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 /* The maximum number of characters preceding the unrecognized one to display */
128 #define UNRECOGNIZED_PRECEDE_COUNT 10
130 /* In variables named $^X, these are the legal values for X.
131 * 1999-02-27 mjd-perl-patch@plover.com */
132 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
134 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
136 /* LEX_* are values for PL_lex_state, the state of the lexer.
137 * They are arranged oddly so that the guard on the switch statement
138 * can get by with a single comparison (if the compiler is smart enough).
141 /* #define LEX_NOTPARSING 11 is done in perl.h. */
143 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
144 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
145 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
146 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
147 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
149 /* at end of code, eg "$x" followed by: */
150 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
151 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
153 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
154 string or after \E, $foo, etc */
155 #define LEX_INTERPCONST 2 /* NOT USED */
156 #define LEX_FORMLINE 1 /* expecting a format line */
157 #define LEX_KNOWNEXT 0 /* next token known; just return it */
161 static const char* const lex_state_names[] = {
180 #include "keywords.h"
182 /* CLINE is a macro that ensures PL_copline has a sane value */
187 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
190 # define SKIPSPACE0(s) skipspace0(s)
191 # define SKIPSPACE1(s) skipspace1(s)
192 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
193 # define PEEKSPACE(s) skipspace2(s,0)
195 # define SKIPSPACE0(s) skipspace(s)
196 # define SKIPSPACE1(s) skipspace(s)
197 # define SKIPSPACE2(s,tsv) skipspace(s)
198 # define PEEKSPACE(s) skipspace(s)
202 * Convenience functions to return different tokens and prime the
203 * lexer for the next token. They all take an argument.
205 * TOKEN : generic token (used for '(', DOLSHARP, etc)
206 * OPERATOR : generic operator
207 * AOPERATOR : assignment operator
208 * PREBLOCK : beginning the block after an if, while, foreach, ...
209 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
210 * PREREF : *EXPR where EXPR is not a simple identifier
211 * TERM : expression term
212 * LOOPX : loop exiting command (goto, last, dump, etc)
213 * FTST : file test operator
214 * FUN0 : zero-argument function
215 * FUN1 : not used, except for not, which isn't a UNIOP
216 * BOop : bitwise or or xor
218 * SHop : shift operator
219 * PWop : power operator
220 * PMop : pattern-matching operator
221 * Aop : addition-level operator
222 * Mop : multiplication-level operator
223 * Eop : equality-testing operator
224 * Rop : relational operator <= != gt
226 * Also see LOP and lop() below.
229 #ifdef DEBUGGING /* Serve -DT. */
230 # define REPORT(retval) tokereport((I32)retval, &pl_yylval)
232 # define REPORT(retval) (retval)
235 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
236 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
237 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
238 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
239 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
240 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
241 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
242 #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
243 #define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
244 #define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
245 #define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
246 #define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
247 #define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
248 #define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
249 #define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
250 #define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
251 #define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
252 #define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
253 #define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
254 #define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
256 /* This bit of chicanery makes a unary function followed by
257 * a parenthesis into a function with one argument, highest precedence.
258 * The UNIDOR macro is for unary functions that can be followed by the //
259 * operator (such as C<shift // 0>).
261 #define UNI2(f,x) { \
262 pl_yylval.ival = f; \
265 PL_last_uni = PL_oldbufptr; \
266 PL_last_lop_op = f; \
268 return REPORT( (int)FUNC1 ); \
270 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
272 #define UNI(f) UNI2(f,XTERM)
273 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
275 #define UNIBRACK(f) { \
276 pl_yylval.ival = f; \
278 PL_last_uni = PL_oldbufptr; \
280 return REPORT( (int)FUNC1 ); \
282 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
285 /* grandfather return to old style */
286 #define OLDLOP(f) return(pl_yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
290 /* how to interpret the pl_yylval associated with the token */
294 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
300 static struct debug_tokens {
302 enum token_type type;
304 } const debug_tokens[] =
306 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
307 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
308 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
309 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
310 { ARROW, TOKENTYPE_NONE, "ARROW" },
311 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
312 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
313 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
314 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
315 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
316 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
317 { DO, TOKENTYPE_NONE, "DO" },
318 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
319 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
320 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
321 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
322 { ELSE, TOKENTYPE_NONE, "ELSE" },
323 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
324 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
325 { FOR, TOKENTYPE_IVAL, "FOR" },
326 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
327 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
328 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
329 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
330 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
331 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
332 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
333 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
334 { IF, TOKENTYPE_IVAL, "IF" },
335 { LABEL, TOKENTYPE_PVAL, "LABEL" },
336 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
337 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
338 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
339 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
340 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
341 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
342 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
343 { MY, TOKENTYPE_IVAL, "MY" },
344 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
345 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
346 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
347 { OROP, TOKENTYPE_IVAL, "OROP" },
348 { OROR, TOKENTYPE_NONE, "OROR" },
349 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
350 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
351 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
352 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
353 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
354 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
355 { PREINC, TOKENTYPE_NONE, "PREINC" },
356 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
357 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
358 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
359 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
360 { SUB, TOKENTYPE_NONE, "SUB" },
361 { THING, TOKENTYPE_OPVAL, "THING" },
362 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
363 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
364 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
365 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
366 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
367 { USE, TOKENTYPE_IVAL, "USE" },
368 { WHEN, TOKENTYPE_IVAL, "WHEN" },
369 { WHILE, TOKENTYPE_IVAL, "WHILE" },
370 { WORD, TOKENTYPE_OPVAL, "WORD" },
371 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
372 { 0, TOKENTYPE_NONE, NULL }
375 /* dump the returned token in rv, plus any optional arg in pl_yylval */
378 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
382 PERL_ARGS_ASSERT_TOKEREPORT;
385 const char *name = NULL;
386 enum token_type type = TOKENTYPE_NONE;
387 const struct debug_tokens *p;
388 SV* const report = newSVpvs("<== ");
390 for (p = debug_tokens; p->token; p++) {
391 if (p->token == (int)rv) {
398 Perl_sv_catpv(aTHX_ report, name);
399 else if ((char)rv > ' ' && (char)rv < '~')
400 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
402 sv_catpvs(report, "EOF");
404 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
407 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
410 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
412 case TOKENTYPE_OPNUM:
413 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
414 PL_op_name[lvalp->ival]);
417 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
419 case TOKENTYPE_OPVAL:
421 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
422 PL_op_name[lvalp->opval->op_type]);
423 if (lvalp->opval->op_type == OP_CONST) {
424 Perl_sv_catpvf(aTHX_ report, " %s",
425 SvPEEK(cSVOPx_sv(lvalp->opval)));
430 sv_catpvs(report, "(opval=null)");
433 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
439 /* print the buffer with suitable escapes */
442 S_printbuf(pTHX_ const char *const fmt, const char *const s)
444 SV* const tmp = newSVpvs("");
446 PERL_ARGS_ASSERT_PRINTBUF;
448 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
457 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
458 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
462 S_ao(pTHX_ int toketype)
465 if (*PL_bufptr == '=') {
467 if (toketype == ANDAND)
468 pl_yylval.ival = OP_ANDASSIGN;
469 else if (toketype == OROR)
470 pl_yylval.ival = OP_ORASSIGN;
471 else if (toketype == DORDOR)
472 pl_yylval.ival = OP_DORASSIGN;
480 * When Perl expects an operator and finds something else, no_op
481 * prints the warning. It always prints "<something> found where
482 * operator expected. It prints "Missing semicolon on previous line?"
483 * if the surprise occurs at the start of the line. "do you need to
484 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
485 * where the compiler doesn't know if foo is a method call or a function.
486 * It prints "Missing operator before end of line" if there's nothing
487 * after the missing operator, or "... before <...>" if there is something
488 * after the missing operator.
492 S_no_op(pTHX_ const char *const what, char *s)
495 char * const oldbp = PL_bufptr;
496 const bool is_first = (PL_oldbufptr == PL_linestart);
498 PERL_ARGS_ASSERT_NO_OP;
504 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
505 if (ckWARN_d(WARN_SYNTAX)) {
507 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
508 "\t(Missing semicolon on previous line?)\n");
509 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
511 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
513 if (t < PL_bufptr && isSPACE(*t))
514 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
515 "\t(Do you need to predeclare %.*s?)\n",
516 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
520 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
521 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
529 * Complain about missing quote/regexp/heredoc terminator.
530 * If it's called with NULL then it cauterizes the line buffer.
531 * If we're in a delimited string and the delimiter is a control
532 * character, it's reformatted into a two-char sequence like ^C.
537 S_missingterm(pTHX_ char *s)
543 char * const nl = strrchr(s,'\n');
547 else if (isCNTRL(PL_multi_close)) {
549 tmpbuf[1] = (char)toCTRL(PL_multi_close);
554 *tmpbuf = (char)PL_multi_close;
558 q = strchr(s,'"') ? '\'' : '"';
559 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
562 #define FEATURE_IS_ENABLED(name) \
563 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
564 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
565 /* The longest string we pass in. */
566 #define MAX_FEATURE_LEN (sizeof("switch")-1)
569 * S_feature_is_enabled
570 * Check whether the named feature is enabled.
573 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
576 HV * const hinthv = GvHV(PL_hintgv);
577 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
579 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
581 assert(namelen <= MAX_FEATURE_LEN);
582 memcpy(&he_name[8], name, namelen);
584 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
592 Perl_deprecate(pTHX_ const char *const s)
594 PERL_ARGS_ASSERT_DEPRECATE;
596 if (ckWARN(WARN_DEPRECATED))
597 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
601 S_deprecate_old(pTHX_ const char *const s)
603 /* This function should NOT be called for any new deprecated warnings */
604 /* Use Perl_deprecate instead */
606 /* It is here to maintain backward compatibility with the pre-5.8 */
607 /* warnings category hierarchy. The "deprecated" category used to */
608 /* live under the "syntax" category. It is now a top-level category */
609 /* in its own right. */
611 PERL_ARGS_ASSERT_DEPRECATE_OLD;
613 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
614 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
615 "Use of %s is deprecated", s);
619 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
620 * utf16-to-utf8-reversed.
623 #ifdef PERL_CR_FILTER
627 register const char *s = SvPVX_const(sv);
628 register const char * const e = s + SvCUR(sv);
630 PERL_ARGS_ASSERT_STRIP_RETURN;
632 /* outer loop optimized to do nothing if there are no CR-LFs */
634 if (*s++ == '\r' && *s == '\n') {
635 /* hit a CR-LF, need to copy the rest */
636 register char *d = s - 1;
639 if (*s == '\r' && s[1] == '\n')
650 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
652 const I32 count = FILTER_READ(idx+1, sv, maxlen);
653 if (count > 0 && !maxlen)
664 * Create a parser object and initialise its parser and lexer fields
666 * rsfp is the opened file handle to read from (if any),
668 * line holds any initial content already read from the file (or in
669 * the case of no file, such as an eval, the whole contents);
671 * new_filter indicates that this is a new file and it shouldn't inherit
672 * the filters from the current parser (ie require).
676 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
679 const char *s = NULL;
681 yy_parser *parser, *oparser;
683 /* create and initialise a parser */
685 Newxz(parser, 1, yy_parser);
686 parser->old_parser = oparser = PL_parser;
689 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
690 parser->ps = parser->stack;
691 parser->stack_size = YYINITDEPTH;
693 parser->stack->state = 0;
694 parser->yyerrstatus = 0;
695 parser->yychar = YYEMPTY; /* Cause a token to be read. */
697 /* on scope exit, free this parser and restore any outer one */
699 parser->saved_curcop = PL_curcop;
701 /* initialise lexer state */
704 parser->curforce = -1;
706 parser->nexttoke = 0;
708 parser->error_count = oparser ? oparser->error_count : 0;
709 parser->copline = NOLINE;
710 parser->lex_state = LEX_NORMAL;
711 parser->expect = XSTATE;
713 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
714 : MUTABLE_AV(SvREFCNT_inc(oparser->rsfp_filters));
716 Newx(parser->lex_brackstack, 120, char);
717 Newx(parser->lex_casestack, 12, char);
718 *parser->lex_casestack = '\0';
721 s = SvPV_const(line, len);
727 parser->linestr = newSVpvs("\n;");
728 } else if (SvREADONLY(line) || s[len-1] != ';') {
729 parser->linestr = newSVsv(line);
731 sv_catpvs(parser->linestr, "\n;");
734 SvREFCNT_inc_simple_void_NN(line);
735 parser->linestr = line;
737 parser->oldoldbufptr =
740 parser->linestart = SvPVX(parser->linestr);
741 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
742 parser->last_lop = parser->last_uni = NULL;
746 /* delete a parser object */
749 Perl_parser_free(pTHX_ const yy_parser *parser)
751 PERL_ARGS_ASSERT_PARSER_FREE;
753 PL_curcop = parser->saved_curcop;
754 SvREFCNT_dec(parser->linestr);
756 if (parser->rsfp == PerlIO_stdin())
757 PerlIO_clearerr(parser->rsfp);
758 else if (parser->rsfp && (!parser->old_parser ||
759 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
760 PerlIO_close(parser->rsfp);
761 SvREFCNT_dec(parser->rsfp_filters);
763 Safefree(parser->stack);
764 Safefree(parser->lex_brackstack);
765 Safefree(parser->lex_casestack);
766 PL_parser = parser->old_parser;
773 * Finalizer for lexing operations. Must be called when the parser is
774 * done with the lexer.
781 PL_doextract = FALSE;
786 * This subroutine has nothing to do with tilting, whether at windmills
787 * or pinball tables. Its name is short for "increment line". It
788 * increments the current line number in CopLINE(PL_curcop) and checks
789 * to see whether the line starts with a comment of the form
790 * # line 500 "foo.pm"
791 * If so, it sets the current line number and file to the values in the comment.
795 S_incline(pTHX_ const char *s)
802 PERL_ARGS_ASSERT_INCLINE;
804 CopLINE_inc(PL_curcop);
807 while (SPACE_OR_TAB(*s))
809 if (strnEQ(s, "line", 4))
813 if (SPACE_OR_TAB(*s))
817 while (SPACE_OR_TAB(*s))
825 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
827 while (SPACE_OR_TAB(*s))
829 if (*s == '"' && (t = strchr(s+1, '"'))) {
839 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
841 if (*e != '\n' && *e != '\0')
842 return; /* false alarm */
845 const STRLEN len = t - s;
847 SV *const temp_sv = CopFILESV(PL_curcop);
853 tmplen = SvCUR(temp_sv);
859 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
860 /* must copy *{"::_<(eval N)[oldfilename:L]"}
861 * to *{"::_<newfilename"} */
862 /* However, the long form of evals is only turned on by the
863 debugger - usually they're "(eval %lu)" */
867 STRLEN tmplen2 = len;
868 if (tmplen + 2 <= sizeof smallbuf)
871 Newx(tmpbuf, tmplen + 2, char);
874 memcpy(tmpbuf + 2, cf, tmplen);
876 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
881 if (tmplen2 + 2 <= sizeof smallbuf)
884 Newx(tmpbuf2, tmplen2 + 2, char);
886 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
887 /* Either they malloc'd it, or we malloc'd it,
888 so no prefix is present in ours. */
893 memcpy(tmpbuf2 + 2, s, tmplen2);
896 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
898 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
899 /* adjust ${"::_<newfilename"} to store the new file name */
900 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
901 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
902 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
905 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
907 if (tmpbuf != smallbuf) Safefree(tmpbuf);
910 CopFILE_free(PL_curcop);
911 CopFILE_setn(PL_curcop, s, len);
913 CopLINE_set(PL_curcop, atoi(n)-1);
917 /* skip space before PL_thistoken */
920 S_skipspace0(pTHX_ register char *s)
922 PERL_ARGS_ASSERT_SKIPSPACE0;
929 PL_thiswhite = newSVpvs("");
930 sv_catsv(PL_thiswhite, PL_skipwhite);
931 sv_free(PL_skipwhite);
934 PL_realtokenstart = s - SvPVX(PL_linestr);
938 /* skip space after PL_thistoken */
941 S_skipspace1(pTHX_ register char *s)
943 const char *start = s;
944 I32 startoff = start - SvPVX(PL_linestr);
946 PERL_ARGS_ASSERT_SKIPSPACE1;
951 start = SvPVX(PL_linestr) + startoff;
952 if (!PL_thistoken && PL_realtokenstart >= 0) {
953 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
954 PL_thistoken = newSVpvn(tstart, start - tstart);
956 PL_realtokenstart = -1;
959 PL_nextwhite = newSVpvs("");
960 sv_catsv(PL_nextwhite, PL_skipwhite);
961 sv_free(PL_skipwhite);
968 S_skipspace2(pTHX_ register char *s, SV **svp)
971 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
972 const I32 startoff = s - SvPVX(PL_linestr);
974 PERL_ARGS_ASSERT_SKIPSPACE2;
977 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
978 if (!PL_madskills || !svp)
980 start = SvPVX(PL_linestr) + startoff;
981 if (!PL_thistoken && PL_realtokenstart >= 0) {
982 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
983 PL_thistoken = newSVpvn(tstart, start - tstart);
984 PL_realtokenstart = -1;
989 sv_setsv(*svp, PL_skipwhite);
990 sv_free(PL_skipwhite);
999 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1001 AV *av = CopFILEAVx(PL_curcop);
1003 SV * const sv = newSV_type(SVt_PVMG);
1005 sv_setsv(sv, orig_sv);
1007 sv_setpvn(sv, buf, len);
1010 av_store(av, (I32)CopLINE(PL_curcop), sv);
1016 * Called to gobble the appropriate amount and type of whitespace.
1017 * Skips comments as well.
1021 S_skipspace(pTHX_ register char *s)
1026 int startoff = s - SvPVX(PL_linestr);
1028 PERL_ARGS_ASSERT_SKIPSPACE;
1031 sv_free(PL_skipwhite);
1035 PERL_ARGS_ASSERT_SKIPSPACE;
1037 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1038 while (s < PL_bufend && SPACE_OR_TAB(*s))
1048 SSize_t oldprevlen, oldoldprevlen;
1049 SSize_t oldloplen = 0, oldunilen = 0;
1050 while (s < PL_bufend && isSPACE(*s)) {
1051 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1056 if (s < PL_bufend && *s == '#') {
1057 while (s < PL_bufend && *s != '\n')
1059 if (s < PL_bufend) {
1061 if (PL_in_eval && !PL_rsfp) {
1068 /* only continue to recharge the buffer if we're at the end
1069 * of the buffer, we're not reading from a source filter, and
1070 * we're in normal lexing mode
1072 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1073 PL_lex_state == LEX_FORMLINE)
1080 /* try to recharge the buffer */
1082 curoff = s - SvPVX(PL_linestr);
1085 if ((s = filter_gets(PL_linestr, PL_rsfp,
1086 (prevlen = SvCUR(PL_linestr)))) == NULL)
1089 if (PL_madskills && curoff != startoff) {
1091 PL_skipwhite = newSVpvs("");
1092 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1096 /* mustn't throw out old stuff yet if madpropping */
1097 SvCUR(PL_linestr) = curoff;
1098 s = SvPVX(PL_linestr) + curoff;
1100 if (curoff && s[-1] == '\n')
1104 /* end of file. Add on the -p or -n magic */
1105 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1108 sv_catpvs(PL_linestr,
1109 ";}continue{print or die qq(-p destination: $!\\n);}");
1111 sv_setpvs(PL_linestr,
1112 ";}continue{print or die qq(-p destination: $!\\n);}");
1114 PL_minus_n = PL_minus_p = 0;
1116 else if (PL_minus_n) {
1118 sv_catpvs(PL_linestr, ";}");
1120 sv_setpvs(PL_linestr, ";}");
1126 sv_catpvs(PL_linestr,";");
1128 sv_setpvs(PL_linestr,";");
1131 /* reset variables for next time we lex */
1132 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1138 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1139 PL_last_lop = PL_last_uni = NULL;
1141 /* Close the filehandle. Could be from
1142 * STDIN, or a regular file. If we were reading code from
1143 * STDIN (because the commandline held no -e or filename)
1144 * then we don't close it, we reset it so the code can
1145 * read from STDIN too.
1148 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1149 PerlIO_clearerr(PL_rsfp);
1151 (void)PerlIO_close(PL_rsfp);
1156 /* not at end of file, so we only read another line */
1157 /* make corresponding updates to old pointers, for yyerror() */
1158 oldprevlen = PL_oldbufptr - PL_bufend;
1159 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1161 oldunilen = PL_last_uni - PL_bufend;
1163 oldloplen = PL_last_lop - PL_bufend;
1164 PL_linestart = PL_bufptr = s + prevlen;
1165 PL_bufend = s + SvCUR(PL_linestr);
1167 PL_oldbufptr = s + oldprevlen;
1168 PL_oldoldbufptr = s + oldoldprevlen;
1170 PL_last_uni = s + oldunilen;
1172 PL_last_lop = s + oldloplen;
1175 /* debugger active and we're not compiling the debugger code,
1176 * so store the line into the debugger's array of lines
1178 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
1179 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1186 PL_skipwhite = newSVpvs("");
1187 curoff = s - SvPVX(PL_linestr);
1188 if (curoff - startoff)
1189 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1198 * Check the unary operators to ensure there's no ambiguity in how they're
1199 * used. An ambiguous piece of code would be:
1201 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1202 * the +5 is its argument.
1212 if (PL_oldoldbufptr != PL_last_uni)
1214 while (isSPACE(*PL_last_uni))
1217 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1219 if ((t = strchr(s, '(')) && t < PL_bufptr)
1222 if (ckWARN_d(WARN_AMBIGUOUS)){
1223 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1224 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1225 (int)(s - PL_last_uni), PL_last_uni);
1230 * LOP : macro to build a list operator. Its behaviour has been replaced
1231 * with a subroutine, S_lop() for which LOP is just another name.
1234 #define LOP(f,x) return lop(f,x,s)
1238 * Build a list operator (or something that might be one). The rules:
1239 * - if we have a next token, then it's a list operator [why?]
1240 * - if the next thing is an opening paren, then it's a function
1241 * - else it's a list operator
1245 S_lop(pTHX_ I32 f, int x, char *s)
1249 PERL_ARGS_ASSERT_LOP;
1255 PL_last_lop = PL_oldbufptr;
1256 PL_last_lop_op = (OPCODE)f;
1259 return REPORT(LSTOP);
1262 return REPORT(LSTOP);
1265 return REPORT(FUNC);
1268 return REPORT(FUNC);
1270 return REPORT(LSTOP);
1276 * Sets up for an eventual force_next(). start_force(0) basically does
1277 * an unshift, while start_force(-1) does a push. yylex removes items
1282 S_start_force(pTHX_ int where)
1286 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1287 where = PL_lasttoke;
1288 assert(PL_curforce < 0 || PL_curforce == where);
1289 if (PL_curforce != where) {
1290 for (i = PL_lasttoke; i > where; --i) {
1291 PL_nexttoke[i] = PL_nexttoke[i-1];
1295 if (PL_curforce < 0) /* in case of duplicate start_force() */
1296 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1297 PL_curforce = where;
1300 curmad('^', newSVpvs(""));
1301 CURMAD('_', PL_nextwhite);
1306 S_curmad(pTHX_ char slot, SV *sv)
1312 if (PL_curforce < 0)
1313 where = &PL_thismad;
1315 where = &PL_nexttoke[PL_curforce].next_mad;
1321 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1323 else if (PL_encoding) {
1324 sv_recode_to_utf8(sv, PL_encoding);
1329 /* keep a slot open for the head of the list? */
1330 if (slot != '_' && *where && (*where)->mad_key == '^') {
1331 (*where)->mad_key = slot;
1332 sv_free(MUTABLE_SV(((*where)->mad_val)));
1333 (*where)->mad_val = (void*)sv;
1336 addmad(newMADsv(slot, sv), where, 0);
1339 # define start_force(where) NOOP
1340 # define curmad(slot, sv) NOOP
1345 * When the lexer realizes it knows the next token (for instance,
1346 * it is reordering tokens for the parser) then it can call S_force_next
1347 * to know what token to return the next time the lexer is called. Caller
1348 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1349 * and possibly PL_expect to ensure the lexer handles the token correctly.
1353 S_force_next(pTHX_ I32 type)
1358 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1359 tokereport(type, &NEXTVAL_NEXTTOKE);
1363 if (PL_curforce < 0)
1364 start_force(PL_lasttoke);
1365 PL_nexttoke[PL_curforce].next_type = type;
1366 if (PL_lex_state != LEX_KNOWNEXT)
1367 PL_lex_defer = PL_lex_state;
1368 PL_lex_state = LEX_KNOWNEXT;
1369 PL_lex_expect = PL_expect;
1372 PL_nexttype[PL_nexttoke] = type;
1374 if (PL_lex_state != LEX_KNOWNEXT) {
1375 PL_lex_defer = PL_lex_state;
1376 PL_lex_expect = PL_expect;
1377 PL_lex_state = LEX_KNOWNEXT;
1383 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1386 SV * const sv = newSVpvn_utf8(start, len,
1389 && !is_ascii_string((const U8*)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 S_newSV_maybe_utf8(aTHX_ 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))
3661 unsigned char c = *s;
3662 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3663 if (len > UNRECOGNIZED_PRECEDE_COUNT) {
3664 d = UTF ? (char *) Perl_utf8_hop(aTHX_ (U8 *) s, -UNRECOGNIZED_PRECEDE_COUNT) : s - UNRECOGNIZED_PRECEDE_COUNT;
3669 Perl_croak(aTHX_ "Unrecognized character \\x%02X; marked by <-- HERE after %s<-- HERE near column %d", c, d, (int) len + 1);
3673 goto fake_eof; /* emulate EOF on ^D or ^Z */
3682 if (PL_lex_brackets) {
3683 yyerror((const char *)
3685 ? "Format not terminated"
3686 : "Missing right curly or square bracket"));
3688 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3689 "### Tokener got EOF\n");
3693 if (s++ < PL_bufend)
3694 goto retry; /* ignore stray nulls */
3697 if (!PL_in_eval && !PL_preambled) {
3698 PL_preambled = TRUE;
3704 /* Generate a string of Perl code to load the debugger.
3705 * If PERL5DB is set, it will return the contents of that,
3706 * otherwise a compile-time require of perl5db.pl. */
3708 const char * const pdb = PerlEnv_getenv("PERL5DB");
3711 sv_setpv(PL_linestr, pdb);
3712 sv_catpvs(PL_linestr,";");
3714 SETERRNO(0,SS_NORMAL);
3715 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3718 sv_setpvs(PL_linestr,"");
3719 if (PL_preambleav) {
3720 SV **svp = AvARRAY(PL_preambleav);
3721 SV **const end = svp + AvFILLp(PL_preambleav);
3723 sv_catsv(PL_linestr, *svp);
3725 sv_catpvs(PL_linestr, ";");
3727 sv_free(MUTABLE_SV(PL_preambleav));
3728 PL_preambleav = NULL;
3731 sv_catpvs(PL_linestr,
3732 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3733 if (PL_minus_n || PL_minus_p) {
3734 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3736 sv_catpvs(PL_linestr,"chomp;");
3739 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3740 || *PL_splitstr == '"')
3741 && strchr(PL_splitstr + 1, *PL_splitstr))
3742 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3744 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3745 bytes can be used as quoting characters. :-) */
3746 const char *splits = PL_splitstr;
3747 sv_catpvs(PL_linestr, "our @F=split(q\0");
3750 if (*splits == '\\')
3751 sv_catpvn(PL_linestr, splits, 1);
3752 sv_catpvn(PL_linestr, splits, 1);
3753 } while (*splits++);
3754 /* This loop will embed the trailing NUL of
3755 PL_linestr as the last thing it does before
3757 sv_catpvs(PL_linestr, ");");
3761 sv_catpvs(PL_linestr,"our @F=split(' ');");
3764 sv_catpvs(PL_linestr, "\n");
3765 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3766 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3767 PL_last_lop = PL_last_uni = NULL;
3768 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3769 update_debugger_info(PL_linestr, NULL, 0);
3773 bof = PL_rsfp ? TRUE : FALSE;
3774 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3777 PL_realtokenstart = -1;
3780 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3781 PerlIO_clearerr(PL_rsfp);
3783 (void)PerlIO_close(PL_rsfp);
3785 PL_doextract = FALSE;
3787 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3793 sv_setpvs(PL_linestr, ";}continue{print;}");
3795 sv_setpvs(PL_linestr, ";}");
3796 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3797 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3798 PL_last_lop = PL_last_uni = NULL;
3799 PL_minus_n = PL_minus_p = 0;
3802 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3803 PL_last_lop = PL_last_uni = NULL;
3804 sv_setpvs(PL_linestr,"");
3805 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3807 /* If it looks like the start of a BOM or raw UTF-16,
3808 * check if it in fact is. */
3814 #ifdef PERLIO_IS_STDIO
3815 # ifdef __GNU_LIBRARY__
3816 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3817 # define FTELL_FOR_PIPE_IS_BROKEN
3821 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3822 # define FTELL_FOR_PIPE_IS_BROKEN
3827 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3829 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3830 s = swallow_bom((U8*)s);
3834 /* Incest with pod. */
3837 sv_catsv(PL_thiswhite, PL_linestr);
3839 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3840 sv_setpvs(PL_linestr, "");
3841 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3842 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3843 PL_last_lop = PL_last_uni = NULL;
3844 PL_doextract = FALSE;
3848 } while (PL_doextract);
3849 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3850 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3851 update_debugger_info(PL_linestr, NULL, 0);
3852 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3853 PL_last_lop = PL_last_uni = NULL;
3854 if (CopLINE(PL_curcop) == 1) {
3855 while (s < PL_bufend && isSPACE(*s))
3857 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3861 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3865 if (*s == '#' && *(s+1) == '!')
3867 #ifdef ALTERNATE_SHEBANG
3869 static char const as[] = ALTERNATE_SHEBANG;
3870 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3871 d = s + (sizeof(as) - 1);
3873 #endif /* ALTERNATE_SHEBANG */
3882 while (*d && !isSPACE(*d))
3886 #ifdef ARG_ZERO_IS_SCRIPT
3887 if (ipathend > ipath) {
3889 * HP-UX (at least) sets argv[0] to the script name,
3890 * which makes $^X incorrect. And Digital UNIX and Linux,
3891 * at least, set argv[0] to the basename of the Perl
3892 * interpreter. So, having found "#!", we'll set it right.
3894 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3896 assert(SvPOK(x) || SvGMAGICAL(x));
3897 if (sv_eq(x, CopFILESV(PL_curcop))) {
3898 sv_setpvn(x, ipath, ipathend - ipath);
3904 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3905 const char * const lstart = SvPV_const(x,llen);
3907 bstart += blen - llen;
3908 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3909 sv_setpvn(x, ipath, ipathend - ipath);
3914 TAINT_NOT; /* $^X is always tainted, but that's OK */
3916 #endif /* ARG_ZERO_IS_SCRIPT */
3921 d = instr(s,"perl -");
3923 d = instr(s,"perl");
3925 /* avoid getting into infinite loops when shebang
3926 * line contains "Perl" rather than "perl" */
3928 for (d = ipathend-4; d >= ipath; --d) {
3929 if ((*d == 'p' || *d == 'P')
3930 && !ibcmp(d, "perl", 4))
3940 #ifdef ALTERNATE_SHEBANG
3942 * If the ALTERNATE_SHEBANG on this system starts with a
3943 * character that can be part of a Perl expression, then if
3944 * we see it but not "perl", we're probably looking at the
3945 * start of Perl code, not a request to hand off to some
3946 * other interpreter. Similarly, if "perl" is there, but
3947 * not in the first 'word' of the line, we assume the line
3948 * contains the start of the Perl program.
3950 if (d && *s != '#') {
3951 const char *c = ipath;
3952 while (*c && !strchr("; \t\r\n\f\v#", *c))
3955 d = NULL; /* "perl" not in first word; ignore */
3957 *s = '#'; /* Don't try to parse shebang line */
3959 #endif /* ALTERNATE_SHEBANG */
3964 !instr(s,"indir") &&
3965 instr(PL_origargv[0],"perl"))
3972 while (s < PL_bufend && isSPACE(*s))
3974 if (s < PL_bufend) {
3975 Newx(newargv,PL_origargc+3,char*);
3977 while (s < PL_bufend && !isSPACE(*s))
3980 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3983 newargv = PL_origargv;
3986 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3988 Perl_croak(aTHX_ "Can't exec %s", ipath);
3991 while (*d && !isSPACE(*d))
3993 while (SPACE_OR_TAB(*d))
3997 const bool switches_done = PL_doswitches;
3998 const U32 oldpdb = PL_perldb;
3999 const bool oldn = PL_minus_n;
4000 const bool oldp = PL_minus_p;
4004 bool baduni = FALSE;
4006 const char *d2 = d1 + 1;
4007 if (parse_unicode_opts((const char **)&d2)
4011 if (baduni || *d1 == 'M' || *d1 == 'm') {
4012 const char * const m = d1;
4013 while (*d1 && !isSPACE(*d1))
4015 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
4018 d1 = moreswitches(d1);
4020 if (PL_doswitches && !switches_done) {
4021 int argc = PL_origargc;
4022 char **argv = PL_origargv;
4025 } while (argc && argv[0][0] == '-' && argv[0][1]);
4026 init_argv_symbols(argc,argv);
4028 if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
4029 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
4030 /* if we have already added "LINE: while (<>) {",
4031 we must not do it again */
4033 sv_setpvs(PL_linestr, "");
4034 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4035 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4036 PL_last_lop = PL_last_uni = NULL;
4037 PL_preambled = FALSE;
4038 if (PERLDB_LINE || PERLDB_SAVESRC)
4039 (void)gv_fetchfile(PL_origfilename);
4046 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4048 PL_lex_state = LEX_FORMLINE;
4053 #ifdef PERL_STRICT_CR
4054 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4056 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4058 case ' ': case '\t': case '\f': case 013:
4060 PL_realtokenstart = -1;
4062 PL_thiswhite = newSVpvs("");
4063 sv_catpvn(PL_thiswhite, s, 1);
4070 PL_realtokenstart = -1;
4074 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4075 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4076 /* handle eval qq[#line 1 "foo"\n ...] */
4077 CopLINE_dec(PL_curcop);
4080 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4082 if (!PL_in_eval || PL_rsfp)
4087 while (d < PL_bufend && *d != '\n')
4091 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4092 Perl_croak(aTHX_ "panic: input overflow");
4095 PL_thiswhite = newSVpvn(s, d - s);
4100 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4102 PL_lex_state = LEX_FORMLINE;
4108 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4109 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4112 TOKEN(PEG); /* make sure any #! line is accessible */
4117 /* if (PL_madskills && PL_lex_formbrack) { */
4119 while (d < PL_bufend && *d != '\n')
4123 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4124 Perl_croak(aTHX_ "panic: input overflow");
4125 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4127 PL_thiswhite = newSVpvs("");
4128 if (CopLINE(PL_curcop) == 1) {
4129 sv_setpvs(PL_thiswhite, "");
4132 sv_catpvn(PL_thiswhite, s, d - s);
4146 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4154 while (s < PL_bufend && SPACE_OR_TAB(*s))
4157 if (strnEQ(s,"=>",2)) {
4158 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4159 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4160 OPERATOR('-'); /* unary minus */
4162 PL_last_uni = PL_oldbufptr;
4164 case 'r': ftst = OP_FTEREAD; break;
4165 case 'w': ftst = OP_FTEWRITE; break;
4166 case 'x': ftst = OP_FTEEXEC; break;
4167 case 'o': ftst = OP_FTEOWNED; break;
4168 case 'R': ftst = OP_FTRREAD; break;
4169 case 'W': ftst = OP_FTRWRITE; break;
4170 case 'X': ftst = OP_FTREXEC; break;
4171 case 'O': ftst = OP_FTROWNED; break;
4172 case 'e': ftst = OP_FTIS; break;
4173 case 'z': ftst = OP_FTZERO; break;
4174 case 's': ftst = OP_FTSIZE; break;
4175 case 'f': ftst = OP_FTFILE; break;
4176 case 'd': ftst = OP_FTDIR; break;
4177 case 'l': ftst = OP_FTLINK; break;
4178 case 'p': ftst = OP_FTPIPE; break;
4179 case 'S': ftst = OP_FTSOCK; break;
4180 case 'u': ftst = OP_FTSUID; break;
4181 case 'g': ftst = OP_FTSGID; break;
4182 case 'k': ftst = OP_FTSVTX; break;
4183 case 'b': ftst = OP_FTBLK; break;
4184 case 'c': ftst = OP_FTCHR; break;
4185 case 't': ftst = OP_FTTTY; break;
4186 case 'T': ftst = OP_FTTEXT; break;
4187 case 'B': ftst = OP_FTBINARY; break;
4188 case 'M': case 'A': case 'C':
4189 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4191 case 'M': ftst = OP_FTMTIME; break;
4192 case 'A': ftst = OP_FTATIME; break;
4193 case 'C': ftst = OP_FTCTIME; break;
4201 PL_last_lop_op = (OPCODE)ftst;
4202 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4203 "### Saw file test %c\n", (int)tmp);
4208 /* Assume it was a minus followed by a one-letter named
4209 * subroutine call (or a -bareword), then. */
4210 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4211 "### '-%c' looked like a file test but was not\n",
4218 const char tmp = *s++;
4221 if (PL_expect == XOPERATOR)
4226 else if (*s == '>') {
4229 if (isIDFIRST_lazy_if(s,UTF)) {
4230 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4238 if (PL_expect == XOPERATOR)
4241 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4243 OPERATOR('-'); /* unary minus */
4249 const char tmp = *s++;
4252 if (PL_expect == XOPERATOR)
4257 if (PL_expect == XOPERATOR)
4260 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4267 if (PL_expect != XOPERATOR) {
4268 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4269 PL_expect = XOPERATOR;
4270 force_ident(PL_tokenbuf, '*');
4283 if (PL_expect == XOPERATOR) {
4287 PL_tokenbuf[0] = '%';
4288 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4289 sizeof PL_tokenbuf - 1, FALSE);
4290 if (!PL_tokenbuf[1]) {
4293 PL_pending_ident = '%';
4302 const char tmp = *s++;
4307 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4314 const char tmp = *s++;
4320 goto just_a_word_zero_gv;
4323 switch (PL_expect) {
4329 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4331 PL_bufptr = s; /* update in case we back off */
4337 PL_expect = XTERMBLOCK;
4340 stuffstart = s - SvPVX(PL_linestr) - 1;
4344 while (isIDFIRST_lazy_if(s,UTF)) {
4347 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4348 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4349 if (tmp < 0) tmp = -tmp;
4364 sv = newSVpvn(s, len);
4366 d = scan_str(d,TRUE,TRUE);
4368 /* MUST advance bufptr here to avoid bogus
4369 "at end of line" context messages from yyerror().
4371 PL_bufptr = s + len;
4372 yyerror("Unterminated attribute parameter in attribute list");
4376 return REPORT(0); /* EOF indicator */
4380 sv_catsv(sv, PL_lex_stuff);
4381 attrs = append_elem(OP_LIST, attrs,
4382 newSVOP(OP_CONST, 0, sv));
4383 SvREFCNT_dec(PL_lex_stuff);
4384 PL_lex_stuff = NULL;
4387 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4389 if (PL_in_my == KEY_our) {
4390 deprecate(":unique");
4393 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4396 /* NOTE: any CV attrs applied here need to be part of
4397 the CVf_BUILTIN_ATTRS define in cv.h! */
4398 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4400 CvLVALUE_on(PL_compcv);
4402 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4404 deprecate(":locked");
4406 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4408 CvMETHOD_on(PL_compcv);
4410 /* After we've set the flags, it could be argued that
4411 we don't need to do the attributes.pm-based setting
4412 process, and shouldn't bother appending recognized
4413 flags. To experiment with that, uncomment the
4414 following "else". (Note that's already been
4415 uncommented. That keeps the above-applied built-in
4416 attributes from being intercepted (and possibly
4417 rejected) by a package's attribute routines, but is
4418 justified by the performance win for the common case
4419 of applying only built-in attributes.) */
4421 attrs = append_elem(OP_LIST, attrs,
4422 newSVOP(OP_CONST, 0,
4426 if (*s == ':' && s[1] != ':')
4429 break; /* require real whitespace or :'s */
4430 /* XXX losing whitespace on sequential attributes here */
4434 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4435 if (*s != ';' && *s != '}' && *s != tmp
4436 && (tmp != '=' || *s != ')')) {
4437 const char q = ((*s == '\'') ? '"' : '\'');
4438 /* If here for an expression, and parsed no attrs, back
4440 if (tmp == '=' && !attrs) {
4444 /* MUST advance bufptr here to avoid bogus "at end of line"
4445 context messages from yyerror().
4448 yyerror( (const char *)
4450 ? Perl_form(aTHX_ "Invalid separator character "
4451 "%c%c%c in attribute list", q, *s, q)
4452 : "Unterminated attribute list" ) );
4460 start_force(PL_curforce);
4461 NEXTVAL_NEXTTOKE.opval = attrs;
4462 CURMAD('_', PL_nextwhite);
4467 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4468 (s - SvPVX(PL_linestr)) - stuffstart);
4476 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4477 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4485 const char tmp = *s++;
4490 const char tmp = *s++;
4498 if (PL_lex_brackets <= 0)
4499 yyerror("Unmatched right square bracket");
4502 if (PL_lex_state == LEX_INTERPNORMAL) {
4503 if (PL_lex_brackets == 0) {
4504 if (*s == '-' && s[1] == '>')
4505 PL_lex_state = LEX_INTERPENDMAYBE;
4506 else if (*s != '[' && *s != '{')
4507 PL_lex_state = LEX_INTERPEND;
4514 if (PL_lex_brackets > 100) {
4515 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4517 switch (PL_expect) {
4519 if (PL_lex_formbrack) {
4523 if (PL_oldoldbufptr == PL_last_lop)
4524 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4526 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4527 OPERATOR(HASHBRACK);
4529 while (s < PL_bufend && SPACE_OR_TAB(*s))
4532 PL_tokenbuf[0] = '\0';
4533 if (d < PL_bufend && *d == '-') {
4534 PL_tokenbuf[0] = '-';
4536 while (d < PL_bufend && SPACE_OR_TAB(*d))
4539 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4540 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4542 while (d < PL_bufend && SPACE_OR_TAB(*d))
4545 const char minus = (PL_tokenbuf[0] == '-');
4546 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4554 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4559 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4564 if (PL_oldoldbufptr == PL_last_lop)
4565 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4567 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4570 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4572 /* This hack is to get the ${} in the message. */
4574 yyerror("syntax error");
4577 OPERATOR(HASHBRACK);
4579 /* This hack serves to disambiguate a pair of curlies
4580 * as being a block or an anon hash. Normally, expectation
4581 * determines that, but in cases where we're not in a
4582 * position to expect anything in particular (like inside
4583 * eval"") we have to resolve the ambiguity. This code
4584 * covers the case where the first term in the curlies is a
4585 * quoted string. Most other cases need to be explicitly
4586 * disambiguated by prepending a "+" before the opening
4587 * curly in order to force resolution as an anon hash.
4589 * XXX should probably propagate the outer expectation
4590 * into eval"" to rely less on this hack, but that could
4591 * potentially break current behavior of eval"".
4595 if (*s == '\'' || *s == '"' || *s == '`') {
4596 /* common case: get past first string, handling escapes */
4597 for (t++; t < PL_bufend && *t != *s;)
4598 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4602 else if (*s == 'q') {
4605 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4608 /* skip q//-like construct */
4610 char open, close, term;
4613 while (t < PL_bufend && isSPACE(*t))
4615 /* check for q => */
4616 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4617 OPERATOR(HASHBRACK);
4621 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4625 for (t++; t < PL_bufend; t++) {
4626 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4628 else if (*t == open)
4632 for (t++; t < PL_bufend; t++) {
4633 if (*t == '\\' && t+1 < PL_bufend)
4635 else if (*t == close && --brackets <= 0)
4637 else if (*t == open)
4644 /* skip plain q word */
4645 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4648 else if (isALNUM_lazy_if(t,UTF)) {
4650 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4653 while (t < PL_bufend && isSPACE(*t))
4655 /* if comma follows first term, call it an anon hash */
4656 /* XXX it could be a comma expression with loop modifiers */
4657 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4658 || (*t == '=' && t[1] == '>')))
4659 OPERATOR(HASHBRACK);
4660 if (PL_expect == XREF)
4663 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4669 pl_yylval.ival = CopLINE(PL_curcop);
4670 if (isSPACE(*s) || *s == '#')
4671 PL_copline = NOLINE; /* invalidate current command line number */
4676 if (PL_lex_brackets <= 0)
4677 yyerror("Unmatched right curly bracket");
4679 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4680 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4681 PL_lex_formbrack = 0;
4682 if (PL_lex_state == LEX_INTERPNORMAL) {
4683 if (PL_lex_brackets == 0) {
4684 if (PL_expect & XFAKEBRACK) {
4685 PL_expect &= XENUMMASK;
4686 PL_lex_state = LEX_INTERPEND;
4691 PL_thiswhite = newSVpvs("");
4692 sv_catpvs(PL_thiswhite,"}");
4695 return yylex(); /* ignore fake brackets */
4697 if (*s == '-' && s[1] == '>')
4698 PL_lex_state = LEX_INTERPENDMAYBE;
4699 else if (*s != '[' && *s != '{')
4700 PL_lex_state = LEX_INTERPEND;
4703 if (PL_expect & XFAKEBRACK) {
4704 PL_expect &= XENUMMASK;
4706 return yylex(); /* ignore fake brackets */
4708 start_force(PL_curforce);
4710 curmad('X', newSVpvn(s-1,1));
4711 CURMAD('_', PL_thiswhite);
4716 PL_thistoken = newSVpvs("");
4724 if (PL_expect == XOPERATOR) {
4725 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4726 && isIDFIRST_lazy_if(s,UTF))
4728 CopLINE_dec(PL_curcop);
4729 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
4730 CopLINE_inc(PL_curcop);
4735 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4737 PL_expect = XOPERATOR;
4738 force_ident(PL_tokenbuf, '&');
4742 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4754 const char tmp = *s++;
4761 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4762 && strchr("+-*/%.^&|<",tmp))
4763 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4764 "Reversed %c= operator",(int)tmp);
4766 if (PL_expect == XSTATE && isALPHA(tmp) &&
4767 (s == PL_linestart+1 || s[-2] == '\n') )
4769 if (PL_in_eval && !PL_rsfp) {
4774 if (strnEQ(s,"=cut",4)) {
4790 PL_thiswhite = newSVpvs("");
4791 sv_catpvn(PL_thiswhite, PL_linestart,
4792 PL_bufend - PL_linestart);
4796 PL_doextract = TRUE;
4800 if (PL_lex_brackets < PL_lex_formbrack) {
4802 #ifdef PERL_STRICT_CR
4803 while (SPACE_OR_TAB(*t))
4805 while (SPACE_OR_TAB(*t) || *t == '\r')
4808 if (*t == '\n' || *t == '#') {
4819 const char tmp = *s++;
4821 /* was this !=~ where !~ was meant?
4822 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4824 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4825 const char *t = s+1;
4827 while (t < PL_bufend && isSPACE(*t))
4830 if (*t == '/' || *t == '?' ||
4831 ((*t == 'm' || *t == 's' || *t == 'y')
4832 && !isALNUM(t[1])) ||
4833 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4834 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4835 "!=~ should be !~");
4845 if (PL_expect != XOPERATOR) {
4846 if (s[1] != '<' && !strchr(s,'>'))
4849 s = scan_heredoc(s);
4851 s = scan_inputsymbol(s);
4852 TERM(sublex_start());
4858 SHop(OP_LEFT_SHIFT);
4872 const char tmp = *s++;
4874 SHop(OP_RIGHT_SHIFT);
4875 else if (tmp == '=')
4884 if (PL_expect == XOPERATOR) {
4885 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4887 deprecate_old(commaless_variable_list);
4888 return REPORT(','); /* grandfather non-comma-format format */
4892 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4893 PL_tokenbuf[0] = '@';
4894 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4895 sizeof PL_tokenbuf - 1, FALSE);
4896 if (PL_expect == XOPERATOR)
4897 no_op("Array length", s);
4898 if (!PL_tokenbuf[1])
4900 PL_expect = XOPERATOR;
4901 PL_pending_ident = '#';
4905 PL_tokenbuf[0] = '$';
4906 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4907 sizeof PL_tokenbuf - 1, FALSE);
4908 if (PL_expect == XOPERATOR)
4910 if (!PL_tokenbuf[1]) {
4912 yyerror("Final $ should be \\$ or $name");
4916 /* This kludge not intended to be bulletproof. */
4917 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4918 pl_yylval.opval = newSVOP(OP_CONST, 0,
4919 newSViv(CopARYBASE_get(&PL_compiling)));
4920 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4926 const char tmp = *s;
4927 if (PL_lex_state == LEX_NORMAL)
4930 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4931 && intuit_more(s)) {
4933 PL_tokenbuf[0] = '@';
4934 if (ckWARN(WARN_SYNTAX)) {
4937 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4940 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4941 while (t < PL_bufend && *t != ']')
4943 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4944 "Multidimensional syntax %.*s not supported",
4945 (int)((t - PL_bufptr) + 1), PL_bufptr);
4949 else if (*s == '{') {
4951 PL_tokenbuf[0] = '%';
4952 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4953 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4955 char tmpbuf[sizeof PL_tokenbuf];
4958 } while (isSPACE(*t));
4959 if (isIDFIRST_lazy_if(t,UTF)) {
4961 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4965 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4966 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4967 "You need to quote \"%s\"",
4974 PL_expect = XOPERATOR;
4975 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4976 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4977 if (!islop || PL_last_lop_op == OP_GREPSTART)
4978 PL_expect = XOPERATOR;
4979 else if (strchr("$@\"'`q", *s))
4980 PL_expect = XTERM; /* e.g. print $fh "foo" */
4981 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4982 PL_expect = XTERM; /* e.g. print $fh &sub */
4983 else if (isIDFIRST_lazy_if(s,UTF)) {
4984 char tmpbuf[sizeof PL_tokenbuf];
4986 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4987 if ((t2 = keyword(tmpbuf, len, 0))) {
4988 /* binary operators exclude handle interpretations */
5000 PL_expect = XTERM; /* e.g. print $fh length() */
5005 PL_expect = XTERM; /* e.g. print $fh subr() */
5008 else if (isDIGIT(*s))
5009 PL_expect = XTERM; /* e.g. print $fh 3 */
5010 else if (*s == '.' && isDIGIT(s[1]))
5011 PL_expect = XTERM; /* e.g. print $fh .3 */
5012 else if ((*s == '?' || *s == '-' || *s == '+')
5013 && !isSPACE(s[1]) && s[1] != '=')
5014 PL_expect = XTERM; /* e.g. print $fh -1 */
5015 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
5017 PL_expect = XTERM; /* e.g. print $fh /.../
5018 XXX except DORDOR operator
5020 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
5022 PL_expect = XTERM; /* print $fh <<"EOF" */
5025 PL_pending_ident = '$';
5029 if (PL_expect == XOPERATOR)
5031 PL_tokenbuf[0] = '@';
5032 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
5033 if (!PL_tokenbuf[1]) {
5036 if (PL_lex_state == LEX_NORMAL)
5038 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5040 PL_tokenbuf[0] = '%';
5042 /* Warn about @ where they meant $. */
5043 if (*s == '[' || *s == '{') {
5044 if (ckWARN(WARN_SYNTAX)) {
5045 const char *t = s + 1;
5046 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5048 if (*t == '}' || *t == ']') {
5050 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5051 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5052 "Scalar value %.*s better written as $%.*s",
5053 (int)(t-PL_bufptr), PL_bufptr,
5054 (int)(t-PL_bufptr-1), PL_bufptr+1);
5059 PL_pending_ident = '@';
5062 case '/': /* may be division, defined-or, or pattern */
5063 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5067 case '?': /* may either be conditional or pattern */
5068 if (PL_expect == XOPERATOR) {
5076 /* A // operator. */
5086 /* Disable warning on "study /blah/" */
5087 if (PL_oldoldbufptr == PL_last_uni
5088 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5089 || memNE(PL_last_uni, "study", 5)
5090 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5093 s = scan_pat(s,OP_MATCH);
5094 TERM(sublex_start());
5098 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5099 #ifdef PERL_STRICT_CR
5102 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5104 && (s == PL_linestart || s[-1] == '\n') )
5106 PL_lex_formbrack = 0;
5110 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5114 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5120 pl_yylval.ival = OPf_SPECIAL;
5126 if (PL_expect != XOPERATOR)
5131 case '0': case '1': case '2': case '3': case '4':
5132 case '5': case '6': case '7': case '8': case '9':
5133 s = scan_num(s, &pl_yylval);
5134 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5135 if (PL_expect == XOPERATOR)
5140 s = scan_str(s,!!PL_madskills,FALSE);
5141 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5142 if (PL_expect == XOPERATOR) {
5143 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5145 deprecate_old(commaless_variable_list);
5146 return REPORT(','); /* grandfather non-comma-format format */
5153 pl_yylval.ival = OP_CONST;
5154 TERM(sublex_start());
5157 s = scan_str(s,!!PL_madskills,FALSE);
5158 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5159 if (PL_expect == XOPERATOR) {
5160 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5162 deprecate_old(commaless_variable_list);
5163 return REPORT(','); /* grandfather non-comma-format format */
5170 pl_yylval.ival = OP_CONST;
5171 /* FIXME. I think that this can be const if char *d is replaced by
5172 more localised variables. */
5173 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5174 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5175 pl_yylval.ival = OP_STRINGIFY;
5179 TERM(sublex_start());
5182 s = scan_str(s,!!PL_madskills,FALSE);
5183 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5184 if (PL_expect == XOPERATOR)
5185 no_op("Backticks",s);
5188 readpipe_override();
5189 TERM(sublex_start());
5193 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5194 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5196 if (PL_expect == XOPERATOR)
5197 no_op("Backslash",s);
5201 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5202 char *start = s + 2;
5203 while (isDIGIT(*start) || *start == '_')
5205 if (*start == '.' && isDIGIT(start[1])) {
5206 s = scan_num(s, &pl_yylval);
5209 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5210 else if (!isALPHA(*start) && (PL_expect == XTERM
5211 || PL_expect == XREF || PL_expect == XSTATE
5212 || PL_expect == XTERMORDORDOR)) {
5213 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5215 s = scan_num(s, &pl_yylval);
5222 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5264 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5266 /* Some keywords can be followed by any delimiter, including ':' */
5267 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5268 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5269 (PL_tokenbuf[0] == 'q' &&
5270 strchr("qwxr", PL_tokenbuf[1])))));
5272 /* x::* is just a word, unless x is "CORE" */
5273 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5277 while (d < PL_bufend && isSPACE(*d))
5278 d++; /* no comments skipped here, or s### is misparsed */
5280 /* Is this a label? */
5281 if (!tmp && PL_expect == XSTATE
5282 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5283 tmp = keyword(PL_tokenbuf, len, 0);
5285 Perl_croak(aTHX_ "Can't use keyword '%s' as a label", PL_tokenbuf);
5287 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5292 /* Check for keywords */
5293 tmp = keyword(PL_tokenbuf, len, 0);
5295 /* Is this a word before a => operator? */
5296 if (*d == '=' && d[1] == '>') {
5299 = (OP*)newSVOP(OP_CONST, 0,
5300 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5301 pl_yylval.opval->op_private = OPpCONST_BARE;
5305 if (tmp < 0) { /* second-class keyword? */
5306 GV *ogv = NULL; /* override (winner) */
5307 GV *hgv = NULL; /* hidden (loser) */
5308 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5310 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5313 if (GvIMPORTED_CV(gv))
5315 else if (! CvMETHOD(cv))
5319 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5320 (gv = *gvp) && isGV_with_GP(gv) &&
5321 GvCVu(gv) && GvIMPORTED_CV(gv))
5328 tmp = 0; /* overridden by import or by GLOBAL */
5331 && -tmp==KEY_lock /* XXX generalizable kludge */
5334 tmp = 0; /* any sub overrides "weak" keyword */
5336 else { /* no override */
5338 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5339 Perl_warner(aTHX_ packWARN(WARN_MISC),
5340 "dump() better written as CORE::dump()");
5344 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5345 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5346 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5347 "Ambiguous call resolved as CORE::%s(), %s",
5348 GvENAME(hgv), "qualify as such or use &");
5355 default: /* not a keyword */
5356 /* Trade off - by using this evil construction we can pull the
5357 variable gv into the block labelled keylookup. If not, then
5358 we have to give it function scope so that the goto from the
5359 earlier ':' case doesn't bypass the initialisation. */
5361 just_a_word_zero_gv:
5369 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5372 SV *nextPL_nextwhite = 0;
5376 /* Get the rest if it looks like a package qualifier */
5378 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5380 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5383 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5384 *s == '\'' ? "'" : "::");
5389 if (PL_expect == XOPERATOR) {
5390 if (PL_bufptr == PL_linestart) {
5391 CopLINE_dec(PL_curcop);
5392 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5393 CopLINE_inc(PL_curcop);
5396 no_op("Bareword",s);
5399 /* Look for a subroutine with this name in current package,
5400 unless name is "Foo::", in which case Foo is a bearword
5401 (and a package name). */
5403 if (len > 2 && !PL_madskills &&
5404 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5406 if (ckWARN(WARN_BAREWORD)
5407 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5408 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5409 "Bareword \"%s\" refers to nonexistent package",
5412 PL_tokenbuf[len] = '\0';
5418 /* Mustn't actually add anything to a symbol table.
5419 But also don't want to "initialise" any placeholder
5420 constants that might already be there into full
5421 blown PVGVs with attached PVCV. */
5422 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5423 GV_NOADD_NOINIT, SVt_PVCV);
5428 /* if we saw a global override before, get the right name */
5431 sv = newSVpvs("CORE::GLOBAL::");
5432 sv_catpv(sv,PL_tokenbuf);
5435 /* If len is 0, newSVpv does strlen(), which is correct.
5436 If len is non-zero, then it will be the true length,
5437 and so the scalar will be created correctly. */
5438 sv = newSVpv(PL_tokenbuf,len);
5441 if (PL_madskills && !PL_thistoken) {
5442 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5443 PL_thistoken = newSVpvn(start,s - start);
5444 PL_realtokenstart = s - SvPVX(PL_linestr);
5448 /* Presume this is going to be a bareword of some sort. */
5451 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5452 pl_yylval.opval->op_private = OPpCONST_BARE;
5453 /* UTF-8 package name? */
5454 if (UTF && !IN_BYTES &&
5455 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5458 /* And if "Foo::", then that's what it certainly is. */
5463 /* Do the explicit type check so that we don't need to force
5464 the initialisation of the symbol table to have a real GV.
5465 Beware - gv may not really be a PVGV, cv may not really be
5466 a PVCV, (because of the space optimisations that gv_init
5467 understands) But they're true if for this symbol there is
5468 respectively a typeglob and a subroutine.
5470 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5471 /* Real typeglob, so get the real subroutine: */
5473 /* A proxy for a subroutine in this package? */
5474 : SvOK(gv) ? MUTABLE_CV(gv) : NULL)
5477 /* See if it's the indirect object for a list operator. */
5479 if (PL_oldoldbufptr &&
5480 PL_oldoldbufptr < PL_bufptr &&
5481 (PL_oldoldbufptr == PL_last_lop
5482 || PL_oldoldbufptr == PL_last_uni) &&
5483 /* NO SKIPSPACE BEFORE HERE! */
5484 (PL_expect == XREF ||
5485 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5487 bool immediate_paren = *s == '(';
5489 /* (Now we can afford to cross potential line boundary.) */
5490 s = SKIPSPACE2(s,nextPL_nextwhite);
5492 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5495 /* Two barewords in a row may indicate method call. */
5497 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5498 (tmp = intuit_method(s, gv, cv)))
5501 /* If not a declared subroutine, it's an indirect object. */
5502 /* (But it's an indir obj regardless for sort.) */
5503 /* Also, if "_" follows a filetest operator, it's a bareword */
5506 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5508 (PL_last_lop_op != OP_MAPSTART &&
5509 PL_last_lop_op != OP_GREPSTART))))
5510 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5511 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5514 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5519 PL_expect = XOPERATOR;
5522 s = SKIPSPACE2(s,nextPL_nextwhite);
5523 PL_nextwhite = nextPL_nextwhite;
5528 /* Is this a word before a => operator? */
5529 if (*s == '=' && s[1] == '>' && !pkgname) {
5531 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5532 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5533 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5537 /* If followed by a paren, it's certainly a subroutine. */
5542 while (SPACE_OR_TAB(*d))
5544 if (*d == ')' && (sv = gv_const_sv(gv))) {
5551 PL_nextwhite = PL_thiswhite;
5554 start_force(PL_curforce);
5556 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5557 PL_expect = XOPERATOR;
5560 PL_nextwhite = nextPL_nextwhite;
5561 curmad('X', PL_thistoken);
5562 PL_thistoken = newSVpvs("");
5570 /* If followed by var or block, call it a method (unless sub) */
5572 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5573 PL_last_lop = PL_oldbufptr;
5574 PL_last_lop_op = OP_METHOD;
5578 /* If followed by a bareword, see if it looks like indir obj. */
5581 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5582 && (tmp = intuit_method(s, gv, cv)))
5585 /* Not a method, so call it a subroutine (if defined) */
5588 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5589 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5590 "Ambiguous use of -%s resolved as -&%s()",
5591 PL_tokenbuf, PL_tokenbuf);
5592 /* Check for a constant sub */
5593 if ((sv = gv_const_sv(gv))) {
5595 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5596 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5597 pl_yylval.opval->op_private = 0;
5601 /* Resolve to GV now. */
5602 if (SvTYPE(gv) != SVt_PVGV) {
5603 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5604 assert (SvTYPE(gv) == SVt_PVGV);
5605 /* cv must have been some sort of placeholder, so
5606 now needs replacing with a real code reference. */
5610 op_free(pl_yylval.opval);
5611 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5612 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5613 PL_last_lop = PL_oldbufptr;
5614 PL_last_lop_op = OP_ENTERSUB;
5615 /* Is there a prototype? */
5623 const char *proto = SvPV_const(MUTABLE_SV(cv), protolen);
5626 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5628 while (*proto == ';')
5630 if (*proto == '&' && *s == '{') {
5632 sv_setpvs(PL_subname, "__ANON__");
5634 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5641 PL_nextwhite = PL_thiswhite;
5644 start_force(PL_curforce);
5645 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5648 PL_nextwhite = nextPL_nextwhite;
5649 curmad('X', PL_thistoken);
5650 PL_thistoken = newSVpvs("");
5657 /* Guess harder when madskills require "best effort". */
5658 if (PL_madskills && (!gv || !GvCVu(gv))) {
5659 int probable_sub = 0;
5660 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5662 else if (isALPHA(*s)) {
5666 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5667 if (!keyword(tmpbuf, tmplen, 0))
5670 while (d < PL_bufend && isSPACE(*d))
5672 if (*d == '=' && d[1] == '>')
5677 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5678 op_free(pl_yylval.opval);
5679 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5680 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5681 PL_last_lop = PL_oldbufptr;
5682 PL_last_lop_op = OP_ENTERSUB;
5683 PL_nextwhite = PL_thiswhite;
5685 start_force(PL_curforce);
5686 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5688 PL_nextwhite = nextPL_nextwhite;
5689 curmad('X', PL_thistoken);
5690 PL_thistoken = newSVpvs("");
5695 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5702 /* Call it a bare word */
5704 if (PL_hints & HINT_STRICT_SUBS)
5705 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5708 /* after "print" and similar functions (corresponding to
5709 * "F? L" in opcode.pl), whatever wasn't already parsed as
5710 * a filehandle should be subject to "strict subs".
5711 * Likewise for the optional indirect-object argument to system
5712 * or exec, which can't be a bareword */
5713 if ((PL_last_lop_op == OP_PRINT
5714 || PL_last_lop_op == OP_PRTF
5715 || PL_last_lop_op == OP_SAY
5716 || PL_last_lop_op == OP_SYSTEM
5717 || PL_last_lop_op == OP_EXEC)
5718 && (PL_hints & HINT_STRICT_SUBS))
5719 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5720 if (lastchar != '-') {
5721 if (ckWARN(WARN_RESERVED)) {
5725 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5726 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5733 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5734 && ckWARN_d(WARN_AMBIGUOUS)) {
5735 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5736 "Operator or semicolon missing before %c%s",
5737 lastchar, PL_tokenbuf);
5738 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5739 "Ambiguous use of %c resolved as operator %c",
5740 lastchar, lastchar);
5746 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5747 newSVpv(CopFILE(PL_curcop),0));
5751 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5752 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5755 case KEY___PACKAGE__:
5756 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5758 ? newSVhek(HvNAME_HEK(PL_curstash))
5765 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5766 const char *pname = "main";
5767 if (PL_tokenbuf[2] == 'D')
5768 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5769 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5773 GvIOp(gv) = newIO();
5774 IoIFP(GvIOp(gv)) = PL_rsfp;
5775 #if defined(HAS_FCNTL) && defined(F_SETFD)
5777 const int fd = PerlIO_fileno(PL_rsfp);
5778 fcntl(fd,F_SETFD,fd >= 3);
5781 /* Mark this internal pseudo-handle as clean */
5782 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5783 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5784 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5786 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5787 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5788 /* if the script was opened in binmode, we need to revert
5789 * it to text mode for compatibility; but only iff it has CRs
5790 * XXX this is a questionable hack at best. */
5791 if (PL_bufend-PL_bufptr > 2
5792 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5795 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5796 loc = PerlIO_tell(PL_rsfp);
5797 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5800 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5802 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5803 #endif /* NETWARE */
5804 #ifdef PERLIO_IS_STDIO /* really? */
5805 # if defined(__BORLANDC__)
5806 /* XXX see note in do_binmode() */
5807 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5811 PerlIO_seek(PL_rsfp, loc, 0);
5815 #ifdef PERLIO_LAYERS
5818 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5819 else if (PL_encoding) {
5826 XPUSHs(PL_encoding);
5828 call_method("name", G_SCALAR);
5832 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5833 Perl_form(aTHX_ ":encoding(%"SVf")",
5842 if (PL_realtokenstart >= 0) {
5843 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5845 PL_endwhite = newSVpvs("");
5846 sv_catsv(PL_endwhite, PL_thiswhite);
5848 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5849 PL_realtokenstart = -1;
5851 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5852 SvCUR(PL_endwhite))) != NULL) ;
5867 if (PL_expect == XSTATE) {
5874 if (*s == ':' && s[1] == ':') {
5877 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5878 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5879 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5882 else if (tmp == KEY_require || tmp == KEY_do)
5883 /* that's a way to remember we saw "CORE::" */
5896 LOP(OP_ACCEPT,XTERM);
5902 LOP(OP_ATAN2,XTERM);
5908 LOP(OP_BINMODE,XTERM);
5911 LOP(OP_BLESS,XTERM);
5920 /* When 'use switch' is in effect, continue has a dual
5921 life as a control operator. */
5923 if (!FEATURE_IS_ENABLED("switch"))
5926 /* We have to disambiguate the two senses of
5927 "continue". If the next token is a '{' then
5928 treat it as the start of a continue block;
5929 otherwise treat it as a control operator.
5941 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5958 if (!PL_cryptseen) {
5959 PL_cryptseen = TRUE;
5963 LOP(OP_CRYPT,XTERM);
5966 LOP(OP_CHMOD,XTERM);
5969 LOP(OP_CHOWN,XTERM);
5972 LOP(OP_CONNECT,XTERM);
5991 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5992 if (orig_keyword == KEY_do) {
6001 PL_hints |= HINT_BLOCK_SCOPE;
6011 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
6012 LOP(OP_DBMOPEN,XTERM);
6018 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6025 pl_yylval.ival = CopLINE(PL_curcop);
6041 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
6042 UNIBRACK(OP_ENTEREVAL);
6056 case KEY_endhostent:
6062 case KEY_endservent:
6065 case KEY_endprotoent:
6076 pl_yylval.ival = CopLINE(PL_curcop);
6078 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6081 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6084 if ((PL_bufend - p) >= 3 &&
6085 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6087 else if ((PL_bufend - p) >= 4 &&
6088 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6091 if (isIDFIRST_lazy_if(p,UTF)) {
6092 p = scan_ident(p, PL_bufend,
6093 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6097 Perl_croak(aTHX_ "Missing $ on loop variable");
6099 s = SvPVX(PL_linestr) + soff;
6105 LOP(OP_FORMLINE,XTERM);
6111 LOP(OP_FCNTL,XTERM);
6117 LOP(OP_FLOCK,XTERM);
6126 LOP(OP_GREPSTART, XREF);
6129 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6144 case KEY_getpriority:
6145 LOP(OP_GETPRIORITY,XTERM);
6147 case KEY_getprotobyname:
6150 case KEY_getprotobynumber:
6151 LOP(OP_GPBYNUMBER,XTERM);
6153 case KEY_getprotoent:
6165 case KEY_getpeername:
6166 UNI(OP_GETPEERNAME);
6168 case KEY_gethostbyname:
6171 case KEY_gethostbyaddr:
6172 LOP(OP_GHBYADDR,XTERM);
6174 case KEY_gethostent:
6177 case KEY_getnetbyname:
6180 case KEY_getnetbyaddr:
6181 LOP(OP_GNBYADDR,XTERM);
6186 case KEY_getservbyname:
6187 LOP(OP_GSBYNAME,XTERM);
6189 case KEY_getservbyport:
6190 LOP(OP_GSBYPORT,XTERM);
6192 case KEY_getservent:
6195 case KEY_getsockname:
6196 UNI(OP_GETSOCKNAME);
6198 case KEY_getsockopt:
6199 LOP(OP_GSOCKOPT,XTERM);
6214 pl_yylval.ival = CopLINE(PL_curcop);
6224 pl_yylval.ival = CopLINE(PL_curcop);
6228 LOP(OP_INDEX,XTERM);
6234 LOP(OP_IOCTL,XTERM);
6246 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6278 LOP(OP_LISTEN,XTERM);
6287 s = scan_pat(s,OP_MATCH);
6288 TERM(sublex_start());
6291 LOP(OP_MAPSTART, XREF);
6294 LOP(OP_MKDIR,XTERM);
6297 LOP(OP_MSGCTL,XTERM);
6300 LOP(OP_MSGGET,XTERM);
6303 LOP(OP_MSGRCV,XTERM);
6306 LOP(OP_MSGSND,XTERM);
6311 PL_in_my = (U16)tmp;
6313 if (isIDFIRST_lazy_if(s,UTF)) {
6317 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6318 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6320 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6321 if (!PL_in_my_stash) {
6324 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6328 if (PL_madskills) { /* just add type to declarator token */
6329 sv_catsv(PL_thistoken, PL_nextwhite);
6331 sv_catpvn(PL_thistoken, start, s - start);
6339 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6346 s = tokenize_use(0, s);
6350 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6357 if (isIDFIRST_lazy_if(s,UTF)) {
6359 for (d = s; isALNUM_lazy_if(d,UTF);)
6361 for (t=d; isSPACE(*t);)
6363 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6365 && !(t[0] == '=' && t[1] == '>')
6367 int parms_len = (int)(d-s);
6368 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6369 "Precedence problem: open %.*s should be open(%.*s)",
6370 parms_len, s, parms_len, s);
6376 pl_yylval.ival = OP_OR;
6386 LOP(OP_OPEN_DIR,XTERM);
6389 checkcomma(s,PL_tokenbuf,"filehandle");
6393 checkcomma(s,PL_tokenbuf,"filehandle");
6412 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6413 s = force_version(s, FALSE);
6417 LOP(OP_PIPE_OP,XTERM);
6420 s = scan_str(s,!!PL_madskills,FALSE);
6423 pl_yylval.ival = OP_CONST;
6424 TERM(sublex_start());
6430 s = scan_str(s,!!PL_madskills,FALSE);
6433 PL_expect = XOPERATOR;
6435 if (SvCUR(PL_lex_stuff)) {
6438 d = SvPV_force(PL_lex_stuff, len);
6440 for (; isSPACE(*d) && len; --len, ++d)
6445 if (!warned && ckWARN(WARN_QW)) {
6446 for (; !isSPACE(*d) && len; --len, ++d) {
6448 Perl_warner(aTHX_ packWARN(WARN_QW),
6449 "Possible attempt to separate words with commas");
6452 else if (*d == '#') {
6453 Perl_warner(aTHX_ packWARN(WARN_QW),
6454 "Possible attempt to put comments in qw() list");
6460 for (; !isSPACE(*d) && len; --len, ++d)
6463 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6464 words = append_elem(OP_LIST, words,
6465 newSVOP(OP_CONST, 0, tokeq(sv)));
6469 start_force(PL_curforce);
6470 NEXTVAL_NEXTTOKE.opval = words;
6475 SvREFCNT_dec(PL_lex_stuff);
6476 PL_lex_stuff = NULL;
6482 s = scan_str(s,!!PL_madskills,FALSE);
6485 pl_yylval.ival = OP_STRINGIFY;
6486 if (SvIVX(PL_lex_stuff) == '\'')
6487 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6488 TERM(sublex_start());
6491 s = scan_pat(s,OP_QR);
6492 TERM(sublex_start());
6495 s = scan_str(s,!!PL_madskills,FALSE);
6498 readpipe_override();
6499 TERM(sublex_start());
6507 s = force_version(s, FALSE);
6509 else if (*s != 'v' || !isDIGIT(s[1])
6510 || (s = force_version(s, TRUE), *s == 'v'))
6512 *PL_tokenbuf = '\0';
6513 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6514 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6515 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6517 yyerror("<> should be quotes");
6519 if (orig_keyword == KEY_require) {
6527 PL_last_uni = PL_oldbufptr;
6528 PL_last_lop_op = OP_REQUIRE;
6530 return REPORT( (int)REQUIRE );
6536 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6540 LOP(OP_RENAME,XTERM);
6549 LOP(OP_RINDEX,XTERM);
6558 UNIDOR(OP_READLINE);
6561 UNIDOR(OP_BACKTICK);
6570 LOP(OP_REVERSE,XTERM);
6573 UNIDOR(OP_READLINK);
6580 if (pl_yylval.opval)
6581 TERM(sublex_start());
6583 TOKEN(1); /* force error */
6586 checkcomma(s,PL_tokenbuf,"filehandle");
6596 LOP(OP_SELECT,XTERM);
6602 LOP(OP_SEMCTL,XTERM);
6605 LOP(OP_SEMGET,XTERM);
6608 LOP(OP_SEMOP,XTERM);
6614 LOP(OP_SETPGRP,XTERM);
6616 case KEY_setpriority:
6617 LOP(OP_SETPRIORITY,XTERM);
6619 case KEY_sethostent:
6625 case KEY_setservent:
6628 case KEY_setprotoent:
6638 LOP(OP_SEEKDIR,XTERM);
6640 case KEY_setsockopt:
6641 LOP(OP_SSOCKOPT,XTERM);
6647 LOP(OP_SHMCTL,XTERM);
6650 LOP(OP_SHMGET,XTERM);
6653 LOP(OP_SHMREAD,XTERM);
6656 LOP(OP_SHMWRITE,XTERM);
6659 LOP(OP_SHUTDOWN,XTERM);
6668 LOP(OP_SOCKET,XTERM);
6670 case KEY_socketpair:
6671 LOP(OP_SOCKPAIR,XTERM);
6674 checkcomma(s,PL_tokenbuf,"subroutine name");
6676 if (*s == ';' || *s == ')') /* probably a close */
6677 Perl_croak(aTHX_ "sort is now a reserved word");
6679 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6683 LOP(OP_SPLIT,XTERM);
6686 LOP(OP_SPRINTF,XTERM);
6689 LOP(OP_SPLICE,XTERM);
6704 LOP(OP_SUBSTR,XTERM);
6710 char tmpbuf[sizeof PL_tokenbuf];
6711 SSize_t tboffset = 0;
6712 expectation attrful;
6713 bool have_name, have_proto;
6714 const int key = tmp;
6719 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6720 SV *subtoken = newSVpvn(tstart, s - tstart);
6724 s = SKIPSPACE2(s,tmpwhite);
6729 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6730 (*s == ':' && s[1] == ':'))
6733 SV *nametoke = NULL;
6737 attrful = XATTRBLOCK;
6738 /* remember buffer pos'n for later force_word */
6739 tboffset = s - PL_oldbufptr;
6740 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6743 nametoke = newSVpvn(s, d - s);
6745 if (memchr(tmpbuf, ':', len))
6746 sv_setpvn(PL_subname, tmpbuf, len);
6748 sv_setsv(PL_subname,PL_curstname);
6749 sv_catpvs(PL_subname,"::");
6750 sv_catpvn(PL_subname,tmpbuf,len);
6757 CURMAD('X', nametoke);
6758 CURMAD('_', tmpwhite);
6759 (void) force_word(PL_oldbufptr + tboffset, WORD,
6762 s = SKIPSPACE2(d,tmpwhite);
6769 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6770 PL_expect = XTERMBLOCK;
6771 attrful = XATTRTERM;
6772 sv_setpvs(PL_subname,"?");
6776 if (key == KEY_format) {
6778 PL_lex_formbrack = PL_lex_brackets + 1;
6780 PL_thistoken = subtoken;
6784 (void) force_word(PL_oldbufptr + tboffset, WORD,
6790 /* Look for a prototype */
6793 bool bad_proto = FALSE;
6794 bool in_brackets = FALSE;
6795 char greedy_proto = ' ';
6796 bool proto_after_greedy_proto = FALSE;
6797 bool must_be_last = FALSE;
6798 bool underscore = FALSE;
6799 bool seen_underscore = FALSE;
6800 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6802 s = scan_str(s,!!PL_madskills,FALSE);
6804 Perl_croak(aTHX_ "Prototype not terminated");
6805 /* strip spaces and check for bad characters */
6806 d = SvPVX(PL_lex_stuff);
6808 for (p = d; *p; ++p) {
6814 proto_after_greedy_proto = TRUE;
6815 if (!strchr("$@%*;[]&\\_", *p)) {
6827 else if ( *p == ']' ) {
6828 in_brackets = FALSE;
6830 else if ( (*p == '@' || *p == '%') &&
6831 ( tmp < 2 || d[tmp-2] != '\\' ) &&
6833 must_be_last = TRUE;
6836 else if ( *p == '_' ) {
6837 underscore = seen_underscore = TRUE;
6844 if (proto_after_greedy_proto)
6845 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6846 "Prototype after '%c' for %"SVf" : %s",
6847 greedy_proto, SVfARG(PL_subname), d);
6849 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6850 "Illegal character %sin prototype for %"SVf" : %s",
6851 seen_underscore ? "after '_' " : "",
6852 SVfARG(PL_subname), d);
6853 SvCUR_set(PL_lex_stuff, tmp);
6858 CURMAD('q', PL_thisopen);
6859 CURMAD('_', tmpwhite);
6860 CURMAD('=', PL_thisstuff);
6861 CURMAD('Q', PL_thisclose);
6862 NEXTVAL_NEXTTOKE.opval =
6863 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6864 PL_lex_stuff = NULL;
6867 s = SKIPSPACE2(s,tmpwhite);
6875 if (*s == ':' && s[1] != ':')
6876 PL_expect = attrful;
6877 else if (*s != '{' && key == KEY_sub) {
6879 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6881 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6888 curmad('^', newSVpvs(""));
6889 CURMAD('_', tmpwhite);
6893 PL_thistoken = subtoken;
6896 NEXTVAL_NEXTTOKE.opval =
6897 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6898 PL_lex_stuff = NULL;
6904 sv_setpvs(PL_subname, "__ANON__");
6906 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6910 (void) force_word(PL_oldbufptr + tboffset, WORD,
6919 LOP(OP_SYSTEM,XREF);
6922 LOP(OP_SYMLINK,XTERM);
6925 LOP(OP_SYSCALL,XTERM);
6928 LOP(OP_SYSOPEN,XTERM);
6931 LOP(OP_SYSSEEK,XTERM);
6934 LOP(OP_SYSREAD,XTERM);
6937 LOP(OP_SYSWRITE,XTERM);
6941 TERM(sublex_start());
6962 LOP(OP_TRUNCATE,XTERM);
6974 pl_yylval.ival = CopLINE(PL_curcop);
6978 pl_yylval.ival = CopLINE(PL_curcop);
6982 LOP(OP_UNLINK,XTERM);
6988 LOP(OP_UNPACK,XTERM);
6991 LOP(OP_UTIME,XTERM);
6997 LOP(OP_UNSHIFT,XTERM);
7000 s = tokenize_use(1, s);
7010 pl_yylval.ival = CopLINE(PL_curcop);
7014 pl_yylval.ival = CopLINE(PL_curcop);
7018 PL_hints |= HINT_BLOCK_SCOPE;
7025 LOP(OP_WAITPID,XTERM);
7034 ctl_l[0] = toCTRL('L');
7036 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
7039 /* Make sure $^L is defined */
7040 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
7045 if (PL_expect == XOPERATOR)
7051 pl_yylval.ival = OP_XOR;
7056 TERM(sublex_start());
7061 #pragma segment Main
7065 S_pending_ident(pTHX)
7070 /* pit holds the identifier we read and pending_ident is reset */
7071 char pit = PL_pending_ident;
7072 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7073 /* All routes through this function want to know if there is a colon. */
7074 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7075 PL_pending_ident = 0;
7077 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7078 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7079 "### Pending identifier '%s'\n", PL_tokenbuf); });
7081 /* if we're in a my(), we can't allow dynamics here.
7082 $foo'bar has already been turned into $foo::bar, so
7083 just check for colons.
7085 if it's a legal name, the OP is a PADANY.
7088 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7090 yyerror(Perl_form(aTHX_ "No package name allowed for "
7091 "variable %s in \"our\"",
7093 tmp = allocmy(PL_tokenbuf);
7097 yyerror(Perl_form(aTHX_ PL_no_myglob,
7098 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7100 pl_yylval.opval = newOP(OP_PADANY, 0);
7101 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7107 build the ops for accesses to a my() variable.
7109 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7110 then used in a comparison. This catches most, but not
7111 all cases. For instance, it catches
7112 sort { my($a); $a <=> $b }
7114 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7115 (although why you'd do that is anyone's guess).
7120 tmp = pad_findmy(PL_tokenbuf);
7121 if (tmp != NOT_IN_PAD) {
7122 /* might be an "our" variable" */
7123 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7124 /* build ops for a bareword */
7125 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7126 HEK * const stashname = HvNAME_HEK(stash);
7127 SV * const sym = newSVhek(stashname);
7128 sv_catpvs(sym, "::");
7129 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7130 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7131 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7134 ? (GV_ADDMULTI | GV_ADDINEVAL)
7137 ((PL_tokenbuf[0] == '$') ? SVt_PV
7138 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7143 /* if it's a sort block and they're naming $a or $b */
7144 if (PL_last_lop_op == OP_SORT &&
7145 PL_tokenbuf[0] == '$' &&
7146 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7149 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7150 d < PL_bufend && *d != '\n';
7153 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7154 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7160 pl_yylval.opval = newOP(OP_PADANY, 0);
7161 pl_yylval.opval->op_targ = tmp;
7167 Whine if they've said @foo in a doublequoted string,
7168 and @foo isn't a variable we can find in the symbol
7171 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7172 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7174 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7175 && ckWARN(WARN_AMBIGUOUS)
7176 /* DO NOT warn for @- and @+ */
7177 && !( PL_tokenbuf[2] == '\0' &&
7178 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7181 /* Downgraded from fatal to warning 20000522 mjd */
7182 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7183 "Possible unintended interpolation of %s in string",
7188 /* build ops for a bareword */
7189 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7191 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7193 PL_tokenbuf + 1, tokenbuf_len - 1,
7194 /* If the identifier refers to a stash, don't autovivify it.
7195 * Change 24660 had the side effect of causing symbol table
7196 * hashes to always be defined, even if they were freshly
7197 * created and the only reference in the entire program was
7198 * the single statement with the defined %foo::bar:: test.
7199 * It appears that all code in the wild doing this actually
7200 * wants to know whether sub-packages have been loaded, so
7201 * by avoiding auto-vivifying symbol tables, we ensure that
7202 * defined %foo::bar:: continues to be false, and the existing
7203 * tests still give the expected answers, even though what
7204 * they're actually testing has now changed subtly.
7206 (*PL_tokenbuf == '%'
7207 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7210 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7211 ((PL_tokenbuf[0] == '$') ? SVt_PV
7212 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7218 * The following code was generated by perl_keyword.pl.
7222 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7226 PERL_ARGS_ASSERT_KEYWORD;
7230 case 1: /* 5 tokens of length 1 */
7262 case 2: /* 18 tokens of length 2 */
7408 case 3: /* 29 tokens of length 3 */
7412 if (name[1] == 'N' &&
7475 if (name[1] == 'i' &&
7507 if (name[1] == 'o' &&
7516 if (name[1] == 'e' &&
7525 if (name[1] == 'n' &&
7534 if (name[1] == 'o' &&
7543 if (name[1] == 'a' &&
7552 if (name[1] == 'o' &&
7614 if (name[1] == 'e' &&
7628 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7654 if (name[1] == 'i' &&
7663 if (name[1] == 's' &&
7672 if (name[1] == 'e' &&
7681 if (name[1] == 'o' &&
7693 case 4: /* 41 tokens of length 4 */
7697 if (name[1] == 'O' &&
7707 if (name[1] == 'N' &&
7717 if (name[1] == 'i' &&
7727 if (name[1] == 'h' &&
7737 if (name[1] == 'u' &&
7750 if (name[2] == 'c' &&
7759 if (name[2] == 's' &&
7768 if (name[2] == 'a' &&
7804 if (name[1] == 'o' &&
7817 if (name[2] == 't' &&
7826 if (name[2] == 'o' &&
7835 if (name[2] == 't' &&
7844 if (name[2] == 'e' &&
7857 if (name[1] == 'o' &&
7870 if (name[2] == 'y' &&
7879 if (name[2] == 'l' &&
7895 if (name[2] == 's' &&
7904 if (name[2] == 'n' &&
7913 if (name[2] == 'c' &&
7926 if (name[1] == 'e' &&
7936 if (name[1] == 'p' &&
7949 if (name[2] == 'c' &&
7958 if (name[2] == 'p' &&
7967 if (name[2] == 's' &&
7983 if (name[2] == 'n' &&
8053 if (name[2] == 'r' &&
8062 if (name[2] == 'r' &&
8071 if (name[2] == 'a' &&
8087 if (name[2] == 'l' &&
8149 if (name[2] == 'e' &&
8152 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8165 case 5: /* 39 tokens of length 5 */
8169 if (name[1] == 'E' &&
8180 if (name[1] == 'H' &&
8194 if (name[2] == 'a' &&
8204 if (name[2] == 'a' &&
8221 if (name[2] == 'e' &&
8231 if (name[2] == 'e' &&
8235 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8251 if (name[3] == 'i' &&
8260 if (name[3] == 'o' &&
8296 if (name[2] == 'o' &&
8306 if (name[2] == 'y' &&
8320 if (name[1] == 'l' &&
8334 if (name[2] == 'n' &&
8344 if (name[2] == 'o' &&
8358 if (name[1] == 'i' &&
8363 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8372 if (name[2] == 'd' &&
8382 if (name[2] == 'c' &&
8399 if (name[2] == 'c' &&
8409 if (name[2] == 't' &&
8423 if (name[1] == 'k' &&
8434 if (name[1] == 'r' &&
8448 if (name[2] == 's' &&
8458 if (name[2] == 'd' &&
8475 if (name[2] == 'm' &&
8485 if (name[2] == 'i' &&
8495 if (name[2] == 'e' &&
8505 if (name[2] == 'l' &&
8515 if (name[2] == 'a' &&
8528 if (name[3] == 't' &&
8531 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8537 if (name[3] == 'd' &&
8554 if (name[1] == 'i' &&
8568 if (name[2] == 'a' &&
8581 if (name[3] == 'e' &&
8616 if (name[2] == 'i' &&
8633 if (name[2] == 'i' &&
8643 if (name[2] == 'i' &&
8660 case 6: /* 33 tokens of length 6 */
8664 if (name[1] == 'c' &&
8679 if (name[2] == 'l' &&
8690 if (name[2] == 'r' &&
8705 if (name[1] == 'e' &&
8720 if (name[2] == 's' &&
8725 if(ckWARN_d(WARN_SYNTAX))
8726 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8732 if (name[2] == 'i' &&
8750 if (name[2] == 'l' &&
8761 if (name[2] == 'r' &&
8776 if (name[1] == 'm' &&
8791 if (name[2] == 'n' &&
8802 if (name[2] == 's' &&
8817 if (name[1] == 's' &&
8823 if (name[4] == 't' &&
8832 if (name[4] == 'e' &&
8841 if (name[4] == 'c' &&
8850 if (name[4] == 'n' &&
8866 if (name[1] == 'r' &&
8884 if (name[3] == 'a' &&
8894 if (name[3] == 'u' &&
8908 if (name[2] == 'n' &&
8926 if (name[2] == 'a' &&
8940 if (name[3] == 'e' &&
8953 if (name[4] == 't' &&
8962 if (name[4] == 'e' &&
8984 if (name[4] == 't' &&
8993 if (name[4] == 'e' &&
9009 if (name[2] == 'c' &&
9020 if (name[2] == 'l' &&
9031 if (name[2] == 'b' &&
9042 if (name[2] == 's' &&
9065 if (name[4] == 's' &&
9074 if (name[4] == 'n' &&
9087 if (name[3] == 'a' &&
9104 if (name[1] == 'a' &&
9119 case 7: /* 29 tokens of length 7 */
9123 if (name[1] == 'E' &&
9136 if (name[1] == '_' &&
9149 if (name[1] == 'i' &&
9156 return -KEY_binmode;
9162 if (name[1] == 'o' &&
9169 return -KEY_connect;
9178 if (name[2] == 'm' &&
9184 return -KEY_dbmopen;
9195 if (name[4] == 'u' &&
9199 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9205 if (name[4] == 'n' &&
9226 if (name[1] == 'o' &&
9239 if (name[1] == 'e' &&
9246 if (name[5] == 'r' &&
9249 return -KEY_getpgrp;
9255 if (name[5] == 'i' &&
9258 return -KEY_getppid;
9271 if (name[1] == 'c' &&
9278 return -KEY_lcfirst;
9284 if (name[1] == 'p' &&
9291 return -KEY_opendir;
9297 if (name[1] == 'a' &&
9315 if (name[3] == 'd' &&
9320 return -KEY_readdir;
9326 if (name[3] == 'u' &&
9337 if (name[3] == 'e' &&
9342 return -KEY_reverse;
9361 if (name[3] == 'k' &&
9366 return -KEY_seekdir;
9372 if (name[3] == 'p' &&
9377 return -KEY_setpgrp;
9387 if (name[2] == 'm' &&
9393 return -KEY_shmread;
9399 if (name[2] == 'r' &&
9405 return -KEY_sprintf;
9414 if (name[3] == 'l' &&
9419 return -KEY_symlink;
9428 if (name[4] == 'a' &&
9432 return -KEY_syscall;
9438 if (name[4] == 'p' &&
9442 return -KEY_sysopen;
9448 if (name[4] == 'e' &&
9452 return -KEY_sysread;
9458 if (name[4] == 'e' &&
9462 return -KEY_sysseek;
9480 if (name[1] == 'e' &&
9487 return -KEY_telldir;
9496 if (name[2] == 'f' &&
9502 return -KEY_ucfirst;
9508 if (name[2] == 's' &&
9514 return -KEY_unshift;
9524 if (name[1] == 'a' &&
9531 return -KEY_waitpid;
9540 case 8: /* 26 tokens of length 8 */
9544 if (name[1] == 'U' &&
9552 return KEY_AUTOLOAD;
9563 if (name[3] == 'A' &&
9569 return KEY___DATA__;
9575 if (name[3] == 'I' &&
9581 return -KEY___FILE__;
9587 if (name[3] == 'I' &&
9593 return -KEY___LINE__;
9609 if (name[2] == 'o' &&
9616 return -KEY_closedir;
9622 if (name[2] == 'n' &&
9629 return -KEY_continue;
9639 if (name[1] == 'b' &&
9647 return -KEY_dbmclose;
9653 if (name[1] == 'n' &&
9659 if (name[4] == 'r' &&
9664 return -KEY_endgrent;
9670 if (name[4] == 'w' &&
9675 return -KEY_endpwent;
9688 if (name[1] == 'o' &&
9696 return -KEY_formline;
9702 if (name[1] == 'e' &&
9713 if (name[6] == 'n' &&
9716 return -KEY_getgrent;
9722 if (name[6] == 'i' &&
9725 return -KEY_getgrgid;
9731 if (name[6] == 'a' &&
9734 return -KEY_getgrnam;
9747 if (name[4] == 'o' &&
9752 return -KEY_getlogin;
9763 if (name[6] == 'n' &&
9766 return -KEY_getpwent;
9772 if (name[6] == 'a' &&
9775 return -KEY_getpwnam;
9781 if (name[6] == 'i' &&
9784 return -KEY_getpwuid;
9804 if (name[1] == 'e' &&
9811 if (name[5] == 'i' &&
9818 return -KEY_readline;
9823 return -KEY_readlink;
9834 if (name[5] == 'i' &&
9838 return -KEY_readpipe;
9859 if (name[4] == 'r' &&
9864 return -KEY_setgrent;
9870 if (name[4] == 'w' &&
9875 return -KEY_setpwent;
9891 if (name[3] == 'w' &&
9897 return -KEY_shmwrite;
9903 if (name[3] == 't' &&
9909 return -KEY_shutdown;
9919 if (name[2] == 's' &&
9926 return -KEY_syswrite;
9936 if (name[1] == 'r' &&
9944 return -KEY_truncate;
9953 case 9: /* 9 tokens of length 9 */
9957 if (name[1] == 'N' &&
9966 return KEY_UNITCHECK;
9972 if (name[1] == 'n' &&
9981 return -KEY_endnetent;
9987 if (name[1] == 'e' &&
9996 return -KEY_getnetent;
10002 if (name[1] == 'o' &&
10011 return -KEY_localtime;
10017 if (name[1] == 'r' &&
10026 return KEY_prototype;
10032 if (name[1] == 'u' &&
10041 return -KEY_quotemeta;
10047 if (name[1] == 'e' &&
10056 return -KEY_rewinddir;
10062 if (name[1] == 'e' &&
10071 return -KEY_setnetent;
10077 if (name[1] == 'a' &&
10086 return -KEY_wantarray;
10095 case 10: /* 9 tokens of length 10 */
10099 if (name[1] == 'n' &&
10105 if (name[4] == 'o' &&
10112 return -KEY_endhostent;
10118 if (name[4] == 'e' &&
10125 return -KEY_endservent;
10138 if (name[1] == 'e' &&
10144 if (name[4] == 'o' &&
10151 return -KEY_gethostent;
10160 if (name[5] == 'r' &&
10166 return -KEY_getservent;
10172 if (name[5] == 'c' &&
10178 return -KEY_getsockopt;
10198 if (name[2] == 't')
10203 if (name[4] == 'o' &&
10210 return -KEY_sethostent;
10219 if (name[5] == 'r' &&
10225 return -KEY_setservent;
10231 if (name[5] == 'c' &&
10237 return -KEY_setsockopt;
10254 if (name[2] == 'c' &&
10263 return -KEY_socketpair;
10276 case 11: /* 8 tokens of length 11 */
10280 if (name[1] == '_' &&
10290 { /* __PACKAGE__ */
10291 return -KEY___PACKAGE__;
10297 if (name[1] == 'n' &&
10307 { /* endprotoent */
10308 return -KEY_endprotoent;
10314 if (name[1] == 'e' &&
10323 if (name[5] == 'e' &&
10329 { /* getpeername */
10330 return -KEY_getpeername;
10339 if (name[6] == 'o' &&
10344 { /* getpriority */
10345 return -KEY_getpriority;
10351 if (name[6] == 't' &&
10356 { /* getprotoent */
10357 return -KEY_getprotoent;
10371 if (name[4] == 'o' &&
10378 { /* getsockname */
10379 return -KEY_getsockname;
10392 if (name[1] == 'e' &&
10400 if (name[6] == 'o' &&
10405 { /* setpriority */
10406 return -KEY_setpriority;
10412 if (name[6] == 't' &&
10417 { /* setprotoent */
10418 return -KEY_setprotoent;
10434 case 12: /* 2 tokens of length 12 */
10435 if (name[0] == 'g' &&
10447 if (name[9] == 'd' &&
10450 { /* getnetbyaddr */
10451 return -KEY_getnetbyaddr;
10457 if (name[9] == 'a' &&
10460 { /* getnetbyname */
10461 return -KEY_getnetbyname;
10473 case 13: /* 4 tokens of length 13 */
10474 if (name[0] == 'g' &&
10481 if (name[4] == 'o' &&
10490 if (name[10] == 'd' &&
10493 { /* gethostbyaddr */
10494 return -KEY_gethostbyaddr;
10500 if (name[10] == 'a' &&
10503 { /* gethostbyname */
10504 return -KEY_gethostbyname;
10517 if (name[4] == 'e' &&
10526 if (name[10] == 'a' &&
10529 { /* getservbyname */
10530 return -KEY_getservbyname;
10536 if (name[10] == 'o' &&
10539 { /* getservbyport */
10540 return -KEY_getservbyport;
10559 case 14: /* 1 tokens of length 14 */
10560 if (name[0] == 'g' &&
10574 { /* getprotobyname */
10575 return -KEY_getprotobyname;
10580 case 16: /* 1 tokens of length 16 */
10581 if (name[0] == 'g' &&
10597 { /* getprotobynumber */
10598 return -KEY_getprotobynumber;
10612 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10616 PERL_ARGS_ASSERT_CHECKCOMMA;
10618 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10619 if (ckWARN(WARN_SYNTAX)) {
10622 for (w = s+2; *w && level; w++) {
10625 else if (*w == ')')
10628 while (isSPACE(*w))
10630 /* the list of chars below is for end of statements or
10631 * block / parens, boolean operators (&&, ||, //) and branch
10632 * constructs (or, and, if, until, unless, while, err, for).
10633 * Not a very solid hack... */
10634 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10635 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10636 "%s (...) interpreted as function",name);
10639 while (s < PL_bufend && isSPACE(*s))
10643 while (s < PL_bufend && isSPACE(*s))
10645 if (isIDFIRST_lazy_if(s,UTF)) {
10646 const char * const w = s++;
10647 while (isALNUM_lazy_if(s,UTF))
10649 while (s < PL_bufend && isSPACE(*s))
10653 if (keyword(w, s - w, 0))
10656 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10657 if (gv && GvCVu(gv))
10659 Perl_croak(aTHX_ "No comma allowed after %s", what);
10664 /* Either returns sv, or mortalizes sv and returns a new SV*.
10665 Best used as sv=new_constant(..., sv, ...).
10666 If s, pv are NULL, calls subroutine with one argument,
10667 and type is used with error messages only. */
10670 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10671 SV *sv, SV *pv, const char *type, STRLEN typelen)
10674 HV * const table = GvHV(PL_hintgv); /* ^H */
10678 const char *why1 = "", *why2 = "", *why3 = "";
10680 PERL_ARGS_ASSERT_NEW_CONSTANT;
10682 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10685 why2 = (const char *)
10686 (strEQ(key,"charnames")
10687 ? "(possibly a missing \"use charnames ...\")"
10689 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10690 (type ? type: "undef"), why2);
10692 /* This is convoluted and evil ("goto considered harmful")
10693 * but I do not understand the intricacies of all the different
10694 * failure modes of %^H in here. The goal here is to make
10695 * the most probable error message user-friendly. --jhi */
10700 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10701 (type ? type: "undef"), why1, why2, why3);
10703 yyerror(SvPVX_const(msg));
10707 cvp = hv_fetch(table, key, keylen, FALSE);
10708 if (!cvp || !SvOK(*cvp)) {
10711 why3 = "} is not defined";
10714 sv_2mortal(sv); /* Parent created it permanently */
10717 pv = newSVpvn_flags(s, len, SVs_TEMP);
10719 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10721 typesv = &PL_sv_undef;
10723 PUSHSTACKi(PERLSI_OVERLOAD);
10735 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10739 /* Check the eval first */
10740 if (!PL_in_eval && SvTRUE(ERRSV)) {
10741 sv_catpvs(ERRSV, "Propagated");
10742 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10744 res = SvREFCNT_inc_simple(sv);
10748 SvREFCNT_inc_simple_void(res);
10757 why1 = "Call to &{$^H{";
10759 why3 = "}} did not return a defined value";
10767 /* Returns a NUL terminated string, with the length of the string written to
10771 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10774 register char *d = dest;
10775 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10777 PERL_ARGS_ASSERT_SCAN_WORD;
10781 Perl_croak(aTHX_ ident_too_long);
10782 if (isALNUM(*s)) /* UTF handled below */
10784 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10789 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10793 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10794 char *t = s + UTF8SKIP(s);
10796 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10800 Perl_croak(aTHX_ ident_too_long);
10801 Copy(s, d, len, char);
10814 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10817 char *bracket = NULL;
10819 register char *d = dest;
10820 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10822 PERL_ARGS_ASSERT_SCAN_IDENT;
10827 while (isDIGIT(*s)) {
10829 Perl_croak(aTHX_ ident_too_long);
10836 Perl_croak(aTHX_ ident_too_long);
10837 if (isALNUM(*s)) /* UTF handled below */
10839 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10844 else if (*s == ':' && s[1] == ':') {
10848 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10849 char *t = s + UTF8SKIP(s);
10850 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10852 if (d + (t - s) > e)
10853 Perl_croak(aTHX_ ident_too_long);
10854 Copy(s, d, t - s, char);
10865 if (PL_lex_state != LEX_NORMAL)
10866 PL_lex_state = LEX_INTERPENDMAYBE;
10869 if (*s == '$' && s[1] &&
10870 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10883 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10888 if (isSPACE(s[-1])) {
10890 const char ch = *s++;
10891 if (!SPACE_OR_TAB(ch)) {
10897 if (isIDFIRST_lazy_if(d,UTF)) {
10901 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10902 end += UTF8SKIP(end);
10903 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10904 end += UTF8SKIP(end);
10906 Copy(s, d, end - s, char);
10911 while ((isALNUM(*s) || *s == ':') && d < e)
10914 Perl_croak(aTHX_ ident_too_long);
10917 while (s < send && SPACE_OR_TAB(*s))
10919 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10920 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10921 const char * const brack =
10923 ((*s == '[') ? "[...]" : "{...}");
10924 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10925 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10926 funny, dest, brack, funny, dest, brack);
10929 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10933 /* Handle extended ${^Foo} variables
10934 * 1999-02-27 mjd-perl-patch@plover.com */
10935 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10939 while (isALNUM(*s) && d < e) {
10943 Perl_croak(aTHX_ ident_too_long);
10948 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10949 PL_lex_state = LEX_INTERPEND;
10952 if (PL_lex_state == LEX_NORMAL) {
10953 if (ckWARN(WARN_AMBIGUOUS) &&
10954 (keyword(dest, d - dest, 0)
10955 || get_cvn_flags(dest, d - dest, 0)))
10959 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10960 "Ambiguous use of %c{%s} resolved to %c%s",
10961 funny, dest, funny, dest);
10966 s = bracket; /* let the parser handle it */
10970 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10971 PL_lex_state = LEX_INTERPEND;
10976 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10978 PERL_ARGS_ASSERT_PMFLAG;
10980 PERL_UNUSED_CONTEXT;
10982 const char c = (char)ch;
10984 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10985 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10986 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10987 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10988 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10994 S_scan_pat(pTHX_ char *start, I32 type)
10998 char *s = scan_str(start,!!PL_madskills,FALSE);
10999 const char * const valid_flags =
11000 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
11005 PERL_ARGS_ASSERT_SCAN_PAT;
11008 const char * const delimiter = skipspace(start);
11012 ? "Search pattern not terminated or ternary operator parsed as search pattern"
11013 : "Search pattern not terminated" ));
11016 pm = (PMOP*)newPMOP(type, 0);
11017 if (PL_multi_open == '?') {
11018 /* This is the only point in the code that sets PMf_ONCE: */
11019 pm->op_pmflags |= PMf_ONCE;
11021 /* Hence it's safe to do this bit of PMOP book-keeping here, which
11022 allows us to restrict the list needed by reset to just the ??
11024 assert(type != OP_TRANS);
11026 MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
11029 mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
11032 elements = mg->mg_len / sizeof(PMOP**);
11033 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
11034 ((PMOP**)mg->mg_ptr) [elements++] = pm;
11035 mg->mg_len = elements * sizeof(PMOP**);
11036 PmopSTASH_set(pm,PL_curstash);
11042 while (*s && strchr(valid_flags, *s))
11043 pmflag(&pm->op_pmflags,*s++);
11045 if (PL_madskills && modstart != s) {
11046 SV* tmptoken = newSVpvn(modstart, s - modstart);
11047 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11050 /* issue a warning if /c is specified,but /g is not */
11051 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
11052 && ckWARN(WARN_REGEXP))
11054 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
11055 "Use of /c modifier is meaningless without /g" );
11058 PL_lex_op = (OP*)pm;
11059 pl_yylval.ival = OP_MATCH;
11064 S_scan_subst(pTHX_ char *start)
11075 PERL_ARGS_ASSERT_SCAN_SUBST;
11077 pl_yylval.ival = OP_NULL;
11079 s = scan_str(start,!!PL_madskills,FALSE);
11082 Perl_croak(aTHX_ "Substitution pattern not terminated");
11084 if (s[-1] == PL_multi_open)
11087 if (PL_madskills) {
11088 CURMAD('q', PL_thisopen);
11089 CURMAD('_', PL_thiswhite);
11090 CURMAD('E', PL_thisstuff);
11091 CURMAD('Q', PL_thisclose);
11092 PL_realtokenstart = s - SvPVX(PL_linestr);
11096 first_start = PL_multi_start;
11097 s = scan_str(s,!!PL_madskills,FALSE);
11099 if (PL_lex_stuff) {
11100 SvREFCNT_dec(PL_lex_stuff);
11101 PL_lex_stuff = NULL;
11103 Perl_croak(aTHX_ "Substitution replacement not terminated");
11105 PL_multi_start = first_start; /* so whole substitution is taken together */
11107 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11110 if (PL_madskills) {
11111 CURMAD('z', PL_thisopen);
11112 CURMAD('R', PL_thisstuff);
11113 CURMAD('Z', PL_thisclose);
11119 if (*s == EXEC_PAT_MOD) {
11123 else if (strchr(S_PAT_MODS, *s))
11124 pmflag(&pm->op_pmflags,*s++);
11130 if (PL_madskills) {
11132 curmad('m', newSVpvn(modstart, s - modstart));
11133 append_madprops(PL_thismad, (OP*)pm, 0);
11137 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11138 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11142 SV * const repl = newSVpvs("");
11144 PL_sublex_info.super_bufptr = s;
11145 PL_sublex_info.super_bufend = PL_bufend;
11147 pm->op_pmflags |= PMf_EVAL;
11150 sv_catpvs(repl, "eval ");
11152 sv_catpvs(repl, "do ");
11154 sv_catpvs(repl, "{");
11155 sv_catsv(repl, PL_lex_repl);
11156 if (strchr(SvPVX(PL_lex_repl), '#'))
11157 sv_catpvs(repl, "\n");
11158 sv_catpvs(repl, "}");
11160 SvREFCNT_dec(PL_lex_repl);
11161 PL_lex_repl = repl;
11164 PL_lex_op = (OP*)pm;
11165 pl_yylval.ival = OP_SUBST;
11170 S_scan_trans(pTHX_ char *start)
11183 PERL_ARGS_ASSERT_SCAN_TRANS;
11185 pl_yylval.ival = OP_NULL;
11187 s = scan_str(start,!!PL_madskills,FALSE);
11189 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11191 if (s[-1] == PL_multi_open)
11194 if (PL_madskills) {
11195 CURMAD('q', PL_thisopen);
11196 CURMAD('_', PL_thiswhite);
11197 CURMAD('E', PL_thisstuff);
11198 CURMAD('Q', PL_thisclose);
11199 PL_realtokenstart = s - SvPVX(PL_linestr);
11203 s = scan_str(s,!!PL_madskills,FALSE);
11205 if (PL_lex_stuff) {
11206 SvREFCNT_dec(PL_lex_stuff);
11207 PL_lex_stuff = NULL;
11209 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11211 if (PL_madskills) {
11212 CURMAD('z', PL_thisopen);
11213 CURMAD('R', PL_thisstuff);
11214 CURMAD('Z', PL_thisclose);
11217 complement = del = squash = 0;
11224 complement = OPpTRANS_COMPLEMENT;
11227 del = OPpTRANS_DELETE;
11230 squash = OPpTRANS_SQUASH;
11239 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11240 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11241 o->op_private &= ~OPpTRANS_ALL;
11242 o->op_private |= del|squash|complement|
11243 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11244 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11247 pl_yylval.ival = OP_TRANS;
11250 if (PL_madskills) {
11252 curmad('m', newSVpvn(modstart, s - modstart));
11253 append_madprops(PL_thismad, o, 0);
11262 S_scan_heredoc(pTHX_ register char *s)
11266 I32 op_type = OP_SCALAR;
11270 const char *found_newline;
11274 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11276 I32 stuffstart = s - SvPVX(PL_linestr);
11279 PL_realtokenstart = -1;
11282 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11286 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11290 while (SPACE_OR_TAB(*peek))
11292 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11295 s = delimcpy(d, e, s, PL_bufend, term, &len);
11305 if (!isALNUM_lazy_if(s,UTF))
11306 deprecate_old("bare << to mean <<\"\"");
11307 for (; isALNUM_lazy_if(s,UTF); s++) {
11312 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11313 Perl_croak(aTHX_ "Delimiter for here document is too long");
11316 len = d - PL_tokenbuf;
11319 if (PL_madskills) {
11320 tstart = PL_tokenbuf + !outer;
11321 PL_thisclose = newSVpvn(tstart, len - !outer);
11322 tstart = SvPVX(PL_linestr) + stuffstart;
11323 PL_thisopen = newSVpvn(tstart, s - tstart);
11324 stuffstart = s - SvPVX(PL_linestr);
11327 #ifndef PERL_STRICT_CR
11328 d = strchr(s, '\r');
11330 char * const olds = s;
11332 while (s < PL_bufend) {
11338 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11347 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11354 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11355 herewas = newSVpvn(s,PL_bufend-s);
11359 herewas = newSVpvn(s-1,found_newline-s+1);
11362 herewas = newSVpvn(s,found_newline-s);
11366 if (PL_madskills) {
11367 tstart = SvPVX(PL_linestr) + stuffstart;
11369 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11371 PL_thisstuff = newSVpvn(tstart, s - tstart);
11374 s += SvCUR(herewas);
11377 stuffstart = s - SvPVX(PL_linestr);
11383 tmpstr = newSV_type(SVt_PVIV);
11384 SvGROW(tmpstr, 80);
11385 if (term == '\'') {
11386 op_type = OP_CONST;
11387 SvIV_set(tmpstr, -1);
11389 else if (term == '`') {
11390 op_type = OP_BACKTICK;
11391 SvIV_set(tmpstr, '\\');
11395 PL_multi_start = CopLINE(PL_curcop);
11396 PL_multi_open = PL_multi_close = '<';
11397 term = *PL_tokenbuf;
11398 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11399 char * const bufptr = PL_sublex_info.super_bufptr;
11400 char * const bufend = PL_sublex_info.super_bufend;
11401 char * const olds = s - SvCUR(herewas);
11402 s = strchr(bufptr, '\n');
11406 while (s < bufend &&
11407 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11409 CopLINE_inc(PL_curcop);
11412 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11413 missingterm(PL_tokenbuf);
11415 sv_setpvn(herewas,bufptr,d-bufptr+1);
11416 sv_setpvn(tmpstr,d+1,s-d);
11418 sv_catpvn(herewas,s,bufend-s);
11419 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11426 while (s < PL_bufend &&
11427 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11429 CopLINE_inc(PL_curcop);
11431 if (s >= PL_bufend) {
11432 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11433 missingterm(PL_tokenbuf);
11435 sv_setpvn(tmpstr,d+1,s-d);
11437 if (PL_madskills) {
11439 sv_catpvn(PL_thisstuff, d + 1, s - d);
11441 PL_thisstuff = newSVpvn(d + 1, s - d);
11442 stuffstart = s - SvPVX(PL_linestr);
11446 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11448 sv_catpvn(herewas,s,PL_bufend-s);
11449 sv_setsv(PL_linestr,herewas);
11450 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11451 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11452 PL_last_lop = PL_last_uni = NULL;
11455 sv_setpvs(tmpstr,""); /* avoid "uninitialized" warning */
11456 while (s >= PL_bufend) { /* multiple line string? */
11458 if (PL_madskills) {
11459 tstart = SvPVX(PL_linestr) + stuffstart;
11461 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11463 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11467 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11468 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11469 missingterm(PL_tokenbuf);
11472 stuffstart = s - SvPVX(PL_linestr);
11474 CopLINE_inc(PL_curcop);
11475 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11476 PL_last_lop = PL_last_uni = NULL;
11477 #ifndef PERL_STRICT_CR
11478 if (PL_bufend - PL_linestart >= 2) {
11479 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11480 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11482 PL_bufend[-2] = '\n';
11484 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11486 else if (PL_bufend[-1] == '\r')
11487 PL_bufend[-1] = '\n';
11489 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11490 PL_bufend[-1] = '\n';
11492 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11493 update_debugger_info(PL_linestr, NULL, 0);
11494 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11495 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11496 *(SvPVX(PL_linestr) + off ) = ' ';
11497 sv_catsv(PL_linestr,herewas);
11498 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11499 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11503 sv_catsv(tmpstr,PL_linestr);
11508 PL_multi_end = CopLINE(PL_curcop);
11509 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11510 SvPV_shrink_to_cur(tmpstr);
11512 SvREFCNT_dec(herewas);
11514 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11516 else if (PL_encoding)
11517 sv_recode_to_utf8(tmpstr, PL_encoding);
11519 PL_lex_stuff = tmpstr;
11520 pl_yylval.ival = op_type;
11524 /* scan_inputsymbol
11525 takes: current position in input buffer
11526 returns: new position in input buffer
11527 side-effects: pl_yylval and lex_op are set.
11532 <FH> read from filehandle
11533 <pkg::FH> read from package qualified filehandle
11534 <pkg'FH> read from package qualified filehandle
11535 <$fh> read from filehandle in $fh
11536 <*.h> filename glob
11541 S_scan_inputsymbol(pTHX_ char *start)
11544 register char *s = start; /* current position in buffer */
11547 char *d = PL_tokenbuf; /* start of temp holding space */
11548 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11550 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11552 end = strchr(s, '\n');
11555 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11557 /* die if we didn't have space for the contents of the <>,
11558 or if it didn't end, or if we see a newline
11561 if (len >= (I32)sizeof PL_tokenbuf)
11562 Perl_croak(aTHX_ "Excessively long <> operator");
11564 Perl_croak(aTHX_ "Unterminated <> operator");
11569 Remember, only scalar variables are interpreted as filehandles by
11570 this code. Anything more complex (e.g., <$fh{$num}>) will be
11571 treated as a glob() call.
11572 This code makes use of the fact that except for the $ at the front,
11573 a scalar variable and a filehandle look the same.
11575 if (*d == '$' && d[1]) d++;
11577 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11578 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11581 /* If we've tried to read what we allow filehandles to look like, and
11582 there's still text left, then it must be a glob() and not a getline.
11583 Use scan_str to pull out the stuff between the <> and treat it
11584 as nothing more than a string.
11587 if (d - PL_tokenbuf != len) {
11588 pl_yylval.ival = OP_GLOB;
11589 s = scan_str(start,!!PL_madskills,FALSE);
11591 Perl_croak(aTHX_ "Glob not terminated");
11595 bool readline_overriden = FALSE;
11598 /* we're in a filehandle read situation */
11601 /* turn <> into <ARGV> */
11603 Copy("ARGV",d,5,char);
11605 /* Check whether readline() is overriden */
11606 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11608 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11610 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11611 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11612 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11613 readline_overriden = TRUE;
11615 /* if <$fh>, create the ops to turn the variable into a
11619 /* try to find it in the pad for this block, otherwise find
11620 add symbol table ops
11622 const PADOFFSET tmp = pad_findmy(d);
11623 if (tmp != NOT_IN_PAD) {
11624 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11625 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11626 HEK * const stashname = HvNAME_HEK(stash);
11627 SV * const sym = sv_2mortal(newSVhek(stashname));
11628 sv_catpvs(sym, "::");
11629 sv_catpv(sym, d+1);
11634 OP * const o = newOP(OP_PADSV, 0);
11636 PL_lex_op = readline_overriden
11637 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11638 append_elem(OP_LIST, o,
11639 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11640 : (OP*)newUNOP(OP_READLINE, 0, o);
11649 ? (GV_ADDMULTI | GV_ADDINEVAL)
11652 PL_lex_op = readline_overriden
11653 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11654 append_elem(OP_LIST,
11655 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11656 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11657 : (OP*)newUNOP(OP_READLINE, 0,
11658 newUNOP(OP_RV2SV, 0,
11659 newGVOP(OP_GV, 0, gv)));
11661 if (!readline_overriden)
11662 PL_lex_op->op_flags |= OPf_SPECIAL;
11663 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11664 pl_yylval.ival = OP_NULL;
11667 /* If it's none of the above, it must be a literal filehandle
11668 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11670 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11671 PL_lex_op = readline_overriden
11672 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11673 append_elem(OP_LIST,
11674 newGVOP(OP_GV, 0, gv),
11675 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11676 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11677 pl_yylval.ival = OP_NULL;
11686 takes: start position in buffer
11687 keep_quoted preserve \ on the embedded delimiter(s)
11688 keep_delims preserve the delimiters around the string
11689 returns: position to continue reading from buffer
11690 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11691 updates the read buffer.
11693 This subroutine pulls a string out of the input. It is called for:
11694 q single quotes q(literal text)
11695 ' single quotes 'literal text'
11696 qq double quotes qq(interpolate $here please)
11697 " double quotes "interpolate $here please"
11698 qx backticks qx(/bin/ls -l)
11699 ` backticks `/bin/ls -l`
11700 qw quote words @EXPORT_OK = qw( func() $spam )
11701 m// regexp match m/this/
11702 s/// regexp substitute s/this/that/
11703 tr/// string transliterate tr/this/that/
11704 y/// string transliterate y/this/that/
11705 ($*@) sub prototypes sub foo ($)
11706 (stuff) sub attr parameters sub foo : attr(stuff)
11707 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11709 In most of these cases (all but <>, patterns and transliterate)
11710 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11711 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11712 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11715 It skips whitespace before the string starts, and treats the first
11716 character as the delimiter. If the delimiter is one of ([{< then
11717 the corresponding "close" character )]}> is used as the closing
11718 delimiter. It allows quoting of delimiters, and if the string has
11719 balanced delimiters ([{<>}]) it allows nesting.
11721 On success, the SV with the resulting string is put into lex_stuff or,
11722 if that is already non-NULL, into lex_repl. The second case occurs only
11723 when parsing the RHS of the special constructs s/// and tr/// (y///).
11724 For convenience, the terminating delimiter character is stuffed into
11729 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11732 SV *sv; /* scalar value: string */
11733 const char *tmps; /* temp string, used for delimiter matching */
11734 register char *s = start; /* current position in the buffer */
11735 register char term; /* terminating character */
11736 register char *to; /* current position in the sv's data */
11737 I32 brackets = 1; /* bracket nesting level */
11738 bool has_utf8 = FALSE; /* is there any utf8 content? */
11739 I32 termcode; /* terminating char. code */
11740 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11741 STRLEN termlen; /* length of terminating string */
11742 int last_off = 0; /* last position for nesting bracket */
11748 PERL_ARGS_ASSERT_SCAN_STR;
11750 /* skip space before the delimiter */
11756 if (PL_realtokenstart >= 0) {
11757 stuffstart = PL_realtokenstart;
11758 PL_realtokenstart = -1;
11761 stuffstart = start - SvPVX(PL_linestr);
11763 /* mark where we are, in case we need to report errors */
11766 /* after skipping whitespace, the next character is the terminator */
11769 termcode = termstr[0] = term;
11773 termcode = utf8_to_uvchr((U8*)s, &termlen);
11774 Copy(s, termstr, termlen, U8);
11775 if (!UTF8_IS_INVARIANT(term))
11779 /* mark where we are */
11780 PL_multi_start = CopLINE(PL_curcop);
11781 PL_multi_open = term;
11783 /* find corresponding closing delimiter */
11784 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11785 termcode = termstr[0] = term = tmps[5];
11787 PL_multi_close = term;
11789 /* create a new SV to hold the contents. 79 is the SV's initial length.
11790 What a random number. */
11791 sv = newSV_type(SVt_PVIV);
11793 SvIV_set(sv, termcode);
11794 (void)SvPOK_only(sv); /* validate pointer */
11796 /* move past delimiter and try to read a complete string */
11798 sv_catpvn(sv, s, termlen);
11801 tstart = SvPVX(PL_linestr) + stuffstart;
11802 if (!PL_thisopen && !keep_delims) {
11803 PL_thisopen = newSVpvn(tstart, s - tstart);
11804 stuffstart = s - SvPVX(PL_linestr);
11808 if (PL_encoding && !UTF) {
11812 int offset = s - SvPVX_const(PL_linestr);
11813 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11814 &offset, (char*)termstr, termlen);
11815 const char * const ns = SvPVX_const(PL_linestr) + offset;
11816 char * const svlast = SvEND(sv) - 1;
11818 for (; s < ns; s++) {
11819 if (*s == '\n' && !PL_rsfp)
11820 CopLINE_inc(PL_curcop);
11823 goto read_more_line;
11825 /* handle quoted delimiters */
11826 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11828 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11830 if ((svlast-1 - t) % 2) {
11831 if (!keep_quoted) {
11832 *(svlast-1) = term;
11834 SvCUR_set(sv, SvCUR(sv) - 1);
11839 if (PL_multi_open == PL_multi_close) {
11845 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11846 /* At here, all closes are "was quoted" one,
11847 so we don't check PL_multi_close. */
11849 if (!keep_quoted && *(t+1) == PL_multi_open)
11854 else if (*t == PL_multi_open)
11862 SvCUR_set(sv, w - SvPVX_const(sv));
11864 last_off = w - SvPVX(sv);
11865 if (--brackets <= 0)
11870 if (!keep_delims) {
11871 SvCUR_set(sv, SvCUR(sv) - 1);
11877 /* extend sv if need be */
11878 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11879 /* set 'to' to the next character in the sv's string */
11880 to = SvPVX(sv)+SvCUR(sv);
11882 /* if open delimiter is the close delimiter read unbridle */
11883 if (PL_multi_open == PL_multi_close) {
11884 for (; s < PL_bufend; s++,to++) {
11885 /* embedded newlines increment the current line number */
11886 if (*s == '\n' && !PL_rsfp)
11887 CopLINE_inc(PL_curcop);
11888 /* handle quoted delimiters */
11889 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11890 if (!keep_quoted && s[1] == term)
11892 /* any other quotes are simply copied straight through */
11896 /* terminate when run out of buffer (the for() condition), or
11897 have found the terminator */
11898 else if (*s == term) {
11901 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11904 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11910 /* if the terminator isn't the same as the start character (e.g.,
11911 matched brackets), we have to allow more in the quoting, and
11912 be prepared for nested brackets.
11915 /* read until we run out of string, or we find the terminator */
11916 for (; s < PL_bufend; s++,to++) {
11917 /* embedded newlines increment the line count */
11918 if (*s == '\n' && !PL_rsfp)
11919 CopLINE_inc(PL_curcop);
11920 /* backslashes can escape the open or closing characters */
11921 if (*s == '\\' && s+1 < PL_bufend) {
11922 if (!keep_quoted &&
11923 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11928 /* allow nested opens and closes */
11929 else if (*s == PL_multi_close && --brackets <= 0)
11931 else if (*s == PL_multi_open)
11933 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11938 /* terminate the copied string and update the sv's end-of-string */
11940 SvCUR_set(sv, to - SvPVX_const(sv));
11943 * this next chunk reads more into the buffer if we're not done yet
11947 break; /* handle case where we are done yet :-) */
11949 #ifndef PERL_STRICT_CR
11950 if (to - SvPVX_const(sv) >= 2) {
11951 if ((to[-2] == '\r' && to[-1] == '\n') ||
11952 (to[-2] == '\n' && to[-1] == '\r'))
11956 SvCUR_set(sv, to - SvPVX_const(sv));
11958 else if (to[-1] == '\r')
11961 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11966 /* if we're out of file, or a read fails, bail and reset the current
11967 line marker so we can report where the unterminated string began
11970 if (PL_madskills) {
11971 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11973 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11975 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11979 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11981 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11987 /* we read a line, so increment our line counter */
11988 CopLINE_inc(PL_curcop);
11990 /* update debugger info */
11991 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11992 update_debugger_info(PL_linestr, NULL, 0);
11994 /* having changed the buffer, we must update PL_bufend */
11995 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11996 PL_last_lop = PL_last_uni = NULL;
11999 /* at this point, we have successfully read the delimited string */
12001 if (!PL_encoding || UTF) {
12003 if (PL_madskills) {
12004 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12005 const int len = s - tstart;
12007 sv_catpvn(PL_thisstuff, tstart, len);
12009 PL_thisstuff = newSVpvn(tstart, len);
12010 if (!PL_thisclose && !keep_delims)
12011 PL_thisclose = newSVpvn(s,termlen);
12016 sv_catpvn(sv, s, termlen);
12021 if (PL_madskills) {
12022 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12023 const int len = s - tstart - termlen;
12025 sv_catpvn(PL_thisstuff, tstart, len);
12027 PL_thisstuff = newSVpvn(tstart, len);
12028 if (!PL_thisclose && !keep_delims)
12029 PL_thisclose = newSVpvn(s - termlen,termlen);
12033 if (has_utf8 || PL_encoding)
12036 PL_multi_end = CopLINE(PL_curcop);
12038 /* if we allocated too much space, give some back */
12039 if (SvCUR(sv) + 5 < SvLEN(sv)) {
12040 SvLEN_set(sv, SvCUR(sv) + 1);
12041 SvPV_renew(sv, SvLEN(sv));
12044 /* decide whether this is the first or second quoted string we've read
12057 takes: pointer to position in buffer
12058 returns: pointer to new position in buffer
12059 side-effects: builds ops for the constant in pl_yylval.op
12061 Read a number in any of the formats that Perl accepts:
12063 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12064 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12067 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12069 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12072 If it reads a number without a decimal point or an exponent, it will
12073 try converting the number to an integer and see if it can do so
12074 without loss of precision.
12078 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12081 register const char *s = start; /* current position in buffer */
12082 register char *d; /* destination in temp buffer */
12083 register char *e; /* end of temp buffer */
12084 NV nv; /* number read, as a double */
12085 SV *sv = NULL; /* place to put the converted number */
12086 bool floatit; /* boolean: int or float? */
12087 const char *lastub = NULL; /* position of last underbar */
12088 static char const number_too_long[] = "Number too long";
12090 PERL_ARGS_ASSERT_SCAN_NUM;
12092 /* We use the first character to decide what type of number this is */
12096 Perl_croak(aTHX_ "panic: scan_num");
12098 /* if it starts with a 0, it could be an octal number, a decimal in
12099 0.13 disguise, or a hexadecimal number, or a binary number. */
12103 u holds the "number so far"
12104 shift the power of 2 of the base
12105 (hex == 4, octal == 3, binary == 1)
12106 overflowed was the number more than we can hold?
12108 Shift is used when we add a digit. It also serves as an "are
12109 we in octal/hex/binary?" indicator to disallow hex characters
12110 when in octal mode.
12115 bool overflowed = FALSE;
12116 bool just_zero = TRUE; /* just plain 0 or binary number? */
12117 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12118 static const char* const bases[5] =
12119 { "", "binary", "", "octal", "hexadecimal" };
12120 static const char* const Bases[5] =
12121 { "", "Binary", "", "Octal", "Hexadecimal" };
12122 static const char* const maxima[5] =
12124 "0b11111111111111111111111111111111",
12128 const char *base, *Base, *max;
12130 /* check for hex */
12135 } else if (s[1] == 'b') {
12140 /* check for a decimal in disguise */
12141 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12143 /* so it must be octal */
12150 if (ckWARN(WARN_SYNTAX))
12151 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12152 "Misplaced _ in number");
12156 base = bases[shift];
12157 Base = Bases[shift];
12158 max = maxima[shift];
12160 /* read the rest of the number */
12162 /* x is used in the overflow test,
12163 b is the digit we're adding on. */
12168 /* if we don't mention it, we're done */
12172 /* _ are ignored -- but warned about if consecutive */
12174 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12175 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12176 "Misplaced _ in number");
12180 /* 8 and 9 are not octal */
12181 case '8': case '9':
12183 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12187 case '2': case '3': case '4':
12188 case '5': case '6': case '7':
12190 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12193 case '0': case '1':
12194 b = *s++ & 15; /* ASCII digit -> value of digit */
12198 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12199 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12200 /* make sure they said 0x */
12203 b = (*s++ & 7) + 9;
12205 /* Prepare to put the digit we have onto the end
12206 of the number so far. We check for overflows.
12212 x = u << shift; /* make room for the digit */
12214 if ((x >> shift) != u
12215 && !(PL_hints & HINT_NEW_BINARY)) {
12218 if (ckWARN_d(WARN_OVERFLOW))
12219 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12220 "Integer overflow in %s number",
12223 u = x | b; /* add the digit to the end */
12226 n *= nvshift[shift];
12227 /* If an NV has not enough bits in its
12228 * mantissa to represent an UV this summing of
12229 * small low-order numbers is a waste of time
12230 * (because the NV cannot preserve the
12231 * low-order bits anyway): we could just
12232 * remember when did we overflow and in the
12233 * end just multiply n by the right
12241 /* if we get here, we had success: make a scalar value from
12246 /* final misplaced underbar check */
12247 if (s[-1] == '_') {
12248 if (ckWARN(WARN_SYNTAX))
12249 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12254 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12255 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12256 "%s number > %s non-portable",
12262 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12263 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12264 "%s number > %s non-portable",
12269 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12270 sv = new_constant(start, s - start, "integer",
12271 sv, NULL, NULL, 0);
12272 else if (PL_hints & HINT_NEW_BINARY)
12273 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12278 handle decimal numbers.
12279 we're also sent here when we read a 0 as the first digit
12281 case '1': case '2': case '3': case '4': case '5':
12282 case '6': case '7': case '8': case '9': case '.':
12285 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12288 /* read next group of digits and _ and copy into d */
12289 while (isDIGIT(*s) || *s == '_') {
12290 /* skip underscores, checking for misplaced ones
12294 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12295 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12296 "Misplaced _ in number");
12300 /* check for end of fixed-length buffer */
12302 Perl_croak(aTHX_ number_too_long);
12303 /* if we're ok, copy the character */
12308 /* final misplaced underbar check */
12309 if (lastub && s == lastub + 1) {
12310 if (ckWARN(WARN_SYNTAX))
12311 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12314 /* read a decimal portion if there is one. avoid
12315 3..5 being interpreted as the number 3. followed
12318 if (*s == '.' && s[1] != '.') {
12323 if (ckWARN(WARN_SYNTAX))
12324 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12325 "Misplaced _ in number");
12329 /* copy, ignoring underbars, until we run out of digits.
12331 for (; isDIGIT(*s) || *s == '_'; s++) {
12332 /* fixed length buffer check */
12334 Perl_croak(aTHX_ number_too_long);
12336 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12337 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12338 "Misplaced _ in number");
12344 /* fractional part ending in underbar? */
12345 if (s[-1] == '_') {
12346 if (ckWARN(WARN_SYNTAX))
12347 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12348 "Misplaced _ in number");
12350 if (*s == '.' && isDIGIT(s[1])) {
12351 /* oops, it's really a v-string, but without the "v" */
12357 /* read exponent part, if present */
12358 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12362 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12363 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12365 /* stray preinitial _ */
12367 if (ckWARN(WARN_SYNTAX))
12368 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12369 "Misplaced _ in number");
12373 /* allow positive or negative exponent */
12374 if (*s == '+' || *s == '-')
12377 /* stray initial _ */
12379 if (ckWARN(WARN_SYNTAX))
12380 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12381 "Misplaced _ in number");
12385 /* read digits of exponent */
12386 while (isDIGIT(*s) || *s == '_') {
12389 Perl_croak(aTHX_ number_too_long);
12393 if (((lastub && s == lastub + 1) ||
12394 (!isDIGIT(s[1]) && s[1] != '_'))
12395 && ckWARN(WARN_SYNTAX))
12396 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12397 "Misplaced _ in number");
12404 /* make an sv from the string */
12408 We try to do an integer conversion first if no characters
12409 indicating "float" have been found.
12414 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12416 if (flags == IS_NUMBER_IN_UV) {
12418 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12421 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12422 if (uv <= (UV) IV_MIN)
12423 sv_setiv(sv, -(IV)uv);
12430 /* terminate the string */
12432 nv = Atof(PL_tokenbuf);
12437 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12438 const char *const key = floatit ? "float" : "integer";
12439 const STRLEN keylen = floatit ? 5 : 7;
12440 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12441 key, keylen, sv, NULL, NULL, 0);
12445 /* if it starts with a v, it could be a v-string */
12448 sv = newSV(5); /* preallocate storage space */
12449 s = scan_vstring(s, PL_bufend, sv);
12453 /* make the op for the constant and return */
12456 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12458 lvalp->opval = NULL;
12464 S_scan_formline(pTHX_ register char *s)
12467 register char *eol;
12469 SV * const stuff = newSVpvs("");
12470 bool needargs = FALSE;
12471 bool eofmt = FALSE;
12473 char *tokenstart = s;
12474 SV* savewhite = NULL;
12476 if (PL_madskills) {
12477 savewhite = PL_thiswhite;
12482 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12484 while (!needargs) {
12487 #ifdef PERL_STRICT_CR
12488 while (SPACE_OR_TAB(*t))
12491 while (SPACE_OR_TAB(*t) || *t == '\r')
12494 if (*t == '\n' || t == PL_bufend) {
12499 if (PL_in_eval && !PL_rsfp) {
12500 eol = (char *) memchr(s,'\n',PL_bufend-s);
12505 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12507 for (t = s; t < eol; t++) {
12508 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12510 goto enough; /* ~~ must be first line in formline */
12512 if (*t == '@' || *t == '^')
12516 sv_catpvn(stuff, s, eol-s);
12517 #ifndef PERL_STRICT_CR
12518 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12519 char *end = SvPVX(stuff) + SvCUR(stuff);
12522 SvCUR_set(stuff, SvCUR(stuff) - 1);
12532 if (PL_madskills) {
12534 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12536 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12539 s = filter_gets(PL_linestr, PL_rsfp, 0);
12541 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12543 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12545 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12546 PL_last_lop = PL_last_uni = NULL;
12555 if (SvCUR(stuff)) {
12558 PL_lex_state = LEX_NORMAL;
12559 start_force(PL_curforce);
12560 NEXTVAL_NEXTTOKE.ival = 0;
12564 PL_lex_state = LEX_FORMLINE;
12566 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12568 else if (PL_encoding)
12569 sv_recode_to_utf8(stuff, PL_encoding);
12571 start_force(PL_curforce);
12572 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12574 start_force(PL_curforce);
12575 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12579 SvREFCNT_dec(stuff);
12581 PL_lex_formbrack = 0;
12585 if (PL_madskills) {
12587 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12589 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12590 PL_thiswhite = savewhite;
12597 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12600 const I32 oldsavestack_ix = PL_savestack_ix;
12601 CV* const outsidecv = PL_compcv;
12604 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12606 SAVEI32(PL_subline);
12607 save_item(PL_subname);
12608 SAVESPTR(PL_compcv);
12610 PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
12611 CvFLAGS(PL_compcv) |= flags;
12613 PL_subline = CopLINE(PL_curcop);
12614 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12615 CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
12616 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12618 return oldsavestack_ix;
12622 #pragma segment Perl_yylex
12625 S_yywarn(pTHX_ const char *const s)
12629 PERL_ARGS_ASSERT_YYWARN;
12631 PL_in_eval |= EVAL_WARNONLY;
12633 PL_in_eval &= ~EVAL_WARNONLY;
12638 Perl_yyerror(pTHX_ const char *const s)
12641 const char *where = NULL;
12642 const char *context = NULL;
12645 int yychar = PL_parser->yychar;
12647 PERL_ARGS_ASSERT_YYERROR;
12649 if (!yychar || (yychar == ';' && !PL_rsfp))
12651 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12652 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12653 PL_oldbufptr != PL_bufptr) {
12656 The code below is removed for NetWare because it abends/crashes on NetWare
12657 when the script has error such as not having the closing quotes like:
12658 if ($var eq "value)
12659 Checking of white spaces is anyway done in NetWare code.
12662 while (isSPACE(*PL_oldoldbufptr))
12665 context = PL_oldoldbufptr;
12666 contlen = PL_bufptr - PL_oldoldbufptr;
12668 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12669 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12672 The code below is removed for NetWare because it abends/crashes on NetWare
12673 when the script has error such as not having the closing quotes like:
12674 if ($var eq "value)
12675 Checking of white spaces is anyway done in NetWare code.
12678 while (isSPACE(*PL_oldbufptr))
12681 context = PL_oldbufptr;
12682 contlen = PL_bufptr - PL_oldbufptr;
12684 else if (yychar > 255)
12685 where = "next token ???";
12686 else if (yychar == -2) { /* YYEMPTY */
12687 if (PL_lex_state == LEX_NORMAL ||
12688 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12689 where = "at end of line";
12690 else if (PL_lex_inpat)
12691 where = "within pattern";
12693 where = "within string";
12696 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12698 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12699 else if (isPRINT_LC(yychar)) {
12700 const char string = yychar;
12701 sv_catpvn(where_sv, &string, 1);
12704 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12705 where = SvPVX_const(where_sv);
12707 msg = sv_2mortal(newSVpv(s, 0));
12708 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12709 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12711 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12713 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12714 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12715 Perl_sv_catpvf(aTHX_ msg,
12716 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12717 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12720 if (PL_in_eval & EVAL_WARNONLY) {
12721 if (ckWARN_d(WARN_SYNTAX))
12722 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12726 if (PL_error_count >= 10) {
12727 if (PL_in_eval && SvCUR(ERRSV))
12728 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12729 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12731 Perl_croak(aTHX_ "%s has too many errors.\n",
12732 OutCopFILE(PL_curcop));
12735 PL_in_my_stash = NULL;
12739 #pragma segment Main
12743 S_swallow_bom(pTHX_ U8 *s)
12746 const STRLEN slen = SvCUR(PL_linestr);
12748 PERL_ARGS_ASSERT_SWALLOW_BOM;
12752 if (s[1] == 0xFE) {
12753 /* UTF-16 little-endian? (or UTF32-LE?) */
12754 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12755 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12756 #ifndef PERL_NO_UTF16_FILTER
12757 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12760 if (PL_bufend > (char*)s) {
12764 filter_add(utf16rev_textfilter, NULL);
12765 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12766 utf16_to_utf8_reversed(s, news,
12767 PL_bufend - (char*)s - 1,
12769 sv_setpvn(PL_linestr, (const char*)news, newlen);
12771 s = (U8*)SvPVX(PL_linestr);
12772 Copy(news, s, newlen, U8);
12776 SvUTF8_on(PL_linestr);
12777 s = (U8*)SvPVX(PL_linestr);
12779 /* FIXME - is this a general bug fix? */
12782 PL_bufend = SvPVX(PL_linestr) + newlen;
12785 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12790 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12791 #ifndef PERL_NO_UTF16_FILTER
12792 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12795 if (PL_bufend > (char *)s) {
12799 filter_add(utf16_textfilter, NULL);
12800 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12801 utf16_to_utf8(s, news,
12802 PL_bufend - (char*)s,
12804 sv_setpvn(PL_linestr, (const char*)news, newlen);
12806 SvUTF8_on(PL_linestr);
12807 s = (U8*)SvPVX(PL_linestr);
12808 PL_bufend = SvPVX(PL_linestr) + newlen;
12811 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12816 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12817 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12818 s += 3; /* UTF-8 */
12824 if (s[2] == 0xFE && s[3] == 0xFF) {
12825 /* UTF-32 big-endian */
12826 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12829 else if (s[2] == 0 && s[3] != 0) {
12832 * are a good indicator of UTF-16BE. */
12833 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12839 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12840 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12841 s += 4; /* UTF-8 */
12847 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12850 * are a good indicator of UTF-16LE. */
12851 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12859 #ifndef PERL_NO_UTF16_FILTER
12861 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12864 const STRLEN old = SvCUR(sv);
12865 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12866 DEBUG_P(PerlIO_printf(Perl_debug_log,
12867 "utf16_textfilter(%p): %d %d (%d)\n",
12868 FPTR2DPTR(void *, utf16_textfilter),
12869 idx, maxlen, (int) count));
12873 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12874 Copy(SvPVX_const(sv), tmps, old, char);
12875 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12876 SvCUR(sv) - old, &newlen);
12877 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12879 DEBUG_P({sv_dump(sv);});
12884 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12887 const STRLEN old = SvCUR(sv);
12888 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12889 DEBUG_P(PerlIO_printf(Perl_debug_log,
12890 "utf16rev_textfilter(%p): %d %d (%d)\n",
12891 FPTR2DPTR(void *, utf16rev_textfilter),
12892 idx, maxlen, (int) count));
12896 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12897 Copy(SvPVX_const(sv), tmps, old, char);
12898 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12899 SvCUR(sv) - old, &newlen);
12900 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12902 DEBUG_P({ sv_dump(sv); });
12908 Returns a pointer to the next character after the parsed
12909 vstring, as well as updating the passed in sv.
12911 Function must be called like
12914 s = scan_vstring(s,e,sv);
12916 where s and e are the start and end of the string.
12917 The sv should already be large enough to store the vstring
12918 passed in, for performance reasons.
12923 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12926 const char *pos = s;
12927 const char *start = s;
12929 PERL_ARGS_ASSERT_SCAN_VSTRING;
12931 if (*pos == 'v') pos++; /* get past 'v' */
12932 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12934 if ( *pos != '.') {
12935 /* this may not be a v-string if followed by => */
12936 const char *next = pos;
12937 while (next < e && isSPACE(*next))
12939 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12940 /* return string not v-string */
12941 sv_setpvn(sv,(char *)s,pos-s);
12942 return (char *)pos;
12946 if (!isALPHA(*pos)) {
12947 U8 tmpbuf[UTF8_MAXBYTES+1];
12950 s++; /* get past 'v' */
12955 /* this is atoi() that tolerates underscores */
12958 const char *end = pos;
12960 while (--end >= s) {
12962 const UV orev = rev;
12963 rev += (*end - '0') * mult;
12965 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12966 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12967 "Integer overflow in decimal number");
12971 if (rev > 0x7FFFFFFF)
12972 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12974 /* Append native character for the rev point */
12975 tmpend = uvchr_to_utf8(tmpbuf, rev);
12976 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12977 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12979 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12985 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12989 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12997 * c-indentation-style: bsd
12998 * c-basic-offset: 4
12999 * indent-tabs-mode: t
13002 * ex: set ts=8 sts=4 sw=4 noet: