3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 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
79 static char* lex_state_names[] = {
100 /* CLINE is a macro that ensures PL_copline has a sane value */
105 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
108 * Convenience functions to return different tokens and prime the
109 * lexer for the next token. They all take an argument.
111 * TOKEN : generic token (used for '(', DOLSHARP, etc)
112 * OPERATOR : generic operator
113 * AOPERATOR : assignment operator
114 * PREBLOCK : beginning the block after an if, while, foreach, ...
115 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
116 * PREREF : *EXPR where EXPR is not a simple identifier
117 * TERM : expression term
118 * LOOPX : loop exiting command (goto, last, dump, etc)
119 * FTST : file test operator
120 * FUN0 : zero-argument function
121 * FUN1 : not used, except for not, which isn't a UNIOP
122 * BOop : bitwise or or xor
124 * SHop : shift operator
125 * PWop : power operator
126 * PMop : pattern-matching operator
127 * Aop : addition-level operator
128 * Mop : multiplication-level operator
129 * Eop : equality-testing operator
130 * Rop : relational operator <= != gt
132 * Also see LOP and lop() below.
135 #ifdef DEBUGGING /* Serve -DT. */
136 # define REPORT(retval) tokereport(s,(int)retval)
138 # define REPORT(retval) (retval)
141 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
142 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
143 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
144 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
145 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
146 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
147 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
148 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
149 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
150 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
151 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
152 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
153 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
154 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
155 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
156 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
157 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
158 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
159 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
160 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
162 /* This bit of chicanery makes a unary function followed by
163 * a parenthesis into a function with one argument, highest precedence.
164 * The UNIDOR macro is for unary functions that can be followed by the //
165 * operator (such as C<shift // 0>).
167 #define UNI2(f,x) return ( \
171 PL_last_uni = PL_oldbufptr, \
172 PL_last_lop_op = f, \
174 (*s == '(' || (s = skipspace(s), *s == '(') \
175 ? (int)FUNC1 : (int)UNIOP)))
176 #define UNI(f) UNI2(f,XTERM)
177 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
179 #define UNIBRACK(f) return ( \
182 PL_last_uni = PL_oldbufptr, \
184 (*s == '(' || (s = skipspace(s), *s == '(') \
185 ? (int)FUNC1 : (int)UNIOP)))
187 /* grandfather return to old style */
188 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
192 /* how to interpret the yylval associated with the token */
196 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
202 static struct debug_tokens { int token, type; char *name;} debug_tokens[] =
204 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
205 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
206 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
207 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
208 { ARROW, TOKENTYPE_NONE, "ARROW" },
209 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
210 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
211 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
212 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
213 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
214 { DO, TOKENTYPE_NONE, "DO" },
215 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
216 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
217 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
218 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
219 { ELSE, TOKENTYPE_NONE, "ELSE" },
220 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
221 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
222 { FOR, TOKENTYPE_IVAL, "FOR" },
223 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
224 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
225 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
226 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
227 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
228 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
229 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
230 { IF, TOKENTYPE_IVAL, "IF" },
231 { LABEL, TOKENTYPE_PVAL, "LABEL" },
232 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
233 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
234 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
235 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
236 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
237 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
238 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
239 { MY, TOKENTYPE_IVAL, "MY" },
240 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
241 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
242 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
243 { OROP, TOKENTYPE_IVAL, "OROP" },
244 { OROR, TOKENTYPE_NONE, "OROR" },
245 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
246 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
247 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
248 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
249 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
250 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
251 { PREINC, TOKENTYPE_NONE, "PREINC" },
252 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
253 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
254 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
255 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
256 { SUB, TOKENTYPE_NONE, "SUB" },
257 { THING, TOKENTYPE_OPVAL, "THING" },
258 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
259 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
260 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
261 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
262 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
263 { USE, TOKENTYPE_IVAL, "USE" },
264 { WHILE, TOKENTYPE_IVAL, "WHILE" },
265 { WORD, TOKENTYPE_OPVAL, "WORD" },
266 { 0, TOKENTYPE_NONE, 0 }
269 /* dump the returned token in rv, plus any optional arg in yylval */
272 S_tokereport(pTHX_ char* s, I32 rv)
276 enum token_type type = TOKENTYPE_NONE;
277 struct debug_tokens *p;
278 SV* report = NEWSV(0, 60);
280 Perl_sv_catpvf(aTHX_ report, "<== ");
282 for (p = debug_tokens; p->token; p++) {
283 if (p->token == (int)rv) {
290 Perl_sv_catpvf(aTHX_ report, "%s", name);
291 else if ((char)rv > ' ' && (char)rv < '~')
292 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
294 Perl_sv_catpvf(aTHX_ report, "EOF");
296 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
299 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
302 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", yylval.ival);
304 case TOKENTYPE_OPNUM:
305 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
306 PL_op_name[yylval.ival]);
309 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
311 case TOKENTYPE_OPVAL:
313 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
314 PL_op_name[yylval.opval->op_type]);
316 Perl_sv_catpv(aTHX_ report, "(opval=null)");
319 Perl_sv_catpvf(aTHX_ report, " at line %d [", CopLINE(PL_curcop));
320 if (s - PL_bufptr > 0)
321 sv_catpvn(report, PL_bufptr, s - PL_bufptr);
323 if (PL_oldbufptr && *PL_oldbufptr)
324 sv_catpv(report, PL_tokenbuf);
326 PerlIO_printf(Perl_debug_log, "### %s]\n", SvPV_nolen(report));
336 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
337 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
341 S_ao(pTHX_ int toketype)
343 if (*PL_bufptr == '=') {
345 if (toketype == ANDAND)
346 yylval.ival = OP_ANDASSIGN;
347 else if (toketype == OROR)
348 yylval.ival = OP_ORASSIGN;
349 else if (toketype == DORDOR)
350 yylval.ival = OP_DORASSIGN;
358 * When Perl expects an operator and finds something else, no_op
359 * prints the warning. It always prints "<something> found where
360 * operator expected. It prints "Missing semicolon on previous line?"
361 * if the surprise occurs at the start of the line. "do you need to
362 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
363 * where the compiler doesn't know if foo is a method call or a function.
364 * It prints "Missing operator before end of line" if there's nothing
365 * after the missing operator, or "... before <...>" if there is something
366 * after the missing operator.
370 S_no_op(pTHX_ char *what, char *s)
372 char *oldbp = PL_bufptr;
373 bool is_first = (PL_oldbufptr == PL_linestart);
379 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
380 if (ckWARN_d(WARN_SYNTAX)) {
382 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
383 "\t(Missing semicolon on previous line?)\n");
384 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
386 for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ;
387 if (t < PL_bufptr && isSPACE(*t))
388 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
389 "\t(Do you need to predeclare %.*s?)\n",
390 t - PL_oldoldbufptr, PL_oldoldbufptr);
394 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
395 "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp);
403 * Complain about missing quote/regexp/heredoc terminator.
404 * If it's called with (char *)NULL then it cauterizes the line buffer.
405 * If we're in a delimited string and the delimiter is a control
406 * character, it's reformatted into a two-char sequence like ^C.
411 S_missingterm(pTHX_ char *s)
416 char *nl = strrchr(s,'\n');
422 iscntrl(PL_multi_close)
424 PL_multi_close < 32 || PL_multi_close == 127
428 tmpbuf[1] = toCTRL(PL_multi_close);
434 *tmpbuf = (char)PL_multi_close;
438 q = strchr(s,'"') ? '\'' : '"';
439 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
447 Perl_deprecate(pTHX_ char *s)
449 if (ckWARN(WARN_DEPRECATED))
450 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
454 Perl_deprecate_old(pTHX_ char *s)
456 /* This function should NOT be called for any new deprecated warnings */
457 /* Use Perl_deprecate instead */
459 /* It is here to maintain backward compatibility with the pre-5.8 */
460 /* warnings category hierarchy. The "deprecated" category used to */
461 /* live under the "syntax" category. It is now a top-level category */
462 /* in its own right. */
464 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
465 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
466 "Use of %s is deprecated", s);
471 * Deprecate a comma-less variable list.
477 deprecate_old("comma-less variable list");
481 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
482 * utf16-to-utf8-reversed.
485 #ifdef PERL_CR_FILTER
489 register char *s = SvPVX(sv);
490 register char *e = s + SvCUR(sv);
491 /* outer loop optimized to do nothing if there are no CR-LFs */
493 if (*s++ == '\r' && *s == '\n') {
494 /* hit a CR-LF, need to copy the rest */
495 register char *d = s - 1;
498 if (*s == '\r' && s[1] == '\n')
509 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
511 I32 count = FILTER_READ(idx+1, sv, maxlen);
512 if (count > 0 && !maxlen)
520 * Initialize variables. Uses the Perl save_stack to save its state (for
521 * recursive calls to the parser).
525 Perl_lex_start(pTHX_ SV *line)
530 SAVEI32(PL_lex_dojoin);
531 SAVEI32(PL_lex_brackets);
532 SAVEI32(PL_lex_casemods);
533 SAVEI32(PL_lex_starts);
534 SAVEI32(PL_lex_state);
535 SAVEVPTR(PL_lex_inpat);
536 SAVEI32(PL_lex_inwhat);
537 if (PL_lex_state == LEX_KNOWNEXT) {
538 I32 toke = PL_nexttoke;
539 while (--toke >= 0) {
540 SAVEI32(PL_nexttype[toke]);
541 SAVEVPTR(PL_nextval[toke]);
543 SAVEI32(PL_nexttoke);
545 SAVECOPLINE(PL_curcop);
548 SAVEPPTR(PL_oldbufptr);
549 SAVEPPTR(PL_oldoldbufptr);
550 SAVEPPTR(PL_last_lop);
551 SAVEPPTR(PL_last_uni);
552 SAVEPPTR(PL_linestart);
553 SAVESPTR(PL_linestr);
554 SAVEGENERICPV(PL_lex_brackstack);
555 SAVEGENERICPV(PL_lex_casestack);
556 SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp);
557 SAVESPTR(PL_lex_stuff);
558 SAVEI32(PL_lex_defer);
559 SAVEI32(PL_sublex_info.sub_inwhat);
560 SAVESPTR(PL_lex_repl);
562 SAVEINT(PL_lex_expect);
564 PL_lex_state = LEX_NORMAL;
568 New(899, PL_lex_brackstack, 120, char);
569 New(899, PL_lex_casestack, 12, char);
571 *PL_lex_casestack = '\0';
574 PL_lex_stuff = Nullsv;
575 PL_lex_repl = Nullsv;
579 PL_sublex_info.sub_inwhat = 0;
581 if (SvREADONLY(PL_linestr))
582 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
583 s = SvPV(PL_linestr, len);
584 if (!len || s[len-1] != ';') {
585 if (!(SvFLAGS(PL_linestr) & SVs_TEMP))
586 PL_linestr = sv_2mortal(newSVsv(PL_linestr));
587 sv_catpvn(PL_linestr, "\n;", 2);
589 SvTEMP_off(PL_linestr);
590 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
591 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
592 PL_last_lop = PL_last_uni = Nullch;
598 * Finalizer for lexing operations. Must be called when the parser is
599 * done with the lexer.
605 PL_doextract = FALSE;
610 * This subroutine has nothing to do with tilting, whether at windmills
611 * or pinball tables. Its name is short for "increment line". It
612 * increments the current line number in CopLINE(PL_curcop) and checks
613 * to see whether the line starts with a comment of the form
614 * # line 500 "foo.pm"
615 * If so, it sets the current line number and file to the values in the comment.
619 S_incline(pTHX_ char *s)
626 CopLINE_inc(PL_curcop);
629 while (SPACE_OR_TAB(*s)) s++;
630 if (strnEQ(s, "line", 4))
634 if (SPACE_OR_TAB(*s))
638 while (SPACE_OR_TAB(*s)) s++;
644 while (SPACE_OR_TAB(*s))
646 if (*s == '"' && (t = strchr(s+1, '"'))) {
651 for (t = s; !isSPACE(*t); t++) ;
654 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
656 if (*e != '\n' && *e != '\0')
657 return; /* false alarm */
662 CopFILE_free(PL_curcop);
663 CopFILE_set(PL_curcop, s);
666 CopLINE_set(PL_curcop, atoi(n)-1);
671 * Called to gobble the appropriate amount and type of whitespace.
672 * Skips comments as well.
676 S_skipspace(pTHX_ register char *s)
678 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
679 while (s < PL_bufend && SPACE_OR_TAB(*s))
685 SSize_t oldprevlen, oldoldprevlen;
686 SSize_t oldloplen = 0, oldunilen = 0;
687 while (s < PL_bufend && isSPACE(*s)) {
688 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
693 if (s < PL_bufend && *s == '#') {
694 while (s < PL_bufend && *s != '\n')
698 if (PL_in_eval && !PL_rsfp) {
705 /* only continue to recharge the buffer if we're at the end
706 * of the buffer, we're not reading from a source filter, and
707 * we're in normal lexing mode
709 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
710 PL_lex_state == LEX_FORMLINE)
713 /* try to recharge the buffer */
714 if ((s = filter_gets(PL_linestr, PL_rsfp,
715 (prevlen = SvCUR(PL_linestr)))) == Nullch)
717 /* end of file. Add on the -p or -n magic */
720 ";}continue{print or die qq(-p destination: $!\\n);}");
721 PL_minus_n = PL_minus_p = 0;
723 else if (PL_minus_n) {
724 sv_setpvn(PL_linestr, ";}", 2);
728 sv_setpvn(PL_linestr,";", 1);
730 /* reset variables for next time we lex */
731 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
733 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
734 PL_last_lop = PL_last_uni = Nullch;
736 /* Close the filehandle. Could be from -P preprocessor,
737 * STDIN, or a regular file. If we were reading code from
738 * STDIN (because the commandline held no -e or filename)
739 * then we don't close it, we reset it so the code can
740 * read from STDIN too.
743 if (PL_preprocess && !PL_in_eval)
744 (void)PerlProc_pclose(PL_rsfp);
745 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
746 PerlIO_clearerr(PL_rsfp);
748 (void)PerlIO_close(PL_rsfp);
753 /* not at end of file, so we only read another line */
754 /* make corresponding updates to old pointers, for yyerror() */
755 oldprevlen = PL_oldbufptr - PL_bufend;
756 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
758 oldunilen = PL_last_uni - PL_bufend;
760 oldloplen = PL_last_lop - PL_bufend;
761 PL_linestart = PL_bufptr = s + prevlen;
762 PL_bufend = s + SvCUR(PL_linestr);
764 PL_oldbufptr = s + oldprevlen;
765 PL_oldoldbufptr = s + oldoldprevlen;
767 PL_last_uni = s + oldunilen;
769 PL_last_lop = s + oldloplen;
772 /* debugger active and we're not compiling the debugger code,
773 * so store the line into the debugger's array of lines
775 if (PERLDB_LINE && PL_curstash != PL_debstash) {
776 SV *sv = NEWSV(85,0);
778 sv_upgrade(sv, SVt_PVMG);
779 sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr);
782 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
789 * Check the unary operators to ensure there's no ambiguity in how they're
790 * used. An ambiguous piece of code would be:
792 * This doesn't mean rand() + 5. Because rand() is a unary operator,
793 * the +5 is its argument.
802 if (PL_oldoldbufptr != PL_last_uni)
804 while (isSPACE(*PL_last_uni))
806 for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ;
807 if ((t = strchr(s, '(')) && t < PL_bufptr)
809 if (ckWARN_d(WARN_AMBIGUOUS)){
812 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
813 "Warning: Use of \"%s\" without parentheses is ambiguous",
820 * LOP : macro to build a list operator. Its behaviour has been replaced
821 * with a subroutine, S_lop() for which LOP is just another name.
824 #define LOP(f,x) return lop(f,x,s)
828 * Build a list operator (or something that might be one). The rules:
829 * - if we have a next token, then it's a list operator [why?]
830 * - if the next thing is an opening paren, then it's a function
831 * - else it's a list operator
835 S_lop(pTHX_ I32 f, int x, char *s)
841 PL_last_lop = PL_oldbufptr;
842 PL_last_lop_op = (OPCODE)f;
844 return REPORT(LSTOP);
851 return REPORT(LSTOP);
856 * When the lexer realizes it knows the next token (for instance,
857 * it is reordering tokens for the parser) then it can call S_force_next
858 * to know what token to return the next time the lexer is called. Caller
859 * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer
860 * handles the token correctly.
864 S_force_next(pTHX_ I32 type)
866 PL_nexttype[PL_nexttoke] = type;
868 if (PL_lex_state != LEX_KNOWNEXT) {
869 PL_lex_defer = PL_lex_state;
870 PL_lex_expect = PL_expect;
871 PL_lex_state = LEX_KNOWNEXT;
876 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
878 SV *sv = newSVpvn(start,len);
879 if (UTF && !IN_BYTES && is_utf8_string((U8*)start, len))
886 * When the lexer knows the next thing is a word (for instance, it has
887 * just seen -> and it knows that the next char is a word char, then
888 * it calls S_force_word to stick the next word into the PL_next lookahead.
891 * char *start : buffer position (must be within PL_linestr)
892 * int token : PL_next will be this type of bare word (e.g., METHOD,WORD)
893 * int check_keyword : if true, Perl checks to make sure the word isn't
894 * a keyword (do this if the word is a label, e.g. goto FOO)
895 * int allow_pack : if true, : characters will also be allowed (require,
897 * int allow_initial_tick : used by the "sub" lexer only.
901 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
906 start = skipspace(start);
908 if (isIDFIRST_lazy_if(s,UTF) ||
909 (allow_pack && *s == ':') ||
910 (allow_initial_tick && *s == '\'') )
912 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
913 if (check_keyword && keyword(PL_tokenbuf, len))
915 if (token == METHOD) {
920 PL_expect = XOPERATOR;
923 PL_nextval[PL_nexttoke].opval
924 = (OP*)newSVOP(OP_CONST,0,
925 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
926 PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE;
934 * Called when the lexer wants $foo *foo &foo etc, but the program
935 * text only contains the "foo" portion. The first argument is a pointer
936 * to the "foo", and the second argument is the type symbol to prefix.
937 * Forces the next token to be a "WORD".
938 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
942 S_force_ident(pTHX_ register char *s, int kind)
945 OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0));
946 PL_nextval[PL_nexttoke].opval = o;
949 o->op_private = OPpCONST_ENTERED;
950 /* XXX see note in pp_entereval() for why we forgo typo
951 warnings if the symbol must be introduced in an eval.
953 gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
954 kind == '$' ? SVt_PV :
955 kind == '@' ? SVt_PVAV :
956 kind == '%' ? SVt_PVHV :
964 Perl_str_to_version(pTHX_ SV *sv)
969 char *start = SvPVx(sv,len);
970 bool utf = SvUTF8(sv) ? TRUE : FALSE;
971 char *end = start + len;
972 while (start < end) {
976 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
981 retval += ((NV)n)/nshift;
990 * Forces the next token to be a version number.
991 * If the next token appears to be an invalid version number, (e.g. "v2b"),
992 * and if "guessing" is TRUE, then no new token is created (and the caller
993 * must use an alternative parsing method).
997 S_force_version(pTHX_ char *s, int guessing)
999 OP *version = Nullop;
1008 while (isDIGIT(*d) || *d == '_' || *d == '.')
1010 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1012 s = scan_num(s, &yylval);
1013 version = yylval.opval;
1014 ver = cSVOPx(version)->op_sv;
1015 if (SvPOK(ver) && !SvNIOK(ver)) {
1016 (void)SvUPGRADE(ver, SVt_PVNV);
1017 SvNVX(ver) = str_to_version(ver);
1018 SvNOK_on(ver); /* hint that it is a version */
1025 /* NOTE: The parser sees the package name and the VERSION swapped */
1026 PL_nextval[PL_nexttoke].opval = version;
1034 * Tokenize a quoted string passed in as an SV. It finds the next
1035 * chunk, up to end of string or a backslash. It may make a new
1036 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1041 S_tokeq(pTHX_ SV *sv)
1044 register char *send;
1052 s = SvPV_force(sv, len);
1053 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1056 while (s < send && *s != '\\')
1061 if ( PL_hints & HINT_NEW_STRING ) {
1062 pv = sv_2mortal(newSVpvn(SvPVX(pv), len));
1068 if (s + 1 < send && (s[1] == '\\'))
1069 s++; /* all that, just for this */
1074 SvCUR_set(sv, d - SvPVX(sv));
1076 if ( PL_hints & HINT_NEW_STRING )
1077 return new_constant(NULL, 0, "q", sv, pv, "q");
1082 * Now come three functions related to double-quote context,
1083 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1084 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1085 * interact with PL_lex_state, and create fake ( ... ) argument lists
1086 * to handle functions and concatenation.
1087 * They assume that whoever calls them will be setting up a fake
1088 * join call, because each subthing puts a ',' after it. This lets
1091 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1093 * (I'm not sure whether the spurious commas at the end of lcfirst's
1094 * arguments and join's arguments are created or not).
1099 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1101 * Pattern matching will set PL_lex_op to the pattern-matching op to
1102 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1104 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1106 * Everything else becomes a FUNC.
1108 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1109 * had an OP_CONST or OP_READLINE). This just sets us up for a
1110 * call to S_sublex_push().
1114 S_sublex_start(pTHX)
1116 register I32 op_type = yylval.ival;
1118 if (op_type == OP_NULL) {
1119 yylval.opval = PL_lex_op;
1123 if (op_type == OP_CONST || op_type == OP_READLINE) {
1124 SV *sv = tokeq(PL_lex_stuff);
1126 if (SvTYPE(sv) == SVt_PVIV) {
1127 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1133 nsv = newSVpvn(p, len);
1139 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1140 PL_lex_stuff = Nullsv;
1141 /* Allow <FH> // "foo" */
1142 if (op_type == OP_READLINE)
1143 PL_expect = XTERMORDORDOR;
1147 PL_sublex_info.super_state = PL_lex_state;
1148 PL_sublex_info.sub_inwhat = op_type;
1149 PL_sublex_info.sub_op = PL_lex_op;
1150 PL_lex_state = LEX_INTERPPUSH;
1154 yylval.opval = PL_lex_op;
1164 * Create a new scope to save the lexing state. The scope will be
1165 * ended in S_sublex_done. Returns a '(', starting the function arguments
1166 * to the uc, lc, etc. found before.
1167 * Sets PL_lex_state to LEX_INTERPCONCAT.
1175 PL_lex_state = PL_sublex_info.super_state;
1176 SAVEI32(PL_lex_dojoin);
1177 SAVEI32(PL_lex_brackets);
1178 SAVEI32(PL_lex_casemods);
1179 SAVEI32(PL_lex_starts);
1180 SAVEI32(PL_lex_state);
1181 SAVEVPTR(PL_lex_inpat);
1182 SAVEI32(PL_lex_inwhat);
1183 SAVECOPLINE(PL_curcop);
1184 SAVEPPTR(PL_bufptr);
1185 SAVEPPTR(PL_bufend);
1186 SAVEPPTR(PL_oldbufptr);
1187 SAVEPPTR(PL_oldoldbufptr);
1188 SAVEPPTR(PL_last_lop);
1189 SAVEPPTR(PL_last_uni);
1190 SAVEPPTR(PL_linestart);
1191 SAVESPTR(PL_linestr);
1192 SAVEGENERICPV(PL_lex_brackstack);
1193 SAVEGENERICPV(PL_lex_casestack);
1195 PL_linestr = PL_lex_stuff;
1196 PL_lex_stuff = Nullsv;
1198 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1199 = SvPVX(PL_linestr);
1200 PL_bufend += SvCUR(PL_linestr);
1201 PL_last_lop = PL_last_uni = Nullch;
1202 SAVEFREESV(PL_linestr);
1204 PL_lex_dojoin = FALSE;
1205 PL_lex_brackets = 0;
1206 New(899, PL_lex_brackstack, 120, char);
1207 New(899, PL_lex_casestack, 12, char);
1208 PL_lex_casemods = 0;
1209 *PL_lex_casestack = '\0';
1211 PL_lex_state = LEX_INTERPCONCAT;
1212 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1214 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1215 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1216 PL_lex_inpat = PL_sublex_info.sub_op;
1218 PL_lex_inpat = Nullop;
1225 * Restores lexer state after a S_sublex_push.
1231 if (!PL_lex_starts++) {
1232 SV *sv = newSVpvn("",0);
1233 if (SvUTF8(PL_linestr))
1235 PL_expect = XOPERATOR;
1236 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1240 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1241 PL_lex_state = LEX_INTERPCASEMOD;
1245 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1246 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1247 PL_linestr = PL_lex_repl;
1249 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1250 PL_bufend += SvCUR(PL_linestr);
1251 PL_last_lop = PL_last_uni = Nullch;
1252 SAVEFREESV(PL_linestr);
1253 PL_lex_dojoin = FALSE;
1254 PL_lex_brackets = 0;
1255 PL_lex_casemods = 0;
1256 *PL_lex_casestack = '\0';
1258 if (SvEVALED(PL_lex_repl)) {
1259 PL_lex_state = LEX_INTERPNORMAL;
1261 /* we don't clear PL_lex_repl here, so that we can check later
1262 whether this is an evalled subst; that means we rely on the
1263 logic to ensure sublex_done() is called again only via the
1264 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1267 PL_lex_state = LEX_INTERPCONCAT;
1268 PL_lex_repl = Nullsv;
1274 PL_bufend = SvPVX(PL_linestr);
1275 PL_bufend += SvCUR(PL_linestr);
1276 PL_expect = XOPERATOR;
1277 PL_sublex_info.sub_inwhat = 0;
1285 Extracts a pattern, double-quoted string, or transliteration. This
1288 It looks at lex_inwhat and PL_lex_inpat to find out whether it's
1289 processing a pattern (PL_lex_inpat is true), a transliteration
1290 (lex_inwhat & OP_TRANS is true), or a double-quoted string.
1292 Returns a pointer to the character scanned up to. Iff this is
1293 advanced from the start pointer supplied (ie if anything was
1294 successfully parsed), will leave an OP for the substring scanned
1295 in yylval. Caller must intuit reason for not parsing further
1296 by looking at the next characters herself.
1300 double-quoted style: \r and \n
1301 regexp special ones: \D \s
1303 backrefs: \1 (deprecated in substitution replacements)
1304 case and quoting: \U \Q \E
1305 stops on @ and $, but not for $ as tail anchor
1307 In transliterations:
1308 characters are VERY literal, except for - not at the start or end
1309 of the string, which indicates a range. scan_const expands the
1310 range to the full set of intermediate characters.
1312 In double-quoted strings:
1314 double-quoted style: \r and \n
1316 backrefs: \1 (deprecated)
1317 case and quoting: \U \Q \E
1320 scan_const does *not* construct ops to handle interpolated strings.
1321 It stops processing as soon as it finds an embedded $ or @ variable
1322 and leaves it to the caller to work out what's going on.
1324 @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @::foo.
1326 $ in pattern could be $foo or could be tail anchor. Assumption:
1327 it's a tail anchor if $ is the last thing in the string, or if it's
1328 followed by one of ")| \n\t"
1330 \1 (backreferences) are turned into $1
1332 The structure of the code is
1333 while (there's a character to process) {
1334 handle transliteration ranges
1335 skip regexp comments
1336 skip # initiated comments in //x patterns
1337 check for embedded @foo
1338 check for embedded scalars
1340 leave intact backslashes from leave (below)
1341 deprecate \1 in strings and sub replacements
1342 handle string-changing backslashes \l \U \Q \E, etc.
1343 switch (what was escaped) {
1344 handle - in a transliteration (becomes a literal -)
1345 handle \132 octal characters
1346 handle 0x15 hex characters
1347 handle \cV (control V)
1348 handle printf backslashes (\f, \r, \n, etc)
1350 } (end if backslash)
1351 } (end while character to read)
1356 S_scan_const(pTHX_ char *start)
1358 register char *send = PL_bufend; /* end of the constant */
1359 SV *sv = NEWSV(93, send - start); /* sv for the constant */
1360 register char *s = start; /* start of the constant */
1361 register char *d = SvPVX(sv); /* destination for copies */
1362 bool dorange = FALSE; /* are we in a translit range? */
1363 bool didrange = FALSE; /* did we just finish a range? */
1364 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1365 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1368 const char *leaveit = /* set of acceptably-backslashed characters */
1370 ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxz0123456789[{]} \t\n\r\f\v#"
1373 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1374 /* If we are doing a trans and we know we want UTF8 set expectation */
1375 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1376 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1380 while (s < send || dorange) {
1381 /* get transliterations out of the way (they're most literal) */
1382 if (PL_lex_inwhat == OP_TRANS) {
1383 /* expand a range A-Z to the full set of characters. AIE! */
1385 I32 i; /* current expanded character */
1386 I32 min; /* first character in range */
1387 I32 max; /* last character in range */
1390 char *c = (char*)utf8_hop((U8*)d, -1);
1394 *c = (char)UTF_TO_NATIVE(0xff);
1395 /* mark the range as done, and continue */
1401 i = d - SvPVX(sv); /* remember current offset */
1402 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1403 d = SvPVX(sv) + i; /* refresh d after realloc */
1404 d -= 2; /* eat the first char and the - */
1406 min = (U8)*d; /* first char in range */
1407 max = (U8)d[1]; /* last char in range */
1411 "Invalid range \"%c-%c\" in transliteration operator",
1412 (char)min, (char)max);
1416 if ((isLOWER(min) && isLOWER(max)) ||
1417 (isUPPER(min) && isUPPER(max))) {
1419 for (i = min; i <= max; i++)
1421 *d++ = NATIVE_TO_NEED(has_utf8,i);
1423 for (i = min; i <= max; i++)
1425 *d++ = NATIVE_TO_NEED(has_utf8,i);
1430 for (i = min; i <= max; i++)
1433 /* mark the range as done, and continue */
1439 /* range begins (ignore - as first or last char) */
1440 else if (*s == '-' && s+1 < send && s != start) {
1442 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
1445 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
1457 /* if we get here, we're not doing a transliteration */
1459 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
1460 except for the last char, which will be done separately. */
1461 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
1463 while (s+1 < send && *s != ')')
1464 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1466 else if (s[2] == '{' /* This should match regcomp.c */
1467 || ((s[2] == 'p' || s[2] == '?') && s[3] == '{'))
1470 char *regparse = s + (s[2] == '{' ? 3 : 4);
1473 while (count && (c = *regparse)) {
1474 if (c == '\\' && regparse[1])
1482 if (*regparse != ')')
1483 regparse--; /* Leave one char for continuation. */
1484 while (s < regparse)
1485 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1489 /* likewise skip #-initiated comments in //x patterns */
1490 else if (*s == '#' && PL_lex_inpat &&
1491 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
1492 while (s+1 < send && *s != '\n')
1493 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1496 /* check for embedded arrays
1497 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
1499 else if (*s == '@' && s[1]
1500 && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1])))
1503 /* check for embedded scalars. only stop if we're sure it's a
1506 else if (*s == '$') {
1507 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
1509 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
1510 break; /* in regexp, $ might be tail anchor */
1513 /* End of else if chain - OP_TRANS rejoin rest */
1516 if (*s == '\\' && s+1 < send) {
1519 /* some backslashes we leave behind */
1520 if (*leaveit && *s && strchr(leaveit, *s)) {
1521 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
1522 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1526 /* deprecate \1 in strings and substitution replacements */
1527 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
1528 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
1530 if (ckWARN(WARN_SYNTAX))
1531 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
1536 /* string-change backslash escapes */
1537 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
1542 /* if we get here, it's either a quoted -, or a digit */
1545 /* quoted - in transliterations */
1547 if (PL_lex_inwhat == OP_TRANS) {
1554 if (ckWARN(WARN_MISC) &&
1557 Perl_warner(aTHX_ packWARN(WARN_MISC),
1558 "Unrecognized escape \\%c passed through",
1560 /* default action is to copy the quoted character */
1561 goto default_action;
1564 /* \132 indicates an octal constant */
1565 case '0': case '1': case '2': case '3':
1566 case '4': case '5': case '6': case '7':
1570 uv = grok_oct(s, &len, &flags, NULL);
1573 goto NUM_ESCAPE_INSERT;
1575 /* \x24 indicates a hex constant */
1579 char* e = strchr(s, '}');
1580 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1581 PERL_SCAN_DISALLOW_PREFIX;
1586 yyerror("Missing right brace on \\x{}");
1590 uv = grok_hex(s, &len, &flags, NULL);
1596 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
1597 uv = grok_hex(s, &len, &flags, NULL);
1603 /* Insert oct or hex escaped character.
1604 * There will always enough room in sv since such
1605 * escapes will be longer than any UTF-8 sequence
1606 * they can end up as. */
1608 /* We need to map to chars to ASCII before doing the tests
1611 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
1612 if (!has_utf8 && uv > 255) {
1613 /* Might need to recode whatever we have
1614 * accumulated so far if it contains any
1617 * (Can't we keep track of that and avoid
1618 * this rescan? --jhi)
1622 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
1623 if (!NATIVE_IS_INVARIANT(*c)) {
1628 STRLEN offset = d - SvPVX(sv);
1630 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
1634 while (src >= (U8 *)SvPVX(sv)) {
1635 if (!NATIVE_IS_INVARIANT(*src)) {
1636 U8 ch = NATIVE_TO_ASCII(*src);
1637 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
1638 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
1648 if (has_utf8 || uv > 255) {
1649 d = (char*)uvchr_to_utf8((U8*)d, uv);
1651 if (PL_lex_inwhat == OP_TRANS &&
1652 PL_sublex_info.sub_op) {
1653 PL_sublex_info.sub_op->op_private |=
1654 (PL_lex_repl ? OPpTRANS_FROM_UTF
1667 /* \N{LATIN SMALL LETTER A} is a named character */
1671 char* e = strchr(s, '}');
1677 yyerror("Missing right brace on \\N{}");
1681 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
1683 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
1684 PERL_SCAN_DISALLOW_PREFIX;
1687 uv = grok_hex(s, &len, &flags, NULL);
1689 goto NUM_ESCAPE_INSERT;
1691 res = newSVpvn(s + 1, e - s - 1);
1692 res = new_constant( Nullch, 0, "charnames",
1693 res, Nullsv, "\\N{...}" );
1695 sv_utf8_upgrade(res);
1696 str = SvPV(res,len);
1697 #ifdef EBCDIC_NEVER_MIND
1698 /* charnames uses pack U and that has been
1699 * recently changed to do the below uni->native
1700 * mapping, so this would be redundant (and wrong,
1701 * the code point would be doubly converted).
1702 * But leave this in just in case the pack U change
1703 * gets revoked, but the semantics is still
1704 * desireable for charnames. --jhi */
1706 UV uv = utf8_to_uvchr((U8*)str, 0);
1709 U8 tmpbuf[UTF8_MAXLEN+1], *d;
1711 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
1712 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
1713 str = SvPV(res, len);
1717 if (!has_utf8 && SvUTF8(res)) {
1718 char *ostart = SvPVX(sv);
1719 SvCUR_set(sv, d - ostart);
1722 sv_utf8_upgrade(sv);
1723 /* this just broke our allocation above... */
1724 SvGROW(sv, (STRLEN)(send - start));
1725 d = SvPVX(sv) + SvCUR(sv);
1728 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
1729 char *odest = SvPVX(sv);
1731 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
1732 d = SvPVX(sv) + (d - odest);
1734 Copy(str, d, len, char);
1741 yyerror("Missing braces on \\N{}");
1744 /* \c is a control character */
1753 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
1756 yyerror("Missing control char name in \\c");
1760 /* printf-style backslashes, formfeeds, newlines, etc */
1762 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
1765 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
1768 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
1771 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
1774 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
1777 *d++ = ASCII_TO_NEED(has_utf8,'\033');
1780 *d++ = ASCII_TO_NEED(has_utf8,'\007');
1786 } /* end if (backslash) */
1789 /* If we started with encoded form, or already know we want it
1790 and then encode the next character */
1791 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
1793 UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
1794 STRLEN need = UNISKIP(NATIVE_TO_UNI(uv));
1797 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
1798 STRLEN off = d - SvPVX(sv);
1799 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
1801 d = (char*)uvchr_to_utf8((U8*)d, uv);
1805 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
1807 } /* while loop to process each character */
1809 /* terminate the string and set up the sv */
1811 SvCUR_set(sv, d - SvPVX(sv));
1812 if (SvCUR(sv) >= SvLEN(sv))
1813 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
1816 if (PL_encoding && !has_utf8) {
1817 sv_recode_to_utf8(sv, PL_encoding);
1823 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1824 PL_sublex_info.sub_op->op_private |=
1825 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1829 /* shrink the sv if we allocated more than we used */
1830 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1831 SvLEN_set(sv, SvCUR(sv) + 1);
1832 Renew(SvPVX(sv), SvLEN(sv), char);
1835 /* return the substring (via yylval) only if we parsed anything */
1836 if (s > PL_bufptr) {
1837 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
1838 sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"),
1840 ( PL_lex_inwhat == OP_TRANS
1842 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
1845 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1852 * Returns TRUE if there's more to the expression (e.g., a subscript),
1855 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
1857 * ->[ and ->{ return TRUE
1858 * { and [ outside a pattern are always subscripts, so return TRUE
1859 * if we're outside a pattern and it's not { or [, then return FALSE
1860 * if we're in a pattern and the first char is a {
1861 * {4,5} (any digits around the comma) returns FALSE
1862 * if we're in a pattern and the first char is a [
1864 * [SOMETHING] has a funky algorithm to decide whether it's a
1865 * character class or not. It has to deal with things like
1866 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
1867 * anything else returns TRUE
1870 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
1873 S_intuit_more(pTHX_ register char *s)
1875 if (PL_lex_brackets)
1877 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
1879 if (*s != '{' && *s != '[')
1884 /* In a pattern, so maybe we have {n,m}. */
1901 /* On the other hand, maybe we have a character class */
1904 if (*s == ']' || *s == '^')
1907 /* this is terrifying, and it works */
1908 int weight = 2; /* let's weigh the evidence */
1910 unsigned char un_char = 255, last_un_char;
1911 char *send = strchr(s,']');
1912 char tmpbuf[sizeof PL_tokenbuf * 4];
1914 if (!send) /* has to be an expression */
1917 Zero(seen,256,char);
1920 else if (isDIGIT(*s)) {
1922 if (isDIGIT(s[1]) && s[2] == ']')
1928 for (; s < send; s++) {
1929 last_un_char = un_char;
1930 un_char = (unsigned char)*s;
1935 weight -= seen[un_char] * 10;
1936 if (isALNUM_lazy_if(s+1,UTF)) {
1937 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
1938 if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV))
1943 else if (*s == '$' && s[1] &&
1944 strchr("[#!%*<>()-=",s[1])) {
1945 if (/*{*/ strchr("])} =",s[2]))
1954 if (strchr("wds]",s[1]))
1956 else if (seen['\''] || seen['"'])
1958 else if (strchr("rnftbxcav",s[1]))
1960 else if (isDIGIT(s[1])) {
1962 while (s[1] && isDIGIT(s[1]))
1972 if (strchr("aA01! ",last_un_char))
1974 if (strchr("zZ79~",s[1]))
1976 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
1977 weight -= 5; /* cope with negative subscript */
1980 if (!isALNUM(last_un_char)
1981 && !(last_un_char == '$' || last_un_char == '@'
1982 || last_un_char == '&')
1983 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
1988 if (keyword(tmpbuf, d - tmpbuf))
1991 if (un_char == last_un_char + 1)
1993 weight -= seen[un_char];
1998 if (weight >= 0) /* probably a character class */
2008 * Does all the checking to disambiguate
2010 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2011 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2013 * First argument is the stuff after the first token, e.g. "bar".
2015 * Not a method if bar is a filehandle.
2016 * Not a method if foo is a subroutine prototyped to take a filehandle.
2017 * Not a method if it's really "Foo $bar"
2018 * Method if it's "foo $bar"
2019 * Not a method if it's really "print foo $bar"
2020 * Method if it's really "foo package::" (interpreted as package->foo)
2021 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2022 * Not a method if bar is a filehandle or package, but is quoted with
2027 S_intuit_method(pTHX_ char *start, GV *gv)
2029 char *s = start + (*start == '$');
2030 char tmpbuf[sizeof PL_tokenbuf];
2038 if ((cv = GvCVu(gv))) {
2039 char *proto = SvPVX(cv);
2049 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2050 /* start is the beginning of the possible filehandle/object,
2051 * and s is the end of it
2052 * tmpbuf is a copy of it
2055 if (*start == '$') {
2056 if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf))
2061 return *s == '(' ? FUNCMETH : METHOD;
2063 if (!keyword(tmpbuf, len)) {
2064 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2069 indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV);
2070 if (indirgv && GvCVu(indirgv))
2072 /* filehandle or package name makes it a method */
2073 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) {
2075 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2076 return 0; /* no assumptions -- "=>" quotes bearword */
2078 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0,
2079 newSVpvn(tmpbuf,len));
2080 PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE;
2084 return *s == '(' ? FUNCMETH : METHOD;
2092 * Return a string of Perl code to load the debugger. If PERL5DB
2093 * is set, it will return the contents of that, otherwise a
2094 * compile-time require of perl5db.pl.
2101 char *pdb = PerlEnv_getenv("PERL5DB");
2105 SETERRNO(0,SS_NORMAL);
2106 return "BEGIN { require 'perl5db.pl' }";
2112 /* Encoded script support. filter_add() effectively inserts a
2113 * 'pre-processing' function into the current source input stream.
2114 * Note that the filter function only applies to the current source file
2115 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2117 * The datasv parameter (which may be NULL) can be used to pass
2118 * private data to this instance of the filter. The filter function
2119 * can recover the SV using the FILTER_DATA macro and use it to
2120 * store private buffers and state information.
2122 * The supplied datasv parameter is upgraded to a PVIO type
2123 * and the IoDIRP/IoANY field is used to store the function pointer,
2124 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2125 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2126 * private use must be set using malloc'd pointers.
2130 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2135 if (!PL_rsfp_filters)
2136 PL_rsfp_filters = newAV();
2138 datasv = NEWSV(255,0);
2139 if (!SvUPGRADE(datasv, SVt_PVIO))
2140 Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO");
2141 IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */
2142 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2143 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2144 (void*)funcp, SvPV_nolen(datasv)));
2145 av_unshift(PL_rsfp_filters, 1);
2146 av_store(PL_rsfp_filters, 0, datasv) ;
2151 /* Delete most recently added instance of this filter function. */
2153 Perl_filter_del(pTHX_ filter_t funcp)
2156 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", (void*)funcp));
2157 if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2159 /* if filter is on top of stack (usual case) just pop it off */
2160 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2161 if (IoANY(datasv) == (void *)funcp) {
2162 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2163 IoANY(datasv) = (void *)NULL;
2164 sv_free(av_pop(PL_rsfp_filters));
2168 /* we need to search for the correct entry and clear it */
2169 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2173 /* Invoke the idxth filter function for the current rsfp. */
2174 /* maxlen 0 = read one text line */
2176 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2181 if (!PL_rsfp_filters)
2183 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2184 /* Provide a default input filter to make life easy. */
2185 /* Note that we append to the line. This is handy. */
2186 DEBUG_P(PerlIO_printf(Perl_debug_log,
2187 "filter_read %d: from rsfp\n", idx));
2191 int old_len = SvCUR(buf_sv) ;
2193 /* ensure buf_sv is large enough */
2194 SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ;
2195 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){
2196 if (PerlIO_error(PL_rsfp))
2197 return -1; /* error */
2199 return 0 ; /* end of file */
2201 SvCUR_set(buf_sv, old_len + len) ;
2204 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2205 if (PerlIO_error(PL_rsfp))
2206 return -1; /* error */
2208 return 0 ; /* end of file */
2211 return SvCUR(buf_sv);
2213 /* Skip this filter slot if filter has been deleted */
2214 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2215 DEBUG_P(PerlIO_printf(Perl_debug_log,
2216 "filter_read %d: skipped (filter deleted)\n",
2218 return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */
2220 /* Get function pointer hidden within datasv */
2221 funcp = (filter_t)IoANY(datasv);
2222 DEBUG_P(PerlIO_printf(Perl_debug_log,
2223 "filter_read %d: via function %p (%s)\n",
2224 idx, (void*)funcp, SvPV_nolen(datasv)));
2225 /* Call function. The function is expected to */
2226 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2227 /* Return: <0:error, =0:eof, >0:not eof */
2228 return (*funcp)(aTHX_ idx, buf_sv, maxlen);
2232 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2234 #ifdef PERL_CR_FILTER
2235 if (!PL_rsfp_filters) {
2236 filter_add(S_cr_textfilter,NULL);
2239 if (PL_rsfp_filters) {
2241 SvCUR_set(sv, 0); /* start with empty line */
2242 if (FILTER_READ(0, sv, 0) > 0)
2243 return ( SvPVX(sv) ) ;
2248 return (sv_gets(sv, fp, append));
2252 S_find_in_my_stash(pTHX_ char *pkgname, I32 len)
2256 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2260 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2261 (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV)))
2263 return GvHV(gv); /* Foo:: */
2266 /* use constant CLASS => 'MyClass' */
2267 if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) {
2269 if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) {
2270 pkgname = SvPV_nolen(sv);
2274 return gv_stashpv(pkgname, FALSE);
2278 static char* exp_name[] =
2279 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
2280 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
2287 Works out what to call the token just pulled out of the input
2288 stream. The yacc parser takes care of taking the ops we return and
2289 stitching them into a tree.
2295 if read an identifier
2296 if we're in a my declaration
2297 croak if they tried to say my($foo::bar)
2298 build the ops for a my() declaration
2299 if it's an access to a my() variable
2300 are we in a sort block?
2301 croak if my($a); $a <=> $b
2302 build ops for access to a my() variable
2303 if in a dq string, and they've said @foo and we can't find @foo
2305 build ops for a bareword
2306 if we already built the token before, use it.
2311 #pragma segment Perl_yylex
2316 register char *s = PL_bufptr;
2323 I32 orig_keyword = 0;
2326 PerlIO_printf(Perl_debug_log, "### LEX_%s\n",
2327 lex_state_names[PL_lex_state]);
2329 /* check if there's an identifier for us to look at */
2330 if (PL_pending_ident)
2331 return REPORT(S_pending_ident(aTHX));
2333 /* no identifier pending identification */
2335 switch (PL_lex_state) {
2337 case LEX_NORMAL: /* Some compilers will produce faster */
2338 case LEX_INTERPNORMAL: /* code if we comment these out. */
2342 /* when we've already built the next token, just pull it out of the queue */
2345 yylval = PL_nextval[PL_nexttoke];
2347 PL_lex_state = PL_lex_defer;
2348 PL_expect = PL_lex_expect;
2349 PL_lex_defer = LEX_NORMAL;
2351 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2352 "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr,
2353 (IV)PL_nexttype[PL_nexttoke]); });
2355 return REPORT(PL_nexttype[PL_nexttoke]);
2357 /* interpolated case modifiers like \L \U, including \Q and \E.
2358 when we get here, PL_bufptr is at the \
2360 case LEX_INTERPCASEMOD:
2362 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
2363 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
2365 /* handle \E or end of string */
2366 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
2370 if (PL_lex_casemods) {
2371 oldmod = PL_lex_casestack[--PL_lex_casemods];
2372 PL_lex_casestack[PL_lex_casemods] = '\0';
2374 if (PL_bufptr != PL_bufend
2375 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
2377 PL_lex_state = LEX_INTERPCONCAT;
2381 if (PL_bufptr != PL_bufend)
2383 PL_lex_state = LEX_INTERPCONCAT;
2387 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2388 "### Saw case modifier at '%s'\n", PL_bufptr); });
2390 if (s[1] == '\\' && s[2] == 'E') {
2392 PL_lex_state = LEX_INTERPCONCAT;
2396 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
2397 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
2398 if ((*s == 'L' || *s == 'U') &&
2399 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
2400 PL_lex_casestack[--PL_lex_casemods] = '\0';
2403 if (PL_lex_casemods > 10)
2404 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
2405 PL_lex_casestack[PL_lex_casemods++] = *s;
2406 PL_lex_casestack[PL_lex_casemods] = '\0';
2407 PL_lex_state = LEX_INTERPCONCAT;
2408 PL_nextval[PL_nexttoke].ival = 0;
2411 PL_nextval[PL_nexttoke].ival = OP_LCFIRST;
2413 PL_nextval[PL_nexttoke].ival = OP_UCFIRST;
2415 PL_nextval[PL_nexttoke].ival = OP_LC;
2417 PL_nextval[PL_nexttoke].ival = OP_UC;
2419 PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA;
2421 Perl_croak(aTHX_ "panic: yylex");
2425 if (PL_lex_starts) {
2428 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2429 if (PL_lex_casemods == 1 && PL_lex_inpat)
2438 case LEX_INTERPPUSH:
2439 return REPORT(sublex_push());
2441 case LEX_INTERPSTART:
2442 if (PL_bufptr == PL_bufend)
2443 return REPORT(sublex_done());
2444 DEBUG_T({ PerlIO_printf(Perl_debug_log,
2445 "### Interpolated variable at '%s'\n", PL_bufptr); });
2447 PL_lex_dojoin = (*PL_bufptr == '@');
2448 PL_lex_state = LEX_INTERPNORMAL;
2449 if (PL_lex_dojoin) {
2450 PL_nextval[PL_nexttoke].ival = 0;
2452 force_ident("\"", '$');
2453 PL_nextval[PL_nexttoke].ival = 0;
2455 PL_nextval[PL_nexttoke].ival = 0;
2457 PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */
2460 if (PL_lex_starts++) {
2462 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2463 if (!PL_lex_casemods && PL_lex_inpat)
2470 case LEX_INTERPENDMAYBE:
2471 if (intuit_more(PL_bufptr)) {
2472 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
2478 if (PL_lex_dojoin) {
2479 PL_lex_dojoin = FALSE;
2480 PL_lex_state = LEX_INTERPCONCAT;
2483 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
2484 && SvEVALED(PL_lex_repl))
2486 if (PL_bufptr != PL_bufend)
2487 Perl_croak(aTHX_ "Bad evalled substitution pattern");
2488 PL_lex_repl = Nullsv;
2491 case LEX_INTERPCONCAT:
2493 if (PL_lex_brackets)
2494 Perl_croak(aTHX_ "panic: INTERPCONCAT");
2496 if (PL_bufptr == PL_bufend)
2497 return REPORT(sublex_done());
2499 if (SvIVX(PL_linestr) == '\'') {
2500 SV *sv = newSVsv(PL_linestr);
2503 else if ( PL_hints & HINT_NEW_RE )
2504 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
2505 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2509 s = scan_const(PL_bufptr);
2511 PL_lex_state = LEX_INTERPCASEMOD;
2513 PL_lex_state = LEX_INTERPSTART;
2516 if (s != PL_bufptr) {
2517 PL_nextval[PL_nexttoke] = yylval;
2520 if (PL_lex_starts++) {
2521 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
2522 if (!PL_lex_casemods && PL_lex_inpat)
2535 PL_lex_state = LEX_NORMAL;
2536 s = scan_formline(PL_bufptr);
2537 if (!PL_lex_formbrack)
2543 PL_oldoldbufptr = PL_oldbufptr;
2546 PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at [%s]\n",
2547 exp_name[PL_expect], s);
2553 if (isIDFIRST_lazy_if(s,UTF))
2555 Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255);
2558 goto fake_eof; /* emulate EOF on ^D or ^Z */
2563 if (PL_lex_brackets) {
2564 if (PL_lex_formbrack)
2565 yyerror("Format not terminated");
2567 yyerror("Missing right curly or square bracket");
2569 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2570 "### Tokener got EOF\n");
2574 if (s++ < PL_bufend)
2575 goto retry; /* ignore stray nulls */
2578 if (!PL_in_eval && !PL_preambled) {
2579 PL_preambled = TRUE;
2580 sv_setpv(PL_linestr,incl_perldb());
2581 if (SvCUR(PL_linestr))
2582 sv_catpvn(PL_linestr,";", 1);
2584 while(AvFILLp(PL_preambleav) >= 0) {
2585 SV *tmpsv = av_shift(PL_preambleav);
2586 sv_catsv(PL_linestr, tmpsv);
2587 sv_catpvn(PL_linestr, ";", 1);
2590 sv_free((SV*)PL_preambleav);
2591 PL_preambleav = NULL;
2593 if (PL_minus_n || PL_minus_p) {
2594 sv_catpv(PL_linestr, "LINE: while (<>) {");
2596 sv_catpv(PL_linestr,"chomp;");
2599 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
2600 || *PL_splitstr == '"')
2601 && strchr(PL_splitstr + 1, *PL_splitstr))
2602 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
2604 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
2605 bytes can be used as quoting characters. :-) */
2606 /* The count here deliberately includes the NUL
2607 that terminates the C string constant. This
2608 embeds the opening NUL into the string. */
2609 sv_catpvn(PL_linestr, "our @F=split(q", 15);
2614 sv_catpvn(PL_linestr, s, 1);
2615 sv_catpvn(PL_linestr, s, 1);
2617 /* This loop will embed the trailing NUL of
2618 PL_linestr as the last thing it does before
2620 sv_catpvn(PL_linestr, ");", 2);
2624 sv_catpv(PL_linestr,"our @F=split(' ');");
2627 sv_catpvn(PL_linestr, "\n", 1);
2628 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2629 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2630 PL_last_lop = PL_last_uni = Nullch;
2631 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2632 SV *sv = NEWSV(85,0);
2634 sv_upgrade(sv, SVt_PVMG);
2635 sv_setsv(sv,PL_linestr);
2638 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2643 bof = PL_rsfp ? TRUE : FALSE;
2644 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) {
2647 if (PL_preprocess && !PL_in_eval)
2648 (void)PerlProc_pclose(PL_rsfp);
2649 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
2650 PerlIO_clearerr(PL_rsfp);
2652 (void)PerlIO_close(PL_rsfp);
2654 PL_doextract = FALSE;
2656 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
2657 sv_setpv(PL_linestr,PL_minus_p
2658 ? ";}continue{print;}" : ";}");
2659 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2660 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2661 PL_last_lop = PL_last_uni = Nullch;
2662 PL_minus_n = PL_minus_p = 0;
2665 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2666 PL_last_lop = PL_last_uni = Nullch;
2667 sv_setpv(PL_linestr,"");
2668 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
2670 /* If it looks like the start of a BOM or raw UTF-16,
2671 * check if it in fact is. */
2677 #ifdef PERLIO_IS_STDIO
2678 # ifdef __GNU_LIBRARY__
2679 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
2680 # define FTELL_FOR_PIPE_IS_BROKEN
2684 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
2685 # define FTELL_FOR_PIPE_IS_BROKEN
2690 #ifdef FTELL_FOR_PIPE_IS_BROKEN
2691 /* This loses the possibility to detect the bof
2692 * situation on perl -P when the libc5 is being used.
2693 * Workaround? Maybe attach some extra state to PL_rsfp?
2696 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
2698 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
2701 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2702 s = swallow_bom((U8*)s);
2706 /* Incest with pod. */
2707 if (*s == '=' && strnEQ(s, "=cut", 4)) {
2708 sv_setpv(PL_linestr, "");
2709 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2710 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2711 PL_last_lop = PL_last_uni = Nullch;
2712 PL_doextract = FALSE;
2716 } while (PL_doextract);
2717 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
2718 if (PERLDB_LINE && PL_curstash != PL_debstash) {
2719 SV *sv = NEWSV(85,0);
2721 sv_upgrade(sv, SVt_PVMG);
2722 sv_setsv(sv,PL_linestr);
2725 av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv);
2727 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2728 PL_last_lop = PL_last_uni = Nullch;
2729 if (CopLINE(PL_curcop) == 1) {
2730 while (s < PL_bufend && isSPACE(*s))
2732 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
2736 if (*s == '#' && *(s+1) == '!')
2738 #ifdef ALTERNATE_SHEBANG
2740 static char as[] = ALTERNATE_SHEBANG;
2741 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
2742 d = s + (sizeof(as) - 1);
2744 #endif /* ALTERNATE_SHEBANG */
2753 while (*d && !isSPACE(*d))
2757 #ifdef ARG_ZERO_IS_SCRIPT
2758 if (ipathend > ipath) {
2760 * HP-UX (at least) sets argv[0] to the script name,
2761 * which makes $^X incorrect. And Digital UNIX and Linux,
2762 * at least, set argv[0] to the basename of the Perl
2763 * interpreter. So, having found "#!", we'll set it right.
2765 SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); /* $^X */
2766 assert(SvPOK(x) || SvGMAGICAL(x));
2767 if (sv_eq(x, CopFILESV(PL_curcop))) {
2768 sv_setpvn(x, ipath, ipathend - ipath);
2774 char *bstart = SvPV(CopFILESV(PL_curcop),blen);
2775 char *lstart = SvPV(x,llen);
2777 bstart += blen - llen;
2778 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
2779 sv_setpvn(x, ipath, ipathend - ipath);
2784 TAINT_NOT; /* $^X is always tainted, but that's OK */
2786 #endif /* ARG_ZERO_IS_SCRIPT */
2791 d = instr(s,"perl -");
2793 d = instr(s,"perl");
2795 /* avoid getting into infinite loops when shebang
2796 * line contains "Perl" rather than "perl" */
2798 for (d = ipathend-4; d >= ipath; --d) {
2799 if ((*d == 'p' || *d == 'P')
2800 && !ibcmp(d, "perl", 4))
2810 #ifdef ALTERNATE_SHEBANG
2812 * If the ALTERNATE_SHEBANG on this system starts with a
2813 * character that can be part of a Perl expression, then if
2814 * we see it but not "perl", we're probably looking at the
2815 * start of Perl code, not a request to hand off to some
2816 * other interpreter. Similarly, if "perl" is there, but
2817 * not in the first 'word' of the line, we assume the line
2818 * contains the start of the Perl program.
2820 if (d && *s != '#') {
2822 while (*c && !strchr("; \t\r\n\f\v#", *c))
2825 d = Nullch; /* "perl" not in first word; ignore */
2827 *s = '#'; /* Don't try to parse shebang line */
2829 #endif /* ALTERNATE_SHEBANG */
2830 #ifndef MACOS_TRADITIONAL
2835 !instr(s,"indir") &&
2836 instr(PL_origargv[0],"perl"))
2842 while (s < PL_bufend && isSPACE(*s))
2844 if (s < PL_bufend) {
2845 Newz(899,newargv,PL_origargc+3,char*);
2847 while (s < PL_bufend && !isSPACE(*s))
2850 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
2853 newargv = PL_origargv;
2856 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
2858 Perl_croak(aTHX_ "Can't exec %s", ipath);
2862 U32 oldpdb = PL_perldb;
2863 bool oldn = PL_minus_n;
2864 bool oldp = PL_minus_p;
2866 while (*d && !isSPACE(*d)) d++;
2867 while (SPACE_OR_TAB(*d)) d++;
2870 bool switches_done = PL_doswitches;
2872 if (*d == 'M' || *d == 'm') {
2874 while (*d && !isSPACE(*d)) d++;
2875 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
2878 d = moreswitches(d);
2880 if (PL_doswitches && !switches_done) {
2881 int argc = PL_origargc;
2882 char **argv = PL_origargv;
2885 } while (argc && argv[0][0] == '-' && argv[0][1]);
2886 init_argv_symbols(argc,argv);
2888 if ((PERLDB_LINE && !oldpdb) ||
2889 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
2890 /* if we have already added "LINE: while (<>) {",
2891 we must not do it again */
2893 sv_setpv(PL_linestr, "");
2894 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
2895 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
2896 PL_last_lop = PL_last_uni = Nullch;
2897 PL_preambled = FALSE;
2899 (void)gv_fetchfile(PL_origfilename);
2902 if (PL_doswitches && !switches_done) {
2903 int argc = PL_origargc;
2904 char **argv = PL_origargv;
2907 } while (argc && argv[0][0] == '-' && argv[0][1]);
2908 init_argv_symbols(argc,argv);
2914 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2916 PL_lex_state = LEX_FORMLINE;
2921 #ifdef PERL_STRICT_CR
2922 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
2924 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
2926 case ' ': case '\t': case '\f': case 013:
2927 #ifdef MACOS_TRADITIONAL
2934 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
2935 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
2936 /* handle eval qq[#line 1 "foo"\n ...] */
2937 CopLINE_dec(PL_curcop);
2941 while (s < d && *s != '\n')
2945 else if (s > d) /* Found by Ilya: feed random input to Perl. */
2946 Perl_croak(aTHX_ "panic: input overflow");
2948 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
2950 PL_lex_state = LEX_FORMLINE;
2960 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
2967 while (s < PL_bufend && SPACE_OR_TAB(*s))
2970 if (strnEQ(s,"=>",2)) {
2971 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
2972 DEBUG_T( { PerlIO_printf(Perl_debug_log,
2973 "### Saw unary minus before =>, forcing word '%s'\n", s);
2975 OPERATOR('-'); /* unary minus */
2977 PL_last_uni = PL_oldbufptr;
2979 case 'r': ftst = OP_FTEREAD; break;
2980 case 'w': ftst = OP_FTEWRITE; break;
2981 case 'x': ftst = OP_FTEEXEC; break;
2982 case 'o': ftst = OP_FTEOWNED; break;
2983 case 'R': ftst = OP_FTRREAD; break;
2984 case 'W': ftst = OP_FTRWRITE; break;
2985 case 'X': ftst = OP_FTREXEC; break;
2986 case 'O': ftst = OP_FTROWNED; break;
2987 case 'e': ftst = OP_FTIS; break;
2988 case 'z': ftst = OP_FTZERO; break;
2989 case 's': ftst = OP_FTSIZE; break;
2990 case 'f': ftst = OP_FTFILE; break;
2991 case 'd': ftst = OP_FTDIR; break;
2992 case 'l': ftst = OP_FTLINK; break;
2993 case 'p': ftst = OP_FTPIPE; break;
2994 case 'S': ftst = OP_FTSOCK; break;
2995 case 'u': ftst = OP_FTSUID; break;
2996 case 'g': ftst = OP_FTSGID; break;
2997 case 'k': ftst = OP_FTSVTX; break;
2998 case 'b': ftst = OP_FTBLK; break;
2999 case 'c': ftst = OP_FTCHR; break;
3000 case 't': ftst = OP_FTTTY; break;
3001 case 'T': ftst = OP_FTTEXT; break;
3002 case 'B': ftst = OP_FTBINARY; break;
3003 case 'M': case 'A': case 'C':
3004 gv_fetchpv("\024",TRUE, SVt_PV);
3006 case 'M': ftst = OP_FTMTIME; break;
3007 case 'A': ftst = OP_FTATIME; break;
3008 case 'C': ftst = OP_FTCTIME; break;
3016 PL_last_lop_op = (OPCODE)ftst;
3017 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3018 "### Saw file test %c\n", (int)ftst);
3023 /* Assume it was a minus followed by a one-letter named
3024 * subroutine call (or a -bareword), then. */
3025 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3026 "### '-%c' looked like a file test but was not\n",
3035 if (PL_expect == XOPERATOR)
3040 else if (*s == '>') {
3043 if (isIDFIRST_lazy_if(s,UTF)) {
3044 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
3052 if (PL_expect == XOPERATOR)
3055 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3057 OPERATOR('-'); /* unary minus */
3064 if (PL_expect == XOPERATOR)
3069 if (PL_expect == XOPERATOR)
3072 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
3078 if (PL_expect != XOPERATOR) {
3079 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3080 PL_expect = XOPERATOR;
3081 force_ident(PL_tokenbuf, '*');
3094 if (PL_expect == XOPERATOR) {
3098 PL_tokenbuf[0] = '%';
3099 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE);
3100 if (!PL_tokenbuf[1]) {
3103 PL_pending_ident = '%';
3122 switch (PL_expect) {
3125 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
3127 PL_bufptr = s; /* update in case we back off */
3133 PL_expect = XTERMBLOCK;
3137 while (isIDFIRST_lazy_if(s,UTF)) {
3138 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3139 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) {
3140 if (tmp < 0) tmp = -tmp;
3156 d = scan_str(d,TRUE,TRUE);
3158 /* MUST advance bufptr here to avoid bogus
3159 "at end of line" context messages from yyerror().
3161 PL_bufptr = s + len;
3162 yyerror("Unterminated attribute parameter in attribute list");
3165 return REPORT(0); /* EOF indicator */
3169 SV *sv = newSVpvn(s, len);
3170 sv_catsv(sv, PL_lex_stuff);
3171 attrs = append_elem(OP_LIST, attrs,
3172 newSVOP(OP_CONST, 0, sv));
3173 SvREFCNT_dec(PL_lex_stuff);
3174 PL_lex_stuff = Nullsv;
3177 if (len == 6 && strnEQ(s, "unique", len)) {
3178 if (PL_in_my == KEY_our)
3180 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
3182 ; /* skip to avoid loading attributes.pm */
3185 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
3188 /* NOTE: any CV attrs applied here need to be part of
3189 the CVf_BUILTIN_ATTRS define in cv.h! */
3190 else if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len))
3191 CvLVALUE_on(PL_compcv);
3192 else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len))
3193 CvLOCKED_on(PL_compcv);
3194 else if (!PL_in_my && len == 6 && strnEQ(s, "method", len))
3195 CvMETHOD_on(PL_compcv);
3196 else if (!PL_in_my && len == 9 && strnEQ(s, "assertion", len))
3197 CvASSERTION_on(PL_compcv);
3198 /* After we've set the flags, it could be argued that
3199 we don't need to do the attributes.pm-based setting
3200 process, and shouldn't bother appending recognized
3201 flags. To experiment with that, uncomment the
3202 following "else". (Note that's already been
3203 uncommented. That keeps the above-applied built-in
3204 attributes from being intercepted (and possibly
3205 rejected) by a package's attribute routines, but is
3206 justified by the performance win for the common case
3207 of applying only built-in attributes.) */
3209 attrs = append_elem(OP_LIST, attrs,
3210 newSVOP(OP_CONST, 0,
3214 if (*s == ':' && s[1] != ':')
3217 break; /* require real whitespace or :'s */
3219 tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
3220 if (*s != ';' && *s != '}' && *s != tmp && (tmp != '=' || *s != ')')) {
3221 char q = ((*s == '\'') ? '"' : '\'');
3222 /* If here for an expression, and parsed no attrs, back off. */
3223 if (tmp == '=' && !attrs) {
3227 /* MUST advance bufptr here to avoid bogus "at end of line"
3228 context messages from yyerror().
3232 yyerror("Unterminated attribute list");
3234 yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list",
3242 PL_nextval[PL_nexttoke].opval = attrs;
3250 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
3251 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
3268 if (PL_lex_brackets <= 0)
3269 yyerror("Unmatched right square bracket");
3272 if (PL_lex_state == LEX_INTERPNORMAL) {
3273 if (PL_lex_brackets == 0) {
3274 if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>'))
3275 PL_lex_state = LEX_INTERPEND;
3282 if (PL_lex_brackets > 100) {
3283 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
3285 switch (PL_expect) {
3287 if (PL_lex_formbrack) {
3291 if (PL_oldoldbufptr == PL_last_lop)
3292 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3294 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3295 OPERATOR(HASHBRACK);
3297 while (s < PL_bufend && SPACE_OR_TAB(*s))
3300 PL_tokenbuf[0] = '\0';
3301 if (d < PL_bufend && *d == '-') {
3302 PL_tokenbuf[0] = '-';
3304 while (d < PL_bufend && SPACE_OR_TAB(*d))
3307 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
3308 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
3310 while (d < PL_bufend && SPACE_OR_TAB(*d))
3313 char minus = (PL_tokenbuf[0] == '-');
3314 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
3322 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
3327 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3332 if (PL_oldoldbufptr == PL_last_lop)
3333 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
3335 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
3338 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
3340 /* This hack is to get the ${} in the message. */
3342 yyerror("syntax error");
3345 OPERATOR(HASHBRACK);
3347 /* This hack serves to disambiguate a pair of curlies
3348 * as being a block or an anon hash. Normally, expectation
3349 * determines that, but in cases where we're not in a
3350 * position to expect anything in particular (like inside
3351 * eval"") we have to resolve the ambiguity. This code
3352 * covers the case where the first term in the curlies is a
3353 * quoted string. Most other cases need to be explicitly
3354 * disambiguated by prepending a `+' before the opening
3355 * curly in order to force resolution as an anon hash.
3357 * XXX should probably propagate the outer expectation
3358 * into eval"" to rely less on this hack, but that could
3359 * potentially break current behavior of eval"".
3363 if (*s == '\'' || *s == '"' || *s == '`') {
3364 /* common case: get past first string, handling escapes */
3365 for (t++; t < PL_bufend && *t != *s;)
3366 if (*t++ == '\\' && (*t == '\\' || *t == *s))
3370 else if (*s == 'q') {
3373 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
3376 /* skip q//-like construct */
3378 char open, close, term;
3381 while (t < PL_bufend && isSPACE(*t))
3383 /* check for q => */
3384 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
3385 OPERATOR(HASHBRACK);
3389 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
3393 for (t++; t < PL_bufend; t++) {
3394 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
3396 else if (*t == open)
3400 for (t++; t < PL_bufend; t++) {
3401 if (*t == '\\' && t+1 < PL_bufend)
3403 else if (*t == close && --brackets <= 0)
3405 else if (*t == open)
3412 /* skip plain q word */
3413 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3416 else if (isALNUM_lazy_if(t,UTF)) {
3418 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
3421 while (t < PL_bufend && isSPACE(*t))
3423 /* if comma follows first term, call it an anon hash */
3424 /* XXX it could be a comma expression with loop modifiers */
3425 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
3426 || (*t == '=' && t[1] == '>')))
3427 OPERATOR(HASHBRACK);
3428 if (PL_expect == XREF)
3431 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
3437 yylval.ival = CopLINE(PL_curcop);
3438 if (isSPACE(*s) || *s == '#')
3439 PL_copline = NOLINE; /* invalidate current command line number */
3444 if (PL_lex_brackets <= 0)
3445 yyerror("Unmatched right curly bracket");
3447 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
3448 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
3449 PL_lex_formbrack = 0;
3450 if (PL_lex_state == LEX_INTERPNORMAL) {
3451 if (PL_lex_brackets == 0) {
3452 if (PL_expect & XFAKEBRACK) {
3453 PL_expect &= XENUMMASK;
3454 PL_lex_state = LEX_INTERPEND;
3456 return yylex(); /* ignore fake brackets */
3458 if (*s == '-' && s[1] == '>')
3459 PL_lex_state = LEX_INTERPENDMAYBE;
3460 else if (*s != '[' && *s != '{')
3461 PL_lex_state = LEX_INTERPEND;
3464 if (PL_expect & XFAKEBRACK) {
3465 PL_expect &= XENUMMASK;
3467 return yylex(); /* ignore fake brackets */
3477 if (PL_expect == XOPERATOR) {
3478 if (ckWARN(WARN_SEMICOLON)
3479 && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart)
3481 CopLINE_dec(PL_curcop);
3482 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
3483 CopLINE_inc(PL_curcop);
3488 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
3490 PL_expect = XOPERATOR;
3491 force_ident(PL_tokenbuf, '&');
3495 yylval.ival = (OPpENTERSUB_AMPER<<8);
3514 if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp))
3515 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Reversed %c= operator",(int)tmp);
3517 if (PL_expect == XSTATE && isALPHA(tmp) &&
3518 (s == PL_linestart+1 || s[-2] == '\n') )
3520 if (PL_in_eval && !PL_rsfp) {
3525 if (strnEQ(s,"=cut",4)) {
3539 PL_doextract = TRUE;
3542 if (PL_lex_brackets < PL_lex_formbrack) {
3544 #ifdef PERL_STRICT_CR
3545 for (t = s; SPACE_OR_TAB(*t); t++) ;
3547 for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
3549 if (*t == '\n' || *t == '#') {
3561 /* was this !=~ where !~ was meant?
3562 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
3564 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
3567 while (t < PL_bufend && isSPACE(*t))
3570 if (*t == '/' || *t == '?' ||
3571 ((*t == 'm' || *t == 's' || *t == 'y') && !isALNUM(t[1])) ||
3572 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
3573 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3574 "!=~ should be !~");
3583 if (PL_expect != XOPERATOR) {
3584 if (s[1] != '<' && !strchr(s,'>'))
3587 s = scan_heredoc(s);
3589 s = scan_inputsymbol(s);
3590 TERM(sublex_start());
3595 SHop(OP_LEFT_SHIFT);
3609 SHop(OP_RIGHT_SHIFT);
3618 if (PL_expect == XOPERATOR) {
3619 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3622 return REPORT(','); /* grandfather non-comma-format format */
3626 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
3627 PL_tokenbuf[0] = '@';
3628 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
3629 sizeof PL_tokenbuf - 1, FALSE);
3630 if (PL_expect == XOPERATOR)
3631 no_op("Array length", s);
3632 if (!PL_tokenbuf[1])
3634 PL_expect = XOPERATOR;
3635 PL_pending_ident = '#';
3639 PL_tokenbuf[0] = '$';
3640 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
3641 sizeof PL_tokenbuf - 1, FALSE);
3642 if (PL_expect == XOPERATOR)
3644 if (!PL_tokenbuf[1]) {
3646 yyerror("Final $ should be \\$ or $name");
3650 /* This kludge not intended to be bulletproof. */
3651 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
3652 yylval.opval = newSVOP(OP_CONST, 0,
3653 newSViv(PL_compiling.cop_arybase));
3654 yylval.opval->op_private = OPpCONST_ARYBASE;
3660 if (PL_lex_state == LEX_NORMAL)
3663 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3666 PL_tokenbuf[0] = '@';
3667 if (ckWARN(WARN_SYNTAX)) {
3669 isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$';
3672 PL_bufptr = skipspace(PL_bufptr);
3673 while (t < PL_bufend && *t != ']')
3675 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3676 "Multidimensional syntax %.*s not supported",
3677 (t - PL_bufptr) + 1, PL_bufptr);
3681 else if (*s == '{') {
3682 PL_tokenbuf[0] = '%';
3683 if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") &&
3684 (t = strchr(s, '}')) && (t = strchr(t, '=')))
3686 char tmpbuf[sizeof PL_tokenbuf];
3688 for (t++; isSPACE(*t); t++) ;
3689 if (isIDFIRST_lazy_if(t,UTF)) {
3690 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len);
3691 for (; isSPACE(*t); t++) ;
3692 if (*t == ';' && get_cv(tmpbuf, FALSE))
3693 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3694 "You need to quote \"%s\"", tmpbuf);
3700 PL_expect = XOPERATOR;
3701 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
3702 bool islop = (PL_last_lop == PL_oldoldbufptr);
3703 if (!islop || PL_last_lop_op == OP_GREPSTART)
3704 PL_expect = XOPERATOR;
3705 else if (strchr("$@\"'`q", *s))
3706 PL_expect = XTERM; /* e.g. print $fh "foo" */
3707 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
3708 PL_expect = XTERM; /* e.g. print $fh &sub */
3709 else if (isIDFIRST_lazy_if(s,UTF)) {
3710 char tmpbuf[sizeof PL_tokenbuf];
3711 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3712 if ((tmp = keyword(tmpbuf, len))) {
3713 /* binary operators exclude handle interpretations */
3725 PL_expect = XTERM; /* e.g. print $fh length() */
3730 PL_expect = XTERM; /* e.g. print $fh subr() */
3733 else if (isDIGIT(*s))
3734 PL_expect = XTERM; /* e.g. print $fh 3 */
3735 else if (*s == '.' && isDIGIT(s[1]))
3736 PL_expect = XTERM; /* e.g. print $fh .3 */
3737 else if ((*s == '?' || *s == '-' || *s == '+')
3738 && !isSPACE(s[1]) && s[1] != '=')
3739 PL_expect = XTERM; /* e.g. print $fh -1 */
3740 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '=' && s[1] != '/')
3741 PL_expect = XTERM; /* e.g. print $fh /.../
3742 XXX except DORDOR operator */
3743 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=')
3744 PL_expect = XTERM; /* print $fh <<"EOF" */
3746 PL_pending_ident = '$';
3750 if (PL_expect == XOPERATOR)
3752 PL_tokenbuf[0] = '@';
3753 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
3754 if (!PL_tokenbuf[1]) {
3757 if (PL_lex_state == LEX_NORMAL)
3759 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
3761 PL_tokenbuf[0] = '%';
3763 /* Warn about @ where they meant $. */
3764 if (ckWARN(WARN_SYNTAX)) {
3765 if (*s == '[' || *s == '{') {
3767 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
3769 if (*t == '}' || *t == ']') {
3771 PL_bufptr = skipspace(PL_bufptr);
3772 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
3773 "Scalar value %.*s better written as $%.*s",
3774 t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1);
3779 PL_pending_ident = '@';
3782 case '/': /* may be division, defined-or, or pattern */
3783 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
3787 case '?': /* may either be conditional or pattern */
3788 if(PL_expect == XOPERATOR) {
3796 /* A // operator. */
3806 /* Disable warning on "study /blah/" */
3807 if (PL_oldoldbufptr == PL_last_uni
3808 && (*PL_last_uni != 's' || s - PL_last_uni < 5
3809 || memNE(PL_last_uni, "study", 5)
3810 || isALNUM_lazy_if(PL_last_uni+5,UTF)
3813 s = scan_pat(s,OP_MATCH);
3814 TERM(sublex_start());
3818 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
3819 #ifdef PERL_STRICT_CR
3822 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
3824 && (s == PL_linestart || s[-1] == '\n') )
3826 PL_lex_formbrack = 0;
3830 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
3836 yylval.ival = OPf_SPECIAL;
3842 if (PL_expect != XOPERATOR)
3847 case '0': case '1': case '2': case '3': case '4':
3848 case '5': case '6': case '7': case '8': case '9':
3849 s = scan_num(s, &yylval);
3850 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3851 "### Saw number in '%s'\n", s);
3853 if (PL_expect == XOPERATOR)
3858 s = scan_str(s,FALSE,FALSE);
3859 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3860 "### Saw string before '%s'\n", s);
3862 if (PL_expect == XOPERATOR) {
3863 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3866 return REPORT(','); /* grandfather non-comma-format format */
3872 missingterm((char*)0);
3873 yylval.ival = OP_CONST;
3874 TERM(sublex_start());
3877 s = scan_str(s,FALSE,FALSE);
3878 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3879 "### Saw string before '%s'\n", s);
3881 if (PL_expect == XOPERATOR) {
3882 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
3885 return REPORT(','); /* grandfather non-comma-format format */
3891 missingterm((char*)0);
3892 yylval.ival = OP_CONST;
3893 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
3894 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
3895 yylval.ival = OP_STRINGIFY;
3899 TERM(sublex_start());
3902 s = scan_str(s,FALSE,FALSE);
3903 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3904 "### Saw backtick string before '%s'\n", s);
3906 if (PL_expect == XOPERATOR)
3907 no_op("Backticks",s);
3909 missingterm((char*)0);
3910 yylval.ival = OP_BACKTICK;
3912 TERM(sublex_start());
3916 if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s))
3917 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
3919 if (PL_expect == XOPERATOR)
3920 no_op("Backslash",s);
3924 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
3928 while (isDIGIT(*start) || *start == '_')
3930 if (*start == '.' && isDIGIT(start[1])) {
3931 s = scan_num(s, &yylval);
3934 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
3935 else if (!isALPHA(*start) && (PL_expect == XTERM
3936 || PL_expect == XREF || PL_expect == XSTATE
3937 || PL_expect == XTERMORDORDOR)) {
3941 gv = gv_fetchpv(s, FALSE, SVt_PVCV);
3944 s = scan_num(s, &yylval);
3951 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
3991 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
3993 /* Some keywords can be followed by any delimiter, including ':' */
3994 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
3995 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
3996 (PL_tokenbuf[0] == 'q' &&
3997 strchr("qwxr", PL_tokenbuf[1])))));
3999 /* x::* is just a word, unless x is "CORE" */
4000 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
4004 while (d < PL_bufend && isSPACE(*d))
4005 d++; /* no comments skipped here, or s### is misparsed */
4007 /* Is this a label? */
4008 if (!tmp && PL_expect == XSTATE
4009 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
4011 yylval.pval = savepv(PL_tokenbuf);
4016 /* Check for keywords */
4017 tmp = keyword(PL_tokenbuf, len);
4019 /* Is this a word before a => operator? */
4020 if (*d == '=' && d[1] == '>') {
4023 = (OP*)newSVOP(OP_CONST, 0,
4024 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
4025 yylval.opval->op_private = OPpCONST_BARE;
4029 if (tmp < 0) { /* second-class keyword? */
4030 GV *ogv = Nullgv; /* override (winner) */
4031 GV *hgv = Nullgv; /* hidden (loser) */
4032 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
4034 if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) &&
4037 if (GvIMPORTED_CV(gv))
4039 else if (! CvMETHOD(cv))
4043 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
4044 (gv = *gvp) != (GV*)&PL_sv_undef &&
4045 GvCVu(gv) && GvIMPORTED_CV(gv))
4052 tmp = 0; /* overridden by import or by GLOBAL */
4055 && -tmp==KEY_lock /* XXX generalizable kludge */
4057 && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE))
4059 tmp = 0; /* any sub overrides "weak" keyword */
4064 && PL_expect != XOPERATOR
4065 && PL_expect != XTERMORDORDOR)
4067 /* any sub overrides the "err" keyword, except when really an
4068 * operator is expected */
4071 else { /* no override */
4073 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
4074 Perl_warner(aTHX_ packWARN(WARN_MISC),
4075 "dump() better written as CORE::dump()");
4079 if (ckWARN(WARN_AMBIGUOUS) && hgv
4080 && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
4081 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4082 "Ambiguous call resolved as CORE::%s(), %s",
4083 GvENAME(hgv), "qualify as such or use &");
4090 default: /* not a keyword */
4094 char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
4096 /* Get the rest if it looks like a package qualifier */
4098 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
4100 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
4103 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
4104 *s == '\'' ? "'" : "::");
4109 if (PL_expect == XOPERATOR) {
4110 if (PL_bufptr == PL_linestart) {
4111 CopLINE_dec(PL_curcop);
4112 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4113 CopLINE_inc(PL_curcop);
4116 no_op("Bareword",s);
4119 /* Look for a subroutine with this name in current package,
4120 unless name is "Foo::", in which case Foo is a bearword
4121 (and a package name). */
4124 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
4126 if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV))
4127 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
4128 "Bareword \"%s\" refers to nonexistent package",
4131 PL_tokenbuf[len] = '\0';
4138 gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV);
4141 /* if we saw a global override before, get the right name */
4144 sv = newSVpvn("CORE::GLOBAL::",14);
4145 sv_catpv(sv,PL_tokenbuf);
4148 /* If len is 0, newSVpv does strlen(), which is correct.
4149 If len is non-zero, then it will be the true length,
4150 and so the scalar will be created correctly. */
4151 sv = newSVpv(PL_tokenbuf,len);
4154 /* Presume this is going to be a bareword of some sort. */
4157 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4158 yylval.opval->op_private = OPpCONST_BARE;
4159 /* UTF-8 package name? */
4160 if (UTF && !IN_BYTES &&
4161 is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
4164 /* And if "Foo::", then that's what it certainly is. */
4169 /* See if it's the indirect object for a list operator. */
4171 if (PL_oldoldbufptr &&
4172 PL_oldoldbufptr < PL_bufptr &&
4173 (PL_oldoldbufptr == PL_last_lop
4174 || PL_oldoldbufptr == PL_last_uni) &&
4175 /* NO SKIPSPACE BEFORE HERE! */
4176 (PL_expect == XREF ||
4177 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
4179 bool immediate_paren = *s == '(';
4181 /* (Now we can afford to cross potential line boundary.) */
4184 /* Two barewords in a row may indicate method call. */
4186 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv)))
4189 /* If not a declared subroutine, it's an indirect object. */
4190 /* (But it's an indir obj regardless for sort.) */
4192 if ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
4193 ((!gv || !GvCVu(gv)) &&
4194 (PL_last_lop_op != OP_MAPSTART &&
4195 PL_last_lop_op != OP_GREPSTART))))
4197 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
4202 PL_expect = XOPERATOR;
4205 /* Is this a word before a => operator? */
4206 if (*s == '=' && s[1] == '>' && !pkgname) {
4208 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
4209 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
4210 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
4214 /* If followed by a paren, it's certainly a subroutine. */
4217 if (gv && GvCVu(gv)) {
4218 for (d = s + 1; SPACE_OR_TAB(*d); d++) ;
4219 if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) {
4224 PL_nextval[PL_nexttoke].opval = yylval.opval;
4225 PL_expect = XOPERATOR;
4231 /* If followed by var or block, call it a method (unless sub) */
4233 if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) {
4234 PL_last_lop = PL_oldbufptr;
4235 PL_last_lop_op = OP_METHOD;
4239 /* If followed by a bareword, see if it looks like indir obj. */
4242 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
4243 && (tmp = intuit_method(s,gv)))
4246 /* Not a method, so call it a subroutine (if defined) */
4248 if (gv && GvCVu(gv)) {
4250 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
4251 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4252 "Ambiguous use of -%s resolved as -&%s()",
4253 PL_tokenbuf, PL_tokenbuf);
4254 /* Check for a constant sub */
4256 if ((sv = cv_const_sv(cv))) {
4258 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
4259 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv);
4260 yylval.opval->op_private = 0;
4264 /* Resolve to GV now. */
4265 op_free(yylval.opval);
4266 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
4267 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
4268 PL_last_lop = PL_oldbufptr;
4269 PL_last_lop_op = OP_ENTERSUB;
4270 /* Is there a prototype? */
4273 char *proto = SvPV((SV*)cv, len);
4276 if (*proto == '$' && proto[1] == '\0')
4278 while (*proto == ';')
4280 if (*proto == '&' && *s == '{') {
4281 sv_setpv(PL_subname, PL_curstash ?
4282 "__ANON__" : "__ANON__::__ANON__");
4286 PL_nextval[PL_nexttoke].opval = yylval.opval;
4292 /* Call it a bare word */
4294 if (PL_hints & HINT_STRICT_SUBS)
4295 yylval.opval->op_private |= OPpCONST_STRICT;
4298 if (ckWARN(WARN_RESERVED)) {
4299 if (lastchar != '-') {
4300 for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ;
4301 if (!*d && !gv_stashpv(PL_tokenbuf,FALSE))
4302 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
4309 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
4310 && ckWARN_d(WARN_AMBIGUOUS)) {
4311 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4312 "Operator or semicolon missing before %c%s",
4313 lastchar, PL_tokenbuf);
4314 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
4315 "Ambiguous use of %c resolved as operator %c",
4316 lastchar, lastchar);
4322 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4323 newSVpv(CopFILE(PL_curcop),0));
4327 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4328 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
4331 case KEY___PACKAGE__:
4332 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
4334 ? newSVpv(HvNAME(PL_curstash), 0)
4343 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
4344 char *pname = "main";
4345 if (PL_tokenbuf[2] == 'D')
4346 pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash);
4347 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO);
4350 GvIOp(gv) = newIO();
4351 IoIFP(GvIOp(gv)) = PL_rsfp;
4352 #if defined(HAS_FCNTL) && defined(F_SETFD)
4354 int fd = PerlIO_fileno(PL_rsfp);
4355 fcntl(fd,F_SETFD,fd >= 3);
4358 /* Mark this internal pseudo-handle as clean */
4359 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
4361 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
4362 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
4363 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
4365 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
4366 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
4367 /* if the script was opened in binmode, we need to revert
4368 * it to text mode for compatibility; but only iff it has CRs
4369 * XXX this is a questionable hack at best. */
4370 if (PL_bufend-PL_bufptr > 2
4371 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
4374 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
4375 loc = PerlIO_tell(PL_rsfp);
4376 (void)PerlIO_seek(PL_rsfp, 0L, 0);
4379 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
4381 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
4382 #endif /* NETWARE */
4383 #ifdef PERLIO_IS_STDIO /* really? */
4384 # if defined(__BORLANDC__)
4385 /* XXX see note in do_binmode() */
4386 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
4390 PerlIO_seek(PL_rsfp, loc, 0);
4394 #ifdef PERLIO_LAYERS
4397 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
4398 else if (PL_encoding) {
4405 XPUSHs(PL_encoding);
4407 call_method("name", G_SCALAR);
4411 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
4412 Perl_form(aTHX_ ":encoding(%"SVf")",
4430 if (PL_expect == XSTATE) {
4437 if (*s == ':' && s[1] == ':') {
4440 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4441 if (!(tmp = keyword(PL_tokenbuf, len)))
4442 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
4456 LOP(OP_ACCEPT,XTERM);
4462 LOP(OP_ATAN2,XTERM);
4468 LOP(OP_BINMODE,XTERM);
4471 LOP(OP_BLESS,XTERM);
4480 (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */
4497 if (!PL_cryptseen) {
4498 PL_cryptseen = TRUE;
4502 LOP(OP_CRYPT,XTERM);
4505 LOP(OP_CHMOD,XTERM);
4508 LOP(OP_CHOWN,XTERM);
4511 LOP(OP_CONNECT,XTERM);
4527 s = force_word(s,WORD,TRUE,TRUE,FALSE);
4531 PL_hints |= HINT_BLOCK_SCOPE;
4541 gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
4542 LOP(OP_DBMOPEN,XTERM);
4548 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4555 yylval.ival = CopLINE(PL_curcop);
4569 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
4570 UNIBRACK(OP_ENTEREVAL);
4588 case KEY_endhostent:
4594 case KEY_endservent:
4597 case KEY_endprotoent:
4608 yylval.ival = CopLINE(PL_curcop);
4610 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
4612 if ((PL_bufend - p) >= 3 &&
4613 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
4615 else if ((PL_bufend - p) >= 4 &&
4616 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
4619 if (isIDFIRST_lazy_if(p,UTF)) {
4620 p = scan_ident(p, PL_bufend,
4621 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4625 Perl_croak(aTHX_ "Missing $ on loop variable");
4630 LOP(OP_FORMLINE,XTERM);
4636 LOP(OP_FCNTL,XTERM);
4642 LOP(OP_FLOCK,XTERM);
4651 LOP(OP_GREPSTART, XREF);
4654 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4669 case KEY_getpriority:
4670 LOP(OP_GETPRIORITY,XTERM);
4672 case KEY_getprotobyname:
4675 case KEY_getprotobynumber:
4676 LOP(OP_GPBYNUMBER,XTERM);
4678 case KEY_getprotoent:
4690 case KEY_getpeername:
4691 UNI(OP_GETPEERNAME);
4693 case KEY_gethostbyname:
4696 case KEY_gethostbyaddr:
4697 LOP(OP_GHBYADDR,XTERM);
4699 case KEY_gethostent:
4702 case KEY_getnetbyname:
4705 case KEY_getnetbyaddr:
4706 LOP(OP_GNBYADDR,XTERM);
4711 case KEY_getservbyname:
4712 LOP(OP_GSBYNAME,XTERM);
4714 case KEY_getservbyport:
4715 LOP(OP_GSBYPORT,XTERM);
4717 case KEY_getservent:
4720 case KEY_getsockname:
4721 UNI(OP_GETSOCKNAME);
4723 case KEY_getsockopt:
4724 LOP(OP_GSOCKOPT,XTERM);
4746 yylval.ival = CopLINE(PL_curcop);
4750 LOP(OP_INDEX,XTERM);
4756 LOP(OP_IOCTL,XTERM);
4768 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4800 LOP(OP_LISTEN,XTERM);
4809 s = scan_pat(s,OP_MATCH);
4810 TERM(sublex_start());
4813 LOP(OP_MAPSTART, XREF);
4816 LOP(OP_MKDIR,XTERM);
4819 LOP(OP_MSGCTL,XTERM);
4822 LOP(OP_MSGGET,XTERM);
4825 LOP(OP_MSGRCV,XTERM);
4828 LOP(OP_MSGSND,XTERM);
4834 if (isIDFIRST_lazy_if(s,UTF)) {
4835 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
4836 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
4838 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
4839 if (!PL_in_my_stash) {
4842 sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf);
4850 s = force_word(s,WORD,TRUE,FALSE,FALSE);
4857 if (PL_expect != XSTATE)
4858 yyerror("\"no\" not allowed in expression");
4859 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4860 s = force_version(s, FALSE);
4865 if (*s == '(' || (s = skipspace(s), *s == '('))
4872 if (isIDFIRST_lazy_if(s,UTF)) {
4874 for (d = s; isALNUM_lazy_if(d,UTF); d++) ;
4875 for (t=d; *t && isSPACE(*t); t++) ;
4876 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
4878 && !(t[0] == '=' && t[1] == '>')
4880 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
4881 "Precedence problem: open %.*s should be open(%.*s)",
4882 d - s, s, d - s, s);
4888 yylval.ival = OP_OR;
4898 LOP(OP_OPEN_DIR,XTERM);
4901 checkcomma(s,PL_tokenbuf,"filehandle");
4905 checkcomma(s,PL_tokenbuf,"filehandle");
4924 s = force_word(s,WORD,FALSE,TRUE,FALSE);
4928 LOP(OP_PIPE_OP,XTERM);
4931 s = scan_str(s,FALSE,FALSE);
4933 missingterm((char*)0);
4934 yylval.ival = OP_CONST;
4935 TERM(sublex_start());
4941 s = scan_str(s,FALSE,FALSE);
4943 missingterm((char*)0);
4945 if (SvCUR(PL_lex_stuff)) {
4948 d = SvPV_force(PL_lex_stuff, len);
4951 for (; isSPACE(*d) && len; --len, ++d) ;
4954 if (!warned && ckWARN(WARN_QW)) {
4955 for (; !isSPACE(*d) && len; --len, ++d) {
4957 Perl_warner(aTHX_ packWARN(WARN_QW),
4958 "Possible attempt to separate words with commas");
4961 else if (*d == '#') {
4962 Perl_warner(aTHX_ packWARN(WARN_QW),
4963 "Possible attempt to put comments in qw() list");
4969 for (; !isSPACE(*d) && len; --len, ++d) ;
4971 sv = newSVpvn(b, d-b);
4972 if (DO_UTF8(PL_lex_stuff))
4974 words = append_elem(OP_LIST, words,
4975 newSVOP(OP_CONST, 0, tokeq(sv)));
4979 PL_nextval[PL_nexttoke].opval = words;
4984 SvREFCNT_dec(PL_lex_stuff);
4985 PL_lex_stuff = Nullsv;
4991 s = scan_str(s,FALSE,FALSE);
4993 missingterm((char*)0);
4994 yylval.ival = OP_STRINGIFY;
4995 if (SvIVX(PL_lex_stuff) == '\'')
4996 SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */
4997 TERM(sublex_start());
5000 s = scan_pat(s,OP_QR);
5001 TERM(sublex_start());
5004 s = scan_str(s,FALSE,FALSE);
5006 missingterm((char*)0);
5007 yylval.ival = OP_BACKTICK;
5009 TERM(sublex_start());
5017 s = force_version(s, FALSE);
5019 else if (*s != 'v' || !isDIGIT(s[1])
5020 || (s = force_version(s, TRUE), *s == 'v'))
5022 *PL_tokenbuf = '\0';
5023 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5024 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
5025 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE);
5027 yyerror("<> should be quotes");
5035 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5039 LOP(OP_RENAME,XTERM);
5048 LOP(OP_RINDEX,XTERM);
5058 UNIDOR(OP_READLINE);
5071 LOP(OP_REVERSE,XTERM);
5074 UNIDOR(OP_READLINK);
5082 TERM(sublex_start());
5084 TOKEN(1); /* force error */
5093 LOP(OP_SELECT,XTERM);
5099 LOP(OP_SEMCTL,XTERM);
5102 LOP(OP_SEMGET,XTERM);
5105 LOP(OP_SEMOP,XTERM);
5111 LOP(OP_SETPGRP,XTERM);
5113 case KEY_setpriority:
5114 LOP(OP_SETPRIORITY,XTERM);
5116 case KEY_sethostent:
5122 case KEY_setservent:
5125 case KEY_setprotoent:
5135 LOP(OP_SEEKDIR,XTERM);
5137 case KEY_setsockopt:
5138 LOP(OP_SSOCKOPT,XTERM);
5144 LOP(OP_SHMCTL,XTERM);
5147 LOP(OP_SHMGET,XTERM);
5150 LOP(OP_SHMREAD,XTERM);
5153 LOP(OP_SHMWRITE,XTERM);
5156 LOP(OP_SHUTDOWN,XTERM);
5165 LOP(OP_SOCKET,XTERM);
5167 case KEY_socketpair:
5168 LOP(OP_SOCKPAIR,XTERM);
5171 checkcomma(s,PL_tokenbuf,"subroutine name");
5173 if (*s == ';' || *s == ')') /* probably a close */
5174 Perl_croak(aTHX_ "sort is now a reserved word");
5176 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5180 LOP(OP_SPLIT,XTERM);
5183 LOP(OP_SPRINTF,XTERM);
5186 LOP(OP_SPLICE,XTERM);
5201 LOP(OP_SUBSTR,XTERM);
5207 char tmpbuf[sizeof PL_tokenbuf];
5208 SSize_t tboffset = 0;
5209 expectation attrful;
5210 bool have_name, have_proto, bad_proto;
5215 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
5216 (*s == ':' && s[1] == ':'))
5219 attrful = XATTRBLOCK;
5220 /* remember buffer pos'n for later force_word */
5221 tboffset = s - PL_oldbufptr;
5222 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5223 if (strchr(tmpbuf, ':'))
5224 sv_setpv(PL_subname, tmpbuf);
5226 sv_setsv(PL_subname,PL_curstname);
5227 sv_catpvn(PL_subname,"::",2);
5228 sv_catpvn(PL_subname,tmpbuf,len);
5235 Perl_croak(aTHX_ "Missing name in \"my sub\"");
5236 PL_expect = XTERMBLOCK;
5237 attrful = XATTRTERM;
5238 sv_setpv(PL_subname,"?");
5242 if (key == KEY_format) {
5244 PL_lex_formbrack = PL_lex_brackets + 1;
5246 (void) force_word(PL_oldbufptr + tboffset, WORD,
5251 /* Look for a prototype */
5255 s = scan_str(s,FALSE,FALSE);
5257 Perl_croak(aTHX_ "Prototype not terminated");
5258 /* strip spaces and check for bad characters */
5259 d = SvPVX(PL_lex_stuff);
5262 for (p = d; *p; ++p) {
5265 if (!strchr("$@%*;[]&\\", *p))
5270 if (bad_proto && ckWARN(WARN_SYNTAX))
5271 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5272 "Illegal character in prototype for %"SVf" : %s",
5274 SvCUR(PL_lex_stuff) = tmp;
5282 if (*s == ':' && s[1] != ':')
5283 PL_expect = attrful;
5284 else if (*s != '{' && key == KEY_sub) {
5286 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
5288 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, PL_subname);
5292 PL_nextval[PL_nexttoke].opval =
5293 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
5294 PL_lex_stuff = Nullsv;
5298 sv_setpv(PL_subname,
5299 PL_curstash ? "__ANON__" : "__ANON__::__ANON__");
5302 (void) force_word(PL_oldbufptr + tboffset, WORD,
5311 LOP(OP_SYSTEM,XREF);
5314 LOP(OP_SYMLINK,XTERM);
5317 LOP(OP_SYSCALL,XTERM);
5320 LOP(OP_SYSOPEN,XTERM);
5323 LOP(OP_SYSSEEK,XTERM);
5326 LOP(OP_SYSREAD,XTERM);
5329 LOP(OP_SYSWRITE,XTERM);
5333 TERM(sublex_start());
5354 LOP(OP_TRUNCATE,XTERM);
5366 yylval.ival = CopLINE(PL_curcop);
5370 yylval.ival = CopLINE(PL_curcop);
5374 LOP(OP_UNLINK,XTERM);
5380 LOP(OP_UNPACK,XTERM);
5383 LOP(OP_UTIME,XTERM);
5389 LOP(OP_UNSHIFT,XTERM);
5392 if (PL_expect != XSTATE)
5393 yyerror("\"use\" not allowed in expression");
5395 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
5396 s = force_version(s, TRUE);
5397 if (*s == ';' || (s = skipspace(s), *s == ';')) {
5398 PL_nextval[PL_nexttoke].opval = Nullop;
5401 else if (*s == 'v') {
5402 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5403 s = force_version(s, FALSE);
5407 s = force_word(s,WORD,FALSE,TRUE,FALSE);
5408 s = force_version(s, FALSE);
5420 yylval.ival = CopLINE(PL_curcop);
5424 PL_hints |= HINT_BLOCK_SCOPE;
5431 LOP(OP_WAITPID,XTERM);
5440 ctl_l[0] = toCTRL('L');
5442 gv_fetchpv(ctl_l,TRUE, SVt_PV);
5445 gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */
5450 if (PL_expect == XOPERATOR)
5456 yylval.ival = OP_XOR;
5461 TERM(sublex_start());
5466 #pragma segment Main
5470 S_pending_ident(pTHX)
5473 register I32 tmp = 0;
5474 /* pit holds the identifier we read and pending_ident is reset */
5475 char pit = PL_pending_ident;
5476 PL_pending_ident = 0;
5478 DEBUG_T({ PerlIO_printf(Perl_debug_log,
5479 "### Tokener saw identifier '%s'\n", PL_tokenbuf); });
5481 /* if we're in a my(), we can't allow dynamics here.
5482 $foo'bar has already been turned into $foo::bar, so
5483 just check for colons.
5485 if it's a legal name, the OP is a PADANY.
5488 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
5489 if (strchr(PL_tokenbuf,':'))
5490 yyerror(Perl_form(aTHX_ "No package name allowed for "
5491 "variable %s in \"our\"",
5493 tmp = allocmy(PL_tokenbuf);
5496 if (strchr(PL_tokenbuf,':'))
5497 yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf));
5499 yylval.opval = newOP(OP_PADANY, 0);
5500 yylval.opval->op_targ = allocmy(PL_tokenbuf);
5506 build the ops for accesses to a my() variable.
5508 Deny my($a) or my($b) in a sort block, *if* $a or $b is
5509 then used in a comparison. This catches most, but not
5510 all cases. For instance, it catches
5511 sort { my($a); $a <=> $b }
5513 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
5514 (although why you'd do that is anyone's guess).
5517 if (!strchr(PL_tokenbuf,':')) {
5519 tmp = pad_findmy(PL_tokenbuf);
5520 if (tmp != NOT_IN_PAD) {
5521 /* might be an "our" variable" */
5522 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
5523 /* build ops for a bareword */
5524 SV *sym = newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)), 0);
5525 sv_catpvn(sym, "::", 2);
5526 sv_catpv(sym, PL_tokenbuf+1);
5527 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
5528 yylval.opval->op_private = OPpCONST_ENTERED;
5531 ? (GV_ADDMULTI | GV_ADDINEVAL)
5534 ((PL_tokenbuf[0] == '$') ? SVt_PV
5535 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5540 /* if it's a sort block and they're naming $a or $b */
5541 if (PL_last_lop_op == OP_SORT &&
5542 PL_tokenbuf[0] == '$' &&
5543 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
5546 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
5547 d < PL_bufend && *d != '\n';
5550 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
5551 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
5557 yylval.opval = newOP(OP_PADANY, 0);
5558 yylval.opval->op_targ = tmp;
5564 Whine if they've said @foo in a doublequoted string,
5565 and @foo isn't a variable we can find in the symbol
5568 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
5569 GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV);
5570 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
5571 && ckWARN(WARN_AMBIGUOUS))
5573 /* Downgraded from fatal to warning 20000522 mjd */
5574 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5575 "Possible unintended interpolation of %s in string",
5580 /* build ops for a bareword */
5581 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
5582 yylval.opval->op_private = OPpCONST_ENTERED;
5583 gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE,
5584 ((PL_tokenbuf[0] == '$') ? SVt_PV
5585 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
5591 Perl_keyword(pTHX_ register char *d, I32 len)
5596 if (strEQ(d,"__FILE__")) return -KEY___FILE__;
5597 if (strEQ(d,"__LINE__")) return -KEY___LINE__;
5598 if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__;
5599 if (strEQ(d,"__DATA__")) return KEY___DATA__;
5600 if (strEQ(d,"__END__")) return KEY___END__;
5604 if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD;
5609 if (strEQ(d,"and")) return -KEY_and;
5610 if (strEQ(d,"abs")) return -KEY_abs;
5613 if (strEQ(d,"alarm")) return -KEY_alarm;
5614 if (strEQ(d,"atan2")) return -KEY_atan2;
5617 if (strEQ(d,"accept")) return -KEY_accept;
5622 if (strEQ(d,"BEGIN")) return KEY_BEGIN;
5625 if (strEQ(d,"bless")) return -KEY_bless;
5626 if (strEQ(d,"bind")) return -KEY_bind;
5627 if (strEQ(d,"binmode")) return -KEY_binmode;
5630 if (strEQ(d,"CORE")) return -KEY_CORE;
5631 if (strEQ(d,"CHECK")) return KEY_CHECK;
5636 if (strEQ(d,"cmp")) return -KEY_cmp;
5637 if (strEQ(d,"chr")) return -KEY_chr;
5638 if (strEQ(d,"cos")) return -KEY_cos;
5641 if (strEQ(d,"chop")) return -KEY_chop;
5644 if (strEQ(d,"close")) return -KEY_close;
5645 if (strEQ(d,"chdir")) return -KEY_chdir;
5646 if (strEQ(d,"chomp")) return -KEY_chomp;
5647 if (strEQ(d,"chmod")) return -KEY_chmod;
5648 if (strEQ(d,"chown")) return -KEY_chown;
5649 if (strEQ(d,"crypt")) return -KEY_crypt;
5652 if (strEQ(d,"chroot")) return -KEY_chroot;
5653 if (strEQ(d,"caller")) return -KEY_caller;
5656 if (strEQ(d,"connect")) return -KEY_connect;
5659 if (strEQ(d,"closedir")) return -KEY_closedir;
5660 if (strEQ(d,"continue")) return -KEY_continue;
5665 if (strEQ(d,"DESTROY")) return KEY_DESTROY;
5670 if (strEQ(d,"do")) return KEY_do;
5673 if (strEQ(d,"die")) return -KEY_die;
5676 if (strEQ(d,"dump")) return -KEY_dump;
5679 if (strEQ(d,"delete")) return KEY_delete;
5682 if (strEQ(d,"defined")) return KEY_defined;
5683 if (strEQ(d,"dbmopen")) return -KEY_dbmopen;
5686 if (strEQ(d,"dbmclose")) return -KEY_dbmclose;
5691 if (strEQ(d,"END")) return KEY_END;
5696 if (strEQ(d,"eq")) return -KEY_eq;
5699 if (strEQ(d,"eof")) return -KEY_eof;
5700 if (strEQ(d,"err")) return -KEY_err;
5701 if (strEQ(d,"exp")) return -KEY_exp;
5704 if (strEQ(d,"else")) return KEY_else;
5705 if (strEQ(d,"exit")) return -KEY_exit;
5706 if (strEQ(d,"eval")) return KEY_eval;
5707 if (strEQ(d,"exec")) return -KEY_exec;
5708 if (strEQ(d,"each")) return -KEY_each;
5711 if (strEQ(d,"elsif")) return KEY_elsif;
5714 if (strEQ(d,"exists")) return KEY_exists;
5715 if (strEQ(d,"elseif") && ckWARN_d(WARN_SYNTAX))
5716 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5717 "elseif should be elsif");
5720 if (strEQ(d,"endgrent")) return -KEY_endgrent;
5721 if (strEQ(d,"endpwent")) return -KEY_endpwent;
5724 if (strEQ(d,"endnetent")) return -KEY_endnetent;
5727 if (strEQ(d,"endhostent")) return -KEY_endhostent;
5728 if (strEQ(d,"endservent")) return -KEY_endservent;
5731 if (strEQ(d,"endprotoent")) return -KEY_endprotoent;
5738 if (strEQ(d,"for")) return KEY_for;
5741 if (strEQ(d,"fork")) return -KEY_fork;
5744 if (strEQ(d,"fcntl")) return -KEY_fcntl;
5745 if (strEQ(d,"flock")) return -KEY_flock;
5748 if (strEQ(d,"format")) return KEY_format;
5749 if (strEQ(d,"fileno")) return -KEY_fileno;
5752 if (strEQ(d,"foreach")) return KEY_foreach;
5755 if (strEQ(d,"formline")) return -KEY_formline;
5760 if (strnEQ(d,"get",3)) {
5765 if (strEQ(d,"ppid")) return -KEY_getppid;
5766 if (strEQ(d,"pgrp")) return -KEY_getpgrp;
5769 if (strEQ(d,"pwent")) return -KEY_getpwent;
5770 if (strEQ(d,"pwnam")) return -KEY_getpwnam;
5771 if (strEQ(d,"pwuid")) return -KEY_getpwuid;
5774 if (strEQ(d,"peername")) return -KEY_getpeername;
5775 if (strEQ(d,"protoent")) return -KEY_getprotoent;
5776 if (strEQ(d,"priority")) return -KEY_getpriority;
5779 if (strEQ(d,"protobyname")) return -KEY_getprotobyname;
5782 if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber;
5786 else if (*d == 'h') {
5787 if (strEQ(d,"hostbyname")) return -KEY_gethostbyname;
5788 if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr;
5789 if (strEQ(d,"hostent")) return -KEY_gethostent;
5791 else if (*d == 'n') {
5792 if (strEQ(d,"netbyname")) return -KEY_getnetbyname;
5793 if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr;
5794 if (strEQ(d,"netent")) return -KEY_getnetent;
5796 else if (*d == 's') {
5797 if (strEQ(d,"servbyname")) return -KEY_getservbyname;
5798 if (strEQ(d,"servbyport")) return -KEY_getservbyport;
5799 if (strEQ(d,"servent")) return -KEY_getservent;
5800 if (strEQ(d,"sockname")) return -KEY_getsockname;
5801 if (strEQ(d,"sockopt")) return -KEY_getsockopt;
5803 else if (*d == 'g') {
5804 if (strEQ(d,"grent")) return -KEY_getgrent;
5805 if (strEQ(d,"grnam")) return -KEY_getgrnam;
5806 if (strEQ(d,"grgid")) return -KEY_getgrgid;
5808 else if (*d == 'l') {
5809 if (strEQ(d,"login")) return -KEY_getlogin;
5811 else if (*d == 'c' && d[1] == '\0') return -KEY_getc;
5816 if (strEQ(d,"gt")) return -KEY_gt;
5817 if (strEQ(d,"ge")) return -KEY_ge;
5820 if (strEQ(d,"grep")) return KEY_grep;
5821 if (strEQ(d,"goto")) return KEY_goto;
5822 if (strEQ(d,"glob")) return KEY_glob;
5825 if (strEQ(d,"gmtime")) return -KEY_gmtime;
5830 if (strEQ(d,"hex")) return -KEY_hex;
5833 if (strEQ(d,"INIT")) return KEY_INIT;
5838 if (strEQ(d,"if")) return KEY_if;
5841 if (strEQ(d,"int")) return -KEY_int;
5844 if (strEQ(d,"index")) return -KEY_index;
5845 if (strEQ(d,"ioctl")) return -KEY_ioctl;
5850 if (strEQ(d,"join")) return -KEY_join;
5854 if (strEQ(d,"keys")) return -KEY_keys;
5855 if (strEQ(d,"kill")) return -KEY_kill;
5861 if (strEQ(d,"lt")) return -KEY_lt;
5862 if (strEQ(d,"le")) return -KEY_le;
5863 if (strEQ(d,"lc")) return -KEY_lc;
5866 if (strEQ(d,"log")) return -KEY_log;
5869 if (strEQ(d,"last")) return KEY_last;
5870 if (strEQ(d,"link")) return -KEY_link;
5871 if (strEQ(d,"lock")) return -KEY_lock;
5874 if (strEQ(d,"local")) return KEY_local;
5875 if (strEQ(d,"lstat")) return -KEY_lstat;
5878 if (strEQ(d,"length")) return -KEY_length;
5879 if (strEQ(d,"listen")) return -KEY_listen;
5882 if (strEQ(d,"lcfirst")) return -KEY_lcfirst;
5885 if (strEQ(d,"localtime")) return -KEY_localtime;
5891 case 1: return KEY_m;
5893 if (strEQ(d,"my")) return KEY_my;
5896 if (strEQ(d,"map")) return KEY_map;
5899 if (strEQ(d,"mkdir")) return -KEY_mkdir;
5902 if (strEQ(d,"msgctl")) return -KEY_msgctl;
5903 if (strEQ(d,"msgget")) return -KEY_msgget;
5904 if (strEQ(d,"msgrcv")) return -KEY_msgrcv;
5905 if (strEQ(d,"msgsnd")) return -KEY_msgsnd;
5910 if (strEQ(d,"next")) return KEY_next;
5911 if (strEQ(d,"ne")) return -KEY_ne;
5912 if (strEQ(d,"not")) return -KEY_not;
5913 if (strEQ(d,"no")) return KEY_no;
5918 if (strEQ(d,"or")) return -KEY_or;
5921 if (strEQ(d,"ord")) return -KEY_ord;
5922 if (strEQ(d,"oct")) return -KEY_oct;
5923 if (strEQ(d,"our")) return KEY_our;
5926 if (strEQ(d,"open")) return -KEY_open;
5929 if (strEQ(d,"opendir")) return -KEY_opendir;
5936 if (strEQ(d,"pop")) return -KEY_pop;
5937 if (strEQ(d,"pos")) return KEY_pos;
5940 if (strEQ(d,"push")) return -KEY_push;
5941 if (strEQ(d,"pack")) return -KEY_pack;
5942 if (strEQ(d,"pipe")) return -KEY_pipe;
5945 if (strEQ(d,"print")) return KEY_print;
5948 if (strEQ(d,"printf")) return KEY_printf;
5951 if (strEQ(d,"package")) return KEY_package;
5954 if (strEQ(d,"prototype")) return KEY_prototype;
5961 else if (len == 2) {
5963 case 'r': return KEY_qr;
5964 case 'q': return KEY_qq;
5965 case 'w': return KEY_qw;
5966 case 'x': return KEY_qx;
5969 else if (strEQ(d,"quotemeta")) return -KEY_quotemeta;
5974 if (strEQ(d,"ref")) return -KEY_ref;
5977 if (strEQ(d,"read")) return -KEY_read;
5978 if (strEQ(d,"rand")) return -KEY_rand;
5979 if (strEQ(d,"recv")) return -KEY_recv;
5980 if (strEQ(d,"redo")) return KEY_redo;
5983 if (strEQ(d,"rmdir")) return -KEY_rmdir;
5984 if (strEQ(d,"reset")) return -KEY_reset;
5987 if (strEQ(d,"return")) return KEY_return;
5988 if (strEQ(d,"rename")) return -KEY_rename;
5989 if (strEQ(d,"rindex")) return -KEY_rindex;
5992 if (strEQ(d,"require")) return KEY_require;
5993 if (strEQ(d,"reverse")) return -KEY_reverse;
5994 if (strEQ(d,"readdir")) return -KEY_readdir;
5997 if (strEQ(d,"readlink")) return -KEY_readlink;
5998 if (strEQ(d,"readline")) return -KEY_readline;
5999 if (strEQ(d,"readpipe")) return -KEY_readpipe;
6002 if (strEQ(d,"rewinddir")) return -KEY_rewinddir;
6008 case 0: return KEY_s;
6010 if (strEQ(d,"scalar")) return KEY_scalar;
6015 if (strEQ(d,"seek")) return -KEY_seek;
6016 if (strEQ(d,"send")) return -KEY_send;
6019 if (strEQ(d,"semop")) return -KEY_semop;
6022 if (strEQ(d,"select")) return -KEY_select;
6023 if (strEQ(d,"semctl")) return -KEY_semctl;
6024 if (strEQ(d,"semget")) return -KEY_semget;
6027 if (strEQ(d,"setpgrp")) return -KEY_setpgrp;
6028 if (strEQ(d,"seekdir")) return -KEY_seekdir;
6031 if (strEQ(d,"setpwent")) return -KEY_setpwent;
6032 if (strEQ(d,"setgrent")) return -KEY_setgrent;
6035 if (strEQ(d,"setnetent")) return -KEY_setnetent;
6038 if (strEQ(d,"setsockopt")) return -KEY_setsockopt;
6039 if (strEQ(d,"sethostent")) return -KEY_sethostent;
6040 if (strEQ(d,"setservent")) return -KEY_setservent;
6043 if (strEQ(d,"setpriority")) return -KEY_setpriority;
6044 if (strEQ(d,"setprotoent")) return -KEY_setprotoent;
6051 if (strEQ(d,"shift")) return -KEY_shift;
6054 if (strEQ(d,"shmctl")) return -KEY_shmctl;
6055 if (strEQ(d,"shmget")) return -KEY_shmget;
6058 if (strEQ(d,"shmread")) return -KEY_shmread;
6061 if (strEQ(d,"shmwrite")) return -KEY_shmwrite;
6062 if (strEQ(d,"shutdown")) return -KEY_shutdown;
6067 if (strEQ(d,"sin")) return -KEY_sin;
6070 if (strEQ(d,"sleep")) return -KEY_sleep;
6073 if (strEQ(d,"sort")) return KEY_sort;
6074 if (strEQ(d,"socket")) return -KEY_socket;
6075 if (strEQ(d,"socketpair")) return -KEY_socketpair;
6078 if (strEQ(d,"split")) return KEY_split;
6079 if (strEQ(d,"sprintf")) return -KEY_sprintf;
6080 if (strEQ(d,"splice")) return -KEY_splice;
6083 if (strEQ(d,"sqrt")) return -KEY_sqrt;
6086 if (strEQ(d,"srand")) return -KEY_srand;
6089 if (strEQ(d,"stat")) return -KEY_stat;
6090 if (strEQ(d,"study")) return KEY_study;
6093 if (strEQ(d,"substr")) return -KEY_substr;
6094 if (strEQ(d,"sub")) return KEY_sub;
6099 if (strEQ(d,"system")) return -KEY_system;
6102 if (strEQ(d,"symlink")) return -KEY_symlink;
6103 if (strEQ(d,"syscall")) return -KEY_syscall;
6104 if (strEQ(d,"sysopen")) return -KEY_sysopen;
6105 if (strEQ(d,"sysread")) return -KEY_sysread;
6106 if (strEQ(d,"sysseek")) return -KEY_sysseek;
6109 if (strEQ(d,"syswrite")) return -KEY_syswrite;
6118 if (strEQ(d,"tr")) return KEY_tr;
6121 if (strEQ(d,"tie")) return KEY_tie;
6124 if (strEQ(d,"tell")) return -KEY_tell;
6125 if (strEQ(d,"tied")) return KEY_tied;
6126 if (strEQ(d,"time")) return -KEY_time;
6129 if (strEQ(d,"times")) return -KEY_times;
6132 if (strEQ(d,"telldir")) return -KEY_telldir;
6135 if (strEQ(d,"truncate")) return -KEY_truncate;
6142 if (strEQ(d,"uc")) return -KEY_uc;
6145 if (strEQ(d,"use")) return KEY_use;
6148 if (strEQ(d,"undef")) return KEY_undef;
6149 if (strEQ(d,"until")) return KEY_until;
6150 if (strEQ(d,"untie")) return KEY_untie;
6151 if (strEQ(d,"utime")) return -KEY_utime;
6152 if (strEQ(d,"umask")) return -KEY_umask;
6155 if (strEQ(d,"unless")) return KEY_unless;
6156 if (strEQ(d,"unpack")) return -KEY_unpack;
6157 if (strEQ(d,"unlink")) return -KEY_unlink;
6160 if (strEQ(d,"unshift")) return -KEY_unshift;
6161 if (strEQ(d,"ucfirst")) return -KEY_ucfirst;
6166 if (strEQ(d,"values")) return -KEY_values;
6167 if (strEQ(d,"vec")) return -KEY_vec;
6172 if (strEQ(d,"warn")) return -KEY_warn;
6173 if (strEQ(d,"wait")) return -KEY_wait;
6176 if (strEQ(d,"while")) return KEY_while;
6177 if (strEQ(d,"write")) return -KEY_write;
6180 if (strEQ(d,"waitpid")) return -KEY_waitpid;
6183 if (strEQ(d,"wantarray")) return -KEY_wantarray;
6188 if (len == 1) return -KEY_x;
6189 if (strEQ(d,"xor")) return -KEY_xor;
6192 if (len == 1) return KEY_y;
6201 S_checkcomma(pTHX_ register char *s, char *name, char *what)
6205 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
6206 if (ckWARN(WARN_SYNTAX)) {
6208 for (w = s+2; *w && level; w++) {
6215 for (; *w && isSPACE(*w); w++) ;
6216 if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */
6217 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6218 "%s (...) interpreted as function",name);
6221 while (s < PL_bufend && isSPACE(*s))
6225 while (s < PL_bufend && isSPACE(*s))
6227 if (isIDFIRST_lazy_if(s,UTF)) {
6229 while (isALNUM_lazy_if(s,UTF))
6231 while (s < PL_bufend && isSPACE(*s))
6236 kw = keyword(w, s - w) || get_cv(w, FALSE) != 0;
6240 Perl_croak(aTHX_ "No comma allowed after %s", what);
6245 /* Either returns sv, or mortalizes sv and returns a new SV*.
6246 Best used as sv=new_constant(..., sv, ...).
6247 If s, pv are NULL, calls subroutine with one argument,
6248 and type is used with error messages only. */
6251 S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv,
6255 HV *table = GvHV(PL_hintgv); /* ^H */
6259 const char *why1, *why2, *why3;
6261 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
6264 why2 = strEQ(key,"charnames")
6265 ? "(possibly a missing \"use charnames ...\")"
6267 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
6268 (type ? type: "undef"), why2);
6270 /* This is convoluted and evil ("goto considered harmful")
6271 * but I do not understand the intricacies of all the different
6272 * failure modes of %^H in here. The goal here is to make
6273 * the most probable error message user-friendly. --jhi */
6278 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
6279 (type ? type: "undef"), why1, why2, why3);
6281 yyerror(SvPVX(msg));
6285 cvp = hv_fetch(table, key, strlen(key), FALSE);
6286 if (!cvp || !SvOK(*cvp)) {
6289 why3 = "} is not defined";
6292 sv_2mortal(sv); /* Parent created it permanently */
6295 pv = sv_2mortal(newSVpvn(s, len));
6297 typesv = sv_2mortal(newSVpv(type, 0));
6299 typesv = &PL_sv_undef;
6301 PUSHSTACKi(PERLSI_OVERLOAD);
6313 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
6317 /* Check the eval first */
6318 if (!PL_in_eval && SvTRUE(ERRSV)) {
6320 sv_catpv(ERRSV, "Propagated");
6321 yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */
6323 res = SvREFCNT_inc(sv);
6327 (void)SvREFCNT_inc(res);
6336 why1 = "Call to &{$^H{";
6338 why3 = "}} did not return a defined value";
6346 /* Returns a NUL terminated string, with the length of the string written to
6350 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
6352 register char *d = dest;
6353 register char *e = d + destlen - 3; /* two-character token, ending NUL */
6356 Perl_croak(aTHX_ ident_too_long);
6357 if (isALNUM(*s)) /* UTF handled below */
6359 else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) {
6364 else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') {
6368 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6369 char *t = s + UTF8SKIP(s);
6370 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6372 if (d + (t - s) > e)
6373 Perl_croak(aTHX_ ident_too_long);
6374 Copy(s, d, t - s, char);
6387 S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni)
6397 e = d + destlen - 3; /* two-character token, ending NUL */
6399 while (isDIGIT(*s)) {
6401 Perl_croak(aTHX_ ident_too_long);
6408 Perl_croak(aTHX_ ident_too_long);
6409 if (isALNUM(*s)) /* UTF handled below */
6411 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
6416 else if (*s == ':' && s[1] == ':') {
6420 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
6421 char *t = s + UTF8SKIP(s);
6422 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
6424 if (d + (t - s) > e)
6425 Perl_croak(aTHX_ ident_too_long);
6426 Copy(s, d, t - s, char);
6437 if (PL_lex_state != LEX_NORMAL)
6438 PL_lex_state = LEX_INTERPENDMAYBE;
6441 if (*s == '$' && s[1] &&
6442 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
6455 if (*d == '^' && *s && isCONTROLVAR(*s)) {
6460 if (isSPACE(s[-1])) {
6463 if (!SPACE_OR_TAB(ch)) {
6469 if (isIDFIRST_lazy_if(d,UTF)) {
6473 while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') {
6475 while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e))
6478 Copy(s, d, e - s, char);
6483 while ((isALNUM(*s) || *s == ':') && d < e)
6486 Perl_croak(aTHX_ ident_too_long);
6489 while (s < send && SPACE_OR_TAB(*s)) s++;
6490 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
6491 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) {
6492 const char *brack = *s == '[' ? "[...]" : "{...}";
6493 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6494 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
6495 funny, dest, brack, funny, dest, brack);
6498 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
6502 /* Handle extended ${^Foo} variables
6503 * 1999-02-27 mjd-perl-patch@plover.com */
6504 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
6508 while (isALNUM(*s) && d < e) {
6512 Perl_croak(aTHX_ ident_too_long);
6517 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
6518 PL_lex_state = LEX_INTERPEND;
6523 if (PL_lex_state == LEX_NORMAL) {
6524 if (ckWARN(WARN_AMBIGUOUS) &&
6525 (keyword(dest, d - dest) || get_cv(dest, FALSE)))
6527 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
6528 "Ambiguous use of %c{%s} resolved to %c%s",
6529 funny, dest, funny, dest);
6534 s = bracket; /* let the parser handle it */
6538 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
6539 PL_lex_state = LEX_INTERPEND;
6544 Perl_pmflag(pTHX_ U32* pmfl, int ch)
6549 *pmfl |= PMf_GLOBAL;
6551 *pmfl |= PMf_CONTINUE;
6555 *pmfl |= PMf_MULTILINE;
6557 *pmfl |= PMf_SINGLELINE;
6559 *pmfl |= PMf_EXTENDED;
6563 S_scan_pat(pTHX_ char *start, I32 type)
6568 s = scan_str(start,FALSE,FALSE);
6570 Perl_croak(aTHX_ "Search pattern not terminated");
6572 pm = (PMOP*)newPMOP(type, 0);
6573 if (PL_multi_open == '?')
6574 pm->op_pmflags |= PMf_ONCE;
6576 while (*s && strchr("iomsx", *s))
6577 pmflag(&pm->op_pmflags,*s++);
6580 while (*s && strchr("iogcmsx", *s))
6581 pmflag(&pm->op_pmflags,*s++);
6583 /* issue a warning if /c is specified,but /g is not */
6584 if (ckWARN(WARN_REGEXP) &&
6585 (pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
6587 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_without_g);
6590 pm->op_pmpermflags = pm->op_pmflags;
6592 PL_lex_op = (OP*)pm;
6593 yylval.ival = OP_MATCH;
6598 S_scan_subst(pTHX_ char *start)
6605 yylval.ival = OP_NULL;
6607 s = scan_str(start,FALSE,FALSE);
6610 Perl_croak(aTHX_ "Substitution pattern not terminated");
6612 if (s[-1] == PL_multi_open)
6615 first_start = PL_multi_start;
6616 s = scan_str(s,FALSE,FALSE);
6619 SvREFCNT_dec(PL_lex_stuff);
6620 PL_lex_stuff = Nullsv;
6622 Perl_croak(aTHX_ "Substitution replacement not terminated");
6624 PL_multi_start = first_start; /* so whole substitution is taken together */
6626 pm = (PMOP*)newPMOP(OP_SUBST, 0);
6632 else if (strchr("iogcmsx", *s))
6633 pmflag(&pm->op_pmflags,*s++);
6638 /* /c is not meaningful with s/// */
6639 if (ckWARN(WARN_REGEXP) && (pm->op_pmflags & PMf_CONTINUE))
6641 Perl_warner(aTHX_ packWARN(WARN_REGEXP), c_in_subst);
6646 PL_sublex_info.super_bufptr = s;
6647 PL_sublex_info.super_bufend = PL_bufend;
6649 pm->op_pmflags |= PMf_EVAL;
6650 repl = newSVpvn("",0);
6652 sv_catpv(repl, es ? "eval " : "do ");
6653 sv_catpvn(repl, "{ ", 2);
6654 sv_catsv(repl, PL_lex_repl);
6655 sv_catpvn(repl, " };", 2);
6657 SvREFCNT_dec(PL_lex_repl);
6661 pm->op_pmpermflags = pm->op_pmflags;
6662 PL_lex_op = (OP*)pm;
6663 yylval.ival = OP_SUBST;
6668 S_scan_trans(pTHX_ char *start)
6677 yylval.ival = OP_NULL;
6679 s = scan_str(start,FALSE,FALSE);
6681 Perl_croak(aTHX_ "Transliteration pattern not terminated");
6682 if (s[-1] == PL_multi_open)
6685 s = scan_str(s,FALSE,FALSE);
6688 SvREFCNT_dec(PL_lex_stuff);
6689 PL_lex_stuff = Nullsv;
6691 Perl_croak(aTHX_ "Transliteration replacement not terminated");
6694 complement = del = squash = 0;
6698 complement = OPpTRANS_COMPLEMENT;
6701 del = OPpTRANS_DELETE;
6704 squash = OPpTRANS_SQUASH;
6713 New(803, tbl, complement&&!del?258:256, short);
6714 o = newPVOP(OP_TRANS, 0, (char*)tbl);
6715 o->op_private &= ~OPpTRANS_ALL;
6716 o->op_private |= del|squash|complement|
6717 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
6718 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
6721 yylval.ival = OP_TRANS;
6726 S_scan_heredoc(pTHX_ register char *s)
6729 I32 op_type = OP_SCALAR;
6736 int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
6740 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
6743 for (peek = s; SPACE_OR_TAB(*peek); peek++) ;
6744 if (*peek == '`' || *peek == '\'' || *peek =='"') {
6747 s = delimcpy(d, e, s, PL_bufend, term, &len);
6757 if (!isALNUM_lazy_if(s,UTF))
6758 deprecate_old("bare << to mean <<\"\"");
6759 for (; isALNUM_lazy_if(s,UTF); s++) {
6764 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
6765 Perl_croak(aTHX_ "Delimiter for here document is too long");
6768 len = d - PL_tokenbuf;
6769 #ifndef PERL_STRICT_CR
6770 d = strchr(s, '\r');
6774 while (s < PL_bufend) {
6780 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
6789 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6794 if (outer || !(d=ninstr(s,PL_bufend,d,d+1)))
6795 herewas = newSVpvn(s,PL_bufend-s);
6797 s--, herewas = newSVpvn(s,d-s);
6798 s += SvCUR(herewas);
6800 tmpstr = NEWSV(87,79);
6801 sv_upgrade(tmpstr, SVt_PVIV);
6806 else if (term == '`') {
6807 op_type = OP_BACKTICK;
6808 SvIVX(tmpstr) = '\\';
6812 PL_multi_start = CopLINE(PL_curcop);
6813 PL_multi_open = PL_multi_close = '<';
6814 term = *PL_tokenbuf;
6815 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
6816 char *bufptr = PL_sublex_info.super_bufptr;
6817 char *bufend = PL_sublex_info.super_bufend;
6818 char *olds = s - SvCUR(herewas);
6819 s = strchr(bufptr, '\n');
6823 while (s < bufend &&
6824 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6826 CopLINE_inc(PL_curcop);
6829 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6830 missingterm(PL_tokenbuf);
6832 sv_setpvn(herewas,bufptr,d-bufptr+1);
6833 sv_setpvn(tmpstr,d+1,s-d);
6835 sv_catpvn(herewas,s,bufend-s);
6836 Copy(SvPVX(herewas),bufptr,SvCUR(herewas) + 1,char);
6843 while (s < PL_bufend &&
6844 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
6846 CopLINE_inc(PL_curcop);
6848 if (s >= PL_bufend) {
6849 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6850 missingterm(PL_tokenbuf);
6852 sv_setpvn(tmpstr,d+1,s-d);
6854 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
6856 sv_catpvn(herewas,s,PL_bufend-s);
6857 sv_setsv(PL_linestr,herewas);
6858 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
6859 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6860 PL_last_lop = PL_last_uni = Nullch;
6863 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
6864 while (s >= PL_bufend) { /* multiple line string? */
6866 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
6867 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
6868 missingterm(PL_tokenbuf);
6870 CopLINE_inc(PL_curcop);
6871 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6872 PL_last_lop = PL_last_uni = Nullch;
6873 #ifndef PERL_STRICT_CR
6874 if (PL_bufend - PL_linestart >= 2) {
6875 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
6876 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
6878 PL_bufend[-2] = '\n';
6880 SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr));
6882 else if (PL_bufend[-1] == '\r')
6883 PL_bufend[-1] = '\n';
6885 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
6886 PL_bufend[-1] = '\n';
6888 if (PERLDB_LINE && PL_curstash != PL_debstash) {
6889 SV *sv = NEWSV(88,0);
6891 sv_upgrade(sv, SVt_PVMG);
6892 sv_setsv(sv,PL_linestr);
6895 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv);
6897 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
6898 STRLEN off = PL_bufend - 1 - SvPVX(PL_linestr);
6899 *(SvPVX(PL_linestr) + off ) = ' ';
6900 sv_catsv(PL_linestr,herewas);
6901 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
6902 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
6906 sv_catsv(tmpstr,PL_linestr);
6911 PL_multi_end = CopLINE(PL_curcop);
6912 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
6913 SvLEN_set(tmpstr, SvCUR(tmpstr) + 1);
6914 Renew(SvPVX(tmpstr), SvLEN(tmpstr), char);
6916 SvREFCNT_dec(herewas);
6918 if (UTF && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr)))
6920 else if (PL_encoding)
6921 sv_recode_to_utf8(tmpstr, PL_encoding);
6923 PL_lex_stuff = tmpstr;
6924 yylval.ival = op_type;
6929 takes: current position in input buffer
6930 returns: new position in input buffer
6931 side-effects: yylval and lex_op are set.
6936 <FH> read from filehandle
6937 <pkg::FH> read from package qualified filehandle
6938 <pkg'FH> read from package qualified filehandle
6939 <$fh> read from filehandle in $fh
6945 S_scan_inputsymbol(pTHX_ char *start)
6947 register char *s = start; /* current position in buffer */
6953 d = PL_tokenbuf; /* start of temp holding space */
6954 e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
6955 end = strchr(s, '\n');
6958 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
6960 /* die if we didn't have space for the contents of the <>,
6961 or if it didn't end, or if we see a newline
6964 if (len >= sizeof PL_tokenbuf)
6965 Perl_croak(aTHX_ "Excessively long <> operator");
6967 Perl_croak(aTHX_ "Unterminated <> operator");
6972 Remember, only scalar variables are interpreted as filehandles by
6973 this code. Anything more complex (e.g., <$fh{$num}>) will be
6974 treated as a glob() call.
6975 This code makes use of the fact that except for the $ at the front,
6976 a scalar variable and a filehandle look the same.
6978 if (*d == '$' && d[1]) d++;
6980 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
6981 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
6984 /* If we've tried to read what we allow filehandles to look like, and
6985 there's still text left, then it must be a glob() and not a getline.
6986 Use scan_str to pull out the stuff between the <> and treat it
6987 as nothing more than a string.
6990 if (d - PL_tokenbuf != len) {
6991 yylval.ival = OP_GLOB;
6993 s = scan_str(start,FALSE,FALSE);
6995 Perl_croak(aTHX_ "Glob not terminated");
6999 bool readline_overriden = FALSE;
7000 GV *gv_readline = Nullgv;
7002 /* we're in a filehandle read situation */
7005 /* turn <> into <ARGV> */
7007 Copy("ARGV",d,5,char);
7009 /* Check whether readline() is overriden */
7010 if (((gv_readline = gv_fetchpv("readline", FALSE, SVt_PVCV))
7011 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
7013 ((gvp = (GV**)hv_fetch(PL_globalstash, "readline", 8, FALSE))
7014 && (gv_readline = *gvp) != (GV*)&PL_sv_undef
7015 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
7016 readline_overriden = TRUE;
7018 /* if <$fh>, create the ops to turn the variable into a
7024 /* try to find it in the pad for this block, otherwise find
7025 add symbol table ops
7027 if ((tmp = pad_findmy(d)) != NOT_IN_PAD) {
7028 if (PAD_COMPNAME_FLAGS(tmp) & SVpad_OUR) {
7029 SV *sym = sv_2mortal(
7030 newSVpv(HvNAME(PAD_COMPNAME_OURSTASH(tmp)),0));
7031 sv_catpvn(sym, "::", 2);
7037 OP *o = newOP(OP_PADSV, 0);
7039 PL_lex_op = readline_overriden
7040 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
7041 append_elem(OP_LIST, o,
7042 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
7043 : (OP*)newUNOP(OP_READLINE, 0, o);
7052 ? (GV_ADDMULTI | GV_ADDINEVAL)
7055 PL_lex_op = readline_overriden
7056 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
7057 append_elem(OP_LIST,
7058 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
7059 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
7060 : (OP*)newUNOP(OP_READLINE, 0,
7061 newUNOP(OP_RV2SV, 0,
7062 newGVOP(OP_GV, 0, gv)));
7064 if (!readline_overriden)
7065 PL_lex_op->op_flags |= OPf_SPECIAL;
7066 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
7067 yylval.ival = OP_NULL;
7070 /* If it's none of the above, it must be a literal filehandle
7071 (<Foo::BAR> or <FOO>) so build a simple readline OP */
7073 GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO);
7074 PL_lex_op = readline_overriden
7075 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
7076 append_elem(OP_LIST,
7077 newGVOP(OP_GV, 0, gv),
7078 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
7079 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
7080 yylval.ival = OP_NULL;
7089 takes: start position in buffer
7090 keep_quoted preserve \ on the embedded delimiter(s)
7091 keep_delims preserve the delimiters around the string
7092 returns: position to continue reading from buffer
7093 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
7094 updates the read buffer.
7096 This subroutine pulls a string out of the input. It is called for:
7097 q single quotes q(literal text)
7098 ' single quotes 'literal text'
7099 qq double quotes qq(interpolate $here please)
7100 " double quotes "interpolate $here please"
7101 qx backticks qx(/bin/ls -l)
7102 ` backticks `/bin/ls -l`
7103 qw quote words @EXPORT_OK = qw( func() $spam )
7104 m// regexp match m/this/
7105 s/// regexp substitute s/this/that/
7106 tr/// string transliterate tr/this/that/
7107 y/// string transliterate y/this/that/
7108 ($*@) sub prototypes sub foo ($)
7109 (stuff) sub attr parameters sub foo : attr(stuff)
7110 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
7112 In most of these cases (all but <>, patterns and transliterate)
7113 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
7114 calls scan_str(). s/// makes yylex() call scan_subst() which calls
7115 scan_str(). tr/// and y/// make yylex() call scan_trans() which
7118 It skips whitespace before the string starts, and treats the first
7119 character as the delimiter. If the delimiter is one of ([{< then
7120 the corresponding "close" character )]}> is used as the closing
7121 delimiter. It allows quoting of delimiters, and if the string has
7122 balanced delimiters ([{<>}]) it allows nesting.
7124 On success, the SV with the resulting string is put into lex_stuff or,
7125 if that is already non-NULL, into lex_repl. The second case occurs only
7126 when parsing the RHS of the special constructs s/// and tr/// (y///).
7127 For convenience, the terminating delimiter character is stuffed into
7132 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
7134 SV *sv; /* scalar value: string */
7135 char *tmps; /* temp string, used for delimiter matching */
7136 register char *s = start; /* current position in the buffer */
7137 register char term; /* terminating character */
7138 register char *to; /* current position in the sv's data */
7139 I32 brackets = 1; /* bracket nesting level */
7140 bool has_utf8 = FALSE; /* is there any utf8 content? */
7141 I32 termcode; /* terminating char. code */
7142 U8 termstr[UTF8_MAXLEN]; /* terminating string */
7143 STRLEN termlen; /* length of terminating string */
7144 char *last = NULL; /* last position for nesting bracket */
7146 /* skip space before the delimiter */
7150 /* mark where we are, in case we need to report errors */
7153 /* after skipping whitespace, the next character is the terminator */
7156 termcode = termstr[0] = term;
7160 termcode = utf8_to_uvchr((U8*)s, &termlen);
7161 Copy(s, termstr, termlen, U8);
7162 if (!UTF8_IS_INVARIANT(term))
7166 /* mark where we are */
7167 PL_multi_start = CopLINE(PL_curcop);
7168 PL_multi_open = term;
7170 /* find corresponding closing delimiter */
7171 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
7172 termcode = termstr[0] = term = tmps[5];
7174 PL_multi_close = term;
7176 /* create a new SV to hold the contents. 87 is leak category, I'm
7177 assuming. 79 is the SV's initial length. What a random number. */
7179 sv_upgrade(sv, SVt_PVIV);
7180 SvIVX(sv) = termcode;
7181 (void)SvPOK_only(sv); /* validate pointer */
7183 /* move past delimiter and try to read a complete string */
7185 sv_catpvn(sv, s, termlen);
7188 if (PL_encoding && !UTF) {
7192 int offset = s - SvPVX(PL_linestr);
7193 bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
7194 &offset, (char*)termstr, termlen);
7195 char *ns = SvPVX(PL_linestr) + offset;
7196 char *svlast = SvEND(sv) - 1;
7198 for (; s < ns; s++) {
7199 if (*s == '\n' && !PL_rsfp)
7200 CopLINE_inc(PL_curcop);
7203 goto read_more_line;
7205 /* handle quoted delimiters */
7206 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
7208 for (t = svlast-2; t >= SvPVX(sv) && *t == '\\';)
7210 if ((svlast-1 - t) % 2) {
7214 SvCUR_set(sv, SvCUR(sv) - 1);
7219 if (PL_multi_open == PL_multi_close) {
7226 for (w = t = last; t < svlast; w++, t++) {
7227 /* At here, all closes are "was quoted" one,
7228 so we don't check PL_multi_close. */
7230 if (!keep_quoted && *(t+1) == PL_multi_open)
7235 else if (*t == PL_multi_open)
7243 SvCUR_set(sv, w - SvPVX(sv));
7246 if (--brackets <= 0)
7252 SvCUR_set(sv, SvCUR(sv) - 1);
7258 /* extend sv if need be */
7259 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
7260 /* set 'to' to the next character in the sv's string */
7261 to = SvPVX(sv)+SvCUR(sv);
7263 /* if open delimiter is the close delimiter read unbridle */
7264 if (PL_multi_open == PL_multi_close) {
7265 for (; s < PL_bufend; s++,to++) {
7266 /* embedded newlines increment the current line number */
7267 if (*s == '\n' && !PL_rsfp)
7268 CopLINE_inc(PL_curcop);
7269 /* handle quoted delimiters */
7270 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
7271 if (!keep_quoted && s[1] == term)
7273 /* any other quotes are simply copied straight through */
7277 /* terminate when run out of buffer (the for() condition), or
7278 have found the terminator */
7279 else if (*s == term) {
7282 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
7285 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7291 /* if the terminator isn't the same as the start character (e.g.,
7292 matched brackets), we have to allow more in the quoting, and
7293 be prepared for nested brackets.
7296 /* read until we run out of string, or we find the terminator */
7297 for (; s < PL_bufend; s++,to++) {
7298 /* embedded newlines increment the line count */
7299 if (*s == '\n' && !PL_rsfp)
7300 CopLINE_inc(PL_curcop);
7301 /* backslashes can escape the open or closing characters */
7302 if (*s == '\\' && s+1 < PL_bufend) {
7304 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
7309 /* allow nested opens and closes */
7310 else if (*s == PL_multi_close && --brackets <= 0)
7312 else if (*s == PL_multi_open)
7314 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
7319 /* terminate the copied string and update the sv's end-of-string */
7321 SvCUR_set(sv, to - SvPVX(sv));
7324 * this next chunk reads more into the buffer if we're not done yet
7328 break; /* handle case where we are done yet :-) */
7330 #ifndef PERL_STRICT_CR
7331 if (to - SvPVX(sv) >= 2) {
7332 if ((to[-2] == '\r' && to[-1] == '\n') ||
7333 (to[-2] == '\n' && to[-1] == '\r'))
7337 SvCUR_set(sv, to - SvPVX(sv));
7339 else if (to[-1] == '\r')
7342 else if (to - SvPVX(sv) == 1 && to[-1] == '\r')
7347 /* if we're out of file, or a read fails, bail and reset the current
7348 line marker so we can report where the unterminated string began
7351 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
7353 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
7356 /* we read a line, so increment our line counter */
7357 CopLINE_inc(PL_curcop);
7359 /* update debugger info */
7360 if (PERLDB_LINE && PL_curstash != PL_debstash) {
7361 SV *sv = NEWSV(88,0);
7363 sv_upgrade(sv, SVt_PVMG);
7364 sv_setsv(sv,PL_linestr);
7367 av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv);
7370 /* having changed the buffer, we must update PL_bufend */
7371 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7372 PL_last_lop = PL_last_uni = Nullch;
7375 /* at this point, we have successfully read the delimited string */
7377 if (!PL_encoding || UTF) {
7379 sv_catpvn(sv, s, termlen);
7382 if (has_utf8 || PL_encoding)
7385 PL_multi_end = CopLINE(PL_curcop);
7387 /* if we allocated too much space, give some back */
7388 if (SvCUR(sv) + 5 < SvLEN(sv)) {
7389 SvLEN_set(sv, SvCUR(sv) + 1);
7390 Renew(SvPVX(sv), SvLEN(sv), char);
7393 /* decide whether this is the first or second quoted string we've read
7406 takes: pointer to position in buffer
7407 returns: pointer to new position in buffer
7408 side-effects: builds ops for the constant in yylval.op
7410 Read a number in any of the formats that Perl accepts:
7412 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
7413 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
7416 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
7418 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
7421 If it reads a number without a decimal point or an exponent, it will
7422 try converting the number to an integer and see if it can do so
7423 without loss of precision.
7427 Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp)
7429 register char *s = start; /* current position in buffer */
7430 register char *d; /* destination in temp buffer */
7431 register char *e; /* end of temp buffer */
7432 NV nv; /* number read, as a double */
7433 SV *sv = Nullsv; /* place to put the converted number */
7434 bool floatit; /* boolean: int or float? */
7435 char *lastub = 0; /* position of last underbar */
7436 static char number_too_long[] = "Number too long";
7438 /* We use the first character to decide what type of number this is */
7442 Perl_croak(aTHX_ "panic: scan_num");
7444 /* if it starts with a 0, it could be an octal number, a decimal in
7445 0.13 disguise, or a hexadecimal number, or a binary number. */
7449 u holds the "number so far"
7450 shift the power of 2 of the base
7451 (hex == 4, octal == 3, binary == 1)
7452 overflowed was the number more than we can hold?
7454 Shift is used when we add a digit. It also serves as an "are
7455 we in octal/hex/binary?" indicator to disallow hex characters
7461 bool overflowed = FALSE;
7462 bool just_zero = TRUE; /* just plain 0 or binary number? */
7463 static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
7464 static char* bases[5] = { "", "binary", "", "octal",
7466 static char* Bases[5] = { "", "Binary", "", "Octal",
7468 static char *maxima[5] = { "",
7469 "0b11111111111111111111111111111111",
7473 char *base, *Base, *max;
7480 } else if (s[1] == 'b') {
7485 /* check for a decimal in disguise */
7486 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
7488 /* so it must be octal */
7495 if (ckWARN(WARN_SYNTAX))
7496 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7497 "Misplaced _ in number");
7501 base = bases[shift];
7502 Base = Bases[shift];
7503 max = maxima[shift];
7505 /* read the rest of the number */
7507 /* x is used in the overflow test,
7508 b is the digit we're adding on. */
7513 /* if we don't mention it, we're done */
7517 /* _ are ignored -- but warned about if consecutive */
7519 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7520 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7521 "Misplaced _ in number");
7525 /* 8 and 9 are not octal */
7528 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
7532 case '2': case '3': case '4':
7533 case '5': case '6': case '7':
7535 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
7539 b = *s++ & 15; /* ASCII digit -> value of digit */
7543 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
7544 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
7545 /* make sure they said 0x */
7550 /* Prepare to put the digit we have onto the end
7551 of the number so far. We check for overflows.
7557 x = u << shift; /* make room for the digit */
7559 if ((x >> shift) != u
7560 && !(PL_hints & HINT_NEW_BINARY)) {
7563 if (ckWARN_d(WARN_OVERFLOW))
7564 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
7565 "Integer overflow in %s number",
7568 u = x | b; /* add the digit to the end */
7571 n *= nvshift[shift];
7572 /* If an NV has not enough bits in its
7573 * mantissa to represent an UV this summing of
7574 * small low-order numbers is a waste of time
7575 * (because the NV cannot preserve the
7576 * low-order bits anyway): we could just
7577 * remember when did we overflow and in the
7578 * end just multiply n by the right
7586 /* if we get here, we had success: make a scalar value from
7591 /* final misplaced underbar check */
7593 if (ckWARN(WARN_SYNTAX))
7594 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7599 if (ckWARN(WARN_PORTABLE) && n > 4294967295.0)
7600 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7601 "%s number > %s non-portable",
7607 if (ckWARN(WARN_PORTABLE) && u > 0xffffffff)
7608 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
7609 "%s number > %s non-portable",
7614 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
7615 sv = new_constant(start, s - start, "integer",
7617 else if (PL_hints & HINT_NEW_BINARY)
7618 sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL);
7623 handle decimal numbers.
7624 we're also sent here when we read a 0 as the first digit
7626 case '1': case '2': case '3': case '4': case '5':
7627 case '6': case '7': case '8': case '9': case '.':
7630 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
7633 /* read next group of digits and _ and copy into d */
7634 while (isDIGIT(*s) || *s == '_') {
7635 /* skip underscores, checking for misplaced ones
7639 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7640 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7641 "Misplaced _ in number");
7645 /* check for end of fixed-length buffer */
7647 Perl_croak(aTHX_ number_too_long);
7648 /* if we're ok, copy the character */
7653 /* final misplaced underbar check */
7654 if (lastub && s == lastub + 1) {
7655 if (ckWARN(WARN_SYNTAX))
7656 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
7659 /* read a decimal portion if there is one. avoid
7660 3..5 being interpreted as the number 3. followed
7663 if (*s == '.' && s[1] != '.') {
7668 if (ckWARN(WARN_SYNTAX))
7669 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7670 "Misplaced _ in number");
7674 /* copy, ignoring underbars, until we run out of digits.
7676 for (; isDIGIT(*s) || *s == '_'; s++) {
7677 /* fixed length buffer check */
7679 Perl_croak(aTHX_ number_too_long);
7681 if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1)
7682 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7683 "Misplaced _ in number");
7689 /* fractional part ending in underbar? */
7691 if (ckWARN(WARN_SYNTAX))
7692 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7693 "Misplaced _ in number");
7695 if (*s == '.' && isDIGIT(s[1])) {
7696 /* oops, it's really a v-string, but without the "v" */
7702 /* read exponent part, if present */
7703 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
7707 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
7708 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
7710 /* stray preinitial _ */
7712 if (ckWARN(WARN_SYNTAX))
7713 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7714 "Misplaced _ in number");
7718 /* allow positive or negative exponent */
7719 if (*s == '+' || *s == '-')
7722 /* stray initial _ */
7724 if (ckWARN(WARN_SYNTAX))
7725 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7726 "Misplaced _ in number");
7730 /* read digits of exponent */
7731 while (isDIGIT(*s) || *s == '_') {
7734 Perl_croak(aTHX_ number_too_long);
7738 if (ckWARN(WARN_SYNTAX) &&
7739 ((lastub && s == lastub + 1) ||
7740 (!isDIGIT(s[1]) && s[1] != '_')))
7741 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
7742 "Misplaced _ in number");
7749 /* make an sv from the string */
7753 We try to do an integer conversion first if no characters
7754 indicating "float" have been found.
7759 int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
7761 if (flags == IS_NUMBER_IN_UV) {
7763 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
7766 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
7767 if (uv <= (UV) IV_MIN)
7768 sv_setiv(sv, -(IV)uv);
7775 /* terminate the string */
7777 nv = Atof(PL_tokenbuf);
7781 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
7782 (PL_hints & HINT_NEW_INTEGER) )
7783 sv = new_constant(PL_tokenbuf, d - PL_tokenbuf,
7784 (floatit ? "float" : "integer"),
7788 /* if it starts with a v, it could be a v-string */
7791 sv = NEWSV(92,5); /* preallocate storage space */
7792 s = scan_vstring(s,sv);
7796 /* make the op for the constant and return */
7799 lvalp->opval = newSVOP(OP_CONST, 0, sv);
7801 lvalp->opval = Nullop;
7807 S_scan_formline(pTHX_ register char *s)
7811 SV *stuff = newSVpvn("",0);
7812 bool needargs = FALSE;
7818 #ifdef PERL_STRICT_CR
7819 for (t = s+1;SPACE_OR_TAB(*t); t++) ;
7821 for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ;
7823 if (*t == '\n' || t == PL_bufend) {
7828 if (PL_in_eval && !PL_rsfp) {
7829 eol = memchr(s,'\n',PL_bufend-s);
7834 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
7836 for (t = s; t < eol; t++) {
7837 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
7839 goto enough; /* ~~ must be first line in formline */
7841 if (*t == '@' || *t == '^')
7845 sv_catpvn(stuff, s, eol-s);
7846 #ifndef PERL_STRICT_CR
7847 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
7848 char *end = SvPVX(stuff) + SvCUR(stuff);
7860 s = filter_gets(PL_linestr, PL_rsfp, 0);
7861 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
7862 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
7863 PL_last_lop = PL_last_uni = Nullch;
7875 PL_lex_state = LEX_NORMAL;
7876 PL_nextval[PL_nexttoke].ival = 0;
7880 PL_lex_state = LEX_FORMLINE;
7882 if (UTF && is_utf8_string((U8*)SvPVX(stuff), SvCUR(stuff)))
7884 else if (PL_encoding)
7885 sv_recode_to_utf8(stuff, PL_encoding);
7887 PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff);
7889 PL_nextval[PL_nexttoke].ival = OP_FORMLINE;
7893 SvREFCNT_dec(stuff);
7895 PL_lex_formbrack = 0;
7906 PL_cshlen = strlen(PL_cshname);
7911 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
7913 I32 oldsavestack_ix = PL_savestack_ix;
7914 CV* outsidecv = PL_compcv;
7917 assert(SvTYPE(PL_compcv) == SVt_PVCV);
7919 SAVEI32(PL_subline);
7920 save_item(PL_subname);
7921 SAVESPTR(PL_compcv);
7923 PL_compcv = (CV*)NEWSV(1104,0);
7924 sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV);
7925 CvFLAGS(PL_compcv) |= flags;
7927 PL_subline = CopLINE(PL_curcop);
7928 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
7929 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv);
7930 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
7932 return oldsavestack_ix;
7936 #pragma segment Perl_yylex
7939 Perl_yywarn(pTHX_ char *s)
7941 PL_in_eval |= EVAL_WARNONLY;
7943 PL_in_eval &= ~EVAL_WARNONLY;
7948 Perl_yyerror(pTHX_ char *s)
7951 char *context = NULL;
7955 if (!yychar || (yychar == ';' && !PL_rsfp))
7957 else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 &&
7958 PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) {
7961 The code below is removed for NetWare because it abends/crashes on NetWare
7962 when the script has error such as not having the closing quotes like:
7964 Checking of white spaces is anyway done in NetWare code.
7967 while (isSPACE(*PL_oldoldbufptr))
7970 context = PL_oldoldbufptr;
7971 contlen = PL_bufptr - PL_oldoldbufptr;
7973 else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 &&
7974 PL_oldbufptr != PL_bufptr) {
7977 The code below is removed for NetWare because it abends/crashes on NetWare
7978 when the script has error such as not having the closing quotes like:
7980 Checking of white spaces is anyway done in NetWare code.
7983 while (isSPACE(*PL_oldbufptr))
7986 context = PL_oldbufptr;
7987 contlen = PL_bufptr - PL_oldbufptr;
7989 else if (yychar > 255)
7990 where = "next token ???";
7991 else if (yychar == -2) { /* YYEMPTY */
7992 if (PL_lex_state == LEX_NORMAL ||
7993 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
7994 where = "at end of line";
7995 else if (PL_lex_inpat)
7996 where = "within pattern";
7998 where = "within string";
8001 SV *where_sv = sv_2mortal(newSVpvn("next char ", 10));
8003 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
8004 else if (isPRINT_LC(yychar))
8005 Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar);
8007 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
8008 where = SvPVX(where_sv);
8010 msg = sv_2mortal(newSVpv(s, 0));
8011 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
8012 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
8014 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
8016 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
8017 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
8018 Perl_sv_catpvf(aTHX_ msg,
8019 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
8020 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
8023 if (PL_in_eval & EVAL_WARNONLY && ckWARN_d(WARN_SYNTAX))
8024 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, msg);
8027 if (PL_error_count >= 10) {
8028 if (PL_in_eval && SvCUR(ERRSV))
8029 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
8030 ERRSV, OutCopFILE(PL_curcop));
8032 Perl_croak(aTHX_ "%s has too many errors.\n",
8033 OutCopFILE(PL_curcop));
8036 PL_in_my_stash = Nullhv;
8040 #pragma segment Main
8044 S_swallow_bom(pTHX_ U8 *s)
8047 slen = SvCUR(PL_linestr);
8051 /* UTF-16 little-endian? (or UTF32-LE?) */
8052 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
8053 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
8054 #ifndef PERL_NO_UTF16_FILTER
8055 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
8058 if (PL_bufend > (char*)s) {
8062 filter_add(utf16rev_textfilter, NULL);
8063 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
8064 utf16_to_utf8_reversed(s, news,
8065 PL_bufend - (char*)s - 1,
8067 sv_setpvn(PL_linestr, (const char*)news, newlen);
8069 SvUTF8_on(PL_linestr);
8070 s = (U8*)SvPVX(PL_linestr);
8071 PL_bufend = SvPVX(PL_linestr) + newlen;
8074 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
8079 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
8080 #ifndef PERL_NO_UTF16_FILTER
8081 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
8084 if (PL_bufend > (char *)s) {
8088 filter_add(utf16_textfilter, NULL);
8089 New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
8090 utf16_to_utf8(s, news,
8091 PL_bufend - (char*)s,
8093 sv_setpvn(PL_linestr, (const char*)news, newlen);
8095 SvUTF8_on(PL_linestr);
8096 s = (U8*)SvPVX(PL_linestr);
8097 PL_bufend = SvPVX(PL_linestr) + newlen;
8100 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
8105 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
8106 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
8113 if (s[2] == 0xFE && s[3] == 0xFF) {
8114 /* UTF-32 big-endian */
8115 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
8118 else if (s[2] == 0 && s[3] != 0) {
8121 * are a good indicator of UTF-16BE. */
8122 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
8127 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
8130 * are a good indicator of UTF-16LE. */
8131 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
8140 * Restore a source filter.
8144 restore_rsfp(pTHX_ void *f)
8146 PerlIO *fp = (PerlIO*)f;
8148 if (PL_rsfp == PerlIO_stdin())
8149 PerlIO_clearerr(PL_rsfp);
8150 else if (PL_rsfp && (PL_rsfp != fp))
8151 PerlIO_close(PL_rsfp);
8155 #ifndef PERL_NO_UTF16_FILTER
8157 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
8159 STRLEN old = SvCUR(sv);
8160 I32 count = FILTER_READ(idx+1, sv, maxlen);
8161 DEBUG_P(PerlIO_printf(Perl_debug_log,
8162 "utf16_textfilter(%p): %d %d (%d)\n",
8163 utf16_textfilter, idx, maxlen, (int) count));
8167 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
8168 Copy(SvPVX(sv), tmps, old, char);
8169 utf16_to_utf8((U8*)SvPVX(sv) + old, tmps + old,
8170 SvCUR(sv) - old, &newlen);
8171 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
8173 DEBUG_P({sv_dump(sv);});
8178 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
8180 STRLEN old = SvCUR(sv);
8181 I32 count = FILTER_READ(idx+1, sv, maxlen);
8182 DEBUG_P(PerlIO_printf(Perl_debug_log,
8183 "utf16rev_textfilter(%p): %d %d (%d)\n",
8184 utf16rev_textfilter, idx, maxlen, (int) count));
8188 New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8);
8189 Copy(SvPVX(sv), tmps, old, char);
8190 utf16_to_utf8((U8*)SvPVX(sv) + old, tmps + old,
8191 SvCUR(sv) - old, &newlen);
8192 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
8194 DEBUG_P({ sv_dump(sv); });
8200 Returns a pointer to the next character after the parsed
8201 vstring, as well as updating the passed in sv.
8203 Function must be called like
8206 s = scan_vstring(s,sv);
8208 The sv should already be large enough to store the vstring
8209 passed in, for performance reasons.
8214 Perl_scan_vstring(pTHX_ char *s, SV *sv)
8218 if (*pos == 'v') pos++; /* get past 'v' */
8219 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8222 /* this may not be a v-string if followed by => */
8224 while (next < PL_bufend && isSPACE(*next))
8226 if ((PL_bufend - next) >= 2 && *next == '=' && next[1] == '>' ) {
8227 /* return string not v-string */
8228 sv_setpvn(sv,(char *)s,pos-s);
8233 if (!isALPHA(*pos)) {
8235 U8 tmpbuf[UTF8_MAXLEN+1];
8238 if (*s == 'v') s++; /* get past 'v' */
8240 sv_setpvn(sv, "", 0);
8245 /* this is atoi() that tolerates underscores */
8248 while (--end >= s) {
8253 rev += (*end - '0') * mult;
8255 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
8256 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
8257 "Integer overflow in decimal number");
8261 if (rev > 0x7FFFFFFF)
8262 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
8264 /* Append native character for the rev point */
8265 tmpend = uvchr_to_utf8(tmpbuf, rev);
8266 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
8267 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
8269 if (pos + 1 < PL_bufend && *pos == '.' && isDIGIT(pos[1]))
8275 while (pos < PL_bufend && (isDIGIT(*pos) || *pos == '_'))
8279 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);