 if (strEQ(d,"seek")) return -KEY_seek;
5822 if (strEQ(d,"send")) return -KEY_send;
5825 if (strEQ(d,"semop")) return -KEY_semop;
5828 if (strEQ(d,"select")) return -KEY_select;
5829 if (strEQ(d,"semctl")) return -KEY_semctl;
5830 if (strEQ(d,"semget")) return -KEY_semget;
5833 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
5834 if (strEQ(d,"seekdir")) return -KEY_seekdir;
5837 if (strEQ(d,"setpwent")) return -KEY_setpwent;
5838 if (strEQ(d,"setgrent")) return -KEY_setgrent;
5841 if (strEQ(d,"setnetent")) return -KEY_setnetent;
5844 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
5845 if (strEQ(d,"sethostent")) return -KEY_sethostent;
5846 if (strEQ(d,"setservent")) return -KEY_setservent;
5849 if (strEQ(d,"setpriority")) return -KEY_setpriority;
5850 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
5857 if (strEQ(d,"shift")) return -KEY_shift;
5860 if (strEQ(d,"shmctl")) return -KEY_shmctl;
5861 if (strEQ(d,"shmget")) return -KEY_shmget;
5864 if (strEQ(d,"shmread")) return -KEY_shmread;
5867 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
5868 if (strEQ(d,"shutdown")) return -KEY_shutdown;
5873 if (strEQ(d,"sin")) return -KEY_sin;
5876 if (strEQ(d,"sleep")) return -KEY_sleep;
5879 if (strEQ(d,"sort")) return KEY_sort;
5880 if (strEQ(d,"socket")) return -KEY_socket;
5881 if (strEQ(d,"socketpair")) return -KEY_socketpair;
5884 if (strEQ(d,"split")) return KEY_split;
5885 if (strEQ(d,"sprintf")) return -KEY_sprintf;
5886 if (strEQ(d,"splice")) return -KEY_splice;
5889 if (strEQ(d,"sqrt")) return -KEY_sqrt;
5892 if (strEQ(d,"srand")) return -KEY_srand;
5895 if (strEQ(d,"stat")) return -KEY_stat;
5896 if (strEQ(d,"study")) return KEY_study;
5899 if (strEQ(d,"substr")) return -KEY_substr;
5900 if (strEQ(d,"sub")) return KEY_sub;
5905 if (strEQ(d,"system")) return -KEY_system;
5908 if (strEQ(d,"symlink")) return -KEY_symlink;
5909 if (strEQ(d,"syscall")) return -KEY_syscall;
5910 if (strEQ(d,"sysopen")) return -KEY_sysopen;
5911 if (strEQ(d,"sysread")) return -KEY_sysread;
5912 if (strEQ(d,"sysseek")) return -KEY_sysseek;
5915 if (strEQ(d,"syswrite")) return -KEY_syswrite;
5924 if (strEQ(d,"tr")) return KEY_tr;
5927 if (strEQ(d,"tie")) return KEY_tie;
5930 if (strEQ(d,"tell")) return -KEY_tell;
5931 if (strEQ(d,"tied")) return KEY_tied;
5932 if (strEQ(d,"time")) return -KEY_time;
5935 if (strEQ(d,"times")) return -KEY_times;
5938 if (strEQ(d,"telldir")) return -KEY_telldir;
5941 if (strEQ(d,"truncate")) return -KEY_truncate;
5948 if (strEQ(d,"uc")) return -KEY_uc;
5951 if (strEQ(d,"use")) return KEY_use;
5954 if (strEQ(d,"undef")) return KEY_undef;
5955 if (strEQ(d,"until")) return KEY_until;
5956 if (strEQ(d,"untie")) return KEY_untie;
5957 if (strEQ(d,"utime")) return -KEY_utime;
5958 if (strEQ(d,"umask")) return -KEY_umask;
5961 if (strEQ(d,"unless")) return KEY_unless;
5962 if (strEQ(d,"unpack")) return -KEY_unpack;
5963 if (strEQ(d,"unlink")) return -KEY_unlink;
5966 if (strEQ(d,"unshift")) return -KEY_unshift;
5967 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
5972 if (strEQ(d,"values")) return -KEY_values;
5973 if (strEQ(d,"vec")) return -KEY_vec;
5978 if (strEQ(d,"warn")) return -KEY_warn;
5979 if (strEQ(d,"wait")) return -KEY_wait;
5982 if (strEQ(d,"while")) return KEY_while;
5983 if (strEQ(d,"write")) return -KEY_write;
5986 if (strEQ(d,"waitpid")) return -KEY_waitpid;
5989 if (strEQ(d,"wantarray")) return -KEY_wantarray;
5994 if (len == 1) return -KEY_x;
5995 if (strEQ(d,"xor")) return -KEY_xor;
5998 if (len == 1) return KEY_y;
6007 S_checkcomma(pTHX_ register char *s, char *name, char *what)
6011 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
6012 if (ckWARN(WARN_SYNTAX)) {
6014 for (w = s+2; *w && level; w++) {
6021 for (; *w && isSPACE(*w); w++) ;
6022 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
6023 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6024 "%s (...) interpreted as function",name);
6027 while (s < PL_bufend && isSPACE(*s))
6031 while (s < PL_bufend && isSPACE(*s))
6033 if (isIDFIRST_lazy_if(s,UTF)) {
6035 while (isALNUM_lazy_if(s,UTF))
6037 while (s < PL_bufend && isSPACE(*s))
6042 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6046 Perl_croak(aTHX_ "No comma allowed after %s", what);
6051 /* Either returns sv, or mortalizes sv and returns a new SV*.
6052 Best used as sv=new_constant(..., sv, ...).
6053 If s, pv are NULL, calls subroutine with one argument,
6054 and type is used with error messages only. */
6057 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6061 HV *table = GvHV(PL_hintgv); /* ^H */
6065 const char *why1, *why2, *why3;
6067 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6070 why2 = strEQ(key,"charnames")
6071 ? "(possibly a missing \"use charnames ...\")"
6073 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6074 (type ? type: "undef"), why2);
6076 /* This is convoluted and evil ("goto considered harmful")
6077 * but I do not understand the intricacies of all the different
6078 * failure modes of %^H in here. The goal here is to make
6079 * the most probable error message user-friendly. --jhi */
6084 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6085 (type ? type: "undef"), why1, why2, why3);
6087 yyerror(SvPVX(msg));
6091 cvp = hv_fetch(table, key, strlen(key), FALSE);
6092 if (!cvp || !SvOK(*cvp)) {
6095 why3 = "} is not defined";
6098 sv_2mortal(sv); /* Parent created it permanently */
6101 pv = sv_2mortal(newSVpvn(s, len));
6103 typesv = sv_2mortal(newSVpv(type, 0));
6105 typesv = &PL_sv_undef;
6107 PUSHSTACKi(PERLSI_OVERLOAD);
6119 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6123 /* Check the eval first */
6124 if (!PL_in_eval && SvTRUE(ERRSV)) {
6126 sv_catpv(ERRSV, "Propagated");
6127 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6129 res = SvREFCNT_inc(sv);
6133 (void)SvREFCNT_inc(res);
6142 why1 = "Call to &{$^H{";
6144 why3 = "}} did not return a defined value";
6153 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6155 register char *d = dest;
6156 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6159 Perl_croak(aTHX_ ident_too_long);
6160 if (isALNUM(*s)) /* UTF handled below */
6162 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6167 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6171 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6172 char *t = s + UTF8SKIP(s);
6173 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6175 if (d + (t - s) > e)
6176 Perl_croak(aTHX_ ident_too_long);
6177 Copy(s, d, t - s, char);
6190 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6200 e = d + destlen - 3; /* two-character token, ending NUL */
6202 while (isDIGIT(*s)) {
6204 Perl_croak(aTHX_ ident_too_long);
6211 Perl_croak(aTHX_ ident_too_long);
6212 if (isALNUM(*s)) /* UTF handled below */
6214 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6219 else if (*s == ':' && s[1] == ':') {
6223 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6224 char *t = s + UTF8SKIP(s);
6225 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6227 if (d + (t - s) > e)
6228 Perl_croak(aTHX_ ident_too_long);
6229 Copy(s, d, t - s, char);
6240 if (PL_lex_state != LEX_NORMAL)
6241 PL_lex_state = LEX_INTERPENDMAYBE;
6244 if (*s == '$' && s[1] &&
6245 (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) )
6258 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6263 if (isSPACE(s[-1])) {
6266 if (!SPACE_OR_TAB(ch)) {
6272 if (isIDFIRST_lazy_if(d,UTF)) {
6276 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6278 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6281 Copy(s, d, e - s, char);
6286 while ((isALNUM(*s) || *s == ':') && d < e)
6289 Perl_croak(aTHX_ ident_too_long);
6292 while (s < send && SPACE_OR_TAB(*s)) s++;
6293 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6294 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6295 const char *brack = *s == '[' ? "[...]" : "{...}";
6296 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6297 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6298 funny, dest, brack, funny, dest, brack);
6301 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6305 /* Handle extended ${^Foo} variables
6306 * 1999-02-27 mjd-perl-patch@plover.com */
6307 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6311 while (isALNUM(*s) && d < e) {
6315 Perl_croak(aTHX_ ident_too_long);
6320 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6321 PL_lex_state = LEX_INTERPEND;
6326 if (PL_lex_state == LEX_NORMAL) {
6327 if (ckWARN(WARN_AMBIGUOUS) &&
6328 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6330 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6331 "Ambiguous use of %c{%s} resolved to %c%s",
6332 funny, dest, funny, dest);
6337 s = bracket; /* let the parser handle it */
6341 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6342 PL_lex_state = LEX_INTERPEND;
6347 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6352 *pmfl |= PMf_GLOBAL;
6354 *pmfl |= PMf_CONTINUE;
6358 *pmfl |= PMf_MULTILINE;
6360 *pmfl |= PMf_SINGLELINE;
6362 *pmfl |= PMf_EXTENDED;
6366 S_scan_pat(pTHX_ char *start, I32 type)
6371 s = scan_str(start,FALSE,FALSE);
6373 Perl_croak(aTHX_ "Search pattern not terminated");
6375 pm = (PMOP*)newPMOP(type, 0);
6376 if (PL_multi_open == '?')
6377 pm->op_pmflags |= PMf_ONCE;
6379 while (*s && strchr("iomsx", *s))
6380 pmflag(&pm->op_pmflags,*s++);
6383 while (*s && strchr("iogcmsx", *s))
6384 pmflag(&pm->op_pmflags,*s++);
6386 /* issue a warning if /c is specified,but /g is not */
6387 if (ckWARN(WARN_REGEXP) &&
6388 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6390 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6393 pm->op_pmpermflags = pm->op_pmflags;
6395 PL_lex_op = (OP*)pm;
6396 yylval.ival = OP_MATCH;
6401 S_scan_subst(pTHX_ char *start)
6408 yylval.ival = OP_NULL;
6410 s = scan_str(start,FALSE,FALSE);
6413 Perl_croak(aTHX_ "Substitution pattern not terminated");
6415 if (s[-1] == PL_multi_open)
6418 first_start = PL_multi_start;
6419 s = scan_str(s,FALSE,FALSE);
6422 SvREFCNT_dec(PL_lex_stuff);
6423 PL_lex_stuff = Nullsv;
6425 Perl_croak(aTHX_ "Substitution replacement not terminated");
6427 PL_multi_start = first_start; /* so whole substitution is taken together */
6429 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6435 else if (strchr("iogcmsx", *s))
6436 pmflag(&pm->op_pmflags,*s++);
6441 /* /c is not meaningful with s/// */
6442 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6444 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6449 PL_sublex_info.super_bufptr = s;
6450 PL_sublex_info.super_bufend = PL_bufend;
6452 pm->op_pmflags |= PMf_EVAL;
6453 repl = newSVpvn("",0);
6455 sv_catpv(repl, es ? "eval " : "do ");
6456 sv_catpvn(repl, "{ ", 2);
6457 sv_catsv(repl, PL_lex_repl);
6458 sv_catpvn(repl, " };", 2);
6460 SvREFCNT_dec(PL_lex_repl);
6464 pm->op_pmpermflags = pm->op_pmflags;
6465 PL_lex_op = (OP*)pm;
6466 yylval.ival = OP_SUBST;
6471 S_scan_trans(pTHX_ char *start)
6480 yylval.ival = OP_NULL;
6482 s = scan_str(start,FALSE,FALSE);
6484 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6485 if (s[-1] == PL_multi_open)
6488 s = scan_str(s,FALSE,FALSE);
6491 SvREFCNT_dec(PL_lex_stuff);
6492 PL_lex_stuff = Nullsv;
6494 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6497 complement = del = squash = 0;
6498 while (strchr("cds", *s)) {
6500 complement = OPpTRANS_COMPLEMENT;
6502 del = OPpTRANS_DELETE;
6504 squash = OPpTRANS_SQUASH;
6508 New(803, tbl, complement&&!del?258:256, short);
6509 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6510 o->op_private &= ~OPpTRANS_ALL;
6511 o->op_private |= del|squash|complement|
6512 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6513 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6516 yylval.ival = OP_TRANS;
6521 S_scan_heredoc(pTHX_ register char *s)
6524 I32 op_type = OP_SCALAR;
6531 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6535 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6538 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6539 if (*peek && strchr("`'\"",*peek)) {
6542 s = delimcpy(d, e, s, PL_bufend, term, &len);
6552 if (!isALNUM_lazy_if(s,UTF))
6553 deprecate_old("bare << to mean <<\"\"");
6554 for (; isALNUM_lazy_if(s,UTF); s++) {
6559 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6560 Perl_croak(aTHX_ "Delimiter for here document is too long");
6563 len = d - PL_tokenbuf;
6564 #ifndef PERL_STRICT_CR
6565 d = strchr(s, '\r');
6569 while (s < PL_bufend) {
6575 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6584 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6589 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6590 herewas = newSVpvn(s,PL_bufend-s);
6592 s--, herewas = newSVpvn(s,d-s);
6593 s += SvCUR(herewas);
6595 tmpstr = NEWSV(87,79);
6596 sv_upgrade(tmpstr, SVt_PVIV);
6601 else if (term == '`') {
6602 op_type = OP_BACKTICK;
6603 SvIVX(tmpstr) = '\\';
6607 PL_multi_start = CopLINE(PL_curcop);
6608 PL_multi_open = PL_multi_close = '<';
6609 term = *PL_tokenbuf;
6610 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6611 char *bufptr = PL_sublex_info.super_bufptr;
6612 char *bufend = PL_sublex_info.super_bufend;
6613 char *olds = s - SvCUR(herewas);
6614 s = strchr(bufptr, '\n');
6618 while (s < bufend &&
6619 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6621 CopLINE_inc(PL_curcop);
6624 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6625 missingterm(PL_tokenbuf);
6627 sv_setpvn(herewas,bufptr,d-bufptr+1);
6628 sv_setpvn(tmpstr,d+1,s-d);
6630 sv_catpvn(herewas,s,bufend-s);
6631 (void)strcpy(bufptr,SvPVX(herewas));
6638 while (s < PL_bufend &&
6639 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6641 CopLINE_inc(PL_curcop);
6643 if (s >= PL_bufend) {
6644 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6645 missingterm(PL_tokenbuf);
6647 sv_setpvn(tmpstr,d+1,s-d);
6649 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6651 sv_catpvn(herewas,s,PL_bufend-s);
6652 sv_setsv(PL_linestr,herewas);
6653 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6654 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6655 PL_last_lop = PL_last_uni = Nullch;
6658 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6659 while (s >= PL_bufend) { /* multiple line string? */
6661 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6662 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6663 missingterm(PL_tokenbuf);
6665 CopLINE_inc(PL_curcop);
6666 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6667 PL_last_lop = PL_last_uni = Nullch;
6668 #ifndef PERL_STRICT_CR
6669 if (PL_bufend - PL_linestart >= 2) {
6670 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6671 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6673 PL_bufend[-2] = '\n';
6675 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6677 else if (PL_bufend[-1] == '\r')
6678 PL_bufend[-1] = '\n';
6680 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6681 PL_bufend[-1] = '\n';
6683 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6684 SV *sv = NEWSV(88,0);
6686 sv_upgrade(sv, SVt_PVMG);
6687 sv_setsv(sv,PL_linestr);
6690 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6692 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6695 sv_catsv(PL_linestr,herewas);
6696 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6700 sv_catsv(tmpstr,PL_linestr);
6705 PL_multi_end = CopLINE(PL_curcop);
6706 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6707 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6708 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6710 SvREFCNT_dec(herewas);
6712 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6714 else if (PL_encoding)
6715 sv_recode_to_utf8(tmpstr, PL_encoding);
6717 PL_lex_stuff = tmpstr;
6718 yylval.ival = op_type;
6723 takes: current position in input buffer
6724 returns: new position in input buffer
6725 side-effects: yylval and lex_op are set.
6730 <FH> read from filehandle
6731 <pkg::FH> read from package qualified filehandle
6732 <pkg'FH> read from package qualified filehandle
6733 <$fh> read from filehandle in $fh
6739 S_scan_inputsymbol(pTHX_ char *start)
6741 register char *s = start; /* current position in buffer */
6747 d = PL_tokenbuf; /* start of temp holding space */
6748 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6749 end = strchr(s, '\n');
6752 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6754 /* die if we didn't have space for the contents of the <>,
6755 or if it didn't end, or if we see a newline
6758 if (len >= sizeof PL_tokenbuf)
6759 Perl_croak(aTHX_ "Excessively long <> operator");
6761 Perl_croak(aTHX_ "Unterminated <> operator");
6766 Remember, only scalar variables are interpreted as filehandles by
6767 this code. Anything more complex (e.g., <$fh{$num}>) will be
6768 treated as a glob() call.
6769 This code makes use of the fact that except for the $ at the front,
6770 a scalar variable and a filehandle look the same.
6772 if (*d == '$' && d[1]) d++;
6774 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6775 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6778 /* If we've tried to read what we allow filehandles to look like, and
6779 there's still text left, then it must be a glob() and not a getline.
6780 Use scan_str to pull out the stuff between the <> and treat it
6781 as nothing more than a string.
6784 if (d - PL_tokenbuf != len) {
6785 yylval.ival = OP_GLOB;
6787 s = scan_str(start,FALSE,FALSE);
6789 Perl_croak(aTHX_ "Glob not terminated");
6793 bool readline_overriden = FALSE;
6794 GV *gv_readline = Nullgv;
6796 /* we're in a filehandle read situation */
6799 /* turn <> into <ARGV> */
6801 (void)strcpy(d,"ARGV");
6803 /* Check whether readline() is overriden */
6804 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
6805 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
6807 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
6808 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
6809 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
6810 readline_overriden = TRUE;
6812 /* if <$fh>, create the ops to turn the variable into a
6818 /* try to find it in the pad for this block, otherwise find
6819 add symbol table ops
6821 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
6822 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
6823 SV *sym = sv_2mortal(
6824 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
6825 sv_catpvn(sym, "::", 2);
6831 OP *o = newOP(OP_PADSV, 0);
6833 PL_lex_op = readline_overriden
6834 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6835 append_elem(OP_LIST, o,
6836 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
6837 : (OP*)newUNOP(OP_READLINE, 0, o);
6846 ? (GV_ADDMULTI | GV_ADDINEVAL)
6849 PL_lex_op = readline_overriden
6850 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6851 append_elem(OP_LIST,
6852 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
6853 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6854 : (OP*)newUNOP(OP_READLINE, 0,
6855 newUNOP(OP_RV2SV, 0,
6856 newGVOP(OP_GV, 0, gv)));
6858 if (!readline_overriden)
6859 PL_lex_op->op_flags |= OPf_SPECIAL;
6860 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
6861 yylval.ival = OP_NULL;
6864 /* If it's none of the above, it must be a literal filehandle
6865 (<Foo::BAR> or <FOO>) so build a simple readline OP */
6867 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
6868 PL_lex_op = readline_overriden
6869 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
6870 append_elem(OP_LIST,
6871 newGVOP(OP_GV, 0, gv),
6872 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
6873 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
6874 yylval.ival = OP_NULL;
6883 takes: start position in buffer
6884 keep_quoted preserve \ on the embedded delimiter(s)
6885 keep_delims preserve the delimiters around the string
6886 returns: position to continue reading from buffer
6887 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
6888 updates the read buffer.
6890 This subroutine pulls a string out of the input. It is called for:
6891 q single quotes q(literal text)
6892 ' single quotes 'literal text'
6893 qq double quotes qq(interpolate $here please)
6894 " double quotes "interpolate $here please"
6895 qx backticks qx(/bin/ls -l)
6896 ` backticks `/bin/ls -l`
6897 qw quote words @EXPORT_OK = qw( func() $spam )
6898 m// regexp match m/this/
6899 s/// regexp substitute s/this/that/
6900 tr/// string transliterate tr/this/that/
6901 y/// string transliterate y/this/that/
6902 ($*@) sub prototypes sub foo ($)
6903 (stuff) sub attr parameters sub foo : attr(stuff)
6904 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
6906 In most of these cases (all but <>, patterns and transliterate)
6907 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
6908 calls scan_str(). s/// makes yylex() call scan_subst() which calls
6909 scan_str(). tr/// and y/// make yylex() call scan_trans() which
6912 It skips whitespace before the string starts, and treats the first
6913 character as the delimiter. If the delimiter is one of ([{< then
6914 the corresponding "close" character )]}> is used as the closing
6915 delimiter. It allows quoting of delimiters, and if the string has
6916 balanced delimiters ([{<>}]) it allows nesting.
6918 On success, the SV with the resulting string is put into lex_stuff or,
6919 if that is already non-NULL, into lex_repl. The second case occurs only
6920 when parsing the RHS of the special constructs s/// and tr/// (y///).
6921 For convenience, the terminating delimiter character is stuffed into
6926 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
6928 SV *sv; /* scalar value: string */
6929 char *tmps; /* temp string, used for delimiter matching */
6930 register char *s = start; /* current position in the buffer */
6931 register char term; /* terminating character */
6932 register char *to; /* current position in the sv's data */
6933 I32 brackets = 1; /* bracket nesting level */
6934 bool has_utf8 = FALSE; /* is there any utf8 content? */
6935 I32 termcode; /* terminating char. code */
6936 U8 termstr[UTF8_MAXLEN]; /* terminating string */
6937 STRLEN termlen; /* length of terminating string */
6938 char *last = NULL; /* last position for nesting bracket */
6940 /* skip space before the delimiter */
6944 /* mark where we are, in case we need to report errors */
6947 /* after skipping whitespace, the next character is the terminator */
6950 termcode = termstr[0] = term;
6954 termcode = utf8_to_uvchr((U8*)s, &termlen);
6955 Copy(s, termstr, termlen, U8);
6956 if (!UTF8_IS_INVARIANT(term))
6960 /* mark where we are */
6961 PL_multi_start = CopLINE(PL_curcop);
6962 PL_multi_open = term;
6964 /* find corresponding closing delimiter */
6965 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
6966 termcode = termstr[0] = term = tmps[5];
6968 PL_multi_close = term;
6970 /* create a new SV to hold the contents. 87 is leak category, I'm
6971 assuming. 79 is the SV's initial length. What a random number. */
6973 sv_upgrade(sv, SVt_PVIV);
6974 SvIVX(sv) = termcode;
6975 (void)SvPOK_only(sv); /* validate pointer */
6977 /* move past delimiter and try to read a complete string */
6979 sv_catpvn(sv, s, termlen);
6982 if (PL_encoding && !UTF) {
6986 int offset = s - SvPVX(PL_linestr);
6987 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
6988 &offset, (char*)termstr, termlen);
6989 char *ns = SvPVX(PL_linestr) + offset;
6990 char *svlast = SvEND(sv) - 1;
6992 for (; s < ns; s++) {
6993 if (*s == '\n' && !PL_rsfp)
6994 CopLINE_inc(PL_curcop);
6997 goto read_more_line;
6999 /* handle quoted delimiters */
7000 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
7002 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
7004 if ((svlast-1 - t) % 2) {
7008 SvCUR_set(sv, SvCUR(sv) - 1);
7013 if (PL_multi_open == PL_multi_close) {
7020 for (w = t = last; t < svlast; w++, t++) {
7021 /* At here, all closes are "was quoted" one,
7022 so we don't check PL_multi_close. */
7024 if (!keep_quoted && *(t+1) == PL_multi_open)
7029 else if (*t == PL_multi_open)
7037 SvCUR_set(sv, w - SvPVX(sv));
7040 if (--brackets <= 0)
7046 SvCUR_set(sv, SvCUR(sv) - 1);
7052 /* extend sv if need be */
7053 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7054 /* set 'to' to the next character in the sv's string */
7055 to = SvPVX(sv)+SvCUR(sv);
7057 /* if open delimiter is the close delimiter read unbridle */
7058 if (PL_multi_open == PL_multi_close) {
7059 for (; s < PL_bufend; s++,to++) {
7060 /* embedded newlines increment the current line number */
7061 if (*s == '\n' && !PL_rsfp)
7062 CopLINE_inc(PL_curcop);
7063 /* handle quoted delimiters */
7064 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7065 if (!keep_quoted && s[1] == term)
7067 /* any other quotes are simply copied straight through */
7071 /* terminate when run out of buffer (the for() condition), or
7072 have found the terminator */
7073 else if (*s == term) {
7076 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7079 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7085 /* if the terminator isn't the same as the start character (e.g.,
7086 matched brackets), we have to allow more in the quoting, and
7087 be prepared for nested brackets.
7090 /* read until we run out of string, or we find the terminator */
7091 for (; s < PL_bufend; s++,to++) {
7092 /* embedded newlines increment the line count */
7093 if (*s == '\n' && !PL_rsfp)
7094 CopLINE_inc(PL_curcop);
7095 /* backslashes can escape the open or closing characters */
7096 if (*s == '\\' && s+1 < PL_bufend) {
7098 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7103 /* allow nested opens and closes */
7104 else if (*s == PL_multi_close && --brackets <= 0)
7106 else if (*s == PL_multi_open)
7108 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7113 /* terminate the copied string and update the sv's end-of-string */
7115 SvCUR_set(sv, to - SvPVX(sv));
7118 * this next chunk reads more into the buffer if we're not done yet
7122 break; /* handle case where we are done yet :-) */
7124 #ifndef PERL_STRICT_CR
7125 if (to - SvPVX(sv) >= 2) {
7126 if ((to[-2] == '\r' && to[-1] == '\n') ||
7127 (to[-2] == '\n' && to[-1] == '\r'))
7131 SvCUR_set(sv, to - SvPVX(sv));
7133 else if (to[-1] == '\r')
7136 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7141 /* if we're out of file, or a read fails, bail and reset the current
7142 line marker so we can report where the unterminated string began
7145 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7147 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7150 /* we read a line, so increment our line counter */
7151 CopLINE_inc(PL_curcop);
7153 /* update debugger info */
7154 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7155 SV *sv = NEWSV(88,0);
7157 sv_upgrade(sv, SVt_PVMG);
7158 sv_setsv(sv,PL_linestr);
7161 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7164 /* having changed the buffer, we must update PL_bufend */
7165 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7166 PL_last_lop = PL_last_uni = Nullch;
7169 /* at this point, we have successfully read the delimited string */
7171 if (!PL_encoding || UTF) {
7173 sv_catpvn(sv, s, termlen);
7176 if (has_utf8 || PL_encoding)
7179 PL_multi_end = CopLINE(PL_curcop);
7181 /* if we allocated too much space, give some back */
7182 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7183 SvLEN_set(sv, SvCUR(sv) + 1);
7184 Renew(SvPVX(sv), SvLEN(sv), char);
7187 /* decide whether this is the first or second quoted string we've read
7200 takes: pointer to position in buffer
7201 returns: pointer to new position in buffer
7202 side-effects: builds ops for the constant in yylval.op
7204 Read a number in any of the formats that Perl accepts:
7206 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7207 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7210 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7212 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7215 If it reads a number without a decimal point or an exponent, it will
7216 try converting the number to an integer and see if it can do so
7217 without loss of precision.
7221 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7223 register char *s = start; /* current position in buffer */
7224 register char *d; /* destination in temp buffer */
7225 register char *e; /* end of temp buffer */
7226 NV nv; /* number read, as a double */
7227 SV *sv = Nullsv; /* place to put the converted number */
7228 bool floatit; /* boolean: int or float? */
7229 char *lastub = 0; /* position of last underbar */
7230 static char number_too_long[] = "Number too long";
7232 /* We use the first character to decide what type of number this is */
7236 Perl_croak(aTHX_ "panic: scan_num");
7238 /* if it starts with a 0, it could be an octal number, a decimal in
7239 0.13 disguise, or a hexadecimal number, or a binary number. */
7243 u holds the "number so far"
7244 shift the power of 2 of the base
7245 (hex == 4, octal == 3, binary == 1)
7246 overflowed was the number more than we can hold?
7248 Shift is used when we add a digit. It also serves as an "are
7249 we in octal/hex/binary?" indicator to disallow hex characters
7255 bool overflowed = FALSE;
7256 bool just_zero = TRUE; /* just plain 0 or binary number? */
7257 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7258 static char* bases[5] = { "", "binary", "", "octal",
7260 static char* Bases[5] = { "", "Binary", "", "Octal",
7262 static char *maxima[5] = { "",
7263 "0b11111111111111111111111111111111",
7267 char *base, *Base, *max;
7274 } else if (s[1] == 'b') {
7279 /* check for a decimal in disguise */
7280 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7282 /* so it must be octal */
7289 if (ckWARN(WARN_SYNTAX))
7290 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7291 "Misplaced _ in number");
7295 base = bases[shift];
7296 Base = Bases[shift];
7297 max = maxima[shift];
7299 /* read the rest of the number */
7301 /* x is used in the overflow test,
7302 b is the digit we're adding on. */
7307 /* if we don't mention it, we're done */
7311 /* _ are ignored -- but warned about if consecutive */
7313 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7314 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7315 "Misplaced _ in number");
7319 /* 8 and 9 are not octal */
7322 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7326 case '2': case '3': case '4':
7327 case '5': case '6': case '7':
7329 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7333 b = *s++ & 15; /* ASCII digit -> value of digit */
7337 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7338 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7339 /* make sure they said 0x */
7344 /* Prepare to put the digit we have onto the end
7345 of the number so far. We check for overflows.
7351 x = u << shift; /* make room for the digit */
7353 if ((x >> shift) != u
7354 && !(PL_hints & HINT_NEW_BINARY)) {
7357 if (ckWARN_d(WARN_OVERFLOW))
7358 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7359 "Integer overflow in %s number",
7362 u = x | b; /* add the digit to the end */
7365 n *= nvshift[shift];
7366 /* If an NV has not enough bits in its
7367 * mantissa to represent an UV this summing of
7368 * small low-order numbers is a waste of time
7369 * (because the NV cannot preserve the
7370 * low-order bits anyway): we could just
7371 * remember when did we overflow and in the
7372 * end just multiply n by the right
7380 /* if we get here, we had success: make a scalar value from
7385 /* final misplaced underbar check */
7387 if (ckWARN(WARN_SYNTAX))
7388 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7393 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7394 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7395 "%s number > %s non-portable",
7401 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7402 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7403 "%s number > %s non-portable",
7408 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
7409 sv = new_constant(start, s - start, "integer",
7411 else if (PL_hints & HINT_NEW_BINARY)
7412 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7417 handle decimal numbers.
7418 we're also sent here when we read a 0 as the first digit
7420 case '1': case '2': case '3': case '4': case '5':
7421 case '6': case '7': case '8': case '9': case '.':
7424 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7427 /* read next group of digits and _ and copy into d */
7428 while (isDIGIT(*s) || *s == '_') {
7429 /* skip underscores, checking for misplaced ones
7433 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7434 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7435 "Misplaced _ in number");
7439 /* check for end of fixed-length buffer */
7441 Perl_croak(aTHX_ number_too_long);
7442 /* if we're ok, copy the character */
7447 /* final misplaced underbar check */
7448 if (lastub && s == lastub + 1) {
7449 if (ckWARN(WARN_SYNTAX))
7450 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7453 /* read a decimal portion if there is one. avoid
7454 3..5 being interpreted as the number 3. followed
7457 if (*s == '.' && s[1] != '.') {
7462 if (ckWARN(WARN_SYNTAX))
7463 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7464 "Misplaced _ in number");
7468 /* copy, ignoring underbars, until we run out of digits.
7470 for (; isDIGIT(*s) || *s == '_'; s++) {
7471 /* fixed length buffer check */
7473 Perl_croak(aTHX_ number_too_long);
7475 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7476 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7477 "Misplaced _ in number");
7483 /* fractional part ending in underbar? */
7485 if (ckWARN(WARN_SYNTAX))
7486 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7487 "Misplaced _ in number");
7489 if (*s == '.' && isDIGIT(s[1])) {
7490 /* oops, it's really a v-string, but without the "v" */
7496 /* read exponent part, if present */
7497 if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) {
7501 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7502 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7504 /* stray preinitial _ */
7506 if (ckWARN(WARN_SYNTAX))
7507 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7508 "Misplaced _ in number");
7512 /* allow positive or negative exponent */
7513 if (*s == '+' || *s == '-')
7516 /* stray initial _ */
7518 if (ckWARN(WARN_SYNTAX))
7519 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7520 "Misplaced _ in number");
7524 /* read digits of exponent */
7525 while (isDIGIT(*s) || *s == '_') {
7528 Perl_croak(aTHX_ number_too_long);
7532 if (ckWARN(WARN_SYNTAX) &&
7533 ((lastub && s == lastub + 1) ||
7534 (!isDIGIT(s[1]) && s[1] != '_')))
7535 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7536 "Misplaced _ in number");
7543 /* make an sv from the string */
7547 We try to do an integer conversion first if no characters
7548 indicating "float" have been found.
7553 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7555 if (flags == IS_NUMBER_IN_UV) {
7557 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7560 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7561 if (uv <= (UV) IV_MIN)
7562 sv_setiv(sv, -(IV)uv);
7569 /* terminate the string */
7571 nv = Atof(PL_tokenbuf);
7575 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7576 (PL_hints & HINT_NEW_INTEGER) )
7577 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7578 (floatit ? "float" : "integer"),
7582 /* if it starts with a v, it could be a v-string */
7585 sv = NEWSV(92,5); /* preallocate storage space */
7586 s = scan_vstring(s,sv);
7590 /* make the op for the constant and return */
7593 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7595 lvalp->opval = Nullop;
7601 S_scan_formline(pTHX_ register char *s)
7605 SV *stuff = newSVpvn("",0);
7606 bool needargs = FALSE;
7612 #ifdef PERL_STRICT_CR
7613 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7615 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7617 if (*t == '\n' || t == PL_bufend) {
7622 if (PL_in_eval && !PL_rsfp) {
7623 eol = memchr(s,'\n',PL_bufend-s);
7628 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7630 for (t = s; t < eol; t++) {
7631 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7633 goto enough; /* ~~ must be first line in formline */
7635 if (*t == '@' || *t == '^')
7639 sv_catpvn(stuff, s, eol-s);
7640 #ifndef PERL_STRICT_CR
7641 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7642 char *end = SvPVX(stuff) + SvCUR(stuff);
7654 s = filter_gets(PL_linestr, PL_rsfp, 0);
7655 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7656 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7657 PL_last_lop = PL_last_uni = Nullch;
7669 PL_lex_state = LEX_NORMAL;
7670 PL_nextval[PL_nexttoke].ival = 0;
7674 PL_lex_state = LEX_FORMLINE;
7676 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7678 else if (PL_encoding)
7679 sv_recode_to_utf8(stuff, PL_encoding);
7681 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7683 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7687 SvREFCNT_dec(stuff);
7689 PL_lex_formbrack = 0;
7700 PL_cshlen = strlen(PL_cshname);
7705 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7707 I32 oldsavestack_ix = PL_savestack_ix;
7708 CV* outsidecv = PL_compcv;
7711 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7713 SAVEI32(PL_subline);
7714 save_item(PL_subname);
7715 SAVESPTR(PL_compcv);
7717 PL_compcv = (CV*)NEWSV(1104,0);
7718 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7719 CvFLAGS(PL_compcv) |= flags;
7721 PL_subline = CopLINE(PL_curcop);
7722 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7723 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7724 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7726 return oldsavestack_ix;
7730 #pragma segment Perl_yylex
7733 Perl_yywarn(pTHX_ char *s)
7735 PL_in_eval |= EVAL_WARNONLY;
7737 PL_in_eval &= ~EVAL_WARNONLY;
7742 Perl_yyerror(pTHX_ char *s)
7745 char *context = NULL;
7749 if (!yychar || (yychar == ';' && !PL_rsfp))
7751 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7752 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7755 The code below is removed for NetWare because it abends/crashes on NetWare
7756 when the script has error such as not having the closing quotes like:
7758 Checking of white spaces is anyway done in NetWare code.
7761 while (isSPACE(*PL_oldoldbufptr))
7764 context = PL_oldoldbufptr;
7765 contlen = PL_bufptr - PL_oldoldbufptr;
7767 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7768 PL_oldbufptr != PL_bufptr) {
7771 The code below is removed for NetWare because it abends/crashes on NetWare
7772 when the script has error such as not having the closing quotes like:
7774 Checking of white spaces is anyway done in NetWare code.
7777 while (isSPACE(*PL_oldbufptr))
7780 context = PL_oldbufptr;
7781 contlen = PL_bufptr - PL_oldbufptr;
7783 else if (yychar > 255)
7784 where = "next token ???";
7785 else if (yychar == -2) { /* YYEMPTY */
7786 if (PL_lex_state == LEX_NORMAL ||
7787 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7788 where = "at end of line";
7789 else if (PL_lex_inpat)
7790 where = "within pattern";
7792 where = "within string";
7795 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
7797 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
7798 else if (isPRINT_LC(yychar))
7799 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
7801 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
7802 where = SvPVX(where_sv);
7804 msg = sv_2mortal(newSVpv(s, 0));
7805 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
7806 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
7808 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
7810 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
7811 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
7812 Perl_sv_catpvf(aTHX_ msg,
7813 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
7814 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
7817 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
7818 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
7821 if (PL_error_count >= 10) {
7822 if (PL_in_eval && SvCUR(ERRSV))
7823 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
7824 ERRSV, OutCopFILE(PL_curcop));
7826 Perl_croak(aTHX_ "%s has too many errors.\n",
7827 OutCopFILE(PL_curcop));
7830 PL_in_my_stash = Nullhv;
7834 #pragma segment Main
7838 S_swallow_bom(pTHX_ U8 *s)
7841 slen = SvCUR(PL_linestr);
7845 /* UTF-16 little-endian? (or UTF32-LE?) */
7846 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
7847 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
7848 #ifndef PERL_NO_UTF16_FILTER
7849 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
7852 if (PL_bufend > (char*)s) {
7856 filter_add(utf16rev_textfilter, NULL);
7857 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7859 (char*)utf16_to_utf8_reversed(s, news,
7860 PL_bufend - (char*)s - 1,
7862 sv_setpvn(PL_linestr, (const char*)news, newlen);
7864 SvUTF8_on(PL_linestr);
7865 s = (U8*)SvPVX(PL_linestr);
7866 PL_bufend = SvPVX(PL_linestr) + newlen;
7869 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
7874 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
7875 #ifndef PERL_NO_UTF16_FILTER
7876 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
7879 if (PL_bufend > (char *)s) {
7883 filter_add(utf16_textfilter, NULL);
7884 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
7886 (char*)utf16_to_utf8(s, news,
7887 PL_bufend - (char*)s,
7889 sv_setpvn(PL_linestr, (const char*)news, newlen);
7891 SvUTF8_on(PL_linestr);
7892 s = (U8*)SvPVX(PL_linestr);
7893 PL_bufend = SvPVX(PL_linestr) + newlen;
7896 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
7901 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
7902 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
7909 if (s[2] == 0xFE && s[3] == 0xFF) {
7910 /* UTF-32 big-endian */
7911 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
7914 else if (s[2] == 0 && s[3] != 0) {
7917 * are a good indicator of UTF-16BE. */
7918 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
7923 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
7926 * are a good indicator of UTF-16LE. */
7927 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
7936 * Restore a source filter.
7940 restore_rsfp(pTHX_ void *f)
7942 PerlIO *fp = (PerlIO*)f;
7944 if (PL_rsfp == PerlIO_stdin())
7945 PerlIO_clearerr(PL_rsfp);
7946 else if (PL_rsfp && (PL_rsfp != fp))
7947 PerlIO_close(PL_rsfp);
7951 #ifndef PERL_NO_UTF16_FILTER
7953 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7955 I32 count = FILTER_READ(idx+1, sv, maxlen);
7960 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7961 if (!*SvPV_nolen(sv))
7962 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7965 tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7966 sv_usepvn(sv, (char*)tmps, tend - tmps);
7972 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
7974 I32 count = FILTER_READ(idx+1, sv, maxlen);
7979 if (!*SvPV_nolen(sv))
7980 /* Game over, but don't feed an odd-length string to utf16_to_utf8 */
7983 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
7984 tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen);
7985 sv_usepvn(sv, (char*)tmps, tend - tmps);
7992 Returns a pointer to the next character after the parsed
7993 vstring, as well as updating the passed in sv.
7995 Function must be called like
7998 s = scan_vstring(s,sv);
8000 The sv should already be large enough to store the vstring
8001 passed in, for performance reasons.
8006 Perl_scan_vstring(pTHX_ char *s, SV *sv)
8010 if (*pos == 'v') pos++; /* get past 'v' */
8011 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8014 /* this may not be a v-string if followed by => */
8016 while (next < PL_bufend && isSPACE(*next))
8018 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
8019 /* return string not v-string */
8020 sv_setpvn(sv,(char *)s,pos-s);
8025 if (!isALPHA(*pos)) {
8027 U8 tmpbuf[UTF8_MAXLEN+1];
8030 if (*s == 'v') s++; /* get past 'v' */
8032 sv_setpvn(sv, "", 0);
8037 /* this is atoi() that tolerates underscores */
8040 while (--end >= s) {
8045 rev += (*end - '0') * mult;
8047 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
8048 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
8049 "Integer overflow in decimal number");
8053 if (rev > 0x7FFFFFFF)
8054 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
8056 /* Append native character for the rev point */
8057 tmpend = uvchr_to_utf8(tmpbuf, rev);
8058 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
8059 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
8061 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
8067 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8071 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);