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 /* See Note on sizing above. */
2310 sv_utf8_upgrade_flags_grow(sv,
2311 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2312 UNISKIP(uv) + (STRLEN)(send - s) + 1);
2313 d = SvPVX(sv) + SvCUR(sv);
2318 d = (char*)uvuni_to_utf8((U8*)d, uv);
2319 if (PL_lex_inwhat == OP_TRANS &&
2320 PL_sublex_info.sub_op) {
2321 PL_sublex_info.sub_op->op_private |=
2322 (PL_lex_repl ? OPpTRANS_FROM_UTF
2326 if (uv > 255 && !dorange)
2327 native_range = FALSE;
2339 /* \N{LATIN SMALL LETTER A} is a named character, and so is
2344 char* e = strchr(s, '}');
2350 yyerror("Missing right brace on \\N{}");
2354 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2355 /* \N{U+...} The ... is a unicode value even on EBCDIC
2357 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2358 PERL_SCAN_DISALLOW_PREFIX;
2361 uv = grok_hex(s, &len, &flags, NULL);
2362 if ( e > s && len != (STRLEN)(e - s) ) {
2366 goto NUM_ESCAPE_INSERT;
2368 res = newSVpvn(s + 1, e - s - 1);
2369 res = new_constant( NULL, 0, "charnames",
2370 res, NULL, s - 2, e - s + 3 );
2372 sv_utf8_upgrade(res);
2373 str = SvPV_const(res,len);
2374 #ifdef EBCDIC_NEVER_MIND
2375 /* charnames uses pack U and that has been
2376 * recently changed to do the below uni->native
2377 * mapping, so this would be redundant (and wrong,
2378 * the code point would be doubly converted).
2379 * But leave this in just in case the pack U change
2380 * gets revoked, but the semantics is still
2381 * desireable for charnames. --jhi */
2383 UV uv = utf8_to_uvchr((const U8*)str, 0);
2386 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2388 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2389 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2390 str = SvPV_const(res, len);
2394 /* If destination is not in utf8 but this new character is,
2395 * recode the dest to utf8 */
2396 if (!has_utf8 && SvUTF8(res)) {
2397 SvCUR_set(sv, d - SvPVX_const(sv));
2400 /* See Note on sizing above. */
2401 sv_utf8_upgrade_flags_grow(sv,
2402 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2403 len + (STRLEN)(send - s) + 1);
2404 d = SvPVX(sv) + SvCUR(sv);
2406 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2408 /* See Note on sizing above. (NOTE: SvCUR() is not set
2409 * correctly here). */
2410 const STRLEN off = d - SvPVX_const(sv);
2411 d = SvGROW(sv, off + len + (STRLEN)(send - s) + 1) + off;
2415 native_range = FALSE; /* \N{} is guessed to be Unicode */
2417 Copy(str, d, len, char);
2424 yyerror("Missing braces on \\N{}");
2427 /* \c is a control character */
2436 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2439 yyerror("Missing control char name in \\c");
2443 /* printf-style backslashes, formfeeds, newlines, etc */
2445 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2448 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2451 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2454 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2457 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2460 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2463 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2469 } /* end if (backslash) */
2476 /* If we started with encoded form, or already know we want it,
2477 then encode the next character */
2478 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
2482 /* One might think that it is wasted effort in the case of the
2483 * source being utf8 (this_utf8 == TRUE) to take the next character
2484 * in the source, convert it to an unsigned value, and then convert
2485 * it back again. But the source has not been validated here. The
2486 * routine that does the conversion checks for errors like
2489 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2490 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2492 SvCUR_set(sv, d - SvPVX_const(sv));
2495 /* See Note on sizing above. */
2496 sv_utf8_upgrade_flags_grow(sv,
2497 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2498 need + (STRLEN)(send - s) + 1);
2499 d = SvPVX(sv) + SvCUR(sv);
2501 } else if (need > len) {
2502 /* encoded value larger than old, may need extra space (NOTE:
2503 * SvCUR() is not set correctly here). See Note on sizing
2505 const STRLEN off = d - SvPVX_const(sv);
2506 d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
2510 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2512 if (uv > 255 && !dorange)
2513 native_range = FALSE;
2517 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2519 } /* while loop to process each character */
2521 /* terminate the string and set up the sv */
2523 SvCUR_set(sv, d - SvPVX_const(sv));
2524 if (SvCUR(sv) >= SvLEN(sv))
2525 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2528 if (PL_encoding && !has_utf8) {
2529 sv_recode_to_utf8(sv, PL_encoding);
2535 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2536 PL_sublex_info.sub_op->op_private |=
2537 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2541 /* shrink the sv if we allocated more than we used */
2542 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2543 SvPV_shrink_to_cur(sv);
2546 /* return the substring (via pl_yylval) only if we parsed anything */
2547 if (s > PL_bufptr) {
2548 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2549 const char *const key = PL_lex_inpat ? "qr" : "q";
2550 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2554 if (PL_lex_inwhat == OP_TRANS) {
2557 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2565 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2568 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2575 * Returns TRUE if there's more to the expression (e.g., a subscript),
2578 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2580 * ->[ and ->{ return TRUE
2581 * { and [ outside a pattern are always subscripts, so return TRUE
2582 * if we're outside a pattern and it's not { or [, then return FALSE
2583 * if we're in a pattern and the first char is a {
2584 * {4,5} (any digits around the comma) returns FALSE
2585 * if we're in a pattern and the first char is a [
2587 * [SOMETHING] has a funky algorithm to decide whether it's a
2588 * character class or not. It has to deal with things like
2589 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2590 * anything else returns TRUE
2593 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2596 S_intuit_more(pTHX_ register char *s)
2600 PERL_ARGS_ASSERT_INTUIT_MORE;
2602 if (PL_lex_brackets)
2604 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2606 if (*s != '{' && *s != '[')
2611 /* In a pattern, so maybe we have {n,m}. */
2628 /* On the other hand, maybe we have a character class */
2631 if (*s == ']' || *s == '^')
2634 /* this is terrifying, and it works */
2635 int weight = 2; /* let's weigh the evidence */
2637 unsigned char un_char = 255, last_un_char;
2638 const char * const send = strchr(s,']');
2639 char tmpbuf[sizeof PL_tokenbuf * 4];
2641 if (!send) /* has to be an expression */
2644 Zero(seen,256,char);
2647 else if (isDIGIT(*s)) {
2649 if (isDIGIT(s[1]) && s[2] == ']')
2655 for (; s < send; s++) {
2656 last_un_char = un_char;
2657 un_char = (unsigned char)*s;
2662 weight -= seen[un_char] * 10;
2663 if (isALNUM_lazy_if(s+1,UTF)) {
2665 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2666 len = (int)strlen(tmpbuf);
2667 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2672 else if (*s == '$' && s[1] &&
2673 strchr("[#!%*<>()-=",s[1])) {
2674 if (/*{*/ strchr("])} =",s[2]))
2683 if (strchr("wds]",s[1]))
2685 else if (seen[(U8)'\''] || seen[(U8)'"'])
2687 else if (strchr("rnftbxcav",s[1]))
2689 else if (isDIGIT(s[1])) {
2691 while (s[1] && isDIGIT(s[1]))
2701 if (strchr("aA01! ",last_un_char))
2703 if (strchr("zZ79~",s[1]))
2705 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2706 weight -= 5; /* cope with negative subscript */
2709 if (!isALNUM(last_un_char)
2710 && !(last_un_char == '$' || last_un_char == '@'
2711 || last_un_char == '&')
2712 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2717 if (keyword(tmpbuf, d - tmpbuf, 0))
2720 if (un_char == last_un_char + 1)
2722 weight -= seen[un_char];
2727 if (weight >= 0) /* probably a character class */
2737 * Does all the checking to disambiguate
2739 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2740 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2742 * First argument is the stuff after the first token, e.g. "bar".
2744 * Not a method if bar is a filehandle.
2745 * Not a method if foo is a subroutine prototyped to take a filehandle.
2746 * Not a method if it's really "Foo $bar"
2747 * Method if it's "foo $bar"
2748 * Not a method if it's really "print foo $bar"
2749 * Method if it's really "foo package::" (interpreted as package->foo)
2750 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2751 * Not a method if bar is a filehandle or package, but is quoted with
2756 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2759 char *s = start + (*start == '$');
2760 char tmpbuf[sizeof PL_tokenbuf];
2767 PERL_ARGS_ASSERT_INTUIT_METHOD;
2770 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2774 const char *proto = SvPVX_const(cv);
2785 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2786 /* start is the beginning of the possible filehandle/object,
2787 * and s is the end of it
2788 * tmpbuf is a copy of it
2791 if (*start == '$') {
2792 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2793 isUPPER(*PL_tokenbuf))
2796 len = start - SvPVX(PL_linestr);
2800 start = SvPVX(PL_linestr) + len;
2804 return *s == '(' ? FUNCMETH : METHOD;
2806 if (!keyword(tmpbuf, len, 0)) {
2807 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2811 soff = s - SvPVX(PL_linestr);
2815 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2816 if (indirgv && GvCVu(indirgv))
2818 /* filehandle or package name makes it a method */
2819 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2821 soff = s - SvPVX(PL_linestr);
2824 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2825 return 0; /* no assumptions -- "=>" quotes bearword */
2827 start_force(PL_curforce);
2828 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2829 newSVpvn(tmpbuf,len));
2830 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2832 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2837 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2839 return *s == '(' ? FUNCMETH : METHOD;
2845 /* Encoded script support. filter_add() effectively inserts a
2846 * 'pre-processing' function into the current source input stream.
2847 * Note that the filter function only applies to the current source file
2848 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2850 * The datasv parameter (which may be NULL) can be used to pass
2851 * private data to this instance of the filter. The filter function
2852 * can recover the SV using the FILTER_DATA macro and use it to
2853 * store private buffers and state information.
2855 * The supplied datasv parameter is upgraded to a PVIO type
2856 * and the IoDIRP/IoANY field is used to store the function pointer,
2857 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2858 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2859 * private use must be set using malloc'd pointers.
2863 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2872 if (!PL_rsfp_filters)
2873 PL_rsfp_filters = newAV();
2876 SvUPGRADE(datasv, SVt_PVIO);
2877 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2878 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2879 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2880 FPTR2DPTR(void *, IoANY(datasv)),
2881 SvPV_nolen(datasv)));
2882 av_unshift(PL_rsfp_filters, 1);
2883 av_store(PL_rsfp_filters, 0, datasv) ;
2888 /* Delete most recently added instance of this filter function. */
2890 Perl_filter_del(pTHX_ filter_t funcp)
2895 PERL_ARGS_ASSERT_FILTER_DEL;
2898 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2899 FPTR2DPTR(void*, funcp)));
2901 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2903 /* if filter is on top of stack (usual case) just pop it off */
2904 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2905 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2906 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2907 IoANY(datasv) = (void *)NULL;
2908 sv_free(av_pop(PL_rsfp_filters));
2912 /* we need to search for the correct entry and clear it */
2913 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2917 /* Invoke the idxth filter function for the current rsfp. */
2918 /* maxlen 0 = read one text line */
2920 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2925 /* This API is bad. It should have been using unsigned int for maxlen.
2926 Not sure if we want to change the API, but if not we should sanity
2927 check the value here. */
2928 const unsigned int correct_length
2937 PERL_ARGS_ASSERT_FILTER_READ;
2939 if (!PL_parser || !PL_rsfp_filters)
2941 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2942 /* Provide a default input filter to make life easy. */
2943 /* Note that we append to the line. This is handy. */
2944 DEBUG_P(PerlIO_printf(Perl_debug_log,
2945 "filter_read %d: from rsfp\n", idx));
2946 if (correct_length) {
2949 const int old_len = SvCUR(buf_sv);
2951 /* ensure buf_sv is large enough */
2952 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2953 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2954 correct_length)) <= 0) {
2955 if (PerlIO_error(PL_rsfp))
2956 return -1; /* error */
2958 return 0 ; /* end of file */
2960 SvCUR_set(buf_sv, old_len + len) ;
2963 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2964 if (PerlIO_error(PL_rsfp))
2965 return -1; /* error */
2967 return 0 ; /* end of file */
2970 return SvCUR(buf_sv);
2972 /* Skip this filter slot if filter has been deleted */
2973 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2974 DEBUG_P(PerlIO_printf(Perl_debug_log,
2975 "filter_read %d: skipped (filter deleted)\n",
2977 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2979 /* Get function pointer hidden within datasv */
2980 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2981 DEBUG_P(PerlIO_printf(Perl_debug_log,
2982 "filter_read %d: via function %p (%s)\n",
2983 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2984 /* Call function. The function is expected to */
2985 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2986 /* Return: <0:error, =0:eof, >0:not eof */
2987 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2991 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2995 PERL_ARGS_ASSERT_FILTER_GETS;
2997 #ifdef PERL_CR_FILTER
2998 if (!PL_rsfp_filters) {
2999 filter_add(S_cr_textfilter,NULL);
3002 if (PL_rsfp_filters) {
3004 SvCUR_set(sv, 0); /* start with empty line */
3005 if (FILTER_READ(0, sv, 0) > 0)
3006 return ( SvPVX(sv) ) ;
3011 return (sv_gets(sv, fp, append));
3015 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
3020 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
3022 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
3026 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
3027 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3029 return GvHV(gv); /* Foo:: */
3032 /* use constant CLASS => 'MyClass' */
3033 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3034 if (gv && GvCV(gv)) {
3035 SV * const sv = cv_const_sv(GvCV(gv));
3037 pkgname = SvPV_const(sv, len);
3040 return gv_stashpvn(pkgname, len, 0);
3044 * S_readpipe_override
3045 * Check whether readpipe() is overriden, and generates the appropriate
3046 * optree, provided sublex_start() is called afterwards.
3049 S_readpipe_override(pTHX)
3052 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3053 pl_yylval.ival = OP_BACKTICK;
3055 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3057 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3058 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3059 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3061 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3062 append_elem(OP_LIST,
3063 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3064 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3071 * The intent of this yylex wrapper is to minimize the changes to the
3072 * tokener when we aren't interested in collecting madprops. It remains
3073 * to be seen how successful this strategy will be...
3080 char *s = PL_bufptr;
3082 /* make sure PL_thiswhite is initialized */
3086 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3087 if (PL_pending_ident)
3088 return S_pending_ident(aTHX);
3090 /* previous token ate up our whitespace? */
3091 if (!PL_lasttoke && PL_nextwhite) {
3092 PL_thiswhite = PL_nextwhite;
3096 /* isolate the token, and figure out where it is without whitespace */
3097 PL_realtokenstart = -1;
3101 assert(PL_curforce < 0);
3103 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3104 if (!PL_thistoken) {
3105 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3106 PL_thistoken = newSVpvs("");
3108 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3109 PL_thistoken = newSVpvn(tstart, s - tstart);
3112 if (PL_thismad) /* install head */
3113 CURMAD('X', PL_thistoken);
3116 /* last whitespace of a sublex? */
3117 if (optype == ')' && PL_endwhite) {
3118 CURMAD('X', PL_endwhite);
3123 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3124 if (!PL_thiswhite && !PL_endwhite && !optype) {
3125 sv_free(PL_thistoken);
3130 /* put off final whitespace till peg */
3131 if (optype == ';' && !PL_rsfp) {
3132 PL_nextwhite = PL_thiswhite;
3135 else if (PL_thisopen) {
3136 CURMAD('q', PL_thisopen);
3138 sv_free(PL_thistoken);
3142 /* Store actual token text as madprop X */
3143 CURMAD('X', PL_thistoken);
3147 /* add preceding whitespace as madprop _ */
3148 CURMAD('_', PL_thiswhite);
3152 /* add quoted material as madprop = */
3153 CURMAD('=', PL_thisstuff);
3157 /* add terminating quote as madprop Q */
3158 CURMAD('Q', PL_thisclose);
3162 /* special processing based on optype */
3166 /* opval doesn't need a TOKEN since it can already store mp */
3176 if (pl_yylval.opval)
3177 append_madprops(PL_thismad, pl_yylval.opval, 0);
3185 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3194 /* remember any fake bracket that lexer is about to discard */
3195 if (PL_lex_brackets == 1 &&
3196 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3199 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3202 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3203 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3206 break; /* don't bother looking for trailing comment */
3215 /* attach a trailing comment to its statement instead of next token */
3219 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3221 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3223 if (*s == '\n' || *s == '#') {
3224 while (s < PL_bufend && *s != '\n')
3228 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3229 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3246 /* Create new token struct. Note: opvals return early above. */
3247 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3254 S_tokenize_use(pTHX_ int is_use, char *s) {
3257 PERL_ARGS_ASSERT_TOKENIZE_USE;
3259 if (PL_expect != XSTATE)
3260 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3261 is_use ? "use" : "no"));
3263 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3264 s = force_version(s, TRUE);
3265 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3266 start_force(PL_curforce);
3267 NEXTVAL_NEXTTOKE.opval = NULL;
3270 else if (*s == 'v') {
3271 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3272 s = force_version(s, FALSE);
3276 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3277 s = force_version(s, FALSE);
3279 pl_yylval.ival = is_use;
3283 static const char* const exp_name[] =
3284 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3285 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3292 Works out what to call the token just pulled out of the input
3293 stream. The yacc parser takes care of taking the ops we return and
3294 stitching them into a tree.
3300 if read an identifier
3301 if we're in a my declaration
3302 croak if they tried to say my($foo::bar)
3303 build the ops for a my() declaration
3304 if it's an access to a my() variable
3305 are we in a sort block?
3306 croak if my($a); $a <=> $b
3307 build ops for access to a my() variable
3308 if in a dq string, and they've said @foo and we can't find @foo
3310 build ops for a bareword
3311 if we already built the token before, use it.
3316 #pragma segment Perl_yylex
3322 register char *s = PL_bufptr;
3327 /* orig_keyword, gvp, and gv are initialized here because
3328 * jump to the label just_a_word_zero can bypass their
3329 * initialization later. */
3330 I32 orig_keyword = 0;
3335 SV* tmp = newSVpvs("");
3336 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3337 (IV)CopLINE(PL_curcop),
3338 lex_state_names[PL_lex_state],
3339 exp_name[PL_expect],
3340 pv_display(tmp, s, strlen(s), 0, 60));
3343 /* check if there's an identifier for us to look at */
3344 if (PL_pending_ident)
3345 return REPORT(S_pending_ident(aTHX));
3347 /* no identifier pending identification */
3349 switch (PL_lex_state) {
3351 case LEX_NORMAL: /* Some compilers will produce faster */
3352 case LEX_INTERPNORMAL: /* code if we comment these out. */
3356 /* when we've already built the next token, just pull it out of the queue */
3360 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3362 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3363 PL_nexttoke[PL_lasttoke].next_mad = 0;
3364 if (PL_thismad && PL_thismad->mad_key == '_') {
3365 PL_thiswhite = MUTABLE_SV(PL_thismad->mad_val);
3366 PL_thismad->mad_val = 0;
3367 mad_free(PL_thismad);
3372 PL_lex_state = PL_lex_defer;
3373 PL_expect = PL_lex_expect;
3374 PL_lex_defer = LEX_NORMAL;
3375 if (!PL_nexttoke[PL_lasttoke].next_type)
3380 pl_yylval = PL_nextval[PL_nexttoke];
3382 PL_lex_state = PL_lex_defer;
3383 PL_expect = PL_lex_expect;
3384 PL_lex_defer = LEX_NORMAL;
3388 /* FIXME - can these be merged? */
3389 return(PL_nexttoke[PL_lasttoke].next_type);
3391 return REPORT(PL_nexttype[PL_nexttoke]);
3394 /* interpolated case modifiers like \L \U, including \Q and \E.
3395 when we get here, PL_bufptr is at the \
3397 case LEX_INTERPCASEMOD:
3399 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3400 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3402 /* handle \E or end of string */
3403 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3405 if (PL_lex_casemods) {
3406 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3407 PL_lex_casestack[PL_lex_casemods] = '\0';
3409 if (PL_bufptr != PL_bufend
3410 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3412 PL_lex_state = LEX_INTERPCONCAT;
3415 PL_thistoken = newSVpvs("\\E");
3421 while (PL_bufptr != PL_bufend &&
3422 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3424 PL_thiswhite = newSVpvs("");
3425 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3429 if (PL_bufptr != PL_bufend)
3432 PL_lex_state = LEX_INTERPCONCAT;
3436 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3437 "### Saw case modifier\n"); });
3439 if (s[1] == '\\' && s[2] == 'E') {
3442 PL_thiswhite = newSVpvs("");
3443 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3446 PL_lex_state = LEX_INTERPCONCAT;
3451 if (!PL_madskills) /* when just compiling don't need correct */
3452 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3453 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3454 if ((*s == 'L' || *s == 'U') &&
3455 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3456 PL_lex_casestack[--PL_lex_casemods] = '\0';
3459 if (PL_lex_casemods > 10)
3460 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3461 PL_lex_casestack[PL_lex_casemods++] = *s;
3462 PL_lex_casestack[PL_lex_casemods] = '\0';
3463 PL_lex_state = LEX_INTERPCONCAT;
3464 start_force(PL_curforce);
3465 NEXTVAL_NEXTTOKE.ival = 0;
3467 start_force(PL_curforce);
3469 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3471 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3473 NEXTVAL_NEXTTOKE.ival = OP_LC;
3475 NEXTVAL_NEXTTOKE.ival = OP_UC;
3477 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3479 Perl_croak(aTHX_ "panic: yylex");
3481 SV* const tmpsv = newSVpvs("\\ ");
3482 /* replace the space with the character we want to escape
3484 SvPVX(tmpsv)[1] = *s;
3490 if (PL_lex_starts) {
3496 sv_free(PL_thistoken);
3497 PL_thistoken = newSVpvs("");
3500 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3501 if (PL_lex_casemods == 1 && PL_lex_inpat)
3510 case LEX_INTERPPUSH:
3511 return REPORT(sublex_push());
3513 case LEX_INTERPSTART:
3514 if (PL_bufptr == PL_bufend)
3515 return REPORT(sublex_done());
3516 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3517 "### Interpolated variable\n"); });
3519 PL_lex_dojoin = (*PL_bufptr == '@');
3520 PL_lex_state = LEX_INTERPNORMAL;
3521 if (PL_lex_dojoin) {
3522 start_force(PL_curforce);
3523 NEXTVAL_NEXTTOKE.ival = 0;
3525 start_force(PL_curforce);
3526 force_ident("\"", '$');
3527 start_force(PL_curforce);
3528 NEXTVAL_NEXTTOKE.ival = 0;
3530 start_force(PL_curforce);
3531 NEXTVAL_NEXTTOKE.ival = 0;
3533 start_force(PL_curforce);
3534 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3537 if (PL_lex_starts++) {
3542 sv_free(PL_thistoken);
3543 PL_thistoken = newSVpvs("");
3546 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3547 if (!PL_lex_casemods && PL_lex_inpat)
3554 case LEX_INTERPENDMAYBE:
3555 if (intuit_more(PL_bufptr)) {
3556 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3562 if (PL_lex_dojoin) {
3563 PL_lex_dojoin = FALSE;
3564 PL_lex_state = LEX_INTERPCONCAT;
3568 sv_free(PL_thistoken);
3569 PL_thistoken = newSVpvs("");
3574 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3575 && SvEVALED(PL_lex_repl))
3577 if (PL_bufptr != PL_bufend)
3578 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3582 case LEX_INTERPCONCAT:
3584 if (PL_lex_brackets)
3585 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3587 if (PL_bufptr == PL_bufend)
3588 return REPORT(sublex_done());
3590 if (SvIVX(PL_linestr) == '\'') {
3591 SV *sv = newSVsv(PL_linestr);
3594 else if ( PL_hints & HINT_NEW_RE )
3595 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3596 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3600 s = scan_const(PL_bufptr);
3602 PL_lex_state = LEX_INTERPCASEMOD;
3604 PL_lex_state = LEX_INTERPSTART;
3607 if (s != PL_bufptr) {
3608 start_force(PL_curforce);
3610 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3612 NEXTVAL_NEXTTOKE = pl_yylval;
3615 if (PL_lex_starts++) {
3619 sv_free(PL_thistoken);
3620 PL_thistoken = newSVpvs("");
3623 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3624 if (!PL_lex_casemods && PL_lex_inpat)
3637 PL_lex_state = LEX_NORMAL;
3638 s = scan_formline(PL_bufptr);
3639 if (!PL_lex_formbrack)
3645 PL_oldoldbufptr = PL_oldbufptr;
3651 sv_free(PL_thistoken);
3654 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3658 if (isIDFIRST_lazy_if(s,UTF))
3660 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3661 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3664 goto fake_eof; /* emulate EOF on ^D or ^Z */
3673 if (PL_lex_brackets) {
3674 yyerror((const char *)
3676 ? "Format not terminated"
3677 : "Missing right curly or square bracket"));
3679 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3680 "### Tokener got EOF\n");
3684 if (s++ < PL_bufend)
3685 goto retry; /* ignore stray nulls */
3688 if (!PL_in_eval && !PL_preambled) {
3689 PL_preambled = TRUE;
3695 /* Generate a string of Perl code to load the debugger.
3696 * If PERL5DB is set, it will return the contents of that,
3697 * otherwise a compile-time require of perl5db.pl. */
3699 const char * const pdb = PerlEnv_getenv("PERL5DB");
3702 sv_setpv(PL_linestr, pdb);
3703 sv_catpvs(PL_linestr,";");
3705 SETERRNO(0,SS_NORMAL);
3706 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3709 sv_setpvs(PL_linestr,"");
3710 if (PL_preambleav) {
3711 SV **svp = AvARRAY(PL_preambleav);
3712 SV **const end = svp + AvFILLp(PL_preambleav);
3714 sv_catsv(PL_linestr, *svp);
3716 sv_catpvs(PL_linestr, ";");
3718 sv_free(MUTABLE_SV(PL_preambleav));
3719 PL_preambleav = NULL;
3722 sv_catpvs(PL_linestr,
3723 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3724 if (PL_minus_n || PL_minus_p) {
3725 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3727 sv_catpvs(PL_linestr,"chomp;");
3730 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3731 || *PL_splitstr == '"')
3732 && strchr(PL_splitstr + 1, *PL_splitstr))
3733 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3735 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3736 bytes can be used as quoting characters. :-) */
3737 const char *splits = PL_splitstr;
3738 sv_catpvs(PL_linestr, "our @F=split(q\0");
3741 if (*splits == '\\')
3742 sv_catpvn(PL_linestr, splits, 1);
3743 sv_catpvn(PL_linestr, splits, 1);
3744 } while (*splits++);
3745 /* This loop will embed the trailing NUL of
3746 PL_linestr as the last thing it does before
3748 sv_catpvs(PL_linestr, ");");
3752 sv_catpvs(PL_linestr,"our @F=split(' ');");
3755 sv_catpvs(PL_linestr, "\n");
3756 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3757 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3758 PL_last_lop = PL_last_uni = NULL;
3759 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3760 update_debugger_info(PL_linestr, NULL, 0);
3764 bof = PL_rsfp ? TRUE : FALSE;
3765 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3768 PL_realtokenstart = -1;
3771 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3772 PerlIO_clearerr(PL_rsfp);
3774 (void)PerlIO_close(PL_rsfp);
3776 PL_doextract = FALSE;
3778 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3784 sv_setpvs(PL_linestr, ";}continue{print;}");
3786 sv_setpvs(PL_linestr, ";}");
3787 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3788 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3789 PL_last_lop = PL_last_uni = NULL;
3790 PL_minus_n = PL_minus_p = 0;
3793 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3794 PL_last_lop = PL_last_uni = NULL;
3795 sv_setpvs(PL_linestr,"");
3796 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3798 /* If it looks like the start of a BOM or raw UTF-16,
3799 * check if it in fact is. */
3805 #ifdef PERLIO_IS_STDIO
3806 # ifdef __GNU_LIBRARY__
3807 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3808 # define FTELL_FOR_PIPE_IS_BROKEN
3812 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3813 # define FTELL_FOR_PIPE_IS_BROKEN
3818 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3820 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3821 s = swallow_bom((U8*)s);
3825 /* Incest with pod. */
3828 sv_catsv(PL_thiswhite, PL_linestr);
3830 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3831 sv_setpvs(PL_linestr, "");
3832 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3833 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3834 PL_last_lop = PL_last_uni = NULL;
3835 PL_doextract = FALSE;
3839 } while (PL_doextract);
3840 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3841 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3842 update_debugger_info(PL_linestr, NULL, 0);
3843 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3844 PL_last_lop = PL_last_uni = NULL;
3845 if (CopLINE(PL_curcop) == 1) {
3846 while (s < PL_bufend && isSPACE(*s))
3848 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3852 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3856 if (*s == '#' && *(s+1) == '!')
3858 #ifdef ALTERNATE_SHEBANG
3860 static char const as[] = ALTERNATE_SHEBANG;
3861 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3862 d = s + (sizeof(as) - 1);
3864 #endif /* ALTERNATE_SHEBANG */
3873 while (*d && !isSPACE(*d))
3877 #ifdef ARG_ZERO_IS_SCRIPT
3878 if (ipathend > ipath) {
3880 * HP-UX (at least) sets argv[0] to the script name,
3881 * which makes $^X incorrect. And Digital UNIX and Linux,
3882 * at least, set argv[0] to the basename of the Perl
3883 * interpreter. So, having found "#!", we'll set it right.
3885 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3887 assert(SvPOK(x) || SvGMAGICAL(x));
3888 if (sv_eq(x, CopFILESV(PL_curcop))) {
3889 sv_setpvn(x, ipath, ipathend - ipath);
3895 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3896 const char * const lstart = SvPV_const(x,llen);
3898 bstart += blen - llen;
3899 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3900 sv_setpvn(x, ipath, ipathend - ipath);
3905 TAINT_NOT; /* $^X is always tainted, but that's OK */
3907 #endif /* ARG_ZERO_IS_SCRIPT */
3912 d = instr(s,"perl -");
3914 d = instr(s,"perl");
3916 /* avoid getting into infinite loops when shebang
3917 * line contains "Perl" rather than "perl" */
3919 for (d = ipathend-4; d >= ipath; --d) {
3920 if ((*d == 'p' || *d == 'P')
3921 && !ibcmp(d, "perl", 4))
3931 #ifdef ALTERNATE_SHEBANG
3933 * If the ALTERNATE_SHEBANG on this system starts with a
3934 * character that can be part of a Perl expression, then if
3935 * we see it but not "perl", we're probably looking at the
3936 * start of Perl code, not a request to hand off to some
3937 * other interpreter. Similarly, if "perl" is there, but
3938 * not in the first 'word' of the line, we assume the line
3939 * contains the start of the Perl program.
3941 if (d && *s != '#') {
3942 const char *c = ipath;
3943 while (*c && !strchr("; \t\r\n\f\v#", *c))
3946 d = NULL; /* "perl" not in first word; ignore */
3948 *s = '#'; /* Don't try to parse shebang line */
3950 #endif /* ALTERNATE_SHEBANG */
3951 #ifndef MACOS_TRADITIONAL
3956 !instr(s,"indir") &&
3957 instr(PL_origargv[0],"perl"))
3964 while (s < PL_bufend && isSPACE(*s))
3966 if (s < PL_bufend) {
3967 Newxz(newargv,PL_origargc+3,char*);
3969 while (s < PL_bufend && !isSPACE(*s))
3972 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3975 newargv = PL_origargv;
3978 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3980 Perl_croak(aTHX_ "Can't exec %s", ipath);
3984 while (*d && !isSPACE(*d))
3986 while (SPACE_OR_TAB(*d))
3990 const bool switches_done = PL_doswitches;
3991 const U32 oldpdb = PL_perldb;
3992 const bool oldn = PL_minus_n;
3993 const bool oldp = PL_minus_p;
3997 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3998 const char * const m = d1;
3999 while (*d1 && !isSPACE(*d1))
4001 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
4004 d1 = moreswitches(d1);
4006 if (PL_doswitches && !switches_done) {
4007 int argc = PL_origargc;
4008 char **argv = PL_origargv;
4011 } while (argc && argv[0][0] == '-' && argv[0][1]);
4012 init_argv_symbols(argc,argv);
4014 if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
4015 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
4016 /* if we have already added "LINE: while (<>) {",
4017 we must not do it again */
4019 sv_setpvs(PL_linestr, "");
4020 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4021 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4022 PL_last_lop = PL_last_uni = NULL;
4023 PL_preambled = FALSE;
4024 if (PERLDB_LINE || PERLDB_SAVESRC)
4025 (void)gv_fetchfile(PL_origfilename);
4032 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4034 PL_lex_state = LEX_FORMLINE;
4039 #ifdef PERL_STRICT_CR
4040 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4042 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4044 case ' ': case '\t': case '\f': case 013:
4045 #ifdef MACOS_TRADITIONAL
4049 PL_realtokenstart = -1;
4051 PL_thiswhite = newSVpvs("");
4052 sv_catpvn(PL_thiswhite, s, 1);
4059 PL_realtokenstart = -1;
4063 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4064 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4065 /* handle eval qq[#line 1 "foo"\n ...] */
4066 CopLINE_dec(PL_curcop);
4069 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4071 if (!PL_in_eval || PL_rsfp)
4076 while (d < PL_bufend && *d != '\n')
4080 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4081 Perl_croak(aTHX_ "panic: input overflow");
4084 PL_thiswhite = newSVpvn(s, d - s);
4089 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4091 PL_lex_state = LEX_FORMLINE;
4097 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4098 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4101 TOKEN(PEG); /* make sure any #! line is accessible */
4106 /* if (PL_madskills && PL_lex_formbrack) { */
4108 while (d < PL_bufend && *d != '\n')
4112 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4113 Perl_croak(aTHX_ "panic: input overflow");
4114 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4116 PL_thiswhite = newSVpvs("");
4117 if (CopLINE(PL_curcop) == 1) {
4118 sv_setpvs(PL_thiswhite, "");
4121 sv_catpvn(PL_thiswhite, s, d - s);
4135 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4143 while (s < PL_bufend && SPACE_OR_TAB(*s))
4146 if (strnEQ(s,"=>",2)) {
4147 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4148 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4149 OPERATOR('-'); /* unary minus */
4151 PL_last_uni = PL_oldbufptr;
4153 case 'r': ftst = OP_FTEREAD; break;
4154 case 'w': ftst = OP_FTEWRITE; break;
4155 case 'x': ftst = OP_FTEEXEC; break;
4156 case 'o': ftst = OP_FTEOWNED; break;
4157 case 'R': ftst = OP_FTRREAD; break;
4158 case 'W': ftst = OP_FTRWRITE; break;
4159 case 'X': ftst = OP_FTREXEC; break;
4160 case 'O': ftst = OP_FTROWNED; break;
4161 case 'e': ftst = OP_FTIS; break;
4162 case 'z': ftst = OP_FTZERO; break;
4163 case 's': ftst = OP_FTSIZE; break;
4164 case 'f': ftst = OP_FTFILE; break;
4165 case 'd': ftst = OP_FTDIR; break;
4166 case 'l': ftst = OP_FTLINK; break;
4167 case 'p': ftst = OP_FTPIPE; break;
4168 case 'S': ftst = OP_FTSOCK; break;
4169 case 'u': ftst = OP_FTSUID; break;
4170 case 'g': ftst = OP_FTSGID; break;
4171 case 'k': ftst = OP_FTSVTX; break;
4172 case 'b': ftst = OP_FTBLK; break;
4173 case 'c': ftst = OP_FTCHR; break;
4174 case 't': ftst = OP_FTTTY; break;
4175 case 'T': ftst = OP_FTTEXT; break;
4176 case 'B': ftst = OP_FTBINARY; break;
4177 case 'M': case 'A': case 'C':
4178 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4180 case 'M': ftst = OP_FTMTIME; break;
4181 case 'A': ftst = OP_FTATIME; break;
4182 case 'C': ftst = OP_FTCTIME; break;
4190 PL_last_lop_op = (OPCODE)ftst;
4191 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4192 "### Saw file test %c\n", (int)tmp);
4197 /* Assume it was a minus followed by a one-letter named
4198 * subroutine call (or a -bareword), then. */
4199 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4200 "### '-%c' looked like a file test but was not\n",
4207 const char tmp = *s++;
4210 if (PL_expect == XOPERATOR)
4215 else if (*s == '>') {
4218 if (isIDFIRST_lazy_if(s,UTF)) {
4219 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4227 if (PL_expect == XOPERATOR)
4230 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4232 OPERATOR('-'); /* unary minus */
4238 const char tmp = *s++;
4241 if (PL_expect == XOPERATOR)
4246 if (PL_expect == XOPERATOR)
4249 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4256 if (PL_expect != XOPERATOR) {
4257 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4258 PL_expect = XOPERATOR;
4259 force_ident(PL_tokenbuf, '*');
4272 if (PL_expect == XOPERATOR) {
4276 PL_tokenbuf[0] = '%';
4277 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4278 sizeof PL_tokenbuf - 1, FALSE);
4279 if (!PL_tokenbuf[1]) {
4282 PL_pending_ident = '%';
4291 const char tmp = *s++;
4296 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4303 const char tmp = *s++;
4309 goto just_a_word_zero_gv;
4312 switch (PL_expect) {
4318 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4320 PL_bufptr = s; /* update in case we back off */
4326 PL_expect = XTERMBLOCK;
4329 stuffstart = s - SvPVX(PL_linestr) - 1;
4333 while (isIDFIRST_lazy_if(s,UTF)) {
4336 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4337 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4338 if (tmp < 0) tmp = -tmp;
4353 sv = newSVpvn(s, len);
4355 d = scan_str(d,TRUE,TRUE);
4357 /* MUST advance bufptr here to avoid bogus
4358 "at end of line" context messages from yyerror().
4360 PL_bufptr = s + len;
4361 yyerror("Unterminated attribute parameter in attribute list");
4365 return REPORT(0); /* EOF indicator */
4369 sv_catsv(sv, PL_lex_stuff);
4370 attrs = append_elem(OP_LIST, attrs,
4371 newSVOP(OP_CONST, 0, sv));
4372 SvREFCNT_dec(PL_lex_stuff);
4373 PL_lex_stuff = NULL;
4376 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4378 if (PL_in_my == KEY_our) {
4380 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4382 /* skip to avoid loading attributes.pm */
4384 deprecate(":unique");
4387 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4390 /* NOTE: any CV attrs applied here need to be part of
4391 the CVf_BUILTIN_ATTRS define in cv.h! */
4392 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4394 CvLVALUE_on(PL_compcv);
4396 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4398 deprecate(":locked");
4400 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4402 CvMETHOD_on(PL_compcv);
4404 /* After we've set the flags, it could be argued that
4405 we don't need to do the attributes.pm-based setting
4406 process, and shouldn't bother appending recognized
4407 flags. To experiment with that, uncomment the
4408 following "else". (Note that's already been
4409 uncommented. That keeps the above-applied built-in
4410 attributes from being intercepted (and possibly
4411 rejected) by a package's attribute routines, but is
4412 justified by the performance win for the common case
4413 of applying only built-in attributes.) */
4415 attrs = append_elem(OP_LIST, attrs,
4416 newSVOP(OP_CONST, 0,
4420 if (*s == ':' && s[1] != ':')
4423 break; /* require real whitespace or :'s */
4424 /* XXX losing whitespace on sequential attributes here */
4428 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4429 if (*s != ';' && *s != '}' && *s != tmp
4430 && (tmp != '=' || *s != ')')) {
4431 const char q = ((*s == '\'') ? '"' : '\'');
4432 /* If here for an expression, and parsed no attrs, back
4434 if (tmp == '=' && !attrs) {
4438 /* MUST advance bufptr here to avoid bogus "at end of line"
4439 context messages from yyerror().
4442 yyerror( (const char *)
4444 ? Perl_form(aTHX_ "Invalid separator character "
4445 "%c%c%c in attribute list", q, *s, q)
4446 : "Unterminated attribute list" ) );
4454 start_force(PL_curforce);
4455 NEXTVAL_NEXTTOKE.opval = attrs;
4456 CURMAD('_', PL_nextwhite);
4461 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4462 (s - SvPVX(PL_linestr)) - stuffstart);
4470 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4471 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4479 const char tmp = *s++;
4484 const char tmp = *s++;
4492 if (PL_lex_brackets <= 0)
4493 yyerror("Unmatched right square bracket");
4496 if (PL_lex_state == LEX_INTERPNORMAL) {
4497 if (PL_lex_brackets == 0) {
4498 if (*s == '-' && s[1] == '>')
4499 PL_lex_state = LEX_INTERPENDMAYBE;
4500 else if (*s != '[' && *s != '{')
4501 PL_lex_state = LEX_INTERPEND;
4508 if (PL_lex_brackets > 100) {
4509 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4511 switch (PL_expect) {
4513 if (PL_lex_formbrack) {
4517 if (PL_oldoldbufptr == PL_last_lop)
4518 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4520 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4521 OPERATOR(HASHBRACK);
4523 while (s < PL_bufend && SPACE_OR_TAB(*s))
4526 PL_tokenbuf[0] = '\0';
4527 if (d < PL_bufend && *d == '-') {
4528 PL_tokenbuf[0] = '-';
4530 while (d < PL_bufend && SPACE_OR_TAB(*d))
4533 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4534 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4536 while (d < PL_bufend && SPACE_OR_TAB(*d))
4539 const char minus = (PL_tokenbuf[0] == '-');
4540 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4548 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4553 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4558 if (PL_oldoldbufptr == PL_last_lop)
4559 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4561 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4564 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4566 /* This hack is to get the ${} in the message. */
4568 yyerror("syntax error");
4571 OPERATOR(HASHBRACK);
4573 /* This hack serves to disambiguate a pair of curlies
4574 * as being a block or an anon hash. Normally, expectation
4575 * determines that, but in cases where we're not in a
4576 * position to expect anything in particular (like inside
4577 * eval"") we have to resolve the ambiguity. This code
4578 * covers the case where the first term in the curlies is a
4579 * quoted string. Most other cases need to be explicitly
4580 * disambiguated by prepending a "+" before the opening
4581 * curly in order to force resolution as an anon hash.
4583 * XXX should probably propagate the outer expectation
4584 * into eval"" to rely less on this hack, but that could
4585 * potentially break current behavior of eval"".
4589 if (*s == '\'' || *s == '"' || *s == '`') {
4590 /* common case: get past first string, handling escapes */
4591 for (t++; t < PL_bufend && *t != *s;)
4592 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4596 else if (*s == 'q') {
4599 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4602 /* skip q//-like construct */
4604 char open, close, term;
4607 while (t < PL_bufend && isSPACE(*t))
4609 /* check for q => */
4610 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4611 OPERATOR(HASHBRACK);
4615 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4619 for (t++; t < PL_bufend; t++) {
4620 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4622 else if (*t == open)
4626 for (t++; t < PL_bufend; t++) {
4627 if (*t == '\\' && t+1 < PL_bufend)
4629 else if (*t == close && --brackets <= 0)
4631 else if (*t == open)
4638 /* skip plain q word */
4639 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4642 else if (isALNUM_lazy_if(t,UTF)) {
4644 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4647 while (t < PL_bufend && isSPACE(*t))
4649 /* if comma follows first term, call it an anon hash */
4650 /* XXX it could be a comma expression with loop modifiers */
4651 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4652 || (*t == '=' && t[1] == '>')))
4653 OPERATOR(HASHBRACK);
4654 if (PL_expect == XREF)
4657 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4663 pl_yylval.ival = CopLINE(PL_curcop);
4664 if (isSPACE(*s) || *s == '#')
4665 PL_copline = NOLINE; /* invalidate current command line number */
4670 if (PL_lex_brackets <= 0)
4671 yyerror("Unmatched right curly bracket");
4673 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4674 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4675 PL_lex_formbrack = 0;
4676 if (PL_lex_state == LEX_INTERPNORMAL) {
4677 if (PL_lex_brackets == 0) {
4678 if (PL_expect & XFAKEBRACK) {
4679 PL_expect &= XENUMMASK;
4680 PL_lex_state = LEX_INTERPEND;
4685 PL_thiswhite = newSVpvs("");
4686 sv_catpvs(PL_thiswhite,"}");
4689 return yylex(); /* ignore fake brackets */
4691 if (*s == '-' && s[1] == '>')
4692 PL_lex_state = LEX_INTERPENDMAYBE;
4693 else if (*s != '[' && *s != '{')
4694 PL_lex_state = LEX_INTERPEND;
4697 if (PL_expect & XFAKEBRACK) {
4698 PL_expect &= XENUMMASK;
4700 return yylex(); /* ignore fake brackets */
4702 start_force(PL_curforce);
4704 curmad('X', newSVpvn(s-1,1));
4705 CURMAD('_', PL_thiswhite);
4710 PL_thistoken = newSVpvs("");
4718 if (PL_expect == XOPERATOR) {
4719 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4720 && isIDFIRST_lazy_if(s,UTF))
4722 CopLINE_dec(PL_curcop);
4723 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
4724 CopLINE_inc(PL_curcop);
4729 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4731 PL_expect = XOPERATOR;
4732 force_ident(PL_tokenbuf, '&');
4736 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4748 const char tmp = *s++;
4755 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4756 && strchr("+-*/%.^&|<",tmp))
4757 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4758 "Reversed %c= operator",(int)tmp);
4760 if (PL_expect == XSTATE && isALPHA(tmp) &&
4761 (s == PL_linestart+1 || s[-2] == '\n') )
4763 if (PL_in_eval && !PL_rsfp) {
4768 if (strnEQ(s,"=cut",4)) {
4784 PL_thiswhite = newSVpvs("");
4785 sv_catpvn(PL_thiswhite, PL_linestart,
4786 PL_bufend - PL_linestart);
4790 PL_doextract = TRUE;
4794 if (PL_lex_brackets < PL_lex_formbrack) {
4796 #ifdef PERL_STRICT_CR
4797 while (SPACE_OR_TAB(*t))
4799 while (SPACE_OR_TAB(*t) || *t == '\r')
4802 if (*t == '\n' || *t == '#') {
4811 if (PL_expect == XSTATE && s[1] == '!' && s[2] == '!') {
4817 const char tmp = *s++;
4819 /* was this !=~ where !~ was meant?
4820 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4822 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4823 const char *t = s+1;
4825 while (t < PL_bufend && isSPACE(*t))
4828 if (*t == '/' || *t == '?' ||
4829 ((*t == 'm' || *t == 's' || *t == 'y')
4830 && !isALNUM(t[1])) ||
4831 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4832 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4833 "!=~ should be !~");
4843 if (PL_expect != XOPERATOR) {
4844 if (s[1] != '<' && !strchr(s,'>'))
4847 s = scan_heredoc(s);
4849 s = scan_inputsymbol(s);
4850 TERM(sublex_start());
4856 SHop(OP_LEFT_SHIFT);
4870 const char tmp = *s++;
4872 SHop(OP_RIGHT_SHIFT);
4873 else if (tmp == '=')
4882 if (PL_expect == XOPERATOR) {
4883 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4885 deprecate_old(commaless_variable_list);
4886 return REPORT(','); /* grandfather non-comma-format format */
4890 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4891 PL_tokenbuf[0] = '@';
4892 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4893 sizeof PL_tokenbuf - 1, FALSE);
4894 if (PL_expect == XOPERATOR)
4895 no_op("Array length", s);
4896 if (!PL_tokenbuf[1])
4898 PL_expect = XOPERATOR;
4899 PL_pending_ident = '#';
4903 PL_tokenbuf[0] = '$';
4904 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4905 sizeof PL_tokenbuf - 1, FALSE);
4906 if (PL_expect == XOPERATOR)
4908 if (!PL_tokenbuf[1]) {
4910 yyerror("Final $ should be \\$ or $name");
4914 /* This kludge not intended to be bulletproof. */
4915 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4916 pl_yylval.opval = newSVOP(OP_CONST, 0,
4917 newSViv(CopARYBASE_get(&PL_compiling)));
4918 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4924 const char tmp = *s;
4925 if (PL_lex_state == LEX_NORMAL)
4928 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4929 && intuit_more(s)) {
4931 PL_tokenbuf[0] = '@';
4932 if (ckWARN(WARN_SYNTAX)) {
4935 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4938 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4939 while (t < PL_bufend && *t != ']')
4941 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4942 "Multidimensional syntax %.*s not supported",
4943 (int)((t - PL_bufptr) + 1), PL_bufptr);
4947 else if (*s == '{') {
4949 PL_tokenbuf[0] = '%';
4950 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4951 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4953 char tmpbuf[sizeof PL_tokenbuf];
4956 } while (isSPACE(*t));
4957 if (isIDFIRST_lazy_if(t,UTF)) {
4959 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4963 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4964 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4965 "You need to quote \"%s\"",
4972 PL_expect = XOPERATOR;
4973 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4974 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4975 if (!islop || PL_last_lop_op == OP_GREPSTART)
4976 PL_expect = XOPERATOR;
4977 else if (strchr("$@\"'`q", *s))
4978 PL_expect = XTERM; /* e.g. print $fh "foo" */
4979 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4980 PL_expect = XTERM; /* e.g. print $fh &sub */
4981 else if (isIDFIRST_lazy_if(s,UTF)) {
4982 char tmpbuf[sizeof PL_tokenbuf];
4984 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4985 if ((t2 = keyword(tmpbuf, len, 0))) {
4986 /* binary operators exclude handle interpretations */
4998 PL_expect = XTERM; /* e.g. print $fh length() */
5003 PL_expect = XTERM; /* e.g. print $fh subr() */
5006 else if (isDIGIT(*s))
5007 PL_expect = XTERM; /* e.g. print $fh 3 */
5008 else if (*s == '.' && isDIGIT(s[1]))
5009 PL_expect = XTERM; /* e.g. print $fh .3 */
5010 else if ((*s == '?' || *s == '-' || *s == '+')
5011 && !isSPACE(s[1]) && s[1] != '=')
5012 PL_expect = XTERM; /* e.g. print $fh -1 */
5013 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
5015 PL_expect = XTERM; /* e.g. print $fh /.../
5016 XXX except DORDOR operator
5018 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
5020 PL_expect = XTERM; /* print $fh <<"EOF" */
5023 PL_pending_ident = '$';
5027 if (PL_expect == XOPERATOR)
5029 PL_tokenbuf[0] = '@';
5030 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
5031 if (!PL_tokenbuf[1]) {
5034 if (PL_lex_state == LEX_NORMAL)
5036 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5038 PL_tokenbuf[0] = '%';
5040 /* Warn about @ where they meant $. */
5041 if (*s == '[' || *s == '{') {
5042 if (ckWARN(WARN_SYNTAX)) {
5043 const char *t = s + 1;
5044 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5046 if (*t == '}' || *t == ']') {
5048 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5049 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5050 "Scalar value %.*s better written as $%.*s",
5051 (int)(t-PL_bufptr), PL_bufptr,
5052 (int)(t-PL_bufptr-1), PL_bufptr+1);
5057 PL_pending_ident = '@';
5060 case '/': /* may be division, defined-or, or pattern */
5061 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5065 case '?': /* may either be conditional or pattern */
5066 if (PL_expect == XSTATE && s[1] == '?' && s[2] == '?') {
5070 if (PL_expect == XOPERATOR) {
5078 /* A // operator. */
5088 /* Disable warning on "study /blah/" */
5089 if (PL_oldoldbufptr == PL_last_uni
5090 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5091 || memNE(PL_last_uni, "study", 5)
5092 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5095 s = scan_pat(s,OP_MATCH);
5096 TERM(sublex_start());
5100 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5101 #ifdef PERL_STRICT_CR
5104 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5106 && (s == PL_linestart || s[-1] == '\n') )
5108 PL_lex_formbrack = 0;
5112 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5116 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5122 pl_yylval.ival = OPf_SPECIAL;
5128 if (PL_expect != XOPERATOR)
5133 case '0': case '1': case '2': case '3': case '4':
5134 case '5': case '6': case '7': case '8': case '9':
5135 s = scan_num(s, &pl_yylval);
5136 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5137 if (PL_expect == XOPERATOR)
5142 s = scan_str(s,!!PL_madskills,FALSE);
5143 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5144 if (PL_expect == XOPERATOR) {
5145 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5147 deprecate_old(commaless_variable_list);
5148 return REPORT(','); /* grandfather non-comma-format format */
5155 pl_yylval.ival = OP_CONST;
5156 TERM(sublex_start());
5159 s = scan_str(s,!!PL_madskills,FALSE);
5160 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5161 if (PL_expect == XOPERATOR) {
5162 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5164 deprecate_old(commaless_variable_list);
5165 return REPORT(','); /* grandfather non-comma-format format */
5172 pl_yylval.ival = OP_CONST;
5173 /* FIXME. I think that this can be const if char *d is replaced by
5174 more localised variables. */
5175 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5176 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5177 pl_yylval.ival = OP_STRINGIFY;
5181 TERM(sublex_start());
5184 s = scan_str(s,!!PL_madskills,FALSE);
5185 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5186 if (PL_expect == XOPERATOR)
5187 no_op("Backticks",s);
5190 readpipe_override();
5191 TERM(sublex_start());
5195 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5196 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5198 if (PL_expect == XOPERATOR)
5199 no_op("Backslash",s);
5203 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5204 char *start = s + 2;
5205 while (isDIGIT(*start) || *start == '_')
5207 if (*start == '.' && isDIGIT(start[1])) {
5208 s = scan_num(s, &pl_yylval);
5211 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5212 else if (!isALPHA(*start) && (PL_expect == XTERM
5213 || PL_expect == XREF || PL_expect == XSTATE
5214 || PL_expect == XTERMORDORDOR)) {
5215 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5217 s = scan_num(s, &pl_yylval);
5224 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5266 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5268 /* Some keywords can be followed by any delimiter, including ':' */
5269 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5270 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5271 (PL_tokenbuf[0] == 'q' &&
5272 strchr("qwxr", PL_tokenbuf[1])))));
5274 /* x::* is just a word, unless x is "CORE" */
5275 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5279 while (d < PL_bufend && isSPACE(*d))
5280 d++; /* no comments skipped here, or s### is misparsed */
5282 /* Is this a label? */
5283 if (!tmp && PL_expect == XSTATE
5284 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5286 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5291 /* Check for keywords */
5292 tmp = keyword(PL_tokenbuf, len, 0);
5294 /* Is this a word before a => operator? */
5295 if (*d == '=' && d[1] == '>') {
5298 = (OP*)newSVOP(OP_CONST, 0,
5299 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5300 pl_yylval.opval->op_private = OPpCONST_BARE;
5304 if (tmp < 0) { /* second-class keyword? */
5305 GV *ogv = NULL; /* override (winner) */
5306 GV *hgv = NULL; /* hidden (loser) */
5307 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5309 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5312 if (GvIMPORTED_CV(gv))
5314 else if (! CvMETHOD(cv))
5318 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5319 (gv = *gvp) && isGV_with_GP(gv) &&
5320 GvCVu(gv) && GvIMPORTED_CV(gv))
5327 tmp = 0; /* overridden by import or by GLOBAL */
5330 && -tmp==KEY_lock /* XXX generalizable kludge */
5333 tmp = 0; /* any sub overrides "weak" keyword */
5335 else { /* no override */
5337 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5338 Perl_warner(aTHX_ packWARN(WARN_MISC),
5339 "dump() better written as CORE::dump()");
5343 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5344 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5345 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5346 "Ambiguous call resolved as CORE::%s(), %s",
5347 GvENAME(hgv), "qualify as such or use &");
5354 default: /* not a keyword */
5355 /* Trade off - by using this evil construction we can pull the
5356 variable gv into the block labelled keylookup. If not, then
5357 we have to give it function scope so that the goto from the
5358 earlier ':' case doesn't bypass the initialisation. */
5360 just_a_word_zero_gv:
5368 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5371 SV *nextPL_nextwhite = 0;
5375 /* Get the rest if it looks like a package qualifier */
5377 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5379 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5382 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5383 *s == '\'' ? "'" : "::");
5388 if (PL_expect == XOPERATOR) {
5389 if (PL_bufptr == PL_linestart) {
5390 CopLINE_dec(PL_curcop);
5391 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5392 CopLINE_inc(PL_curcop);
5395 no_op("Bareword",s);
5398 /* Look for a subroutine with this name in current package,
5399 unless name is "Foo::", in which case Foo is a bearword
5400 (and a package name). */
5402 if (len > 2 && !PL_madskills &&
5403 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5405 if (ckWARN(WARN_BAREWORD)
5406 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5407 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5408 "Bareword \"%s\" refers to nonexistent package",
5411 PL_tokenbuf[len] = '\0';
5417 /* Mustn't actually add anything to a symbol table.
5418 But also don't want to "initialise" any placeholder
5419 constants that might already be there into full
5420 blown PVGVs with attached PVCV. */
5421 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5422 GV_NOADD_NOINIT, SVt_PVCV);
5427 /* if we saw a global override before, get the right name */
5430 sv = newSVpvs("CORE::GLOBAL::");
5431 sv_catpv(sv,PL_tokenbuf);
5434 /* If len is 0, newSVpv does strlen(), which is correct.
5435 If len is non-zero, then it will be the true length,
5436 and so the scalar will be created correctly. */
5437 sv = newSVpv(PL_tokenbuf,len);
5440 if (PL_madskills && !PL_thistoken) {
5441 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5442 PL_thistoken = newSVpvn(start,s - start);
5443 PL_realtokenstart = s - SvPVX(PL_linestr);
5447 /* Presume this is going to be a bareword of some sort. */
5450 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5451 pl_yylval.opval->op_private = OPpCONST_BARE;
5452 /* UTF-8 package name? */
5453 if (UTF && !IN_BYTES &&
5454 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5457 /* And if "Foo::", then that's what it certainly is. */
5462 /* Do the explicit type check so that we don't need to force
5463 the initialisation of the symbol table to have a real GV.
5464 Beware - gv may not really be a PVGV, cv may not really be
5465 a PVCV, (because of the space optimisations that gv_init
5466 understands) But they're true if for this symbol there is
5467 respectively a typeglob and a subroutine.
5469 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5470 /* Real typeglob, so get the real subroutine: */
5472 /* A proxy for a subroutine in this package? */
5473 : SvOK(gv) ? MUTABLE_CV(gv) : NULL)
5476 /* See if it's the indirect object for a list operator. */
5478 if (PL_oldoldbufptr &&
5479 PL_oldoldbufptr < PL_bufptr &&
5480 (PL_oldoldbufptr == PL_last_lop
5481 || PL_oldoldbufptr == PL_last_uni) &&
5482 /* NO SKIPSPACE BEFORE HERE! */
5483 (PL_expect == XREF ||
5484 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5486 bool immediate_paren = *s == '(';
5488 /* (Now we can afford to cross potential line boundary.) */
5489 s = SKIPSPACE2(s,nextPL_nextwhite);
5491 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5494 /* Two barewords in a row may indicate method call. */
5496 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5497 (tmp = intuit_method(s, gv, cv)))
5500 /* If not a declared subroutine, it's an indirect object. */
5501 /* (But it's an indir obj regardless for sort.) */
5502 /* Also, if "_" follows a filetest operator, it's a bareword */
5505 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5507 (PL_last_lop_op != OP_MAPSTART &&
5508 PL_last_lop_op != OP_GREPSTART))))
5509 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5510 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5513 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5518 PL_expect = XOPERATOR;
5521 s = SKIPSPACE2(s,nextPL_nextwhite);
5522 PL_nextwhite = nextPL_nextwhite;
5527 /* Is this a word before a => operator? */
5528 if (*s == '=' && s[1] == '>' && !pkgname) {
5530 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5531 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5532 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5536 /* If followed by a paren, it's certainly a subroutine. */
5541 while (SPACE_OR_TAB(*d))
5543 if (*d == ')' && (sv = gv_const_sv(gv))) {
5550 PL_nextwhite = PL_thiswhite;
5553 start_force(PL_curforce);
5555 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5556 PL_expect = XOPERATOR;
5559 PL_nextwhite = nextPL_nextwhite;
5560 curmad('X', PL_thistoken);
5561 PL_thistoken = newSVpvs("");
5569 /* If followed by var or block, call it a method (unless sub) */
5571 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5572 PL_last_lop = PL_oldbufptr;
5573 PL_last_lop_op = OP_METHOD;
5577 /* If followed by a bareword, see if it looks like indir obj. */
5580 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5581 && (tmp = intuit_method(s, gv, cv)))
5584 /* Not a method, so call it a subroutine (if defined) */
5587 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5588 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5589 "Ambiguous use of -%s resolved as -&%s()",
5590 PL_tokenbuf, PL_tokenbuf);
5591 /* Check for a constant sub */
5592 if ((sv = gv_const_sv(gv))) {
5594 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5595 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5596 pl_yylval.opval->op_private = 0;
5600 /* Resolve to GV now. */
5601 if (SvTYPE(gv) != SVt_PVGV) {
5602 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5603 assert (SvTYPE(gv) == SVt_PVGV);
5604 /* cv must have been some sort of placeholder, so
5605 now needs replacing with a real code reference. */
5609 op_free(pl_yylval.opval);
5610 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5611 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5612 PL_last_lop = PL_oldbufptr;
5613 PL_last_lop_op = OP_ENTERSUB;
5614 /* Is there a prototype? */
5622 const char *proto = SvPV_const(MUTABLE_SV(cv), protolen);
5625 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5627 while (*proto == ';')
5629 if (*proto == '&' && *s == '{') {
5631 sv_setpvs(PL_subname, "__ANON__");
5633 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5640 PL_nextwhite = PL_thiswhite;
5643 start_force(PL_curforce);
5644 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5647 PL_nextwhite = nextPL_nextwhite;
5648 curmad('X', PL_thistoken);
5649 PL_thistoken = newSVpvs("");
5656 /* Guess harder when madskills require "best effort". */
5657 if (PL_madskills && (!gv || !GvCVu(gv))) {
5658 int probable_sub = 0;
5659 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5661 else if (isALPHA(*s)) {
5665 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5666 if (!keyword(tmpbuf, tmplen, 0))
5669 while (d < PL_bufend && isSPACE(*d))
5671 if (*d == '=' && d[1] == '>')
5676 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5677 op_free(pl_yylval.opval);
5678 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5679 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5680 PL_last_lop = PL_oldbufptr;
5681 PL_last_lop_op = OP_ENTERSUB;
5682 PL_nextwhite = PL_thiswhite;
5684 start_force(PL_curforce);
5685 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5687 PL_nextwhite = nextPL_nextwhite;
5688 curmad('X', PL_thistoken);
5689 PL_thistoken = newSVpvs("");
5694 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5701 /* Call it a bare word */
5704 if (PL_hints & HINT_STRICT_SUBS)
5705 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5707 if (lastchar != '-') {
5708 if (ckWARN(WARN_RESERVED)) {
5712 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5713 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5720 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5721 && ckWARN_d(WARN_AMBIGUOUS)) {
5722 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5723 "Operator or semicolon missing before %c%s",
5724 lastchar, PL_tokenbuf);
5725 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5726 "Ambiguous use of %c resolved as operator %c",
5727 lastchar, lastchar);
5733 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5734 newSVpv(CopFILE(PL_curcop),0));
5738 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5739 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5742 case KEY___PACKAGE__:
5743 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5745 ? newSVhek(HvNAME_HEK(PL_curstash))
5752 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5753 const char *pname = "main";
5754 if (PL_tokenbuf[2] == 'D')
5755 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5756 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5760 GvIOp(gv) = newIO();
5761 IoIFP(GvIOp(gv)) = PL_rsfp;
5762 #if defined(HAS_FCNTL) && defined(F_SETFD)
5764 const int fd = PerlIO_fileno(PL_rsfp);
5765 fcntl(fd,F_SETFD,fd >= 3);
5768 /* Mark this internal pseudo-handle as clean */
5769 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5770 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5771 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5773 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5774 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5775 /* if the script was opened in binmode, we need to revert
5776 * it to text mode for compatibility; but only iff it has CRs
5777 * XXX this is a questionable hack at best. */
5778 if (PL_bufend-PL_bufptr > 2
5779 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5782 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5783 loc = PerlIO_tell(PL_rsfp);
5784 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5787 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5789 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5790 #endif /* NETWARE */
5791 #ifdef PERLIO_IS_STDIO /* really? */
5792 # if defined(__BORLANDC__)
5793 /* XXX see note in do_binmode() */
5794 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5798 PerlIO_seek(PL_rsfp, loc, 0);
5802 #ifdef PERLIO_LAYERS
5805 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5806 else if (PL_encoding) {
5813 XPUSHs(PL_encoding);
5815 call_method("name", G_SCALAR);
5819 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5820 Perl_form(aTHX_ ":encoding(%"SVf")",
5829 if (PL_realtokenstart >= 0) {
5830 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5832 PL_endwhite = newSVpvs("");
5833 sv_catsv(PL_endwhite, PL_thiswhite);
5835 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5836 PL_realtokenstart = -1;
5838 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5839 SvCUR(PL_endwhite))) != NULL) ;
5854 if (PL_expect == XSTATE) {
5861 if (*s == ':' && s[1] == ':') {
5864 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5865 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5866 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5869 else if (tmp == KEY_require || tmp == KEY_do)
5870 /* that's a way to remember we saw "CORE::" */
5883 LOP(OP_ACCEPT,XTERM);
5889 LOP(OP_ATAN2,XTERM);
5895 LOP(OP_BINMODE,XTERM);
5898 LOP(OP_BLESS,XTERM);
5907 /* When 'use switch' is in effect, continue has a dual
5908 life as a control operator. */
5910 if (!FEATURE_IS_ENABLED("switch"))
5913 /* We have to disambiguate the two senses of
5914 "continue". If the next token is a '{' then
5915 treat it as the start of a continue block;
5916 otherwise treat it as a control operator.
5928 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5945 if (!PL_cryptseen) {
5946 PL_cryptseen = TRUE;
5950 LOP(OP_CRYPT,XTERM);
5953 LOP(OP_CHMOD,XTERM);
5956 LOP(OP_CHOWN,XTERM);
5959 LOP(OP_CONNECT,XTERM);
5978 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5979 if (orig_keyword == KEY_do) {
5988 PL_hints |= HINT_BLOCK_SCOPE;
5998 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5999 LOP(OP_DBMOPEN,XTERM);
6005 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6012 pl_yylval.ival = CopLINE(PL_curcop);
6028 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
6029 UNIBRACK(OP_ENTEREVAL);
6043 case KEY_endhostent:
6049 case KEY_endservent:
6052 case KEY_endprotoent:
6063 pl_yylval.ival = CopLINE(PL_curcop);
6065 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6068 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6071 if ((PL_bufend - p) >= 3 &&
6072 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6074 else if ((PL_bufend - p) >= 4 &&
6075 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6078 if (isIDFIRST_lazy_if(p,UTF)) {
6079 p = scan_ident(p, PL_bufend,
6080 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6084 Perl_croak(aTHX_ "Missing $ on loop variable");
6086 s = SvPVX(PL_linestr) + soff;
6092 LOP(OP_FORMLINE,XTERM);
6098 LOP(OP_FCNTL,XTERM);
6104 LOP(OP_FLOCK,XTERM);
6113 LOP(OP_GREPSTART, XREF);
6116 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6131 case KEY_getpriority:
6132 LOP(OP_GETPRIORITY,XTERM);
6134 case KEY_getprotobyname:
6137 case KEY_getprotobynumber:
6138 LOP(OP_GPBYNUMBER,XTERM);
6140 case KEY_getprotoent:
6152 case KEY_getpeername:
6153 UNI(OP_GETPEERNAME);
6155 case KEY_gethostbyname:
6158 case KEY_gethostbyaddr:
6159 LOP(OP_GHBYADDR,XTERM);
6161 case KEY_gethostent:
6164 case KEY_getnetbyname:
6167 case KEY_getnetbyaddr:
6168 LOP(OP_GNBYADDR,XTERM);
6173 case KEY_getservbyname:
6174 LOP(OP_GSBYNAME,XTERM);
6176 case KEY_getservbyport:
6177 LOP(OP_GSBYPORT,XTERM);
6179 case KEY_getservent:
6182 case KEY_getsockname:
6183 UNI(OP_GETSOCKNAME);
6185 case KEY_getsockopt:
6186 LOP(OP_GSOCKOPT,XTERM);
6201 pl_yylval.ival = CopLINE(PL_curcop);
6211 pl_yylval.ival = CopLINE(PL_curcop);
6215 LOP(OP_INDEX,XTERM);
6221 LOP(OP_IOCTL,XTERM);
6233 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6265 LOP(OP_LISTEN,XTERM);
6274 s = scan_pat(s,OP_MATCH);
6275 TERM(sublex_start());
6278 LOP(OP_MAPSTART, XREF);
6281 LOP(OP_MKDIR,XTERM);
6284 LOP(OP_MSGCTL,XTERM);
6287 LOP(OP_MSGGET,XTERM);
6290 LOP(OP_MSGRCV,XTERM);
6293 LOP(OP_MSGSND,XTERM);
6298 PL_in_my = (U16)tmp;
6300 if (isIDFIRST_lazy_if(s,UTF)) {
6304 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6305 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6307 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6308 if (!PL_in_my_stash) {
6311 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6315 if (PL_madskills) { /* just add type to declarator token */
6316 sv_catsv(PL_thistoken, PL_nextwhite);
6318 sv_catpvn(PL_thistoken, start, s - start);
6326 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6333 s = tokenize_use(0, s);
6337 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6344 if (isIDFIRST_lazy_if(s,UTF)) {
6346 for (d = s; isALNUM_lazy_if(d,UTF);)
6348 for (t=d; isSPACE(*t);)
6350 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6352 && !(t[0] == '=' && t[1] == '>')
6354 int parms_len = (int)(d-s);
6355 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6356 "Precedence problem: open %.*s should be open(%.*s)",
6357 parms_len, s, parms_len, s);
6363 pl_yylval.ival = OP_OR;
6373 LOP(OP_OPEN_DIR,XTERM);
6376 checkcomma(s,PL_tokenbuf,"filehandle");
6380 checkcomma(s,PL_tokenbuf,"filehandle");
6399 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6403 LOP(OP_PIPE_OP,XTERM);
6406 s = scan_str(s,!!PL_madskills,FALSE);
6409 pl_yylval.ival = OP_CONST;
6410 TERM(sublex_start());
6416 s = scan_str(s,!!PL_madskills,FALSE);
6419 PL_expect = XOPERATOR;
6421 if (SvCUR(PL_lex_stuff)) {
6424 d = SvPV_force(PL_lex_stuff, len);
6426 for (; isSPACE(*d) && len; --len, ++d)
6431 if (!warned && ckWARN(WARN_QW)) {
6432 for (; !isSPACE(*d) && len; --len, ++d) {
6434 Perl_warner(aTHX_ packWARN(WARN_QW),
6435 "Possible attempt to separate words with commas");
6438 else if (*d == '#') {
6439 Perl_warner(aTHX_ packWARN(WARN_QW),
6440 "Possible attempt to put comments in qw() list");
6446 for (; !isSPACE(*d) && len; --len, ++d)
6449 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6450 words = append_elem(OP_LIST, words,
6451 newSVOP(OP_CONST, 0, tokeq(sv)));
6455 start_force(PL_curforce);
6456 NEXTVAL_NEXTTOKE.opval = words;
6461 SvREFCNT_dec(PL_lex_stuff);
6462 PL_lex_stuff = NULL;
6468 s = scan_str(s,!!PL_madskills,FALSE);
6471 pl_yylval.ival = OP_STRINGIFY;
6472 if (SvIVX(PL_lex_stuff) == '\'')
6473 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6474 TERM(sublex_start());
6477 s = scan_pat(s,OP_QR);
6478 TERM(sublex_start());
6481 s = scan_str(s,!!PL_madskills,FALSE);
6484 readpipe_override();
6485 TERM(sublex_start());
6493 s = force_version(s, FALSE);
6495 else if (*s != 'v' || !isDIGIT(s[1])
6496 || (s = force_version(s, TRUE), *s == 'v'))
6498 *PL_tokenbuf = '\0';
6499 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6500 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6501 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6503 yyerror("<> should be quotes");
6505 if (orig_keyword == KEY_require) {
6513 PL_last_uni = PL_oldbufptr;
6514 PL_last_lop_op = OP_REQUIRE;
6516 return REPORT( (int)REQUIRE );
6522 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6526 LOP(OP_RENAME,XTERM);
6535 LOP(OP_RINDEX,XTERM);
6544 UNIDOR(OP_READLINE);
6547 UNIDOR(OP_BACKTICK);
6556 LOP(OP_REVERSE,XTERM);
6559 UNIDOR(OP_READLINK);
6566 if (pl_yylval.opval)
6567 TERM(sublex_start());
6569 TOKEN(1); /* force error */
6572 checkcomma(s,PL_tokenbuf,"filehandle");
6582 LOP(OP_SELECT,XTERM);
6588 LOP(OP_SEMCTL,XTERM);
6591 LOP(OP_SEMGET,XTERM);
6594 LOP(OP_SEMOP,XTERM);
6600 LOP(OP_SETPGRP,XTERM);
6602 case KEY_setpriority:
6603 LOP(OP_SETPRIORITY,XTERM);
6605 case KEY_sethostent:
6611 case KEY_setservent:
6614 case KEY_setprotoent:
6624 LOP(OP_SEEKDIR,XTERM);
6626 case KEY_setsockopt:
6627 LOP(OP_SSOCKOPT,XTERM);
6633 LOP(OP_SHMCTL,XTERM);
6636 LOP(OP_SHMGET,XTERM);
6639 LOP(OP_SHMREAD,XTERM);
6642 LOP(OP_SHMWRITE,XTERM);
6645 LOP(OP_SHUTDOWN,XTERM);
6654 LOP(OP_SOCKET,XTERM);
6656 case KEY_socketpair:
6657 LOP(OP_SOCKPAIR,XTERM);
6660 checkcomma(s,PL_tokenbuf,"subroutine name");
6662 if (*s == ';' || *s == ')') /* probably a close */
6663 Perl_croak(aTHX_ "sort is now a reserved word");
6665 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6669 LOP(OP_SPLIT,XTERM);
6672 LOP(OP_SPRINTF,XTERM);
6675 LOP(OP_SPLICE,XTERM);
6690 LOP(OP_SUBSTR,XTERM);
6696 char tmpbuf[sizeof PL_tokenbuf];
6697 SSize_t tboffset = 0;
6698 expectation attrful;
6699 bool have_name, have_proto;
6700 const int key = tmp;
6705 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6706 SV *subtoken = newSVpvn(tstart, s - tstart);
6710 s = SKIPSPACE2(s,tmpwhite);
6715 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6716 (*s == ':' && s[1] == ':'))
6719 SV *nametoke = NULL;
6723 attrful = XATTRBLOCK;
6724 /* remember buffer pos'n for later force_word */
6725 tboffset = s - PL_oldbufptr;
6726 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6729 nametoke = newSVpvn(s, d - s);
6731 if (memchr(tmpbuf, ':', len))
6732 sv_setpvn(PL_subname, tmpbuf, len);
6734 sv_setsv(PL_subname,PL_curstname);
6735 sv_catpvs(PL_subname,"::");
6736 sv_catpvn(PL_subname,tmpbuf,len);
6743 CURMAD('X', nametoke);
6744 CURMAD('_', tmpwhite);
6745 (void) force_word(PL_oldbufptr + tboffset, WORD,
6748 s = SKIPSPACE2(d,tmpwhite);
6755 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6756 PL_expect = XTERMBLOCK;
6757 attrful = XATTRTERM;
6758 sv_setpvs(PL_subname,"?");
6762 if (key == KEY_format) {
6764 PL_lex_formbrack = PL_lex_brackets + 1;
6766 PL_thistoken = subtoken;
6770 (void) force_word(PL_oldbufptr + tboffset, WORD,
6776 /* Look for a prototype */
6779 bool bad_proto = FALSE;
6780 bool in_brackets = FALSE;
6781 char greedy_proto = ' ';
6782 bool proto_after_greedy_proto = FALSE;
6783 bool must_be_last = FALSE;
6784 bool underscore = FALSE;
6785 bool seen_underscore = FALSE;
6786 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6788 s = scan_str(s,!!PL_madskills,FALSE);
6790 Perl_croak(aTHX_ "Prototype not terminated");
6791 /* strip spaces and check for bad characters */
6792 d = SvPVX(PL_lex_stuff);
6794 for (p = d; *p; ++p) {
6800 proto_after_greedy_proto = TRUE;
6801 if (!strchr("$@%*;[]&\\_", *p)) {
6813 else if ( *p == ']' ) {
6814 in_brackets = FALSE;
6816 else if ( (*p == '@' || *p == '%') &&
6817 ( tmp < 2 || d[tmp-2] != '\\' ) &&
6819 must_be_last = TRUE;
6822 else if ( *p == '_' ) {
6823 underscore = seen_underscore = TRUE;
6830 if (proto_after_greedy_proto)
6831 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6832 "Prototype after '%c' for %"SVf" : %s",
6833 greedy_proto, SVfARG(PL_subname), d);
6835 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6836 "Illegal character %sin prototype for %"SVf" : %s",
6837 seen_underscore ? "after '_' " : "",
6838 SVfARG(PL_subname), d);
6839 SvCUR_set(PL_lex_stuff, tmp);
6844 CURMAD('q', PL_thisopen);
6845 CURMAD('_', tmpwhite);
6846 CURMAD('=', PL_thisstuff);
6847 CURMAD('Q', PL_thisclose);
6848 NEXTVAL_NEXTTOKE.opval =
6849 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6850 PL_lex_stuff = NULL;
6853 s = SKIPSPACE2(s,tmpwhite);
6861 if (*s == ':' && s[1] != ':')
6862 PL_expect = attrful;
6863 else if (*s != '{' && key == KEY_sub) {
6865 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6867 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6874 curmad('^', newSVpvs(""));
6875 CURMAD('_', tmpwhite);
6879 PL_thistoken = subtoken;
6882 NEXTVAL_NEXTTOKE.opval =
6883 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6884 PL_lex_stuff = NULL;
6890 sv_setpvs(PL_subname, "__ANON__");
6892 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6896 (void) force_word(PL_oldbufptr + tboffset, WORD,
6905 LOP(OP_SYSTEM,XREF);
6908 LOP(OP_SYMLINK,XTERM);
6911 LOP(OP_SYSCALL,XTERM);
6914 LOP(OP_SYSOPEN,XTERM);
6917 LOP(OP_SYSSEEK,XTERM);
6920 LOP(OP_SYSREAD,XTERM);
6923 LOP(OP_SYSWRITE,XTERM);
6927 TERM(sublex_start());
6948 LOP(OP_TRUNCATE,XTERM);
6960 pl_yylval.ival = CopLINE(PL_curcop);
6964 pl_yylval.ival = CopLINE(PL_curcop);
6968 LOP(OP_UNLINK,XTERM);
6974 LOP(OP_UNPACK,XTERM);
6977 LOP(OP_UTIME,XTERM);
6983 LOP(OP_UNSHIFT,XTERM);
6986 s = tokenize_use(1, s);
6996 pl_yylval.ival = CopLINE(PL_curcop);
7000 pl_yylval.ival = CopLINE(PL_curcop);
7004 PL_hints |= HINT_BLOCK_SCOPE;
7011 LOP(OP_WAITPID,XTERM);
7020 ctl_l[0] = toCTRL('L');
7022 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
7025 /* Make sure $^L is defined */
7026 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
7031 if (PL_expect == XOPERATOR)
7037 pl_yylval.ival = OP_XOR;
7042 TERM(sublex_start());
7047 #pragma segment Main
7051 S_pending_ident(pTHX)
7056 /* pit holds the identifier we read and pending_ident is reset */
7057 char pit = PL_pending_ident;
7058 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7059 /* All routes through this function want to know if there is a colon. */
7060 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7061 PL_pending_ident = 0;
7063 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7064 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7065 "### Pending identifier '%s'\n", PL_tokenbuf); });
7067 /* if we're in a my(), we can't allow dynamics here.
7068 $foo'bar has already been turned into $foo::bar, so
7069 just check for colons.
7071 if it's a legal name, the OP is a PADANY.
7074 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7076 yyerror(Perl_form(aTHX_ "No package name allowed for "
7077 "variable %s in \"our\"",
7079 tmp = allocmy(PL_tokenbuf);
7083 yyerror(Perl_form(aTHX_ PL_no_myglob,
7084 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7086 pl_yylval.opval = newOP(OP_PADANY, 0);
7087 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7093 build the ops for accesses to a my() variable.
7095 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7096 then used in a comparison. This catches most, but not
7097 all cases. For instance, it catches
7098 sort { my($a); $a <=> $b }
7100 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7101 (although why you'd do that is anyone's guess).
7106 tmp = pad_findmy(PL_tokenbuf);
7107 if (tmp != NOT_IN_PAD) {
7108 /* might be an "our" variable" */
7109 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7110 /* build ops for a bareword */
7111 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7112 HEK * const stashname = HvNAME_HEK(stash);
7113 SV * const sym = newSVhek(stashname);
7114 sv_catpvs(sym, "::");
7115 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7116 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7117 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7120 ? (GV_ADDMULTI | GV_ADDINEVAL)
7123 ((PL_tokenbuf[0] == '$') ? SVt_PV
7124 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7129 /* if it's a sort block and they're naming $a or $b */
7130 if (PL_last_lop_op == OP_SORT &&
7131 PL_tokenbuf[0] == '$' &&
7132 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7135 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7136 d < PL_bufend && *d != '\n';
7139 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7140 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7146 pl_yylval.opval = newOP(OP_PADANY, 0);
7147 pl_yylval.opval->op_targ = tmp;
7153 Whine if they've said @foo in a doublequoted string,
7154 and @foo isn't a variable we can find in the symbol
7157 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7158 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7160 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7161 && ckWARN(WARN_AMBIGUOUS)
7162 /* DO NOT warn for @- and @+ */
7163 && !( PL_tokenbuf[2] == '\0' &&
7164 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7167 /* Downgraded from fatal to warning 20000522 mjd */
7168 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7169 "Possible unintended interpolation of %s in string",
7174 /* build ops for a bareword */
7175 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7177 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7179 PL_tokenbuf + 1, tokenbuf_len - 1,
7180 /* If the identifier refers to a stash, don't autovivify it.
7181 * Change 24660 had the side effect of causing symbol table
7182 * hashes to always be defined, even if they were freshly
7183 * created and the only reference in the entire program was
7184 * the single statement with the defined %foo::bar:: test.
7185 * It appears that all code in the wild doing this actually
7186 * wants to know whether sub-packages have been loaded, so
7187 * by avoiding auto-vivifying symbol tables, we ensure that
7188 * defined %foo::bar:: continues to be false, and the existing
7189 * tests still give the expected answers, even though what
7190 * they're actually testing has now changed subtly.
7192 (*PL_tokenbuf == '%'
7193 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7196 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7197 ((PL_tokenbuf[0] == '$') ? SVt_PV
7198 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7204 * The following code was generated by perl_keyword.pl.
7208 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7212 PERL_ARGS_ASSERT_KEYWORD;
7216 case 1: /* 5 tokens of length 1 */
7248 case 2: /* 18 tokens of length 2 */
7394 case 3: /* 29 tokens of length 3 */
7398 if (name[1] == 'N' &&
7461 if (name[1] == 'i' &&
7493 if (name[1] == 'o' &&
7502 if (name[1] == 'e' &&
7511 if (name[1] == 'n' &&
7520 if (name[1] == 'o' &&
7529 if (name[1] == 'a' &&
7538 if (name[1] == 'o' &&
7600 if (name[1] == 'e' &&
7614 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7640 if (name[1] == 'i' &&
7649 if (name[1] == 's' &&
7658 if (name[1] == 'e' &&
7667 if (name[1] == 'o' &&
7679 case 4: /* 41 tokens of length 4 */
7683 if (name[1] == 'O' &&
7693 if (name[1] == 'N' &&
7703 if (name[1] == 'i' &&
7713 if (name[1] == 'h' &&
7723 if (name[1] == 'u' &&
7736 if (name[2] == 'c' &&
7745 if (name[2] == 's' &&
7754 if (name[2] == 'a' &&
7790 if (name[1] == 'o' &&
7803 if (name[2] == 't' &&
7812 if (name[2] == 'o' &&
7821 if (name[2] == 't' &&
7830 if (name[2] == 'e' &&
7843 if (name[1] == 'o' &&
7856 if (name[2] == 'y' &&
7865 if (name[2] == 'l' &&
7881 if (name[2] == 's' &&
7890 if (name[2] == 'n' &&
7899 if (name[2] == 'c' &&
7912 if (name[1] == 'e' &&
7922 if (name[1] == 'p' &&
7935 if (name[2] == 'c' &&
7944 if (name[2] == 'p' &&
7953 if (name[2] == 's' &&
7969 if (name[2] == 'n' &&
8039 if (name[2] == 'r' &&
8048 if (name[2] == 'r' &&
8057 if (name[2] == 'a' &&
8073 if (name[2] == 'l' &&
8135 if (name[2] == 'e' &&
8138 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8151 case 5: /* 39 tokens of length 5 */
8155 if (name[1] == 'E' &&
8166 if (name[1] == 'H' &&
8180 if (name[2] == 'a' &&
8190 if (name[2] == 'a' &&
8207 if (name[2] == 'e' &&
8217 if (name[2] == 'e' &&
8221 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8237 if (name[3] == 'i' &&
8246 if (name[3] == 'o' &&
8282 if (name[2] == 'o' &&
8292 if (name[2] == 'y' &&
8306 if (name[1] == 'l' &&
8320 if (name[2] == 'n' &&
8330 if (name[2] == 'o' &&
8344 if (name[1] == 'i' &&
8349 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8358 if (name[2] == 'd' &&
8368 if (name[2] == 'c' &&
8385 if (name[2] == 'c' &&
8395 if (name[2] == 't' &&
8409 if (name[1] == 'k' &&
8420 if (name[1] == 'r' &&
8434 if (name[2] == 's' &&
8444 if (name[2] == 'd' &&
8461 if (name[2] == 'm' &&
8471 if (name[2] == 'i' &&
8481 if (name[2] == 'e' &&
8491 if (name[2] == 'l' &&
8501 if (name[2] == 'a' &&
8514 if (name[3] == 't' &&
8517 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8523 if (name[3] == 'd' &&
8540 if (name[1] == 'i' &&
8554 if (name[2] == 'a' &&
8567 if (name[3] == 'e' &&
8602 if (name[2] == 'i' &&
8619 if (name[2] == 'i' &&
8629 if (name[2] == 'i' &&
8646 case 6: /* 33 tokens of length 6 */
8650 if (name[1] == 'c' &&
8665 if (name[2] == 'l' &&
8676 if (name[2] == 'r' &&
8691 if (name[1] == 'e' &&
8706 if (name[2] == 's' &&
8711 if(ckWARN_d(WARN_SYNTAX))
8712 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8718 if (name[2] == 'i' &&
8736 if (name[2] == 'l' &&
8747 if (name[2] == 'r' &&
8762 if (name[1] == 'm' &&
8777 if (name[2] == 'n' &&
8788 if (name[2] == 's' &&
8803 if (name[1] == 's' &&
8809 if (name[4] == 't' &&
8818 if (name[4] == 'e' &&
8827 if (name[4] == 'c' &&
8836 if (name[4] == 'n' &&
8852 if (name[1] == 'r' &&
8870 if (name[3] == 'a' &&
8880 if (name[3] == 'u' &&
8894 if (name[2] == 'n' &&
8912 if (name[2] == 'a' &&
8926 if (name[3] == 'e' &&
8939 if (name[4] == 't' &&
8948 if (name[4] == 'e' &&
8970 if (name[4] == 't' &&
8979 if (name[4] == 'e' &&
8995 if (name[2] == 'c' &&
9006 if (name[2] == 'l' &&
9017 if (name[2] == 'b' &&
9028 if (name[2] == 's' &&
9051 if (name[4] == 's' &&
9060 if (name[4] == 'n' &&
9073 if (name[3] == 'a' &&
9090 if (name[1] == 'a' &&
9105 case 7: /* 29 tokens of length 7 */
9109 if (name[1] == 'E' &&
9122 if (name[1] == '_' &&
9135 if (name[1] == 'i' &&
9142 return -KEY_binmode;
9148 if (name[1] == 'o' &&
9155 return -KEY_connect;
9164 if (name[2] == 'm' &&
9170 return -KEY_dbmopen;
9181 if (name[4] == 'u' &&
9185 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9191 if (name[4] == 'n' &&
9212 if (name[1] == 'o' &&
9225 if (name[1] == 'e' &&
9232 if (name[5] == 'r' &&
9235 return -KEY_getpgrp;
9241 if (name[5] == 'i' &&
9244 return -KEY_getppid;
9257 if (name[1] == 'c' &&
9264 return -KEY_lcfirst;
9270 if (name[1] == 'p' &&
9277 return -KEY_opendir;
9283 if (name[1] == 'a' &&
9301 if (name[3] == 'd' &&
9306 return -KEY_readdir;
9312 if (name[3] == 'u' &&
9323 if (name[3] == 'e' &&
9328 return -KEY_reverse;
9347 if (name[3] == 'k' &&
9352 return -KEY_seekdir;
9358 if (name[3] == 'p' &&
9363 return -KEY_setpgrp;
9373 if (name[2] == 'm' &&
9379 return -KEY_shmread;
9385 if (name[2] == 'r' &&
9391 return -KEY_sprintf;
9400 if (name[3] == 'l' &&
9405 return -KEY_symlink;
9414 if (name[4] == 'a' &&
9418 return -KEY_syscall;
9424 if (name[4] == 'p' &&
9428 return -KEY_sysopen;
9434 if (name[4] == 'e' &&
9438 return -KEY_sysread;
9444 if (name[4] == 'e' &&
9448 return -KEY_sysseek;
9466 if (name[1] == 'e' &&
9473 return -KEY_telldir;
9482 if (name[2] == 'f' &&
9488 return -KEY_ucfirst;
9494 if (name[2] == 's' &&
9500 return -KEY_unshift;
9510 if (name[1] == 'a' &&
9517 return -KEY_waitpid;
9526 case 8: /* 26 tokens of length 8 */
9530 if (name[1] == 'U' &&
9538 return KEY_AUTOLOAD;
9549 if (name[3] == 'A' &&
9555 return KEY___DATA__;
9561 if (name[3] == 'I' &&
9567 return -KEY___FILE__;
9573 if (name[3] == 'I' &&
9579 return -KEY___LINE__;
9595 if (name[2] == 'o' &&
9602 return -KEY_closedir;
9608 if (name[2] == 'n' &&
9615 return -KEY_continue;
9625 if (name[1] == 'b' &&
9633 return -KEY_dbmclose;
9639 if (name[1] == 'n' &&
9645 if (name[4] == 'r' &&
9650 return -KEY_endgrent;
9656 if (name[4] == 'w' &&
9661 return -KEY_endpwent;
9674 if (name[1] == 'o' &&
9682 return -KEY_formline;
9688 if (name[1] == 'e' &&
9699 if (name[6] == 'n' &&
9702 return -KEY_getgrent;
9708 if (name[6] == 'i' &&
9711 return -KEY_getgrgid;
9717 if (name[6] == 'a' &&
9720 return -KEY_getgrnam;
9733 if (name[4] == 'o' &&
9738 return -KEY_getlogin;
9749 if (name[6] == 'n' &&
9752 return -KEY_getpwent;
9758 if (name[6] == 'a' &&
9761 return -KEY_getpwnam;
9767 if (name[6] == 'i' &&
9770 return -KEY_getpwuid;
9790 if (name[1] == 'e' &&
9797 if (name[5] == 'i' &&
9804 return -KEY_readline;
9809 return -KEY_readlink;
9820 if (name[5] == 'i' &&
9824 return -KEY_readpipe;
9845 if (name[4] == 'r' &&
9850 return -KEY_setgrent;
9856 if (name[4] == 'w' &&
9861 return -KEY_setpwent;
9877 if (name[3] == 'w' &&
9883 return -KEY_shmwrite;
9889 if (name[3] == 't' &&
9895 return -KEY_shutdown;
9905 if (name[2] == 's' &&
9912 return -KEY_syswrite;
9922 if (name[1] == 'r' &&
9930 return -KEY_truncate;
9939 case 9: /* 9 tokens of length 9 */
9943 if (name[1] == 'N' &&
9952 return KEY_UNITCHECK;
9958 if (name[1] == 'n' &&
9967 return -KEY_endnetent;
9973 if (name[1] == 'e' &&
9982 return -KEY_getnetent;
9988 if (name[1] == 'o' &&
9997 return -KEY_localtime;
10003 if (name[1] == 'r' &&
10012 return KEY_prototype;
10018 if (name[1] == 'u' &&
10027 return -KEY_quotemeta;
10033 if (name[1] == 'e' &&
10042 return -KEY_rewinddir;
10048 if (name[1] == 'e' &&
10057 return -KEY_setnetent;
10063 if (name[1] == 'a' &&
10072 return -KEY_wantarray;
10081 case 10: /* 9 tokens of length 10 */
10085 if (name[1] == 'n' &&
10091 if (name[4] == 'o' &&
10098 return -KEY_endhostent;
10104 if (name[4] == 'e' &&
10111 return -KEY_endservent;
10124 if (name[1] == 'e' &&
10130 if (name[4] == 'o' &&
10137 return -KEY_gethostent;
10146 if (name[5] == 'r' &&
10152 return -KEY_getservent;
10158 if (name[5] == 'c' &&
10164 return -KEY_getsockopt;
10184 if (name[2] == 't')
10189 if (name[4] == 'o' &&
10196 return -KEY_sethostent;
10205 if (name[5] == 'r' &&
10211 return -KEY_setservent;
10217 if (name[5] == 'c' &&
10223 return -KEY_setsockopt;
10240 if (name[2] == 'c' &&
10249 return -KEY_socketpair;
10262 case 11: /* 8 tokens of length 11 */
10266 if (name[1] == '_' &&
10276 { /* __PACKAGE__ */
10277 return -KEY___PACKAGE__;
10283 if (name[1] == 'n' &&
10293 { /* endprotoent */
10294 return -KEY_endprotoent;
10300 if (name[1] == 'e' &&
10309 if (name[5] == 'e' &&
10315 { /* getpeername */
10316 return -KEY_getpeername;
10325 if (name[6] == 'o' &&
10330 { /* getpriority */
10331 return -KEY_getpriority;
10337 if (name[6] == 't' &&
10342 { /* getprotoent */
10343 return -KEY_getprotoent;
10357 if (name[4] == 'o' &&
10364 { /* getsockname */
10365 return -KEY_getsockname;
10378 if (name[1] == 'e' &&
10386 if (name[6] == 'o' &&
10391 { /* setpriority */
10392 return -KEY_setpriority;
10398 if (name[6] == 't' &&
10403 { /* setprotoent */
10404 return -KEY_setprotoent;
10420 case 12: /* 2 tokens of length 12 */
10421 if (name[0] == 'g' &&
10433 if (name[9] == 'd' &&
10436 { /* getnetbyaddr */
10437 return -KEY_getnetbyaddr;
10443 if (name[9] == 'a' &&
10446 { /* getnetbyname */
10447 return -KEY_getnetbyname;
10459 case 13: /* 4 tokens of length 13 */
10460 if (name[0] == 'g' &&
10467 if (name[4] == 'o' &&
10476 if (name[10] == 'd' &&
10479 { /* gethostbyaddr */
10480 return -KEY_gethostbyaddr;
10486 if (name[10] == 'a' &&
10489 { /* gethostbyname */
10490 return -KEY_gethostbyname;
10503 if (name[4] == 'e' &&
10512 if (name[10] == 'a' &&
10515 { /* getservbyname */
10516 return -KEY_getservbyname;
10522 if (name[10] == 'o' &&
10525 { /* getservbyport */
10526 return -KEY_getservbyport;
10545 case 14: /* 1 tokens of length 14 */
10546 if (name[0] == 'g' &&
10560 { /* getprotobyname */
10561 return -KEY_getprotobyname;
10566 case 16: /* 1 tokens of length 16 */
10567 if (name[0] == 'g' &&
10583 { /* getprotobynumber */
10584 return -KEY_getprotobynumber;
10598 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10602 PERL_ARGS_ASSERT_CHECKCOMMA;
10604 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10605 if (ckWARN(WARN_SYNTAX)) {
10608 for (w = s+2; *w && level; w++) {
10611 else if (*w == ')')
10614 while (isSPACE(*w))
10616 /* the list of chars below is for end of statements or
10617 * block / parens, boolean operators (&&, ||, //) and branch
10618 * constructs (or, and, if, until, unless, while, err, for).
10619 * Not a very solid hack... */
10620 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10621 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10622 "%s (...) interpreted as function",name);
10625 while (s < PL_bufend && isSPACE(*s))
10629 while (s < PL_bufend && isSPACE(*s))
10631 if (isIDFIRST_lazy_if(s,UTF)) {
10632 const char * const w = s++;
10633 while (isALNUM_lazy_if(s,UTF))
10635 while (s < PL_bufend && isSPACE(*s))
10639 if (keyword(w, s - w, 0))
10642 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10643 if (gv && GvCVu(gv))
10645 Perl_croak(aTHX_ "No comma allowed after %s", what);
10650 /* Either returns sv, or mortalizes sv and returns a new SV*.
10651 Best used as sv=new_constant(..., sv, ...).
10652 If s, pv are NULL, calls subroutine with one argument,
10653 and type is used with error messages only. */
10656 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10657 SV *sv, SV *pv, const char *type, STRLEN typelen)
10660 HV * const table = GvHV(PL_hintgv); /* ^H */
10664 const char *why1 = "", *why2 = "", *why3 = "";
10666 PERL_ARGS_ASSERT_NEW_CONSTANT;
10668 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10671 why2 = (const char *)
10672 (strEQ(key,"charnames")
10673 ? "(possibly a missing \"use charnames ...\")"
10675 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10676 (type ? type: "undef"), why2);
10678 /* This is convoluted and evil ("goto considered harmful")
10679 * but I do not understand the intricacies of all the different
10680 * failure modes of %^H in here. The goal here is to make
10681 * the most probable error message user-friendly. --jhi */
10686 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10687 (type ? type: "undef"), why1, why2, why3);
10689 yyerror(SvPVX_const(msg));
10693 cvp = hv_fetch(table, key, keylen, FALSE);
10694 if (!cvp || !SvOK(*cvp)) {
10697 why3 = "} is not defined";
10700 sv_2mortal(sv); /* Parent created it permanently */
10703 pv = newSVpvn_flags(s, len, SVs_TEMP);
10705 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10707 typesv = &PL_sv_undef;
10709 PUSHSTACKi(PERLSI_OVERLOAD);
10721 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10725 /* Check the eval first */
10726 if (!PL_in_eval && SvTRUE(ERRSV)) {
10727 sv_catpvs(ERRSV, "Propagated");
10728 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10730 res = SvREFCNT_inc_simple(sv);
10734 SvREFCNT_inc_simple_void(res);
10743 why1 = "Call to &{$^H{";
10745 why3 = "}} did not return a defined value";
10753 /* Returns a NUL terminated string, with the length of the string written to
10757 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10760 register char *d = dest;
10761 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10763 PERL_ARGS_ASSERT_SCAN_WORD;
10767 Perl_croak(aTHX_ ident_too_long);
10768 if (isALNUM(*s)) /* UTF handled below */
10770 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10775 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10779 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10780 char *t = s + UTF8SKIP(s);
10782 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10786 Perl_croak(aTHX_ ident_too_long);
10787 Copy(s, d, len, char);
10800 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10803 char *bracket = NULL;
10805 register char *d = dest;
10806 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10808 PERL_ARGS_ASSERT_SCAN_IDENT;
10813 while (isDIGIT(*s)) {
10815 Perl_croak(aTHX_ ident_too_long);
10822 Perl_croak(aTHX_ ident_too_long);
10823 if (isALNUM(*s)) /* UTF handled below */
10825 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10830 else if (*s == ':' && s[1] == ':') {
10834 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10835 char *t = s + UTF8SKIP(s);
10836 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10838 if (d + (t - s) > e)
10839 Perl_croak(aTHX_ ident_too_long);
10840 Copy(s, d, t - s, char);
10851 if (PL_lex_state != LEX_NORMAL)
10852 PL_lex_state = LEX_INTERPENDMAYBE;
10855 if (*s == '$' && s[1] &&
10856 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10869 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10874 if (isSPACE(s[-1])) {
10876 const char ch = *s++;
10877 if (!SPACE_OR_TAB(ch)) {
10883 if (isIDFIRST_lazy_if(d,UTF)) {
10887 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10888 end += UTF8SKIP(end);
10889 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10890 end += UTF8SKIP(end);
10892 Copy(s, d, end - s, char);
10897 while ((isALNUM(*s) || *s == ':') && d < e)
10900 Perl_croak(aTHX_ ident_too_long);
10903 while (s < send && SPACE_OR_TAB(*s))
10905 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10906 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10907 const char * const brack =
10909 ((*s == '[') ? "[...]" : "{...}");
10910 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10911 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10912 funny, dest, brack, funny, dest, brack);
10915 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10919 /* Handle extended ${^Foo} variables
10920 * 1999-02-27 mjd-perl-patch@plover.com */
10921 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10925 while (isALNUM(*s) && d < e) {
10929 Perl_croak(aTHX_ ident_too_long);
10934 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10935 PL_lex_state = LEX_INTERPEND;
10938 if (PL_lex_state == LEX_NORMAL) {
10939 if (ckWARN(WARN_AMBIGUOUS) &&
10940 (keyword(dest, d - dest, 0)
10941 || get_cvn_flags(dest, d - dest, 0)))
10945 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10946 "Ambiguous use of %c{%s} resolved to %c%s",
10947 funny, dest, funny, dest);
10952 s = bracket; /* let the parser handle it */
10956 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10957 PL_lex_state = LEX_INTERPEND;
10962 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10964 PERL_ARGS_ASSERT_PMFLAG;
10966 PERL_UNUSED_CONTEXT;
10968 const char c = (char)ch;
10970 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10971 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10972 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10973 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10974 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10980 S_scan_pat(pTHX_ char *start, I32 type)
10984 char *s = scan_str(start,!!PL_madskills,FALSE);
10985 const char * const valid_flags =
10986 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10991 PERL_ARGS_ASSERT_SCAN_PAT;
10994 const char * const delimiter = skipspace(start);
10998 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10999 : "Search pattern not terminated" ));
11002 pm = (PMOP*)newPMOP(type, 0);
11003 if (PL_multi_open == '?') {
11004 /* This is the only point in the code that sets PMf_ONCE: */
11005 pm->op_pmflags |= PMf_ONCE;
11007 /* Hence it's safe to do this bit of PMOP book-keeping here, which
11008 allows us to restrict the list needed by reset to just the ??
11010 assert(type != OP_TRANS);
11012 MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
11015 mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
11018 elements = mg->mg_len / sizeof(PMOP**);
11019 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
11020 ((PMOP**)mg->mg_ptr) [elements++] = pm;
11021 mg->mg_len = elements * sizeof(PMOP**);
11022 PmopSTASH_set(pm,PL_curstash);
11028 while (*s && strchr(valid_flags, *s))
11029 pmflag(&pm->op_pmflags,*s++);
11031 if (PL_madskills && modstart != s) {
11032 SV* tmptoken = newSVpvn(modstart, s - modstart);
11033 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11036 /* issue a warning if /c is specified,but /g is not */
11037 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
11038 && ckWARN(WARN_REGEXP))
11040 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
11041 "Use of /c modifier is meaningless without /g" );
11044 PL_lex_op = (OP*)pm;
11045 pl_yylval.ival = OP_MATCH;
11050 S_scan_subst(pTHX_ char *start)
11061 PERL_ARGS_ASSERT_SCAN_SUBST;
11063 pl_yylval.ival = OP_NULL;
11065 s = scan_str(start,!!PL_madskills,FALSE);
11068 Perl_croak(aTHX_ "Substitution pattern not terminated");
11070 if (s[-1] == PL_multi_open)
11073 if (PL_madskills) {
11074 CURMAD('q', PL_thisopen);
11075 CURMAD('_', PL_thiswhite);
11076 CURMAD('E', PL_thisstuff);
11077 CURMAD('Q', PL_thisclose);
11078 PL_realtokenstart = s - SvPVX(PL_linestr);
11082 first_start = PL_multi_start;
11083 s = scan_str(s,!!PL_madskills,FALSE);
11085 if (PL_lex_stuff) {
11086 SvREFCNT_dec(PL_lex_stuff);
11087 PL_lex_stuff = NULL;
11089 Perl_croak(aTHX_ "Substitution replacement not terminated");
11091 PL_multi_start = first_start; /* so whole substitution is taken together */
11093 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11096 if (PL_madskills) {
11097 CURMAD('z', PL_thisopen);
11098 CURMAD('R', PL_thisstuff);
11099 CURMAD('Z', PL_thisclose);
11105 if (*s == EXEC_PAT_MOD) {
11109 else if (strchr(S_PAT_MODS, *s))
11110 pmflag(&pm->op_pmflags,*s++);
11116 if (PL_madskills) {
11118 curmad('m', newSVpvn(modstart, s - modstart));
11119 append_madprops(PL_thismad, (OP*)pm, 0);
11123 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11124 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11128 SV * const repl = newSVpvs("");
11130 PL_sublex_info.super_bufptr = s;
11131 PL_sublex_info.super_bufend = PL_bufend;
11133 pm->op_pmflags |= PMf_EVAL;
11136 sv_catpvs(repl, "eval ");
11138 sv_catpvs(repl, "do ");
11140 sv_catpvs(repl, "{");
11141 sv_catsv(repl, PL_lex_repl);
11142 if (strchr(SvPVX(PL_lex_repl), '#'))
11143 sv_catpvs(repl, "\n");
11144 sv_catpvs(repl, "}");
11146 SvREFCNT_dec(PL_lex_repl);
11147 PL_lex_repl = repl;
11150 PL_lex_op = (OP*)pm;
11151 pl_yylval.ival = OP_SUBST;
11156 S_scan_trans(pTHX_ char *start)
11169 PERL_ARGS_ASSERT_SCAN_TRANS;
11171 pl_yylval.ival = OP_NULL;
11173 s = scan_str(start,!!PL_madskills,FALSE);
11175 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11177 if (s[-1] == PL_multi_open)
11180 if (PL_madskills) {
11181 CURMAD('q', PL_thisopen);
11182 CURMAD('_', PL_thiswhite);
11183 CURMAD('E', PL_thisstuff);
11184 CURMAD('Q', PL_thisclose);
11185 PL_realtokenstart = s - SvPVX(PL_linestr);
11189 s = scan_str(s,!!PL_madskills,FALSE);
11191 if (PL_lex_stuff) {
11192 SvREFCNT_dec(PL_lex_stuff);
11193 PL_lex_stuff = NULL;
11195 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11197 if (PL_madskills) {
11198 CURMAD('z', PL_thisopen);
11199 CURMAD('R', PL_thisstuff);
11200 CURMAD('Z', PL_thisclose);
11203 complement = del = squash = 0;
11210 complement = OPpTRANS_COMPLEMENT;
11213 del = OPpTRANS_DELETE;
11216 squash = OPpTRANS_SQUASH;
11225 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11226 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11227 o->op_private &= ~OPpTRANS_ALL;
11228 o->op_private |= del|squash|complement|
11229 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11230 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11233 pl_yylval.ival = OP_TRANS;
11236 if (PL_madskills) {
11238 curmad('m', newSVpvn(modstart, s - modstart));
11239 append_madprops(PL_thismad, o, 0);
11248 S_scan_heredoc(pTHX_ register char *s)
11252 I32 op_type = OP_SCALAR;
11256 const char *found_newline;
11260 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11262 I32 stuffstart = s - SvPVX(PL_linestr);
11265 PL_realtokenstart = -1;
11268 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11272 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11276 while (SPACE_OR_TAB(*peek))
11278 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11281 s = delimcpy(d, e, s, PL_bufend, term, &len);
11291 if (!isALNUM_lazy_if(s,UTF))
11292 deprecate_old("bare << to mean <<\"\"");
11293 for (; isALNUM_lazy_if(s,UTF); s++) {
11298 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11299 Perl_croak(aTHX_ "Delimiter for here document is too long");
11302 len = d - PL_tokenbuf;
11305 if (PL_madskills) {
11306 tstart = PL_tokenbuf + !outer;
11307 PL_thisclose = newSVpvn(tstart, len - !outer);
11308 tstart = SvPVX(PL_linestr) + stuffstart;
11309 PL_thisopen = newSVpvn(tstart, s - tstart);
11310 stuffstart = s - SvPVX(PL_linestr);
11313 #ifndef PERL_STRICT_CR
11314 d = strchr(s, '\r');
11316 char * const olds = s;
11318 while (s < PL_bufend) {
11324 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11333 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11340 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11341 herewas = newSVpvn(s,PL_bufend-s);
11345 herewas = newSVpvn(s-1,found_newline-s+1);
11348 herewas = newSVpvn(s,found_newline-s);
11352 if (PL_madskills) {
11353 tstart = SvPVX(PL_linestr) + stuffstart;
11355 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11357 PL_thisstuff = newSVpvn(tstart, s - tstart);
11360 s += SvCUR(herewas);
11363 stuffstart = s - SvPVX(PL_linestr);
11369 tmpstr = newSV_type(SVt_PVIV);
11370 SvGROW(tmpstr, 80);
11371 if (term == '\'') {
11372 op_type = OP_CONST;
11373 SvIV_set(tmpstr, -1);
11375 else if (term == '`') {
11376 op_type = OP_BACKTICK;
11377 SvIV_set(tmpstr, '\\');
11381 PL_multi_start = CopLINE(PL_curcop);
11382 PL_multi_open = PL_multi_close = '<';
11383 term = *PL_tokenbuf;
11384 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11385 char * const bufptr = PL_sublex_info.super_bufptr;
11386 char * const bufend = PL_sublex_info.super_bufend;
11387 char * const olds = s - SvCUR(herewas);
11388 s = strchr(bufptr, '\n');
11392 while (s < bufend &&
11393 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11395 CopLINE_inc(PL_curcop);
11398 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11399 missingterm(PL_tokenbuf);
11401 sv_setpvn(herewas,bufptr,d-bufptr+1);
11402 sv_setpvn(tmpstr,d+1,s-d);
11404 sv_catpvn(herewas,s,bufend-s);
11405 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11412 while (s < PL_bufend &&
11413 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11415 CopLINE_inc(PL_curcop);
11417 if (s >= PL_bufend) {
11418 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11419 missingterm(PL_tokenbuf);
11421 sv_setpvn(tmpstr,d+1,s-d);
11423 if (PL_madskills) {
11425 sv_catpvn(PL_thisstuff, d + 1, s - d);
11427 PL_thisstuff = newSVpvn(d + 1, s - d);
11428 stuffstart = s - SvPVX(PL_linestr);
11432 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11434 sv_catpvn(herewas,s,PL_bufend-s);
11435 sv_setsv(PL_linestr,herewas);
11436 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11437 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11438 PL_last_lop = PL_last_uni = NULL;
11441 sv_setpvs(tmpstr,""); /* avoid "uninitialized" warning */
11442 while (s >= PL_bufend) { /* multiple line string? */
11444 if (PL_madskills) {
11445 tstart = SvPVX(PL_linestr) + stuffstart;
11447 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11449 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11453 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11454 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11455 missingterm(PL_tokenbuf);
11458 stuffstart = s - SvPVX(PL_linestr);
11460 CopLINE_inc(PL_curcop);
11461 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11462 PL_last_lop = PL_last_uni = NULL;
11463 #ifndef PERL_STRICT_CR
11464 if (PL_bufend - PL_linestart >= 2) {
11465 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11466 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11468 PL_bufend[-2] = '\n';
11470 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11472 else if (PL_bufend[-1] == '\r')
11473 PL_bufend[-1] = '\n';
11475 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11476 PL_bufend[-1] = '\n';
11478 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11479 update_debugger_info(PL_linestr, NULL, 0);
11480 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11481 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11482 *(SvPVX(PL_linestr) + off ) = ' ';
11483 sv_catsv(PL_linestr,herewas);
11484 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11485 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11489 sv_catsv(tmpstr,PL_linestr);
11494 PL_multi_end = CopLINE(PL_curcop);
11495 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11496 SvPV_shrink_to_cur(tmpstr);
11498 SvREFCNT_dec(herewas);
11500 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11502 else if (PL_encoding)
11503 sv_recode_to_utf8(tmpstr, PL_encoding);
11505 PL_lex_stuff = tmpstr;
11506 pl_yylval.ival = op_type;
11510 /* scan_inputsymbol
11511 takes: current position in input buffer
11512 returns: new position in input buffer
11513 side-effects: pl_yylval and lex_op are set.
11518 <FH> read from filehandle
11519 <pkg::FH> read from package qualified filehandle
11520 <pkg'FH> read from package qualified filehandle
11521 <$fh> read from filehandle in $fh
11522 <*.h> filename glob
11527 S_scan_inputsymbol(pTHX_ char *start)
11530 register char *s = start; /* current position in buffer */
11533 char *d = PL_tokenbuf; /* start of temp holding space */
11534 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11536 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11538 end = strchr(s, '\n');
11541 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11543 /* die if we didn't have space for the contents of the <>,
11544 or if it didn't end, or if we see a newline
11547 if (len >= (I32)sizeof PL_tokenbuf)
11548 Perl_croak(aTHX_ "Excessively long <> operator");
11550 Perl_croak(aTHX_ "Unterminated <> operator");
11555 Remember, only scalar variables are interpreted as filehandles by
11556 this code. Anything more complex (e.g., <$fh{$num}>) will be
11557 treated as a glob() call.
11558 This code makes use of the fact that except for the $ at the front,
11559 a scalar variable and a filehandle look the same.
11561 if (*d == '$' && d[1]) d++;
11563 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11564 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11567 /* If we've tried to read what we allow filehandles to look like, and
11568 there's still text left, then it must be a glob() and not a getline.
11569 Use scan_str to pull out the stuff between the <> and treat it
11570 as nothing more than a string.
11573 if (d - PL_tokenbuf != len) {
11574 pl_yylval.ival = OP_GLOB;
11575 s = scan_str(start,!!PL_madskills,FALSE);
11577 Perl_croak(aTHX_ "Glob not terminated");
11581 bool readline_overriden = FALSE;
11584 /* we're in a filehandle read situation */
11587 /* turn <> into <ARGV> */
11589 Copy("ARGV",d,5,char);
11591 /* Check whether readline() is overriden */
11592 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11594 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11596 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11597 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11598 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11599 readline_overriden = TRUE;
11601 /* if <$fh>, create the ops to turn the variable into a
11605 /* try to find it in the pad for this block, otherwise find
11606 add symbol table ops
11608 const PADOFFSET tmp = pad_findmy(d);
11609 if (tmp != NOT_IN_PAD) {
11610 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11611 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11612 HEK * const stashname = HvNAME_HEK(stash);
11613 SV * const sym = sv_2mortal(newSVhek(stashname));
11614 sv_catpvs(sym, "::");
11615 sv_catpv(sym, d+1);
11620 OP * const o = newOP(OP_PADSV, 0);
11622 PL_lex_op = readline_overriden
11623 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11624 append_elem(OP_LIST, o,
11625 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11626 : (OP*)newUNOP(OP_READLINE, 0, o);
11635 ? (GV_ADDMULTI | GV_ADDINEVAL)
11638 PL_lex_op = readline_overriden
11639 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11640 append_elem(OP_LIST,
11641 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11642 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11643 : (OP*)newUNOP(OP_READLINE, 0,
11644 newUNOP(OP_RV2SV, 0,
11645 newGVOP(OP_GV, 0, gv)));
11647 if (!readline_overriden)
11648 PL_lex_op->op_flags |= OPf_SPECIAL;
11649 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11650 pl_yylval.ival = OP_NULL;
11653 /* If it's none of the above, it must be a literal filehandle
11654 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11656 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11657 PL_lex_op = readline_overriden
11658 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11659 append_elem(OP_LIST,
11660 newGVOP(OP_GV, 0, gv),
11661 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11662 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11663 pl_yylval.ival = OP_NULL;
11672 takes: start position in buffer
11673 keep_quoted preserve \ on the embedded delimiter(s)
11674 keep_delims preserve the delimiters around the string
11675 returns: position to continue reading from buffer
11676 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11677 updates the read buffer.
11679 This subroutine pulls a string out of the input. It is called for:
11680 q single quotes q(literal text)
11681 ' single quotes 'literal text'
11682 qq double quotes qq(interpolate $here please)
11683 " double quotes "interpolate $here please"
11684 qx backticks qx(/bin/ls -l)
11685 ` backticks `/bin/ls -l`
11686 qw quote words @EXPORT_OK = qw( func() $spam )
11687 m// regexp match m/this/
11688 s/// regexp substitute s/this/that/
11689 tr/// string transliterate tr/this/that/
11690 y/// string transliterate y/this/that/
11691 ($*@) sub prototypes sub foo ($)
11692 (stuff) sub attr parameters sub foo : attr(stuff)
11693 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11695 In most of these cases (all but <>, patterns and transliterate)
11696 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11697 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11698 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11701 It skips whitespace before the string starts, and treats the first
11702 character as the delimiter. If the delimiter is one of ([{< then
11703 the corresponding "close" character )]}> is used as the closing
11704 delimiter. It allows quoting of delimiters, and if the string has
11705 balanced delimiters ([{<>}]) it allows nesting.
11707 On success, the SV with the resulting string is put into lex_stuff or,
11708 if that is already non-NULL, into lex_repl. The second case occurs only
11709 when parsing the RHS of the special constructs s/// and tr/// (y///).
11710 For convenience, the terminating delimiter character is stuffed into
11715 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11718 SV *sv; /* scalar value: string */
11719 const char *tmps; /* temp string, used for delimiter matching */
11720 register char *s = start; /* current position in the buffer */
11721 register char term; /* terminating character */
11722 register char *to; /* current position in the sv's data */
11723 I32 brackets = 1; /* bracket nesting level */
11724 bool has_utf8 = FALSE; /* is there any utf8 content? */
11725 I32 termcode; /* terminating char. code */
11726 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11727 STRLEN termlen; /* length of terminating string */
11728 int last_off = 0; /* last position for nesting bracket */
11734 PERL_ARGS_ASSERT_SCAN_STR;
11736 /* skip space before the delimiter */
11742 if (PL_realtokenstart >= 0) {
11743 stuffstart = PL_realtokenstart;
11744 PL_realtokenstart = -1;
11747 stuffstart = start - SvPVX(PL_linestr);
11749 /* mark where we are, in case we need to report errors */
11752 /* after skipping whitespace, the next character is the terminator */
11755 termcode = termstr[0] = term;
11759 termcode = utf8_to_uvchr((U8*)s, &termlen);
11760 Copy(s, termstr, termlen, U8);
11761 if (!UTF8_IS_INVARIANT(term))
11765 /* mark where we are */
11766 PL_multi_start = CopLINE(PL_curcop);
11767 PL_multi_open = term;
11769 /* find corresponding closing delimiter */
11770 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11771 termcode = termstr[0] = term = tmps[5];
11773 PL_multi_close = term;
11775 /* create a new SV to hold the contents. 79 is the SV's initial length.
11776 What a random number. */
11777 sv = newSV_type(SVt_PVIV);
11779 SvIV_set(sv, termcode);
11780 (void)SvPOK_only(sv); /* validate pointer */
11782 /* move past delimiter and try to read a complete string */
11784 sv_catpvn(sv, s, termlen);
11787 tstart = SvPVX(PL_linestr) + stuffstart;
11788 if (!PL_thisopen && !keep_delims) {
11789 PL_thisopen = newSVpvn(tstart, s - tstart);
11790 stuffstart = s - SvPVX(PL_linestr);
11794 if (PL_encoding && !UTF) {
11798 int offset = s - SvPVX_const(PL_linestr);
11799 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11800 &offset, (char*)termstr, termlen);
11801 const char * const ns = SvPVX_const(PL_linestr) + offset;
11802 char * const svlast = SvEND(sv) - 1;
11804 for (; s < ns; s++) {
11805 if (*s == '\n' && !PL_rsfp)
11806 CopLINE_inc(PL_curcop);
11809 goto read_more_line;
11811 /* handle quoted delimiters */
11812 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11814 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11816 if ((svlast-1 - t) % 2) {
11817 if (!keep_quoted) {
11818 *(svlast-1) = term;
11820 SvCUR_set(sv, SvCUR(sv) - 1);
11825 if (PL_multi_open == PL_multi_close) {
11831 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11832 /* At here, all closes are "was quoted" one,
11833 so we don't check PL_multi_close. */
11835 if (!keep_quoted && *(t+1) == PL_multi_open)
11840 else if (*t == PL_multi_open)
11848 SvCUR_set(sv, w - SvPVX_const(sv));
11850 last_off = w - SvPVX(sv);
11851 if (--brackets <= 0)
11856 if (!keep_delims) {
11857 SvCUR_set(sv, SvCUR(sv) - 1);
11863 /* extend sv if need be */
11864 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11865 /* set 'to' to the next character in the sv's string */
11866 to = SvPVX(sv)+SvCUR(sv);
11868 /* if open delimiter is the close delimiter read unbridle */
11869 if (PL_multi_open == PL_multi_close) {
11870 for (; s < PL_bufend; s++,to++) {
11871 /* embedded newlines increment the current line number */
11872 if (*s == '\n' && !PL_rsfp)
11873 CopLINE_inc(PL_curcop);
11874 /* handle quoted delimiters */
11875 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11876 if (!keep_quoted && s[1] == term)
11878 /* any other quotes are simply copied straight through */
11882 /* terminate when run out of buffer (the for() condition), or
11883 have found the terminator */
11884 else if (*s == term) {
11887 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11890 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11896 /* if the terminator isn't the same as the start character (e.g.,
11897 matched brackets), we have to allow more in the quoting, and
11898 be prepared for nested brackets.
11901 /* read until we run out of string, or we find the terminator */
11902 for (; s < PL_bufend; s++,to++) {
11903 /* embedded newlines increment the line count */
11904 if (*s == '\n' && !PL_rsfp)
11905 CopLINE_inc(PL_curcop);
11906 /* backslashes can escape the open or closing characters */
11907 if (*s == '\\' && s+1 < PL_bufend) {
11908 if (!keep_quoted &&
11909 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11914 /* allow nested opens and closes */
11915 else if (*s == PL_multi_close && --brackets <= 0)
11917 else if (*s == PL_multi_open)
11919 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11924 /* terminate the copied string and update the sv's end-of-string */
11926 SvCUR_set(sv, to - SvPVX_const(sv));
11929 * this next chunk reads more into the buffer if we're not done yet
11933 break; /* handle case where we are done yet :-) */
11935 #ifndef PERL_STRICT_CR
11936 if (to - SvPVX_const(sv) >= 2) {
11937 if ((to[-2] == '\r' && to[-1] == '\n') ||
11938 (to[-2] == '\n' && to[-1] == '\r'))
11942 SvCUR_set(sv, to - SvPVX_const(sv));
11944 else if (to[-1] == '\r')
11947 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11952 /* if we're out of file, or a read fails, bail and reset the current
11953 line marker so we can report where the unterminated string began
11956 if (PL_madskills) {
11957 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11959 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11961 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11965 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11967 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11973 /* we read a line, so increment our line counter */
11974 CopLINE_inc(PL_curcop);
11976 /* update debugger info */
11977 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11978 update_debugger_info(PL_linestr, NULL, 0);
11980 /* having changed the buffer, we must update PL_bufend */
11981 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11982 PL_last_lop = PL_last_uni = NULL;
11985 /* at this point, we have successfully read the delimited string */
11987 if (!PL_encoding || UTF) {
11989 if (PL_madskills) {
11990 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11991 const int len = s - tstart;
11993 sv_catpvn(PL_thisstuff, tstart, len);
11995 PL_thisstuff = newSVpvn(tstart, len);
11996 if (!PL_thisclose && !keep_delims)
11997 PL_thisclose = newSVpvn(s,termlen);
12002 sv_catpvn(sv, s, termlen);
12007 if (PL_madskills) {
12008 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12009 const int len = s - tstart - termlen;
12011 sv_catpvn(PL_thisstuff, tstart, len);
12013 PL_thisstuff = newSVpvn(tstart, len);
12014 if (!PL_thisclose && !keep_delims)
12015 PL_thisclose = newSVpvn(s - termlen,termlen);
12019 if (has_utf8 || PL_encoding)
12022 PL_multi_end = CopLINE(PL_curcop);
12024 /* if we allocated too much space, give some back */
12025 if (SvCUR(sv) + 5 < SvLEN(sv)) {
12026 SvLEN_set(sv, SvCUR(sv) + 1);
12027 SvPV_renew(sv, SvLEN(sv));
12030 /* decide whether this is the first or second quoted string we've read
12043 takes: pointer to position in buffer
12044 returns: pointer to new position in buffer
12045 side-effects: builds ops for the constant in pl_yylval.op
12047 Read a number in any of the formats that Perl accepts:
12049 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12050 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12053 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12055 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12058 If it reads a number without a decimal point or an exponent, it will
12059 try converting the number to an integer and see if it can do so
12060 without loss of precision.
12064 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12067 register const char *s = start; /* current position in buffer */
12068 register char *d; /* destination in temp buffer */
12069 register char *e; /* end of temp buffer */
12070 NV nv; /* number read, as a double */
12071 SV *sv = NULL; /* place to put the converted number */
12072 bool floatit; /* boolean: int or float? */
12073 const char *lastub = NULL; /* position of last underbar */
12074 static char const number_too_long[] = "Number too long";
12076 PERL_ARGS_ASSERT_SCAN_NUM;
12078 /* We use the first character to decide what type of number this is */
12082 Perl_croak(aTHX_ "panic: scan_num");
12084 /* if it starts with a 0, it could be an octal number, a decimal in
12085 0.13 disguise, or a hexadecimal number, or a binary number. */
12089 u holds the "number so far"
12090 shift the power of 2 of the base
12091 (hex == 4, octal == 3, binary == 1)
12092 overflowed was the number more than we can hold?
12094 Shift is used when we add a digit. It also serves as an "are
12095 we in octal/hex/binary?" indicator to disallow hex characters
12096 when in octal mode.
12101 bool overflowed = FALSE;
12102 bool just_zero = TRUE; /* just plain 0 or binary number? */
12103 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12104 static const char* const bases[5] =
12105 { "", "binary", "", "octal", "hexadecimal" };
12106 static const char* const Bases[5] =
12107 { "", "Binary", "", "Octal", "Hexadecimal" };
12108 static const char* const maxima[5] =
12110 "0b11111111111111111111111111111111",
12114 const char *base, *Base, *max;
12116 /* check for hex */
12121 } else if (s[1] == 'b') {
12126 /* check for a decimal in disguise */
12127 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12129 /* so it must be octal */
12136 if (ckWARN(WARN_SYNTAX))
12137 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12138 "Misplaced _ in number");
12142 base = bases[shift];
12143 Base = Bases[shift];
12144 max = maxima[shift];
12146 /* read the rest of the number */
12148 /* x is used in the overflow test,
12149 b is the digit we're adding on. */
12154 /* if we don't mention it, we're done */
12158 /* _ are ignored -- but warned about if consecutive */
12160 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12161 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12162 "Misplaced _ in number");
12166 /* 8 and 9 are not octal */
12167 case '8': case '9':
12169 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12173 case '2': case '3': case '4':
12174 case '5': case '6': case '7':
12176 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12179 case '0': case '1':
12180 b = *s++ & 15; /* ASCII digit -> value of digit */
12184 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12185 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12186 /* make sure they said 0x */
12189 b = (*s++ & 7) + 9;
12191 /* Prepare to put the digit we have onto the end
12192 of the number so far. We check for overflows.
12198 x = u << shift; /* make room for the digit */
12200 if ((x >> shift) != u
12201 && !(PL_hints & HINT_NEW_BINARY)) {
12204 if (ckWARN_d(WARN_OVERFLOW))
12205 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12206 "Integer overflow in %s number",
12209 u = x | b; /* add the digit to the end */
12212 n *= nvshift[shift];
12213 /* If an NV has not enough bits in its
12214 * mantissa to represent an UV this summing of
12215 * small low-order numbers is a waste of time
12216 * (because the NV cannot preserve the
12217 * low-order bits anyway): we could just
12218 * remember when did we overflow and in the
12219 * end just multiply n by the right
12227 /* if we get here, we had success: make a scalar value from
12232 /* final misplaced underbar check */
12233 if (s[-1] == '_') {
12234 if (ckWARN(WARN_SYNTAX))
12235 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12240 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12241 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12242 "%s number > %s non-portable",
12248 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12249 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12250 "%s number > %s non-portable",
12255 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12256 sv = new_constant(start, s - start, "integer",
12257 sv, NULL, NULL, 0);
12258 else if (PL_hints & HINT_NEW_BINARY)
12259 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12264 handle decimal numbers.
12265 we're also sent here when we read a 0 as the first digit
12267 case '1': case '2': case '3': case '4': case '5':
12268 case '6': case '7': case '8': case '9': case '.':
12271 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12274 /* read next group of digits and _ and copy into d */
12275 while (isDIGIT(*s) || *s == '_') {
12276 /* skip underscores, checking for misplaced ones
12280 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12281 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12282 "Misplaced _ in number");
12286 /* check for end of fixed-length buffer */
12288 Perl_croak(aTHX_ number_too_long);
12289 /* if we're ok, copy the character */
12294 /* final misplaced underbar check */
12295 if (lastub && s == lastub + 1) {
12296 if (ckWARN(WARN_SYNTAX))
12297 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12300 /* read a decimal portion if there is one. avoid
12301 3..5 being interpreted as the number 3. followed
12304 if (*s == '.' && s[1] != '.') {
12309 if (ckWARN(WARN_SYNTAX))
12310 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12311 "Misplaced _ in number");
12315 /* copy, ignoring underbars, until we run out of digits.
12317 for (; isDIGIT(*s) || *s == '_'; s++) {
12318 /* fixed length buffer check */
12320 Perl_croak(aTHX_ number_too_long);
12322 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12323 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12324 "Misplaced _ in number");
12330 /* fractional part ending in underbar? */
12331 if (s[-1] == '_') {
12332 if (ckWARN(WARN_SYNTAX))
12333 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12334 "Misplaced _ in number");
12336 if (*s == '.' && isDIGIT(s[1])) {
12337 /* oops, it's really a v-string, but without the "v" */
12343 /* read exponent part, if present */
12344 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12348 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12349 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12351 /* stray preinitial _ */
12353 if (ckWARN(WARN_SYNTAX))
12354 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12355 "Misplaced _ in number");
12359 /* allow positive or negative exponent */
12360 if (*s == '+' || *s == '-')
12363 /* stray initial _ */
12365 if (ckWARN(WARN_SYNTAX))
12366 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12367 "Misplaced _ in number");
12371 /* read digits of exponent */
12372 while (isDIGIT(*s) || *s == '_') {
12375 Perl_croak(aTHX_ number_too_long);
12379 if (((lastub && s == lastub + 1) ||
12380 (!isDIGIT(s[1]) && s[1] != '_'))
12381 && ckWARN(WARN_SYNTAX))
12382 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12383 "Misplaced _ in number");
12390 /* make an sv from the string */
12394 We try to do an integer conversion first if no characters
12395 indicating "float" have been found.
12400 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12402 if (flags == IS_NUMBER_IN_UV) {
12404 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12407 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12408 if (uv <= (UV) IV_MIN)
12409 sv_setiv(sv, -(IV)uv);
12416 /* terminate the string */
12418 nv = Atof(PL_tokenbuf);
12423 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12424 const char *const key = floatit ? "float" : "integer";
12425 const STRLEN keylen = floatit ? 5 : 7;
12426 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12427 key, keylen, sv, NULL, NULL, 0);
12431 /* if it starts with a v, it could be a v-string */
12434 sv = newSV(5); /* preallocate storage space */
12435 s = scan_vstring(s, PL_bufend, sv);
12439 /* make the op for the constant and return */
12442 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12444 lvalp->opval = NULL;
12450 S_scan_formline(pTHX_ register char *s)
12453 register char *eol;
12455 SV * const stuff = newSVpvs("");
12456 bool needargs = FALSE;
12457 bool eofmt = FALSE;
12459 char *tokenstart = s;
12460 SV* savewhite = NULL;
12462 if (PL_madskills) {
12463 savewhite = PL_thiswhite;
12468 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12470 while (!needargs) {
12473 #ifdef PERL_STRICT_CR
12474 while (SPACE_OR_TAB(*t))
12477 while (SPACE_OR_TAB(*t) || *t == '\r')
12480 if (*t == '\n' || t == PL_bufend) {
12485 if (PL_in_eval && !PL_rsfp) {
12486 eol = (char *) memchr(s,'\n',PL_bufend-s);
12491 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12493 for (t = s; t < eol; t++) {
12494 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12496 goto enough; /* ~~ must be first line in formline */
12498 if (*t == '@' || *t == '^')
12502 sv_catpvn(stuff, s, eol-s);
12503 #ifndef PERL_STRICT_CR
12504 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12505 char *end = SvPVX(stuff) + SvCUR(stuff);
12508 SvCUR_set(stuff, SvCUR(stuff) - 1);
12518 if (PL_madskills) {
12520 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12522 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12525 s = filter_gets(PL_linestr, PL_rsfp, 0);
12527 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12529 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12531 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12532 PL_last_lop = PL_last_uni = NULL;
12541 if (SvCUR(stuff)) {
12544 PL_lex_state = LEX_NORMAL;
12545 start_force(PL_curforce);
12546 NEXTVAL_NEXTTOKE.ival = 0;
12550 PL_lex_state = LEX_FORMLINE;
12552 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12554 else if (PL_encoding)
12555 sv_recode_to_utf8(stuff, PL_encoding);
12557 start_force(PL_curforce);
12558 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12560 start_force(PL_curforce);
12561 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12565 SvREFCNT_dec(stuff);
12567 PL_lex_formbrack = 0;
12571 if (PL_madskills) {
12573 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12575 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12576 PL_thiswhite = savewhite;
12583 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12586 const I32 oldsavestack_ix = PL_savestack_ix;
12587 CV* const outsidecv = PL_compcv;
12590 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12592 SAVEI32(PL_subline);
12593 save_item(PL_subname);
12594 SAVESPTR(PL_compcv);
12596 PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
12597 CvFLAGS(PL_compcv) |= flags;
12599 PL_subline = CopLINE(PL_curcop);
12600 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12601 CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
12602 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12604 return oldsavestack_ix;
12608 #pragma segment Perl_yylex
12611 S_yywarn(pTHX_ const char *const s)
12615 PERL_ARGS_ASSERT_YYWARN;
12617 PL_in_eval |= EVAL_WARNONLY;
12619 PL_in_eval &= ~EVAL_WARNONLY;
12624 Perl_yyerror(pTHX_ const char *const s)
12627 const char *where = NULL;
12628 const char *context = NULL;
12631 int yychar = PL_parser->yychar;
12633 PERL_ARGS_ASSERT_YYERROR;
12635 if (!yychar || (yychar == ';' && !PL_rsfp))
12637 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12638 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12639 PL_oldbufptr != PL_bufptr) {
12642 The code below is removed for NetWare because it abends/crashes on NetWare
12643 when the script has error such as not having the closing quotes like:
12644 if ($var eq "value)
12645 Checking of white spaces is anyway done in NetWare code.
12648 while (isSPACE(*PL_oldoldbufptr))
12651 context = PL_oldoldbufptr;
12652 contlen = PL_bufptr - PL_oldoldbufptr;
12654 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12655 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12658 The code below is removed for NetWare because it abends/crashes on NetWare
12659 when the script has error such as not having the closing quotes like:
12660 if ($var eq "value)
12661 Checking of white spaces is anyway done in NetWare code.
12664 while (isSPACE(*PL_oldbufptr))
12667 context = PL_oldbufptr;
12668 contlen = PL_bufptr - PL_oldbufptr;
12670 else if (yychar > 255)
12671 where = "next token ???";
12672 else if (yychar == -2) { /* YYEMPTY */
12673 if (PL_lex_state == LEX_NORMAL ||
12674 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12675 where = "at end of line";
12676 else if (PL_lex_inpat)
12677 where = "within pattern";
12679 where = "within string";
12682 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12684 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12685 else if (isPRINT_LC(yychar)) {
12686 const char string = yychar;
12687 sv_catpvn(where_sv, &string, 1);
12690 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12691 where = SvPVX_const(where_sv);
12693 msg = sv_2mortal(newSVpv(s, 0));
12694 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12695 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12697 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12699 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12700 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12701 Perl_sv_catpvf(aTHX_ msg,
12702 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12703 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12706 if (PL_in_eval & EVAL_WARNONLY) {
12707 if (ckWARN_d(WARN_SYNTAX))
12708 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12712 if (PL_error_count >= 10) {
12713 if (PL_in_eval && SvCUR(ERRSV))
12714 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12715 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12717 Perl_croak(aTHX_ "%s has too many errors.\n",
12718 OutCopFILE(PL_curcop));
12721 PL_in_my_stash = NULL;
12725 #pragma segment Main
12729 S_swallow_bom(pTHX_ U8 *s)
12732 const STRLEN slen = SvCUR(PL_linestr);
12734 PERL_ARGS_ASSERT_SWALLOW_BOM;
12738 if (s[1] == 0xFE) {
12739 /* UTF-16 little-endian? (or UTF32-LE?) */
12740 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12741 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12742 #ifndef PERL_NO_UTF16_FILTER
12743 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12746 if (PL_bufend > (char*)s) {
12750 filter_add(utf16rev_textfilter, NULL);
12751 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12752 utf16_to_utf8_reversed(s, news,
12753 PL_bufend - (char*)s - 1,
12755 sv_setpvn(PL_linestr, (const char*)news, newlen);
12757 s = (U8*)SvPVX(PL_linestr);
12758 Copy(news, s, newlen, U8);
12762 SvUTF8_on(PL_linestr);
12763 s = (U8*)SvPVX(PL_linestr);
12765 /* FIXME - is this a general bug fix? */
12768 PL_bufend = SvPVX(PL_linestr) + newlen;
12771 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12776 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12777 #ifndef PERL_NO_UTF16_FILTER
12778 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12781 if (PL_bufend > (char *)s) {
12785 filter_add(utf16_textfilter, NULL);
12786 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12787 utf16_to_utf8(s, news,
12788 PL_bufend - (char*)s,
12790 sv_setpvn(PL_linestr, (const char*)news, newlen);
12792 SvUTF8_on(PL_linestr);
12793 s = (U8*)SvPVX(PL_linestr);
12794 PL_bufend = SvPVX(PL_linestr) + newlen;
12797 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12802 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12803 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12804 s += 3; /* UTF-8 */
12810 if (s[2] == 0xFE && s[3] == 0xFF) {
12811 /* UTF-32 big-endian */
12812 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12815 else if (s[2] == 0 && s[3] != 0) {
12818 * are a good indicator of UTF-16BE. */
12819 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12825 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12826 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12827 s += 4; /* UTF-8 */
12833 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12836 * are a good indicator of UTF-16LE. */
12837 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12845 #ifndef PERL_NO_UTF16_FILTER
12847 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12850 const STRLEN old = SvCUR(sv);
12851 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12852 DEBUG_P(PerlIO_printf(Perl_debug_log,
12853 "utf16_textfilter(%p): %d %d (%d)\n",
12854 FPTR2DPTR(void *, utf16_textfilter),
12855 idx, maxlen, (int) count));
12859 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12860 Copy(SvPVX_const(sv), tmps, old, char);
12861 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12862 SvCUR(sv) - old, &newlen);
12863 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12865 DEBUG_P({sv_dump(sv);});
12870 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12873 const STRLEN old = SvCUR(sv);
12874 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12875 DEBUG_P(PerlIO_printf(Perl_debug_log,
12876 "utf16rev_textfilter(%p): %d %d (%d)\n",
12877 FPTR2DPTR(void *, utf16rev_textfilter),
12878 idx, maxlen, (int) count));
12882 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12883 Copy(SvPVX_const(sv), tmps, old, char);
12884 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12885 SvCUR(sv) - old, &newlen);
12886 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12888 DEBUG_P({ sv_dump(sv); });
12894 Returns a pointer to the next character after the parsed
12895 vstring, as well as updating the passed in sv.
12897 Function must be called like
12900 s = scan_vstring(s,e,sv);
12902 where s and e are the start and end of the string.
12903 The sv should already be large enough to store the vstring
12904 passed in, for performance reasons.
12909 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12912 const char *pos = s;
12913 const char *start = s;
12915 PERL_ARGS_ASSERT_SCAN_VSTRING;
12917 if (*pos == 'v') pos++; /* get past 'v' */
12918 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12920 if ( *pos != '.') {
12921 /* this may not be a v-string if followed by => */
12922 const char *next = pos;
12923 while (next < e && isSPACE(*next))
12925 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12926 /* return string not v-string */
12927 sv_setpvn(sv,(char *)s,pos-s);
12928 return (char *)pos;
12932 if (!isALPHA(*pos)) {
12933 U8 tmpbuf[UTF8_MAXBYTES+1];
12936 s++; /* get past 'v' */
12941 /* this is atoi() that tolerates underscores */
12944 const char *end = pos;
12946 while (--end >= s) {
12948 const UV orev = rev;
12949 rev += (*end - '0') * mult;
12951 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12952 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12953 "Integer overflow in decimal number");
12957 if (rev > 0x7FFFFFFF)
12958 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12960 /* Append native character for the rev point */
12961 tmpend = uvchr_to_utf8(tmpbuf, rev);
12962 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12963 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12965 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12971 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12975 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12983 * c-indentation-style: bsd
12984 * c-basic-offset: 4
12985 * indent-tabs-mode: t
12988 * ex: set ts=8 sts=4 sw=4 noet: