3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define new_constant(a,b,c,d,e,f,g) \
27 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
29 #define pl_yylval (PL_parser->yylval)
31 /* YYINITDEPTH -- initial size of the parser's stacks. */
32 #define YYINITDEPTH 200
34 /* XXX temporary backwards compatibility */
35 #define PL_lex_brackets (PL_parser->lex_brackets)
36 #define PL_lex_brackstack (PL_parser->lex_brackstack)
37 #define PL_lex_casemods (PL_parser->lex_casemods)
38 #define PL_lex_casestack (PL_parser->lex_casestack)
39 #define PL_lex_defer (PL_parser->lex_defer)
40 #define PL_lex_dojoin (PL_parser->lex_dojoin)
41 #define PL_lex_expect (PL_parser->lex_expect)
42 #define PL_lex_formbrack (PL_parser->lex_formbrack)
43 #define PL_lex_inpat (PL_parser->lex_inpat)
44 #define PL_lex_inwhat (PL_parser->lex_inwhat)
45 #define PL_lex_op (PL_parser->lex_op)
46 #define PL_lex_repl (PL_parser->lex_repl)
47 #define PL_lex_starts (PL_parser->lex_starts)
48 #define PL_lex_stuff (PL_parser->lex_stuff)
49 #define PL_multi_start (PL_parser->multi_start)
50 #define PL_multi_open (PL_parser->multi_open)
51 #define PL_multi_close (PL_parser->multi_close)
52 #define PL_pending_ident (PL_parser->pending_ident)
53 #define PL_preambled (PL_parser->preambled)
54 #define PL_sublex_info (PL_parser->sublex_info)
55 #define PL_linestr (PL_parser->linestr)
56 #define PL_expect (PL_parser->expect)
57 #define PL_copline (PL_parser->copline)
58 #define PL_bufptr (PL_parser->bufptr)
59 #define PL_oldbufptr (PL_parser->oldbufptr)
60 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
61 #define PL_linestart (PL_parser->linestart)
62 #define PL_bufend (PL_parser->bufend)
63 #define PL_last_uni (PL_parser->last_uni)
64 #define PL_last_lop (PL_parser->last_lop)
65 #define PL_last_lop_op (PL_parser->last_lop_op)
66 #define PL_lex_state (PL_parser->lex_state)
67 #define PL_rsfp (PL_parser->rsfp)
68 #define PL_rsfp_filters (PL_parser->rsfp_filters)
69 #define PL_in_my (PL_parser->in_my)
70 #define PL_in_my_stash (PL_parser->in_my_stash)
71 #define PL_tokenbuf (PL_parser->tokenbuf)
72 #define PL_multi_end (PL_parser->multi_end)
73 #define PL_error_count (PL_parser->error_count)
76 # define PL_endwhite (PL_parser->endwhite)
77 # define PL_faketokens (PL_parser->faketokens)
78 # define PL_lasttoke (PL_parser->lasttoke)
79 # define PL_nextwhite (PL_parser->nextwhite)
80 # define PL_realtokenstart (PL_parser->realtokenstart)
81 # define PL_skipwhite (PL_parser->skipwhite)
82 # define PL_thisclose (PL_parser->thisclose)
83 # define PL_thismad (PL_parser->thismad)
84 # define PL_thisopen (PL_parser->thisopen)
85 # define PL_thisstuff (PL_parser->thisstuff)
86 # define PL_thistoken (PL_parser->thistoken)
87 # define PL_thiswhite (PL_parser->thiswhite)
88 # define PL_thiswhite (PL_parser->thiswhite)
89 # define PL_nexttoke (PL_parser->nexttoke)
90 # define PL_curforce (PL_parser->curforce)
92 # define PL_nexttoke (PL_parser->nexttoke)
93 # define PL_nexttype (PL_parser->nexttype)
94 # define PL_nextval (PL_parser->nextval)
98 S_pending_ident(pTHX);
100 static const char ident_too_long[] = "Identifier too long";
101 static const char commaless_variable_list[] = "comma-less variable list";
103 #ifndef PERL_NO_UTF16_FILTER
104 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
105 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
109 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
110 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
112 # define CURMAD(slot,sv)
113 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
116 #define XFAKEBRACK 128
117 #define XENUMMASK 127
119 #ifdef USE_UTF8_SCRIPTS
120 # define UTF (!IN_BYTES)
122 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
125 /* In variables named $^X, these are the legal values for X.
126 * 1999-02-27 mjd-perl-patch@plover.com */
127 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
129 /* On MacOS, respect nonbreaking spaces */
130 #ifdef MACOS_TRADITIONAL
131 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
133 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
136 /* LEX_* are values for PL_lex_state, the state of the lexer.
137 * They are arranged oddly so that the guard on the switch statement
138 * can get by with a single comparison (if the compiler is smart enough).
141 /* #define LEX_NOTPARSING 11 is done in perl.h. */
143 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
144 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$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');
549 iscntrl(PL_multi_close)
551 PL_multi_close < 32 || PL_multi_close == 127
555 tmpbuf[1] = (char)toCTRL(PL_multi_close);
560 *tmpbuf = (char)PL_multi_close;
564 q = strchr(s,'"') ? '\'' : '"';
565 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
568 #define FEATURE_IS_ENABLED(name) \
569 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
570 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
571 /* The longest string we pass in. */
572 #define MAX_FEATURE_LEN (sizeof("switch")-1)
575 * S_feature_is_enabled
576 * Check whether the named feature is enabled.
579 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
582 HV * const hinthv = GvHV(PL_hintgv);
583 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
585 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
587 assert(namelen <= MAX_FEATURE_LEN);
588 memcpy(&he_name[8], name, namelen);
590 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
598 Perl_deprecate(pTHX_ const char *const s)
600 PERL_ARGS_ASSERT_DEPRECATE;
602 if (ckWARN(WARN_DEPRECATED))
603 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
607 Perl_deprecate_old(pTHX_ const char *const s)
609 /* This function should NOT be called for any new deprecated warnings */
610 /* Use Perl_deprecate instead */
612 /* It is here to maintain backward compatibility with the pre-5.8 */
613 /* warnings category hierarchy. The "deprecated" category used to */
614 /* live under the "syntax" category. It is now a top-level category */
615 /* in its own right. */
617 PERL_ARGS_ASSERT_DEPRECATE_OLD;
619 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
620 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
621 "Use of %s is deprecated", s);
625 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
626 * utf16-to-utf8-reversed.
629 #ifdef PERL_CR_FILTER
633 register const char *s = SvPVX_const(sv);
634 register const char * const e = s + SvCUR(sv);
636 PERL_ARGS_ASSERT_STRIP_RETURN;
638 /* outer loop optimized to do nothing if there are no CR-LFs */
640 if (*s++ == '\r' && *s == '\n') {
641 /* hit a CR-LF, need to copy the rest */
642 register char *d = s - 1;
645 if (*s == '\r' && s[1] == '\n')
656 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
658 const I32 count = FILTER_READ(idx+1, sv, maxlen);
659 if (count > 0 && !maxlen)
670 * Create a parser object and initialise its parser and lexer fields
672 * rsfp is the opened file handle to read from (if any),
674 * line holds any initial content already read from the file (or in
675 * the case of no file, such as an eval, the whole contents);
677 * new_filter indicates that this is a new file and it shouldn't inherit
678 * the filters from the current parser (ie require).
682 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
685 const char *s = NULL;
687 yy_parser *parser, *oparser;
689 /* create and initialise a parser */
691 Newxz(parser, 1, yy_parser);
692 parser->old_parser = oparser = PL_parser;
695 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
696 parser->ps = parser->stack;
697 parser->stack_size = YYINITDEPTH;
699 parser->stack->state = 0;
700 parser->yyerrstatus = 0;
701 parser->yychar = YYEMPTY; /* Cause a token to be read. */
703 /* on scope exit, free this parser and restore any outer one */
705 parser->saved_curcop = PL_curcop;
707 /* initialise lexer state */
710 parser->curforce = -1;
712 parser->nexttoke = 0;
714 parser->error_count = oparser ? oparser->error_count : 0;
715 parser->copline = NOLINE;
716 parser->lex_state = LEX_NORMAL;
717 parser->expect = XSTATE;
719 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
720 : (AV*)SvREFCNT_inc(oparser->rsfp_filters);
722 Newx(parser->lex_brackstack, 120, char);
723 Newx(parser->lex_casestack, 12, char);
724 *parser->lex_casestack = '\0';
727 s = SvPV_const(line, len);
733 parser->linestr = newSVpvs("\n;");
734 } else if (SvREADONLY(line) || s[len-1] != ';') {
735 parser->linestr = newSVsv(line);
737 sv_catpvs(parser->linestr, "\n;");
740 SvREFCNT_inc_simple_void_NN(line);
741 parser->linestr = line;
743 parser->oldoldbufptr =
746 parser->linestart = SvPVX(parser->linestr);
747 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
748 parser->last_lop = parser->last_uni = NULL;
752 /* delete a parser object */
755 Perl_parser_free(pTHX_ const yy_parser *parser)
757 PERL_ARGS_ASSERT_PARSER_FREE;
759 PL_curcop = parser->saved_curcop;
760 SvREFCNT_dec(parser->linestr);
762 if (parser->rsfp == PerlIO_stdin())
763 PerlIO_clearerr(parser->rsfp);
764 else if (parser->rsfp && (!parser->old_parser ||
765 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
766 PerlIO_close(parser->rsfp);
767 SvREFCNT_dec(parser->rsfp_filters);
769 Safefree(parser->stack);
770 Safefree(parser->lex_brackstack);
771 Safefree(parser->lex_casestack);
772 PL_parser = parser->old_parser;
779 * Finalizer for lexing operations. Must be called when the parser is
780 * done with the lexer.
787 PL_doextract = FALSE;
792 * This subroutine has nothing to do with tilting, whether at windmills
793 * or pinball tables. Its name is short for "increment line". It
794 * increments the current line number in CopLINE(PL_curcop) and checks
795 * to see whether the line starts with a comment of the form
796 * # line 500 "foo.pm"
797 * If so, it sets the current line number and file to the values in the comment.
801 S_incline(pTHX_ const char *s)
808 PERL_ARGS_ASSERT_INCLINE;
810 CopLINE_inc(PL_curcop);
813 while (SPACE_OR_TAB(*s))
815 if (strnEQ(s, "line", 4))
819 if (SPACE_OR_TAB(*s))
823 while (SPACE_OR_TAB(*s))
831 while (SPACE_OR_TAB(*s))
833 if (*s == '"' && (t = strchr(s+1, '"'))) {
843 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
845 if (*e != '\n' && *e != '\0')
846 return; /* false alarm */
849 const STRLEN len = t - s;
851 SV *const temp_sv = CopFILESV(PL_curcop);
857 tmplen = SvCUR(temp_sv);
863 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
864 /* must copy *{"::_<(eval N)[oldfilename:L]"}
865 * to *{"::_<newfilename"} */
866 /* However, the long form of evals is only turned on by the
867 debugger - usually they're "(eval %lu)" */
871 STRLEN tmplen2 = len;
872 if (tmplen + 2 <= sizeof smallbuf)
875 Newx(tmpbuf, tmplen + 2, char);
878 memcpy(tmpbuf + 2, cf, tmplen);
880 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
885 if (tmplen2 + 2 <= sizeof smallbuf)
888 Newx(tmpbuf2, tmplen2 + 2, char);
890 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
891 /* Either they malloc'd it, or we malloc'd it,
892 so no prefix is present in ours. */
897 memcpy(tmpbuf2 + 2, s, tmplen2);
900 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
902 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
903 /* adjust ${"::_<newfilename"} to store the new file name */
904 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
905 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
906 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
909 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
911 if (tmpbuf != smallbuf) Safefree(tmpbuf);
914 CopFILE_free(PL_curcop);
915 CopFILE_setn(PL_curcop, s, len);
917 CopLINE_set(PL_curcop, atoi(n)-1);
921 /* skip space before PL_thistoken */
924 S_skipspace0(pTHX_ register char *s)
926 PERL_ARGS_ASSERT_SKIPSPACE0;
933 PL_thiswhite = newSVpvs("");
934 sv_catsv(PL_thiswhite, PL_skipwhite);
935 sv_free(PL_skipwhite);
938 PL_realtokenstart = s - SvPVX(PL_linestr);
942 /* skip space after PL_thistoken */
945 S_skipspace1(pTHX_ register char *s)
947 const char *start = s;
948 I32 startoff = start - SvPVX(PL_linestr);
950 PERL_ARGS_ASSERT_SKIPSPACE1;
955 start = SvPVX(PL_linestr) + startoff;
956 if (!PL_thistoken && PL_realtokenstart >= 0) {
957 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
958 PL_thistoken = newSVpvn(tstart, start - tstart);
960 PL_realtokenstart = -1;
963 PL_nextwhite = newSVpvs("");
964 sv_catsv(PL_nextwhite, PL_skipwhite);
965 sv_free(PL_skipwhite);
972 S_skipspace2(pTHX_ register char *s, SV **svp)
975 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
976 const I32 startoff = s - SvPVX(PL_linestr);
978 PERL_ARGS_ASSERT_SKIPSPACE2;
981 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
982 if (!PL_madskills || !svp)
984 start = SvPVX(PL_linestr) + startoff;
985 if (!PL_thistoken && PL_realtokenstart >= 0) {
986 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
987 PL_thistoken = newSVpvn(tstart, start - tstart);
988 PL_realtokenstart = -1;
993 sv_setsv(*svp, PL_skipwhite);
994 sv_free(PL_skipwhite);
1003 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1005 AV *av = CopFILEAVx(PL_curcop);
1007 SV * const sv = newSV_type(SVt_PVMG);
1009 sv_setsv(sv, orig_sv);
1011 sv_setpvn(sv, buf, len);
1014 av_store(av, (I32)CopLINE(PL_curcop), sv);
1020 * Called to gobble the appropriate amount and type of whitespace.
1021 * Skips comments as well.
1025 S_skipspace(pTHX_ register char *s)
1030 int startoff = s - SvPVX(PL_linestr);
1032 PERL_ARGS_ASSERT_SKIPSPACE;
1035 sv_free(PL_skipwhite);
1039 PERL_ARGS_ASSERT_SKIPSPACE;
1041 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1042 while (s < PL_bufend && SPACE_OR_TAB(*s))
1052 SSize_t oldprevlen, oldoldprevlen;
1053 SSize_t oldloplen = 0, oldunilen = 0;
1054 while (s < PL_bufend && isSPACE(*s)) {
1055 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1060 if (s < PL_bufend && *s == '#') {
1061 while (s < PL_bufend && *s != '\n')
1063 if (s < PL_bufend) {
1065 if (PL_in_eval && !PL_rsfp) {
1072 /* only continue to recharge the buffer if we're at the end
1073 * of the buffer, we're not reading from a source filter, and
1074 * we're in normal lexing mode
1076 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1077 PL_lex_state == LEX_FORMLINE)
1084 /* try to recharge the buffer */
1086 curoff = s - SvPVX(PL_linestr);
1089 if ((s = filter_gets(PL_linestr, PL_rsfp,
1090 (prevlen = SvCUR(PL_linestr)))) == NULL)
1093 if (PL_madskills && curoff != startoff) {
1095 PL_skipwhite = newSVpvs("");
1096 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1100 /* mustn't throw out old stuff yet if madpropping */
1101 SvCUR(PL_linestr) = curoff;
1102 s = SvPVX(PL_linestr) + curoff;
1104 if (curoff && s[-1] == '\n')
1108 /* end of file. Add on the -p or -n magic */
1109 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1112 sv_catpvs(PL_linestr,
1113 ";}continue{print or die qq(-p destination: $!\\n);}");
1115 sv_setpvs(PL_linestr,
1116 ";}continue{print or die qq(-p destination: $!\\n);}");
1118 PL_minus_n = PL_minus_p = 0;
1120 else if (PL_minus_n) {
1122 sv_catpvn(PL_linestr, ";}", 2);
1124 sv_setpvn(PL_linestr, ";}", 2);
1130 sv_catpvn(PL_linestr,";", 1);
1132 sv_setpvn(PL_linestr,";", 1);
1135 /* reset variables for next time we lex */
1136 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1142 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1143 PL_last_lop = PL_last_uni = NULL;
1145 /* Close the filehandle. Could be from
1146 * STDIN, or a regular file. If we were reading code from
1147 * STDIN (because the commandline held no -e or filename)
1148 * then we don't close it, we reset it so the code can
1149 * read from STDIN too.
1152 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1153 PerlIO_clearerr(PL_rsfp);
1155 (void)PerlIO_close(PL_rsfp);
1160 /* not at end of file, so we only read another line */
1161 /* make corresponding updates to old pointers, for yyerror() */
1162 oldprevlen = PL_oldbufptr - PL_bufend;
1163 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1165 oldunilen = PL_last_uni - PL_bufend;
1167 oldloplen = PL_last_lop - PL_bufend;
1168 PL_linestart = PL_bufptr = s + prevlen;
1169 PL_bufend = s + SvCUR(PL_linestr);
1171 PL_oldbufptr = s + oldprevlen;
1172 PL_oldoldbufptr = s + oldoldprevlen;
1174 PL_last_uni = s + oldunilen;
1176 PL_last_lop = s + oldloplen;
1179 /* debugger active and we're not compiling the debugger code,
1180 * so store the line into the debugger's array of lines
1182 if (PERLDB_LINE && PL_curstash != PL_debstash)
1183 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1190 PL_skipwhite = newSVpvs("");
1191 curoff = s - SvPVX(PL_linestr);
1192 if (curoff - startoff)
1193 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1202 * Check the unary operators to ensure there's no ambiguity in how they're
1203 * used. An ambiguous piece of code would be:
1205 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1206 * the +5 is its argument.
1216 if (PL_oldoldbufptr != PL_last_uni)
1218 while (isSPACE(*PL_last_uni))
1221 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1223 if ((t = strchr(s, '(')) && t < PL_bufptr)
1226 if (ckWARN_d(WARN_AMBIGUOUS)){
1227 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1228 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1229 (int)(s - PL_last_uni), PL_last_uni);
1234 * LOP : macro to build a list operator. Its behaviour has been replaced
1235 * with a subroutine, S_lop() for which LOP is just another name.
1238 #define LOP(f,x) return lop(f,x,s)
1242 * Build a list operator (or something that might be one). The rules:
1243 * - if we have a next token, then it's a list operator [why?]
1244 * - if the next thing is an opening paren, then it's a function
1245 * - else it's a list operator
1249 S_lop(pTHX_ I32 f, int x, char *s)
1253 PERL_ARGS_ASSERT_LOP;
1259 PL_last_lop = PL_oldbufptr;
1260 PL_last_lop_op = (OPCODE)f;
1263 return REPORT(LSTOP);
1266 return REPORT(LSTOP);
1269 return REPORT(FUNC);
1272 return REPORT(FUNC);
1274 return REPORT(LSTOP);
1280 * Sets up for an eventual force_next(). start_force(0) basically does
1281 * an unshift, while start_force(-1) does a push. yylex removes items
1286 S_start_force(pTHX_ int where)
1290 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1291 where = PL_lasttoke;
1292 assert(PL_curforce < 0 || PL_curforce == where);
1293 if (PL_curforce != where) {
1294 for (i = PL_lasttoke; i > where; --i) {
1295 PL_nexttoke[i] = PL_nexttoke[i-1];
1299 if (PL_curforce < 0) /* in case of duplicate start_force() */
1300 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1301 PL_curforce = where;
1304 curmad('^', newSVpvs(""));
1305 CURMAD('_', PL_nextwhite);
1310 S_curmad(pTHX_ char slot, SV *sv)
1316 if (PL_curforce < 0)
1317 where = &PL_thismad;
1319 where = &PL_nexttoke[PL_curforce].next_mad;
1322 sv_setpvn(sv, "", 0);
1325 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1327 else if (PL_encoding) {
1328 sv_recode_to_utf8(sv, PL_encoding);
1333 /* keep a slot open for the head of the list? */
1334 if (slot != '_' && *where && (*where)->mad_key == '^') {
1335 (*where)->mad_key = slot;
1336 sv_free((SV*)((*where)->mad_val));
1337 (*where)->mad_val = (void*)sv;
1340 addmad(newMADsv(slot, sv), where, 0);
1343 # define start_force(where) NOOP
1344 # define curmad(slot, sv) NOOP
1349 * When the lexer realizes it knows the next token (for instance,
1350 * it is reordering tokens for the parser) then it can call S_force_next
1351 * to know what token to return the next time the lexer is called. Caller
1352 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1353 * and possibly PL_expect to ensure the lexer handles the token correctly.
1357 S_force_next(pTHX_ I32 type)
1362 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1363 tokereport(type, &NEXTVAL_NEXTTOKE);
1367 if (PL_curforce < 0)
1368 start_force(PL_lasttoke);
1369 PL_nexttoke[PL_curforce].next_type = type;
1370 if (PL_lex_state != LEX_KNOWNEXT)
1371 PL_lex_defer = PL_lex_state;
1372 PL_lex_state = LEX_KNOWNEXT;
1373 PL_lex_expect = PL_expect;
1376 PL_nexttype[PL_nexttoke] = type;
1378 if (PL_lex_state != LEX_KNOWNEXT) {
1379 PL_lex_defer = PL_lex_state;
1380 PL_lex_expect = PL_expect;
1381 PL_lex_state = LEX_KNOWNEXT;
1387 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1390 SV * const sv = newSVpvn_utf8(start, len,
1392 && is_utf8_string((const U8*)start, len));
1398 * When the lexer knows the next thing is a word (for instance, it has
1399 * just seen -> and it knows that the next char is a word char, then
1400 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1404 * char *start : buffer position (must be within PL_linestr)
1405 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1406 * int check_keyword : if true, Perl checks to make sure the word isn't
1407 * a keyword (do this if the word is a label, e.g. goto FOO)
1408 * int allow_pack : if true, : characters will also be allowed (require,
1409 * use, etc. do this)
1410 * int allow_initial_tick : used by the "sub" lexer only.
1414 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1420 PERL_ARGS_ASSERT_FORCE_WORD;
1422 start = SKIPSPACE1(start);
1424 if (isIDFIRST_lazy_if(s,UTF) ||
1425 (allow_pack && *s == ':') ||
1426 (allow_initial_tick && *s == '\'') )
1428 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1429 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1431 start_force(PL_curforce);
1433 curmad('X', newSVpvn(start,s-start));
1434 if (token == METHOD) {
1439 PL_expect = XOPERATOR;
1443 curmad('g', newSVpvs( "forced" ));
1444 NEXTVAL_NEXTTOKE.opval
1445 = (OP*)newSVOP(OP_CONST,0,
1446 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1447 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1455 * Called when the lexer wants $foo *foo &foo etc, but the program
1456 * text only contains the "foo" portion. The first argument is a pointer
1457 * to the "foo", and the second argument is the type symbol to prefix.
1458 * Forces the next token to be a "WORD".
1459 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1463 S_force_ident(pTHX_ register const char *s, int kind)
1467 PERL_ARGS_ASSERT_FORCE_IDENT;
1470 const STRLEN len = strlen(s);
1471 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1472 start_force(PL_curforce);
1473 NEXTVAL_NEXTTOKE.opval = o;
1476 o->op_private = OPpCONST_ENTERED;
1477 /* XXX see note in pp_entereval() for why we forgo typo
1478 warnings if the symbol must be introduced in an eval.
1480 gv_fetchpvn_flags(s, len,
1481 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1483 kind == '$' ? SVt_PV :
1484 kind == '@' ? SVt_PVAV :
1485 kind == '%' ? SVt_PVHV :
1493 Perl_str_to_version(pTHX_ SV *sv)
1498 const char *start = SvPV_const(sv,len);
1499 const char * const end = start + len;
1500 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1502 PERL_ARGS_ASSERT_STR_TO_VERSION;
1504 while (start < end) {
1508 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1513 retval += ((NV)n)/nshift;
1522 * Forces the next token to be a version number.
1523 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1524 * and if "guessing" is TRUE, then no new token is created (and the caller
1525 * must use an alternative parsing method).
1529 S_force_version(pTHX_ char *s, int guessing)
1535 I32 startoff = s - SvPVX(PL_linestr);
1538 PERL_ARGS_ASSERT_FORCE_VERSION;
1546 while (isDIGIT(*d) || *d == '_' || *d == '.')
1550 start_force(PL_curforce);
1551 curmad('X', newSVpvn(s,d-s));
1554 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1556 s = scan_num(s, &pl_yylval);
1557 version = pl_yylval.opval;
1558 ver = cSVOPx(version)->op_sv;
1559 if (SvPOK(ver) && !SvNIOK(ver)) {
1560 SvUPGRADE(ver, SVt_PVNV);
1561 SvNV_set(ver, str_to_version(ver));
1562 SvNOK_on(ver); /* hint that it is a version */
1565 else if (guessing) {
1568 sv_free(PL_nextwhite); /* let next token collect whitespace */
1570 s = SvPVX(PL_linestr) + startoff;
1578 if (PL_madskills && !version) {
1579 sv_free(PL_nextwhite); /* let next token collect whitespace */
1581 s = SvPVX(PL_linestr) + startoff;
1584 /* NOTE: The parser sees the package name and the VERSION swapped */
1585 start_force(PL_curforce);
1586 NEXTVAL_NEXTTOKE.opval = version;
1594 * Tokenize a quoted string passed in as an SV. It finds the next
1595 * chunk, up to end of string or a backslash. It may make a new
1596 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1601 S_tokeq(pTHX_ SV *sv)
1605 register char *send;
1610 PERL_ARGS_ASSERT_TOKEQ;
1615 s = SvPV_force(sv, len);
1616 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1619 while (s < send && *s != '\\')
1624 if ( PL_hints & HINT_NEW_STRING ) {
1625 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
1629 if (s + 1 < send && (s[1] == '\\'))
1630 s++; /* all that, just for this */
1635 SvCUR_set(sv, d - SvPVX_const(sv));
1637 if ( PL_hints & HINT_NEW_STRING )
1638 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
1643 * Now come three functions related to double-quote context,
1644 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1645 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1646 * interact with PL_lex_state, and create fake ( ... ) argument lists
1647 * to handle functions and concatenation.
1648 * They assume that whoever calls them will be setting up a fake
1649 * join call, because each subthing puts a ',' after it. This lets
1652 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1654 * (I'm not sure whether the spurious commas at the end of lcfirst's
1655 * arguments and join's arguments are created or not).
1660 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1662 * Pattern matching will set PL_lex_op to the pattern-matching op to
1663 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
1665 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1667 * Everything else becomes a FUNC.
1669 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1670 * had an OP_CONST or OP_READLINE). This just sets us up for a
1671 * call to S_sublex_push().
1675 S_sublex_start(pTHX)
1678 register const I32 op_type = pl_yylval.ival;
1680 if (op_type == OP_NULL) {
1681 pl_yylval.opval = PL_lex_op;
1685 if (op_type == OP_CONST || op_type == OP_READLINE) {
1686 SV *sv = tokeq(PL_lex_stuff);
1688 if (SvTYPE(sv) == SVt_PVIV) {
1689 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1691 const char * const p = SvPV_const(sv, len);
1692 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
1696 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1697 PL_lex_stuff = NULL;
1698 /* Allow <FH> // "foo" */
1699 if (op_type == OP_READLINE)
1700 PL_expect = XTERMORDORDOR;
1703 else if (op_type == OP_BACKTICK && PL_lex_op) {
1704 /* readpipe() vas overriden */
1705 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
1706 pl_yylval.opval = PL_lex_op;
1708 PL_lex_stuff = NULL;
1712 PL_sublex_info.super_state = PL_lex_state;
1713 PL_sublex_info.sub_inwhat = (U16)op_type;
1714 PL_sublex_info.sub_op = PL_lex_op;
1715 PL_lex_state = LEX_INTERPPUSH;
1719 pl_yylval.opval = PL_lex_op;
1729 * Create a new scope to save the lexing state. The scope will be
1730 * ended in S_sublex_done. Returns a '(', starting the function arguments
1731 * to the uc, lc, etc. found before.
1732 * Sets PL_lex_state to LEX_INTERPCONCAT.
1741 PL_lex_state = PL_sublex_info.super_state;
1742 SAVEBOOL(PL_lex_dojoin);
1743 SAVEI32(PL_lex_brackets);
1744 SAVEI32(PL_lex_casemods);
1745 SAVEI32(PL_lex_starts);
1746 SAVEI8(PL_lex_state);
1747 SAVEVPTR(PL_lex_inpat);
1748 SAVEI16(PL_lex_inwhat);
1749 SAVECOPLINE(PL_curcop);
1750 SAVEPPTR(PL_bufptr);
1751 SAVEPPTR(PL_bufend);
1752 SAVEPPTR(PL_oldbufptr);
1753 SAVEPPTR(PL_oldoldbufptr);
1754 SAVEPPTR(PL_last_lop);
1755 SAVEPPTR(PL_last_uni);
1756 SAVEPPTR(PL_linestart);
1757 SAVESPTR(PL_linestr);
1758 SAVEGENERICPV(PL_lex_brackstack);
1759 SAVEGENERICPV(PL_lex_casestack);
1761 PL_linestr = PL_lex_stuff;
1762 PL_lex_stuff = NULL;
1764 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1765 = SvPVX(PL_linestr);
1766 PL_bufend += SvCUR(PL_linestr);
1767 PL_last_lop = PL_last_uni = NULL;
1768 SAVEFREESV(PL_linestr);
1770 PL_lex_dojoin = FALSE;
1771 PL_lex_brackets = 0;
1772 Newx(PL_lex_brackstack, 120, char);
1773 Newx(PL_lex_casestack, 12, char);
1774 PL_lex_casemods = 0;
1775 *PL_lex_casestack = '\0';
1777 PL_lex_state = LEX_INTERPCONCAT;
1778 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1780 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1781 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1782 PL_lex_inpat = PL_sublex_info.sub_op;
1784 PL_lex_inpat = NULL;
1791 * Restores lexer state after a S_sublex_push.
1798 if (!PL_lex_starts++) {
1799 SV * const sv = newSVpvs("");
1800 if (SvUTF8(PL_linestr))
1802 PL_expect = XOPERATOR;
1803 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1807 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1808 PL_lex_state = LEX_INTERPCASEMOD;
1812 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1813 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1814 PL_linestr = PL_lex_repl;
1816 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1817 PL_bufend += SvCUR(PL_linestr);
1818 PL_last_lop = PL_last_uni = NULL;
1819 SAVEFREESV(PL_linestr);
1820 PL_lex_dojoin = FALSE;
1821 PL_lex_brackets = 0;
1822 PL_lex_casemods = 0;
1823 *PL_lex_casestack = '\0';
1825 if (SvEVALED(PL_lex_repl)) {
1826 PL_lex_state = LEX_INTERPNORMAL;
1828 /* we don't clear PL_lex_repl here, so that we can check later
1829 whether this is an evalled subst; that means we rely on the
1830 logic to ensure sublex_done() is called again only via the
1831 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1834 PL_lex_state = LEX_INTERPCONCAT;
1844 PL_endwhite = newSVpvs("");
1845 sv_catsv(PL_endwhite, PL_thiswhite);
1849 sv_setpvn(PL_thistoken,"",0);
1851 PL_realtokenstart = -1;
1855 PL_bufend = SvPVX(PL_linestr);
1856 PL_bufend += SvCUR(PL_linestr);
1857 PL_expect = XOPERATOR;
1858 PL_sublex_info.sub_inwhat = 0;
1866 Extracts a pattern, double-quoted string, or transliteration. This
1869 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1870 processing a pattern (PL_lex_inpat is true), a transliteration
1871 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1873 Returns a pointer to the character scanned up to. If this is
1874 advanced from the start pointer supplied (i.e. if anything was
1875 successfully parsed), will leave an OP for the substring scanned
1876 in pl_yylval. Caller must intuit reason for not parsing further
1877 by looking at the next characters herself.
1881 double-quoted style: \r and \n
1882 regexp special ones: \D \s
1885 case and quoting: \U \Q \E
1886 stops on @ and $, but not for $ as tail anchor
1888 In transliterations:
1889 characters are VERY literal, except for - not at the start or end
1890 of the string, which indicates a range. If the range is in bytes,
1891 scan_const expands the range to the full set of intermediate
1892 characters. If the range is in utf8, the hyphen is replaced with
1893 a certain range mark which will be handled by pmtrans() in op.c.
1895 In double-quoted strings:
1897 double-quoted style: \r and \n
1899 deprecated backrefs: \1 (in substitution replacements)
1900 case and quoting: \U \Q \E
1903 scan_const does *not* construct ops to handle interpolated strings.
1904 It stops processing as soon as it finds an embedded $ or @ variable
1905 and leaves it to the caller to work out what's going on.
1907 embedded arrays (whether in pattern or not) could be:
1908 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1910 $ in double-quoted strings must be the symbol of an embedded scalar.
1912 $ in pattern could be $foo or could be tail anchor. Assumption:
1913 it's a tail anchor if $ is the last thing in the string, or if it's
1914 followed by one of "()| \r\n\t"
1916 \1 (backreferences) are turned into $1
1918 The structure of the code is
1919 while (there's a character to process) {
1920 handle transliteration ranges
1921 skip regexp comments /(?#comment)/ and codes /(?{code})/
1922 skip #-initiated comments in //x patterns
1923 check for embedded arrays
1924 check for embedded scalars
1926 leave intact backslashes from leaveit (below)
1927 deprecate \1 in substitution replacements
1928 handle string-changing backslashes \l \U \Q \E, etc.
1929 switch (what was escaped) {
1930 handle \- in a transliteration (becomes a literal -)
1931 handle \132 (octal characters)
1932 handle \x15 and \x{1234} (hex characters)
1933 handle \N{name} (named characters)
1934 handle \cV (control characters)
1935 handle printf-style backslashes (\f, \r, \n, etc)
1937 } (end if backslash)
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 */
1948 register char *s = start; /* start of the constant */
1949 register char *d = SvPVX(sv); /* destination for copies */
1950 bool dorange = FALSE; /* are we in a translit range? */
1951 bool didrange = FALSE; /* did we just finish a range? */
1952 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1953 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1956 UV literal_endpoint = 0;
1957 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1960 PERL_ARGS_ASSERT_SCAN_CONST;
1962 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1963 /* If we are doing a trans and we know we want UTF8 set expectation */
1964 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1965 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1969 while (s < send || dorange) {
1970 /* get transliterations out of the way (they're most literal) */
1971 if (PL_lex_inwhat == OP_TRANS) {
1972 /* expand a range A-Z to the full set of characters. AIE! */
1974 I32 i; /* current expanded character */
1975 I32 min; /* first character in range */
1976 I32 max; /* last character in range */
1987 char * const c = (char*)utf8_hop((U8*)d, -1);
1991 *c = (char)UTF_TO_NATIVE(0xff);
1992 /* mark the range as done, and continue */
1998 i = d - SvPVX_const(sv); /* remember current offset */
2001 SvLEN(sv) + (has_utf8 ?
2002 (512 - UTF_CONTINUATION_MARK +
2005 /* How many two-byte within 0..255: 128 in UTF-8,
2006 * 96 in UTF-8-mod. */
2008 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
2010 d = SvPVX(sv) + i; /* refresh d after realloc */
2014 for (j = 0; j <= 1; j++) {
2015 char * const c = (char*)utf8_hop((U8*)d, -1);
2016 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2022 max = (U8)0xff; /* only to \xff */
2023 uvmax = uv; /* \x{100} to uvmax */
2025 d = c; /* eat endpoint chars */
2030 d -= 2; /* eat the first char and the - */
2031 min = (U8)*d; /* first char in range */
2032 max = (U8)d[1]; /* last char in range */
2039 "Invalid range \"%c-%c\" in transliteration operator",
2040 (char)min, (char)max);
2044 if (literal_endpoint == 2 &&
2045 ((isLOWER(min) && isLOWER(max)) ||
2046 (isUPPER(min) && isUPPER(max)))) {
2048 for (i = min; i <= max; i++)
2050 *d++ = NATIVE_TO_NEED(has_utf8,i);
2052 for (i = min; i <= max; i++)
2054 *d++ = NATIVE_TO_NEED(has_utf8,i);
2059 for (i = min; i <= max; i++)
2062 const U8 ch = (U8)NATIVE_TO_UTF(i);
2063 if (UNI_IS_INVARIANT(ch))
2066 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2067 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2076 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2078 *d++ = (char)UTF_TO_NATIVE(0xff);
2080 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2084 /* mark the range as done, and continue */
2088 literal_endpoint = 0;
2093 /* range begins (ignore - as first or last char) */
2094 else if (*s == '-' && s+1 < send && s != start) {
2096 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2103 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2113 literal_endpoint = 0;
2114 native_range = TRUE;
2119 /* if we get here, we're not doing a transliteration */
2121 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2122 except for the last char, which will be done separately. */
2123 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2125 while (s+1 < send && *s != ')')
2126 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2128 else if (s[2] == '{' /* This should match regcomp.c */
2129 || (s[2] == '?' && s[3] == '{'))
2132 char *regparse = s + (s[2] == '{' ? 3 : 4);
2135 while (count && (c = *regparse)) {
2136 if (c == '\\' && regparse[1])
2144 if (*regparse != ')')
2145 regparse--; /* Leave one char for continuation. */
2146 while (s < regparse)
2147 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2151 /* likewise skip #-initiated comments in //x patterns */
2152 else if (*s == '#' && PL_lex_inpat &&
2153 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2154 while (s+1 < send && *s != '\n')
2155 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2158 /* check for embedded arrays
2159 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2161 else if (*s == '@' && s[1]) {
2162 if (isALNUM_lazy_if(s+1,UTF))
2164 if (strchr(":'{$", s[1]))
2166 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2167 break; /* in regexp, neither @+ nor @- are interpolated */
2170 /* check for embedded scalars. only stop if we're sure it's a
2173 else if (*s == '$') {
2174 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2176 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
2177 if (s[1] == '\\' && ckWARN(WARN_AMBIGUOUS)) {
2178 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2179 "Possible unintended interpolation of $\\ in regex");
2181 break; /* in regexp, $ might be tail anchor */
2185 /* End of else if chain - OP_TRANS rejoin rest */
2188 if (*s == '\\' && s+1 < send) {
2191 /* deprecate \1 in strings and substitution replacements */
2192 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2193 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2195 if (ckWARN(WARN_SYNTAX))
2196 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2201 /* string-change backslash escapes */
2202 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2206 /* skip any other backslash escapes in a pattern */
2207 else if (PL_lex_inpat) {
2208 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2209 goto default_action;
2212 /* if we get here, it's either a quoted -, or a digit */
2215 /* quoted - in transliterations */
2217 if (PL_lex_inwhat == OP_TRANS) {
2224 if ((isALPHA(*s) || isDIGIT(*s)) &&
2226 Perl_warner(aTHX_ packWARN(WARN_MISC),
2227 "Unrecognized escape \\%c passed through",
2229 /* default action is to copy the quoted character */
2230 goto default_action;
2233 /* \132 indicates an octal constant */
2234 case '0': case '1': case '2': case '3':
2235 case '4': case '5': case '6': case '7':
2239 uv = grok_oct(s, &len, &flags, NULL);
2242 goto NUM_ESCAPE_INSERT;
2244 /* \x24 indicates a hex constant */
2248 char* const e = strchr(s, '}');
2249 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2250 PERL_SCAN_DISALLOW_PREFIX;
2255 yyerror("Missing right brace on \\x{}");
2259 uv = grok_hex(s, &len, &flags, NULL);
2265 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2266 uv = grok_hex(s, &len, &flags, NULL);
2272 /* Insert oct or hex escaped character.
2273 * There will always enough room in sv since such
2274 * escapes will be longer than any UTF-8 sequence
2275 * they can end up as. */
2277 /* We need to map to chars to ASCII before doing the tests
2280 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2281 if (!has_utf8 && uv > 255) {
2282 /* Might need to recode whatever we have
2283 * accumulated so far if it contains any
2286 * (Can't we keep track of that and avoid
2287 * this rescan? --jhi)
2291 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2292 if (!NATIVE_IS_INVARIANT(*c)) {
2297 const STRLEN offset = d - SvPVX_const(sv);
2299 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2303 while (src >= (const U8 *)SvPVX_const(sv)) {
2304 if (!NATIVE_IS_INVARIANT(*src)) {
2305 const U8 ch = NATIVE_TO_ASCII(*src);
2306 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2307 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2317 if (has_utf8 || uv > 255) {
2318 d = (char*)uvchr_to_utf8((U8*)d, uv);
2320 if (PL_lex_inwhat == OP_TRANS &&
2321 PL_sublex_info.sub_op) {
2322 PL_sublex_info.sub_op->op_private |=
2323 (PL_lex_repl ? OPpTRANS_FROM_UTF
2327 if (uv > 255 && !dorange)
2328 native_range = FALSE;
2340 /* \N{LATIN SMALL LETTER A} is a named character */
2344 char* e = strchr(s, '}');
2350 yyerror("Missing right brace on \\N{}");
2354 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2356 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2357 PERL_SCAN_DISALLOW_PREFIX;
2360 uv = grok_hex(s, &len, &flags, NULL);
2361 if ( e > s && len != (STRLEN)(e - s) ) {
2365 goto NUM_ESCAPE_INSERT;
2367 res = newSVpvn(s + 1, e - s - 1);
2368 res = new_constant( NULL, 0, "charnames",
2369 res, NULL, s - 2, e - s + 3 );
2371 sv_utf8_upgrade(res);
2372 str = SvPV_const(res,len);
2373 #ifdef EBCDIC_NEVER_MIND
2374 /* charnames uses pack U and that has been
2375 * recently changed to do the below uni->native
2376 * mapping, so this would be redundant (and wrong,
2377 * the code point would be doubly converted).
2378 * But leave this in just in case the pack U change
2379 * gets revoked, but the semantics is still
2380 * desireable for charnames. --jhi */
2382 UV uv = utf8_to_uvchr((const U8*)str, 0);
2385 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2387 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2388 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2389 str = SvPV_const(res, len);
2393 if (!has_utf8 && SvUTF8(res)) {
2394 const char * const ostart = SvPVX_const(sv);
2395 SvCUR_set(sv, d - ostart);
2398 sv_utf8_upgrade(sv);
2399 /* this just broke our allocation above... */
2400 SvGROW(sv, (STRLEN)(send - start));
2401 d = SvPVX(sv) + SvCUR(sv);
2404 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2405 const char * const odest = SvPVX_const(sv);
2407 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2408 d = SvPVX(sv) + (d - odest);
2412 native_range = FALSE; /* \N{} is guessed to be Unicode */
2414 Copy(str, d, len, char);
2421 yyerror("Missing braces on \\N{}");
2424 /* \c is a control character */
2433 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2436 yyerror("Missing control char name in \\c");
2440 /* printf-style backslashes, formfeeds, newlines, etc */
2442 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2445 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2448 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2451 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2454 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2457 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2460 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2466 } /* end if (backslash) */
2473 /* If we started with encoded form, or already know we want it
2474 and then encode the next character */
2475 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2477 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2478 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2481 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2482 const STRLEN off = d - SvPVX_const(sv);
2483 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2485 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2488 if (uv > 255 && !dorange)
2489 native_range = FALSE;
2493 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2495 } /* while loop to process each character */
2497 /* terminate the string and set up the sv */
2499 SvCUR_set(sv, d - SvPVX_const(sv));
2500 if (SvCUR(sv) >= SvLEN(sv))
2501 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2504 if (PL_encoding && !has_utf8) {
2505 sv_recode_to_utf8(sv, PL_encoding);
2511 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2512 PL_sublex_info.sub_op->op_private |=
2513 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2517 /* shrink the sv if we allocated more than we used */
2518 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2519 SvPV_shrink_to_cur(sv);
2522 /* return the substring (via pl_yylval) only if we parsed anything */
2523 if (s > PL_bufptr) {
2524 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2525 const char *const key = PL_lex_inpat ? "qr" : "q";
2526 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2530 if (PL_lex_inwhat == OP_TRANS) {
2533 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2541 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2544 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2551 * Returns TRUE if there's more to the expression (e.g., a subscript),
2554 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2556 * ->[ and ->{ return TRUE
2557 * { and [ outside a pattern are always subscripts, so return TRUE
2558 * if we're outside a pattern and it's not { or [, then return FALSE
2559 * if we're in a pattern and the first char is a {
2560 * {4,5} (any digits around the comma) returns FALSE
2561 * if we're in a pattern and the first char is a [
2563 * [SOMETHING] has a funky algorithm to decide whether it's a
2564 * character class or not. It has to deal with things like
2565 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2566 * anything else returns TRUE
2569 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2572 S_intuit_more(pTHX_ register char *s)
2576 PERL_ARGS_ASSERT_INTUIT_MORE;
2578 if (PL_lex_brackets)
2580 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2582 if (*s != '{' && *s != '[')
2587 /* In a pattern, so maybe we have {n,m}. */
2604 /* On the other hand, maybe we have a character class */
2607 if (*s == ']' || *s == '^')
2610 /* this is terrifying, and it works */
2611 int weight = 2; /* let's weigh the evidence */
2613 unsigned char un_char = 255, last_un_char;
2614 const char * const send = strchr(s,']');
2615 char tmpbuf[sizeof PL_tokenbuf * 4];
2617 if (!send) /* has to be an expression */
2620 Zero(seen,256,char);
2623 else if (isDIGIT(*s)) {
2625 if (isDIGIT(s[1]) && s[2] == ']')
2631 for (; s < send; s++) {
2632 last_un_char = un_char;
2633 un_char = (unsigned char)*s;
2638 weight -= seen[un_char] * 10;
2639 if (isALNUM_lazy_if(s+1,UTF)) {
2641 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2642 len = (int)strlen(tmpbuf);
2643 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2648 else if (*s == '$' && s[1] &&
2649 strchr("[#!%*<>()-=",s[1])) {
2650 if (/*{*/ strchr("])} =",s[2]))
2659 if (strchr("wds]",s[1]))
2661 else if (seen[(U8)'\''] || seen[(U8)'"'])
2663 else if (strchr("rnftbxcav",s[1]))
2665 else if (isDIGIT(s[1])) {
2667 while (s[1] && isDIGIT(s[1]))
2677 if (strchr("aA01! ",last_un_char))
2679 if (strchr("zZ79~",s[1]))
2681 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2682 weight -= 5; /* cope with negative subscript */
2685 if (!isALNUM(last_un_char)
2686 && !(last_un_char == '$' || last_un_char == '@'
2687 || last_un_char == '&')
2688 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2693 if (keyword(tmpbuf, d - tmpbuf, 0))
2696 if (un_char == last_un_char + 1)
2698 weight -= seen[un_char];
2703 if (weight >= 0) /* probably a character class */
2713 * Does all the checking to disambiguate
2715 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2716 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2718 * First argument is the stuff after the first token, e.g. "bar".
2720 * Not a method if bar is a filehandle.
2721 * Not a method if foo is a subroutine prototyped to take a filehandle.
2722 * Not a method if it's really "Foo $bar"
2723 * Method if it's "foo $bar"
2724 * Not a method if it's really "print foo $bar"
2725 * Method if it's really "foo package::" (interpreted as package->foo)
2726 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2727 * Not a method if bar is a filehandle or package, but is quoted with
2732 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2735 char *s = start + (*start == '$');
2736 char tmpbuf[sizeof PL_tokenbuf];
2743 PERL_ARGS_ASSERT_INTUIT_METHOD;
2746 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2750 const char *proto = SvPVX_const(cv);
2761 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2762 /* start is the beginning of the possible filehandle/object,
2763 * and s is the end of it
2764 * tmpbuf is a copy of it
2767 if (*start == '$') {
2768 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2769 isUPPER(*PL_tokenbuf))
2772 len = start - SvPVX(PL_linestr);
2776 start = SvPVX(PL_linestr) + len;
2780 return *s == '(' ? FUNCMETH : METHOD;
2782 if (!keyword(tmpbuf, len, 0)) {
2783 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2787 soff = s - SvPVX(PL_linestr);
2791 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2792 if (indirgv && GvCVu(indirgv))
2794 /* filehandle or package name makes it a method */
2795 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2797 soff = s - SvPVX(PL_linestr);
2800 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2801 return 0; /* no assumptions -- "=>" quotes bearword */
2803 start_force(PL_curforce);
2804 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2805 newSVpvn(tmpbuf,len));
2806 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2808 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2813 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2815 return *s == '(' ? FUNCMETH : METHOD;
2821 /* Encoded script support. filter_add() effectively inserts a
2822 * 'pre-processing' function into the current source input stream.
2823 * Note that the filter function only applies to the current source file
2824 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2826 * The datasv parameter (which may be NULL) can be used to pass
2827 * private data to this instance of the filter. The filter function
2828 * can recover the SV using the FILTER_DATA macro and use it to
2829 * store private buffers and state information.
2831 * The supplied datasv parameter is upgraded to a PVIO type
2832 * and the IoDIRP/IoANY field is used to store the function pointer,
2833 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2834 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2835 * private use must be set using malloc'd pointers.
2839 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2848 if (!PL_rsfp_filters)
2849 PL_rsfp_filters = newAV();
2852 SvUPGRADE(datasv, SVt_PVIO);
2853 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2854 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2855 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2856 FPTR2DPTR(void *, IoANY(datasv)),
2857 SvPV_nolen(datasv)));
2858 av_unshift(PL_rsfp_filters, 1);
2859 av_store(PL_rsfp_filters, 0, datasv) ;
2864 /* Delete most recently added instance of this filter function. */
2866 Perl_filter_del(pTHX_ filter_t funcp)
2871 PERL_ARGS_ASSERT_FILTER_DEL;
2874 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2875 FPTR2DPTR(void*, funcp)));
2877 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2879 /* if filter is on top of stack (usual case) just pop it off */
2880 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2881 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2882 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2883 IoANY(datasv) = (void *)NULL;
2884 sv_free(av_pop(PL_rsfp_filters));
2888 /* we need to search for the correct entry and clear it */
2889 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2893 /* Invoke the idxth filter function for the current rsfp. */
2894 /* maxlen 0 = read one text line */
2896 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2901 /* This API is bad. It should have been using unsigned int for maxlen.
2902 Not sure if we want to change the API, but if not we should sanity
2903 check the value here. */
2904 const unsigned int correct_length
2913 PERL_ARGS_ASSERT_FILTER_READ;
2915 if (!PL_parser || !PL_rsfp_filters)
2917 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2918 /* Provide a default input filter to make life easy. */
2919 /* Note that we append to the line. This is handy. */
2920 DEBUG_P(PerlIO_printf(Perl_debug_log,
2921 "filter_read %d: from rsfp\n", idx));
2922 if (correct_length) {
2925 const int old_len = SvCUR(buf_sv);
2927 /* ensure buf_sv is large enough */
2928 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2929 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2930 correct_length)) <= 0) {
2931 if (PerlIO_error(PL_rsfp))
2932 return -1; /* error */
2934 return 0 ; /* end of file */
2936 SvCUR_set(buf_sv, old_len + len) ;
2939 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2940 if (PerlIO_error(PL_rsfp))
2941 return -1; /* error */
2943 return 0 ; /* end of file */
2946 return SvCUR(buf_sv);
2948 /* Skip this filter slot if filter has been deleted */
2949 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2950 DEBUG_P(PerlIO_printf(Perl_debug_log,
2951 "filter_read %d: skipped (filter deleted)\n",
2953 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2955 /* Get function pointer hidden within datasv */
2956 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2957 DEBUG_P(PerlIO_printf(Perl_debug_log,
2958 "filter_read %d: via function %p (%s)\n",
2959 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2960 /* Call function. The function is expected to */
2961 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2962 /* Return: <0:error, =0:eof, >0:not eof */
2963 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2967 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2971 PERL_ARGS_ASSERT_FILTER_GETS;
2973 #ifdef PERL_CR_FILTER
2974 if (!PL_rsfp_filters) {
2975 filter_add(S_cr_textfilter,NULL);
2978 if (PL_rsfp_filters) {
2980 SvCUR_set(sv, 0); /* start with empty line */
2981 if (FILTER_READ(0, sv, 0) > 0)
2982 return ( SvPVX(sv) ) ;
2987 return (sv_gets(sv, fp, append));
2991 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
2996 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
2998 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
3002 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
3003 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3005 return GvHV(gv); /* Foo:: */
3008 /* use constant CLASS => 'MyClass' */
3009 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3010 if (gv && GvCV(gv)) {
3011 SV * const sv = cv_const_sv(GvCV(gv));
3013 pkgname = SvPV_const(sv, len);
3016 return gv_stashpvn(pkgname, len, 0);
3020 * S_readpipe_override
3021 * Check whether readpipe() is overriden, and generates the appropriate
3022 * optree, provided sublex_start() is called afterwards.
3025 S_readpipe_override(pTHX)
3028 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3029 pl_yylval.ival = OP_BACKTICK;
3031 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3033 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3034 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3035 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3037 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3038 append_elem(OP_LIST,
3039 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3040 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3047 * The intent of this yylex wrapper is to minimize the changes to the
3048 * tokener when we aren't interested in collecting madprops. It remains
3049 * to be seen how successful this strategy will be...
3056 char *s = PL_bufptr;
3058 /* make sure PL_thiswhite is initialized */
3062 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3063 if (PL_pending_ident)
3064 return S_pending_ident(aTHX);
3066 /* previous token ate up our whitespace? */
3067 if (!PL_lasttoke && PL_nextwhite) {
3068 PL_thiswhite = PL_nextwhite;
3072 /* isolate the token, and figure out where it is without whitespace */
3073 PL_realtokenstart = -1;
3077 assert(PL_curforce < 0);
3079 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3080 if (!PL_thistoken) {
3081 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3082 PL_thistoken = newSVpvs("");
3084 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3085 PL_thistoken = newSVpvn(tstart, s - tstart);
3088 if (PL_thismad) /* install head */
3089 CURMAD('X', PL_thistoken);
3092 /* last whitespace of a sublex? */
3093 if (optype == ')' && PL_endwhite) {
3094 CURMAD('X', PL_endwhite);
3099 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3100 if (!PL_thiswhite && !PL_endwhite && !optype) {
3101 sv_free(PL_thistoken);
3106 /* put off final whitespace till peg */
3107 if (optype == ';' && !PL_rsfp) {
3108 PL_nextwhite = PL_thiswhite;
3111 else if (PL_thisopen) {
3112 CURMAD('q', PL_thisopen);
3114 sv_free(PL_thistoken);
3118 /* Store actual token text as madprop X */
3119 CURMAD('X', PL_thistoken);
3123 /* add preceding whitespace as madprop _ */
3124 CURMAD('_', PL_thiswhite);
3128 /* add quoted material as madprop = */
3129 CURMAD('=', PL_thisstuff);
3133 /* add terminating quote as madprop Q */
3134 CURMAD('Q', PL_thisclose);
3138 /* special processing based on optype */
3142 /* opval doesn't need a TOKEN since it can already store mp */
3152 if (pl_yylval.opval)
3153 append_madprops(PL_thismad, pl_yylval.opval, 0);
3161 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3170 /* remember any fake bracket that lexer is about to discard */
3171 if (PL_lex_brackets == 1 &&
3172 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3175 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3178 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3179 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3182 break; /* don't bother looking for trailing comment */
3191 /* attach a trailing comment to its statement instead of next token */
3195 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3197 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3199 if (*s == '\n' || *s == '#') {
3200 while (s < PL_bufend && *s != '\n')
3204 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3205 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3222 /* Create new token struct. Note: opvals return early above. */
3223 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3230 S_tokenize_use(pTHX_ int is_use, char *s) {
3233 PERL_ARGS_ASSERT_TOKENIZE_USE;
3235 if (PL_expect != XSTATE)
3236 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3237 is_use ? "use" : "no"));
3239 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3240 s = force_version(s, TRUE);
3241 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3242 start_force(PL_curforce);
3243 NEXTVAL_NEXTTOKE.opval = NULL;
3246 else if (*s == 'v') {
3247 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3248 s = force_version(s, FALSE);
3252 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3253 s = force_version(s, FALSE);
3255 pl_yylval.ival = is_use;
3259 static const char* const exp_name[] =
3260 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3261 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3268 Works out what to call the token just pulled out of the input
3269 stream. The yacc parser takes care of taking the ops we return and
3270 stitching them into a tree.
3276 if read an identifier
3277 if we're in a my declaration
3278 croak if they tried to say my($foo::bar)
3279 build the ops for a my() declaration
3280 if it's an access to a my() variable
3281 are we in a sort block?
3282 croak if my($a); $a <=> $b
3283 build ops for access to a my() variable
3284 if in a dq string, and they've said @foo and we can't find @foo
3286 build ops for a bareword
3287 if we already built the token before, use it.
3292 #pragma segment Perl_yylex
3298 register char *s = PL_bufptr;
3303 /* orig_keyword, gvp, and gv are initialized here because
3304 * jump to the label just_a_word_zero can bypass their
3305 * initialization later. */
3306 I32 orig_keyword = 0;
3311 SV* tmp = newSVpvs("");
3312 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3313 (IV)CopLINE(PL_curcop),
3314 lex_state_names[PL_lex_state],
3315 exp_name[PL_expect],
3316 pv_display(tmp, s, strlen(s), 0, 60));
3319 /* check if there's an identifier for us to look at */
3320 if (PL_pending_ident)
3321 return REPORT(S_pending_ident(aTHX));
3323 /* no identifier pending identification */
3325 switch (PL_lex_state) {
3327 case LEX_NORMAL: /* Some compilers will produce faster */
3328 case LEX_INTERPNORMAL: /* code if we comment these out. */
3332 /* when we've already built the next token, just pull it out of the queue */
3336 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3338 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3339 PL_nexttoke[PL_lasttoke].next_mad = 0;
3340 if (PL_thismad && PL_thismad->mad_key == '_') {
3341 PL_thiswhite = (SV*)PL_thismad->mad_val;
3342 PL_thismad->mad_val = 0;
3343 mad_free(PL_thismad);
3348 PL_lex_state = PL_lex_defer;
3349 PL_expect = PL_lex_expect;
3350 PL_lex_defer = LEX_NORMAL;
3351 if (!PL_nexttoke[PL_lasttoke].next_type)
3356 pl_yylval = PL_nextval[PL_nexttoke];
3358 PL_lex_state = PL_lex_defer;
3359 PL_expect = PL_lex_expect;
3360 PL_lex_defer = LEX_NORMAL;
3364 /* FIXME - can these be merged? */
3365 return(PL_nexttoke[PL_lasttoke].next_type);
3367 return REPORT(PL_nexttype[PL_nexttoke]);
3370 /* interpolated case modifiers like \L \U, including \Q and \E.
3371 when we get here, PL_bufptr is at the \
3373 case LEX_INTERPCASEMOD:
3375 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3376 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3378 /* handle \E or end of string */
3379 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3381 if (PL_lex_casemods) {
3382 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3383 PL_lex_casestack[PL_lex_casemods] = '\0';
3385 if (PL_bufptr != PL_bufend
3386 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3388 PL_lex_state = LEX_INTERPCONCAT;
3391 PL_thistoken = newSVpvs("\\E");
3397 while (PL_bufptr != PL_bufend &&
3398 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3400 PL_thiswhite = newSVpvs("");
3401 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3405 if (PL_bufptr != PL_bufend)
3408 PL_lex_state = LEX_INTERPCONCAT;
3412 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3413 "### Saw case modifier\n"); });
3415 if (s[1] == '\\' && s[2] == 'E') {
3418 PL_thiswhite = newSVpvs("");
3419 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3422 PL_lex_state = LEX_INTERPCONCAT;
3427 if (!PL_madskills) /* when just compiling don't need correct */
3428 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3429 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3430 if ((*s == 'L' || *s == 'U') &&
3431 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3432 PL_lex_casestack[--PL_lex_casemods] = '\0';
3435 if (PL_lex_casemods > 10)
3436 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3437 PL_lex_casestack[PL_lex_casemods++] = *s;
3438 PL_lex_casestack[PL_lex_casemods] = '\0';
3439 PL_lex_state = LEX_INTERPCONCAT;
3440 start_force(PL_curforce);
3441 NEXTVAL_NEXTTOKE.ival = 0;
3443 start_force(PL_curforce);
3445 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3447 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3449 NEXTVAL_NEXTTOKE.ival = OP_LC;
3451 NEXTVAL_NEXTTOKE.ival = OP_UC;
3453 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3455 Perl_croak(aTHX_ "panic: yylex");
3457 SV* const tmpsv = newSVpvs("\\ ");
3458 /* replace the space with the character we want to escape
3460 SvPVX(tmpsv)[1] = *s;
3466 if (PL_lex_starts) {
3472 sv_free(PL_thistoken);
3473 PL_thistoken = newSVpvs("");
3476 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3477 if (PL_lex_casemods == 1 && PL_lex_inpat)
3486 case LEX_INTERPPUSH:
3487 return REPORT(sublex_push());
3489 case LEX_INTERPSTART:
3490 if (PL_bufptr == PL_bufend)
3491 return REPORT(sublex_done());
3492 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3493 "### Interpolated variable\n"); });
3495 PL_lex_dojoin = (*PL_bufptr == '@');
3496 PL_lex_state = LEX_INTERPNORMAL;
3497 if (PL_lex_dojoin) {
3498 start_force(PL_curforce);
3499 NEXTVAL_NEXTTOKE.ival = 0;
3501 start_force(PL_curforce);
3502 force_ident("\"", '$');
3503 start_force(PL_curforce);
3504 NEXTVAL_NEXTTOKE.ival = 0;
3506 start_force(PL_curforce);
3507 NEXTVAL_NEXTTOKE.ival = 0;
3509 start_force(PL_curforce);
3510 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3513 if (PL_lex_starts++) {
3518 sv_free(PL_thistoken);
3519 PL_thistoken = newSVpvs("");
3522 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3523 if (!PL_lex_casemods && PL_lex_inpat)
3530 case LEX_INTERPENDMAYBE:
3531 if (intuit_more(PL_bufptr)) {
3532 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3538 if (PL_lex_dojoin) {
3539 PL_lex_dojoin = FALSE;
3540 PL_lex_state = LEX_INTERPCONCAT;
3544 sv_free(PL_thistoken);
3545 PL_thistoken = newSVpvs("");
3550 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3551 && SvEVALED(PL_lex_repl))
3553 if (PL_bufptr != PL_bufend)
3554 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3558 case LEX_INTERPCONCAT:
3560 if (PL_lex_brackets)
3561 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3563 if (PL_bufptr == PL_bufend)
3564 return REPORT(sublex_done());
3566 if (SvIVX(PL_linestr) == '\'') {
3567 SV *sv = newSVsv(PL_linestr);
3570 else if ( PL_hints & HINT_NEW_RE )
3571 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3572 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3576 s = scan_const(PL_bufptr);
3578 PL_lex_state = LEX_INTERPCASEMOD;
3580 PL_lex_state = LEX_INTERPSTART;
3583 if (s != PL_bufptr) {
3584 start_force(PL_curforce);
3586 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3588 NEXTVAL_NEXTTOKE = pl_yylval;
3591 if (PL_lex_starts++) {
3595 sv_free(PL_thistoken);
3596 PL_thistoken = newSVpvs("");
3599 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3600 if (!PL_lex_casemods && PL_lex_inpat)
3613 PL_lex_state = LEX_NORMAL;
3614 s = scan_formline(PL_bufptr);
3615 if (!PL_lex_formbrack)
3621 PL_oldoldbufptr = PL_oldbufptr;
3627 sv_free(PL_thistoken);
3630 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3634 if (isIDFIRST_lazy_if(s,UTF))
3636 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3637 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3640 goto fake_eof; /* emulate EOF on ^D or ^Z */
3649 if (PL_lex_brackets) {
3650 yyerror((const char *)
3652 ? "Format not terminated"
3653 : "Missing right curly or square bracket"));
3655 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3656 "### Tokener got EOF\n");
3660 if (s++ < PL_bufend)
3661 goto retry; /* ignore stray nulls */
3664 if (!PL_in_eval && !PL_preambled) {
3665 PL_preambled = TRUE;
3671 /* Generate a string of Perl code to load the debugger.
3672 * If PERL5DB is set, it will return the contents of that,
3673 * otherwise a compile-time require of perl5db.pl. */
3675 const char * const pdb = PerlEnv_getenv("PERL5DB");
3678 sv_setpv(PL_linestr, pdb);
3679 sv_catpvs(PL_linestr,";");
3681 SETERRNO(0,SS_NORMAL);
3682 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3685 sv_setpvs(PL_linestr,"");
3686 if (PL_preambleav) {
3687 SV **svp = AvARRAY(PL_preambleav);
3688 SV **const end = svp + AvFILLp(PL_preambleav);
3690 sv_catsv(PL_linestr, *svp);
3692 sv_catpvs(PL_linestr, ";");
3694 sv_free((SV*)PL_preambleav);
3695 PL_preambleav = NULL;
3698 sv_catpvs(PL_linestr,
3699 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3700 if (PL_minus_n || PL_minus_p) {
3701 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3703 sv_catpvs(PL_linestr,"chomp;");
3706 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3707 || *PL_splitstr == '"')
3708 && strchr(PL_splitstr + 1, *PL_splitstr))
3709 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3711 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3712 bytes can be used as quoting characters. :-) */
3713 const char *splits = PL_splitstr;
3714 sv_catpvs(PL_linestr, "our @F=split(q\0");
3717 if (*splits == '\\')
3718 sv_catpvn(PL_linestr, splits, 1);
3719 sv_catpvn(PL_linestr, splits, 1);
3720 } while (*splits++);
3721 /* This loop will embed the trailing NUL of
3722 PL_linestr as the last thing it does before
3724 sv_catpvs(PL_linestr, ");");
3728 sv_catpvs(PL_linestr,"our @F=split(' ');");
3731 sv_catpvs(PL_linestr, "\n");
3732 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3733 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3734 PL_last_lop = PL_last_uni = NULL;
3735 if (PERLDB_LINE && PL_curstash != PL_debstash)
3736 update_debugger_info(PL_linestr, NULL, 0);
3740 bof = PL_rsfp ? TRUE : FALSE;
3741 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3744 PL_realtokenstart = -1;
3747 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3748 PerlIO_clearerr(PL_rsfp);
3750 (void)PerlIO_close(PL_rsfp);
3752 PL_doextract = FALSE;
3754 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3760 sv_setpvs(PL_linestr, ";}continue{print;}");
3762 sv_setpvs(PL_linestr, ";}");
3763 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3764 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3765 PL_last_lop = PL_last_uni = NULL;
3766 PL_minus_n = PL_minus_p = 0;
3769 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3770 PL_last_lop = PL_last_uni = NULL;
3771 sv_setpvn(PL_linestr,"",0);
3772 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3774 /* If it looks like the start of a BOM or raw UTF-16,
3775 * check if it in fact is. */
3781 #ifdef PERLIO_IS_STDIO
3782 # ifdef __GNU_LIBRARY__
3783 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3784 # define FTELL_FOR_PIPE_IS_BROKEN
3788 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3789 # define FTELL_FOR_PIPE_IS_BROKEN
3794 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3796 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3797 s = swallow_bom((U8*)s);
3801 /* Incest with pod. */
3804 sv_catsv(PL_thiswhite, PL_linestr);
3806 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3807 sv_setpvn(PL_linestr, "", 0);
3808 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3809 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3810 PL_last_lop = PL_last_uni = NULL;
3811 PL_doextract = FALSE;
3815 } while (PL_doextract);
3816 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3817 if (PERLDB_LINE && PL_curstash != PL_debstash)
3818 update_debugger_info(PL_linestr, NULL, 0);
3819 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3820 PL_last_lop = PL_last_uni = NULL;
3821 if (CopLINE(PL_curcop) == 1) {
3822 while (s < PL_bufend && isSPACE(*s))
3824 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3828 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3832 if (*s == '#' && *(s+1) == '!')
3834 #ifdef ALTERNATE_SHEBANG
3836 static char const as[] = ALTERNATE_SHEBANG;
3837 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3838 d = s + (sizeof(as) - 1);
3840 #endif /* ALTERNATE_SHEBANG */
3849 while (*d && !isSPACE(*d))
3853 #ifdef ARG_ZERO_IS_SCRIPT
3854 if (ipathend > ipath) {
3856 * HP-UX (at least) sets argv[0] to the script name,
3857 * which makes $^X incorrect. And Digital UNIX and Linux,
3858 * at least, set argv[0] to the basename of the Perl
3859 * interpreter. So, having found "#!", we'll set it right.
3861 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3863 assert(SvPOK(x) || SvGMAGICAL(x));
3864 if (sv_eq(x, CopFILESV(PL_curcop))) {
3865 sv_setpvn(x, ipath, ipathend - ipath);
3871 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3872 const char * const lstart = SvPV_const(x,llen);
3874 bstart += blen - llen;
3875 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3876 sv_setpvn(x, ipath, ipathend - ipath);
3881 TAINT_NOT; /* $^X is always tainted, but that's OK */
3883 #endif /* ARG_ZERO_IS_SCRIPT */
3888 d = instr(s,"perl -");
3890 d = instr(s,"perl");
3892 /* avoid getting into infinite loops when shebang
3893 * line contains "Perl" rather than "perl" */
3895 for (d = ipathend-4; d >= ipath; --d) {
3896 if ((*d == 'p' || *d == 'P')
3897 && !ibcmp(d, "perl", 4))
3907 #ifdef ALTERNATE_SHEBANG
3909 * If the ALTERNATE_SHEBANG on this system starts with a
3910 * character that can be part of a Perl expression, then if
3911 * we see it but not "perl", we're probably looking at the
3912 * start of Perl code, not a request to hand off to some
3913 * other interpreter. Similarly, if "perl" is there, but
3914 * not in the first 'word' of the line, we assume the line
3915 * contains the start of the Perl program.
3917 if (d && *s != '#') {
3918 const char *c = ipath;
3919 while (*c && !strchr("; \t\r\n\f\v#", *c))
3922 d = NULL; /* "perl" not in first word; ignore */
3924 *s = '#'; /* Don't try to parse shebang line */
3926 #endif /* ALTERNATE_SHEBANG */
3927 #ifndef MACOS_TRADITIONAL
3932 !instr(s,"indir") &&
3933 instr(PL_origargv[0],"perl"))
3940 while (s < PL_bufend && isSPACE(*s))
3942 if (s < PL_bufend) {
3943 Newxz(newargv,PL_origargc+3,char*);
3945 while (s < PL_bufend && !isSPACE(*s))
3948 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3951 newargv = PL_origargv;
3954 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3956 Perl_croak(aTHX_ "Can't exec %s", ipath);
3960 while (*d && !isSPACE(*d))
3962 while (SPACE_OR_TAB(*d))
3966 const bool switches_done = PL_doswitches;
3967 const U32 oldpdb = PL_perldb;
3968 const bool oldn = PL_minus_n;
3969 const bool oldp = PL_minus_p;
3973 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3974 const char * const m = d1;
3975 while (*d1 && !isSPACE(*d1))
3977 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3980 d1 = moreswitches(d1);
3982 if (PL_doswitches && !switches_done) {
3983 int argc = PL_origargc;
3984 char **argv = PL_origargv;
3987 } while (argc && argv[0][0] == '-' && argv[0][1]);
3988 init_argv_symbols(argc,argv);
3990 if ((PERLDB_LINE && !oldpdb) ||
3991 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3992 /* if we have already added "LINE: while (<>) {",
3993 we must not do it again */
3995 sv_setpvn(PL_linestr, "", 0);
3996 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3997 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3998 PL_last_lop = PL_last_uni = NULL;
3999 PL_preambled = FALSE;
4001 (void)gv_fetchfile(PL_origfilename);
4008 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4010 PL_lex_state = LEX_FORMLINE;
4015 #ifdef PERL_STRICT_CR
4016 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4018 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4020 case ' ': case '\t': case '\f': case 013:
4021 #ifdef MACOS_TRADITIONAL
4025 PL_realtokenstart = -1;
4027 PL_thiswhite = newSVpvs("");
4028 sv_catpvn(PL_thiswhite, s, 1);
4035 PL_realtokenstart = -1;
4039 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4040 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4041 /* handle eval qq[#line 1 "foo"\n ...] */
4042 CopLINE_dec(PL_curcop);
4045 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4047 if (!PL_in_eval || PL_rsfp)
4052 while (d < PL_bufend && *d != '\n')
4056 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4057 Perl_croak(aTHX_ "panic: input overflow");
4060 PL_thiswhite = newSVpvn(s, d - s);
4065 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4067 PL_lex_state = LEX_FORMLINE;
4073 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4074 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4077 TOKEN(PEG); /* make sure any #! line is accessible */
4082 /* if (PL_madskills && PL_lex_formbrack) { */
4084 while (d < PL_bufend && *d != '\n')
4088 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4089 Perl_croak(aTHX_ "panic: input overflow");
4090 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4092 PL_thiswhite = newSVpvs("");
4093 if (CopLINE(PL_curcop) == 1) {
4094 sv_setpvn(PL_thiswhite, "", 0);
4097 sv_catpvn(PL_thiswhite, s, d - s);
4111 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4119 while (s < PL_bufend && SPACE_OR_TAB(*s))
4122 if (strnEQ(s,"=>",2)) {
4123 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4124 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4125 OPERATOR('-'); /* unary minus */
4127 PL_last_uni = PL_oldbufptr;
4129 case 'r': ftst = OP_FTEREAD; break;
4130 case 'w': ftst = OP_FTEWRITE; break;
4131 case 'x': ftst = OP_FTEEXEC; break;
4132 case 'o': ftst = OP_FTEOWNED; break;
4133 case 'R': ftst = OP_FTRREAD; break;
4134 case 'W': ftst = OP_FTRWRITE; break;
4135 case 'X': ftst = OP_FTREXEC; break;
4136 case 'O': ftst = OP_FTROWNED; break;
4137 case 'e': ftst = OP_FTIS; break;
4138 case 'z': ftst = OP_FTZERO; break;
4139 case 's': ftst = OP_FTSIZE; break;
4140 case 'f': ftst = OP_FTFILE; break;
4141 case 'd': ftst = OP_FTDIR; break;
4142 case 'l': ftst = OP_FTLINK; break;
4143 case 'p': ftst = OP_FTPIPE; break;
4144 case 'S': ftst = OP_FTSOCK; break;
4145 case 'u': ftst = OP_FTSUID; break;
4146 case 'g': ftst = OP_FTSGID; break;
4147 case 'k': ftst = OP_FTSVTX; break;
4148 case 'b': ftst = OP_FTBLK; break;
4149 case 'c': ftst = OP_FTCHR; break;
4150 case 't': ftst = OP_FTTTY; break;
4151 case 'T': ftst = OP_FTTEXT; break;
4152 case 'B': ftst = OP_FTBINARY; break;
4153 case 'M': case 'A': case 'C':
4154 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4156 case 'M': ftst = OP_FTMTIME; break;
4157 case 'A': ftst = OP_FTATIME; break;
4158 case 'C': ftst = OP_FTCTIME; break;
4166 PL_last_lop_op = (OPCODE)ftst;
4167 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4168 "### Saw file test %c\n", (int)tmp);
4173 /* Assume it was a minus followed by a one-letter named
4174 * subroutine call (or a -bareword), then. */
4175 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4176 "### '-%c' looked like a file test but was not\n",
4183 const char tmp = *s++;
4186 if (PL_expect == XOPERATOR)
4191 else if (*s == '>') {
4194 if (isIDFIRST_lazy_if(s,UTF)) {
4195 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4203 if (PL_expect == XOPERATOR)
4206 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4208 OPERATOR('-'); /* unary minus */
4214 const char tmp = *s++;
4217 if (PL_expect == XOPERATOR)
4222 if (PL_expect == XOPERATOR)
4225 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4232 if (PL_expect != XOPERATOR) {
4233 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4234 PL_expect = XOPERATOR;
4235 force_ident(PL_tokenbuf, '*');
4248 if (PL_expect == XOPERATOR) {
4252 PL_tokenbuf[0] = '%';
4253 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4254 sizeof PL_tokenbuf - 1, FALSE);
4255 if (!PL_tokenbuf[1]) {
4258 PL_pending_ident = '%';
4269 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4276 const char tmp = *s++;
4282 goto just_a_word_zero_gv;
4285 switch (PL_expect) {
4291 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4293 PL_bufptr = s; /* update in case we back off */
4299 PL_expect = XTERMBLOCK;
4302 stuffstart = s - SvPVX(PL_linestr) - 1;
4306 while (isIDFIRST_lazy_if(s,UTF)) {
4309 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4310 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4311 if (tmp < 0) tmp = -tmp;
4325 sv = newSVpvn(s, len);
4327 d = scan_str(d,TRUE,TRUE);
4329 /* MUST advance bufptr here to avoid bogus
4330 "at end of line" context messages from yyerror().
4332 PL_bufptr = s + len;
4333 yyerror("Unterminated attribute parameter in attribute list");
4337 return REPORT(0); /* EOF indicator */
4341 sv_catsv(sv, PL_lex_stuff);
4342 attrs = append_elem(OP_LIST, attrs,
4343 newSVOP(OP_CONST, 0, sv));
4344 SvREFCNT_dec(PL_lex_stuff);
4345 PL_lex_stuff = NULL;
4348 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4350 if (PL_in_my == KEY_our) {
4352 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4354 /* skip to avoid loading attributes.pm */
4356 deprecate(":unique");
4359 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4362 /* NOTE: any CV attrs applied here need to be part of
4363 the CVf_BUILTIN_ATTRS define in cv.h! */
4364 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4366 CvLVALUE_on(PL_compcv);
4368 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4370 CvLOCKED_on(PL_compcv);
4372 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4374 CvMETHOD_on(PL_compcv);
4376 /* After we've set the flags, it could be argued that
4377 we don't need to do the attributes.pm-based setting
4378 process, and shouldn't bother appending recognized
4379 flags. To experiment with that, uncomment the
4380 following "else". (Note that's already been
4381 uncommented. That keeps the above-applied built-in
4382 attributes from being intercepted (and possibly
4383 rejected) by a package's attribute routines, but is
4384 justified by the performance win for the common case
4385 of applying only built-in attributes.) */
4387 attrs = append_elem(OP_LIST, attrs,
4388 newSVOP(OP_CONST, 0,
4392 if (*s == ':' && s[1] != ':')
4395 break; /* require real whitespace or :'s */
4396 /* XXX losing whitespace on sequential attributes here */
4400 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4401 if (*s != ';' && *s != '}' && *s != tmp
4402 && (tmp != '=' || *s != ')')) {
4403 const char q = ((*s == '\'') ? '"' : '\'');
4404 /* If here for an expression, and parsed no attrs, back
4406 if (tmp == '=' && !attrs) {
4410 /* MUST advance bufptr here to avoid bogus "at end of line"
4411 context messages from yyerror().
4414 yyerror( (const char *)
4416 ? Perl_form(aTHX_ "Invalid separator character "
4417 "%c%c%c in attribute list", q, *s, q)
4418 : "Unterminated attribute list" ) );
4426 start_force(PL_curforce);
4427 NEXTVAL_NEXTTOKE.opval = attrs;
4428 CURMAD('_', PL_nextwhite);
4433 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4434 (s - SvPVX(PL_linestr)) - stuffstart);
4442 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4443 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4451 const char tmp = *s++;
4456 const char tmp = *s++;
4464 if (PL_lex_brackets <= 0)
4465 yyerror("Unmatched right square bracket");
4468 if (PL_lex_state == LEX_INTERPNORMAL) {
4469 if (PL_lex_brackets == 0) {
4470 if (*s == '-' && s[1] == '>')
4471 PL_lex_state = LEX_INTERPENDMAYBE;
4472 else if (*s != '[' && *s != '{')
4473 PL_lex_state = LEX_INTERPEND;
4480 if (PL_lex_brackets > 100) {
4481 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4483 switch (PL_expect) {
4485 if (PL_lex_formbrack) {
4489 if (PL_oldoldbufptr == PL_last_lop)
4490 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4492 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4493 OPERATOR(HASHBRACK);
4495 while (s < PL_bufend && SPACE_OR_TAB(*s))
4498 PL_tokenbuf[0] = '\0';
4499 if (d < PL_bufend && *d == '-') {
4500 PL_tokenbuf[0] = '-';
4502 while (d < PL_bufend && SPACE_OR_TAB(*d))
4505 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4506 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4508 while (d < PL_bufend && SPACE_OR_TAB(*d))
4511 const char minus = (PL_tokenbuf[0] == '-');
4512 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4520 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4525 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4530 if (PL_oldoldbufptr == PL_last_lop)
4531 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4533 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4536 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4538 /* This hack is to get the ${} in the message. */
4540 yyerror("syntax error");
4543 OPERATOR(HASHBRACK);
4545 /* This hack serves to disambiguate a pair of curlies
4546 * as being a block or an anon hash. Normally, expectation
4547 * determines that, but in cases where we're not in a
4548 * position to expect anything in particular (like inside
4549 * eval"") we have to resolve the ambiguity. This code
4550 * covers the case where the first term in the curlies is a
4551 * quoted string. Most other cases need to be explicitly
4552 * disambiguated by prepending a "+" before the opening
4553 * curly in order to force resolution as an anon hash.
4555 * XXX should probably propagate the outer expectation
4556 * into eval"" to rely less on this hack, but that could
4557 * potentially break current behavior of eval"".
4561 if (*s == '\'' || *s == '"' || *s == '`') {
4562 /* common case: get past first string, handling escapes */
4563 for (t++; t < PL_bufend && *t != *s;)
4564 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4568 else if (*s == 'q') {
4571 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4574 /* skip q//-like construct */
4576 char open, close, term;
4579 while (t < PL_bufend && isSPACE(*t))
4581 /* check for q => */
4582 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4583 OPERATOR(HASHBRACK);
4587 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4591 for (t++; t < PL_bufend; t++) {
4592 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4594 else if (*t == open)
4598 for (t++; t < PL_bufend; t++) {
4599 if (*t == '\\' && t+1 < PL_bufend)
4601 else if (*t == close && --brackets <= 0)
4603 else if (*t == open)
4610 /* skip plain q word */
4611 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4614 else if (isALNUM_lazy_if(t,UTF)) {
4616 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4619 while (t < PL_bufend && isSPACE(*t))
4621 /* if comma follows first term, call it an anon hash */
4622 /* XXX it could be a comma expression with loop modifiers */
4623 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4624 || (*t == '=' && t[1] == '>')))
4625 OPERATOR(HASHBRACK);
4626 if (PL_expect == XREF)
4629 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4635 pl_yylval.ival = CopLINE(PL_curcop);
4636 if (isSPACE(*s) || *s == '#')
4637 PL_copline = NOLINE; /* invalidate current command line number */
4642 if (PL_lex_brackets <= 0)
4643 yyerror("Unmatched right curly bracket");
4645 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4646 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4647 PL_lex_formbrack = 0;
4648 if (PL_lex_state == LEX_INTERPNORMAL) {
4649 if (PL_lex_brackets == 0) {
4650 if (PL_expect & XFAKEBRACK) {
4651 PL_expect &= XENUMMASK;
4652 PL_lex_state = LEX_INTERPEND;
4657 PL_thiswhite = newSVpvs("");
4658 sv_catpvn(PL_thiswhite,"}",1);
4661 return yylex(); /* ignore fake brackets */
4663 if (*s == '-' && s[1] == '>')
4664 PL_lex_state = LEX_INTERPENDMAYBE;
4665 else if (*s != '[' && *s != '{')
4666 PL_lex_state = LEX_INTERPEND;
4669 if (PL_expect & XFAKEBRACK) {
4670 PL_expect &= XENUMMASK;
4672 return yylex(); /* ignore fake brackets */
4674 start_force(PL_curforce);
4676 curmad('X', newSVpvn(s-1,1));
4677 CURMAD('_', PL_thiswhite);
4682 PL_thistoken = newSVpvs("");
4690 if (PL_expect == XOPERATOR) {
4691 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4692 && isIDFIRST_lazy_if(s,UTF))
4694 CopLINE_dec(PL_curcop);
4695 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4696 CopLINE_inc(PL_curcop);
4701 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4703 PL_expect = XOPERATOR;
4704 force_ident(PL_tokenbuf, '&');
4708 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4720 const char tmp = *s++;
4727 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4728 && strchr("+-*/%.^&|<",tmp))
4729 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4730 "Reversed %c= operator",(int)tmp);
4732 if (PL_expect == XSTATE && isALPHA(tmp) &&
4733 (s == PL_linestart+1 || s[-2] == '\n') )
4735 if (PL_in_eval && !PL_rsfp) {
4740 if (strnEQ(s,"=cut",4)) {
4756 PL_thiswhite = newSVpvs("");
4757 sv_catpvn(PL_thiswhite, PL_linestart,
4758 PL_bufend - PL_linestart);
4762 PL_doextract = TRUE;
4766 if (PL_lex_brackets < PL_lex_formbrack) {
4768 #ifdef PERL_STRICT_CR
4769 while (SPACE_OR_TAB(*t))
4771 while (SPACE_OR_TAB(*t) || *t == '\r')
4774 if (*t == '\n' || *t == '#') {
4783 if (PL_expect == XSTATE && s[1] == '!' && s[2] == '!') {
4789 const char tmp = *s++;
4791 /* was this !=~ where !~ was meant?
4792 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4794 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4795 const char *t = s+1;
4797 while (t < PL_bufend && isSPACE(*t))
4800 if (*t == '/' || *t == '?' ||
4801 ((*t == 'm' || *t == 's' || *t == 'y')
4802 && !isALNUM(t[1])) ||
4803 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4804 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4805 "!=~ should be !~");
4815 if (PL_expect != XOPERATOR) {
4816 if (s[1] != '<' && !strchr(s,'>'))
4819 s = scan_heredoc(s);
4821 s = scan_inputsymbol(s);
4822 TERM(sublex_start());
4828 SHop(OP_LEFT_SHIFT);
4842 const char tmp = *s++;
4844 SHop(OP_RIGHT_SHIFT);
4845 else if (tmp == '=')
4854 if (PL_expect == XOPERATOR) {
4855 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4857 deprecate_old(commaless_variable_list);
4858 return REPORT(','); /* grandfather non-comma-format format */
4862 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4863 PL_tokenbuf[0] = '@';
4864 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4865 sizeof PL_tokenbuf - 1, FALSE);
4866 if (PL_expect == XOPERATOR)
4867 no_op("Array length", s);
4868 if (!PL_tokenbuf[1])
4870 PL_expect = XOPERATOR;
4871 PL_pending_ident = '#';
4875 PL_tokenbuf[0] = '$';
4876 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4877 sizeof PL_tokenbuf - 1, FALSE);
4878 if (PL_expect == XOPERATOR)
4880 if (!PL_tokenbuf[1]) {
4882 yyerror("Final $ should be \\$ or $name");
4886 /* This kludge not intended to be bulletproof. */
4887 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4888 pl_yylval.opval = newSVOP(OP_CONST, 0,
4889 newSViv(CopARYBASE_get(&PL_compiling)));
4890 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4896 const char tmp = *s;
4897 if (PL_lex_state == LEX_NORMAL)
4900 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4901 && intuit_more(s)) {
4903 PL_tokenbuf[0] = '@';
4904 if (ckWARN(WARN_SYNTAX)) {
4907 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4910 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4911 while (t < PL_bufend && *t != ']')
4913 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4914 "Multidimensional syntax %.*s not supported",
4915 (int)((t - PL_bufptr) + 1), PL_bufptr);
4919 else if (*s == '{') {
4921 PL_tokenbuf[0] = '%';
4922 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4923 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4925 char tmpbuf[sizeof PL_tokenbuf];
4928 } while (isSPACE(*t));
4929 if (isIDFIRST_lazy_if(t,UTF)) {
4931 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4935 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4936 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4937 "You need to quote \"%s\"",
4944 PL_expect = XOPERATOR;
4945 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4946 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4947 if (!islop || PL_last_lop_op == OP_GREPSTART)
4948 PL_expect = XOPERATOR;
4949 else if (strchr("$@\"'`q", *s))
4950 PL_expect = XTERM; /* e.g. print $fh "foo" */
4951 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4952 PL_expect = XTERM; /* e.g. print $fh &sub */
4953 else if (isIDFIRST_lazy_if(s,UTF)) {
4954 char tmpbuf[sizeof PL_tokenbuf];
4956 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4957 if ((t2 = keyword(tmpbuf, len, 0))) {
4958 /* binary operators exclude handle interpretations */
4970 PL_expect = XTERM; /* e.g. print $fh length() */
4975 PL_expect = XTERM; /* e.g. print $fh subr() */
4978 else if (isDIGIT(*s))
4979 PL_expect = XTERM; /* e.g. print $fh 3 */
4980 else if (*s == '.' && isDIGIT(s[1]))
4981 PL_expect = XTERM; /* e.g. print $fh .3 */
4982 else if ((*s == '?' || *s == '-' || *s == '+')
4983 && !isSPACE(s[1]) && s[1] != '=')
4984 PL_expect = XTERM; /* e.g. print $fh -1 */
4985 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4987 PL_expect = XTERM; /* e.g. print $fh /.../
4988 XXX except DORDOR operator
4990 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4992 PL_expect = XTERM; /* print $fh <<"EOF" */
4995 PL_pending_ident = '$';
4999 if (PL_expect == XOPERATOR)
5001 PL_tokenbuf[0] = '@';
5002 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
5003 if (!PL_tokenbuf[1]) {
5006 if (PL_lex_state == LEX_NORMAL)
5008 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5010 PL_tokenbuf[0] = '%';
5012 /* Warn about @ where they meant $. */
5013 if (*s == '[' || *s == '{') {
5014 if (ckWARN(WARN_SYNTAX)) {
5015 const char *t = s + 1;
5016 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5018 if (*t == '}' || *t == ']') {
5020 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5021 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5022 "Scalar value %.*s better written as $%.*s",
5023 (int)(t-PL_bufptr), PL_bufptr,
5024 (int)(t-PL_bufptr-1), PL_bufptr+1);
5029 PL_pending_ident = '@';
5032 case '/': /* may be division, defined-or, or pattern */
5033 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5037 case '?': /* may either be conditional or pattern */
5038 if (PL_expect == XSTATE && s[1] == '?' && s[2] == '?') {
5042 if (PL_expect == XOPERATOR) {
5050 /* A // operator. */
5060 /* Disable warning on "study /blah/" */
5061 if (PL_oldoldbufptr == PL_last_uni
5062 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5063 || memNE(PL_last_uni, "study", 5)
5064 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5067 s = scan_pat(s,OP_MATCH);
5068 TERM(sublex_start());
5072 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5073 #ifdef PERL_STRICT_CR
5076 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5078 && (s == PL_linestart || s[-1] == '\n') )
5080 PL_lex_formbrack = 0;
5084 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5088 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5094 pl_yylval.ival = OPf_SPECIAL;
5100 if (PL_expect != XOPERATOR)
5105 case '0': case '1': case '2': case '3': case '4':
5106 case '5': case '6': case '7': case '8': case '9':
5107 s = scan_num(s, &pl_yylval);
5108 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5109 if (PL_expect == XOPERATOR)
5114 s = scan_str(s,!!PL_madskills,FALSE);
5115 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5116 if (PL_expect == XOPERATOR) {
5117 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5119 deprecate_old(commaless_variable_list);
5120 return REPORT(','); /* grandfather non-comma-format format */
5127 pl_yylval.ival = OP_CONST;
5128 TERM(sublex_start());
5131 s = scan_str(s,!!PL_madskills,FALSE);
5132 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5133 if (PL_expect == XOPERATOR) {
5134 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5136 deprecate_old(commaless_variable_list);
5137 return REPORT(','); /* grandfather non-comma-format format */
5144 pl_yylval.ival = OP_CONST;
5145 /* FIXME. I think that this can be const if char *d is replaced by
5146 more localised variables. */
5147 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5148 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5149 pl_yylval.ival = OP_STRINGIFY;
5153 TERM(sublex_start());
5156 s = scan_str(s,!!PL_madskills,FALSE);
5157 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5158 if (PL_expect == XOPERATOR)
5159 no_op("Backticks",s);
5162 readpipe_override();
5163 TERM(sublex_start());
5167 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5168 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5170 if (PL_expect == XOPERATOR)
5171 no_op("Backslash",s);
5175 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5176 char *start = s + 2;
5177 while (isDIGIT(*start) || *start == '_')
5179 if (*start == '.' && isDIGIT(start[1])) {
5180 s = scan_num(s, &pl_yylval);
5183 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5184 else if (!isALPHA(*start) && (PL_expect == XTERM
5185 || PL_expect == XREF || PL_expect == XSTATE
5186 || PL_expect == XTERMORDORDOR)) {
5187 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5189 s = scan_num(s, &pl_yylval);
5196 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5238 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5240 /* Some keywords can be followed by any delimiter, including ':' */
5241 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5242 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5243 (PL_tokenbuf[0] == 'q' &&
5244 strchr("qwxr", PL_tokenbuf[1])))));
5246 /* x::* is just a word, unless x is "CORE" */
5247 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5251 while (d < PL_bufend && isSPACE(*d))
5252 d++; /* no comments skipped here, or s### is misparsed */
5254 /* Is this a label? */
5255 if (!tmp && PL_expect == XSTATE
5256 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5258 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5263 /* Check for keywords */
5264 tmp = keyword(PL_tokenbuf, len, 0);
5266 /* Is this a word before a => operator? */
5267 if (*d == '=' && d[1] == '>') {
5270 = (OP*)newSVOP(OP_CONST, 0,
5271 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5272 pl_yylval.opval->op_private = OPpCONST_BARE;
5276 if (tmp < 0) { /* second-class keyword? */
5277 GV *ogv = NULL; /* override (winner) */
5278 GV *hgv = NULL; /* hidden (loser) */
5279 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5281 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5284 if (GvIMPORTED_CV(gv))
5286 else if (! CvMETHOD(cv))
5290 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5291 (gv = *gvp) && isGV_with_GP(gv) &&
5292 GvCVu(gv) && GvIMPORTED_CV(gv))
5299 tmp = 0; /* overridden by import or by GLOBAL */
5302 && -tmp==KEY_lock /* XXX generalizable kludge */
5305 tmp = 0; /* any sub overrides "weak" keyword */
5307 else { /* no override */
5309 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5310 Perl_warner(aTHX_ packWARN(WARN_MISC),
5311 "dump() better written as CORE::dump()");
5315 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5316 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5317 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5318 "Ambiguous call resolved as CORE::%s(), %s",
5319 GvENAME(hgv), "qualify as such or use &");
5326 default: /* not a keyword */
5327 /* Trade off - by using this evil construction we can pull the
5328 variable gv into the block labelled keylookup. If not, then
5329 we have to give it function scope so that the goto from the
5330 earlier ':' case doesn't bypass the initialisation. */
5332 just_a_word_zero_gv:
5340 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5343 SV *nextPL_nextwhite = 0;
5347 /* Get the rest if it looks like a package qualifier */
5349 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5351 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5354 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5355 *s == '\'' ? "'" : "::");
5360 if (PL_expect == XOPERATOR) {
5361 if (PL_bufptr == PL_linestart) {
5362 CopLINE_dec(PL_curcop);
5363 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
5364 CopLINE_inc(PL_curcop);
5367 no_op("Bareword",s);
5370 /* Look for a subroutine with this name in current package,
5371 unless name is "Foo::", in which case Foo is a bearword
5372 (and a package name). */
5374 if (len > 2 && !PL_madskills &&
5375 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5377 if (ckWARN(WARN_BAREWORD)
5378 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5379 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5380 "Bareword \"%s\" refers to nonexistent package",
5383 PL_tokenbuf[len] = '\0';
5389 /* Mustn't actually add anything to a symbol table.
5390 But also don't want to "initialise" any placeholder
5391 constants that might already be there into full
5392 blown PVGVs with attached PVCV. */
5393 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5394 GV_NOADD_NOINIT, SVt_PVCV);
5399 /* if we saw a global override before, get the right name */
5402 sv = newSVpvs("CORE::GLOBAL::");
5403 sv_catpv(sv,PL_tokenbuf);
5406 /* If len is 0, newSVpv does strlen(), which is correct.
5407 If len is non-zero, then it will be the true length,
5408 and so the scalar will be created correctly. */
5409 sv = newSVpv(PL_tokenbuf,len);
5412 if (PL_madskills && !PL_thistoken) {
5413 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5414 PL_thistoken = newSVpvn(start,s - start);
5415 PL_realtokenstart = s - SvPVX(PL_linestr);
5419 /* Presume this is going to be a bareword of some sort. */
5422 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5423 pl_yylval.opval->op_private = OPpCONST_BARE;
5424 /* UTF-8 package name? */
5425 if (UTF && !IN_BYTES &&
5426 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5429 /* And if "Foo::", then that's what it certainly is. */
5434 /* Do the explicit type check so that we don't need to force
5435 the initialisation of the symbol table to have a real GV.
5436 Beware - gv may not really be a PVGV, cv may not really be
5437 a PVCV, (because of the space optimisations that gv_init
5438 understands) But they're true if for this symbol there is
5439 respectively a typeglob and a subroutine.
5441 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5442 /* Real typeglob, so get the real subroutine: */
5444 /* A proxy for a subroutine in this package? */
5445 : SvOK(gv) ? (CV *) gv : NULL)
5448 /* See if it's the indirect object for a list operator. */
5450 if (PL_oldoldbufptr &&
5451 PL_oldoldbufptr < PL_bufptr &&
5452 (PL_oldoldbufptr == PL_last_lop
5453 || PL_oldoldbufptr == PL_last_uni) &&
5454 /* NO SKIPSPACE BEFORE HERE! */
5455 (PL_expect == XREF ||
5456 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5458 bool immediate_paren = *s == '(';
5460 /* (Now we can afford to cross potential line boundary.) */
5461 s = SKIPSPACE2(s,nextPL_nextwhite);
5463 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5466 /* Two barewords in a row may indicate method call. */
5468 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5469 (tmp = intuit_method(s, gv, cv)))
5472 /* If not a declared subroutine, it's an indirect object. */
5473 /* (But it's an indir obj regardless for sort.) */
5474 /* Also, if "_" follows a filetest operator, it's a bareword */
5477 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5479 (PL_last_lop_op != OP_MAPSTART &&
5480 PL_last_lop_op != OP_GREPSTART))))
5481 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5482 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5485 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5490 PL_expect = XOPERATOR;
5493 s = SKIPSPACE2(s,nextPL_nextwhite);
5494 PL_nextwhite = nextPL_nextwhite;
5499 /* Is this a word before a => operator? */
5500 if (*s == '=' && s[1] == '>' && !pkgname) {
5502 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5503 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5504 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5508 /* If followed by a paren, it's certainly a subroutine. */
5513 while (SPACE_OR_TAB(*d))
5515 if (*d == ')' && (sv = gv_const_sv(gv))) {
5522 PL_nextwhite = PL_thiswhite;
5525 start_force(PL_curforce);
5527 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5528 PL_expect = XOPERATOR;
5531 PL_nextwhite = nextPL_nextwhite;
5532 curmad('X', PL_thistoken);
5533 PL_thistoken = newSVpvs("");
5541 /* If followed by var or block, call it a method (unless sub) */
5543 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5544 PL_last_lop = PL_oldbufptr;
5545 PL_last_lop_op = OP_METHOD;
5549 /* If followed by a bareword, see if it looks like indir obj. */
5552 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5553 && (tmp = intuit_method(s, gv, cv)))
5556 /* Not a method, so call it a subroutine (if defined) */
5559 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5560 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5561 "Ambiguous use of -%s resolved as -&%s()",
5562 PL_tokenbuf, PL_tokenbuf);
5563 /* Check for a constant sub */
5564 if ((sv = gv_const_sv(gv))) {
5566 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5567 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5568 pl_yylval.opval->op_private = 0;
5572 /* Resolve to GV now. */
5573 if (SvTYPE(gv) != SVt_PVGV) {
5574 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5575 assert (SvTYPE(gv) == SVt_PVGV);
5576 /* cv must have been some sort of placeholder, so
5577 now needs replacing with a real code reference. */
5581 op_free(pl_yylval.opval);
5582 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5583 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5584 PL_last_lop = PL_oldbufptr;
5585 PL_last_lop_op = OP_ENTERSUB;
5586 /* Is there a prototype? */
5594 const char *proto = SvPV_const((SV*)cv, protolen);
5597 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5599 while (*proto == ';')
5601 if (*proto == '&' && *s == '{') {
5603 sv_setpvs(PL_subname, "__ANON__");
5605 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5612 PL_nextwhite = PL_thiswhite;
5615 start_force(PL_curforce);
5616 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5619 PL_nextwhite = nextPL_nextwhite;
5620 curmad('X', PL_thistoken);
5621 PL_thistoken = newSVpvs("");
5628 /* Guess harder when madskills require "best effort". */
5629 if (PL_madskills && (!gv || !GvCVu(gv))) {
5630 int probable_sub = 0;
5631 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5633 else if (isALPHA(*s)) {
5637 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5638 if (!keyword(tmpbuf, tmplen, 0))
5641 while (d < PL_bufend && isSPACE(*d))
5643 if (*d == '=' && d[1] == '>')
5648 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5649 op_free(pl_yylval.opval);
5650 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5651 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5652 PL_last_lop = PL_oldbufptr;
5653 PL_last_lop_op = OP_ENTERSUB;
5654 PL_nextwhite = PL_thiswhite;
5656 start_force(PL_curforce);
5657 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5659 PL_nextwhite = nextPL_nextwhite;
5660 curmad('X', PL_thistoken);
5661 PL_thistoken = newSVpvs("");
5666 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5673 /* Call it a bare word */
5676 if (PL_hints & HINT_STRICT_SUBS)
5677 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5679 if (lastchar != '-') {
5680 if (ckWARN(WARN_RESERVED)) {
5684 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5685 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5692 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5693 && ckWARN_d(WARN_AMBIGUOUS)) {
5694 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5695 "Operator or semicolon missing before %c%s",
5696 lastchar, PL_tokenbuf);
5697 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5698 "Ambiguous use of %c resolved as operator %c",
5699 lastchar, lastchar);
5705 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5706 newSVpv(CopFILE(PL_curcop),0));
5710 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5711 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5714 case KEY___PACKAGE__:
5715 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5717 ? newSVhek(HvNAME_HEK(PL_curstash))
5724 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5725 const char *pname = "main";
5726 if (PL_tokenbuf[2] == 'D')
5727 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5728 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5732 GvIOp(gv) = newIO();
5733 IoIFP(GvIOp(gv)) = PL_rsfp;
5734 #if defined(HAS_FCNTL) && defined(F_SETFD)
5736 const int fd = PerlIO_fileno(PL_rsfp);
5737 fcntl(fd,F_SETFD,fd >= 3);
5740 /* Mark this internal pseudo-handle as clean */
5741 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5742 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5743 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5745 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5746 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5747 /* if the script was opened in binmode, we need to revert
5748 * it to text mode for compatibility; but only iff it has CRs
5749 * XXX this is a questionable hack at best. */
5750 if (PL_bufend-PL_bufptr > 2
5751 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5754 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5755 loc = PerlIO_tell(PL_rsfp);
5756 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5759 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5761 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5762 #endif /* NETWARE */
5763 #ifdef PERLIO_IS_STDIO /* really? */
5764 # if defined(__BORLANDC__)
5765 /* XXX see note in do_binmode() */
5766 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5770 PerlIO_seek(PL_rsfp, loc, 0);
5774 #ifdef PERLIO_LAYERS
5777 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5778 else if (PL_encoding) {
5785 XPUSHs(PL_encoding);
5787 call_method("name", G_SCALAR);
5791 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5792 Perl_form(aTHX_ ":encoding(%"SVf")",
5801 if (PL_realtokenstart >= 0) {
5802 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5804 PL_endwhite = newSVpvs("");
5805 sv_catsv(PL_endwhite, PL_thiswhite);
5807 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5808 PL_realtokenstart = -1;
5810 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5811 SvCUR(PL_endwhite))) != NULL) ;
5826 if (PL_expect == XSTATE) {
5833 if (*s == ':' && s[1] == ':') {
5836 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5837 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5838 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5841 else if (tmp == KEY_require || tmp == KEY_do)
5842 /* that's a way to remember we saw "CORE::" */
5855 LOP(OP_ACCEPT,XTERM);
5861 LOP(OP_ATAN2,XTERM);
5867 LOP(OP_BINMODE,XTERM);
5870 LOP(OP_BLESS,XTERM);
5879 /* When 'use switch' is in effect, continue has a dual
5880 life as a control operator. */
5882 if (!FEATURE_IS_ENABLED("switch"))
5885 /* We have to disambiguate the two senses of
5886 "continue". If the next token is a '{' then
5887 treat it as the start of a continue block;
5888 otherwise treat it as a control operator.
5900 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5917 if (!PL_cryptseen) {
5918 PL_cryptseen = TRUE;
5922 LOP(OP_CRYPT,XTERM);
5925 LOP(OP_CHMOD,XTERM);
5928 LOP(OP_CHOWN,XTERM);
5931 LOP(OP_CONNECT,XTERM);
5950 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5951 if (orig_keyword == KEY_do) {
5960 PL_hints |= HINT_BLOCK_SCOPE;
5970 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5971 LOP(OP_DBMOPEN,XTERM);
5977 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5984 pl_yylval.ival = CopLINE(PL_curcop);
6000 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
6001 UNIBRACK(OP_ENTEREVAL);
6015 case KEY_endhostent:
6021 case KEY_endservent:
6024 case KEY_endprotoent:
6035 pl_yylval.ival = CopLINE(PL_curcop);
6037 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6040 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6043 if ((PL_bufend - p) >= 3 &&
6044 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6046 else if ((PL_bufend - p) >= 4 &&
6047 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6050 if (isIDFIRST_lazy_if(p,UTF)) {
6051 p = scan_ident(p, PL_bufend,
6052 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6056 Perl_croak(aTHX_ "Missing $ on loop variable");
6058 s = SvPVX(PL_linestr) + soff;
6064 LOP(OP_FORMLINE,XTERM);
6070 LOP(OP_FCNTL,XTERM);
6076 LOP(OP_FLOCK,XTERM);
6085 LOP(OP_GREPSTART, XREF);
6088 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6103 case KEY_getpriority:
6104 LOP(OP_GETPRIORITY,XTERM);
6106 case KEY_getprotobyname:
6109 case KEY_getprotobynumber:
6110 LOP(OP_GPBYNUMBER,XTERM);
6112 case KEY_getprotoent:
6124 case KEY_getpeername:
6125 UNI(OP_GETPEERNAME);
6127 case KEY_gethostbyname:
6130 case KEY_gethostbyaddr:
6131 LOP(OP_GHBYADDR,XTERM);
6133 case KEY_gethostent:
6136 case KEY_getnetbyname:
6139 case KEY_getnetbyaddr:
6140 LOP(OP_GNBYADDR,XTERM);
6145 case KEY_getservbyname:
6146 LOP(OP_GSBYNAME,XTERM);
6148 case KEY_getservbyport:
6149 LOP(OP_GSBYPORT,XTERM);
6151 case KEY_getservent:
6154 case KEY_getsockname:
6155 UNI(OP_GETSOCKNAME);
6157 case KEY_getsockopt:
6158 LOP(OP_GSOCKOPT,XTERM);
6173 pl_yylval.ival = CopLINE(PL_curcop);
6183 pl_yylval.ival = CopLINE(PL_curcop);
6187 LOP(OP_INDEX,XTERM);
6193 LOP(OP_IOCTL,XTERM);
6205 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6237 LOP(OP_LISTEN,XTERM);
6246 s = scan_pat(s,OP_MATCH);
6247 TERM(sublex_start());
6250 LOP(OP_MAPSTART, XREF);
6253 LOP(OP_MKDIR,XTERM);
6256 LOP(OP_MSGCTL,XTERM);
6259 LOP(OP_MSGGET,XTERM);
6262 LOP(OP_MSGRCV,XTERM);
6265 LOP(OP_MSGSND,XTERM);
6270 PL_in_my = (U16)tmp;
6272 if (isIDFIRST_lazy_if(s,UTF)) {
6276 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6277 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6279 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6280 if (!PL_in_my_stash) {
6283 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6287 if (PL_madskills) { /* just add type to declarator token */
6288 sv_catsv(PL_thistoken, PL_nextwhite);
6290 sv_catpvn(PL_thistoken, start, s - start);
6298 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6305 s = tokenize_use(0, s);
6309 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6316 if (isIDFIRST_lazy_if(s,UTF)) {
6318 for (d = s; isALNUM_lazy_if(d,UTF);)
6320 for (t=d; isSPACE(*t);)
6322 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6324 && !(t[0] == '=' && t[1] == '>')
6326 int parms_len = (int)(d-s);
6327 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6328 "Precedence problem: open %.*s should be open(%.*s)",
6329 parms_len, s, parms_len, s);
6335 pl_yylval.ival = OP_OR;
6345 LOP(OP_OPEN_DIR,XTERM);
6348 checkcomma(s,PL_tokenbuf,"filehandle");
6352 checkcomma(s,PL_tokenbuf,"filehandle");
6371 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6375 LOP(OP_PIPE_OP,XTERM);
6378 s = scan_str(s,!!PL_madskills,FALSE);
6381 pl_yylval.ival = OP_CONST;
6382 TERM(sublex_start());
6388 s = scan_str(s,!!PL_madskills,FALSE);
6391 PL_expect = XOPERATOR;
6393 if (SvCUR(PL_lex_stuff)) {
6396 d = SvPV_force(PL_lex_stuff, len);
6398 for (; isSPACE(*d) && len; --len, ++d)
6403 if (!warned && ckWARN(WARN_QW)) {
6404 for (; !isSPACE(*d) && len; --len, ++d) {
6406 Perl_warner(aTHX_ packWARN(WARN_QW),
6407 "Possible attempt to separate words with commas");
6410 else if (*d == '#') {
6411 Perl_warner(aTHX_ packWARN(WARN_QW),
6412 "Possible attempt to put comments in qw() list");
6418 for (; !isSPACE(*d) && len; --len, ++d)
6421 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6422 words = append_elem(OP_LIST, words,
6423 newSVOP(OP_CONST, 0, tokeq(sv)));
6427 start_force(PL_curforce);
6428 NEXTVAL_NEXTTOKE.opval = words;
6433 SvREFCNT_dec(PL_lex_stuff);
6434 PL_lex_stuff = NULL;
6440 s = scan_str(s,!!PL_madskills,FALSE);
6443 pl_yylval.ival = OP_STRINGIFY;
6444 if (SvIVX(PL_lex_stuff) == '\'')
6445 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6446 TERM(sublex_start());
6449 s = scan_pat(s,OP_QR);
6450 TERM(sublex_start());
6453 s = scan_str(s,!!PL_madskills,FALSE);
6456 readpipe_override();
6457 TERM(sublex_start());
6465 s = force_version(s, FALSE);
6467 else if (*s != 'v' || !isDIGIT(s[1])
6468 || (s = force_version(s, TRUE), *s == 'v'))
6470 *PL_tokenbuf = '\0';
6471 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6472 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6473 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6475 yyerror("<> should be quotes");
6477 if (orig_keyword == KEY_require) {
6485 PL_last_uni = PL_oldbufptr;
6486 PL_last_lop_op = OP_REQUIRE;
6488 return REPORT( (int)REQUIRE );
6494 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6498 LOP(OP_RENAME,XTERM);
6507 LOP(OP_RINDEX,XTERM);
6516 UNIDOR(OP_READLINE);
6519 UNIDOR(OP_BACKTICK);
6528 LOP(OP_REVERSE,XTERM);
6531 UNIDOR(OP_READLINK);
6538 if (pl_yylval.opval)
6539 TERM(sublex_start());
6541 TOKEN(1); /* force error */
6544 checkcomma(s,PL_tokenbuf,"filehandle");
6554 LOP(OP_SELECT,XTERM);
6560 LOP(OP_SEMCTL,XTERM);
6563 LOP(OP_SEMGET,XTERM);
6566 LOP(OP_SEMOP,XTERM);
6572 LOP(OP_SETPGRP,XTERM);
6574 case KEY_setpriority:
6575 LOP(OP_SETPRIORITY,XTERM);
6577 case KEY_sethostent:
6583 case KEY_setservent:
6586 case KEY_setprotoent:
6596 LOP(OP_SEEKDIR,XTERM);
6598 case KEY_setsockopt:
6599 LOP(OP_SSOCKOPT,XTERM);
6605 LOP(OP_SHMCTL,XTERM);
6608 LOP(OP_SHMGET,XTERM);
6611 LOP(OP_SHMREAD,XTERM);
6614 LOP(OP_SHMWRITE,XTERM);
6617 LOP(OP_SHUTDOWN,XTERM);
6626 LOP(OP_SOCKET,XTERM);
6628 case KEY_socketpair:
6629 LOP(OP_SOCKPAIR,XTERM);
6632 checkcomma(s,PL_tokenbuf,"subroutine name");
6634 if (*s == ';' || *s == ')') /* probably a close */
6635 Perl_croak(aTHX_ "sort is now a reserved word");
6637 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6641 LOP(OP_SPLIT,XTERM);
6644 LOP(OP_SPRINTF,XTERM);
6647 LOP(OP_SPLICE,XTERM);
6662 LOP(OP_SUBSTR,XTERM);
6668 char tmpbuf[sizeof PL_tokenbuf];
6669 SSize_t tboffset = 0;
6670 expectation attrful;
6671 bool have_name, have_proto;
6672 const int key = tmp;
6677 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6678 SV *subtoken = newSVpvn(tstart, s - tstart);
6682 s = SKIPSPACE2(s,tmpwhite);
6687 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6688 (*s == ':' && s[1] == ':'))
6691 SV *nametoke = NULL;
6695 attrful = XATTRBLOCK;
6696 /* remember buffer pos'n for later force_word */
6697 tboffset = s - PL_oldbufptr;
6698 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6701 nametoke = newSVpvn(s, d - s);
6703 if (memchr(tmpbuf, ':', len))
6704 sv_setpvn(PL_subname, tmpbuf, len);
6706 sv_setsv(PL_subname,PL_curstname);
6707 sv_catpvs(PL_subname,"::");
6708 sv_catpvn(PL_subname,tmpbuf,len);
6715 CURMAD('X', nametoke);
6716 CURMAD('_', tmpwhite);
6717 (void) force_word(PL_oldbufptr + tboffset, WORD,
6720 s = SKIPSPACE2(d,tmpwhite);
6727 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6728 PL_expect = XTERMBLOCK;
6729 attrful = XATTRTERM;
6730 sv_setpvn(PL_subname,"?",1);
6734 if (key == KEY_format) {
6736 PL_lex_formbrack = PL_lex_brackets + 1;
6738 PL_thistoken = subtoken;
6742 (void) force_word(PL_oldbufptr + tboffset, WORD,
6748 /* Look for a prototype */
6751 bool bad_proto = FALSE;
6752 bool in_brackets = FALSE;
6753 char greedy_proto = ' ';
6754 bool proto_after_greedy_proto = FALSE;
6755 bool must_be_last = FALSE;
6756 bool underscore = FALSE;
6757 bool seen_underscore = FALSE;
6758 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6760 s = scan_str(s,!!PL_madskills,FALSE);
6762 Perl_croak(aTHX_ "Prototype not terminated");
6763 /* strip spaces and check for bad characters */
6764 d = SvPVX(PL_lex_stuff);
6766 for (p = d; *p; ++p) {
6772 proto_after_greedy_proto = TRUE;
6773 if (!strchr("$@%*;[]&\\_", *p)) {
6785 else if ( *p == ']' ) {
6786 in_brackets = FALSE;
6788 else if ( (*p == '@' || *p == '%') &&
6789 ( tmp < 2 || d[tmp-2] != '\\' ) &&
6791 must_be_last = TRUE;
6794 else if ( *p == '_' ) {
6795 underscore = seen_underscore = TRUE;
6802 if (proto_after_greedy_proto)
6803 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6804 "Prototype after '%c' for %"SVf" : %s",
6805 greedy_proto, SVfARG(PL_subname), d);
6807 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6808 "Illegal character %sin prototype for %"SVf" : %s",
6809 seen_underscore ? "after '_' " : "",
6810 SVfARG(PL_subname), d);
6811 SvCUR_set(PL_lex_stuff, tmp);
6816 CURMAD('q', PL_thisopen);
6817 CURMAD('_', tmpwhite);
6818 CURMAD('=', PL_thisstuff);
6819 CURMAD('Q', PL_thisclose);
6820 NEXTVAL_NEXTTOKE.opval =
6821 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6822 PL_lex_stuff = NULL;
6825 s = SKIPSPACE2(s,tmpwhite);
6833 if (*s == ':' && s[1] != ':')
6834 PL_expect = attrful;
6835 else if (*s != '{' && key == KEY_sub) {
6837 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6839 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6846 curmad('^', newSVpvs(""));
6847 CURMAD('_', tmpwhite);
6851 PL_thistoken = subtoken;
6854 NEXTVAL_NEXTTOKE.opval =
6855 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6856 PL_lex_stuff = NULL;
6862 sv_setpvs(PL_subname, "__ANON__");
6864 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6868 (void) force_word(PL_oldbufptr + tboffset, WORD,
6877 LOP(OP_SYSTEM,XREF);
6880 LOP(OP_SYMLINK,XTERM);
6883 LOP(OP_SYSCALL,XTERM);
6886 LOP(OP_SYSOPEN,XTERM);
6889 LOP(OP_SYSSEEK,XTERM);
6892 LOP(OP_SYSREAD,XTERM);
6895 LOP(OP_SYSWRITE,XTERM);
6899 TERM(sublex_start());
6920 LOP(OP_TRUNCATE,XTERM);
6932 pl_yylval.ival = CopLINE(PL_curcop);
6936 pl_yylval.ival = CopLINE(PL_curcop);
6940 LOP(OP_UNLINK,XTERM);
6946 LOP(OP_UNPACK,XTERM);
6949 LOP(OP_UTIME,XTERM);
6955 LOP(OP_UNSHIFT,XTERM);
6958 s = tokenize_use(1, s);
6968 pl_yylval.ival = CopLINE(PL_curcop);
6972 pl_yylval.ival = CopLINE(PL_curcop);
6976 PL_hints |= HINT_BLOCK_SCOPE;
6983 LOP(OP_WAITPID,XTERM);
6992 ctl_l[0] = toCTRL('L');
6994 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6997 /* Make sure $^L is defined */
6998 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
7003 if (PL_expect == XOPERATOR)
7009 pl_yylval.ival = OP_XOR;
7014 TERM(sublex_start());
7019 #pragma segment Main
7023 S_pending_ident(pTHX)
7028 /* pit holds the identifier we read and pending_ident is reset */
7029 char pit = PL_pending_ident;
7030 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7031 /* All routes through this function want to know if there is a colon. */
7032 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7033 PL_pending_ident = 0;
7035 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7036 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7037 "### Pending identifier '%s'\n", PL_tokenbuf); });
7039 /* if we're in a my(), we can't allow dynamics here.
7040 $foo'bar has already been turned into $foo::bar, so
7041 just check for colons.
7043 if it's a legal name, the OP is a PADANY.
7046 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7048 yyerror(Perl_form(aTHX_ "No package name allowed for "
7049 "variable %s in \"our\"",
7051 tmp = allocmy(PL_tokenbuf);
7055 yyerror(Perl_form(aTHX_ PL_no_myglob,
7056 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7058 pl_yylval.opval = newOP(OP_PADANY, 0);
7059 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7065 build the ops for accesses to a my() variable.
7067 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7068 then used in a comparison. This catches most, but not
7069 all cases. For instance, it catches
7070 sort { my($a); $a <=> $b }
7072 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7073 (although why you'd do that is anyone's guess).
7078 tmp = pad_findmy(PL_tokenbuf);
7079 if (tmp != NOT_IN_PAD) {
7080 /* might be an "our" variable" */
7081 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7082 /* build ops for a bareword */
7083 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7084 HEK * const stashname = HvNAME_HEK(stash);
7085 SV * const sym = newSVhek(stashname);
7086 sv_catpvs(sym, "::");
7087 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7088 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7089 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7092 ? (GV_ADDMULTI | GV_ADDINEVAL)
7095 ((PL_tokenbuf[0] == '$') ? SVt_PV
7096 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7101 /* if it's a sort block and they're naming $a or $b */
7102 if (PL_last_lop_op == OP_SORT &&
7103 PL_tokenbuf[0] == '$' &&
7104 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7107 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7108 d < PL_bufend && *d != '\n';
7111 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7112 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7118 pl_yylval.opval = newOP(OP_PADANY, 0);
7119 pl_yylval.opval->op_targ = tmp;
7125 Whine if they've said @foo in a doublequoted string,
7126 and @foo isn't a variable we can find in the symbol
7129 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7130 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7132 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7133 && ckWARN(WARN_AMBIGUOUS)
7134 /* DO NOT warn for @- and @+ */
7135 && !( PL_tokenbuf[2] == '\0' &&
7136 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7139 /* Downgraded from fatal to warning 20000522 mjd */
7140 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7141 "Possible unintended interpolation of %s in string",
7146 /* build ops for a bareword */
7147 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7149 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7151 PL_tokenbuf + 1, tokenbuf_len - 1,
7152 /* If the identifier refers to a stash, don't autovivify it.
7153 * Change 24660 had the side effect of causing symbol table
7154 * hashes to always be defined, even if they were freshly
7155 * created and the only reference in the entire program was
7156 * the single statement with the defined %foo::bar:: test.
7157 * It appears that all code in the wild doing this actually
7158 * wants to know whether sub-packages have been loaded, so
7159 * by avoiding auto-vivifying symbol tables, we ensure that
7160 * defined %foo::bar:: continues to be false, and the existing
7161 * tests still give the expected answers, even though what
7162 * they're actually testing has now changed subtly.
7164 (*PL_tokenbuf == '%'
7165 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7168 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7169 ((PL_tokenbuf[0] == '$') ? SVt_PV
7170 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7176 * The following code was generated by perl_keyword.pl.
7180 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7184 PERL_ARGS_ASSERT_KEYWORD;
7188 case 1: /* 5 tokens of length 1 */
7220 case 2: /* 18 tokens of length 2 */
7366 case 3: /* 29 tokens of length 3 */
7370 if (name[1] == 'N' &&
7433 if (name[1] == 'i' &&
7465 if (name[1] == 'o' &&
7474 if (name[1] == 'e' &&
7483 if (name[1] == 'n' &&
7492 if (name[1] == 'o' &&
7501 if (name[1] == 'a' &&
7510 if (name[1] == 'o' &&
7572 if (name[1] == 'e' &&
7586 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7612 if (name[1] == 'i' &&
7621 if (name[1] == 's' &&
7630 if (name[1] == 'e' &&
7639 if (name[1] == 'o' &&
7651 case 4: /* 41 tokens of length 4 */
7655 if (name[1] == 'O' &&
7665 if (name[1] == 'N' &&
7675 if (name[1] == 'i' &&
7685 if (name[1] == 'h' &&
7695 if (name[1] == 'u' &&
7708 if (name[2] == 'c' &&
7717 if (name[2] == 's' &&
7726 if (name[2] == 'a' &&
7762 if (name[1] == 'o' &&
7775 if (name[2] == 't' &&
7784 if (name[2] == 'o' &&
7793 if (name[2] == 't' &&
7802 if (name[2] == 'e' &&
7815 if (name[1] == 'o' &&
7828 if (name[2] == 'y' &&
7837 if (name[2] == 'l' &&
7853 if (name[2] == 's' &&
7862 if (name[2] == 'n' &&
7871 if (name[2] == 'c' &&
7884 if (name[1] == 'e' &&
7894 if (name[1] == 'p' &&
7907 if (name[2] == 'c' &&
7916 if (name[2] == 'p' &&
7925 if (name[2] == 's' &&
7941 if (name[2] == 'n' &&
8011 if (name[2] == 'r' &&
8020 if (name[2] == 'r' &&
8029 if (name[2] == 'a' &&
8045 if (name[2] == 'l' &&
8107 if (name[2] == 'e' &&
8110 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8123 case 5: /* 39 tokens of length 5 */
8127 if (name[1] == 'E' &&
8138 if (name[1] == 'H' &&
8152 if (name[2] == 'a' &&
8162 if (name[2] == 'a' &&
8179 if (name[2] == 'e' &&
8189 if (name[2] == 'e' &&
8193 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8209 if (name[3] == 'i' &&
8218 if (name[3] == 'o' &&
8254 if (name[2] == 'o' &&
8264 if (name[2] == 'y' &&
8278 if (name[1] == 'l' &&
8292 if (name[2] == 'n' &&
8302 if (name[2] == 'o' &&
8316 if (name[1] == 'i' &&
8321 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8330 if (name[2] == 'd' &&
8340 if (name[2] == 'c' &&
8357 if (name[2] == 'c' &&
8367 if (name[2] == 't' &&
8381 if (name[1] == 'k' &&
8392 if (name[1] == 'r' &&
8406 if (name[2] == 's' &&
8416 if (name[2] == 'd' &&
8433 if (name[2] == 'm' &&
8443 if (name[2] == 'i' &&
8453 if (name[2] == 'e' &&
8463 if (name[2] == 'l' &&
8473 if (name[2] == 'a' &&
8486 if (name[3] == 't' &&
8489 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8495 if (name[3] == 'd' &&
8512 if (name[1] == 'i' &&
8526 if (name[2] == 'a' &&
8539 if (name[3] == 'e' &&
8574 if (name[2] == 'i' &&
8591 if (name[2] == 'i' &&
8601 if (name[2] == 'i' &&
8618 case 6: /* 33 tokens of length 6 */
8622 if (name[1] == 'c' &&
8637 if (name[2] == 'l' &&
8648 if (name[2] == 'r' &&
8663 if (name[1] == 'e' &&
8678 if (name[2] == 's' &&
8683 if(ckWARN_d(WARN_SYNTAX))
8684 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8690 if (name[2] == 'i' &&
8708 if (name[2] == 'l' &&
8719 if (name[2] == 'r' &&
8734 if (name[1] == 'm' &&
8749 if (name[2] == 'n' &&
8760 if (name[2] == 's' &&
8775 if (name[1] == 's' &&
8781 if (name[4] == 't' &&
8790 if (name[4] == 'e' &&
8799 if (name[4] == 'c' &&
8808 if (name[4] == 'n' &&
8824 if (name[1] == 'r' &&
8842 if (name[3] == 'a' &&
8852 if (name[3] == 'u' &&
8866 if (name[2] == 'n' &&
8884 if (name[2] == 'a' &&
8898 if (name[3] == 'e' &&
8911 if (name[4] == 't' &&
8920 if (name[4] == 'e' &&
8942 if (name[4] == 't' &&
8951 if (name[4] == 'e' &&
8967 if (name[2] == 'c' &&
8978 if (name[2] == 'l' &&
8989 if (name[2] == 'b' &&
9000 if (name[2] == 's' &&
9023 if (name[4] == 's' &&
9032 if (name[4] == 'n' &&
9045 if (name[3] == 'a' &&
9062 if (name[1] == 'a' &&
9077 case 7: /* 29 tokens of length 7 */
9081 if (name[1] == 'E' &&
9094 if (name[1] == '_' &&
9107 if (name[1] == 'i' &&
9114 return -KEY_binmode;
9120 if (name[1] == 'o' &&
9127 return -KEY_connect;
9136 if (name[2] == 'm' &&
9142 return -KEY_dbmopen;
9153 if (name[4] == 'u' &&
9157 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9163 if (name[4] == 'n' &&
9184 if (name[1] == 'o' &&
9197 if (name[1] == 'e' &&
9204 if (name[5] == 'r' &&
9207 return -KEY_getpgrp;
9213 if (name[5] == 'i' &&
9216 return -KEY_getppid;
9229 if (name[1] == 'c' &&
9236 return -KEY_lcfirst;
9242 if (name[1] == 'p' &&
9249 return -KEY_opendir;
9255 if (name[1] == 'a' &&
9273 if (name[3] == 'd' &&
9278 return -KEY_readdir;
9284 if (name[3] == 'u' &&
9295 if (name[3] == 'e' &&
9300 return -KEY_reverse;
9319 if (name[3] == 'k' &&
9324 return -KEY_seekdir;
9330 if (name[3] == 'p' &&
9335 return -KEY_setpgrp;
9345 if (name[2] == 'm' &&
9351 return -KEY_shmread;
9357 if (name[2] == 'r' &&
9363 return -KEY_sprintf;
9372 if (name[3] == 'l' &&
9377 return -KEY_symlink;
9386 if (name[4] == 'a' &&
9390 return -KEY_syscall;
9396 if (name[4] == 'p' &&
9400 return -KEY_sysopen;
9406 if (name[4] == 'e' &&
9410 return -KEY_sysread;
9416 if (name[4] == 'e' &&
9420 return -KEY_sysseek;
9438 if (name[1] == 'e' &&
9445 return -KEY_telldir;
9454 if (name[2] == 'f' &&
9460 return -KEY_ucfirst;
9466 if (name[2] == 's' &&
9472 return -KEY_unshift;
9482 if (name[1] == 'a' &&
9489 return -KEY_waitpid;
9498 case 8: /* 26 tokens of length 8 */
9502 if (name[1] == 'U' &&
9510 return KEY_AUTOLOAD;
9521 if (name[3] == 'A' &&
9527 return KEY___DATA__;
9533 if (name[3] == 'I' &&
9539 return -KEY___FILE__;
9545 if (name[3] == 'I' &&
9551 return -KEY___LINE__;
9567 if (name[2] == 'o' &&
9574 return -KEY_closedir;
9580 if (name[2] == 'n' &&
9587 return -KEY_continue;
9597 if (name[1] == 'b' &&
9605 return -KEY_dbmclose;
9611 if (name[1] == 'n' &&
9617 if (name[4] == 'r' &&
9622 return -KEY_endgrent;
9628 if (name[4] == 'w' &&
9633 return -KEY_endpwent;
9646 if (name[1] == 'o' &&
9654 return -KEY_formline;
9660 if (name[1] == 'e' &&
9671 if (name[6] == 'n' &&
9674 return -KEY_getgrent;
9680 if (name[6] == 'i' &&
9683 return -KEY_getgrgid;
9689 if (name[6] == 'a' &&
9692 return -KEY_getgrnam;
9705 if (name[4] == 'o' &&
9710 return -KEY_getlogin;
9721 if (name[6] == 'n' &&
9724 return -KEY_getpwent;
9730 if (name[6] == 'a' &&
9733 return -KEY_getpwnam;
9739 if (name[6] == 'i' &&
9742 return -KEY_getpwuid;
9762 if (name[1] == 'e' &&
9769 if (name[5] == 'i' &&
9776 return -KEY_readline;
9781 return -KEY_readlink;
9792 if (name[5] == 'i' &&
9796 return -KEY_readpipe;
9817 if (name[4] == 'r' &&
9822 return -KEY_setgrent;
9828 if (name[4] == 'w' &&
9833 return -KEY_setpwent;
9849 if (name[3] == 'w' &&
9855 return -KEY_shmwrite;
9861 if (name[3] == 't' &&
9867 return -KEY_shutdown;
9877 if (name[2] == 's' &&
9884 return -KEY_syswrite;
9894 if (name[1] == 'r' &&
9902 return -KEY_truncate;
9911 case 9: /* 9 tokens of length 9 */
9915 if (name[1] == 'N' &&
9924 return KEY_UNITCHECK;
9930 if (name[1] == 'n' &&
9939 return -KEY_endnetent;
9945 if (name[1] == 'e' &&
9954 return -KEY_getnetent;
9960 if (name[1] == 'o' &&
9969 return -KEY_localtime;
9975 if (name[1] == 'r' &&
9984 return KEY_prototype;
9990 if (name[1] == 'u' &&
9999 return -KEY_quotemeta;
10005 if (name[1] == 'e' &&
10014 return -KEY_rewinddir;
10020 if (name[1] == 'e' &&
10029 return -KEY_setnetent;
10035 if (name[1] == 'a' &&
10044 return -KEY_wantarray;
10053 case 10: /* 9 tokens of length 10 */
10057 if (name[1] == 'n' &&
10063 if (name[4] == 'o' &&
10070 return -KEY_endhostent;
10076 if (name[4] == 'e' &&
10083 return -KEY_endservent;
10096 if (name[1] == 'e' &&
10102 if (name[4] == 'o' &&
10109 return -KEY_gethostent;
10118 if (name[5] == 'r' &&
10124 return -KEY_getservent;
10130 if (name[5] == 'c' &&
10136 return -KEY_getsockopt;
10156 if (name[2] == 't')
10161 if (name[4] == 'o' &&
10168 return -KEY_sethostent;
10177 if (name[5] == 'r' &&
10183 return -KEY_setservent;
10189 if (name[5] == 'c' &&
10195 return -KEY_setsockopt;
10212 if (name[2] == 'c' &&
10221 return -KEY_socketpair;
10234 case 11: /* 8 tokens of length 11 */
10238 if (name[1] == '_' &&
10248 { /* __PACKAGE__ */
10249 return -KEY___PACKAGE__;
10255 if (name[1] == 'n' &&
10265 { /* endprotoent */
10266 return -KEY_endprotoent;
10272 if (name[1] == 'e' &&
10281 if (name[5] == 'e' &&
10287 { /* getpeername */
10288 return -KEY_getpeername;
10297 if (name[6] == 'o' &&
10302 { /* getpriority */
10303 return -KEY_getpriority;
10309 if (name[6] == 't' &&
10314 { /* getprotoent */
10315 return -KEY_getprotoent;
10329 if (name[4] == 'o' &&
10336 { /* getsockname */
10337 return -KEY_getsockname;
10350 if (name[1] == 'e' &&
10358 if (name[6] == 'o' &&
10363 { /* setpriority */
10364 return -KEY_setpriority;
10370 if (name[6] == 't' &&
10375 { /* setprotoent */
10376 return -KEY_setprotoent;
10392 case 12: /* 2 tokens of length 12 */
10393 if (name[0] == 'g' &&
10405 if (name[9] == 'd' &&
10408 { /* getnetbyaddr */
10409 return -KEY_getnetbyaddr;
10415 if (name[9] == 'a' &&
10418 { /* getnetbyname */
10419 return -KEY_getnetbyname;
10431 case 13: /* 4 tokens of length 13 */
10432 if (name[0] == 'g' &&
10439 if (name[4] == 'o' &&
10448 if (name[10] == 'd' &&
10451 { /* gethostbyaddr */
10452 return -KEY_gethostbyaddr;
10458 if (name[10] == 'a' &&
10461 { /* gethostbyname */
10462 return -KEY_gethostbyname;
10475 if (name[4] == 'e' &&
10484 if (name[10] == 'a' &&
10487 { /* getservbyname */
10488 return -KEY_getservbyname;
10494 if (name[10] == 'o' &&
10497 { /* getservbyport */
10498 return -KEY_getservbyport;
10517 case 14: /* 1 tokens of length 14 */
10518 if (name[0] == 'g' &&
10532 { /* getprotobyname */
10533 return -KEY_getprotobyname;
10538 case 16: /* 1 tokens of length 16 */
10539 if (name[0] == 'g' &&
10555 { /* getprotobynumber */
10556 return -KEY_getprotobynumber;
10570 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10574 PERL_ARGS_ASSERT_CHECKCOMMA;
10576 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10577 if (ckWARN(WARN_SYNTAX)) {
10580 for (w = s+2; *w && level; w++) {
10583 else if (*w == ')')
10586 while (isSPACE(*w))
10588 /* the list of chars below is for end of statements or
10589 * block / parens, boolean operators (&&, ||, //) and branch
10590 * constructs (or, and, if, until, unless, while, err, for).
10591 * Not a very solid hack... */
10592 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10593 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10594 "%s (...) interpreted as function",name);
10597 while (s < PL_bufend && isSPACE(*s))
10601 while (s < PL_bufend && isSPACE(*s))
10603 if (isIDFIRST_lazy_if(s,UTF)) {
10604 const char * const w = s++;
10605 while (isALNUM_lazy_if(s,UTF))
10607 while (s < PL_bufend && isSPACE(*s))
10611 if (keyword(w, s - w, 0))
10614 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10615 if (gv && GvCVu(gv))
10617 Perl_croak(aTHX_ "No comma allowed after %s", what);
10622 /* Either returns sv, or mortalizes sv and returns a new SV*.
10623 Best used as sv=new_constant(..., sv, ...).
10624 If s, pv are NULL, calls subroutine with one argument,
10625 and type is used with error messages only. */
10628 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10629 SV *sv, SV *pv, const char *type, STRLEN typelen)
10632 HV * const table = GvHV(PL_hintgv); /* ^H */
10636 const char *why1 = "", *why2 = "", *why3 = "";
10638 PERL_ARGS_ASSERT_NEW_CONSTANT;
10640 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10643 why2 = (const char *)
10644 (strEQ(key,"charnames")
10645 ? "(possibly a missing \"use charnames ...\")"
10647 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10648 (type ? type: "undef"), why2);
10650 /* This is convoluted and evil ("goto considered harmful")
10651 * but I do not understand the intricacies of all the different
10652 * failure modes of %^H in here. The goal here is to make
10653 * the most probable error message user-friendly. --jhi */
10658 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10659 (type ? type: "undef"), why1, why2, why3);
10661 yyerror(SvPVX_const(msg));
10665 cvp = hv_fetch(table, key, keylen, FALSE);
10666 if (!cvp || !SvOK(*cvp)) {
10669 why3 = "} is not defined";
10672 sv_2mortal(sv); /* Parent created it permanently */
10675 pv = newSVpvn_flags(s, len, SVs_TEMP);
10677 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10679 typesv = &PL_sv_undef;
10681 PUSHSTACKi(PERLSI_OVERLOAD);
10693 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10697 /* Check the eval first */
10698 if (!PL_in_eval && SvTRUE(ERRSV)) {
10699 sv_catpvs(ERRSV, "Propagated");
10700 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10702 res = SvREFCNT_inc_simple(sv);
10706 SvREFCNT_inc_simple_void(res);
10715 why1 = "Call to &{$^H{";
10717 why3 = "}} did not return a defined value";
10725 /* Returns a NUL terminated string, with the length of the string written to
10729 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10732 register char *d = dest;
10733 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10735 PERL_ARGS_ASSERT_SCAN_WORD;
10739 Perl_croak(aTHX_ ident_too_long);
10740 if (isALNUM(*s)) /* UTF handled below */
10742 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10747 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10751 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10752 char *t = s + UTF8SKIP(s);
10754 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10758 Perl_croak(aTHX_ ident_too_long);
10759 Copy(s, d, len, char);
10772 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10775 char *bracket = NULL;
10777 register char *d = dest;
10778 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10780 PERL_ARGS_ASSERT_SCAN_IDENT;
10785 while (isDIGIT(*s)) {
10787 Perl_croak(aTHX_ ident_too_long);
10794 Perl_croak(aTHX_ ident_too_long);
10795 if (isALNUM(*s)) /* UTF handled below */
10797 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10802 else if (*s == ':' && s[1] == ':') {
10806 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10807 char *t = s + UTF8SKIP(s);
10808 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10810 if (d + (t - s) > e)
10811 Perl_croak(aTHX_ ident_too_long);
10812 Copy(s, d, t - s, char);
10823 if (PL_lex_state != LEX_NORMAL)
10824 PL_lex_state = LEX_INTERPENDMAYBE;
10827 if (*s == '$' && s[1] &&
10828 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10841 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10846 if (isSPACE(s[-1])) {
10848 const char ch = *s++;
10849 if (!SPACE_OR_TAB(ch)) {
10855 if (isIDFIRST_lazy_if(d,UTF)) {
10859 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10860 end += UTF8SKIP(end);
10861 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10862 end += UTF8SKIP(end);
10864 Copy(s, d, end - s, char);
10869 while ((isALNUM(*s) || *s == ':') && d < e)
10872 Perl_croak(aTHX_ ident_too_long);
10875 while (s < send && SPACE_OR_TAB(*s))
10877 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10878 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10879 const char * const brack =
10881 ((*s == '[') ? "[...]" : "{...}");
10882 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10883 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10884 funny, dest, brack, funny, dest, brack);
10887 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10891 /* Handle extended ${^Foo} variables
10892 * 1999-02-27 mjd-perl-patch@plover.com */
10893 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10897 while (isALNUM(*s) && d < e) {
10901 Perl_croak(aTHX_ ident_too_long);
10906 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10907 PL_lex_state = LEX_INTERPEND;
10910 if (PL_lex_state == LEX_NORMAL) {
10911 if (ckWARN(WARN_AMBIGUOUS) &&
10912 (keyword(dest, d - dest, 0)
10913 || get_cvn_flags(dest, d - dest, 0)))
10917 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10918 "Ambiguous use of %c{%s} resolved to %c%s",
10919 funny, dest, funny, dest);
10924 s = bracket; /* let the parser handle it */
10928 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10929 PL_lex_state = LEX_INTERPEND;
10934 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10936 PERL_ARGS_ASSERT_PMFLAG;
10938 PERL_UNUSED_CONTEXT;
10940 const char c = (char)ch;
10942 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10943 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10944 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10945 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10946 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10952 S_scan_pat(pTHX_ char *start, I32 type)
10956 char *s = scan_str(start,!!PL_madskills,FALSE);
10957 const char * const valid_flags =
10958 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10963 PERL_ARGS_ASSERT_SCAN_PAT;
10966 const char * const delimiter = skipspace(start);
10970 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10971 : "Search pattern not terminated" ));
10974 pm = (PMOP*)newPMOP(type, 0);
10975 if (PL_multi_open == '?') {
10976 /* This is the only point in the code that sets PMf_ONCE: */
10977 pm->op_pmflags |= PMf_ONCE;
10979 /* Hence it's safe to do this bit of PMOP book-keeping here, which
10980 allows us to restrict the list needed by reset to just the ??
10982 assert(type != OP_TRANS);
10984 MAGIC *mg = mg_find((SV*)PL_curstash, PERL_MAGIC_symtab);
10987 mg = sv_magicext((SV*)PL_curstash, 0, PERL_MAGIC_symtab, 0, 0,
10990 elements = mg->mg_len / sizeof(PMOP**);
10991 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
10992 ((PMOP**)mg->mg_ptr) [elements++] = pm;
10993 mg->mg_len = elements * sizeof(PMOP**);
10994 PmopSTASH_set(pm,PL_curstash);
11000 while (*s && strchr(valid_flags, *s))
11001 pmflag(&pm->op_pmflags,*s++);
11003 if (PL_madskills && modstart != s) {
11004 SV* tmptoken = newSVpvn(modstart, s - modstart);
11005 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11008 /* issue a warning if /c is specified,but /g is not */
11009 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
11010 && ckWARN(WARN_REGEXP))
11012 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
11013 "Use of /c modifier is meaningless without /g" );
11016 PL_lex_op = (OP*)pm;
11017 pl_yylval.ival = OP_MATCH;
11022 S_scan_subst(pTHX_ char *start)
11033 PERL_ARGS_ASSERT_SCAN_SUBST;
11035 pl_yylval.ival = OP_NULL;
11037 s = scan_str(start,!!PL_madskills,FALSE);
11040 Perl_croak(aTHX_ "Substitution pattern not terminated");
11042 if (s[-1] == PL_multi_open)
11045 if (PL_madskills) {
11046 CURMAD('q', PL_thisopen);
11047 CURMAD('_', PL_thiswhite);
11048 CURMAD('E', PL_thisstuff);
11049 CURMAD('Q', PL_thisclose);
11050 PL_realtokenstart = s - SvPVX(PL_linestr);
11054 first_start = PL_multi_start;
11055 s = scan_str(s,!!PL_madskills,FALSE);
11057 if (PL_lex_stuff) {
11058 SvREFCNT_dec(PL_lex_stuff);
11059 PL_lex_stuff = NULL;
11061 Perl_croak(aTHX_ "Substitution replacement not terminated");
11063 PL_multi_start = first_start; /* so whole substitution is taken together */
11065 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11068 if (PL_madskills) {
11069 CURMAD('z', PL_thisopen);
11070 CURMAD('R', PL_thisstuff);
11071 CURMAD('Z', PL_thisclose);
11077 if (*s == EXEC_PAT_MOD) {
11081 else if (strchr(S_PAT_MODS, *s))
11082 pmflag(&pm->op_pmflags,*s++);
11088 if (PL_madskills) {
11090 curmad('m', newSVpvn(modstart, s - modstart));
11091 append_madprops(PL_thismad, (OP*)pm, 0);
11095 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11096 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11100 SV * const repl = newSVpvs("");
11102 PL_sublex_info.super_bufptr = s;
11103 PL_sublex_info.super_bufend = PL_bufend;
11105 pm->op_pmflags |= PMf_EVAL;
11108 sv_catpvs(repl, "eval ");
11110 sv_catpvs(repl, "do ");
11112 sv_catpvs(repl, "{");
11113 sv_catsv(repl, PL_lex_repl);
11114 if (strchr(SvPVX(PL_lex_repl), '#'))
11115 sv_catpvs(repl, "\n");
11116 sv_catpvs(repl, "}");
11118 SvREFCNT_dec(PL_lex_repl);
11119 PL_lex_repl = repl;
11122 PL_lex_op = (OP*)pm;
11123 pl_yylval.ival = OP_SUBST;
11128 S_scan_trans(pTHX_ char *start)
11141 PERL_ARGS_ASSERT_SCAN_TRANS;
11143 pl_yylval.ival = OP_NULL;
11145 s = scan_str(start,!!PL_madskills,FALSE);
11147 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11149 if (s[-1] == PL_multi_open)
11152 if (PL_madskills) {
11153 CURMAD('q', PL_thisopen);
11154 CURMAD('_', PL_thiswhite);
11155 CURMAD('E', PL_thisstuff);
11156 CURMAD('Q', PL_thisclose);
11157 PL_realtokenstart = s - SvPVX(PL_linestr);
11161 s = scan_str(s,!!PL_madskills,FALSE);
11163 if (PL_lex_stuff) {
11164 SvREFCNT_dec(PL_lex_stuff);
11165 PL_lex_stuff = NULL;
11167 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11169 if (PL_madskills) {
11170 CURMAD('z', PL_thisopen);
11171 CURMAD('R', PL_thisstuff);
11172 CURMAD('Z', PL_thisclose);
11175 complement = del = squash = 0;
11182 complement = OPpTRANS_COMPLEMENT;
11185 del = OPpTRANS_DELETE;
11188 squash = OPpTRANS_SQUASH;
11197 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11198 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11199 o->op_private &= ~OPpTRANS_ALL;
11200 o->op_private |= del|squash|complement|
11201 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11202 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11205 pl_yylval.ival = OP_TRANS;
11208 if (PL_madskills) {
11210 curmad('m', newSVpvn(modstart, s - modstart));
11211 append_madprops(PL_thismad, o, 0);
11220 S_scan_heredoc(pTHX_ register char *s)
11224 I32 op_type = OP_SCALAR;
11228 const char *found_newline;
11232 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11234 I32 stuffstart = s - SvPVX(PL_linestr);
11237 PL_realtokenstart = -1;
11240 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11244 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11248 while (SPACE_OR_TAB(*peek))
11250 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11253 s = delimcpy(d, e, s, PL_bufend, term, &len);
11263 if (!isALNUM_lazy_if(s,UTF))
11264 deprecate_old("bare << to mean <<\"\"");
11265 for (; isALNUM_lazy_if(s,UTF); s++) {
11270 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11271 Perl_croak(aTHX_ "Delimiter for here document is too long");
11274 len = d - PL_tokenbuf;
11277 if (PL_madskills) {
11278 tstart = PL_tokenbuf + !outer;
11279 PL_thisclose = newSVpvn(tstart, len - !outer);
11280 tstart = SvPVX(PL_linestr) + stuffstart;
11281 PL_thisopen = newSVpvn(tstart, s - tstart);
11282 stuffstart = s - SvPVX(PL_linestr);
11285 #ifndef PERL_STRICT_CR
11286 d = strchr(s, '\r');
11288 char * const olds = s;
11290 while (s < PL_bufend) {
11296 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11305 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11312 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11313 herewas = newSVpvn(s,PL_bufend-s);
11317 herewas = newSVpvn(s-1,found_newline-s+1);
11320 herewas = newSVpvn(s,found_newline-s);
11324 if (PL_madskills) {
11325 tstart = SvPVX(PL_linestr) + stuffstart;
11327 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11329 PL_thisstuff = newSVpvn(tstart, s - tstart);
11332 s += SvCUR(herewas);
11335 stuffstart = s - SvPVX(PL_linestr);
11341 tmpstr = newSV_type(SVt_PVIV);
11342 SvGROW(tmpstr, 80);
11343 if (term == '\'') {
11344 op_type = OP_CONST;
11345 SvIV_set(tmpstr, -1);
11347 else if (term == '`') {
11348 op_type = OP_BACKTICK;
11349 SvIV_set(tmpstr, '\\');
11353 PL_multi_start = CopLINE(PL_curcop);
11354 PL_multi_open = PL_multi_close = '<';
11355 term = *PL_tokenbuf;
11356 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11357 char * const bufptr = PL_sublex_info.super_bufptr;
11358 char * const bufend = PL_sublex_info.super_bufend;
11359 char * const olds = s - SvCUR(herewas);
11360 s = strchr(bufptr, '\n');
11364 while (s < bufend &&
11365 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11367 CopLINE_inc(PL_curcop);
11370 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11371 missingterm(PL_tokenbuf);
11373 sv_setpvn(herewas,bufptr,d-bufptr+1);
11374 sv_setpvn(tmpstr,d+1,s-d);
11376 sv_catpvn(herewas,s,bufend-s);
11377 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11384 while (s < PL_bufend &&
11385 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11387 CopLINE_inc(PL_curcop);
11389 if (s >= PL_bufend) {
11390 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11391 missingterm(PL_tokenbuf);
11393 sv_setpvn(tmpstr,d+1,s-d);
11395 if (PL_madskills) {
11397 sv_catpvn(PL_thisstuff, d + 1, s - d);
11399 PL_thisstuff = newSVpvn(d + 1, s - d);
11400 stuffstart = s - SvPVX(PL_linestr);
11404 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11406 sv_catpvn(herewas,s,PL_bufend-s);
11407 sv_setsv(PL_linestr,herewas);
11408 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11409 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11410 PL_last_lop = PL_last_uni = NULL;
11413 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
11414 while (s >= PL_bufend) { /* multiple line string? */
11416 if (PL_madskills) {
11417 tstart = SvPVX(PL_linestr) + stuffstart;
11419 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11421 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11425 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11426 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11427 missingterm(PL_tokenbuf);
11430 stuffstart = s - SvPVX(PL_linestr);
11432 CopLINE_inc(PL_curcop);
11433 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11434 PL_last_lop = PL_last_uni = NULL;
11435 #ifndef PERL_STRICT_CR
11436 if (PL_bufend - PL_linestart >= 2) {
11437 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11438 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11440 PL_bufend[-2] = '\n';
11442 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11444 else if (PL_bufend[-1] == '\r')
11445 PL_bufend[-1] = '\n';
11447 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11448 PL_bufend[-1] = '\n';
11450 if (PERLDB_LINE && PL_curstash != PL_debstash)
11451 update_debugger_info(PL_linestr, NULL, 0);
11452 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11453 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11454 *(SvPVX(PL_linestr) + off ) = ' ';
11455 sv_catsv(PL_linestr,herewas);
11456 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11457 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11461 sv_catsv(tmpstr,PL_linestr);
11466 PL_multi_end = CopLINE(PL_curcop);
11467 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11468 SvPV_shrink_to_cur(tmpstr);
11470 SvREFCNT_dec(herewas);
11472 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11474 else if (PL_encoding)
11475 sv_recode_to_utf8(tmpstr, PL_encoding);
11477 PL_lex_stuff = tmpstr;
11478 pl_yylval.ival = op_type;
11482 /* scan_inputsymbol
11483 takes: current position in input buffer
11484 returns: new position in input buffer
11485 side-effects: pl_yylval and lex_op are set.
11490 <FH> read from filehandle
11491 <pkg::FH> read from package qualified filehandle
11492 <pkg'FH> read from package qualified filehandle
11493 <$fh> read from filehandle in $fh
11494 <*.h> filename glob
11499 S_scan_inputsymbol(pTHX_ char *start)
11502 register char *s = start; /* current position in buffer */
11505 char *d = PL_tokenbuf; /* start of temp holding space */
11506 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11508 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11510 end = strchr(s, '\n');
11513 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11515 /* die if we didn't have space for the contents of the <>,
11516 or if it didn't end, or if we see a newline
11519 if (len >= (I32)sizeof PL_tokenbuf)
11520 Perl_croak(aTHX_ "Excessively long <> operator");
11522 Perl_croak(aTHX_ "Unterminated <> operator");
11527 Remember, only scalar variables are interpreted as filehandles by
11528 this code. Anything more complex (e.g., <$fh{$num}>) will be
11529 treated as a glob() call.
11530 This code makes use of the fact that except for the $ at the front,
11531 a scalar variable and a filehandle look the same.
11533 if (*d == '$' && d[1]) d++;
11535 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11536 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11539 /* If we've tried to read what we allow filehandles to look like, and
11540 there's still text left, then it must be a glob() and not a getline.
11541 Use scan_str to pull out the stuff between the <> and treat it
11542 as nothing more than a string.
11545 if (d - PL_tokenbuf != len) {
11546 pl_yylval.ival = OP_GLOB;
11547 s = scan_str(start,!!PL_madskills,FALSE);
11549 Perl_croak(aTHX_ "Glob not terminated");
11553 bool readline_overriden = FALSE;
11556 /* we're in a filehandle read situation */
11559 /* turn <> into <ARGV> */
11561 Copy("ARGV",d,5,char);
11563 /* Check whether readline() is overriden */
11564 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11566 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11568 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11569 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11570 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11571 readline_overriden = TRUE;
11573 /* if <$fh>, create the ops to turn the variable into a
11577 /* try to find it in the pad for this block, otherwise find
11578 add symbol table ops
11580 const PADOFFSET tmp = pad_findmy(d);
11581 if (tmp != NOT_IN_PAD) {
11582 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11583 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11584 HEK * const stashname = HvNAME_HEK(stash);
11585 SV * const sym = sv_2mortal(newSVhek(stashname));
11586 sv_catpvs(sym, "::");
11587 sv_catpv(sym, d+1);
11592 OP * const o = newOP(OP_PADSV, 0);
11594 PL_lex_op = readline_overriden
11595 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11596 append_elem(OP_LIST, o,
11597 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11598 : (OP*)newUNOP(OP_READLINE, 0, o);
11607 ? (GV_ADDMULTI | GV_ADDINEVAL)
11610 PL_lex_op = readline_overriden
11611 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11612 append_elem(OP_LIST,
11613 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11614 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11615 : (OP*)newUNOP(OP_READLINE, 0,
11616 newUNOP(OP_RV2SV, 0,
11617 newGVOP(OP_GV, 0, gv)));
11619 if (!readline_overriden)
11620 PL_lex_op->op_flags |= OPf_SPECIAL;
11621 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11622 pl_yylval.ival = OP_NULL;
11625 /* If it's none of the above, it must be a literal filehandle
11626 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11628 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11629 PL_lex_op = readline_overriden
11630 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11631 append_elem(OP_LIST,
11632 newGVOP(OP_GV, 0, gv),
11633 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11634 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11635 pl_yylval.ival = OP_NULL;
11644 takes: start position in buffer
11645 keep_quoted preserve \ on the embedded delimiter(s)
11646 keep_delims preserve the delimiters around the string
11647 returns: position to continue reading from buffer
11648 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11649 updates the read buffer.
11651 This subroutine pulls a string out of the input. It is called for:
11652 q single quotes q(literal text)
11653 ' single quotes 'literal text'
11654 qq double quotes qq(interpolate $here please)
11655 " double quotes "interpolate $here please"
11656 qx backticks qx(/bin/ls -l)
11657 ` backticks `/bin/ls -l`
11658 qw quote words @EXPORT_OK = qw( func() $spam )
11659 m// regexp match m/this/
11660 s/// regexp substitute s/this/that/
11661 tr/// string transliterate tr/this/that/
11662 y/// string transliterate y/this/that/
11663 ($*@) sub prototypes sub foo ($)
11664 (stuff) sub attr parameters sub foo : attr(stuff)
11665 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11667 In most of these cases (all but <>, patterns and transliterate)
11668 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11669 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11670 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11673 It skips whitespace before the string starts, and treats the first
11674 character as the delimiter. If the delimiter is one of ([{< then
11675 the corresponding "close" character )]}> is used as the closing
11676 delimiter. It allows quoting of delimiters, and if the string has
11677 balanced delimiters ([{<>}]) it allows nesting.
11679 On success, the SV with the resulting string is put into lex_stuff or,
11680 if that is already non-NULL, into lex_repl. The second case occurs only
11681 when parsing the RHS of the special constructs s/// and tr/// (y///).
11682 For convenience, the terminating delimiter character is stuffed into
11687 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11690 SV *sv; /* scalar value: string */
11691 const char *tmps; /* temp string, used for delimiter matching */
11692 register char *s = start; /* current position in the buffer */
11693 register char term; /* terminating character */
11694 register char *to; /* current position in the sv's data */
11695 I32 brackets = 1; /* bracket nesting level */
11696 bool has_utf8 = FALSE; /* is there any utf8 content? */
11697 I32 termcode; /* terminating char. code */
11698 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11699 STRLEN termlen; /* length of terminating string */
11700 int last_off = 0; /* last position for nesting bracket */
11706 PERL_ARGS_ASSERT_SCAN_STR;
11708 /* skip space before the delimiter */
11714 if (PL_realtokenstart >= 0) {
11715 stuffstart = PL_realtokenstart;
11716 PL_realtokenstart = -1;
11719 stuffstart = start - SvPVX(PL_linestr);
11721 /* mark where we are, in case we need to report errors */
11724 /* after skipping whitespace, the next character is the terminator */
11727 termcode = termstr[0] = term;
11731 termcode = utf8_to_uvchr((U8*)s, &termlen);
11732 Copy(s, termstr, termlen, U8);
11733 if (!UTF8_IS_INVARIANT(term))
11737 /* mark where we are */
11738 PL_multi_start = CopLINE(PL_curcop);
11739 PL_multi_open = term;
11741 /* find corresponding closing delimiter */
11742 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11743 termcode = termstr[0] = term = tmps[5];
11745 PL_multi_close = term;
11747 /* create a new SV to hold the contents. 79 is the SV's initial length.
11748 What a random number. */
11749 sv = newSV_type(SVt_PVIV);
11751 SvIV_set(sv, termcode);
11752 (void)SvPOK_only(sv); /* validate pointer */
11754 /* move past delimiter and try to read a complete string */
11756 sv_catpvn(sv, s, termlen);
11759 tstart = SvPVX(PL_linestr) + stuffstart;
11760 if (!PL_thisopen && !keep_delims) {
11761 PL_thisopen = newSVpvn(tstart, s - tstart);
11762 stuffstart = s - SvPVX(PL_linestr);
11766 if (PL_encoding && !UTF) {
11770 int offset = s - SvPVX_const(PL_linestr);
11771 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11772 &offset, (char*)termstr, termlen);
11773 const char * const ns = SvPVX_const(PL_linestr) + offset;
11774 char * const svlast = SvEND(sv) - 1;
11776 for (; s < ns; s++) {
11777 if (*s == '\n' && !PL_rsfp)
11778 CopLINE_inc(PL_curcop);
11781 goto read_more_line;
11783 /* handle quoted delimiters */
11784 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11786 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11788 if ((svlast-1 - t) % 2) {
11789 if (!keep_quoted) {
11790 *(svlast-1) = term;
11792 SvCUR_set(sv, SvCUR(sv) - 1);
11797 if (PL_multi_open == PL_multi_close) {
11803 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11804 /* At here, all closes are "was quoted" one,
11805 so we don't check PL_multi_close. */
11807 if (!keep_quoted && *(t+1) == PL_multi_open)
11812 else if (*t == PL_multi_open)
11820 SvCUR_set(sv, w - SvPVX_const(sv));
11822 last_off = w - SvPVX(sv);
11823 if (--brackets <= 0)
11828 if (!keep_delims) {
11829 SvCUR_set(sv, SvCUR(sv) - 1);
11835 /* extend sv if need be */
11836 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11837 /* set 'to' to the next character in the sv's string */
11838 to = SvPVX(sv)+SvCUR(sv);
11840 /* if open delimiter is the close delimiter read unbridle */
11841 if (PL_multi_open == PL_multi_close) {
11842 for (; s < PL_bufend; s++,to++) {
11843 /* embedded newlines increment the current line number */
11844 if (*s == '\n' && !PL_rsfp)
11845 CopLINE_inc(PL_curcop);
11846 /* handle quoted delimiters */
11847 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11848 if (!keep_quoted && s[1] == term)
11850 /* any other quotes are simply copied straight through */
11854 /* terminate when run out of buffer (the for() condition), or
11855 have found the terminator */
11856 else if (*s == term) {
11859 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11862 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11868 /* if the terminator isn't the same as the start character (e.g.,
11869 matched brackets), we have to allow more in the quoting, and
11870 be prepared for nested brackets.
11873 /* read until we run out of string, or we find the terminator */
11874 for (; s < PL_bufend; s++,to++) {
11875 /* embedded newlines increment the line count */
11876 if (*s == '\n' && !PL_rsfp)
11877 CopLINE_inc(PL_curcop);
11878 /* backslashes can escape the open or closing characters */
11879 if (*s == '\\' && s+1 < PL_bufend) {
11880 if (!keep_quoted &&
11881 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11886 /* allow nested opens and closes */
11887 else if (*s == PL_multi_close && --brackets <= 0)
11889 else if (*s == PL_multi_open)
11891 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11896 /* terminate the copied string and update the sv's end-of-string */
11898 SvCUR_set(sv, to - SvPVX_const(sv));
11901 * this next chunk reads more into the buffer if we're not done yet
11905 break; /* handle case where we are done yet :-) */
11907 #ifndef PERL_STRICT_CR
11908 if (to - SvPVX_const(sv) >= 2) {
11909 if ((to[-2] == '\r' && to[-1] == '\n') ||
11910 (to[-2] == '\n' && to[-1] == '\r'))
11914 SvCUR_set(sv, to - SvPVX_const(sv));
11916 else if (to[-1] == '\r')
11919 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11924 /* if we're out of file, or a read fails, bail and reset the current
11925 line marker so we can report where the unterminated string began
11928 if (PL_madskills) {
11929 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11931 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11933 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11937 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11939 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11945 /* we read a line, so increment our line counter */
11946 CopLINE_inc(PL_curcop);
11948 /* update debugger info */
11949 if (PERLDB_LINE && PL_curstash != PL_debstash)
11950 update_debugger_info(PL_linestr, NULL, 0);
11952 /* having changed the buffer, we must update PL_bufend */
11953 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11954 PL_last_lop = PL_last_uni = NULL;
11957 /* at this point, we have successfully read the delimited string */
11959 if (!PL_encoding || UTF) {
11961 if (PL_madskills) {
11962 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11963 const int len = s - tstart;
11965 sv_catpvn(PL_thisstuff, tstart, len);
11967 PL_thisstuff = newSVpvn(tstart, len);
11968 if (!PL_thisclose && !keep_delims)
11969 PL_thisclose = newSVpvn(s,termlen);
11974 sv_catpvn(sv, s, termlen);
11979 if (PL_madskills) {
11980 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11981 const int len = s - tstart - termlen;
11983 sv_catpvn(PL_thisstuff, tstart, len);
11985 PL_thisstuff = newSVpvn(tstart, len);
11986 if (!PL_thisclose && !keep_delims)
11987 PL_thisclose = newSVpvn(s - termlen,termlen);
11991 if (has_utf8 || PL_encoding)
11994 PL_multi_end = CopLINE(PL_curcop);
11996 /* if we allocated too much space, give some back */
11997 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11998 SvLEN_set(sv, SvCUR(sv) + 1);
11999 SvPV_renew(sv, SvLEN(sv));
12002 /* decide whether this is the first or second quoted string we've read
12015 takes: pointer to position in buffer
12016 returns: pointer to new position in buffer
12017 side-effects: builds ops for the constant in pl_yylval.op
12019 Read a number in any of the formats that Perl accepts:
12021 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12022 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12025 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12027 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12030 If it reads a number without a decimal point or an exponent, it will
12031 try converting the number to an integer and see if it can do so
12032 without loss of precision.
12036 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12039 register const char *s = start; /* current position in buffer */
12040 register char *d; /* destination in temp buffer */
12041 register char *e; /* end of temp buffer */
12042 NV nv; /* number read, as a double */
12043 SV *sv = NULL; /* place to put the converted number */
12044 bool floatit; /* boolean: int or float? */
12045 const char *lastub = NULL; /* position of last underbar */
12046 static char const number_too_long[] = "Number too long";
12048 PERL_ARGS_ASSERT_SCAN_NUM;
12050 /* We use the first character to decide what type of number this is */
12054 Perl_croak(aTHX_ "panic: scan_num");
12056 /* if it starts with a 0, it could be an octal number, a decimal in
12057 0.13 disguise, or a hexadecimal number, or a binary number. */
12061 u holds the "number so far"
12062 shift the power of 2 of the base
12063 (hex == 4, octal == 3, binary == 1)
12064 overflowed was the number more than we can hold?
12066 Shift is used when we add a digit. It also serves as an "are
12067 we in octal/hex/binary?" indicator to disallow hex characters
12068 when in octal mode.
12073 bool overflowed = FALSE;
12074 bool just_zero = TRUE; /* just plain 0 or binary number? */
12075 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12076 static const char* const bases[5] =
12077 { "", "binary", "", "octal", "hexadecimal" };
12078 static const char* const Bases[5] =
12079 { "", "Binary", "", "Octal", "Hexadecimal" };
12080 static const char* const maxima[5] =
12082 "0b11111111111111111111111111111111",
12086 const char *base, *Base, *max;
12088 /* check for hex */
12093 } else if (s[1] == 'b') {
12098 /* check for a decimal in disguise */
12099 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12101 /* so it must be octal */
12108 if (ckWARN(WARN_SYNTAX))
12109 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12110 "Misplaced _ in number");
12114 base = bases[shift];
12115 Base = Bases[shift];
12116 max = maxima[shift];
12118 /* read the rest of the number */
12120 /* x is used in the overflow test,
12121 b is the digit we're adding on. */
12126 /* if we don't mention it, we're done */
12130 /* _ are ignored -- but warned about if consecutive */
12132 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12133 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12134 "Misplaced _ in number");
12138 /* 8 and 9 are not octal */
12139 case '8': case '9':
12141 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12145 case '2': case '3': case '4':
12146 case '5': case '6': case '7':
12148 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12151 case '0': case '1':
12152 b = *s++ & 15; /* ASCII digit -> value of digit */
12156 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12157 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12158 /* make sure they said 0x */
12161 b = (*s++ & 7) + 9;
12163 /* Prepare to put the digit we have onto the end
12164 of the number so far. We check for overflows.
12170 x = u << shift; /* make room for the digit */
12172 if ((x >> shift) != u
12173 && !(PL_hints & HINT_NEW_BINARY)) {
12176 if (ckWARN_d(WARN_OVERFLOW))
12177 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12178 "Integer overflow in %s number",
12181 u = x | b; /* add the digit to the end */
12184 n *= nvshift[shift];
12185 /* If an NV has not enough bits in its
12186 * mantissa to represent an UV this summing of
12187 * small low-order numbers is a waste of time
12188 * (because the NV cannot preserve the
12189 * low-order bits anyway): we could just
12190 * remember when did we overflow and in the
12191 * end just multiply n by the right
12199 /* if we get here, we had success: make a scalar value from
12204 /* final misplaced underbar check */
12205 if (s[-1] == '_') {
12206 if (ckWARN(WARN_SYNTAX))
12207 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12212 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12213 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12214 "%s number > %s non-portable",
12220 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12221 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12222 "%s number > %s non-portable",
12227 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12228 sv = new_constant(start, s - start, "integer",
12229 sv, NULL, NULL, 0);
12230 else if (PL_hints & HINT_NEW_BINARY)
12231 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12236 handle decimal numbers.
12237 we're also sent here when we read a 0 as the first digit
12239 case '1': case '2': case '3': case '4': case '5':
12240 case '6': case '7': case '8': case '9': case '.':
12243 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12246 /* read next group of digits and _ and copy into d */
12247 while (isDIGIT(*s) || *s == '_') {
12248 /* skip underscores, checking for misplaced ones
12252 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12253 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12254 "Misplaced _ in number");
12258 /* check for end of fixed-length buffer */
12260 Perl_croak(aTHX_ number_too_long);
12261 /* if we're ok, copy the character */
12266 /* final misplaced underbar check */
12267 if (lastub && s == lastub + 1) {
12268 if (ckWARN(WARN_SYNTAX))
12269 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12272 /* read a decimal portion if there is one. avoid
12273 3..5 being interpreted as the number 3. followed
12276 if (*s == '.' && s[1] != '.') {
12281 if (ckWARN(WARN_SYNTAX))
12282 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12283 "Misplaced _ in number");
12287 /* copy, ignoring underbars, until we run out of digits.
12289 for (; isDIGIT(*s) || *s == '_'; s++) {
12290 /* fixed length buffer check */
12292 Perl_croak(aTHX_ number_too_long);
12294 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12295 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12296 "Misplaced _ in number");
12302 /* fractional part ending in underbar? */
12303 if (s[-1] == '_') {
12304 if (ckWARN(WARN_SYNTAX))
12305 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12306 "Misplaced _ in number");
12308 if (*s == '.' && isDIGIT(s[1])) {
12309 /* oops, it's really a v-string, but without the "v" */
12315 /* read exponent part, if present */
12316 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12320 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12321 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12323 /* stray preinitial _ */
12325 if (ckWARN(WARN_SYNTAX))
12326 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12327 "Misplaced _ in number");
12331 /* allow positive or negative exponent */
12332 if (*s == '+' || *s == '-')
12335 /* stray initial _ */
12337 if (ckWARN(WARN_SYNTAX))
12338 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12339 "Misplaced _ in number");
12343 /* read digits of exponent */
12344 while (isDIGIT(*s) || *s == '_') {
12347 Perl_croak(aTHX_ number_too_long);
12351 if (((lastub && s == lastub + 1) ||
12352 (!isDIGIT(s[1]) && s[1] != '_'))
12353 && ckWARN(WARN_SYNTAX))
12354 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12355 "Misplaced _ in number");
12362 /* make an sv from the string */
12366 We try to do an integer conversion first if no characters
12367 indicating "float" have been found.
12372 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12374 if (flags == IS_NUMBER_IN_UV) {
12376 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12379 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12380 if (uv <= (UV) IV_MIN)
12381 sv_setiv(sv, -(IV)uv);
12388 /* terminate the string */
12390 nv = Atof(PL_tokenbuf);
12395 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12396 const char *const key = floatit ? "float" : "integer";
12397 const STRLEN keylen = floatit ? 5 : 7;
12398 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12399 key, keylen, sv, NULL, NULL, 0);
12403 /* if it starts with a v, it could be a v-string */
12406 sv = newSV(5); /* preallocate storage space */
12407 s = scan_vstring(s, PL_bufend, sv);
12411 /* make the op for the constant and return */
12414 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12416 lvalp->opval = NULL;
12422 S_scan_formline(pTHX_ register char *s)
12425 register char *eol;
12427 SV * const stuff = newSVpvs("");
12428 bool needargs = FALSE;
12429 bool eofmt = FALSE;
12431 char *tokenstart = s;
12432 SV* savewhite = NULL;
12434 if (PL_madskills) {
12435 savewhite = PL_thiswhite;
12440 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12442 while (!needargs) {
12445 #ifdef PERL_STRICT_CR
12446 while (SPACE_OR_TAB(*t))
12449 while (SPACE_OR_TAB(*t) || *t == '\r')
12452 if (*t == '\n' || t == PL_bufend) {
12457 if (PL_in_eval && !PL_rsfp) {
12458 eol = (char *) memchr(s,'\n',PL_bufend-s);
12463 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12465 for (t = s; t < eol; t++) {
12466 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12468 goto enough; /* ~~ must be first line in formline */
12470 if (*t == '@' || *t == '^')
12474 sv_catpvn(stuff, s, eol-s);
12475 #ifndef PERL_STRICT_CR
12476 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12477 char *end = SvPVX(stuff) + SvCUR(stuff);
12480 SvCUR_set(stuff, SvCUR(stuff) - 1);
12490 if (PL_madskills) {
12492 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12494 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12497 s = filter_gets(PL_linestr, PL_rsfp, 0);
12499 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12501 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12503 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12504 PL_last_lop = PL_last_uni = NULL;
12513 if (SvCUR(stuff)) {
12516 PL_lex_state = LEX_NORMAL;
12517 start_force(PL_curforce);
12518 NEXTVAL_NEXTTOKE.ival = 0;
12522 PL_lex_state = LEX_FORMLINE;
12524 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12526 else if (PL_encoding)
12527 sv_recode_to_utf8(stuff, PL_encoding);
12529 start_force(PL_curforce);
12530 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12532 start_force(PL_curforce);
12533 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12537 SvREFCNT_dec(stuff);
12539 PL_lex_formbrack = 0;
12543 if (PL_madskills) {
12545 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12547 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12548 PL_thiswhite = savewhite;
12555 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12558 const I32 oldsavestack_ix = PL_savestack_ix;
12559 CV* const outsidecv = PL_compcv;
12562 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12564 SAVEI32(PL_subline);
12565 save_item(PL_subname);
12566 SAVESPTR(PL_compcv);
12568 PL_compcv = (CV*)newSV_type(is_format ? SVt_PVFM : SVt_PVCV);
12569 CvFLAGS(PL_compcv) |= flags;
12571 PL_subline = CopLINE(PL_curcop);
12572 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12573 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc_simple(outsidecv);
12574 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12576 return oldsavestack_ix;
12580 #pragma segment Perl_yylex
12583 Perl_yywarn(pTHX_ const char *const s)
12587 PERL_ARGS_ASSERT_YYWARN;
12589 PL_in_eval |= EVAL_WARNONLY;
12591 PL_in_eval &= ~EVAL_WARNONLY;
12596 Perl_yyerror(pTHX_ const char *const s)
12599 const char *where = NULL;
12600 const char *context = NULL;
12603 int yychar = PL_parser->yychar;
12605 PERL_ARGS_ASSERT_YYERROR;
12607 if (!yychar || (yychar == ';' && !PL_rsfp))
12609 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12610 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12611 PL_oldbufptr != PL_bufptr) {
12614 The code below is removed for NetWare because it abends/crashes on NetWare
12615 when the script has error such as not having the closing quotes like:
12616 if ($var eq "value)
12617 Checking of white spaces is anyway done in NetWare code.
12620 while (isSPACE(*PL_oldoldbufptr))
12623 context = PL_oldoldbufptr;
12624 contlen = PL_bufptr - PL_oldoldbufptr;
12626 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12627 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12630 The code below is removed for NetWare because it abends/crashes on NetWare
12631 when the script has error such as not having the closing quotes like:
12632 if ($var eq "value)
12633 Checking of white spaces is anyway done in NetWare code.
12636 while (isSPACE(*PL_oldbufptr))
12639 context = PL_oldbufptr;
12640 contlen = PL_bufptr - PL_oldbufptr;
12642 else if (yychar > 255)
12643 where = "next token ???";
12644 else if (yychar == -2) { /* YYEMPTY */
12645 if (PL_lex_state == LEX_NORMAL ||
12646 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12647 where = "at end of line";
12648 else if (PL_lex_inpat)
12649 where = "within pattern";
12651 where = "within string";
12654 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12656 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12657 else if (isPRINT_LC(yychar)) {
12658 const char string = yychar;
12659 sv_catpvn(where_sv, &string, 1);
12662 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12663 where = SvPVX_const(where_sv);
12665 msg = sv_2mortal(newSVpv(s, 0));
12666 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12667 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12669 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12671 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12672 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12673 Perl_sv_catpvf(aTHX_ msg,
12674 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12675 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12678 if (PL_in_eval & EVAL_WARNONLY) {
12679 if (ckWARN_d(WARN_SYNTAX))
12680 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12684 if (PL_error_count >= 10) {
12685 if (PL_in_eval && SvCUR(ERRSV))
12686 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12687 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12689 Perl_croak(aTHX_ "%s has too many errors.\n",
12690 OutCopFILE(PL_curcop));
12693 PL_in_my_stash = NULL;
12697 #pragma segment Main
12701 S_swallow_bom(pTHX_ U8 *s)
12704 const STRLEN slen = SvCUR(PL_linestr);
12706 PERL_ARGS_ASSERT_SWALLOW_BOM;
12710 if (s[1] == 0xFE) {
12711 /* UTF-16 little-endian? (or UTF32-LE?) */
12712 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12713 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12714 #ifndef PERL_NO_UTF16_FILTER
12715 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12718 if (PL_bufend > (char*)s) {
12722 filter_add(utf16rev_textfilter, NULL);
12723 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12724 utf16_to_utf8_reversed(s, news,
12725 PL_bufend - (char*)s - 1,
12727 sv_setpvn(PL_linestr, (const char*)news, newlen);
12729 s = (U8*)SvPVX(PL_linestr);
12730 Copy(news, s, newlen, U8);
12734 SvUTF8_on(PL_linestr);
12735 s = (U8*)SvPVX(PL_linestr);
12737 /* FIXME - is this a general bug fix? */
12740 PL_bufend = SvPVX(PL_linestr) + newlen;
12743 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12748 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12749 #ifndef PERL_NO_UTF16_FILTER
12750 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12753 if (PL_bufend > (char *)s) {
12757 filter_add(utf16_textfilter, NULL);
12758 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12759 utf16_to_utf8(s, news,
12760 PL_bufend - (char*)s,
12762 sv_setpvn(PL_linestr, (const char*)news, newlen);
12764 SvUTF8_on(PL_linestr);
12765 s = (U8*)SvPVX(PL_linestr);
12766 PL_bufend = SvPVX(PL_linestr) + newlen;
12769 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12774 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12775 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12776 s += 3; /* UTF-8 */
12782 if (s[2] == 0xFE && s[3] == 0xFF) {
12783 /* UTF-32 big-endian */
12784 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12787 else if (s[2] == 0 && s[3] != 0) {
12790 * are a good indicator of UTF-16BE. */
12791 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12797 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12798 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12799 s += 4; /* UTF-8 */
12805 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12808 * are a good indicator of UTF-16LE. */
12809 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12817 #ifndef PERL_NO_UTF16_FILTER
12819 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12822 const STRLEN old = SvCUR(sv);
12823 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12824 DEBUG_P(PerlIO_printf(Perl_debug_log,
12825 "utf16_textfilter(%p): %d %d (%d)\n",
12826 FPTR2DPTR(void *, utf16_textfilter),
12827 idx, maxlen, (int) count));
12831 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12832 Copy(SvPVX_const(sv), tmps, old, char);
12833 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12834 SvCUR(sv) - old, &newlen);
12835 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12837 DEBUG_P({sv_dump(sv);});
12842 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12845 const STRLEN old = SvCUR(sv);
12846 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12847 DEBUG_P(PerlIO_printf(Perl_debug_log,
12848 "utf16rev_textfilter(%p): %d %d (%d)\n",
12849 FPTR2DPTR(void *, utf16rev_textfilter),
12850 idx, maxlen, (int) count));
12854 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12855 Copy(SvPVX_const(sv), tmps, old, char);
12856 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12857 SvCUR(sv) - old, &newlen);
12858 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12860 DEBUG_P({ sv_dump(sv); });
12866 Returns a pointer to the next character after the parsed
12867 vstring, as well as updating the passed in sv.
12869 Function must be called like
12872 s = scan_vstring(s,e,sv);
12874 where s and e are the start and end of the string.
12875 The sv should already be large enough to store the vstring
12876 passed in, for performance reasons.
12881 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12884 const char *pos = s;
12885 const char *start = s;
12887 PERL_ARGS_ASSERT_SCAN_VSTRING;
12889 if (*pos == 'v') pos++; /* get past 'v' */
12890 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12892 if ( *pos != '.') {
12893 /* this may not be a v-string if followed by => */
12894 const char *next = pos;
12895 while (next < e && isSPACE(*next))
12897 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12898 /* return string not v-string */
12899 sv_setpvn(sv,(char *)s,pos-s);
12900 return (char *)pos;
12904 if (!isALPHA(*pos)) {
12905 U8 tmpbuf[UTF8_MAXBYTES+1];
12908 s++; /* get past 'v' */
12910 sv_setpvn(sv, "", 0);
12913 /* this is atoi() that tolerates underscores */
12916 const char *end = pos;
12918 while (--end >= s) {
12920 const UV orev = rev;
12921 rev += (*end - '0') * mult;
12923 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12924 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12925 "Integer overflow in decimal number");
12929 if (rev > 0x7FFFFFFF)
12930 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12932 /* Append native character for the rev point */
12933 tmpend = uvchr_to_utf8(tmpbuf, rev);
12934 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12935 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12937 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12943 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12947 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12955 * c-indentation-style: bsd
12956 * c-basic-offset: 4
12957 * indent-tabs-mode: t
12960 * ex: set ts=8 sts=4 sw=4 noet: