3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 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
14 * [p.719 of _The Lord of the Rings_, IV/ix: "Shelob's Lair"]
18 * This file is the lexer for Perl. It's closely linked to the
21 * The main routine is yylex(), which returns the next token.
25 #define PERL_IN_TOKE_C
28 #define new_constant(a,b,c,d,e,f,g) \
29 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
31 #define pl_yylval (PL_parser->yylval)
33 /* YYINITDEPTH -- initial size of the parser's stacks. */
34 #define YYINITDEPTH 200
36 /* XXX temporary backwards compatibility */
37 #define PL_lex_brackets (PL_parser->lex_brackets)
38 #define PL_lex_brackstack (PL_parser->lex_brackstack)
39 #define PL_lex_casemods (PL_parser->lex_casemods)
40 #define PL_lex_casestack (PL_parser->lex_casestack)
41 #define PL_lex_defer (PL_parser->lex_defer)
42 #define PL_lex_dojoin (PL_parser->lex_dojoin)
43 #define PL_lex_expect (PL_parser->lex_expect)
44 #define PL_lex_formbrack (PL_parser->lex_formbrack)
45 #define PL_lex_inpat (PL_parser->lex_inpat)
46 #define PL_lex_inwhat (PL_parser->lex_inwhat)
47 #define PL_lex_op (PL_parser->lex_op)
48 #define PL_lex_repl (PL_parser->lex_repl)
49 #define PL_lex_starts (PL_parser->lex_starts)
50 #define PL_lex_stuff (PL_parser->lex_stuff)
51 #define PL_multi_start (PL_parser->multi_start)
52 #define PL_multi_open (PL_parser->multi_open)
53 #define PL_multi_close (PL_parser->multi_close)
54 #define PL_pending_ident (PL_parser->pending_ident)
55 #define PL_preambled (PL_parser->preambled)
56 #define PL_sublex_info (PL_parser->sublex_info)
57 #define PL_linestr (PL_parser->linestr)
58 #define PL_expect (PL_parser->expect)
59 #define PL_copline (PL_parser->copline)
60 #define PL_bufptr (PL_parser->bufptr)
61 #define PL_oldbufptr (PL_parser->oldbufptr)
62 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
63 #define PL_linestart (PL_parser->linestart)
64 #define PL_bufend (PL_parser->bufend)
65 #define PL_last_uni (PL_parser->last_uni)
66 #define PL_last_lop (PL_parser->last_lop)
67 #define PL_last_lop_op (PL_parser->last_lop_op)
68 #define PL_lex_state (PL_parser->lex_state)
69 #define PL_rsfp (PL_parser->rsfp)
70 #define PL_rsfp_filters (PL_parser->rsfp_filters)
71 #define PL_in_my (PL_parser->in_my)
72 #define PL_in_my_stash (PL_parser->in_my_stash)
73 #define PL_tokenbuf (PL_parser->tokenbuf)
74 #define PL_multi_end (PL_parser->multi_end)
75 #define PL_error_count (PL_parser->error_count)
78 # define PL_endwhite (PL_parser->endwhite)
79 # define PL_faketokens (PL_parser->faketokens)
80 # define PL_lasttoke (PL_parser->lasttoke)
81 # define PL_nextwhite (PL_parser->nextwhite)
82 # define PL_realtokenstart (PL_parser->realtokenstart)
83 # define PL_skipwhite (PL_parser->skipwhite)
84 # define PL_thisclose (PL_parser->thisclose)
85 # define PL_thismad (PL_parser->thismad)
86 # define PL_thisopen (PL_parser->thisopen)
87 # define PL_thisstuff (PL_parser->thisstuff)
88 # define PL_thistoken (PL_parser->thistoken)
89 # define PL_thiswhite (PL_parser->thiswhite)
90 # define PL_thiswhite (PL_parser->thiswhite)
91 # define PL_nexttoke (PL_parser->nexttoke)
92 # define PL_curforce (PL_parser->curforce)
94 # define PL_nexttoke (PL_parser->nexttoke)
95 # define PL_nexttype (PL_parser->nexttype)
96 # define PL_nextval (PL_parser->nextval)
100 S_pending_ident(pTHX);
102 static const char ident_too_long[] = "Identifier too long";
103 static const char commaless_variable_list[] = "comma-less variable list";
105 #ifndef PERL_NO_UTF16_FILTER
106 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
107 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
111 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
112 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
114 # define CURMAD(slot,sv)
115 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
118 #define XFAKEBRACK 128
119 #define XENUMMASK 127
121 #ifdef USE_UTF8_SCRIPTS
122 # define UTF (!IN_BYTES)
124 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
127 /* In variables named $^X, these are the legal values for X.
128 * 1999-02-27 mjd-perl-patch@plover.com */
129 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
131 /* On MacOS, respect nonbreaking spaces */
132 #ifdef MACOS_TRADITIONAL
133 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
135 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
138 /* LEX_* are values for PL_lex_state, the state of the lexer.
139 * They are arranged oddly so that the guard on the switch statement
140 * can get by with a single comparison (if the compiler is smart enough).
143 /* #define LEX_NOTPARSING 11 is done in perl.h. */
145 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
146 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
147 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
148 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
149 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
151 /* at end of code, eg "$x" followed by: */
152 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
153 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
155 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
156 string or after \E, $foo, etc */
157 #define LEX_INTERPCONST 2 /* NOT USED */
158 #define LEX_FORMLINE 1 /* expecting a format line */
159 #define LEX_KNOWNEXT 0 /* next token known; just return it */
163 static const char* const lex_state_names[] = {
182 #include "keywords.h"
184 /* CLINE is a macro that ensures PL_copline has a sane value */
189 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
192 # define SKIPSPACE0(s) skipspace0(s)
193 # define SKIPSPACE1(s) skipspace1(s)
194 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
195 # define PEEKSPACE(s) skipspace2(s,0)
197 # define SKIPSPACE0(s) skipspace(s)
198 # define SKIPSPACE1(s) skipspace(s)
199 # define SKIPSPACE2(s,tsv) skipspace(s)
200 # define PEEKSPACE(s) skipspace(s)
204 * Convenience functions to return different tokens and prime the
205 * lexer for the next token. They all take an argument.
207 * TOKEN : generic token (used for '(', DOLSHARP, etc)
208 * OPERATOR : generic operator
209 * AOPERATOR : assignment operator
210 * PREBLOCK : beginning the block after an if, while, foreach, ...
211 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
212 * PREREF : *EXPR where EXPR is not a simple identifier
213 * TERM : expression term
214 * LOOPX : loop exiting command (goto, last, dump, etc)
215 * FTST : file test operator
216 * FUN0 : zero-argument function
217 * FUN1 : not used, except for not, which isn't a UNIOP
218 * BOop : bitwise or or xor
220 * SHop : shift operator
221 * PWop : power operator
222 * PMop : pattern-matching operator
223 * Aop : addition-level operator
224 * Mop : multiplication-level operator
225 * Eop : equality-testing operator
226 * Rop : relational operator <= != gt
228 * Also see LOP and lop() below.
231 #ifdef DEBUGGING /* Serve -DT. */
232 # define REPORT(retval) tokereport((I32)retval, &pl_yylval)
234 # define REPORT(retval) (retval)
237 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
238 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
239 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
240 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
241 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
242 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
243 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
244 #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
245 #define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
246 #define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
247 #define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
248 #define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
249 #define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
250 #define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
251 #define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
252 #define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
253 #define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
254 #define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
255 #define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
256 #define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
258 /* This bit of chicanery makes a unary function followed by
259 * a parenthesis into a function with one argument, highest precedence.
260 * The UNIDOR macro is for unary functions that can be followed by the //
261 * operator (such as C<shift // 0>).
263 #define UNI2(f,x) { \
264 pl_yylval.ival = f; \
267 PL_last_uni = PL_oldbufptr; \
268 PL_last_lop_op = f; \
270 return REPORT( (int)FUNC1 ); \
272 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
274 #define UNI(f) UNI2(f,XTERM)
275 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
277 #define UNIBRACK(f) { \
278 pl_yylval.ival = f; \
280 PL_last_uni = PL_oldbufptr; \
282 return REPORT( (int)FUNC1 ); \
284 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
287 /* grandfather return to old style */
288 #define OLDLOP(f) return(pl_yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
292 /* how to interpret the pl_yylval associated with the token */
296 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
302 static struct debug_tokens {
304 enum token_type type;
306 } const debug_tokens[] =
308 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
309 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
310 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
311 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
312 { ARROW, TOKENTYPE_NONE, "ARROW" },
313 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
314 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
315 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
316 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
317 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
318 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
319 { DO, TOKENTYPE_NONE, "DO" },
320 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
321 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
322 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
323 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
324 { ELSE, TOKENTYPE_NONE, "ELSE" },
325 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
326 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
327 { FOR, TOKENTYPE_IVAL, "FOR" },
328 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
329 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
330 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
331 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
332 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
333 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
334 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
335 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
336 { IF, TOKENTYPE_IVAL, "IF" },
337 { LABEL, TOKENTYPE_PVAL, "LABEL" },
338 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
339 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
340 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
341 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
342 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
343 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
344 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
345 { MY, TOKENTYPE_IVAL, "MY" },
346 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
347 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
348 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
349 { OROP, TOKENTYPE_IVAL, "OROP" },
350 { OROR, TOKENTYPE_NONE, "OROR" },
351 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
352 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
353 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
354 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
355 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
356 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
357 { PREINC, TOKENTYPE_NONE, "PREINC" },
358 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
359 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
360 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
361 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
362 { SUB, TOKENTYPE_NONE, "SUB" },
363 { THING, TOKENTYPE_OPVAL, "THING" },
364 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
365 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
366 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
367 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
368 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
369 { USE, TOKENTYPE_IVAL, "USE" },
370 { WHEN, TOKENTYPE_IVAL, "WHEN" },
371 { WHILE, TOKENTYPE_IVAL, "WHILE" },
372 { WORD, TOKENTYPE_OPVAL, "WORD" },
373 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
374 { 0, TOKENTYPE_NONE, NULL }
377 /* dump the returned token in rv, plus any optional arg in pl_yylval */
380 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
384 PERL_ARGS_ASSERT_TOKEREPORT;
387 const char *name = NULL;
388 enum token_type type = TOKENTYPE_NONE;
389 const struct debug_tokens *p;
390 SV* const report = newSVpvs("<== ");
392 for (p = debug_tokens; p->token; p++) {
393 if (p->token == (int)rv) {
400 Perl_sv_catpv(aTHX_ report, name);
401 else if ((char)rv > ' ' && (char)rv < '~')
402 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
404 sv_catpvs(report, "EOF");
406 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
409 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
412 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
414 case TOKENTYPE_OPNUM:
415 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
416 PL_op_name[lvalp->ival]);
419 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
421 case TOKENTYPE_OPVAL:
423 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
424 PL_op_name[lvalp->opval->op_type]);
425 if (lvalp->opval->op_type == OP_CONST) {
426 Perl_sv_catpvf(aTHX_ report, " %s",
427 SvPEEK(cSVOPx_sv(lvalp->opval)));
432 sv_catpvs(report, "(opval=null)");
435 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
441 /* print the buffer with suitable escapes */
444 S_printbuf(pTHX_ const char *const fmt, const char *const s)
446 SV* const tmp = newSVpvs("");
448 PERL_ARGS_ASSERT_PRINTBUF;
450 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
459 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
460 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
464 S_ao(pTHX_ int toketype)
467 if (*PL_bufptr == '=') {
469 if (toketype == ANDAND)
470 pl_yylval.ival = OP_ANDASSIGN;
471 else if (toketype == OROR)
472 pl_yylval.ival = OP_ORASSIGN;
473 else if (toketype == DORDOR)
474 pl_yylval.ival = OP_DORASSIGN;
482 * When Perl expects an operator and finds something else, no_op
483 * prints the warning. It always prints "<something> found where
484 * operator expected. It prints "Missing semicolon on previous line?"
485 * if the surprise occurs at the start of the line. "do you need to
486 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
487 * where the compiler doesn't know if foo is a method call or a function.
488 * It prints "Missing operator before end of line" if there's nothing
489 * after the missing operator, or "... before <...>" if there is something
490 * after the missing operator.
494 S_no_op(pTHX_ const char *const what, char *s)
497 char * const oldbp = PL_bufptr;
498 const bool is_first = (PL_oldbufptr == PL_linestart);
500 PERL_ARGS_ASSERT_NO_OP;
506 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
507 if (ckWARN_d(WARN_SYNTAX)) {
509 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
510 "\t(Missing semicolon on previous line?)\n");
511 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
513 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
515 if (t < PL_bufptr && isSPACE(*t))
516 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
517 "\t(Do you need to predeclare %.*s?)\n",
518 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
522 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
523 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
531 * Complain about missing quote/regexp/heredoc terminator.
532 * If it's called with NULL then it cauterizes the line buffer.
533 * If we're in a delimited string and the delimiter is a control
534 * character, it's reformatted into a two-char sequence like ^C.
539 S_missingterm(pTHX_ char *s)
545 char * const nl = strrchr(s,'\n');
549 else if (isCNTRL(PL_multi_close)) {
551 tmpbuf[1] = (char)toCTRL(PL_multi_close);
556 *tmpbuf = (char)PL_multi_close;
560 q = strchr(s,'"') ? '\'' : '"';
561 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
564 #define FEATURE_IS_ENABLED(name) \
565 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
566 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
567 /* The longest string we pass in. */
568 #define MAX_FEATURE_LEN (sizeof("switch")-1)
571 * S_feature_is_enabled
572 * Check whether the named feature is enabled.
575 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
578 HV * const hinthv = GvHV(PL_hintgv);
579 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
581 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
583 assert(namelen <= MAX_FEATURE_LEN);
584 memcpy(&he_name[8], name, namelen);
586 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
594 Perl_deprecate(pTHX_ const char *const s)
596 PERL_ARGS_ASSERT_DEPRECATE;
598 if (ckWARN(WARN_DEPRECATED))
599 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
603 Perl_deprecate_old(pTHX_ const char *const s)
605 /* This function should NOT be called for any new deprecated warnings */
606 /* Use Perl_deprecate instead */
608 /* It is here to maintain backward compatibility with the pre-5.8 */
609 /* warnings category hierarchy. The "deprecated" category used to */
610 /* live under the "syntax" category. It is now a top-level category */
611 /* in its own right. */
613 PERL_ARGS_ASSERT_DEPRECATE_OLD;
615 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
616 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
617 "Use of %s is deprecated", s);
621 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
622 * utf16-to-utf8-reversed.
625 #ifdef PERL_CR_FILTER
629 register const char *s = SvPVX_const(sv);
630 register const char * const e = s + SvCUR(sv);
632 PERL_ARGS_ASSERT_STRIP_RETURN;
634 /* outer loop optimized to do nothing if there are no CR-LFs */
636 if (*s++ == '\r' && *s == '\n') {
637 /* hit a CR-LF, need to copy the rest */
638 register char *d = s - 1;
641 if (*s == '\r' && s[1] == '\n')
652 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
654 const I32 count = FILTER_READ(idx+1, sv, maxlen);
655 if (count > 0 && !maxlen)
666 * Create a parser object and initialise its parser and lexer fields
668 * rsfp is the opened file handle to read from (if any),
670 * line holds any initial content already read from the file (or in
671 * the case of no file, such as an eval, the whole contents);
673 * new_filter indicates that this is a new file and it shouldn't inherit
674 * the filters from the current parser (ie require).
678 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
681 const char *s = NULL;
683 yy_parser *parser, *oparser;
685 /* create and initialise a parser */
687 Newxz(parser, 1, yy_parser);
688 parser->old_parser = oparser = PL_parser;
691 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
692 parser->ps = parser->stack;
693 parser->stack_size = YYINITDEPTH;
695 parser->stack->state = 0;
696 parser->yyerrstatus = 0;
697 parser->yychar = YYEMPTY; /* Cause a token to be read. */
699 /* on scope exit, free this parser and restore any outer one */
701 parser->saved_curcop = PL_curcop;
703 /* initialise lexer state */
706 parser->curforce = -1;
708 parser->nexttoke = 0;
710 parser->error_count = oparser ? oparser->error_count : 0;
711 parser->copline = NOLINE;
712 parser->lex_state = LEX_NORMAL;
713 parser->expect = XSTATE;
715 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
716 : MUTABLE_AV(SvREFCNT_inc(oparser->rsfp_filters));
718 Newx(parser->lex_brackstack, 120, char);
719 Newx(parser->lex_casestack, 12, char);
720 *parser->lex_casestack = '\0';
723 s = SvPV_const(line, len);
729 parser->linestr = newSVpvs("\n;");
730 } else if (SvREADONLY(line) || s[len-1] != ';') {
731 parser->linestr = newSVsv(line);
733 sv_catpvs(parser->linestr, "\n;");
736 SvREFCNT_inc_simple_void_NN(line);
737 parser->linestr = line;
739 parser->oldoldbufptr =
742 parser->linestart = SvPVX(parser->linestr);
743 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
744 parser->last_lop = parser->last_uni = NULL;
748 /* delete a parser object */
751 Perl_parser_free(pTHX_ const yy_parser *parser)
753 PERL_ARGS_ASSERT_PARSER_FREE;
755 PL_curcop = parser->saved_curcop;
756 SvREFCNT_dec(parser->linestr);
758 if (parser->rsfp == PerlIO_stdin())
759 PerlIO_clearerr(parser->rsfp);
760 else if (parser->rsfp && (!parser->old_parser ||
761 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
762 PerlIO_close(parser->rsfp);
763 SvREFCNT_dec(parser->rsfp_filters);
765 Safefree(parser->stack);
766 Safefree(parser->lex_brackstack);
767 Safefree(parser->lex_casestack);
768 PL_parser = parser->old_parser;
775 * Finalizer for lexing operations. Must be called when the parser is
776 * done with the lexer.
783 PL_doextract = FALSE;
788 * This subroutine has nothing to do with tilting, whether at windmills
789 * or pinball tables. Its name is short for "increment line". It
790 * increments the current line number in CopLINE(PL_curcop) and checks
791 * to see whether the line starts with a comment of the form
792 * # line 500 "foo.pm"
793 * If so, it sets the current line number and file to the values in the comment.
797 S_incline(pTHX_ const char *s)
804 PERL_ARGS_ASSERT_INCLINE;
806 CopLINE_inc(PL_curcop);
809 while (SPACE_OR_TAB(*s))
811 if (strnEQ(s, "line", 4))
815 if (SPACE_OR_TAB(*s))
819 while (SPACE_OR_TAB(*s))
827 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
829 while (SPACE_OR_TAB(*s))
831 if (*s == '"' && (t = strchr(s+1, '"'))) {
841 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
843 if (*e != '\n' && *e != '\0')
844 return; /* false alarm */
847 const STRLEN len = t - s;
849 SV *const temp_sv = CopFILESV(PL_curcop);
855 tmplen = SvCUR(temp_sv);
861 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
862 /* must copy *{"::_<(eval N)[oldfilename:L]"}
863 * to *{"::_<newfilename"} */
864 /* However, the long form of evals is only turned on by the
865 debugger - usually they're "(eval %lu)" */
869 STRLEN tmplen2 = len;
870 if (tmplen + 2 <= sizeof smallbuf)
873 Newx(tmpbuf, tmplen + 2, char);
876 memcpy(tmpbuf + 2, cf, tmplen);
878 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
883 if (tmplen2 + 2 <= sizeof smallbuf)
886 Newx(tmpbuf2, tmplen2 + 2, char);
888 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
889 /* Either they malloc'd it, or we malloc'd it,
890 so no prefix is present in ours. */
895 memcpy(tmpbuf2 + 2, s, tmplen2);
898 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
900 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
901 /* adjust ${"::_<newfilename"} to store the new file name */
902 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
903 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
904 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
907 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
909 if (tmpbuf != smallbuf) Safefree(tmpbuf);
912 CopFILE_free(PL_curcop);
913 CopFILE_setn(PL_curcop, s, len);
915 CopLINE_set(PL_curcop, atoi(n)-1);
919 /* skip space before PL_thistoken */
922 S_skipspace0(pTHX_ register char *s)
924 PERL_ARGS_ASSERT_SKIPSPACE0;
931 PL_thiswhite = newSVpvs("");
932 sv_catsv(PL_thiswhite, PL_skipwhite);
933 sv_free(PL_skipwhite);
936 PL_realtokenstart = s - SvPVX(PL_linestr);
940 /* skip space after PL_thistoken */
943 S_skipspace1(pTHX_ register char *s)
945 const char *start = s;
946 I32 startoff = start - SvPVX(PL_linestr);
948 PERL_ARGS_ASSERT_SKIPSPACE1;
953 start = SvPVX(PL_linestr) + startoff;
954 if (!PL_thistoken && PL_realtokenstart >= 0) {
955 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
956 PL_thistoken = newSVpvn(tstart, start - tstart);
958 PL_realtokenstart = -1;
961 PL_nextwhite = newSVpvs("");
962 sv_catsv(PL_nextwhite, PL_skipwhite);
963 sv_free(PL_skipwhite);
970 S_skipspace2(pTHX_ register char *s, SV **svp)
973 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
974 const I32 startoff = s - SvPVX(PL_linestr);
976 PERL_ARGS_ASSERT_SKIPSPACE2;
979 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
980 if (!PL_madskills || !svp)
982 start = SvPVX(PL_linestr) + startoff;
983 if (!PL_thistoken && PL_realtokenstart >= 0) {
984 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
985 PL_thistoken = newSVpvn(tstart, start - tstart);
986 PL_realtokenstart = -1;
991 sv_setsv(*svp, PL_skipwhite);
992 sv_free(PL_skipwhite);
1001 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1003 AV *av = CopFILEAVx(PL_curcop);
1005 SV * const sv = newSV_type(SVt_PVMG);
1007 sv_setsv(sv, orig_sv);
1009 sv_setpvn(sv, buf, len);
1012 av_store(av, (I32)CopLINE(PL_curcop), sv);
1018 * Called to gobble the appropriate amount and type of whitespace.
1019 * Skips comments as well.
1023 S_skipspace(pTHX_ register char *s)
1028 int startoff = s - SvPVX(PL_linestr);
1030 PERL_ARGS_ASSERT_SKIPSPACE;
1033 sv_free(PL_skipwhite);
1037 PERL_ARGS_ASSERT_SKIPSPACE;
1039 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1040 while (s < PL_bufend && SPACE_OR_TAB(*s))
1050 SSize_t oldprevlen, oldoldprevlen;
1051 SSize_t oldloplen = 0, oldunilen = 0;
1052 while (s < PL_bufend && isSPACE(*s)) {
1053 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1058 if (s < PL_bufend && *s == '#') {
1059 while (s < PL_bufend && *s != '\n')
1061 if (s < PL_bufend) {
1063 if (PL_in_eval && !PL_rsfp) {
1070 /* only continue to recharge the buffer if we're at the end
1071 * of the buffer, we're not reading from a source filter, and
1072 * we're in normal lexing mode
1074 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1075 PL_lex_state == LEX_FORMLINE)
1082 /* try to recharge the buffer */
1084 curoff = s - SvPVX(PL_linestr);
1087 if ((s = filter_gets(PL_linestr, PL_rsfp,
1088 (prevlen = SvCUR(PL_linestr)))) == NULL)
1091 if (PL_madskills && curoff != startoff) {
1093 PL_skipwhite = newSVpvs("");
1094 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1098 /* mustn't throw out old stuff yet if madpropping */
1099 SvCUR(PL_linestr) = curoff;
1100 s = SvPVX(PL_linestr) + curoff;
1102 if (curoff && s[-1] == '\n')
1106 /* end of file. Add on the -p or -n magic */
1107 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1110 sv_catpvs(PL_linestr,
1111 ";}continue{print or die qq(-p destination: $!\\n);}");
1113 sv_setpvs(PL_linestr,
1114 ";}continue{print or die qq(-p destination: $!\\n);}");
1116 PL_minus_n = PL_minus_p = 0;
1118 else if (PL_minus_n) {
1120 sv_catpvs(PL_linestr, ";}");
1122 sv_setpvs(PL_linestr, ";}");
1128 sv_catpvs(PL_linestr,";");
1130 sv_setpvs(PL_linestr,";");
1133 /* reset variables for next time we lex */
1134 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1140 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1141 PL_last_lop = PL_last_uni = NULL;
1143 /* Close the filehandle. Could be from
1144 * STDIN, or a regular file. If we were reading code from
1145 * STDIN (because the commandline held no -e or filename)
1146 * then we don't close it, we reset it so the code can
1147 * read from STDIN too.
1150 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1151 PerlIO_clearerr(PL_rsfp);
1153 (void)PerlIO_close(PL_rsfp);
1158 /* not at end of file, so we only read another line */
1159 /* make corresponding updates to old pointers, for yyerror() */
1160 oldprevlen = PL_oldbufptr - PL_bufend;
1161 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1163 oldunilen = PL_last_uni - PL_bufend;
1165 oldloplen = PL_last_lop - PL_bufend;
1166 PL_linestart = PL_bufptr = s + prevlen;
1167 PL_bufend = s + SvCUR(PL_linestr);
1169 PL_oldbufptr = s + oldprevlen;
1170 PL_oldoldbufptr = s + oldoldprevlen;
1172 PL_last_uni = s + oldunilen;
1174 PL_last_lop = s + oldloplen;
1177 /* debugger active and we're not compiling the debugger code,
1178 * so store the line into the debugger's array of lines
1180 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
1181 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1188 PL_skipwhite = newSVpvs("");
1189 curoff = s - SvPVX(PL_linestr);
1190 if (curoff - startoff)
1191 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1200 * Check the unary operators to ensure there's no ambiguity in how they're
1201 * used. An ambiguous piece of code would be:
1203 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1204 * the +5 is its argument.
1214 if (PL_oldoldbufptr != PL_last_uni)
1216 while (isSPACE(*PL_last_uni))
1219 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1221 if ((t = strchr(s, '(')) && t < PL_bufptr)
1224 if (ckWARN_d(WARN_AMBIGUOUS)){
1225 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1226 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1227 (int)(s - PL_last_uni), PL_last_uni);
1232 * LOP : macro to build a list operator. Its behaviour has been replaced
1233 * with a subroutine, S_lop() for which LOP is just another name.
1236 #define LOP(f,x) return lop(f,x,s)
1240 * Build a list operator (or something that might be one). The rules:
1241 * - if we have a next token, then it's a list operator [why?]
1242 * - if the next thing is an opening paren, then it's a function
1243 * - else it's a list operator
1247 S_lop(pTHX_ I32 f, int x, char *s)
1251 PERL_ARGS_ASSERT_LOP;
1257 PL_last_lop = PL_oldbufptr;
1258 PL_last_lop_op = (OPCODE)f;
1261 return REPORT(LSTOP);
1264 return REPORT(LSTOP);
1267 return REPORT(FUNC);
1270 return REPORT(FUNC);
1272 return REPORT(LSTOP);
1278 * Sets up for an eventual force_next(). start_force(0) basically does
1279 * an unshift, while start_force(-1) does a push. yylex removes items
1284 S_start_force(pTHX_ int where)
1288 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1289 where = PL_lasttoke;
1290 assert(PL_curforce < 0 || PL_curforce == where);
1291 if (PL_curforce != where) {
1292 for (i = PL_lasttoke; i > where; --i) {
1293 PL_nexttoke[i] = PL_nexttoke[i-1];
1297 if (PL_curforce < 0) /* in case of duplicate start_force() */
1298 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1299 PL_curforce = where;
1302 curmad('^', newSVpvs(""));
1303 CURMAD('_', PL_nextwhite);
1308 S_curmad(pTHX_ char slot, SV *sv)
1314 if (PL_curforce < 0)
1315 where = &PL_thismad;
1317 where = &PL_nexttoke[PL_curforce].next_mad;
1323 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1325 else if (PL_encoding) {
1326 sv_recode_to_utf8(sv, PL_encoding);
1331 /* keep a slot open for the head of the list? */
1332 if (slot != '_' && *where && (*where)->mad_key == '^') {
1333 (*where)->mad_key = slot;
1334 sv_free(MUTABLE_SV(((*where)->mad_val)));
1335 (*where)->mad_val = (void*)sv;
1338 addmad(newMADsv(slot, sv), where, 0);
1341 # define start_force(where) NOOP
1342 # define curmad(slot, sv) NOOP
1347 * When the lexer realizes it knows the next token (for instance,
1348 * it is reordering tokens for the parser) then it can call S_force_next
1349 * to know what token to return the next time the lexer is called. Caller
1350 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1351 * and possibly PL_expect to ensure the lexer handles the token correctly.
1355 S_force_next(pTHX_ I32 type)
1360 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1361 tokereport(type, &NEXTVAL_NEXTTOKE);
1365 if (PL_curforce < 0)
1366 start_force(PL_lasttoke);
1367 PL_nexttoke[PL_curforce].next_type = type;
1368 if (PL_lex_state != LEX_KNOWNEXT)
1369 PL_lex_defer = PL_lex_state;
1370 PL_lex_state = LEX_KNOWNEXT;
1371 PL_lex_expect = PL_expect;
1374 PL_nexttype[PL_nexttoke] = type;
1376 if (PL_lex_state != LEX_KNOWNEXT) {
1377 PL_lex_defer = PL_lex_state;
1378 PL_lex_expect = PL_expect;
1379 PL_lex_state = LEX_KNOWNEXT;
1385 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1388 SV * const sv = newSVpvn_utf8(start, len,
1390 && is_utf8_string((const U8*)start, len));
1396 * When the lexer knows the next thing is a word (for instance, it has
1397 * just seen -> and it knows that the next char is a word char, then
1398 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1402 * char *start : buffer position (must be within PL_linestr)
1403 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1404 * int check_keyword : if true, Perl checks to make sure the word isn't
1405 * a keyword (do this if the word is a label, e.g. goto FOO)
1406 * int allow_pack : if true, : characters will also be allowed (require,
1407 * use, etc. do this)
1408 * int allow_initial_tick : used by the "sub" lexer only.
1412 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1418 PERL_ARGS_ASSERT_FORCE_WORD;
1420 start = SKIPSPACE1(start);
1422 if (isIDFIRST_lazy_if(s,UTF) ||
1423 (allow_pack && *s == ':') ||
1424 (allow_initial_tick && *s == '\'') )
1426 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1427 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1429 start_force(PL_curforce);
1431 curmad('X', newSVpvn(start,s-start));
1432 if (token == METHOD) {
1437 PL_expect = XOPERATOR;
1441 curmad('g', newSVpvs( "forced" ));
1442 NEXTVAL_NEXTTOKE.opval
1443 = (OP*)newSVOP(OP_CONST,0,
1444 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1445 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1453 * Called when the lexer wants $foo *foo &foo etc, but the program
1454 * text only contains the "foo" portion. The first argument is a pointer
1455 * to the "foo", and the second argument is the type symbol to prefix.
1456 * Forces the next token to be a "WORD".
1457 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1461 S_force_ident(pTHX_ register const char *s, int kind)
1465 PERL_ARGS_ASSERT_FORCE_IDENT;
1468 const STRLEN len = strlen(s);
1469 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1470 start_force(PL_curforce);
1471 NEXTVAL_NEXTTOKE.opval = o;
1474 o->op_private = OPpCONST_ENTERED;
1475 /* XXX see note in pp_entereval() for why we forgo typo
1476 warnings if the symbol must be introduced in an eval.
1478 gv_fetchpvn_flags(s, len,
1479 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1481 kind == '$' ? SVt_PV :
1482 kind == '@' ? SVt_PVAV :
1483 kind == '%' ? SVt_PVHV :
1491 Perl_str_to_version(pTHX_ SV *sv)
1496 const char *start = SvPV_const(sv,len);
1497 const char * const end = start + len;
1498 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1500 PERL_ARGS_ASSERT_STR_TO_VERSION;
1502 while (start < end) {
1506 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1511 retval += ((NV)n)/nshift;
1520 * Forces the next token to be a version number.
1521 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1522 * and if "guessing" is TRUE, then no new token is created (and the caller
1523 * must use an alternative parsing method).
1527 S_force_version(pTHX_ char *s, int guessing)
1533 I32 startoff = s - SvPVX(PL_linestr);
1536 PERL_ARGS_ASSERT_FORCE_VERSION;
1544 while (isDIGIT(*d) || *d == '_' || *d == '.')
1548 start_force(PL_curforce);
1549 curmad('X', newSVpvn(s,d-s));
1552 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1554 s = scan_num(s, &pl_yylval);
1555 version = pl_yylval.opval;
1556 ver = cSVOPx(version)->op_sv;
1557 if (SvPOK(ver) && !SvNIOK(ver)) {
1558 SvUPGRADE(ver, SVt_PVNV);
1559 SvNV_set(ver, str_to_version(ver));
1560 SvNOK_on(ver); /* hint that it is a version */
1563 else if (guessing) {
1566 sv_free(PL_nextwhite); /* let next token collect whitespace */
1568 s = SvPVX(PL_linestr) + startoff;
1576 if (PL_madskills && !version) {
1577 sv_free(PL_nextwhite); /* let next token collect whitespace */
1579 s = SvPVX(PL_linestr) + startoff;
1582 /* NOTE: The parser sees the package name and the VERSION swapped */
1583 start_force(PL_curforce);
1584 NEXTVAL_NEXTTOKE.opval = version;
1592 * Tokenize a quoted string passed in as an SV. It finds the next
1593 * chunk, up to end of string or a backslash. It may make a new
1594 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1599 S_tokeq(pTHX_ SV *sv)
1603 register char *send;
1608 PERL_ARGS_ASSERT_TOKEQ;
1613 s = SvPV_force(sv, len);
1614 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1617 while (s < send && *s != '\\')
1622 if ( PL_hints & HINT_NEW_STRING ) {
1623 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
1627 if (s + 1 < send && (s[1] == '\\'))
1628 s++; /* all that, just for this */
1633 SvCUR_set(sv, d - SvPVX_const(sv));
1635 if ( PL_hints & HINT_NEW_STRING )
1636 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
1641 * Now come three functions related to double-quote context,
1642 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1643 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1644 * interact with PL_lex_state, and create fake ( ... ) argument lists
1645 * to handle functions and concatenation.
1646 * They assume that whoever calls them will be setting up a fake
1647 * join call, because each subthing puts a ',' after it. This lets
1650 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1652 * (I'm not sure whether the spurious commas at the end of lcfirst's
1653 * arguments and join's arguments are created or not).
1658 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1660 * Pattern matching will set PL_lex_op to the pattern-matching op to
1661 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
1663 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1665 * Everything else becomes a FUNC.
1667 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1668 * had an OP_CONST or OP_READLINE). This just sets us up for a
1669 * call to S_sublex_push().
1673 S_sublex_start(pTHX)
1676 register const I32 op_type = pl_yylval.ival;
1678 if (op_type == OP_NULL) {
1679 pl_yylval.opval = PL_lex_op;
1683 if (op_type == OP_CONST || op_type == OP_READLINE) {
1684 SV *sv = tokeq(PL_lex_stuff);
1686 if (SvTYPE(sv) == SVt_PVIV) {
1687 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1689 const char * const p = SvPV_const(sv, len);
1690 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
1694 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1695 PL_lex_stuff = NULL;
1696 /* Allow <FH> // "foo" */
1697 if (op_type == OP_READLINE)
1698 PL_expect = XTERMORDORDOR;
1701 else if (op_type == OP_BACKTICK && PL_lex_op) {
1702 /* readpipe() vas overriden */
1703 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
1704 pl_yylval.opval = PL_lex_op;
1706 PL_lex_stuff = NULL;
1710 PL_sublex_info.super_state = PL_lex_state;
1711 PL_sublex_info.sub_inwhat = (U16)op_type;
1712 PL_sublex_info.sub_op = PL_lex_op;
1713 PL_lex_state = LEX_INTERPPUSH;
1717 pl_yylval.opval = PL_lex_op;
1727 * Create a new scope to save the lexing state. The scope will be
1728 * ended in S_sublex_done. Returns a '(', starting the function arguments
1729 * to the uc, lc, etc. found before.
1730 * Sets PL_lex_state to LEX_INTERPCONCAT.
1739 PL_lex_state = PL_sublex_info.super_state;
1740 SAVEBOOL(PL_lex_dojoin);
1741 SAVEI32(PL_lex_brackets);
1742 SAVEI32(PL_lex_casemods);
1743 SAVEI32(PL_lex_starts);
1744 SAVEI8(PL_lex_state);
1745 SAVEVPTR(PL_lex_inpat);
1746 SAVEI16(PL_lex_inwhat);
1747 SAVECOPLINE(PL_curcop);
1748 SAVEPPTR(PL_bufptr);
1749 SAVEPPTR(PL_bufend);
1750 SAVEPPTR(PL_oldbufptr);
1751 SAVEPPTR(PL_oldoldbufptr);
1752 SAVEPPTR(PL_last_lop);
1753 SAVEPPTR(PL_last_uni);
1754 SAVEPPTR(PL_linestart);
1755 SAVESPTR(PL_linestr);
1756 SAVEGENERICPV(PL_lex_brackstack);
1757 SAVEGENERICPV(PL_lex_casestack);
1759 PL_linestr = PL_lex_stuff;
1760 PL_lex_stuff = NULL;
1762 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1763 = SvPVX(PL_linestr);
1764 PL_bufend += SvCUR(PL_linestr);
1765 PL_last_lop = PL_last_uni = NULL;
1766 SAVEFREESV(PL_linestr);
1768 PL_lex_dojoin = FALSE;
1769 PL_lex_brackets = 0;
1770 Newx(PL_lex_brackstack, 120, char);
1771 Newx(PL_lex_casestack, 12, char);
1772 PL_lex_casemods = 0;
1773 *PL_lex_casestack = '\0';
1775 PL_lex_state = LEX_INTERPCONCAT;
1776 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1778 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1779 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1780 PL_lex_inpat = PL_sublex_info.sub_op;
1782 PL_lex_inpat = NULL;
1789 * Restores lexer state after a S_sublex_push.
1796 if (!PL_lex_starts++) {
1797 SV * const sv = newSVpvs("");
1798 if (SvUTF8(PL_linestr))
1800 PL_expect = XOPERATOR;
1801 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1805 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1806 PL_lex_state = LEX_INTERPCASEMOD;
1810 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1811 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1812 PL_linestr = PL_lex_repl;
1814 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1815 PL_bufend += SvCUR(PL_linestr);
1816 PL_last_lop = PL_last_uni = NULL;
1817 SAVEFREESV(PL_linestr);
1818 PL_lex_dojoin = FALSE;
1819 PL_lex_brackets = 0;
1820 PL_lex_casemods = 0;
1821 *PL_lex_casestack = '\0';
1823 if (SvEVALED(PL_lex_repl)) {
1824 PL_lex_state = LEX_INTERPNORMAL;
1826 /* we don't clear PL_lex_repl here, so that we can check later
1827 whether this is an evalled subst; that means we rely on the
1828 logic to ensure sublex_done() is called again only via the
1829 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1832 PL_lex_state = LEX_INTERPCONCAT;
1842 PL_endwhite = newSVpvs("");
1843 sv_catsv(PL_endwhite, PL_thiswhite);
1847 sv_setpvs(PL_thistoken,"");
1849 PL_realtokenstart = -1;
1853 PL_bufend = SvPVX(PL_linestr);
1854 PL_bufend += SvCUR(PL_linestr);
1855 PL_expect = XOPERATOR;
1856 PL_sublex_info.sub_inwhat = 0;
1864 Extracts a pattern, double-quoted string, or transliteration. This
1867 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1868 processing a pattern (PL_lex_inpat is true), a transliteration
1869 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1871 Returns a pointer to the character scanned up to. If this is
1872 advanced from the start pointer supplied (i.e. if anything was
1873 successfully parsed), will leave an OP for the substring scanned
1874 in pl_yylval. Caller must intuit reason for not parsing further
1875 by looking at the next characters herself.
1879 double-quoted style: \r and \n
1880 regexp special ones: \D \s
1883 case and quoting: \U \Q \E
1884 stops on @ and $, but not for $ as tail anchor
1886 In transliterations:
1887 characters are VERY literal, except for - not at the start or end
1888 of the string, which indicates a range. If the range is in bytes,
1889 scan_const expands the range to the full set of intermediate
1890 characters. If the range is in utf8, the hyphen is replaced with
1891 a certain range mark which will be handled by pmtrans() in op.c.
1893 In double-quoted strings:
1895 double-quoted style: \r and \n
1897 deprecated backrefs: \1 (in substitution replacements)
1898 case and quoting: \U \Q \E
1901 scan_const does *not* construct ops to handle interpolated strings.
1902 It stops processing as soon as it finds an embedded $ or @ variable
1903 and leaves it to the caller to work out what's going on.
1905 embedded arrays (whether in pattern or not) could be:
1906 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1908 $ in double-quoted strings must be the symbol of an embedded scalar.
1910 $ in pattern could be $foo or could be tail anchor. Assumption:
1911 it's a tail anchor if $ is the last thing in the string, or if it's
1912 followed by one of "()| \r\n\t"
1914 \1 (backreferences) are turned into $1
1916 The structure of the code is
1917 while (there's a character to process) {
1918 handle transliteration ranges
1919 skip regexp comments /(?#comment)/ and codes /(?{code})/
1920 skip #-initiated comments in //x patterns
1921 check for embedded arrays
1922 check for embedded scalars
1924 leave intact backslashes from leaveit (below)
1925 deprecate \1 in substitution replacements
1926 handle string-changing backslashes \l \U \Q \E, etc.
1927 switch (what was escaped) {
1928 handle \- in a transliteration (becomes a literal -)
1929 handle \132 (octal characters)
1930 handle \x15 and \x{1234} (hex characters)
1931 handle \N{name} (named characters)
1932 handle \cV (control characters)
1933 handle printf-style backslashes (\f, \r, \n, etc)
1936 } (end if backslash)
1937 handle regular character
1938 } (end while character to read)
1943 S_scan_const(pTHX_ char *start)
1946 register char *send = PL_bufend; /* end of the constant */
1947 SV *sv = newSV(send - start); /* sv for the constant. See
1948 note below on sizing. */
1949 register char *s = start; /* start of the constant */
1950 register char *d = SvPVX(sv); /* destination for copies */
1951 bool dorange = FALSE; /* are we in a translit range? */
1952 bool didrange = FALSE; /* did we just finish a range? */
1953 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1954 I32 this_utf8 = UTF; /* Is the source string assumed
1955 to be UTF8? But, this can
1956 show as true when the source
1957 isn't utf8, as for example
1958 when it is entirely composed
1961 /* Note on sizing: The scanned constant is placed into sv, which is
1962 * initialized by newSV() assuming one byte of output for every byte of
1963 * input. This routine expects newSV() to allocate an extra byte for a
1964 * trailing NUL, which this routine will append if it gets to the end of
1965 * the input. There may be more bytes of input than output (eg., \N{LATIN
1966 * CAPITAL LETTER A}), or more output than input if the constant ends up
1967 * recoded to utf8, but each time a construct is found that might increase
1968 * the needed size, SvGROW() is called. Its size parameter each time is
1969 * based on the best guess estimate at the time, namely the length used so
1970 * far, plus the length the current construct will occupy, plus room for
1971 * the trailing NUL, plus one byte for every input byte still unscanned */
1975 UV literal_endpoint = 0;
1976 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1979 PERL_ARGS_ASSERT_SCAN_CONST;
1981 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1982 /* If we are doing a trans and we know we want UTF8 set expectation */
1983 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1984 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1988 while (s < send || dorange) {
1989 /* get transliterations out of the way (they're most literal) */
1990 if (PL_lex_inwhat == OP_TRANS) {
1991 /* expand a range A-Z to the full set of characters. AIE! */
1993 I32 i; /* current expanded character */
1994 I32 min; /* first character in range */
1995 I32 max; /* last character in range */
2006 char * const c = (char*)utf8_hop((U8*)d, -1);
2010 *c = (char)UTF_TO_NATIVE(0xff);
2011 /* mark the range as done, and continue */
2017 i = d - SvPVX_const(sv); /* remember current offset */
2020 SvLEN(sv) + (has_utf8 ?
2021 (512 - UTF_CONTINUATION_MARK +
2024 /* How many two-byte within 0..255: 128 in UTF-8,
2025 * 96 in UTF-8-mod. */
2027 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
2029 d = SvPVX(sv) + i; /* refresh d after realloc */
2033 for (j = 0; j <= 1; j++) {
2034 char * const c = (char*)utf8_hop((U8*)d, -1);
2035 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2041 max = (U8)0xff; /* only to \xff */
2042 uvmax = uv; /* \x{100} to uvmax */
2044 d = c; /* eat endpoint chars */
2049 d -= 2; /* eat the first char and the - */
2050 min = (U8)*d; /* first char in range */
2051 max = (U8)d[1]; /* last char in range */
2058 "Invalid range \"%c-%c\" in transliteration operator",
2059 (char)min, (char)max);
2063 if (literal_endpoint == 2 &&
2064 ((isLOWER(min) && isLOWER(max)) ||
2065 (isUPPER(min) && isUPPER(max)))) {
2067 for (i = min; i <= max; i++)
2069 *d++ = NATIVE_TO_NEED(has_utf8,i);
2071 for (i = min; i <= max; i++)
2073 *d++ = NATIVE_TO_NEED(has_utf8,i);
2078 for (i = min; i <= max; i++)
2081 const U8 ch = (U8)NATIVE_TO_UTF(i);
2082 if (UNI_IS_INVARIANT(ch))
2085 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2086 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2095 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2097 *d++ = (char)UTF_TO_NATIVE(0xff);
2099 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2103 /* mark the range as done, and continue */
2107 literal_endpoint = 0;
2112 /* range begins (ignore - as first or last char) */
2113 else if (*s == '-' && s+1 < send && s != start) {
2115 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2122 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2132 literal_endpoint = 0;
2133 native_range = TRUE;
2138 /* if we get here, we're not doing a transliteration */
2140 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2141 except for the last char, which will be done separately. */
2142 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2144 while (s+1 < send && *s != ')')
2145 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2147 else if (s[2] == '{' /* This should match regcomp.c */
2148 || (s[2] == '?' && s[3] == '{'))
2151 char *regparse = s + (s[2] == '{' ? 3 : 4);
2154 while (count && (c = *regparse)) {
2155 if (c == '\\' && regparse[1])
2163 if (*regparse != ')')
2164 regparse--; /* Leave one char for continuation. */
2165 while (s < regparse)
2166 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2170 /* likewise skip #-initiated comments in //x patterns */
2171 else if (*s == '#' && PL_lex_inpat &&
2172 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2173 while (s+1 < send && *s != '\n')
2174 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2177 /* check for embedded arrays
2178 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2180 else if (*s == '@' && s[1]) {
2181 if (isALNUM_lazy_if(s+1,UTF))
2183 if (strchr(":'{$", s[1]))
2185 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2186 break; /* in regexp, neither @+ nor @- are interpolated */
2189 /* check for embedded scalars. only stop if we're sure it's a
2192 else if (*s == '$') {
2193 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2195 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
2196 if (s[1] == '\\' && ckWARN(WARN_AMBIGUOUS)) {
2197 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2198 "Possible unintended interpolation of $\\ in regex");
2200 break; /* in regexp, $ might be tail anchor */
2204 /* End of else if chain - OP_TRANS rejoin rest */
2207 if (*s == '\\' && s+1 < send) {
2210 /* deprecate \1 in strings and substitution replacements */
2211 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2212 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2214 if (ckWARN(WARN_SYNTAX))
2215 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2220 /* string-change backslash escapes */
2221 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2225 /* skip any other backslash escapes in a pattern */
2226 else if (PL_lex_inpat) {
2227 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2228 goto default_action;
2231 /* if we get here, it's either a quoted -, or a digit */
2234 /* quoted - in transliterations */
2236 if (PL_lex_inwhat == OP_TRANS) {
2243 if ((isALPHA(*s) || isDIGIT(*s)) &&
2245 Perl_warner(aTHX_ packWARN(WARN_MISC),
2246 "Unrecognized escape \\%c passed through",
2248 /* default action is to copy the quoted character */
2249 goto default_action;
2252 /* eg. \132 indicates the octal constant 0x132 */
2253 case '0': case '1': case '2': case '3':
2254 case '4': case '5': case '6': case '7':
2258 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
2261 goto NUM_ESCAPE_INSERT;
2263 /* eg. \x24 indicates the hex constant 0x24 */
2267 char* const e = strchr(s, '}');
2268 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2269 PERL_SCAN_DISALLOW_PREFIX;
2274 yyerror("Missing right brace on \\x{}");
2278 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
2284 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2285 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
2291 /* Insert oct, hex, or \N{U+...} escaped character. There will
2292 * always be enough room in sv since such escapes will be
2293 * longer than any UTF-8 sequence they can end up as, except if
2294 * they force us to recode the rest of the string into utf8 */
2296 /* Here uv is the ordinal of the next character being added in
2297 * unicode (converted from native). (It has to be done before
2298 * here because \N is interpreted as unicode, and oct and hex
2300 if (!UNI_IS_INVARIANT(uv)) {
2301 if (!has_utf8 && uv > 255) {
2302 /* Might need to recode whatever we have accumulated so
2303 * far if it contains any chars variant in utf8 or
2306 SvCUR_set(sv, d - SvPVX_const(sv));
2309 sv_utf8_upgrade(sv);
2310 /* See Note on sizing above. */
2311 SvGROW(sv, SvCUR(sv) + UNISKIP(uv) + (STRLEN)(send - s) + 1);
2312 d = SvPVX(sv) + SvCUR(sv);
2317 d = (char*)uvuni_to_utf8((U8*)d, uv);
2318 if (PL_lex_inwhat == OP_TRANS &&
2319 PL_sublex_info.sub_op) {
2320 PL_sublex_info.sub_op->op_private |=
2321 (PL_lex_repl ? OPpTRANS_FROM_UTF
2325 if (uv > 255 && !dorange)
2326 native_range = FALSE;
2338 /* \N{LATIN SMALL LETTER A} is a named character, and so is
2343 char* e = strchr(s, '}');
2349 yyerror("Missing right brace on \\N{}");
2353 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2354 /* \N{U+...} The ... is a unicode value even on EBCDIC
2356 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2357 PERL_SCAN_DISALLOW_PREFIX;
2360 uv = grok_hex(s, &len, &flags, NULL);
2361 if ( e > s && len != (STRLEN)(e - s) ) {
2365 goto NUM_ESCAPE_INSERT;
2367 res = newSVpvn(s + 1, e - s - 1);
2368 res = new_constant( NULL, 0, "charnames",
2369 res, NULL, s - 2, e - s + 3 );
2371 sv_utf8_upgrade(res);
2372 str = SvPV_const(res,len);
2373 #ifdef EBCDIC_NEVER_MIND
2374 /* charnames uses pack U and that has been
2375 * recently changed to do the below uni->native
2376 * mapping, so this would be redundant (and wrong,
2377 * the code point would be doubly converted).
2378 * But leave this in just in case the pack U change
2379 * gets revoked, but the semantics is still
2380 * desireable for charnames. --jhi */
2382 UV uv = utf8_to_uvchr((const U8*)str, 0);
2385 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2387 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2388 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2389 str = SvPV_const(res, len);
2393 /* If destination is not in utf8 but this new character is,
2394 * recode the dest to utf8 */
2395 if (!has_utf8 && SvUTF8(res)) {
2396 SvCUR_set(sv, d - SvPVX_const(sv));
2399 sv_utf8_upgrade(sv);
2400 /* See Note on sizing above. */
2401 SvGROW(sv, SvCUR(sv) + len + (STRLEN)(send - s) + 1);
2402 d = SvPVX(sv) + SvCUR(sv);
2404 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2406 /* See Note on sizing above. (NOTE: SvCUR() is not set
2407 * correctly here). */
2408 const STRLEN off = d - SvPVX_const(sv);
2409 d = SvGROW(sv, off + len + (STRLEN)(send - s) + 1) + off;
2413 native_range = FALSE; /* \N{} is guessed to be Unicode */
2415 Copy(str, d, len, char);
2422 yyerror("Missing braces on \\N{}");
2425 /* \c is a control character */
2434 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2437 yyerror("Missing control char name in \\c");
2441 /* printf-style backslashes, formfeeds, newlines, etc */
2443 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2446 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2449 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2452 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2455 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2458 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2461 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2467 } /* end if (backslash) */
2474 /* If we started with encoded form, or already know we want it,
2475 then encode the next character */
2476 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
2480 /* One might think that it is wasted effort in the case of the
2481 * source being utf8 (this_utf8 == TRUE) to take the next character
2482 * in the source, convert it to an unsigned value, and then convert
2483 * it back again. But the source has not been validated here. The
2484 * routine that does the conversion checks for errors like
2487 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2488 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2490 SvCUR_set(sv, d - SvPVX_const(sv));
2493 sv_utf8_upgrade(sv);
2495 /* See Note on sizing above. */
2496 SvGROW(sv, SvCUR(sv) + need + (STRLEN)(send - s) + 1);
2497 d = SvPVX(sv) + SvCUR(sv);
2499 } else if (need > len) {
2500 /* encoded value larger than old, may need extra space (NOTE:
2501 * SvCUR() is not set correctly here). See Note on sizing
2503 const STRLEN off = d - SvPVX_const(sv);
2504 d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
2508 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2510 if (uv > 255 && !dorange)
2511 native_range = FALSE;
2515 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2517 } /* while loop to process each character */
2519 /* terminate the string and set up the sv */
2521 SvCUR_set(sv, d - SvPVX_const(sv));
2522 if (SvCUR(sv) >= SvLEN(sv))
2523 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2526 if (PL_encoding && !has_utf8) {
2527 sv_recode_to_utf8(sv, PL_encoding);
2533 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2534 PL_sublex_info.sub_op->op_private |=
2535 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2539 /* shrink the sv if we allocated more than we used */
2540 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2541 SvPV_shrink_to_cur(sv);
2544 /* return the substring (via pl_yylval) only if we parsed anything */
2545 if (s > PL_bufptr) {
2546 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2547 const char *const key = PL_lex_inpat ? "qr" : "q";
2548 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2552 if (PL_lex_inwhat == OP_TRANS) {
2555 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2563 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2566 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2573 * Returns TRUE if there's more to the expression (e.g., a subscript),
2576 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2578 * ->[ and ->{ return TRUE
2579 * { and [ outside a pattern are always subscripts, so return TRUE
2580 * if we're outside a pattern and it's not { or [, then return FALSE
2581 * if we're in a pattern and the first char is a {
2582 * {4,5} (any digits around the comma) returns FALSE
2583 * if we're in a pattern and the first char is a [
2585 * [SOMETHING] has a funky algorithm to decide whether it's a
2586 * character class or not. It has to deal with things like
2587 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2588 * anything else returns TRUE
2591 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2594 S_intuit_more(pTHX_ register char *s)
2598 PERL_ARGS_ASSERT_INTUIT_MORE;
2600 if (PL_lex_brackets)
2602 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2604 if (*s != '{' && *s != '[')
2609 /* In a pattern, so maybe we have {n,m}. */
2626 /* On the other hand, maybe we have a character class */
2629 if (*s == ']' || *s == '^')
2632 /* this is terrifying, and it works */
2633 int weight = 2; /* let's weigh the evidence */
2635 unsigned char un_char = 255, last_un_char;
2636 const char * const send = strchr(s,']');
2637 char tmpbuf[sizeof PL_tokenbuf * 4];
2639 if (!send) /* has to be an expression */
2642 Zero(seen,256,char);
2645 else if (isDIGIT(*s)) {
2647 if (isDIGIT(s[1]) && s[2] == ']')
2653 for (; s < send; s++) {
2654 last_un_char = un_char;
2655 un_char = (unsigned char)*s;
2660 weight -= seen[un_char] * 10;
2661 if (isALNUM_lazy_if(s+1,UTF)) {
2663 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2664 len = (int)strlen(tmpbuf);
2665 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2670 else if (*s == '$' && s[1] &&
2671 strchr("[#!%*<>()-=",s[1])) {
2672 if (/*{*/ strchr("])} =",s[2]))
2681 if (strchr("wds]",s[1]))
2683 else if (seen[(U8)'\''] || seen[(U8)'"'])
2685 else if (strchr("rnftbxcav",s[1]))
2687 else if (isDIGIT(s[1])) {
2689 while (s[1] && isDIGIT(s[1]))
2699 if (strchr("aA01! ",last_un_char))
2701 if (strchr("zZ79~",s[1]))
2703 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2704 weight -= 5; /* cope with negative subscript */
2707 if (!isALNUM(last_un_char)
2708 && !(last_un_char == '$' || last_un_char == '@'
2709 || last_un_char == '&')
2710 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2715 if (keyword(tmpbuf, d - tmpbuf, 0))
2718 if (un_char == last_un_char + 1)
2720 weight -= seen[un_char];
2725 if (weight >= 0) /* probably a character class */
2735 * Does all the checking to disambiguate
2737 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2738 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2740 * First argument is the stuff after the first token, e.g. "bar".
2742 * Not a method if bar is a filehandle.
2743 * Not a method if foo is a subroutine prototyped to take a filehandle.
2744 * Not a method if it's really "Foo $bar"
2745 * Method if it's "foo $bar"
2746 * Not a method if it's really "print foo $bar"
2747 * Method if it's really "foo package::" (interpreted as package->foo)
2748 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2749 * Not a method if bar is a filehandle or package, but is quoted with
2754 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2757 char *s = start + (*start == '$');
2758 char tmpbuf[sizeof PL_tokenbuf];
2765 PERL_ARGS_ASSERT_INTUIT_METHOD;
2768 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2772 const char *proto = SvPVX_const(cv);
2783 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2784 /* start is the beginning of the possible filehandle/object,
2785 * and s is the end of it
2786 * tmpbuf is a copy of it
2789 if (*start == '$') {
2790 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2791 isUPPER(*PL_tokenbuf))
2794 len = start - SvPVX(PL_linestr);
2798 start = SvPVX(PL_linestr) + len;
2802 return *s == '(' ? FUNCMETH : METHOD;
2804 if (!keyword(tmpbuf, len, 0)) {
2805 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2809 soff = s - SvPVX(PL_linestr);
2813 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2814 if (indirgv && GvCVu(indirgv))
2816 /* filehandle or package name makes it a method */
2817 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2819 soff = s - SvPVX(PL_linestr);
2822 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2823 return 0; /* no assumptions -- "=>" quotes bearword */
2825 start_force(PL_curforce);
2826 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2827 newSVpvn(tmpbuf,len));
2828 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2830 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2835 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2837 return *s == '(' ? FUNCMETH : METHOD;
2843 /* Encoded script support. filter_add() effectively inserts a
2844 * 'pre-processing' function into the current source input stream.
2845 * Note that the filter function only applies to the current source file
2846 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2848 * The datasv parameter (which may be NULL) can be used to pass
2849 * private data to this instance of the filter. The filter function
2850 * can recover the SV using the FILTER_DATA macro and use it to
2851 * store private buffers and state information.
2853 * The supplied datasv parameter is upgraded to a PVIO type
2854 * and the IoDIRP/IoANY field is used to store the function pointer,
2855 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2856 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2857 * private use must be set using malloc'd pointers.
2861 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2870 if (!PL_rsfp_filters)
2871 PL_rsfp_filters = newAV();
2874 SvUPGRADE(datasv, SVt_PVIO);
2875 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2876 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2877 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2878 FPTR2DPTR(void *, IoANY(datasv)),
2879 SvPV_nolen(datasv)));
2880 av_unshift(PL_rsfp_filters, 1);
2881 av_store(PL_rsfp_filters, 0, datasv) ;
2886 /* Delete most recently added instance of this filter function. */
2888 Perl_filter_del(pTHX_ filter_t funcp)
2893 PERL_ARGS_ASSERT_FILTER_DEL;
2896 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2897 FPTR2DPTR(void*, funcp)));
2899 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2901 /* if filter is on top of stack (usual case) just pop it off */
2902 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2903 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2904 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2905 IoANY(datasv) = (void *)NULL;
2906 sv_free(av_pop(PL_rsfp_filters));
2910 /* we need to search for the correct entry and clear it */
2911 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2915 /* Invoke the idxth filter function for the current rsfp. */
2916 /* maxlen 0 = read one text line */
2918 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2923 /* This API is bad. It should have been using unsigned int for maxlen.
2924 Not sure if we want to change the API, but if not we should sanity
2925 check the value here. */
2926 const unsigned int correct_length
2935 PERL_ARGS_ASSERT_FILTER_READ;
2937 if (!PL_parser || !PL_rsfp_filters)
2939 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2940 /* Provide a default input filter to make life easy. */
2941 /* Note that we append to the line. This is handy. */
2942 DEBUG_P(PerlIO_printf(Perl_debug_log,
2943 "filter_read %d: from rsfp\n", idx));
2944 if (correct_length) {
2947 const int old_len = SvCUR(buf_sv);
2949 /* ensure buf_sv is large enough */
2950 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2951 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2952 correct_length)) <= 0) {
2953 if (PerlIO_error(PL_rsfp))
2954 return -1; /* error */
2956 return 0 ; /* end of file */
2958 SvCUR_set(buf_sv, old_len + len) ;
2961 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2962 if (PerlIO_error(PL_rsfp))
2963 return -1; /* error */
2965 return 0 ; /* end of file */
2968 return SvCUR(buf_sv);
2970 /* Skip this filter slot if filter has been deleted */
2971 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2972 DEBUG_P(PerlIO_printf(Perl_debug_log,
2973 "filter_read %d: skipped (filter deleted)\n",
2975 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2977 /* Get function pointer hidden within datasv */
2978 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2979 DEBUG_P(PerlIO_printf(Perl_debug_log,
2980 "filter_read %d: via function %p (%s)\n",
2981 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2982 /* Call function. The function is expected to */
2983 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2984 /* Return: <0:error, =0:eof, >0:not eof */
2985 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2989 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2993 PERL_ARGS_ASSERT_FILTER_GETS;
2995 #ifdef PERL_CR_FILTER
2996 if (!PL_rsfp_filters) {
2997 filter_add(S_cr_textfilter,NULL);
3000 if (PL_rsfp_filters) {
3002 SvCUR_set(sv, 0); /* start with empty line */
3003 if (FILTER_READ(0, sv, 0) > 0)
3004 return ( SvPVX(sv) ) ;
3009 return (sv_gets(sv, fp, append));
3013 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
3018 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
3020 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
3024 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
3025 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3027 return GvHV(gv); /* Foo:: */
3030 /* use constant CLASS => 'MyClass' */
3031 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3032 if (gv && GvCV(gv)) {
3033 SV * const sv = cv_const_sv(GvCV(gv));
3035 pkgname = SvPV_const(sv, len);
3038 return gv_stashpvn(pkgname, len, 0);
3042 * S_readpipe_override
3043 * Check whether readpipe() is overriden, and generates the appropriate
3044 * optree, provided sublex_start() is called afterwards.
3047 S_readpipe_override(pTHX)
3050 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3051 pl_yylval.ival = OP_BACKTICK;
3053 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3055 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3056 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3057 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3059 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3060 append_elem(OP_LIST,
3061 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3062 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3069 * The intent of this yylex wrapper is to minimize the changes to the
3070 * tokener when we aren't interested in collecting madprops. It remains
3071 * to be seen how successful this strategy will be...
3078 char *s = PL_bufptr;
3080 /* make sure PL_thiswhite is initialized */
3084 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3085 if (PL_pending_ident)
3086 return S_pending_ident(aTHX);
3088 /* previous token ate up our whitespace? */
3089 if (!PL_lasttoke && PL_nextwhite) {
3090 PL_thiswhite = PL_nextwhite;
3094 /* isolate the token, and figure out where it is without whitespace */
3095 PL_realtokenstart = -1;
3099 assert(PL_curforce < 0);
3101 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3102 if (!PL_thistoken) {
3103 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3104 PL_thistoken = newSVpvs("");
3106 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3107 PL_thistoken = newSVpvn(tstart, s - tstart);
3110 if (PL_thismad) /* install head */
3111 CURMAD('X', PL_thistoken);
3114 /* last whitespace of a sublex? */
3115 if (optype == ')' && PL_endwhite) {
3116 CURMAD('X', PL_endwhite);
3121 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3122 if (!PL_thiswhite && !PL_endwhite && !optype) {
3123 sv_free(PL_thistoken);
3128 /* put off final whitespace till peg */
3129 if (optype == ';' && !PL_rsfp) {
3130 PL_nextwhite = PL_thiswhite;
3133 else if (PL_thisopen) {
3134 CURMAD('q', PL_thisopen);
3136 sv_free(PL_thistoken);
3140 /* Store actual token text as madprop X */
3141 CURMAD('X', PL_thistoken);
3145 /* add preceding whitespace as madprop _ */
3146 CURMAD('_', PL_thiswhite);
3150 /* add quoted material as madprop = */
3151 CURMAD('=', PL_thisstuff);
3155 /* add terminating quote as madprop Q */
3156 CURMAD('Q', PL_thisclose);
3160 /* special processing based on optype */
3164 /* opval doesn't need a TOKEN since it can already store mp */
3174 if (pl_yylval.opval)
3175 append_madprops(PL_thismad, pl_yylval.opval, 0);
3183 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3192 /* remember any fake bracket that lexer is about to discard */
3193 if (PL_lex_brackets == 1 &&
3194 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3197 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3200 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3201 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3204 break; /* don't bother looking for trailing comment */
3213 /* attach a trailing comment to its statement instead of next token */
3217 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3219 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3221 if (*s == '\n' || *s == '#') {
3222 while (s < PL_bufend && *s != '\n')
3226 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3227 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3244 /* Create new token struct. Note: opvals return early above. */
3245 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3252 S_tokenize_use(pTHX_ int is_use, char *s) {
3255 PERL_ARGS_ASSERT_TOKENIZE_USE;
3257 if (PL_expect != XSTATE)
3258 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3259 is_use ? "use" : "no"));
3261 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3262 s = force_version(s, TRUE);
3263 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3264 start_force(PL_curforce);
3265 NEXTVAL_NEXTTOKE.opval = NULL;
3268 else if (*s == 'v') {
3269 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3270 s = force_version(s, FALSE);
3274 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3275 s = force_version(s, FALSE);
3277 pl_yylval.ival = is_use;
3281 static const char* const exp_name[] =
3282 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3283 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3290 Works out what to call the token just pulled out of the input
3291 stream. The yacc parser takes care of taking the ops we return and
3292 stitching them into a tree.
3298 if read an identifier
3299 if we're in a my declaration
3300 croak if they tried to say my($foo::bar)
3301 build the ops for a my() declaration
3302 if it's an access to a my() variable
3303 are we in a sort block?
3304 croak if my($a); $a <=> $b
3305 build ops for access to a my() variable
3306 if in a dq string, and they've said @foo and we can't find @foo
3308 build ops for a bareword
3309 if we already built the token before, use it.
3314 #pragma segment Perl_yylex
3320 register char *s = PL_bufptr;
3325 /* orig_keyword, gvp, and gv are initialized here because
3326 * jump to the label just_a_word_zero can bypass their
3327 * initialization later. */
3328 I32 orig_keyword = 0;
3333 SV* tmp = newSVpvs("");
3334 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3335 (IV)CopLINE(PL_curcop),
3336 lex_state_names[PL_lex_state],
3337 exp_name[PL_expect],
3338 pv_display(tmp, s, strlen(s), 0, 60));
3341 /* check if there's an identifier for us to look at */
3342 if (PL_pending_ident)
3343 return REPORT(S_pending_ident(aTHX));
3345 /* no identifier pending identification */
3347 switch (PL_lex_state) {
3349 case LEX_NORMAL: /* Some compilers will produce faster */
3350 case LEX_INTERPNORMAL: /* code if we comment these out. */
3354 /* when we've already built the next token, just pull it out of the queue */
3358 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3360 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3361 PL_nexttoke[PL_lasttoke].next_mad = 0;
3362 if (PL_thismad && PL_thismad->mad_key == '_') {
3363 PL_thiswhite = MUTABLE_SV(PL_thismad->mad_val);
3364 PL_thismad->mad_val = 0;
3365 mad_free(PL_thismad);
3370 PL_lex_state = PL_lex_defer;
3371 PL_expect = PL_lex_expect;
3372 PL_lex_defer = LEX_NORMAL;
3373 if (!PL_nexttoke[PL_lasttoke].next_type)
3378 pl_yylval = PL_nextval[PL_nexttoke];
3380 PL_lex_state = PL_lex_defer;
3381 PL_expect = PL_lex_expect;
3382 PL_lex_defer = LEX_NORMAL;
3386 /* FIXME - can these be merged? */
3387 return(PL_nexttoke[PL_lasttoke].next_type);
3389 return REPORT(PL_nexttype[PL_nexttoke]);
3392 /* interpolated case modifiers like \L \U, including \Q and \E.
3393 when we get here, PL_bufptr is at the \
3395 case LEX_INTERPCASEMOD:
3397 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3398 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3400 /* handle \E or end of string */
3401 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3403 if (PL_lex_casemods) {
3404 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3405 PL_lex_casestack[PL_lex_casemods] = '\0';
3407 if (PL_bufptr != PL_bufend
3408 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3410 PL_lex_state = LEX_INTERPCONCAT;
3413 PL_thistoken = newSVpvs("\\E");
3419 while (PL_bufptr != PL_bufend &&
3420 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3422 PL_thiswhite = newSVpvs("");
3423 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3427 if (PL_bufptr != PL_bufend)
3430 PL_lex_state = LEX_INTERPCONCAT;
3434 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3435 "### Saw case modifier\n"); });
3437 if (s[1] == '\\' && s[2] == 'E') {
3440 PL_thiswhite = newSVpvs("");
3441 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3444 PL_lex_state = LEX_INTERPCONCAT;
3449 if (!PL_madskills) /* when just compiling don't need correct */
3450 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3451 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3452 if ((*s == 'L' || *s == 'U') &&
3453 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3454 PL_lex_casestack[--PL_lex_casemods] = '\0';
3457 if (PL_lex_casemods > 10)
3458 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3459 PL_lex_casestack[PL_lex_casemods++] = *s;
3460 PL_lex_casestack[PL_lex_casemods] = '\0';
3461 PL_lex_state = LEX_INTERPCONCAT;
3462 start_force(PL_curforce);
3463 NEXTVAL_NEXTTOKE.ival = 0;
3465 start_force(PL_curforce);
3467 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3469 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3471 NEXTVAL_NEXTTOKE.ival = OP_LC;
3473 NEXTVAL_NEXTTOKE.ival = OP_UC;
3475 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3477 Perl_croak(aTHX_ "panic: yylex");
3479 SV* const tmpsv = newSVpvs("\\ ");
3480 /* replace the space with the character we want to escape
3482 SvPVX(tmpsv)[1] = *s;
3488 if (PL_lex_starts) {
3494 sv_free(PL_thistoken);
3495 PL_thistoken = newSVpvs("");
3498 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3499 if (PL_lex_casemods == 1 && PL_lex_inpat)
3508 case LEX_INTERPPUSH:
3509 return REPORT(sublex_push());
3511 case LEX_INTERPSTART:
3512 if (PL_bufptr == PL_bufend)
3513 return REPORT(sublex_done());
3514 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3515 "### Interpolated variable\n"); });
3517 PL_lex_dojoin = (*PL_bufptr == '@');
3518 PL_lex_state = LEX_INTERPNORMAL;
3519 if (PL_lex_dojoin) {
3520 start_force(PL_curforce);
3521 NEXTVAL_NEXTTOKE.ival = 0;
3523 start_force(PL_curforce);
3524 force_ident("\"", '$');
3525 start_force(PL_curforce);
3526 NEXTVAL_NEXTTOKE.ival = 0;
3528 start_force(PL_curforce);
3529 NEXTVAL_NEXTTOKE.ival = 0;
3531 start_force(PL_curforce);
3532 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3535 if (PL_lex_starts++) {
3540 sv_free(PL_thistoken);
3541 PL_thistoken = newSVpvs("");
3544 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3545 if (!PL_lex_casemods && PL_lex_inpat)
3552 case LEX_INTERPENDMAYBE:
3553 if (intuit_more(PL_bufptr)) {
3554 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3560 if (PL_lex_dojoin) {
3561 PL_lex_dojoin = FALSE;
3562 PL_lex_state = LEX_INTERPCONCAT;
3566 sv_free(PL_thistoken);
3567 PL_thistoken = newSVpvs("");
3572 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3573 && SvEVALED(PL_lex_repl))
3575 if (PL_bufptr != PL_bufend)
3576 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3580 case LEX_INTERPCONCAT:
3582 if (PL_lex_brackets)
3583 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3585 if (PL_bufptr == PL_bufend)
3586 return REPORT(sublex_done());
3588 if (SvIVX(PL_linestr) == '\'') {
3589 SV *sv = newSVsv(PL_linestr);
3592 else if ( PL_hints & HINT_NEW_RE )
3593 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3594 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3598 s = scan_const(PL_bufptr);
3600 PL_lex_state = LEX_INTERPCASEMOD;
3602 PL_lex_state = LEX_INTERPSTART;
3605 if (s != PL_bufptr) {
3606 start_force(PL_curforce);
3608 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3610 NEXTVAL_NEXTTOKE = pl_yylval;
3613 if (PL_lex_starts++) {
3617 sv_free(PL_thistoken);
3618 PL_thistoken = newSVpvs("");
3621 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3622 if (!PL_lex_casemods && PL_lex_inpat)
3635 PL_lex_state = LEX_NORMAL;
3636 s = scan_formline(PL_bufptr);
3637 if (!PL_lex_formbrack)
3643 PL_oldoldbufptr = PL_oldbufptr;
3649 sv_free(PL_thistoken);
3652 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3656 if (isIDFIRST_lazy_if(s,UTF))
3658 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3659 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3662 goto fake_eof; /* emulate EOF on ^D or ^Z */
3671 if (PL_lex_brackets) {
3672 yyerror((const char *)
3674 ? "Format not terminated"
3675 : "Missing right curly or square bracket"));
3677 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3678 "### Tokener got EOF\n");
3682 if (s++ < PL_bufend)
3683 goto retry; /* ignore stray nulls */
3686 if (!PL_in_eval && !PL_preambled) {
3687 PL_preambled = TRUE;
3693 /* Generate a string of Perl code to load the debugger.
3694 * If PERL5DB is set, it will return the contents of that,
3695 * otherwise a compile-time require of perl5db.pl. */
3697 const char * const pdb = PerlEnv_getenv("PERL5DB");
3700 sv_setpv(PL_linestr, pdb);
3701 sv_catpvs(PL_linestr,";");
3703 SETERRNO(0,SS_NORMAL);
3704 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3707 sv_setpvs(PL_linestr,"");
3708 if (PL_preambleav) {
3709 SV **svp = AvARRAY(PL_preambleav);
3710 SV **const end = svp + AvFILLp(PL_preambleav);
3712 sv_catsv(PL_linestr, *svp);
3714 sv_catpvs(PL_linestr, ";");
3716 sv_free(MUTABLE_SV(PL_preambleav));
3717 PL_preambleav = NULL;
3720 sv_catpvs(PL_linestr,
3721 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3722 if (PL_minus_n || PL_minus_p) {
3723 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3725 sv_catpvs(PL_linestr,"chomp;");
3728 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3729 || *PL_splitstr == '"')
3730 && strchr(PL_splitstr + 1, *PL_splitstr))
3731 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3733 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3734 bytes can be used as quoting characters. :-) */
3735 const char *splits = PL_splitstr;
3736 sv_catpvs(PL_linestr, "our @F=split(q\0");
3739 if (*splits == '\\')
3740 sv_catpvn(PL_linestr, splits, 1);
3741 sv_catpvn(PL_linestr, splits, 1);
3742 } while (*splits++);
3743 /* This loop will embed the trailing NUL of
3744 PL_linestr as the last thing it does before
3746 sv_catpvs(PL_linestr, ");");
3750 sv_catpvs(PL_linestr,"our @F=split(' ');");
3753 sv_catpvs(PL_linestr, "\n");
3754 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3755 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3756 PL_last_lop = PL_last_uni = NULL;
3757 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3758 update_debugger_info(PL_linestr, NULL, 0);
3762 bof = PL_rsfp ? TRUE : FALSE;
3763 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3766 PL_realtokenstart = -1;
3769 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3770 PerlIO_clearerr(PL_rsfp);
3772 (void)PerlIO_close(PL_rsfp);
3774 PL_doextract = FALSE;
3776 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3782 sv_setpvs(PL_linestr, ";}continue{print;}");
3784 sv_setpvs(PL_linestr, ";}");
3785 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3786 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3787 PL_last_lop = PL_last_uni = NULL;
3788 PL_minus_n = PL_minus_p = 0;
3791 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3792 PL_last_lop = PL_last_uni = NULL;
3793 sv_setpvs(PL_linestr,"");
3794 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3796 /* If it looks like the start of a BOM or raw UTF-16,
3797 * check if it in fact is. */
3803 #ifdef PERLIO_IS_STDIO
3804 # ifdef __GNU_LIBRARY__
3805 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3806 # define FTELL_FOR_PIPE_IS_BROKEN
3810 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3811 # define FTELL_FOR_PIPE_IS_BROKEN
3816 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3818 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3819 s = swallow_bom((U8*)s);
3823 /* Incest with pod. */
3826 sv_catsv(PL_thiswhite, PL_linestr);
3828 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3829 sv_setpvs(PL_linestr, "");
3830 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3831 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3832 PL_last_lop = PL_last_uni = NULL;
3833 PL_doextract = FALSE;
3837 } while (PL_doextract);
3838 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3839 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3840 update_debugger_info(PL_linestr, NULL, 0);
3841 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3842 PL_last_lop = PL_last_uni = NULL;
3843 if (CopLINE(PL_curcop) == 1) {
3844 while (s < PL_bufend && isSPACE(*s))
3846 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3850 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3854 if (*s == '#' && *(s+1) == '!')
3856 #ifdef ALTERNATE_SHEBANG
3858 static char const as[] = ALTERNATE_SHEBANG;
3859 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3860 d = s + (sizeof(as) - 1);
3862 #endif /* ALTERNATE_SHEBANG */
3871 while (*d && !isSPACE(*d))
3875 #ifdef ARG_ZERO_IS_SCRIPT
3876 if (ipathend > ipath) {
3878 * HP-UX (at least) sets argv[0] to the script name,
3879 * which makes $^X incorrect. And Digital UNIX and Linux,
3880 * at least, set argv[0] to the basename of the Perl
3881 * interpreter. So, having found "#!", we'll set it right.
3883 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3885 assert(SvPOK(x) || SvGMAGICAL(x));
3886 if (sv_eq(x, CopFILESV(PL_curcop))) {
3887 sv_setpvn(x, ipath, ipathend - ipath);
3893 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3894 const char * const lstart = SvPV_const(x,llen);
3896 bstart += blen - llen;
3897 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3898 sv_setpvn(x, ipath, ipathend - ipath);
3903 TAINT_NOT; /* $^X is always tainted, but that's OK */
3905 #endif /* ARG_ZERO_IS_SCRIPT */
3910 d = instr(s,"perl -");
3912 d = instr(s,"perl");
3914 /* avoid getting into infinite loops when shebang
3915 * line contains "Perl" rather than "perl" */
3917 for (d = ipathend-4; d >= ipath; --d) {
3918 if ((*d == 'p' || *d == 'P')
3919 && !ibcmp(d, "perl", 4))
3929 #ifdef ALTERNATE_SHEBANG
3931 * If the ALTERNATE_SHEBANG on this system starts with a
3932 * character that can be part of a Perl expression, then if
3933 * we see it but not "perl", we're probably looking at the
3934 * start of Perl code, not a request to hand off to some
3935 * other interpreter. Similarly, if "perl" is there, but
3936 * not in the first 'word' of the line, we assume the line
3937 * contains the start of the Perl program.
3939 if (d && *s != '#') {
3940 const char *c = ipath;
3941 while (*c && !strchr("; \t\r\n\f\v#", *c))
3944 d = NULL; /* "perl" not in first word; ignore */
3946 *s = '#'; /* Don't try to parse shebang line */
3948 #endif /* ALTERNATE_SHEBANG */
3949 #ifndef MACOS_TRADITIONAL
3954 !instr(s,"indir") &&
3955 instr(PL_origargv[0],"perl"))
3962 while (s < PL_bufend && isSPACE(*s))
3964 if (s < PL_bufend) {
3965 Newxz(newargv,PL_origargc+3,char*);
3967 while (s < PL_bufend && !isSPACE(*s))
3970 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3973 newargv = PL_origargv;
3976 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3978 Perl_croak(aTHX_ "Can't exec %s", ipath);
3982 while (*d && !isSPACE(*d))
3984 while (SPACE_OR_TAB(*d))
3988 const bool switches_done = PL_doswitches;
3989 const U32 oldpdb = PL_perldb;
3990 const bool oldn = PL_minus_n;
3991 const bool oldp = PL_minus_p;
3995 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3996 const char * const m = d1;
3997 while (*d1 && !isSPACE(*d1))
3999 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
4002 d1 = moreswitches(d1);
4004 if (PL_doswitches && !switches_done) {
4005 int argc = PL_origargc;
4006 char **argv = PL_origargv;
4009 } while (argc && argv[0][0] == '-' && argv[0][1]);
4010 init_argv_symbols(argc,argv);
4012 if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
4013 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
4014 /* if we have already added "LINE: while (<>) {",
4015 we must not do it again */
4017 sv_setpvs(PL_linestr, "");
4018 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4019 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4020 PL_last_lop = PL_last_uni = NULL;
4021 PL_preambled = FALSE;
4022 if (PERLDB_LINE || PERLDB_SAVESRC)
4023 (void)gv_fetchfile(PL_origfilename);
4030 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4032 PL_lex_state = LEX_FORMLINE;
4037 #ifdef PERL_STRICT_CR
4038 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4040 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4042 case ' ': case '\t': case '\f': case 013:
4043 #ifdef MACOS_TRADITIONAL
4047 PL_realtokenstart = -1;
4049 PL_thiswhite = newSVpvs("");
4050 sv_catpvn(PL_thiswhite, s, 1);
4057 PL_realtokenstart = -1;
4061 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4062 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4063 /* handle eval qq[#line 1 "foo"\n ...] */
4064 CopLINE_dec(PL_curcop);
4067 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4069 if (!PL_in_eval || PL_rsfp)
4074 while (d < PL_bufend && *d != '\n')
4078 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4079 Perl_croak(aTHX_ "panic: input overflow");
4082 PL_thiswhite = newSVpvn(s, d - s);
4087 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4089 PL_lex_state = LEX_FORMLINE;
4095 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4096 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4099 TOKEN(PEG); /* make sure any #! line is accessible */
4104 /* if (PL_madskills && PL_lex_formbrack) { */
4106 while (d < PL_bufend && *d != '\n')
4110 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4111 Perl_croak(aTHX_ "panic: input overflow");
4112 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4114 PL_thiswhite = newSVpvs("");
4115 if (CopLINE(PL_curcop) == 1) {
4116 sv_setpvs(PL_thiswhite, "");
4119 sv_catpvn(PL_thiswhite, s, d - s);
4133 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4141 while (s < PL_bufend && SPACE_OR_TAB(*s))
4144 if (strnEQ(s,"=>",2)) {
4145 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4146 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4147 OPERATOR('-'); /* unary minus */
4149 PL_last_uni = PL_oldbufptr;
4151 case 'r': ftst = OP_FTEREAD; break;
4152 case 'w': ftst = OP_FTEWRITE; break;
4153 case 'x': ftst = OP_FTEEXEC; break;
4154 case 'o': ftst = OP_FTEOWNED; break;
4155 case 'R': ftst = OP_FTRREAD; break;
4156 case 'W': ftst = OP_FTRWRITE; break;
4157 case 'X': ftst = OP_FTREXEC; break;
4158 case 'O': ftst = OP_FTROWNED; break;
4159 case 'e': ftst = OP_FTIS; break;
4160 case 'z': ftst = OP_FTZERO; break;
4161 case 's': ftst = OP_FTSIZE; break;
4162 case 'f': ftst = OP_FTFILE; break;
4163 case 'd': ftst = OP_FTDIR; break;
4164 case 'l': ftst = OP_FTLINK; break;
4165 case 'p': ftst = OP_FTPIPE; break;
4166 case 'S': ftst = OP_FTSOCK; break;
4167 case 'u': ftst = OP_FTSUID; break;
4168 case 'g': ftst = OP_FTSGID; break;
4169 case 'k': ftst = OP_FTSVTX; break;
4170 case 'b': ftst = OP_FTBLK; break;
4171 case 'c': ftst = OP_FTCHR; break;
4172 case 't': ftst = OP_FTTTY; break;
4173 case 'T': ftst = OP_FTTEXT; break;
4174 case 'B': ftst = OP_FTBINARY; break;
4175 case 'M': case 'A': case 'C':
4176 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4178 case 'M': ftst = OP_FTMTIME; break;
4179 case 'A': ftst = OP_FTATIME; break;
4180 case 'C': ftst = OP_FTCTIME; break;
4188 PL_last_lop_op = (OPCODE)ftst;
4189 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4190 "### Saw file test %c\n", (int)tmp);
4195 /* Assume it was a minus followed by a one-letter named
4196 * subroutine call (or a -bareword), then. */
4197 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4198 "### '-%c' looked like a file test but was not\n",
4205 const char tmp = *s++;
4208 if (PL_expect == XOPERATOR)
4213 else if (*s == '>') {
4216 if (isIDFIRST_lazy_if(s,UTF)) {
4217 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4225 if (PL_expect == XOPERATOR)
4228 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4230 OPERATOR('-'); /* unary minus */
4236 const char tmp = *s++;
4239 if (PL_expect == XOPERATOR)
4244 if (PL_expect == XOPERATOR)
4247 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4254 if (PL_expect != XOPERATOR) {
4255 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4256 PL_expect = XOPERATOR;
4257 force_ident(PL_tokenbuf, '*');
4270 if (PL_expect == XOPERATOR) {
4274 PL_tokenbuf[0] = '%';
4275 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4276 sizeof PL_tokenbuf - 1, FALSE);
4277 if (!PL_tokenbuf[1]) {
4280 PL_pending_ident = '%';
4291 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4298 const char tmp = *s++;
4304 goto just_a_word_zero_gv;
4307 switch (PL_expect) {
4313 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4315 PL_bufptr = s; /* update in case we back off */
4321 PL_expect = XTERMBLOCK;
4324 stuffstart = s - SvPVX(PL_linestr) - 1;
4328 while (isIDFIRST_lazy_if(s,UTF)) {
4331 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4332 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4333 if (tmp < 0) tmp = -tmp;
4347 sv = newSVpvn(s, len);
4349 d = scan_str(d,TRUE,TRUE);
4351 /* MUST advance bufptr here to avoid bogus
4352 "at end of line" context messages from yyerror().
4354 PL_bufptr = s + len;
4355 yyerror("Unterminated attribute parameter in attribute list");
4359 return REPORT(0); /* EOF indicator */
4363 sv_catsv(sv, PL_lex_stuff);
4364 attrs = append_elem(OP_LIST, attrs,
4365 newSVOP(OP_CONST, 0, sv));
4366 SvREFCNT_dec(PL_lex_stuff);
4367 PL_lex_stuff = NULL;
4370 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4372 if (PL_in_my == KEY_our) {
4374 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4376 /* skip to avoid loading attributes.pm */
4378 deprecate(":unique");
4381 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4384 /* NOTE: any CV attrs applied here need to be part of
4385 the CVf_BUILTIN_ATTRS define in cv.h! */
4386 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4388 CvLVALUE_on(PL_compcv);
4390 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4392 CvLOCKED_on(PL_compcv);
4394 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4396 CvMETHOD_on(PL_compcv);
4398 /* After we've set the flags, it could be argued that
4399 we don't need to do the attributes.pm-based setting
4400 process, and shouldn't bother appending recognized
4401 flags. To experiment with that, uncomment the
4402 following "else". (Note that's already been
4403 uncommented. That keeps the above-applied built-in
4404 attributes from being intercepted (and possibly
4405 rejected) by a package's attribute routines, but is
4406 justified by the performance win for the common case
4407 of applying only built-in attributes.) */
4409 attrs = append_elem(OP_LIST, attrs,
4410 newSVOP(OP_CONST, 0,
4414 if (*s == ':' && s[1] != ':')
4417 break; /* require real whitespace or :'s */
4418 /* XXX losing whitespace on sequential attributes here */
4422 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4423 if (*s != ';' && *s != '}' && *s != tmp
4424 && (tmp != '=' || *s != ')')) {
4425 const char q = ((*s == '\'') ? '"' : '\'');
4426 /* If here for an expression, and parsed no attrs, back
4428 if (tmp == '=' && !attrs) {
4432 /* MUST advance bufptr here to avoid bogus "at end of line"
4433 context messages from yyerror().
4436 yyerror( (const char *)
4438 ? Perl_form(aTHX_ "Invalid separator character "
4439 "%c%c%c in attribute list", q, *s, q)
4440 : "Unterminated attribute list" ) );
4448 start_force(PL_curforce);
4449 NEXTVAL_NEXTTOKE.opval = attrs;
4450 CURMAD('_', PL_nextwhite);
4455 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4456 (s - SvPVX(PL_linestr)) - stuffstart);
4464 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4465 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4473 const char tmp = *s++;
4478 const char tmp = *s++;
4486 if (PL_lex_brackets <= 0)
4487 yyerror("Unmatched right square bracket");
4490 if (PL_lex_state == LEX_INTERPNORMAL) {
4491 if (PL_lex_brackets == 0) {
4492 if (*s == '-' && s[1] == '>')
4493 PL_lex_state = LEX_INTERPENDMAYBE;
4494 else if (*s != '[' && *s != '{')
4495 PL_lex_state = LEX_INTERPEND;
4502 if (PL_lex_brackets > 100) {
4503 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4505 switch (PL_expect) {
4507 if (PL_lex_formbrack) {
4511 if (PL_oldoldbufptr == PL_last_lop)
4512 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4514 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4515 OPERATOR(HASHBRACK);
4517 while (s < PL_bufend && SPACE_OR_TAB(*s))
4520 PL_tokenbuf[0] = '\0';
4521 if (d < PL_bufend && *d == '-') {
4522 PL_tokenbuf[0] = '-';
4524 while (d < PL_bufend && SPACE_OR_TAB(*d))
4527 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4528 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4530 while (d < PL_bufend && SPACE_OR_TAB(*d))
4533 const char minus = (PL_tokenbuf[0] == '-');
4534 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4542 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4547 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4552 if (PL_oldoldbufptr == PL_last_lop)
4553 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4555 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4558 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4560 /* This hack is to get the ${} in the message. */
4562 yyerror("syntax error");
4565 OPERATOR(HASHBRACK);
4567 /* This hack serves to disambiguate a pair of curlies
4568 * as being a block or an anon hash. Normally, expectation
4569 * determines that, but in cases where we're not in a
4570 * position to expect anything in particular (like inside
4571 * eval"") we have to resolve the ambiguity. This code
4572 * covers the case where the first term in the curlies is a
4573 * quoted string. Most other cases need to be explicitly
4574 * disambiguated by prepending a "+" before the opening
4575 * curly in order to force resolution as an anon hash.
4577 * XXX should probably propagate the outer expectation
4578 * into eval"" to rely less on this hack, but that could
4579 * potentially break current behavior of eval"".
4583 if (*s == '\'' || *s == '"' || *s == '`') {
4584 /* common case: get past first string, handling escapes */
4585 for (t++; t < PL_bufend && *t != *s;)
4586 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4590 else if (*s == 'q') {
4593 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4596 /* skip q//-like construct */
4598 char open, close, term;
4601 while (t < PL_bufend && isSPACE(*t))
4603 /* check for q => */
4604 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4605 OPERATOR(HASHBRACK);
4609 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4613 for (t++; t < PL_bufend; t++) {
4614 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4616 else if (*t == open)
4620 for (t++; t < PL_bufend; t++) {
4621 if (*t == '\\' && t+1 < PL_bufend)
4623 else if (*t == close && --brackets <= 0)
4625 else if (*t == open)
4632 /* skip plain q word */
4633 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4636 else if (isALNUM_lazy_if(t,UTF)) {
4638 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4641 while (t < PL_bufend && isSPACE(*t))
4643 /* if comma follows first term, call it an anon hash */
4644 /* XXX it could be a comma expression with loop modifiers */
4645 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4646 || (*t == '=' && t[1] == '>')))
4647 OPERATOR(HASHBRACK);
4648 if (PL_expect == XREF)
4651 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4657 pl_yylval.ival = CopLINE(PL_curcop);
4658 if (isSPACE(*s) || *s == '#')
4659 PL_copline = NOLINE; /* invalidate current command line number */
4664 if (PL_lex_brackets <= 0)
4665 yyerror("Unmatched right curly bracket");
4667 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4668 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4669 PL_lex_formbrack = 0;
4670 if (PL_lex_state == LEX_INTERPNORMAL) {
4671 if (PL_lex_brackets == 0) {
4672 if (PL_expect & XFAKEBRACK) {
4673 PL_expect &= XENUMMASK;
4674 PL_lex_state = LEX_INTERPEND;
4679 PL_thiswhite = newSVpvs("");
4680 sv_catpvs(PL_thiswhite,"}");
4683 return yylex(); /* ignore fake brackets */
4685 if (*s == '-' && s[1] == '>')
4686 PL_lex_state = LEX_INTERPENDMAYBE;
4687 else if (*s != '[' && *s != '{')
4688 PL_lex_state = LEX_INTERPEND;
4691 if (PL_expect & XFAKEBRACK) {
4692 PL_expect &= XENUMMASK;
4694 return yylex(); /* ignore fake brackets */
4696 start_force(PL_curforce);
4698 curmad('X', newSVpvn(s-1,1));
4699 CURMAD('_', PL_thiswhite);
4704 PL_thistoken = newSVpvs("");
4712 if (PL_expect == XOPERATOR) {
4713 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4714 && isIDFIRST_lazy_if(s,UTF))
4716 CopLINE_dec(PL_curcop);
4717 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
4718 CopLINE_inc(PL_curcop);
4723 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4725 PL_expect = XOPERATOR;
4726 force_ident(PL_tokenbuf, '&');
4730 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4742 const char tmp = *s++;
4749 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4750 && strchr("+-*/%.^&|<",tmp))
4751 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4752 "Reversed %c= operator",(int)tmp);
4754 if (PL_expect == XSTATE && isALPHA(tmp) &&
4755 (s == PL_linestart+1 || s[-2] == '\n') )
4757 if (PL_in_eval && !PL_rsfp) {
4762 if (strnEQ(s,"=cut",4)) {
4778 PL_thiswhite = newSVpvs("");
4779 sv_catpvn(PL_thiswhite, PL_linestart,
4780 PL_bufend - PL_linestart);
4784 PL_doextract = TRUE;
4788 if (PL_lex_brackets < PL_lex_formbrack) {
4790 #ifdef PERL_STRICT_CR
4791 while (SPACE_OR_TAB(*t))
4793 while (SPACE_OR_TAB(*t) || *t == '\r')
4796 if (*t == '\n' || *t == '#') {
4805 if (PL_expect == XSTATE && s[1] == '!' && s[2] == '!') {
4811 const char tmp = *s++;
4813 /* was this !=~ where !~ was meant?
4814 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4816 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4817 const char *t = s+1;
4819 while (t < PL_bufend && isSPACE(*t))
4822 if (*t == '/' || *t == '?' ||
4823 ((*t == 'm' || *t == 's' || *t == 'y')
4824 && !isALNUM(t[1])) ||
4825 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4826 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4827 "!=~ should be !~");
4837 if (PL_expect != XOPERATOR) {
4838 if (s[1] != '<' && !strchr(s,'>'))
4841 s = scan_heredoc(s);
4843 s = scan_inputsymbol(s);
4844 TERM(sublex_start());
4850 SHop(OP_LEFT_SHIFT);
4864 const char tmp = *s++;
4866 SHop(OP_RIGHT_SHIFT);
4867 else if (tmp == '=')
4876 if (PL_expect == XOPERATOR) {
4877 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4879 deprecate_old(commaless_variable_list);
4880 return REPORT(','); /* grandfather non-comma-format format */
4884 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4885 PL_tokenbuf[0] = '@';
4886 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4887 sizeof PL_tokenbuf - 1, FALSE);
4888 if (PL_expect == XOPERATOR)
4889 no_op("Array length", s);
4890 if (!PL_tokenbuf[1])
4892 PL_expect = XOPERATOR;
4893 PL_pending_ident = '#';
4897 PL_tokenbuf[0] = '$';
4898 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4899 sizeof PL_tokenbuf - 1, FALSE);
4900 if (PL_expect == XOPERATOR)
4902 if (!PL_tokenbuf[1]) {
4904 yyerror("Final $ should be \\$ or $name");
4908 /* This kludge not intended to be bulletproof. */
4909 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4910 pl_yylval.opval = newSVOP(OP_CONST, 0,
4911 newSViv(CopARYBASE_get(&PL_compiling)));
4912 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4918 const char tmp = *s;
4919 if (PL_lex_state == LEX_NORMAL)
4922 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4923 && intuit_more(s)) {
4925 PL_tokenbuf[0] = '@';
4926 if (ckWARN(WARN_SYNTAX)) {
4929 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4932 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4933 while (t < PL_bufend && *t != ']')
4935 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4936 "Multidimensional syntax %.*s not supported",
4937 (int)((t - PL_bufptr) + 1), PL_bufptr);
4941 else if (*s == '{') {
4943 PL_tokenbuf[0] = '%';
4944 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4945 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4947 char tmpbuf[sizeof PL_tokenbuf];
4950 } while (isSPACE(*t));
4951 if (isIDFIRST_lazy_if(t,UTF)) {
4953 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4957 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4958 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4959 "You need to quote \"%s\"",
4966 PL_expect = XOPERATOR;
4967 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4968 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4969 if (!islop || PL_last_lop_op == OP_GREPSTART)
4970 PL_expect = XOPERATOR;
4971 else if (strchr("$@\"'`q", *s))
4972 PL_expect = XTERM; /* e.g. print $fh "foo" */
4973 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4974 PL_expect = XTERM; /* e.g. print $fh &sub */
4975 else if (isIDFIRST_lazy_if(s,UTF)) {
4976 char tmpbuf[sizeof PL_tokenbuf];
4978 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4979 if ((t2 = keyword(tmpbuf, len, 0))) {
4980 /* binary operators exclude handle interpretations */
4992 PL_expect = XTERM; /* e.g. print $fh length() */
4997 PL_expect = XTERM; /* e.g. print $fh subr() */
5000 else if (isDIGIT(*s))
5001 PL_expect = XTERM; /* e.g. print $fh 3 */
5002 else if (*s == '.' && isDIGIT(s[1]))
5003 PL_expect = XTERM; /* e.g. print $fh .3 */
5004 else if ((*s == '?' || *s == '-' || *s == '+')
5005 && !isSPACE(s[1]) && s[1] != '=')
5006 PL_expect = XTERM; /* e.g. print $fh -1 */
5007 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
5009 PL_expect = XTERM; /* e.g. print $fh /.../
5010 XXX except DORDOR operator
5012 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
5014 PL_expect = XTERM; /* print $fh <<"EOF" */
5017 PL_pending_ident = '$';
5021 if (PL_expect == XOPERATOR)
5023 PL_tokenbuf[0] = '@';
5024 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
5025 if (!PL_tokenbuf[1]) {
5028 if (PL_lex_state == LEX_NORMAL)
5030 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5032 PL_tokenbuf[0] = '%';
5034 /* Warn about @ where they meant $. */
5035 if (*s == '[' || *s == '{') {
5036 if (ckWARN(WARN_SYNTAX)) {
5037 const char *t = s + 1;
5038 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5040 if (*t == '}' || *t == ']') {
5042 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5043 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5044 "Scalar value %.*s better written as $%.*s",
5045 (int)(t-PL_bufptr), PL_bufptr,
5046 (int)(t-PL_bufptr-1), PL_bufptr+1);
5051 PL_pending_ident = '@';
5054 case '/': /* may be division, defined-or, or pattern */
5055 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5059 case '?': /* may either be conditional or pattern */
5060 if (PL_expect == XSTATE && s[1] == '?' && s[2] == '?') {
5064 if (PL_expect == XOPERATOR) {
5072 /* A // operator. */
5082 /* Disable warning on "study /blah/" */
5083 if (PL_oldoldbufptr == PL_last_uni
5084 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5085 || memNE(PL_last_uni, "study", 5)
5086 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5089 s = scan_pat(s,OP_MATCH);
5090 TERM(sublex_start());
5094 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5095 #ifdef PERL_STRICT_CR
5098 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5100 && (s == PL_linestart || s[-1] == '\n') )
5102 PL_lex_formbrack = 0;
5106 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5110 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5116 pl_yylval.ival = OPf_SPECIAL;
5122 if (PL_expect != XOPERATOR)
5127 case '0': case '1': case '2': case '3': case '4':
5128 case '5': case '6': case '7': case '8': case '9':
5129 s = scan_num(s, &pl_yylval);
5130 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5131 if (PL_expect == XOPERATOR)
5136 s = scan_str(s,!!PL_madskills,FALSE);
5137 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5138 if (PL_expect == XOPERATOR) {
5139 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5141 deprecate_old(commaless_variable_list);
5142 return REPORT(','); /* grandfather non-comma-format format */
5149 pl_yylval.ival = OP_CONST;
5150 TERM(sublex_start());
5153 s = scan_str(s,!!PL_madskills,FALSE);
5154 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5155 if (PL_expect == XOPERATOR) {
5156 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5158 deprecate_old(commaless_variable_list);
5159 return REPORT(','); /* grandfather non-comma-format format */
5166 pl_yylval.ival = OP_CONST;
5167 /* FIXME. I think that this can be const if char *d is replaced by
5168 more localised variables. */
5169 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5170 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5171 pl_yylval.ival = OP_STRINGIFY;
5175 TERM(sublex_start());
5178 s = scan_str(s,!!PL_madskills,FALSE);
5179 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5180 if (PL_expect == XOPERATOR)
5181 no_op("Backticks",s);
5184 readpipe_override();
5185 TERM(sublex_start());
5189 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5190 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5192 if (PL_expect == XOPERATOR)
5193 no_op("Backslash",s);
5197 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5198 char *start = s + 2;
5199 while (isDIGIT(*start) || *start == '_')
5201 if (*start == '.' && isDIGIT(start[1])) {
5202 s = scan_num(s, &pl_yylval);
5205 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5206 else if (!isALPHA(*start) && (PL_expect == XTERM
5207 || PL_expect == XREF || PL_expect == XSTATE
5208 || PL_expect == XTERMORDORDOR)) {
5209 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5211 s = scan_num(s, &pl_yylval);
5218 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5260 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5262 /* Some keywords can be followed by any delimiter, including ':' */
5263 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5264 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5265 (PL_tokenbuf[0] == 'q' &&
5266 strchr("qwxr", PL_tokenbuf[1])))));
5268 /* x::* is just a word, unless x is "CORE" */
5269 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5273 while (d < PL_bufend && isSPACE(*d))
5274 d++; /* no comments skipped here, or s### is misparsed */
5276 /* Is this a label? */
5277 if (!tmp && PL_expect == XSTATE
5278 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5280 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5285 /* Check for keywords */
5286 tmp = keyword(PL_tokenbuf, len, 0);
5288 /* Is this a word before a => operator? */
5289 if (*d == '=' && d[1] == '>') {
5292 = (OP*)newSVOP(OP_CONST, 0,
5293 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5294 pl_yylval.opval->op_private = OPpCONST_BARE;
5298 if (tmp < 0) { /* second-class keyword? */
5299 GV *ogv = NULL; /* override (winner) */
5300 GV *hgv = NULL; /* hidden (loser) */
5301 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5303 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5306 if (GvIMPORTED_CV(gv))
5308 else if (! CvMETHOD(cv))
5312 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5313 (gv = *gvp) && isGV_with_GP(gv) &&
5314 GvCVu(gv) && GvIMPORTED_CV(gv))
5321 tmp = 0; /* overridden by import or by GLOBAL */
5324 && -tmp==KEY_lock /* XXX generalizable kludge */
5327 tmp = 0; /* any sub overrides "weak" keyword */
5329 else { /* no override */
5331 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5332 Perl_warner(aTHX_ packWARN(WARN_MISC),
5333 "dump() better written as CORE::dump()");
5337 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5338 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5339 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5340 "Ambiguous call resolved as CORE::%s(), %s",
5341 GvENAME(hgv), "qualify as such or use &");
5348 default: /* not a keyword */
5349 /* Trade off - by using this evil construction we can pull the
5350 variable gv into the block labelled keylookup. If not, then
5351 we have to give it function scope so that the goto from the
5352 earlier ':' case doesn't bypass the initialisation. */
5354 just_a_word_zero_gv:
5362 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5365 SV *nextPL_nextwhite = 0;
5369 /* Get the rest if it looks like a package qualifier */
5371 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5373 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5376 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5377 *s == '\'' ? "'" : "::");
5382 if (PL_expect == XOPERATOR) {
5383 if (PL_bufptr == PL_linestart) {
5384 CopLINE_dec(PL_curcop);
5385 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5386 CopLINE_inc(PL_curcop);
5389 no_op("Bareword",s);
5392 /* Look for a subroutine with this name in current package,
5393 unless name is "Foo::", in which case Foo is a bearword
5394 (and a package name). */
5396 if (len > 2 && !PL_madskills &&
5397 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5399 if (ckWARN(WARN_BAREWORD)
5400 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5401 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5402 "Bareword \"%s\" refers to nonexistent package",
5405 PL_tokenbuf[len] = '\0';
5411 /* Mustn't actually add anything to a symbol table.
5412 But also don't want to "initialise" any placeholder
5413 constants that might already be there into full
5414 blown PVGVs with attached PVCV. */
5415 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5416 GV_NOADD_NOINIT, SVt_PVCV);
5421 /* if we saw a global override before, get the right name */
5424 sv = newSVpvs("CORE::GLOBAL::");
5425 sv_catpv(sv,PL_tokenbuf);
5428 /* If len is 0, newSVpv does strlen(), which is correct.
5429 If len is non-zero, then it will be the true length,
5430 and so the scalar will be created correctly. */
5431 sv = newSVpv(PL_tokenbuf,len);
5434 if (PL_madskills && !PL_thistoken) {
5435 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5436 PL_thistoken = newSVpvn(start,s - start);
5437 PL_realtokenstart = s - SvPVX(PL_linestr);
5441 /* Presume this is going to be a bareword of some sort. */
5444 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5445 pl_yylval.opval->op_private = OPpCONST_BARE;
5446 /* UTF-8 package name? */
5447 if (UTF && !IN_BYTES &&
5448 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5451 /* And if "Foo::", then that's what it certainly is. */
5456 /* Do the explicit type check so that we don't need to force
5457 the initialisation of the symbol table to have a real GV.
5458 Beware - gv may not really be a PVGV, cv may not really be
5459 a PVCV, (because of the space optimisations that gv_init
5460 understands) But they're true if for this symbol there is
5461 respectively a typeglob and a subroutine.
5463 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5464 /* Real typeglob, so get the real subroutine: */
5466 /* A proxy for a subroutine in this package? */
5467 : SvOK(gv) ? MUTABLE_CV(gv) : NULL)
5470 /* See if it's the indirect object for a list operator. */
5472 if (PL_oldoldbufptr &&
5473 PL_oldoldbufptr < PL_bufptr &&
5474 (PL_oldoldbufptr == PL_last_lop
5475 || PL_oldoldbufptr == PL_last_uni) &&
5476 /* NO SKIPSPACE BEFORE HERE! */
5477 (PL_expect == XREF ||
5478 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5480 bool immediate_paren = *s == '(';
5482 /* (Now we can afford to cross potential line boundary.) */
5483 s = SKIPSPACE2(s,nextPL_nextwhite);
5485 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5488 /* Two barewords in a row may indicate method call. */
5490 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5491 (tmp = intuit_method(s, gv, cv)))
5494 /* If not a declared subroutine, it's an indirect object. */
5495 /* (But it's an indir obj regardless for sort.) */
5496 /* Also, if "_" follows a filetest operator, it's a bareword */
5499 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5501 (PL_last_lop_op != OP_MAPSTART &&
5502 PL_last_lop_op != OP_GREPSTART))))
5503 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5504 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5507 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5512 PL_expect = XOPERATOR;
5515 s = SKIPSPACE2(s,nextPL_nextwhite);
5516 PL_nextwhite = nextPL_nextwhite;
5521 /* Is this a word before a => operator? */
5522 if (*s == '=' && s[1] == '>' && !pkgname) {
5524 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5525 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5526 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5530 /* If followed by a paren, it's certainly a subroutine. */
5535 while (SPACE_OR_TAB(*d))
5537 if (*d == ')' && (sv = gv_const_sv(gv))) {
5544 PL_nextwhite = PL_thiswhite;
5547 start_force(PL_curforce);
5549 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5550 PL_expect = XOPERATOR;
5553 PL_nextwhite = nextPL_nextwhite;
5554 curmad('X', PL_thistoken);
5555 PL_thistoken = newSVpvs("");
5563 /* If followed by var or block, call it a method (unless sub) */
5565 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5566 PL_last_lop = PL_oldbufptr;
5567 PL_last_lop_op = OP_METHOD;
5571 /* If followed by a bareword, see if it looks like indir obj. */
5574 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5575 && (tmp = intuit_method(s, gv, cv)))
5578 /* Not a method, so call it a subroutine (if defined) */
5581 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5582 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5583 "Ambiguous use of -%s resolved as -&%s()",
5584 PL_tokenbuf, PL_tokenbuf);
5585 /* Check for a constant sub */
5586 if ((sv = gv_const_sv(gv))) {
5588 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5589 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5590 pl_yylval.opval->op_private = 0;
5594 /* Resolve to GV now. */
5595 if (SvTYPE(gv) != SVt_PVGV) {
5596 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5597 assert (SvTYPE(gv) == SVt_PVGV);
5598 /* cv must have been some sort of placeholder, so
5599 now needs replacing with a real code reference. */
5603 op_free(pl_yylval.opval);
5604 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5605 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5606 PL_last_lop = PL_oldbufptr;
5607 PL_last_lop_op = OP_ENTERSUB;
5608 /* Is there a prototype? */
5616 const char *proto = SvPV_const(MUTABLE_SV(cv), protolen);
5619 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5621 while (*proto == ';')
5623 if (*proto == '&' && *s == '{') {
5625 sv_setpvs(PL_subname, "__ANON__");
5627 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5634 PL_nextwhite = PL_thiswhite;
5637 start_force(PL_curforce);
5638 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5641 PL_nextwhite = nextPL_nextwhite;
5642 curmad('X', PL_thistoken);
5643 PL_thistoken = newSVpvs("");
5650 /* Guess harder when madskills require "best effort". */
5651 if (PL_madskills && (!gv || !GvCVu(gv))) {
5652 int probable_sub = 0;
5653 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5655 else if (isALPHA(*s)) {
5659 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5660 if (!keyword(tmpbuf, tmplen, 0))
5663 while (d < PL_bufend && isSPACE(*d))
5665 if (*d == '=' && d[1] == '>')
5670 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5671 op_free(pl_yylval.opval);
5672 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5673 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5674 PL_last_lop = PL_oldbufptr;
5675 PL_last_lop_op = OP_ENTERSUB;
5676 PL_nextwhite = PL_thiswhite;
5678 start_force(PL_curforce);
5679 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5681 PL_nextwhite = nextPL_nextwhite;
5682 curmad('X', PL_thistoken);
5683 PL_thistoken = newSVpvs("");
5688 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5695 /* Call it a bare word */
5698 if (PL_hints & HINT_STRICT_SUBS)
5699 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5701 if (lastchar != '-') {
5702 if (ckWARN(WARN_RESERVED)) {
5706 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5707 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5714 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5715 && ckWARN_d(WARN_AMBIGUOUS)) {
5716 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5717 "Operator or semicolon missing before %c%s",
5718 lastchar, PL_tokenbuf);
5719 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5720 "Ambiguous use of %c resolved as operator %c",
5721 lastchar, lastchar);
5727 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5728 newSVpv(CopFILE(PL_curcop),0));
5732 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5733 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5736 case KEY___PACKAGE__:
5737 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5739 ? newSVhek(HvNAME_HEK(PL_curstash))
5746 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5747 const char *pname = "main";
5748 if (PL_tokenbuf[2] == 'D')
5749 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5750 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5754 GvIOp(gv) = newIO();
5755 IoIFP(GvIOp(gv)) = PL_rsfp;
5756 #if defined(HAS_FCNTL) && defined(F_SETFD)
5758 const int fd = PerlIO_fileno(PL_rsfp);
5759 fcntl(fd,F_SETFD,fd >= 3);
5762 /* Mark this internal pseudo-handle as clean */
5763 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5764 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5765 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5767 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5768 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5769 /* if the script was opened in binmode, we need to revert
5770 * it to text mode for compatibility; but only iff it has CRs
5771 * XXX this is a questionable hack at best. */
5772 if (PL_bufend-PL_bufptr > 2
5773 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5776 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5777 loc = PerlIO_tell(PL_rsfp);
5778 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5781 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5783 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5784 #endif /* NETWARE */
5785 #ifdef PERLIO_IS_STDIO /* really? */
5786 # if defined(__BORLANDC__)
5787 /* XXX see note in do_binmode() */
5788 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5792 PerlIO_seek(PL_rsfp, loc, 0);
5796 #ifdef PERLIO_LAYERS
5799 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5800 else if (PL_encoding) {
5807 XPUSHs(PL_encoding);
5809 call_method("name", G_SCALAR);
5813 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5814 Perl_form(aTHX_ ":encoding(%"SVf")",
5823 if (PL_realtokenstart >= 0) {
5824 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5826 PL_endwhite = newSVpvs("");
5827 sv_catsv(PL_endwhite, PL_thiswhite);
5829 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5830 PL_realtokenstart = -1;
5832 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5833 SvCUR(PL_endwhite))) != NULL) ;
5848 if (PL_expect == XSTATE) {
5855 if (*s == ':' && s[1] == ':') {
5858 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5859 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5860 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5863 else if (tmp == KEY_require || tmp == KEY_do)
5864 /* that's a way to remember we saw "CORE::" */
5877 LOP(OP_ACCEPT,XTERM);
5883 LOP(OP_ATAN2,XTERM);
5889 LOP(OP_BINMODE,XTERM);
5892 LOP(OP_BLESS,XTERM);
5901 /* When 'use switch' is in effect, continue has a dual
5902 life as a control operator. */
5904 if (!FEATURE_IS_ENABLED("switch"))
5907 /* We have to disambiguate the two senses of
5908 "continue". If the next token is a '{' then
5909 treat it as the start of a continue block;
5910 otherwise treat it as a control operator.
5922 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5939 if (!PL_cryptseen) {
5940 PL_cryptseen = TRUE;
5944 LOP(OP_CRYPT,XTERM);
5947 LOP(OP_CHMOD,XTERM);
5950 LOP(OP_CHOWN,XTERM);
5953 LOP(OP_CONNECT,XTERM);
5972 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5973 if (orig_keyword == KEY_do) {
5982 PL_hints |= HINT_BLOCK_SCOPE;
5992 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5993 LOP(OP_DBMOPEN,XTERM);
5999 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6006 pl_yylval.ival = CopLINE(PL_curcop);
6022 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
6023 UNIBRACK(OP_ENTEREVAL);
6037 case KEY_endhostent:
6043 case KEY_endservent:
6046 case KEY_endprotoent:
6057 pl_yylval.ival = CopLINE(PL_curcop);
6059 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6062 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6065 if ((PL_bufend - p) >= 3 &&
6066 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6068 else if ((PL_bufend - p) >= 4 &&
6069 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6072 if (isIDFIRST_lazy_if(p,UTF)) {
6073 p = scan_ident(p, PL_bufend,
6074 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6078 Perl_croak(aTHX_ "Missing $ on loop variable");
6080 s = SvPVX(PL_linestr) + soff;
6086 LOP(OP_FORMLINE,XTERM);
6092 LOP(OP_FCNTL,XTERM);
6098 LOP(OP_FLOCK,XTERM);
6107 LOP(OP_GREPSTART, XREF);
6110 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6125 case KEY_getpriority:
6126 LOP(OP_GETPRIORITY,XTERM);
6128 case KEY_getprotobyname:
6131 case KEY_getprotobynumber:
6132 LOP(OP_GPBYNUMBER,XTERM);
6134 case KEY_getprotoent:
6146 case KEY_getpeername:
6147 UNI(OP_GETPEERNAME);
6149 case KEY_gethostbyname:
6152 case KEY_gethostbyaddr:
6153 LOP(OP_GHBYADDR,XTERM);
6155 case KEY_gethostent:
6158 case KEY_getnetbyname:
6161 case KEY_getnetbyaddr:
6162 LOP(OP_GNBYADDR,XTERM);
6167 case KEY_getservbyname:
6168 LOP(OP_GSBYNAME,XTERM);
6170 case KEY_getservbyport:
6171 LOP(OP_GSBYPORT,XTERM);
6173 case KEY_getservent:
6176 case KEY_getsockname:
6177 UNI(OP_GETSOCKNAME);
6179 case KEY_getsockopt:
6180 LOP(OP_GSOCKOPT,XTERM);
6195 pl_yylval.ival = CopLINE(PL_curcop);
6205 pl_yylval.ival = CopLINE(PL_curcop);
6209 LOP(OP_INDEX,XTERM);
6215 LOP(OP_IOCTL,XTERM);
6227 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6259 LOP(OP_LISTEN,XTERM);
6268 s = scan_pat(s,OP_MATCH);
6269 TERM(sublex_start());
6272 LOP(OP_MAPSTART, XREF);
6275 LOP(OP_MKDIR,XTERM);
6278 LOP(OP_MSGCTL,XTERM);
6281 LOP(OP_MSGGET,XTERM);
6284 LOP(OP_MSGRCV,XTERM);
6287 LOP(OP_MSGSND,XTERM);
6292 PL_in_my = (U16)tmp;
6294 if (isIDFIRST_lazy_if(s,UTF)) {
6298 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6299 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6301 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6302 if (!PL_in_my_stash) {
6305 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6309 if (PL_madskills) { /* just add type to declarator token */
6310 sv_catsv(PL_thistoken, PL_nextwhite);
6312 sv_catpvn(PL_thistoken, start, s - start);
6320 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6327 s = tokenize_use(0, s);
6331 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6338 if (isIDFIRST_lazy_if(s,UTF)) {
6340 for (d = s; isALNUM_lazy_if(d,UTF);)
6342 for (t=d; isSPACE(*t);)
6344 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6346 && !(t[0] == '=' && t[1] == '>')
6348 int parms_len = (int)(d-s);
6349 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6350 "Precedence problem: open %.*s should be open(%.*s)",
6351 parms_len, s, parms_len, s);
6357 pl_yylval.ival = OP_OR;
6367 LOP(OP_OPEN_DIR,XTERM);
6370 checkcomma(s,PL_tokenbuf,"filehandle");
6374 checkcomma(s,PL_tokenbuf,"filehandle");
6393 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6397 LOP(OP_PIPE_OP,XTERM);
6400 s = scan_str(s,!!PL_madskills,FALSE);
6403 pl_yylval.ival = OP_CONST;
6404 TERM(sublex_start());
6410 s = scan_str(s,!!PL_madskills,FALSE);
6413 PL_expect = XOPERATOR;
6415 if (SvCUR(PL_lex_stuff)) {
6418 d = SvPV_force(PL_lex_stuff, len);
6420 for (; isSPACE(*d) && len; --len, ++d)
6425 if (!warned && ckWARN(WARN_QW)) {
6426 for (; !isSPACE(*d) && len; --len, ++d) {
6428 Perl_warner(aTHX_ packWARN(WARN_QW),
6429 "Possible attempt to separate words with commas");
6432 else if (*d == '#') {
6433 Perl_warner(aTHX_ packWARN(WARN_QW),
6434 "Possible attempt to put comments in qw() list");
6440 for (; !isSPACE(*d) && len; --len, ++d)
6443 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6444 words = append_elem(OP_LIST, words,
6445 newSVOP(OP_CONST, 0, tokeq(sv)));
6449 start_force(PL_curforce);
6450 NEXTVAL_NEXTTOKE.opval = words;
6455 SvREFCNT_dec(PL_lex_stuff);
6456 PL_lex_stuff = NULL;
6462 s = scan_str(s,!!PL_madskills,FALSE);
6465 pl_yylval.ival = OP_STRINGIFY;
6466 if (SvIVX(PL_lex_stuff) == '\'')
6467 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6468 TERM(sublex_start());
6471 s = scan_pat(s,OP_QR);
6472 TERM(sublex_start());
6475 s = scan_str(s,!!PL_madskills,FALSE);
6478 readpipe_override();
6479 TERM(sublex_start());
6487 s = force_version(s, FALSE);
6489 else if (*s != 'v' || !isDIGIT(s[1])
6490 || (s = force_version(s, TRUE), *s == 'v'))
6492 *PL_tokenbuf = '\0';
6493 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6494 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6495 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6497 yyerror("<> should be quotes");
6499 if (orig_keyword == KEY_require) {
6507 PL_last_uni = PL_oldbufptr;
6508 PL_last_lop_op = OP_REQUIRE;
6510 return REPORT( (int)REQUIRE );
6516 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6520 LOP(OP_RENAME,XTERM);
6529 LOP(OP_RINDEX,XTERM);
6538 UNIDOR(OP_READLINE);
6541 UNIDOR(OP_BACKTICK);
6550 LOP(OP_REVERSE,XTERM);
6553 UNIDOR(OP_READLINK);
6560 if (pl_yylval.opval)
6561 TERM(sublex_start());
6563 TOKEN(1); /* force error */
6566 checkcomma(s,PL_tokenbuf,"filehandle");
6576 LOP(OP_SELECT,XTERM);
6582 LOP(OP_SEMCTL,XTERM);
6585 LOP(OP_SEMGET,XTERM);
6588 LOP(OP_SEMOP,XTERM);
6594 LOP(OP_SETPGRP,XTERM);
6596 case KEY_setpriority:
6597 LOP(OP_SETPRIORITY,XTERM);
6599 case KEY_sethostent:
6605 case KEY_setservent:
6608 case KEY_setprotoent:
6618 LOP(OP_SEEKDIR,XTERM);
6620 case KEY_setsockopt:
6621 LOP(OP_SSOCKOPT,XTERM);
6627 LOP(OP_SHMCTL,XTERM);
6630 LOP(OP_SHMGET,XTERM);
6633 LOP(OP_SHMREAD,XTERM);
6636 LOP(OP_SHMWRITE,XTERM);
6639 LOP(OP_SHUTDOWN,XTERM);
6648 LOP(OP_SOCKET,XTERM);
6650 case KEY_socketpair:
6651 LOP(OP_SOCKPAIR,XTERM);
6654 checkcomma(s,PL_tokenbuf,"subroutine name");
6656 if (*s == ';' || *s == ')') /* probably a close */
6657 Perl_croak(aTHX_ "sort is now a reserved word");
6659 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6663 LOP(OP_SPLIT,XTERM);
6666 LOP(OP_SPRINTF,XTERM);
6669 LOP(OP_SPLICE,XTERM);
6684 LOP(OP_SUBSTR,XTERM);
6690 char tmpbuf[sizeof PL_tokenbuf];
6691 SSize_t tboffset = 0;
6692 expectation attrful;
6693 bool have_name, have_proto;
6694 const int key = tmp;
6699 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6700 SV *subtoken = newSVpvn(tstart, s - tstart);
6704 s = SKIPSPACE2(s,tmpwhite);
6709 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6710 (*s == ':' && s[1] == ':'))
6713 SV *nametoke = NULL;
6717 attrful = XATTRBLOCK;
6718 /* remember buffer pos'n for later force_word */
6719 tboffset = s - PL_oldbufptr;
6720 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6723 nametoke = newSVpvn(s, d - s);
6725 if (memchr(tmpbuf, ':', len))
6726 sv_setpvn(PL_subname, tmpbuf, len);
6728 sv_setsv(PL_subname,PL_curstname);
6729 sv_catpvs(PL_subname,"::");
6730 sv_catpvn(PL_subname,tmpbuf,len);
6737 CURMAD('X', nametoke);
6738 CURMAD('_', tmpwhite);
6739 (void) force_word(PL_oldbufptr + tboffset, WORD,
6742 s = SKIPSPACE2(d,tmpwhite);
6749 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6750 PL_expect = XTERMBLOCK;
6751 attrful = XATTRTERM;
6752 sv_setpvs(PL_subname,"?");
6756 if (key == KEY_format) {
6758 PL_lex_formbrack = PL_lex_brackets + 1;
6760 PL_thistoken = subtoken;
6764 (void) force_word(PL_oldbufptr + tboffset, WORD,
6770 /* Look for a prototype */
6773 bool bad_proto = FALSE;
6774 bool in_brackets = FALSE;
6775 char greedy_proto = ' ';
6776 bool proto_after_greedy_proto = FALSE;
6777 bool must_be_last = FALSE;
6778 bool underscore = FALSE;
6779 bool seen_underscore = FALSE;
6780 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6782 s = scan_str(s,!!PL_madskills,FALSE);
6784 Perl_croak(aTHX_ "Prototype not terminated");
6785 /* strip spaces and check for bad characters */
6786 d = SvPVX(PL_lex_stuff);
6788 for (p = d; *p; ++p) {
6794 proto_after_greedy_proto = TRUE;
6795 if (!strchr("$@%*;[]&\\_", *p)) {
6807 else if ( *p == ']' ) {
6808 in_brackets = FALSE;
6810 else if ( (*p == '@' || *p == '%') &&
6811 ( tmp < 2 || d[tmp-2] != '\\' ) &&
6813 must_be_last = TRUE;
6816 else if ( *p == '_' ) {
6817 underscore = seen_underscore = TRUE;
6824 if (proto_after_greedy_proto)
6825 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6826 "Prototype after '%c' for %"SVf" : %s",
6827 greedy_proto, SVfARG(PL_subname), d);
6829 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6830 "Illegal character %sin prototype for %"SVf" : %s",
6831 seen_underscore ? "after '_' " : "",
6832 SVfARG(PL_subname), d);
6833 SvCUR_set(PL_lex_stuff, tmp);
6838 CURMAD('q', PL_thisopen);
6839 CURMAD('_', tmpwhite);
6840 CURMAD('=', PL_thisstuff);
6841 CURMAD('Q', PL_thisclose);
6842 NEXTVAL_NEXTTOKE.opval =
6843 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6844 PL_lex_stuff = NULL;
6847 s = SKIPSPACE2(s,tmpwhite);
6855 if (*s == ':' && s[1] != ':')
6856 PL_expect = attrful;
6857 else if (*s != '{' && key == KEY_sub) {
6859 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6861 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6868 curmad('^', newSVpvs(""));
6869 CURMAD('_', tmpwhite);
6873 PL_thistoken = subtoken;
6876 NEXTVAL_NEXTTOKE.opval =
6877 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6878 PL_lex_stuff = NULL;
6884 sv_setpvs(PL_subname, "__ANON__");
6886 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6890 (void) force_word(PL_oldbufptr + tboffset, WORD,
6899 LOP(OP_SYSTEM,XREF);
6902 LOP(OP_SYMLINK,XTERM);
6905 LOP(OP_SYSCALL,XTERM);
6908 LOP(OP_SYSOPEN,XTERM);
6911 LOP(OP_SYSSEEK,XTERM);
6914 LOP(OP_SYSREAD,XTERM);
6917 LOP(OP_SYSWRITE,XTERM);
6921 TERM(sublex_start());
6942 LOP(OP_TRUNCATE,XTERM);
6954 pl_yylval.ival = CopLINE(PL_curcop);
6958 pl_yylval.ival = CopLINE(PL_curcop);
6962 LOP(OP_UNLINK,XTERM);
6968 LOP(OP_UNPACK,XTERM);
6971 LOP(OP_UTIME,XTERM);
6977 LOP(OP_UNSHIFT,XTERM);
6980 s = tokenize_use(1, s);
6990 pl_yylval.ival = CopLINE(PL_curcop);
6994 pl_yylval.ival = CopLINE(PL_curcop);
6998 PL_hints |= HINT_BLOCK_SCOPE;
7005 LOP(OP_WAITPID,XTERM);
7014 ctl_l[0] = toCTRL('L');
7016 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
7019 /* Make sure $^L is defined */
7020 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
7025 if (PL_expect == XOPERATOR)
7031 pl_yylval.ival = OP_XOR;
7036 TERM(sublex_start());
7041 #pragma segment Main
7045 S_pending_ident(pTHX)
7050 /* pit holds the identifier we read and pending_ident is reset */
7051 char pit = PL_pending_ident;
7052 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7053 /* All routes through this function want to know if there is a colon. */
7054 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7055 PL_pending_ident = 0;
7057 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7058 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7059 "### Pending identifier '%s'\n", PL_tokenbuf); });
7061 /* if we're in a my(), we can't allow dynamics here.
7062 $foo'bar has already been turned into $foo::bar, so
7063 just check for colons.
7065 if it's a legal name, the OP is a PADANY.
7068 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7070 yyerror(Perl_form(aTHX_ "No package name allowed for "
7071 "variable %s in \"our\"",
7073 tmp = allocmy(PL_tokenbuf);
7077 yyerror(Perl_form(aTHX_ PL_no_myglob,
7078 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7080 pl_yylval.opval = newOP(OP_PADANY, 0);
7081 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7087 build the ops for accesses to a my() variable.
7089 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7090 then used in a comparison. This catches most, but not
7091 all cases. For instance, it catches
7092 sort { my($a); $a <=> $b }
7094 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7095 (although why you'd do that is anyone's guess).
7100 tmp = pad_findmy(PL_tokenbuf);
7101 if (tmp != NOT_IN_PAD) {
7102 /* might be an "our" variable" */
7103 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7104 /* build ops for a bareword */
7105 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7106 HEK * const stashname = HvNAME_HEK(stash);
7107 SV * const sym = newSVhek(stashname);
7108 sv_catpvs(sym, "::");
7109 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7110 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7111 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7114 ? (GV_ADDMULTI | GV_ADDINEVAL)
7117 ((PL_tokenbuf[0] == '$') ? SVt_PV
7118 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7123 /* if it's a sort block and they're naming $a or $b */
7124 if (PL_last_lop_op == OP_SORT &&
7125 PL_tokenbuf[0] == '$' &&
7126 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7129 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7130 d < PL_bufend && *d != '\n';
7133 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7134 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7140 pl_yylval.opval = newOP(OP_PADANY, 0);
7141 pl_yylval.opval->op_targ = tmp;
7147 Whine if they've said @foo in a doublequoted string,
7148 and @foo isn't a variable we can find in the symbol
7151 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7152 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7154 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7155 && ckWARN(WARN_AMBIGUOUS)
7156 /* DO NOT warn for @- and @+ */
7157 && !( PL_tokenbuf[2] == '\0' &&
7158 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7161 /* Downgraded from fatal to warning 20000522 mjd */
7162 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7163 "Possible unintended interpolation of %s in string",
7168 /* build ops for a bareword */
7169 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7171 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7173 PL_tokenbuf + 1, tokenbuf_len - 1,
7174 /* If the identifier refers to a stash, don't autovivify it.
7175 * Change 24660 had the side effect of causing symbol table
7176 * hashes to always be defined, even if they were freshly
7177 * created and the only reference in the entire program was
7178 * the single statement with the defined %foo::bar:: test.
7179 * It appears that all code in the wild doing this actually
7180 * wants to know whether sub-packages have been loaded, so
7181 * by avoiding auto-vivifying symbol tables, we ensure that
7182 * defined %foo::bar:: continues to be false, and the existing
7183 * tests still give the expected answers, even though what
7184 * they're actually testing has now changed subtly.
7186 (*PL_tokenbuf == '%'
7187 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7190 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7191 ((PL_tokenbuf[0] == '$') ? SVt_PV
7192 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7198 * The following code was generated by perl_keyword.pl.
7202 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7206 PERL_ARGS_ASSERT_KEYWORD;
7210 case 1: /* 5 tokens of length 1 */
7242 case 2: /* 18 tokens of length 2 */
7388 case 3: /* 29 tokens of length 3 */
7392 if (name[1] == 'N' &&
7455 if (name[1] == 'i' &&
7487 if (name[1] == 'o' &&
7496 if (name[1] == 'e' &&
7505 if (name[1] == 'n' &&
7514 if (name[1] == 'o' &&
7523 if (name[1] == 'a' &&
7532 if (name[1] == 'o' &&
7594 if (name[1] == 'e' &&
7608 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7634 if (name[1] == 'i' &&
7643 if (name[1] == 's' &&
7652 if (name[1] == 'e' &&
7661 if (name[1] == 'o' &&
7673 case 4: /* 41 tokens of length 4 */
7677 if (name[1] == 'O' &&
7687 if (name[1] == 'N' &&
7697 if (name[1] == 'i' &&
7707 if (name[1] == 'h' &&
7717 if (name[1] == 'u' &&
7730 if (name[2] == 'c' &&
7739 if (name[2] == 's' &&
7748 if (name[2] == 'a' &&
7784 if (name[1] == 'o' &&
7797 if (name[2] == 't' &&
7806 if (name[2] == 'o' &&
7815 if (name[2] == 't' &&
7824 if (name[2] == 'e' &&
7837 if (name[1] == 'o' &&
7850 if (name[2] == 'y' &&
7859 if (name[2] == 'l' &&
7875 if (name[2] == 's' &&
7884 if (name[2] == 'n' &&
7893 if (name[2] == 'c' &&
7906 if (name[1] == 'e' &&
7916 if (name[1] == 'p' &&
7929 if (name[2] == 'c' &&
7938 if (name[2] == 'p' &&
7947 if (name[2] == 's' &&
7963 if (name[2] == 'n' &&
8033 if (name[2] == 'r' &&
8042 if (name[2] == 'r' &&
8051 if (name[2] == 'a' &&
8067 if (name[2] == 'l' &&
8129 if (name[2] == 'e' &&
8132 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8145 case 5: /* 39 tokens of length 5 */
8149 if (name[1] == 'E' &&
8160 if (name[1] == 'H' &&
8174 if (name[2] == 'a' &&
8184 if (name[2] == 'a' &&
8201 if (name[2] == 'e' &&
8211 if (name[2] == 'e' &&
8215 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8231 if (name[3] == 'i' &&
8240 if (name[3] == 'o' &&
8276 if (name[2] == 'o' &&
8286 if (name[2] == 'y' &&
8300 if (name[1] == 'l' &&
8314 if (name[2] == 'n' &&
8324 if (name[2] == 'o' &&
8338 if (name[1] == 'i' &&
8343 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8352 if (name[2] == 'd' &&
8362 if (name[2] == 'c' &&
8379 if (name[2] == 'c' &&
8389 if (name[2] == 't' &&
8403 if (name[1] == 'k' &&
8414 if (name[1] == 'r' &&
8428 if (name[2] == 's' &&
8438 if (name[2] == 'd' &&
8455 if (name[2] == 'm' &&
8465 if (name[2] == 'i' &&
8475 if (name[2] == 'e' &&
8485 if (name[2] == 'l' &&
8495 if (name[2] == 'a' &&
8508 if (name[3] == 't' &&
8511 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8517 if (name[3] == 'd' &&
8534 if (name[1] == 'i' &&
8548 if (name[2] == 'a' &&
8561 if (name[3] == 'e' &&
8596 if (name[2] == 'i' &&
8613 if (name[2] == 'i' &&
8623 if (name[2] == 'i' &&
8640 case 6: /* 33 tokens of length 6 */
8644 if (name[1] == 'c' &&
8659 if (name[2] == 'l' &&
8670 if (name[2] == 'r' &&
8685 if (name[1] == 'e' &&
8700 if (name[2] == 's' &&
8705 if(ckWARN_d(WARN_SYNTAX))
8706 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8712 if (name[2] == 'i' &&
8730 if (name[2] == 'l' &&
8741 if (name[2] == 'r' &&
8756 if (name[1] == 'm' &&
8771 if (name[2] == 'n' &&
8782 if (name[2] == 's' &&
8797 if (name[1] == 's' &&
8803 if (name[4] == 't' &&
8812 if (name[4] == 'e' &&
8821 if (name[4] == 'c' &&
8830 if (name[4] == 'n' &&
8846 if (name[1] == 'r' &&
8864 if (name[3] == 'a' &&
8874 if (name[3] == 'u' &&
8888 if (name[2] == 'n' &&
8906 if (name[2] == 'a' &&
8920 if (name[3] == 'e' &&
8933 if (name[4] == 't' &&
8942 if (name[4] == 'e' &&
8964 if (name[4] == 't' &&
8973 if (name[4] == 'e' &&
8989 if (name[2] == 'c' &&
9000 if (name[2] == 'l' &&
9011 if (name[2] == 'b' &&
9022 if (name[2] == 's' &&
9045 if (name[4] == 's' &&
9054 if (name[4] == 'n' &&
9067 if (name[3] == 'a' &&
9084 if (name[1] == 'a' &&
9099 case 7: /* 29 tokens of length 7 */
9103 if (name[1] == 'E' &&
9116 if (name[1] == '_' &&
9129 if (name[1] == 'i' &&
9136 return -KEY_binmode;
9142 if (name[1] == 'o' &&
9149 return -KEY_connect;
9158 if (name[2] == 'm' &&
9164 return -KEY_dbmopen;
9175 if (name[4] == 'u' &&
9179 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9185 if (name[4] == 'n' &&
9206 if (name[1] == 'o' &&
9219 if (name[1] == 'e' &&
9226 if (name[5] == 'r' &&
9229 return -KEY_getpgrp;
9235 if (name[5] == 'i' &&
9238 return -KEY_getppid;
9251 if (name[1] == 'c' &&
9258 return -KEY_lcfirst;
9264 if (name[1] == 'p' &&
9271 return -KEY_opendir;
9277 if (name[1] == 'a' &&
9295 if (name[3] == 'd' &&
9300 return -KEY_readdir;
9306 if (name[3] == 'u' &&
9317 if (name[3] == 'e' &&
9322 return -KEY_reverse;
9341 if (name[3] == 'k' &&
9346 return -KEY_seekdir;
9352 if (name[3] == 'p' &&
9357 return -KEY_setpgrp;
9367 if (name[2] == 'm' &&
9373 return -KEY_shmread;
9379 if (name[2] == 'r' &&
9385 return -KEY_sprintf;
9394 if (name[3] == 'l' &&
9399 return -KEY_symlink;
9408 if (name[4] == 'a' &&
9412 return -KEY_syscall;
9418 if (name[4] == 'p' &&
9422 return -KEY_sysopen;
9428 if (name[4] == 'e' &&
9432 return -KEY_sysread;
9438 if (name[4] == 'e' &&
9442 return -KEY_sysseek;
9460 if (name[1] == 'e' &&
9467 return -KEY_telldir;
9476 if (name[2] == 'f' &&
9482 return -KEY_ucfirst;
9488 if (name[2] == 's' &&
9494 return -KEY_unshift;
9504 if (name[1] == 'a' &&
9511 return -KEY_waitpid;
9520 case 8: /* 26 tokens of length 8 */
9524 if (name[1] == 'U' &&
9532 return KEY_AUTOLOAD;
9543 if (name[3] == 'A' &&
9549 return KEY___DATA__;
9555 if (name[3] == 'I' &&
9561 return -KEY___FILE__;
9567 if (name[3] == 'I' &&
9573 return -KEY___LINE__;
9589 if (name[2] == 'o' &&
9596 return -KEY_closedir;
9602 if (name[2] == 'n' &&
9609 return -KEY_continue;
9619 if (name[1] == 'b' &&
9627 return -KEY_dbmclose;
9633 if (name[1] == 'n' &&
9639 if (name[4] == 'r' &&
9644 return -KEY_endgrent;
9650 if (name[4] == 'w' &&
9655 return -KEY_endpwent;
9668 if (name[1] == 'o' &&
9676 return -KEY_formline;
9682 if (name[1] == 'e' &&
9693 if (name[6] == 'n' &&
9696 return -KEY_getgrent;
9702 if (name[6] == 'i' &&
9705 return -KEY_getgrgid;
9711 if (name[6] == 'a' &&
9714 return -KEY_getgrnam;
9727 if (name[4] == 'o' &&
9732 return -KEY_getlogin;
9743 if (name[6] == 'n' &&
9746 return -KEY_getpwent;
9752 if (name[6] == 'a' &&
9755 return -KEY_getpwnam;
9761 if (name[6] == 'i' &&
9764 return -KEY_getpwuid;
9784 if (name[1] == 'e' &&
9791 if (name[5] == 'i' &&
9798 return -KEY_readline;
9803 return -KEY_readlink;
9814 if (name[5] == 'i' &&
9818 return -KEY_readpipe;
9839 if (name[4] == 'r' &&
9844 return -KEY_setgrent;
9850 if (name[4] == 'w' &&
9855 return -KEY_setpwent;
9871 if (name[3] == 'w' &&
9877 return -KEY_shmwrite;
9883 if (name[3] == 't' &&
9889 return -KEY_shutdown;
9899 if (name[2] == 's' &&
9906 return -KEY_syswrite;
9916 if (name[1] == 'r' &&
9924 return -KEY_truncate;
9933 case 9: /* 9 tokens of length 9 */
9937 if (name[1] == 'N' &&
9946 return KEY_UNITCHECK;
9952 if (name[1] == 'n' &&
9961 return -KEY_endnetent;
9967 if (name[1] == 'e' &&
9976 return -KEY_getnetent;
9982 if (name[1] == 'o' &&
9991 return -KEY_localtime;
9997 if (name[1] == 'r' &&
10006 return KEY_prototype;
10012 if (name[1] == 'u' &&
10021 return -KEY_quotemeta;
10027 if (name[1] == 'e' &&
10036 return -KEY_rewinddir;
10042 if (name[1] == 'e' &&
10051 return -KEY_setnetent;
10057 if (name[1] == 'a' &&
10066 return -KEY_wantarray;
10075 case 10: /* 9 tokens of length 10 */
10079 if (name[1] == 'n' &&
10085 if (name[4] == 'o' &&
10092 return -KEY_endhostent;
10098 if (name[4] == 'e' &&
10105 return -KEY_endservent;
10118 if (name[1] == 'e' &&
10124 if (name[4] == 'o' &&
10131 return -KEY_gethostent;
10140 if (name[5] == 'r' &&
10146 return -KEY_getservent;
10152 if (name[5] == 'c' &&
10158 return -KEY_getsockopt;
10178 if (name[2] == 't')
10183 if (name[4] == 'o' &&
10190 return -KEY_sethostent;
10199 if (name[5] == 'r' &&
10205 return -KEY_setservent;
10211 if (name[5] == 'c' &&
10217 return -KEY_setsockopt;
10234 if (name[2] == 'c' &&
10243 return -KEY_socketpair;
10256 case 11: /* 8 tokens of length 11 */
10260 if (name[1] == '_' &&
10270 { /* __PACKAGE__ */
10271 return -KEY___PACKAGE__;
10277 if (name[1] == 'n' &&
10287 { /* endprotoent */
10288 return -KEY_endprotoent;
10294 if (name[1] == 'e' &&
10303 if (name[5] == 'e' &&
10309 { /* getpeername */
10310 return -KEY_getpeername;
10319 if (name[6] == 'o' &&
10324 { /* getpriority */
10325 return -KEY_getpriority;
10331 if (name[6] == 't' &&
10336 { /* getprotoent */
10337 return -KEY_getprotoent;
10351 if (name[4] == 'o' &&
10358 { /* getsockname */
10359 return -KEY_getsockname;
10372 if (name[1] == 'e' &&
10380 if (name[6] == 'o' &&
10385 { /* setpriority */
10386 return -KEY_setpriority;
10392 if (name[6] == 't' &&
10397 { /* setprotoent */
10398 return -KEY_setprotoent;
10414 case 12: /* 2 tokens of length 12 */
10415 if (name[0] == 'g' &&
10427 if (name[9] == 'd' &&
10430 { /* getnetbyaddr */
10431 return -KEY_getnetbyaddr;
10437 if (name[9] == 'a' &&
10440 { /* getnetbyname */
10441 return -KEY_getnetbyname;
10453 case 13: /* 4 tokens of length 13 */
10454 if (name[0] == 'g' &&
10461 if (name[4] == 'o' &&
10470 if (name[10] == 'd' &&
10473 { /* gethostbyaddr */
10474 return -KEY_gethostbyaddr;
10480 if (name[10] == 'a' &&
10483 { /* gethostbyname */
10484 return -KEY_gethostbyname;
10497 if (name[4] == 'e' &&
10506 if (name[10] == 'a' &&
10509 { /* getservbyname */
10510 return -KEY_getservbyname;
10516 if (name[10] == 'o' &&
10519 { /* getservbyport */
10520 return -KEY_getservbyport;
10539 case 14: /* 1 tokens of length 14 */
10540 if (name[0] == 'g' &&
10554 { /* getprotobyname */
10555 return -KEY_getprotobyname;
10560 case 16: /* 1 tokens of length 16 */
10561 if (name[0] == 'g' &&
10577 { /* getprotobynumber */
10578 return -KEY_getprotobynumber;
10592 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10596 PERL_ARGS_ASSERT_CHECKCOMMA;
10598 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10599 if (ckWARN(WARN_SYNTAX)) {
10602 for (w = s+2; *w && level; w++) {
10605 else if (*w == ')')
10608 while (isSPACE(*w))
10610 /* the list of chars below is for end of statements or
10611 * block / parens, boolean operators (&&, ||, //) and branch
10612 * constructs (or, and, if, until, unless, while, err, for).
10613 * Not a very solid hack... */
10614 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10615 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10616 "%s (...) interpreted as function",name);
10619 while (s < PL_bufend && isSPACE(*s))
10623 while (s < PL_bufend && isSPACE(*s))
10625 if (isIDFIRST_lazy_if(s,UTF)) {
10626 const char * const w = s++;
10627 while (isALNUM_lazy_if(s,UTF))
10629 while (s < PL_bufend && isSPACE(*s))
10633 if (keyword(w, s - w, 0))
10636 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10637 if (gv && GvCVu(gv))
10639 Perl_croak(aTHX_ "No comma allowed after %s", what);
10644 /* Either returns sv, or mortalizes sv and returns a new SV*.
10645 Best used as sv=new_constant(..., sv, ...).
10646 If s, pv are NULL, calls subroutine with one argument,
10647 and type is used with error messages only. */
10650 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10651 SV *sv, SV *pv, const char *type, STRLEN typelen)
10654 HV * const table = GvHV(PL_hintgv); /* ^H */
10658 const char *why1 = "", *why2 = "", *why3 = "";
10660 PERL_ARGS_ASSERT_NEW_CONSTANT;
10662 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10665 why2 = (const char *)
10666 (strEQ(key,"charnames")
10667 ? "(possibly a missing \"use charnames ...\")"
10669 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10670 (type ? type: "undef"), why2);
10672 /* This is convoluted and evil ("goto considered harmful")
10673 * but I do not understand the intricacies of all the different
10674 * failure modes of %^H in here. The goal here is to make
10675 * the most probable error message user-friendly. --jhi */
10680 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10681 (type ? type: "undef"), why1, why2, why3);
10683 yyerror(SvPVX_const(msg));
10687 cvp = hv_fetch(table, key, keylen, FALSE);
10688 if (!cvp || !SvOK(*cvp)) {
10691 why3 = "} is not defined";
10694 sv_2mortal(sv); /* Parent created it permanently */
10697 pv = newSVpvn_flags(s, len, SVs_TEMP);
10699 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10701 typesv = &PL_sv_undef;
10703 PUSHSTACKi(PERLSI_OVERLOAD);
10715 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10719 /* Check the eval first */
10720 if (!PL_in_eval && SvTRUE(ERRSV)) {
10721 sv_catpvs(ERRSV, "Propagated");
10722 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10724 res = SvREFCNT_inc_simple(sv);
10728 SvREFCNT_inc_simple_void(res);
10737 why1 = "Call to &{$^H{";
10739 why3 = "}} did not return a defined value";
10747 /* Returns a NUL terminated string, with the length of the string written to
10751 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10754 register char *d = dest;
10755 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10757 PERL_ARGS_ASSERT_SCAN_WORD;
10761 Perl_croak(aTHX_ ident_too_long);
10762 if (isALNUM(*s)) /* UTF handled below */
10764 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10769 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10773 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10774 char *t = s + UTF8SKIP(s);
10776 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10780 Perl_croak(aTHX_ ident_too_long);
10781 Copy(s, d, len, char);
10794 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10797 char *bracket = NULL;
10799 register char *d = dest;
10800 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10802 PERL_ARGS_ASSERT_SCAN_IDENT;
10807 while (isDIGIT(*s)) {
10809 Perl_croak(aTHX_ ident_too_long);
10816 Perl_croak(aTHX_ ident_too_long);
10817 if (isALNUM(*s)) /* UTF handled below */
10819 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10824 else if (*s == ':' && s[1] == ':') {
10828 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10829 char *t = s + UTF8SKIP(s);
10830 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10832 if (d + (t - s) > e)
10833 Perl_croak(aTHX_ ident_too_long);
10834 Copy(s, d, t - s, char);
10845 if (PL_lex_state != LEX_NORMAL)
10846 PL_lex_state = LEX_INTERPENDMAYBE;
10849 if (*s == '$' && s[1] &&
10850 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10863 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10868 if (isSPACE(s[-1])) {
10870 const char ch = *s++;
10871 if (!SPACE_OR_TAB(ch)) {
10877 if (isIDFIRST_lazy_if(d,UTF)) {
10881 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10882 end += UTF8SKIP(end);
10883 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10884 end += UTF8SKIP(end);
10886 Copy(s, d, end - s, char);
10891 while ((isALNUM(*s) || *s == ':') && d < e)
10894 Perl_croak(aTHX_ ident_too_long);
10897 while (s < send && SPACE_OR_TAB(*s))
10899 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10900 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10901 const char * const brack =
10903 ((*s == '[') ? "[...]" : "{...}");
10904 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10905 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10906 funny, dest, brack, funny, dest, brack);
10909 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10913 /* Handle extended ${^Foo} variables
10914 * 1999-02-27 mjd-perl-patch@plover.com */
10915 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10919 while (isALNUM(*s) && d < e) {
10923 Perl_croak(aTHX_ ident_too_long);
10928 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10929 PL_lex_state = LEX_INTERPEND;
10932 if (PL_lex_state == LEX_NORMAL) {
10933 if (ckWARN(WARN_AMBIGUOUS) &&
10934 (keyword(dest, d - dest, 0)
10935 || get_cvn_flags(dest, d - dest, 0)))
10939 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10940 "Ambiguous use of %c{%s} resolved to %c%s",
10941 funny, dest, funny, dest);
10946 s = bracket; /* let the parser handle it */
10950 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10951 PL_lex_state = LEX_INTERPEND;
10956 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10958 PERL_ARGS_ASSERT_PMFLAG;
10960 PERL_UNUSED_CONTEXT;
10962 const char c = (char)ch;
10964 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10965 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10966 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10967 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10968 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10974 S_scan_pat(pTHX_ char *start, I32 type)
10978 char *s = scan_str(start,!!PL_madskills,FALSE);
10979 const char * const valid_flags =
10980 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10985 PERL_ARGS_ASSERT_SCAN_PAT;
10988 const char * const delimiter = skipspace(start);
10992 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10993 : "Search pattern not terminated" ));
10996 pm = (PMOP*)newPMOP(type, 0);
10997 if (PL_multi_open == '?') {
10998 /* This is the only point in the code that sets PMf_ONCE: */
10999 pm->op_pmflags |= PMf_ONCE;
11001 /* Hence it's safe to do this bit of PMOP book-keeping here, which
11002 allows us to restrict the list needed by reset to just the ??
11004 assert(type != OP_TRANS);
11006 MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
11009 mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
11012 elements = mg->mg_len / sizeof(PMOP**);
11013 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
11014 ((PMOP**)mg->mg_ptr) [elements++] = pm;
11015 mg->mg_len = elements * sizeof(PMOP**);
11016 PmopSTASH_set(pm,PL_curstash);
11022 while (*s && strchr(valid_flags, *s))
11023 pmflag(&pm->op_pmflags,*s++);
11025 if (PL_madskills && modstart != s) {
11026 SV* tmptoken = newSVpvn(modstart, s - modstart);
11027 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11030 /* issue a warning if /c is specified,but /g is not */
11031 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
11032 && ckWARN(WARN_REGEXP))
11034 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
11035 "Use of /c modifier is meaningless without /g" );
11038 PL_lex_op = (OP*)pm;
11039 pl_yylval.ival = OP_MATCH;
11044 S_scan_subst(pTHX_ char *start)
11055 PERL_ARGS_ASSERT_SCAN_SUBST;
11057 pl_yylval.ival = OP_NULL;
11059 s = scan_str(start,!!PL_madskills,FALSE);
11062 Perl_croak(aTHX_ "Substitution pattern not terminated");
11064 if (s[-1] == PL_multi_open)
11067 if (PL_madskills) {
11068 CURMAD('q', PL_thisopen);
11069 CURMAD('_', PL_thiswhite);
11070 CURMAD('E', PL_thisstuff);
11071 CURMAD('Q', PL_thisclose);
11072 PL_realtokenstart = s - SvPVX(PL_linestr);
11076 first_start = PL_multi_start;
11077 s = scan_str(s,!!PL_madskills,FALSE);
11079 if (PL_lex_stuff) {
11080 SvREFCNT_dec(PL_lex_stuff);
11081 PL_lex_stuff = NULL;
11083 Perl_croak(aTHX_ "Substitution replacement not terminated");
11085 PL_multi_start = first_start; /* so whole substitution is taken together */
11087 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11090 if (PL_madskills) {
11091 CURMAD('z', PL_thisopen);
11092 CURMAD('R', PL_thisstuff);
11093 CURMAD('Z', PL_thisclose);
11099 if (*s == EXEC_PAT_MOD) {
11103 else if (strchr(S_PAT_MODS, *s))
11104 pmflag(&pm->op_pmflags,*s++);
11110 if (PL_madskills) {
11112 curmad('m', newSVpvn(modstart, s - modstart));
11113 append_madprops(PL_thismad, (OP*)pm, 0);
11117 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11118 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11122 SV * const repl = newSVpvs("");
11124 PL_sublex_info.super_bufptr = s;
11125 PL_sublex_info.super_bufend = PL_bufend;
11127 pm->op_pmflags |= PMf_EVAL;
11130 sv_catpvs(repl, "eval ");
11132 sv_catpvs(repl, "do ");
11134 sv_catpvs(repl, "{");
11135 sv_catsv(repl, PL_lex_repl);
11136 if (strchr(SvPVX(PL_lex_repl), '#'))
11137 sv_catpvs(repl, "\n");
11138 sv_catpvs(repl, "}");
11140 SvREFCNT_dec(PL_lex_repl);
11141 PL_lex_repl = repl;
11144 PL_lex_op = (OP*)pm;
11145 pl_yylval.ival = OP_SUBST;
11150 S_scan_trans(pTHX_ char *start)
11163 PERL_ARGS_ASSERT_SCAN_TRANS;
11165 pl_yylval.ival = OP_NULL;
11167 s = scan_str(start,!!PL_madskills,FALSE);
11169 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11171 if (s[-1] == PL_multi_open)
11174 if (PL_madskills) {
11175 CURMAD('q', PL_thisopen);
11176 CURMAD('_', PL_thiswhite);
11177 CURMAD('E', PL_thisstuff);
11178 CURMAD('Q', PL_thisclose);
11179 PL_realtokenstart = s - SvPVX(PL_linestr);
11183 s = scan_str(s,!!PL_madskills,FALSE);
11185 if (PL_lex_stuff) {
11186 SvREFCNT_dec(PL_lex_stuff);
11187 PL_lex_stuff = NULL;
11189 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11191 if (PL_madskills) {
11192 CURMAD('z', PL_thisopen);
11193 CURMAD('R', PL_thisstuff);
11194 CURMAD('Z', PL_thisclose);
11197 complement = del = squash = 0;
11204 complement = OPpTRANS_COMPLEMENT;
11207 del = OPpTRANS_DELETE;
11210 squash = OPpTRANS_SQUASH;
11219 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11220 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11221 o->op_private &= ~OPpTRANS_ALL;
11222 o->op_private |= del|squash|complement|
11223 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11224 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11227 pl_yylval.ival = OP_TRANS;
11230 if (PL_madskills) {
11232 curmad('m', newSVpvn(modstart, s - modstart));
11233 append_madprops(PL_thismad, o, 0);
11242 S_scan_heredoc(pTHX_ register char *s)
11246 I32 op_type = OP_SCALAR;
11250 const char *found_newline;
11254 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11256 I32 stuffstart = s - SvPVX(PL_linestr);
11259 PL_realtokenstart = -1;
11262 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11266 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11270 while (SPACE_OR_TAB(*peek))
11272 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11275 s = delimcpy(d, e, s, PL_bufend, term, &len);
11285 if (!isALNUM_lazy_if(s,UTF))
11286 deprecate_old("bare << to mean <<\"\"");
11287 for (; isALNUM_lazy_if(s,UTF); s++) {
11292 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11293 Perl_croak(aTHX_ "Delimiter for here document is too long");
11296 len = d - PL_tokenbuf;
11299 if (PL_madskills) {
11300 tstart = PL_tokenbuf + !outer;
11301 PL_thisclose = newSVpvn(tstart, len - !outer);
11302 tstart = SvPVX(PL_linestr) + stuffstart;
11303 PL_thisopen = newSVpvn(tstart, s - tstart);
11304 stuffstart = s - SvPVX(PL_linestr);
11307 #ifndef PERL_STRICT_CR
11308 d = strchr(s, '\r');
11310 char * const olds = s;
11312 while (s < PL_bufend) {
11318 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11327 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11334 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11335 herewas = newSVpvn(s,PL_bufend-s);
11339 herewas = newSVpvn(s-1,found_newline-s+1);
11342 herewas = newSVpvn(s,found_newline-s);
11346 if (PL_madskills) {
11347 tstart = SvPVX(PL_linestr) + stuffstart;
11349 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11351 PL_thisstuff = newSVpvn(tstart, s - tstart);
11354 s += SvCUR(herewas);
11357 stuffstart = s - SvPVX(PL_linestr);
11363 tmpstr = newSV_type(SVt_PVIV);
11364 SvGROW(tmpstr, 80);
11365 if (term == '\'') {
11366 op_type = OP_CONST;
11367 SvIV_set(tmpstr, -1);
11369 else if (term == '`') {
11370 op_type = OP_BACKTICK;
11371 SvIV_set(tmpstr, '\\');
11375 PL_multi_start = CopLINE(PL_curcop);
11376 PL_multi_open = PL_multi_close = '<';
11377 term = *PL_tokenbuf;
11378 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11379 char * const bufptr = PL_sublex_info.super_bufptr;
11380 char * const bufend = PL_sublex_info.super_bufend;
11381 char * const olds = s - SvCUR(herewas);
11382 s = strchr(bufptr, '\n');
11386 while (s < bufend &&
11387 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11389 CopLINE_inc(PL_curcop);
11392 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11393 missingterm(PL_tokenbuf);
11395 sv_setpvn(herewas,bufptr,d-bufptr+1);
11396 sv_setpvn(tmpstr,d+1,s-d);
11398 sv_catpvn(herewas,s,bufend-s);
11399 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11406 while (s < PL_bufend &&
11407 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11409 CopLINE_inc(PL_curcop);
11411 if (s >= PL_bufend) {
11412 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11413 missingterm(PL_tokenbuf);
11415 sv_setpvn(tmpstr,d+1,s-d);
11417 if (PL_madskills) {
11419 sv_catpvn(PL_thisstuff, d + 1, s - d);
11421 PL_thisstuff = newSVpvn(d + 1, s - d);
11422 stuffstart = s - SvPVX(PL_linestr);
11426 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11428 sv_catpvn(herewas,s,PL_bufend-s);
11429 sv_setsv(PL_linestr,herewas);
11430 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11431 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11432 PL_last_lop = PL_last_uni = NULL;
11435 sv_setpvs(tmpstr,""); /* avoid "uninitialized" warning */
11436 while (s >= PL_bufend) { /* multiple line string? */
11438 if (PL_madskills) {
11439 tstart = SvPVX(PL_linestr) + stuffstart;
11441 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11443 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11447 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11448 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11449 missingterm(PL_tokenbuf);
11452 stuffstart = s - SvPVX(PL_linestr);
11454 CopLINE_inc(PL_curcop);
11455 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11456 PL_last_lop = PL_last_uni = NULL;
11457 #ifndef PERL_STRICT_CR
11458 if (PL_bufend - PL_linestart >= 2) {
11459 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11460 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11462 PL_bufend[-2] = '\n';
11464 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11466 else if (PL_bufend[-1] == '\r')
11467 PL_bufend[-1] = '\n';
11469 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11470 PL_bufend[-1] = '\n';
11472 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11473 update_debugger_info(PL_linestr, NULL, 0);
11474 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11475 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11476 *(SvPVX(PL_linestr) + off ) = ' ';
11477 sv_catsv(PL_linestr,herewas);
11478 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11479 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11483 sv_catsv(tmpstr,PL_linestr);
11488 PL_multi_end = CopLINE(PL_curcop);
11489 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11490 SvPV_shrink_to_cur(tmpstr);
11492 SvREFCNT_dec(herewas);
11494 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11496 else if (PL_encoding)
11497 sv_recode_to_utf8(tmpstr, PL_encoding);
11499 PL_lex_stuff = tmpstr;
11500 pl_yylval.ival = op_type;
11504 /* scan_inputsymbol
11505 takes: current position in input buffer
11506 returns: new position in input buffer
11507 side-effects: pl_yylval and lex_op are set.
11512 <FH> read from filehandle
11513 <pkg::FH> read from package qualified filehandle
11514 <pkg'FH> read from package qualified filehandle
11515 <$fh> read from filehandle in $fh
11516 <*.h> filename glob
11521 S_scan_inputsymbol(pTHX_ char *start)
11524 register char *s = start; /* current position in buffer */
11527 char *d = PL_tokenbuf; /* start of temp holding space */
11528 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11530 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11532 end = strchr(s, '\n');
11535 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11537 /* die if we didn't have space for the contents of the <>,
11538 or if it didn't end, or if we see a newline
11541 if (len >= (I32)sizeof PL_tokenbuf)
11542 Perl_croak(aTHX_ "Excessively long <> operator");
11544 Perl_croak(aTHX_ "Unterminated <> operator");
11549 Remember, only scalar variables are interpreted as filehandles by
11550 this code. Anything more complex (e.g., <$fh{$num}>) will be
11551 treated as a glob() call.
11552 This code makes use of the fact that except for the $ at the front,
11553 a scalar variable and a filehandle look the same.
11555 if (*d == '$' && d[1]) d++;
11557 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11558 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11561 /* If we've tried to read what we allow filehandles to look like, and
11562 there's still text left, then it must be a glob() and not a getline.
11563 Use scan_str to pull out the stuff between the <> and treat it
11564 as nothing more than a string.
11567 if (d - PL_tokenbuf != len) {
11568 pl_yylval.ival = OP_GLOB;
11569 s = scan_str(start,!!PL_madskills,FALSE);
11571 Perl_croak(aTHX_ "Glob not terminated");
11575 bool readline_overriden = FALSE;
11578 /* we're in a filehandle read situation */
11581 /* turn <> into <ARGV> */
11583 Copy("ARGV",d,5,char);
11585 /* Check whether readline() is overriden */
11586 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11588 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11590 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11591 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11592 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11593 readline_overriden = TRUE;
11595 /* if <$fh>, create the ops to turn the variable into a
11599 /* try to find it in the pad for this block, otherwise find
11600 add symbol table ops
11602 const PADOFFSET tmp = pad_findmy(d);
11603 if (tmp != NOT_IN_PAD) {
11604 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11605 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11606 HEK * const stashname = HvNAME_HEK(stash);
11607 SV * const sym = sv_2mortal(newSVhek(stashname));
11608 sv_catpvs(sym, "::");
11609 sv_catpv(sym, d+1);
11614 OP * const o = newOP(OP_PADSV, 0);
11616 PL_lex_op = readline_overriden
11617 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11618 append_elem(OP_LIST, o,
11619 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11620 : (OP*)newUNOP(OP_READLINE, 0, o);
11629 ? (GV_ADDMULTI | GV_ADDINEVAL)
11632 PL_lex_op = readline_overriden
11633 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11634 append_elem(OP_LIST,
11635 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11636 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11637 : (OP*)newUNOP(OP_READLINE, 0,
11638 newUNOP(OP_RV2SV, 0,
11639 newGVOP(OP_GV, 0, gv)));
11641 if (!readline_overriden)
11642 PL_lex_op->op_flags |= OPf_SPECIAL;
11643 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11644 pl_yylval.ival = OP_NULL;
11647 /* If it's none of the above, it must be a literal filehandle
11648 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11650 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11651 PL_lex_op = readline_overriden
11652 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11653 append_elem(OP_LIST,
11654 newGVOP(OP_GV, 0, gv),
11655 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11656 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11657 pl_yylval.ival = OP_NULL;
11666 takes: start position in buffer
11667 keep_quoted preserve \ on the embedded delimiter(s)
11668 keep_delims preserve the delimiters around the string
11669 returns: position to continue reading from buffer
11670 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11671 updates the read buffer.
11673 This subroutine pulls a string out of the input. It is called for:
11674 q single quotes q(literal text)
11675 ' single quotes 'literal text'
11676 qq double quotes qq(interpolate $here please)
11677 " double quotes "interpolate $here please"
11678 qx backticks qx(/bin/ls -l)
11679 ` backticks `/bin/ls -l`
11680 qw quote words @EXPORT_OK = qw( func() $spam )
11681 m// regexp match m/this/
11682 s/// regexp substitute s/this/that/
11683 tr/// string transliterate tr/this/that/
11684 y/// string transliterate y/this/that/
11685 ($*@) sub prototypes sub foo ($)
11686 (stuff) sub attr parameters sub foo : attr(stuff)
11687 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11689 In most of these cases (all but <>, patterns and transliterate)
11690 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11691 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11692 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11695 It skips whitespace before the string starts, and treats the first
11696 character as the delimiter. If the delimiter is one of ([{< then
11697 the corresponding "close" character )]}> is used as the closing
11698 delimiter. It allows quoting of delimiters, and if the string has
11699 balanced delimiters ([{<>}]) it allows nesting.
11701 On success, the SV with the resulting string is put into lex_stuff or,
11702 if that is already non-NULL, into lex_repl. The second case occurs only
11703 when parsing the RHS of the special constructs s/// and tr/// (y///).
11704 For convenience, the terminating delimiter character is stuffed into
11709 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11712 SV *sv; /* scalar value: string */
11713 const char *tmps; /* temp string, used for delimiter matching */
11714 register char *s = start; /* current position in the buffer */
11715 register char term; /* terminating character */
11716 register char *to; /* current position in the sv's data */
11717 I32 brackets = 1; /* bracket nesting level */
11718 bool has_utf8 = FALSE; /* is there any utf8 content? */
11719 I32 termcode; /* terminating char. code */
11720 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11721 STRLEN termlen; /* length of terminating string */
11722 int last_off = 0; /* last position for nesting bracket */
11728 PERL_ARGS_ASSERT_SCAN_STR;
11730 /* skip space before the delimiter */
11736 if (PL_realtokenstart >= 0) {
11737 stuffstart = PL_realtokenstart;
11738 PL_realtokenstart = -1;
11741 stuffstart = start - SvPVX(PL_linestr);
11743 /* mark where we are, in case we need to report errors */
11746 /* after skipping whitespace, the next character is the terminator */
11749 termcode = termstr[0] = term;
11753 termcode = utf8_to_uvchr((U8*)s, &termlen);
11754 Copy(s, termstr, termlen, U8);
11755 if (!UTF8_IS_INVARIANT(term))
11759 /* mark where we are */
11760 PL_multi_start = CopLINE(PL_curcop);
11761 PL_multi_open = term;
11763 /* find corresponding closing delimiter */
11764 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11765 termcode = termstr[0] = term = tmps[5];
11767 PL_multi_close = term;
11769 /* create a new SV to hold the contents. 79 is the SV's initial length.
11770 What a random number. */
11771 sv = newSV_type(SVt_PVIV);
11773 SvIV_set(sv, termcode);
11774 (void)SvPOK_only(sv); /* validate pointer */
11776 /* move past delimiter and try to read a complete string */
11778 sv_catpvn(sv, s, termlen);
11781 tstart = SvPVX(PL_linestr) + stuffstart;
11782 if (!PL_thisopen && !keep_delims) {
11783 PL_thisopen = newSVpvn(tstart, s - tstart);
11784 stuffstart = s - SvPVX(PL_linestr);
11788 if (PL_encoding && !UTF) {
11792 int offset = s - SvPVX_const(PL_linestr);
11793 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11794 &offset, (char*)termstr, termlen);
11795 const char * const ns = SvPVX_const(PL_linestr) + offset;
11796 char * const svlast = SvEND(sv) - 1;
11798 for (; s < ns; s++) {
11799 if (*s == '\n' && !PL_rsfp)
11800 CopLINE_inc(PL_curcop);
11803 goto read_more_line;
11805 /* handle quoted delimiters */
11806 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11808 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11810 if ((svlast-1 - t) % 2) {
11811 if (!keep_quoted) {
11812 *(svlast-1) = term;
11814 SvCUR_set(sv, SvCUR(sv) - 1);
11819 if (PL_multi_open == PL_multi_close) {
11825 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11826 /* At here, all closes are "was quoted" one,
11827 so we don't check PL_multi_close. */
11829 if (!keep_quoted && *(t+1) == PL_multi_open)
11834 else if (*t == PL_multi_open)
11842 SvCUR_set(sv, w - SvPVX_const(sv));
11844 last_off = w - SvPVX(sv);
11845 if (--brackets <= 0)
11850 if (!keep_delims) {
11851 SvCUR_set(sv, SvCUR(sv) - 1);
11857 /* extend sv if need be */
11858 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11859 /* set 'to' to the next character in the sv's string */
11860 to = SvPVX(sv)+SvCUR(sv);
11862 /* if open delimiter is the close delimiter read unbridle */
11863 if (PL_multi_open == PL_multi_close) {
11864 for (; s < PL_bufend; s++,to++) {
11865 /* embedded newlines increment the current line number */
11866 if (*s == '\n' && !PL_rsfp)
11867 CopLINE_inc(PL_curcop);
11868 /* handle quoted delimiters */
11869 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11870 if (!keep_quoted && s[1] == term)
11872 /* any other quotes are simply copied straight through */
11876 /* terminate when run out of buffer (the for() condition), or
11877 have found the terminator */
11878 else if (*s == term) {
11881 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11884 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11890 /* if the terminator isn't the same as the start character (e.g.,
11891 matched brackets), we have to allow more in the quoting, and
11892 be prepared for nested brackets.
11895 /* read until we run out of string, or we find the terminator */
11896 for (; s < PL_bufend; s++,to++) {
11897 /* embedded newlines increment the line count */
11898 if (*s == '\n' && !PL_rsfp)
11899 CopLINE_inc(PL_curcop);
11900 /* backslashes can escape the open or closing characters */
11901 if (*s == '\\' && s+1 < PL_bufend) {
11902 if (!keep_quoted &&
11903 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11908 /* allow nested opens and closes */
11909 else if (*s == PL_multi_close && --brackets <= 0)
11911 else if (*s == PL_multi_open)
11913 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11918 /* terminate the copied string and update the sv's end-of-string */
11920 SvCUR_set(sv, to - SvPVX_const(sv));
11923 * this next chunk reads more into the buffer if we're not done yet
11927 break; /* handle case where we are done yet :-) */
11929 #ifndef PERL_STRICT_CR
11930 if (to - SvPVX_const(sv) >= 2) {
11931 if ((to[-2] == '\r' && to[-1] == '\n') ||
11932 (to[-2] == '\n' && to[-1] == '\r'))
11936 SvCUR_set(sv, to - SvPVX_const(sv));
11938 else if (to[-1] == '\r')
11941 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11946 /* if we're out of file, or a read fails, bail and reset the current
11947 line marker so we can report where the unterminated string began
11950 if (PL_madskills) {
11951 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11953 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11955 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11959 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11961 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11967 /* we read a line, so increment our line counter */
11968 CopLINE_inc(PL_curcop);
11970 /* update debugger info */
11971 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11972 update_debugger_info(PL_linestr, NULL, 0);
11974 /* having changed the buffer, we must update PL_bufend */
11975 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11976 PL_last_lop = PL_last_uni = NULL;
11979 /* at this point, we have successfully read the delimited string */
11981 if (!PL_encoding || UTF) {
11983 if (PL_madskills) {
11984 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11985 const int len = s - tstart;
11987 sv_catpvn(PL_thisstuff, tstart, len);
11989 PL_thisstuff = newSVpvn(tstart, len);
11990 if (!PL_thisclose && !keep_delims)
11991 PL_thisclose = newSVpvn(s,termlen);
11996 sv_catpvn(sv, s, termlen);
12001 if (PL_madskills) {
12002 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12003 const int len = s - tstart - termlen;
12005 sv_catpvn(PL_thisstuff, tstart, len);
12007 PL_thisstuff = newSVpvn(tstart, len);
12008 if (!PL_thisclose && !keep_delims)
12009 PL_thisclose = newSVpvn(s - termlen,termlen);
12013 if (has_utf8 || PL_encoding)
12016 PL_multi_end = CopLINE(PL_curcop);
12018 /* if we allocated too much space, give some back */
12019 if (SvCUR(sv) + 5 < SvLEN(sv)) {
12020 SvLEN_set(sv, SvCUR(sv) + 1);
12021 SvPV_renew(sv, SvLEN(sv));
12024 /* decide whether this is the first or second quoted string we've read
12037 takes: pointer to position in buffer
12038 returns: pointer to new position in buffer
12039 side-effects: builds ops for the constant in pl_yylval.op
12041 Read a number in any of the formats that Perl accepts:
12043 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12044 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12047 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12049 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12052 If it reads a number without a decimal point or an exponent, it will
12053 try converting the number to an integer and see if it can do so
12054 without loss of precision.
12058 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12061 register const char *s = start; /* current position in buffer */
12062 register char *d; /* destination in temp buffer */
12063 register char *e; /* end of temp buffer */
12064 NV nv; /* number read, as a double */
12065 SV *sv = NULL; /* place to put the converted number */
12066 bool floatit; /* boolean: int or float? */
12067 const char *lastub = NULL; /* position of last underbar */
12068 static char const number_too_long[] = "Number too long";
12070 PERL_ARGS_ASSERT_SCAN_NUM;
12072 /* We use the first character to decide what type of number this is */
12076 Perl_croak(aTHX_ "panic: scan_num");
12078 /* if it starts with a 0, it could be an octal number, a decimal in
12079 0.13 disguise, or a hexadecimal number, or a binary number. */
12083 u holds the "number so far"
12084 shift the power of 2 of the base
12085 (hex == 4, octal == 3, binary == 1)
12086 overflowed was the number more than we can hold?
12088 Shift is used when we add a digit. It also serves as an "are
12089 we in octal/hex/binary?" indicator to disallow hex characters
12090 when in octal mode.
12095 bool overflowed = FALSE;
12096 bool just_zero = TRUE; /* just plain 0 or binary number? */
12097 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12098 static const char* const bases[5] =
12099 { "", "binary", "", "octal", "hexadecimal" };
12100 static const char* const Bases[5] =
12101 { "", "Binary", "", "Octal", "Hexadecimal" };
12102 static const char* const maxima[5] =
12104 "0b11111111111111111111111111111111",
12108 const char *base, *Base, *max;
12110 /* check for hex */
12115 } else if (s[1] == 'b') {
12120 /* check for a decimal in disguise */
12121 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12123 /* so it must be octal */
12130 if (ckWARN(WARN_SYNTAX))
12131 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12132 "Misplaced _ in number");
12136 base = bases[shift];
12137 Base = Bases[shift];
12138 max = maxima[shift];
12140 /* read the rest of the number */
12142 /* x is used in the overflow test,
12143 b is the digit we're adding on. */
12148 /* if we don't mention it, we're done */
12152 /* _ are ignored -- but warned about if consecutive */
12154 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12155 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12156 "Misplaced _ in number");
12160 /* 8 and 9 are not octal */
12161 case '8': case '9':
12163 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12167 case '2': case '3': case '4':
12168 case '5': case '6': case '7':
12170 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12173 case '0': case '1':
12174 b = *s++ & 15; /* ASCII digit -> value of digit */
12178 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12179 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12180 /* make sure they said 0x */
12183 b = (*s++ & 7) + 9;
12185 /* Prepare to put the digit we have onto the end
12186 of the number so far. We check for overflows.
12192 x = u << shift; /* make room for the digit */
12194 if ((x >> shift) != u
12195 && !(PL_hints & HINT_NEW_BINARY)) {
12198 if (ckWARN_d(WARN_OVERFLOW))
12199 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12200 "Integer overflow in %s number",
12203 u = x | b; /* add the digit to the end */
12206 n *= nvshift[shift];
12207 /* If an NV has not enough bits in its
12208 * mantissa to represent an UV this summing of
12209 * small low-order numbers is a waste of time
12210 * (because the NV cannot preserve the
12211 * low-order bits anyway): we could just
12212 * remember when did we overflow and in the
12213 * end just multiply n by the right
12221 /* if we get here, we had success: make a scalar value from
12226 /* final misplaced underbar check */
12227 if (s[-1] == '_') {
12228 if (ckWARN(WARN_SYNTAX))
12229 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12234 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12235 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12236 "%s number > %s non-portable",
12242 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12243 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12244 "%s number > %s non-portable",
12249 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12250 sv = new_constant(start, s - start, "integer",
12251 sv, NULL, NULL, 0);
12252 else if (PL_hints & HINT_NEW_BINARY)
12253 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12258 handle decimal numbers.
12259 we're also sent here when we read a 0 as the first digit
12261 case '1': case '2': case '3': case '4': case '5':
12262 case '6': case '7': case '8': case '9': case '.':
12265 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12268 /* read next group of digits and _ and copy into d */
12269 while (isDIGIT(*s) || *s == '_') {
12270 /* skip underscores, checking for misplaced ones
12274 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12275 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12276 "Misplaced _ in number");
12280 /* check for end of fixed-length buffer */
12282 Perl_croak(aTHX_ number_too_long);
12283 /* if we're ok, copy the character */
12288 /* final misplaced underbar check */
12289 if (lastub && s == lastub + 1) {
12290 if (ckWARN(WARN_SYNTAX))
12291 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12294 /* read a decimal portion if there is one. avoid
12295 3..5 being interpreted as the number 3. followed
12298 if (*s == '.' && s[1] != '.') {
12303 if (ckWARN(WARN_SYNTAX))
12304 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12305 "Misplaced _ in number");
12309 /* copy, ignoring underbars, until we run out of digits.
12311 for (; isDIGIT(*s) || *s == '_'; s++) {
12312 /* fixed length buffer check */
12314 Perl_croak(aTHX_ number_too_long);
12316 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12317 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12318 "Misplaced _ in number");
12324 /* fractional part ending in underbar? */
12325 if (s[-1] == '_') {
12326 if (ckWARN(WARN_SYNTAX))
12327 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12328 "Misplaced _ in number");
12330 if (*s == '.' && isDIGIT(s[1])) {
12331 /* oops, it's really a v-string, but without the "v" */
12337 /* read exponent part, if present */
12338 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12342 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12343 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12345 /* stray preinitial _ */
12347 if (ckWARN(WARN_SYNTAX))
12348 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12349 "Misplaced _ in number");
12353 /* allow positive or negative exponent */
12354 if (*s == '+' || *s == '-')
12357 /* stray initial _ */
12359 if (ckWARN(WARN_SYNTAX))
12360 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12361 "Misplaced _ in number");
12365 /* read digits of exponent */
12366 while (isDIGIT(*s) || *s == '_') {
12369 Perl_croak(aTHX_ number_too_long);
12373 if (((lastub && s == lastub + 1) ||
12374 (!isDIGIT(s[1]) && s[1] != '_'))
12375 && ckWARN(WARN_SYNTAX))
12376 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12377 "Misplaced _ in number");
12384 /* make an sv from the string */
12388 We try to do an integer conversion first if no characters
12389 indicating "float" have been found.
12394 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12396 if (flags == IS_NUMBER_IN_UV) {
12398 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12401 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12402 if (uv <= (UV) IV_MIN)
12403 sv_setiv(sv, -(IV)uv);
12410 /* terminate the string */
12412 nv = Atof(PL_tokenbuf);
12417 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12418 const char *const key = floatit ? "float" : "integer";
12419 const STRLEN keylen = floatit ? 5 : 7;
12420 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12421 key, keylen, sv, NULL, NULL, 0);
12425 /* if it starts with a v, it could be a v-string */
12428 sv = newSV(5); /* preallocate storage space */
12429 s = scan_vstring(s, PL_bufend, sv);
12433 /* make the op for the constant and return */
12436 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12438 lvalp->opval = NULL;
12444 S_scan_formline(pTHX_ register char *s)
12447 register char *eol;
12449 SV * const stuff = newSVpvs("");
12450 bool needargs = FALSE;
12451 bool eofmt = FALSE;
12453 char *tokenstart = s;
12454 SV* savewhite = NULL;
12456 if (PL_madskills) {
12457 savewhite = PL_thiswhite;
12462 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12464 while (!needargs) {
12467 #ifdef PERL_STRICT_CR
12468 while (SPACE_OR_TAB(*t))
12471 while (SPACE_OR_TAB(*t) || *t == '\r')
12474 if (*t == '\n' || t == PL_bufend) {
12479 if (PL_in_eval && !PL_rsfp) {
12480 eol = (char *) memchr(s,'\n',PL_bufend-s);
12485 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12487 for (t = s; t < eol; t++) {
12488 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12490 goto enough; /* ~~ must be first line in formline */
12492 if (*t == '@' || *t == '^')
12496 sv_catpvn(stuff, s, eol-s);
12497 #ifndef PERL_STRICT_CR
12498 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12499 char *end = SvPVX(stuff) + SvCUR(stuff);
12502 SvCUR_set(stuff, SvCUR(stuff) - 1);
12512 if (PL_madskills) {
12514 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12516 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12519 s = filter_gets(PL_linestr, PL_rsfp, 0);
12521 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12523 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12525 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12526 PL_last_lop = PL_last_uni = NULL;
12535 if (SvCUR(stuff)) {
12538 PL_lex_state = LEX_NORMAL;
12539 start_force(PL_curforce);
12540 NEXTVAL_NEXTTOKE.ival = 0;
12544 PL_lex_state = LEX_FORMLINE;
12546 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12548 else if (PL_encoding)
12549 sv_recode_to_utf8(stuff, PL_encoding);
12551 start_force(PL_curforce);
12552 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12554 start_force(PL_curforce);
12555 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12559 SvREFCNT_dec(stuff);
12561 PL_lex_formbrack = 0;
12565 if (PL_madskills) {
12567 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12569 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12570 PL_thiswhite = savewhite;
12577 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12580 const I32 oldsavestack_ix = PL_savestack_ix;
12581 CV* const outsidecv = PL_compcv;
12584 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12586 SAVEI32(PL_subline);
12587 save_item(PL_subname);
12588 SAVESPTR(PL_compcv);
12590 PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
12591 CvFLAGS(PL_compcv) |= flags;
12593 PL_subline = CopLINE(PL_curcop);
12594 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12595 CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
12596 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12598 return oldsavestack_ix;
12602 #pragma segment Perl_yylex
12605 S_yywarn(pTHX_ const char *const s)
12609 PERL_ARGS_ASSERT_YYWARN;
12611 PL_in_eval |= EVAL_WARNONLY;
12613 PL_in_eval &= ~EVAL_WARNONLY;
12618 Perl_yyerror(pTHX_ const char *const s)
12621 const char *where = NULL;
12622 const char *context = NULL;
12625 int yychar = PL_parser->yychar;
12627 PERL_ARGS_ASSERT_YYERROR;
12629 if (!yychar || (yychar == ';' && !PL_rsfp))
12631 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12632 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12633 PL_oldbufptr != PL_bufptr) {
12636 The code below is removed for NetWare because it abends/crashes on NetWare
12637 when the script has error such as not having the closing quotes like:
12638 if ($var eq "value)
12639 Checking of white spaces is anyway done in NetWare code.
12642 while (isSPACE(*PL_oldoldbufptr))
12645 context = PL_oldoldbufptr;
12646 contlen = PL_bufptr - PL_oldoldbufptr;
12648 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12649 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12652 The code below is removed for NetWare because it abends/crashes on NetWare
12653 when the script has error such as not having the closing quotes like:
12654 if ($var eq "value)
12655 Checking of white spaces is anyway done in NetWare code.
12658 while (isSPACE(*PL_oldbufptr))
12661 context = PL_oldbufptr;
12662 contlen = PL_bufptr - PL_oldbufptr;
12664 else if (yychar > 255)
12665 where = "next token ???";
12666 else if (yychar == -2) { /* YYEMPTY */
12667 if (PL_lex_state == LEX_NORMAL ||
12668 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12669 where = "at end of line";
12670 else if (PL_lex_inpat)
12671 where = "within pattern";
12673 where = "within string";
12676 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12678 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12679 else if (isPRINT_LC(yychar)) {
12680 const char string = yychar;
12681 sv_catpvn(where_sv, &string, 1);
12684 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12685 where = SvPVX_const(where_sv);
12687 msg = sv_2mortal(newSVpv(s, 0));
12688 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12689 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12691 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12693 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12694 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12695 Perl_sv_catpvf(aTHX_ msg,
12696 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12697 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12700 if (PL_in_eval & EVAL_WARNONLY) {
12701 if (ckWARN_d(WARN_SYNTAX))
12702 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12706 if (PL_error_count >= 10) {
12707 if (PL_in_eval && SvCUR(ERRSV))
12708 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12709 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12711 Perl_croak(aTHX_ "%s has too many errors.\n",
12712 OutCopFILE(PL_curcop));
12715 PL_in_my_stash = NULL;
12719 #pragma segment Main
12723 S_swallow_bom(pTHX_ U8 *s)
12726 const STRLEN slen = SvCUR(PL_linestr);
12728 PERL_ARGS_ASSERT_SWALLOW_BOM;
12732 if (s[1] == 0xFE) {
12733 /* UTF-16 little-endian? (or UTF32-LE?) */
12734 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12735 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12736 #ifndef PERL_NO_UTF16_FILTER
12737 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12740 if (PL_bufend > (char*)s) {
12744 filter_add(utf16rev_textfilter, NULL);
12745 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12746 utf16_to_utf8_reversed(s, news,
12747 PL_bufend - (char*)s - 1,
12749 sv_setpvn(PL_linestr, (const char*)news, newlen);
12751 s = (U8*)SvPVX(PL_linestr);
12752 Copy(news, s, newlen, U8);
12756 SvUTF8_on(PL_linestr);
12757 s = (U8*)SvPVX(PL_linestr);
12759 /* FIXME - is this a general bug fix? */
12762 PL_bufend = SvPVX(PL_linestr) + newlen;
12765 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12770 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12771 #ifndef PERL_NO_UTF16_FILTER
12772 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12775 if (PL_bufend > (char *)s) {
12779 filter_add(utf16_textfilter, NULL);
12780 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12781 utf16_to_utf8(s, news,
12782 PL_bufend - (char*)s,
12784 sv_setpvn(PL_linestr, (const char*)news, newlen);
12786 SvUTF8_on(PL_linestr);
12787 s = (U8*)SvPVX(PL_linestr);
12788 PL_bufend = SvPVX(PL_linestr) + newlen;
12791 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12796 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12797 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12798 s += 3; /* UTF-8 */
12804 if (s[2] == 0xFE && s[3] == 0xFF) {
12805 /* UTF-32 big-endian */
12806 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12809 else if (s[2] == 0 && s[3] != 0) {
12812 * are a good indicator of UTF-16BE. */
12813 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12819 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12820 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12821 s += 4; /* UTF-8 */
12827 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12830 * are a good indicator of UTF-16LE. */
12831 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12839 #ifndef PERL_NO_UTF16_FILTER
12841 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12844 const STRLEN old = SvCUR(sv);
12845 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12846 DEBUG_P(PerlIO_printf(Perl_debug_log,
12847 "utf16_textfilter(%p): %d %d (%d)\n",
12848 FPTR2DPTR(void *, utf16_textfilter),
12849 idx, maxlen, (int) count));
12853 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12854 Copy(SvPVX_const(sv), tmps, old, char);
12855 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12856 SvCUR(sv) - old, &newlen);
12857 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12859 DEBUG_P({sv_dump(sv);});
12864 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12867 const STRLEN old = SvCUR(sv);
12868 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12869 DEBUG_P(PerlIO_printf(Perl_debug_log,
12870 "utf16rev_textfilter(%p): %d %d (%d)\n",
12871 FPTR2DPTR(void *, utf16rev_textfilter),
12872 idx, maxlen, (int) count));
12876 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12877 Copy(SvPVX_const(sv), tmps, old, char);
12878 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12879 SvCUR(sv) - old, &newlen);
12880 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12882 DEBUG_P({ sv_dump(sv); });
12888 Returns a pointer to the next character after the parsed
12889 vstring, as well as updating the passed in sv.
12891 Function must be called like
12894 s = scan_vstring(s,e,sv);
12896 where s and e are the start and end of the string.
12897 The sv should already be large enough to store the vstring
12898 passed in, for performance reasons.
12903 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12906 const char *pos = s;
12907 const char *start = s;
12909 PERL_ARGS_ASSERT_SCAN_VSTRING;
12911 if (*pos == 'v') pos++; /* get past 'v' */
12912 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12914 if ( *pos != '.') {
12915 /* this may not be a v-string if followed by => */
12916 const char *next = pos;
12917 while (next < e && isSPACE(*next))
12919 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12920 /* return string not v-string */
12921 sv_setpvn(sv,(char *)s,pos-s);
12922 return (char *)pos;
12926 if (!isALPHA(*pos)) {
12927 U8 tmpbuf[UTF8_MAXBYTES+1];
12930 s++; /* get past 'v' */
12935 /* this is atoi() that tolerates underscores */
12938 const char *end = pos;
12940 while (--end >= s) {
12942 const UV orev = rev;
12943 rev += (*end - '0') * mult;
12945 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12946 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12947 "Integer overflow in decimal number");
12951 if (rev > 0x7FFFFFFF)
12952 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12954 /* Append native character for the rev point */
12955 tmpend = uvchr_to_utf8(tmpbuf, rev);
12956 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12957 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12959 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12965 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12969 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12977 * c-indentation-style: bsd
12978 * c-basic-offset: 4
12979 * indent-tabs-mode: t
12982 * ex: set ts=8 sts=4 sw=4 noet: