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->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 break; /* in regexp, $ might be tail anchor */
2180 /* End of else if chain - OP_TRANS rejoin rest */
2183 if (*s == '\\' && s+1 < send) {
2186 /* deprecate \1 in strings and substitution replacements */
2187 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2188 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2190 if (ckWARN(WARN_SYNTAX))
2191 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2196 /* string-change backslash escapes */
2197 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2201 /* skip any other backslash escapes in a pattern */
2202 else if (PL_lex_inpat) {
2203 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2204 goto default_action;
2207 /* if we get here, it's either a quoted -, or a digit */
2210 /* quoted - in transliterations */
2212 if (PL_lex_inwhat == OP_TRANS) {
2219 if ((isALPHA(*s) || isDIGIT(*s)) &&
2221 Perl_warner(aTHX_ packWARN(WARN_MISC),
2222 "Unrecognized escape \\%c passed through",
2224 /* default action is to copy the quoted character */
2225 goto default_action;
2228 /* \132 indicates an octal constant */
2229 case '0': case '1': case '2': case '3':
2230 case '4': case '5': case '6': case '7':
2234 uv = grok_oct(s, &len, &flags, NULL);
2237 goto NUM_ESCAPE_INSERT;
2239 /* \x24 indicates a hex constant */
2243 char* const e = strchr(s, '}');
2244 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2245 PERL_SCAN_DISALLOW_PREFIX;
2250 yyerror("Missing right brace on \\x{}");
2254 uv = grok_hex(s, &len, &flags, NULL);
2260 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2261 uv = grok_hex(s, &len, &flags, NULL);
2267 /* Insert oct or hex escaped character.
2268 * There will always enough room in sv since such
2269 * escapes will be longer than any UTF-8 sequence
2270 * they can end up as. */
2272 /* We need to map to chars to ASCII before doing the tests
2275 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2276 if (!has_utf8 && uv > 255) {
2277 /* Might need to recode whatever we have
2278 * accumulated so far if it contains any
2281 * (Can't we keep track of that and avoid
2282 * this rescan? --jhi)
2286 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2287 if (!NATIVE_IS_INVARIANT(*c)) {
2292 const STRLEN offset = d - SvPVX_const(sv);
2294 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2298 while (src >= (const U8 *)SvPVX_const(sv)) {
2299 if (!NATIVE_IS_INVARIANT(*src)) {
2300 const U8 ch = NATIVE_TO_ASCII(*src);
2301 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2302 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2312 if (has_utf8 || uv > 255) {
2313 d = (char*)uvchr_to_utf8((U8*)d, uv);
2315 if (PL_lex_inwhat == OP_TRANS &&
2316 PL_sublex_info.sub_op) {
2317 PL_sublex_info.sub_op->op_private |=
2318 (PL_lex_repl ? OPpTRANS_FROM_UTF
2322 if (uv > 255 && !dorange)
2323 native_range = FALSE;
2335 /* \N{LATIN SMALL LETTER A} is a named character */
2339 char* e = strchr(s, '}');
2345 yyerror("Missing right brace on \\N{}");
2349 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2351 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2352 PERL_SCAN_DISALLOW_PREFIX;
2355 uv = grok_hex(s, &len, &flags, NULL);
2356 if ( e > s && len != (STRLEN)(e - s) ) {
2360 goto NUM_ESCAPE_INSERT;
2362 res = newSVpvn(s + 1, e - s - 1);
2363 res = new_constant( NULL, 0, "charnames",
2364 res, NULL, s - 2, e - s + 3 );
2366 sv_utf8_upgrade(res);
2367 str = SvPV_const(res,len);
2368 #ifdef EBCDIC_NEVER_MIND
2369 /* charnames uses pack U and that has been
2370 * recently changed to do the below uni->native
2371 * mapping, so this would be redundant (and wrong,
2372 * the code point would be doubly converted).
2373 * But leave this in just in case the pack U change
2374 * gets revoked, but the semantics is still
2375 * desireable for charnames. --jhi */
2377 UV uv = utf8_to_uvchr((const U8*)str, 0);
2380 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2382 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2383 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2384 str = SvPV_const(res, len);
2388 if (!has_utf8 && SvUTF8(res)) {
2389 const char * const ostart = SvPVX_const(sv);
2390 SvCUR_set(sv, d - ostart);
2393 sv_utf8_upgrade(sv);
2394 /* this just broke our allocation above... */
2395 SvGROW(sv, (STRLEN)(send - start));
2396 d = SvPVX(sv) + SvCUR(sv);
2399 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2400 const char * const odest = SvPVX_const(sv);
2402 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2403 d = SvPVX(sv) + (d - odest);
2407 native_range = FALSE; /* \N{} is guessed to be Unicode */
2409 Copy(str, d, len, char);
2416 yyerror("Missing braces on \\N{}");
2419 /* \c is a control character */
2428 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2431 yyerror("Missing control char name in \\c");
2435 /* printf-style backslashes, formfeeds, newlines, etc */
2437 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2440 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2443 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2446 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2449 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2452 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2455 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2461 } /* end if (backslash) */
2468 /* If we started with encoded form, or already know we want it
2469 and then encode the next character */
2470 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2472 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2473 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2476 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2477 const STRLEN off = d - SvPVX_const(sv);
2478 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2480 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2483 if (uv > 255 && !dorange)
2484 native_range = FALSE;
2488 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2490 } /* while loop to process each character */
2492 /* terminate the string and set up the sv */
2494 SvCUR_set(sv, d - SvPVX_const(sv));
2495 if (SvCUR(sv) >= SvLEN(sv))
2496 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2499 if (PL_encoding && !has_utf8) {
2500 sv_recode_to_utf8(sv, PL_encoding);
2506 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2507 PL_sublex_info.sub_op->op_private |=
2508 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2512 /* shrink the sv if we allocated more than we used */
2513 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2514 SvPV_shrink_to_cur(sv);
2517 /* return the substring (via pl_yylval) only if we parsed anything */
2518 if (s > PL_bufptr) {
2519 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2520 const char *const key = PL_lex_inpat ? "qr" : "q";
2521 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2525 if (PL_lex_inwhat == OP_TRANS) {
2528 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2536 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2539 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2546 * Returns TRUE if there's more to the expression (e.g., a subscript),
2549 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2551 * ->[ and ->{ return TRUE
2552 * { and [ outside a pattern are always subscripts, so return TRUE
2553 * if we're outside a pattern and it's not { or [, then return FALSE
2554 * if we're in a pattern and the first char is a {
2555 * {4,5} (any digits around the comma) returns FALSE
2556 * if we're in a pattern and the first char is a [
2558 * [SOMETHING] has a funky algorithm to decide whether it's a
2559 * character class or not. It has to deal with things like
2560 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2561 * anything else returns TRUE
2564 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2567 S_intuit_more(pTHX_ register char *s)
2571 PERL_ARGS_ASSERT_INTUIT_MORE;
2573 if (PL_lex_brackets)
2575 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2577 if (*s != '{' && *s != '[')
2582 /* In a pattern, so maybe we have {n,m}. */
2599 /* On the other hand, maybe we have a character class */
2602 if (*s == ']' || *s == '^')
2605 /* this is terrifying, and it works */
2606 int weight = 2; /* let's weigh the evidence */
2608 unsigned char un_char = 255, last_un_char;
2609 const char * const send = strchr(s,']');
2610 char tmpbuf[sizeof PL_tokenbuf * 4];
2612 if (!send) /* has to be an expression */
2615 Zero(seen,256,char);
2618 else if (isDIGIT(*s)) {
2620 if (isDIGIT(s[1]) && s[2] == ']')
2626 for (; s < send; s++) {
2627 last_un_char = un_char;
2628 un_char = (unsigned char)*s;
2633 weight -= seen[un_char] * 10;
2634 if (isALNUM_lazy_if(s+1,UTF)) {
2636 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2637 len = (int)strlen(tmpbuf);
2638 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2643 else if (*s == '$' && s[1] &&
2644 strchr("[#!%*<>()-=",s[1])) {
2645 if (/*{*/ strchr("])} =",s[2]))
2654 if (strchr("wds]",s[1]))
2656 else if (seen[(U8)'\''] || seen[(U8)'"'])
2658 else if (strchr("rnftbxcav",s[1]))
2660 else if (isDIGIT(s[1])) {
2662 while (s[1] && isDIGIT(s[1]))
2672 if (strchr("aA01! ",last_un_char))
2674 if (strchr("zZ79~",s[1]))
2676 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2677 weight -= 5; /* cope with negative subscript */
2680 if (!isALNUM(last_un_char)
2681 && !(last_un_char == '$' || last_un_char == '@'
2682 || last_un_char == '&')
2683 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2688 if (keyword(tmpbuf, d - tmpbuf, 0))
2691 if (un_char == last_un_char + 1)
2693 weight -= seen[un_char];
2698 if (weight >= 0) /* probably a character class */
2708 * Does all the checking to disambiguate
2710 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2711 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2713 * First argument is the stuff after the first token, e.g. "bar".
2715 * Not a method if bar is a filehandle.
2716 * Not a method if foo is a subroutine prototyped to take a filehandle.
2717 * Not a method if it's really "Foo $bar"
2718 * Method if it's "foo $bar"
2719 * Not a method if it's really "print foo $bar"
2720 * Method if it's really "foo package::" (interpreted as package->foo)
2721 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2722 * Not a method if bar is a filehandle or package, but is quoted with
2727 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2730 char *s = start + (*start == '$');
2731 char tmpbuf[sizeof PL_tokenbuf];
2738 PERL_ARGS_ASSERT_INTUIT_METHOD;
2741 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2745 const char *proto = SvPVX_const(cv);
2756 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2757 /* start is the beginning of the possible filehandle/object,
2758 * and s is the end of it
2759 * tmpbuf is a copy of it
2762 if (*start == '$') {
2763 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2764 isUPPER(*PL_tokenbuf))
2767 len = start - SvPVX(PL_linestr);
2771 start = SvPVX(PL_linestr) + len;
2775 return *s == '(' ? FUNCMETH : METHOD;
2777 if (!keyword(tmpbuf, len, 0)) {
2778 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2782 soff = s - SvPVX(PL_linestr);
2786 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2787 if (indirgv && GvCVu(indirgv))
2789 /* filehandle or package name makes it a method */
2790 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2792 soff = s - SvPVX(PL_linestr);
2795 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2796 return 0; /* no assumptions -- "=>" quotes bearword */
2798 start_force(PL_curforce);
2799 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2800 newSVpvn(tmpbuf,len));
2801 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2803 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2808 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2810 return *s == '(' ? FUNCMETH : METHOD;
2816 /* Encoded script support. filter_add() effectively inserts a
2817 * 'pre-processing' function into the current source input stream.
2818 * Note that the filter function only applies to the current source file
2819 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2821 * The datasv parameter (which may be NULL) can be used to pass
2822 * private data to this instance of the filter. The filter function
2823 * can recover the SV using the FILTER_DATA macro and use it to
2824 * store private buffers and state information.
2826 * The supplied datasv parameter is upgraded to a PVIO type
2827 * and the IoDIRP/IoANY field is used to store the function pointer,
2828 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2829 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2830 * private use must be set using malloc'd pointers.
2834 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2843 if (!PL_rsfp_filters)
2844 PL_rsfp_filters = newAV();
2847 SvUPGRADE(datasv, SVt_PVIO);
2848 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2849 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2850 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2851 FPTR2DPTR(void *, IoANY(datasv)),
2852 SvPV_nolen(datasv)));
2853 av_unshift(PL_rsfp_filters, 1);
2854 av_store(PL_rsfp_filters, 0, datasv) ;
2859 /* Delete most recently added instance of this filter function. */
2861 Perl_filter_del(pTHX_ filter_t funcp)
2866 PERL_ARGS_ASSERT_FILTER_DEL;
2869 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2870 FPTR2DPTR(void*, funcp)));
2872 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2874 /* if filter is on top of stack (usual case) just pop it off */
2875 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2876 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2877 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2878 IoANY(datasv) = (void *)NULL;
2879 sv_free(av_pop(PL_rsfp_filters));
2883 /* we need to search for the correct entry and clear it */
2884 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2888 /* Invoke the idxth filter function for the current rsfp. */
2889 /* maxlen 0 = read one text line */
2891 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2896 /* This API is bad. It should have been using unsigned int for maxlen.
2897 Not sure if we want to change the API, but if not we should sanity
2898 check the value here. */
2899 const unsigned int correct_length
2908 PERL_ARGS_ASSERT_FILTER_READ;
2910 if (!PL_parser || !PL_rsfp_filters)
2912 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2913 /* Provide a default input filter to make life easy. */
2914 /* Note that we append to the line. This is handy. */
2915 DEBUG_P(PerlIO_printf(Perl_debug_log,
2916 "filter_read %d: from rsfp\n", idx));
2917 if (correct_length) {
2920 const int old_len = SvCUR(buf_sv);
2922 /* ensure buf_sv is large enough */
2923 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2924 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2925 correct_length)) <= 0) {
2926 if (PerlIO_error(PL_rsfp))
2927 return -1; /* error */
2929 return 0 ; /* end of file */
2931 SvCUR_set(buf_sv, old_len + len) ;
2934 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2935 if (PerlIO_error(PL_rsfp))
2936 return -1; /* error */
2938 return 0 ; /* end of file */
2941 return SvCUR(buf_sv);
2943 /* Skip this filter slot if filter has been deleted */
2944 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2945 DEBUG_P(PerlIO_printf(Perl_debug_log,
2946 "filter_read %d: skipped (filter deleted)\n",
2948 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2950 /* Get function pointer hidden within datasv */
2951 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2952 DEBUG_P(PerlIO_printf(Perl_debug_log,
2953 "filter_read %d: via function %p (%s)\n",
2954 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2955 /* Call function. The function is expected to */
2956 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2957 /* Return: <0:error, =0:eof, >0:not eof */
2958 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2962 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2966 PERL_ARGS_ASSERT_FILTER_GETS;
2968 #ifdef PERL_CR_FILTER
2969 if (!PL_rsfp_filters) {
2970 filter_add(S_cr_textfilter,NULL);
2973 if (PL_rsfp_filters) {
2975 SvCUR_set(sv, 0); /* start with empty line */
2976 if (FILTER_READ(0, sv, 0) > 0)
2977 return ( SvPVX(sv) ) ;
2982 return (sv_gets(sv, fp, append));
2986 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
2991 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
2993 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2997 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2998 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3000 return GvHV(gv); /* Foo:: */
3003 /* use constant CLASS => 'MyClass' */
3004 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3005 if (gv && GvCV(gv)) {
3006 SV * const sv = cv_const_sv(GvCV(gv));
3008 pkgname = SvPV_const(sv, len);
3011 return gv_stashpvn(pkgname, len, 0);
3015 * S_readpipe_override
3016 * Check whether readpipe() is overriden, and generates the appropriate
3017 * optree, provided sublex_start() is called afterwards.
3020 S_readpipe_override(pTHX)
3023 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3024 pl_yylval.ival = OP_BACKTICK;
3026 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3028 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3029 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3030 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3032 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3033 append_elem(OP_LIST,
3034 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3035 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3042 * The intent of this yylex wrapper is to minimize the changes to the
3043 * tokener when we aren't interested in collecting madprops. It remains
3044 * to be seen how successful this strategy will be...
3051 char *s = PL_bufptr;
3053 /* make sure PL_thiswhite is initialized */
3057 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3058 if (PL_pending_ident)
3059 return S_pending_ident(aTHX);
3061 /* previous token ate up our whitespace? */
3062 if (!PL_lasttoke && PL_nextwhite) {
3063 PL_thiswhite = PL_nextwhite;
3067 /* isolate the token, and figure out where it is without whitespace */
3068 PL_realtokenstart = -1;
3072 assert(PL_curforce < 0);
3074 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3075 if (!PL_thistoken) {
3076 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3077 PL_thistoken = newSVpvs("");
3079 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3080 PL_thistoken = newSVpvn(tstart, s - tstart);
3083 if (PL_thismad) /* install head */
3084 CURMAD('X', PL_thistoken);
3087 /* last whitespace of a sublex? */
3088 if (optype == ')' && PL_endwhite) {
3089 CURMAD('X', PL_endwhite);
3094 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3095 if (!PL_thiswhite && !PL_endwhite && !optype) {
3096 sv_free(PL_thistoken);
3101 /* put off final whitespace till peg */
3102 if (optype == ';' && !PL_rsfp) {
3103 PL_nextwhite = PL_thiswhite;
3106 else if (PL_thisopen) {
3107 CURMAD('q', PL_thisopen);
3109 sv_free(PL_thistoken);
3113 /* Store actual token text as madprop X */
3114 CURMAD('X', PL_thistoken);
3118 /* add preceding whitespace as madprop _ */
3119 CURMAD('_', PL_thiswhite);
3123 /* add quoted material as madprop = */
3124 CURMAD('=', PL_thisstuff);
3128 /* add terminating quote as madprop Q */
3129 CURMAD('Q', PL_thisclose);
3133 /* special processing based on optype */
3137 /* opval doesn't need a TOKEN since it can already store mp */
3147 if (pl_yylval.opval)
3148 append_madprops(PL_thismad, pl_yylval.opval, 0);
3156 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3165 /* remember any fake bracket that lexer is about to discard */
3166 if (PL_lex_brackets == 1 &&
3167 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3170 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3173 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3174 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3177 break; /* don't bother looking for trailing comment */
3186 /* attach a trailing comment to its statement instead of next token */
3190 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3192 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3194 if (*s == '\n' || *s == '#') {
3195 while (s < PL_bufend && *s != '\n')
3199 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3200 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3217 /* Create new token struct. Note: opvals return early above. */
3218 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3225 S_tokenize_use(pTHX_ int is_use, char *s) {
3228 PERL_ARGS_ASSERT_TOKENIZE_USE;
3230 if (PL_expect != XSTATE)
3231 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3232 is_use ? "use" : "no"));
3234 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3235 s = force_version(s, TRUE);
3236 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3237 start_force(PL_curforce);
3238 NEXTVAL_NEXTTOKE.opval = NULL;
3241 else if (*s == 'v') {
3242 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3243 s = force_version(s, FALSE);
3247 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3248 s = force_version(s, FALSE);
3250 pl_yylval.ival = is_use;
3254 static const char* const exp_name[] =
3255 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3256 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3263 Works out what to call the token just pulled out of the input
3264 stream. The yacc parser takes care of taking the ops we return and
3265 stitching them into a tree.
3271 if read an identifier
3272 if we're in a my declaration
3273 croak if they tried to say my($foo::bar)
3274 build the ops for a my() declaration
3275 if it's an access to a my() variable
3276 are we in a sort block?
3277 croak if my($a); $a <=> $b
3278 build ops for access to a my() variable
3279 if in a dq string, and they've said @foo and we can't find @foo
3281 build ops for a bareword
3282 if we already built the token before, use it.
3287 #pragma segment Perl_yylex
3293 register char *s = PL_bufptr;
3298 /* orig_keyword, gvp, and gv are initialized here because
3299 * jump to the label just_a_word_zero can bypass their
3300 * initialization later. */
3301 I32 orig_keyword = 0;
3306 SV* tmp = newSVpvs("");
3307 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3308 (IV)CopLINE(PL_curcop),
3309 lex_state_names[PL_lex_state],
3310 exp_name[PL_expect],
3311 pv_display(tmp, s, strlen(s), 0, 60));
3314 /* check if there's an identifier for us to look at */
3315 if (PL_pending_ident)
3316 return REPORT(S_pending_ident(aTHX));
3318 /* no identifier pending identification */
3320 switch (PL_lex_state) {
3322 case LEX_NORMAL: /* Some compilers will produce faster */
3323 case LEX_INTERPNORMAL: /* code if we comment these out. */
3327 /* when we've already built the next token, just pull it out of the queue */
3331 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3333 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3334 PL_nexttoke[PL_lasttoke].next_mad = 0;
3335 if (PL_thismad && PL_thismad->mad_key == '_') {
3336 PL_thiswhite = (SV*)PL_thismad->mad_val;
3337 PL_thismad->mad_val = 0;
3338 mad_free(PL_thismad);
3343 PL_lex_state = PL_lex_defer;
3344 PL_expect = PL_lex_expect;
3345 PL_lex_defer = LEX_NORMAL;
3346 if (!PL_nexttoke[PL_lasttoke].next_type)
3351 pl_yylval = PL_nextval[PL_nexttoke];
3353 PL_lex_state = PL_lex_defer;
3354 PL_expect = PL_lex_expect;
3355 PL_lex_defer = LEX_NORMAL;
3359 /* FIXME - can these be merged? */
3360 return(PL_nexttoke[PL_lasttoke].next_type);
3362 return REPORT(PL_nexttype[PL_nexttoke]);
3365 /* interpolated case modifiers like \L \U, including \Q and \E.
3366 when we get here, PL_bufptr is at the \
3368 case LEX_INTERPCASEMOD:
3370 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3371 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3373 /* handle \E or end of string */
3374 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3376 if (PL_lex_casemods) {
3377 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3378 PL_lex_casestack[PL_lex_casemods] = '\0';
3380 if (PL_bufptr != PL_bufend
3381 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3383 PL_lex_state = LEX_INTERPCONCAT;
3386 PL_thistoken = newSVpvs("\\E");
3392 while (PL_bufptr != PL_bufend &&
3393 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3395 PL_thiswhite = newSVpvs("");
3396 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3400 if (PL_bufptr != PL_bufend)
3403 PL_lex_state = LEX_INTERPCONCAT;
3407 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3408 "### Saw case modifier\n"); });
3410 if (s[1] == '\\' && s[2] == 'E') {
3413 PL_thiswhite = newSVpvs("");
3414 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3417 PL_lex_state = LEX_INTERPCONCAT;
3422 if (!PL_madskills) /* when just compiling don't need correct */
3423 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3424 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3425 if ((*s == 'L' || *s == 'U') &&
3426 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3427 PL_lex_casestack[--PL_lex_casemods] = '\0';
3430 if (PL_lex_casemods > 10)
3431 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3432 PL_lex_casestack[PL_lex_casemods++] = *s;
3433 PL_lex_casestack[PL_lex_casemods] = '\0';
3434 PL_lex_state = LEX_INTERPCONCAT;
3435 start_force(PL_curforce);
3436 NEXTVAL_NEXTTOKE.ival = 0;
3438 start_force(PL_curforce);
3440 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3442 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3444 NEXTVAL_NEXTTOKE.ival = OP_LC;
3446 NEXTVAL_NEXTTOKE.ival = OP_UC;
3448 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3450 Perl_croak(aTHX_ "panic: yylex");
3452 SV* const tmpsv = newSVpvs("\\ ");
3453 /* replace the space with the character we want to escape
3455 SvPVX(tmpsv)[1] = *s;
3461 if (PL_lex_starts) {
3467 sv_free(PL_thistoken);
3468 PL_thistoken = newSVpvs("");
3471 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3472 if (PL_lex_casemods == 1 && PL_lex_inpat)
3481 case LEX_INTERPPUSH:
3482 return REPORT(sublex_push());
3484 case LEX_INTERPSTART:
3485 if (PL_bufptr == PL_bufend)
3486 return REPORT(sublex_done());
3487 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3488 "### Interpolated variable\n"); });
3490 PL_lex_dojoin = (*PL_bufptr == '@');
3491 PL_lex_state = LEX_INTERPNORMAL;
3492 if (PL_lex_dojoin) {
3493 start_force(PL_curforce);
3494 NEXTVAL_NEXTTOKE.ival = 0;
3496 start_force(PL_curforce);
3497 force_ident("\"", '$');
3498 start_force(PL_curforce);
3499 NEXTVAL_NEXTTOKE.ival = 0;
3501 start_force(PL_curforce);
3502 NEXTVAL_NEXTTOKE.ival = 0;
3504 start_force(PL_curforce);
3505 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3508 if (PL_lex_starts++) {
3513 sv_free(PL_thistoken);
3514 PL_thistoken = newSVpvs("");
3517 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3518 if (!PL_lex_casemods && PL_lex_inpat)
3525 case LEX_INTERPENDMAYBE:
3526 if (intuit_more(PL_bufptr)) {
3527 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3533 if (PL_lex_dojoin) {
3534 PL_lex_dojoin = FALSE;
3535 PL_lex_state = LEX_INTERPCONCAT;
3539 sv_free(PL_thistoken);
3540 PL_thistoken = newSVpvs("");
3545 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3546 && SvEVALED(PL_lex_repl))
3548 if (PL_bufptr != PL_bufend)
3549 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3553 case LEX_INTERPCONCAT:
3555 if (PL_lex_brackets)
3556 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3558 if (PL_bufptr == PL_bufend)
3559 return REPORT(sublex_done());
3561 if (SvIVX(PL_linestr) == '\'') {
3562 SV *sv = newSVsv(PL_linestr);
3565 else if ( PL_hints & HINT_NEW_RE )
3566 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3567 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3571 s = scan_const(PL_bufptr);
3573 PL_lex_state = LEX_INTERPCASEMOD;
3575 PL_lex_state = LEX_INTERPSTART;
3578 if (s != PL_bufptr) {
3579 start_force(PL_curforce);
3581 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3583 NEXTVAL_NEXTTOKE = pl_yylval;
3586 if (PL_lex_starts++) {
3590 sv_free(PL_thistoken);
3591 PL_thistoken = newSVpvs("");
3594 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3595 if (!PL_lex_casemods && PL_lex_inpat)
3608 PL_lex_state = LEX_NORMAL;
3609 s = scan_formline(PL_bufptr);
3610 if (!PL_lex_formbrack)
3616 PL_oldoldbufptr = PL_oldbufptr;
3622 sv_free(PL_thistoken);
3625 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3629 if (isIDFIRST_lazy_if(s,UTF))
3631 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3632 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3635 goto fake_eof; /* emulate EOF on ^D or ^Z */
3644 if (PL_lex_brackets) {
3645 yyerror((const char *)
3647 ? "Format not terminated"
3648 : "Missing right curly or square bracket"));
3650 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3651 "### Tokener got EOF\n");
3655 if (s++ < PL_bufend)
3656 goto retry; /* ignore stray nulls */
3659 if (!PL_in_eval && !PL_preambled) {
3660 PL_preambled = TRUE;
3666 /* Generate a string of Perl code to load the debugger.
3667 * If PERL5DB is set, it will return the contents of that,
3668 * otherwise a compile-time require of perl5db.pl. */
3670 const char * const pdb = PerlEnv_getenv("PERL5DB");
3673 sv_setpv(PL_linestr, pdb);
3674 sv_catpvs(PL_linestr,";");
3676 SETERRNO(0,SS_NORMAL);
3677 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3680 sv_setpvs(PL_linestr,"");
3681 if (PL_preambleav) {
3682 SV **svp = AvARRAY(PL_preambleav);
3683 SV **const end = svp + AvFILLp(PL_preambleav);
3685 sv_catsv(PL_linestr, *svp);
3687 sv_catpvs(PL_linestr, ";");
3689 sv_free((SV*)PL_preambleav);
3690 PL_preambleav = NULL;
3693 sv_catpvs(PL_linestr,
3694 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3695 if (PL_minus_n || PL_minus_p) {
3696 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3698 sv_catpvs(PL_linestr,"chomp;");
3701 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3702 || *PL_splitstr == '"')
3703 && strchr(PL_splitstr + 1, *PL_splitstr))
3704 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3706 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3707 bytes can be used as quoting characters. :-) */
3708 const char *splits = PL_splitstr;
3709 sv_catpvs(PL_linestr, "our @F=split(q\0");
3712 if (*splits == '\\')
3713 sv_catpvn(PL_linestr, splits, 1);
3714 sv_catpvn(PL_linestr, splits, 1);
3715 } while (*splits++);
3716 /* This loop will embed the trailing NUL of
3717 PL_linestr as the last thing it does before
3719 sv_catpvs(PL_linestr, ");");
3723 sv_catpvs(PL_linestr,"our @F=split(' ');");
3726 sv_catpvs(PL_linestr, "\n");
3727 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3728 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3729 PL_last_lop = PL_last_uni = NULL;
3730 if (PERLDB_LINE && PL_curstash != PL_debstash)
3731 update_debugger_info(PL_linestr, NULL, 0);
3735 bof = PL_rsfp ? TRUE : FALSE;
3736 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3739 PL_realtokenstart = -1;
3742 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3743 PerlIO_clearerr(PL_rsfp);
3745 (void)PerlIO_close(PL_rsfp);
3747 PL_doextract = FALSE;
3749 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3755 sv_setpvs(PL_linestr, ";}continue{print;}");
3757 sv_setpvs(PL_linestr, ";}");
3758 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3759 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3760 PL_last_lop = PL_last_uni = NULL;
3761 PL_minus_n = PL_minus_p = 0;
3764 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3765 PL_last_lop = PL_last_uni = NULL;
3766 sv_setpvn(PL_linestr,"",0);
3767 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3769 /* If it looks like the start of a BOM or raw UTF-16,
3770 * check if it in fact is. */
3776 #ifdef PERLIO_IS_STDIO
3777 # ifdef __GNU_LIBRARY__
3778 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3779 # define FTELL_FOR_PIPE_IS_BROKEN
3783 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3784 # define FTELL_FOR_PIPE_IS_BROKEN
3789 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3791 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3792 s = swallow_bom((U8*)s);
3796 /* Incest with pod. */
3799 sv_catsv(PL_thiswhite, PL_linestr);
3801 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3802 sv_setpvn(PL_linestr, "", 0);
3803 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3804 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3805 PL_last_lop = PL_last_uni = NULL;
3806 PL_doextract = FALSE;
3810 } while (PL_doextract);
3811 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3812 if (PERLDB_LINE && PL_curstash != PL_debstash)
3813 update_debugger_info(PL_linestr, NULL, 0);
3814 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3815 PL_last_lop = PL_last_uni = NULL;
3816 if (CopLINE(PL_curcop) == 1) {
3817 while (s < PL_bufend && isSPACE(*s))
3819 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3823 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3827 if (*s == '#' && *(s+1) == '!')
3829 #ifdef ALTERNATE_SHEBANG
3831 static char const as[] = ALTERNATE_SHEBANG;
3832 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3833 d = s + (sizeof(as) - 1);
3835 #endif /* ALTERNATE_SHEBANG */
3844 while (*d && !isSPACE(*d))
3848 #ifdef ARG_ZERO_IS_SCRIPT
3849 if (ipathend > ipath) {
3851 * HP-UX (at least) sets argv[0] to the script name,
3852 * which makes $^X incorrect. And Digital UNIX and Linux,
3853 * at least, set argv[0] to the basename of the Perl
3854 * interpreter. So, having found "#!", we'll set it right.
3856 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3858 assert(SvPOK(x) || SvGMAGICAL(x));
3859 if (sv_eq(x, CopFILESV(PL_curcop))) {
3860 sv_setpvn(x, ipath, ipathend - ipath);
3866 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3867 const char * const lstart = SvPV_const(x,llen);
3869 bstart += blen - llen;
3870 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3871 sv_setpvn(x, ipath, ipathend - ipath);
3876 TAINT_NOT; /* $^X is always tainted, but that's OK */
3878 #endif /* ARG_ZERO_IS_SCRIPT */
3883 d = instr(s,"perl -");
3885 d = instr(s,"perl");
3887 /* avoid getting into infinite loops when shebang
3888 * line contains "Perl" rather than "perl" */
3890 for (d = ipathend-4; d >= ipath; --d) {
3891 if ((*d == 'p' || *d == 'P')
3892 && !ibcmp(d, "perl", 4))
3902 #ifdef ALTERNATE_SHEBANG
3904 * If the ALTERNATE_SHEBANG on this system starts with a
3905 * character that can be part of a Perl expression, then if
3906 * we see it but not "perl", we're probably looking at the
3907 * start of Perl code, not a request to hand off to some
3908 * other interpreter. Similarly, if "perl" is there, but
3909 * not in the first 'word' of the line, we assume the line
3910 * contains the start of the Perl program.
3912 if (d && *s != '#') {
3913 const char *c = ipath;
3914 while (*c && !strchr("; \t\r\n\f\v#", *c))
3917 d = NULL; /* "perl" not in first word; ignore */
3919 *s = '#'; /* Don't try to parse shebang line */
3921 #endif /* ALTERNATE_SHEBANG */
3922 #ifndef MACOS_TRADITIONAL
3927 !instr(s,"indir") &&
3928 instr(PL_origargv[0],"perl"))
3935 while (s < PL_bufend && isSPACE(*s))
3937 if (s < PL_bufend) {
3938 Newxz(newargv,PL_origargc+3,char*);
3940 while (s < PL_bufend && !isSPACE(*s))
3943 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3946 newargv = PL_origargv;
3949 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3951 Perl_croak(aTHX_ "Can't exec %s", ipath);
3955 while (*d && !isSPACE(*d))
3957 while (SPACE_OR_TAB(*d))
3961 const bool switches_done = PL_doswitches;
3962 const U32 oldpdb = PL_perldb;
3963 const bool oldn = PL_minus_n;
3964 const bool oldp = PL_minus_p;
3968 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3969 const char * const m = d1;
3970 while (*d1 && !isSPACE(*d1))
3972 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3975 d1 = moreswitches(d1);
3977 if (PL_doswitches && !switches_done) {
3978 int argc = PL_origargc;
3979 char **argv = PL_origargv;
3982 } while (argc && argv[0][0] == '-' && argv[0][1]);
3983 init_argv_symbols(argc,argv);
3985 if ((PERLDB_LINE && !oldpdb) ||
3986 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3987 /* if we have already added "LINE: while (<>) {",
3988 we must not do it again */
3990 sv_setpvn(PL_linestr, "", 0);
3991 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3992 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3993 PL_last_lop = PL_last_uni = NULL;
3994 PL_preambled = FALSE;
3996 (void)gv_fetchfile(PL_origfilename);
4003 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4005 PL_lex_state = LEX_FORMLINE;
4010 #ifdef PERL_STRICT_CR
4011 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4013 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4015 case ' ': case '\t': case '\f': case 013:
4016 #ifdef MACOS_TRADITIONAL
4020 PL_realtokenstart = -1;
4022 PL_thiswhite = newSVpvs("");
4023 sv_catpvn(PL_thiswhite, s, 1);
4030 PL_realtokenstart = -1;
4034 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4035 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4036 /* handle eval qq[#line 1 "foo"\n ...] */
4037 CopLINE_dec(PL_curcop);
4040 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4042 if (!PL_in_eval || PL_rsfp)
4047 while (d < PL_bufend && *d != '\n')
4051 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4052 Perl_croak(aTHX_ "panic: input overflow");
4055 PL_thiswhite = newSVpvn(s, d - s);
4060 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4062 PL_lex_state = LEX_FORMLINE;
4068 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4069 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4072 TOKEN(PEG); /* make sure any #! line is accessible */
4077 /* if (PL_madskills && PL_lex_formbrack) { */
4079 while (d < PL_bufend && *d != '\n')
4083 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4084 Perl_croak(aTHX_ "panic: input overflow");
4085 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4087 PL_thiswhite = newSVpvs("");
4088 if (CopLINE(PL_curcop) == 1) {
4089 sv_setpvn(PL_thiswhite, "", 0);
4092 sv_catpvn(PL_thiswhite, s, d - s);
4106 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4114 while (s < PL_bufend && SPACE_OR_TAB(*s))
4117 if (strnEQ(s,"=>",2)) {
4118 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4119 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4120 OPERATOR('-'); /* unary minus */
4122 PL_last_uni = PL_oldbufptr;
4124 case 'r': ftst = OP_FTEREAD; break;
4125 case 'w': ftst = OP_FTEWRITE; break;
4126 case 'x': ftst = OP_FTEEXEC; break;
4127 case 'o': ftst = OP_FTEOWNED; break;
4128 case 'R': ftst = OP_FTRREAD; break;
4129 case 'W': ftst = OP_FTRWRITE; break;
4130 case 'X': ftst = OP_FTREXEC; break;
4131 case 'O': ftst = OP_FTROWNED; break;
4132 case 'e': ftst = OP_FTIS; break;
4133 case 'z': ftst = OP_FTZERO; break;
4134 case 's': ftst = OP_FTSIZE; break;
4135 case 'f': ftst = OP_FTFILE; break;
4136 case 'd': ftst = OP_FTDIR; break;
4137 case 'l': ftst = OP_FTLINK; break;
4138 case 'p': ftst = OP_FTPIPE; break;
4139 case 'S': ftst = OP_FTSOCK; break;
4140 case 'u': ftst = OP_FTSUID; break;
4141 case 'g': ftst = OP_FTSGID; break;
4142 case 'k': ftst = OP_FTSVTX; break;
4143 case 'b': ftst = OP_FTBLK; break;
4144 case 'c': ftst = OP_FTCHR; break;
4145 case 't': ftst = OP_FTTTY; break;
4146 case 'T': ftst = OP_FTTEXT; break;
4147 case 'B': ftst = OP_FTBINARY; break;
4148 case 'M': case 'A': case 'C':
4149 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4151 case 'M': ftst = OP_FTMTIME; break;
4152 case 'A': ftst = OP_FTATIME; break;
4153 case 'C': ftst = OP_FTCTIME; break;
4161 PL_last_lop_op = (OPCODE)ftst;
4162 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4163 "### Saw file test %c\n", (int)tmp);
4168 /* Assume it was a minus followed by a one-letter named
4169 * subroutine call (or a -bareword), then. */
4170 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4171 "### '-%c' looked like a file test but was not\n",
4178 const char tmp = *s++;
4181 if (PL_expect == XOPERATOR)
4186 else if (*s == '>') {
4189 if (isIDFIRST_lazy_if(s,UTF)) {
4190 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4198 if (PL_expect == XOPERATOR)
4201 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4203 OPERATOR('-'); /* unary minus */
4209 const char tmp = *s++;
4212 if (PL_expect == XOPERATOR)
4217 if (PL_expect == XOPERATOR)
4220 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4227 if (PL_expect != XOPERATOR) {
4228 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4229 PL_expect = XOPERATOR;
4230 force_ident(PL_tokenbuf, '*');
4243 if (PL_expect == XOPERATOR) {
4247 PL_tokenbuf[0] = '%';
4248 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4249 sizeof PL_tokenbuf - 1, FALSE);
4250 if (!PL_tokenbuf[1]) {
4253 PL_pending_ident = '%';
4264 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4271 const char tmp = *s++;
4277 goto just_a_word_zero_gv;
4280 switch (PL_expect) {
4286 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4288 PL_bufptr = s; /* update in case we back off */
4294 PL_expect = XTERMBLOCK;
4297 stuffstart = s - SvPVX(PL_linestr) - 1;
4301 while (isIDFIRST_lazy_if(s,UTF)) {
4304 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4305 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4306 if (tmp < 0) tmp = -tmp;
4320 sv = newSVpvn(s, len);
4322 d = scan_str(d,TRUE,TRUE);
4324 /* MUST advance bufptr here to avoid bogus
4325 "at end of line" context messages from yyerror().
4327 PL_bufptr = s + len;
4328 yyerror("Unterminated attribute parameter in attribute list");
4332 return REPORT(0); /* EOF indicator */
4336 sv_catsv(sv, PL_lex_stuff);
4337 attrs = append_elem(OP_LIST, attrs,
4338 newSVOP(OP_CONST, 0, sv));
4339 SvREFCNT_dec(PL_lex_stuff);
4340 PL_lex_stuff = NULL;
4343 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4345 if (PL_in_my == KEY_our) {
4347 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4349 /* skip to avoid loading attributes.pm */
4351 deprecate(":unique");
4354 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4357 /* NOTE: any CV attrs applied here need to be part of
4358 the CVf_BUILTIN_ATTRS define in cv.h! */
4359 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4361 CvLVALUE_on(PL_compcv);
4363 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4365 CvLOCKED_on(PL_compcv);
4367 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4369 CvMETHOD_on(PL_compcv);
4371 /* After we've set the flags, it could be argued that
4372 we don't need to do the attributes.pm-based setting
4373 process, and shouldn't bother appending recognized
4374 flags. To experiment with that, uncomment the
4375 following "else". (Note that's already been
4376 uncommented. That keeps the above-applied built-in
4377 attributes from being intercepted (and possibly
4378 rejected) by a package's attribute routines, but is
4379 justified by the performance win for the common case
4380 of applying only built-in attributes.) */
4382 attrs = append_elem(OP_LIST, attrs,
4383 newSVOP(OP_CONST, 0,
4387 if (*s == ':' && s[1] != ':')
4390 break; /* require real whitespace or :'s */
4391 /* XXX losing whitespace on sequential attributes here */
4395 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4396 if (*s != ';' && *s != '}' && *s != tmp
4397 && (tmp != '=' || *s != ')')) {
4398 const char q = ((*s == '\'') ? '"' : '\'');
4399 /* If here for an expression, and parsed no attrs, back
4401 if (tmp == '=' && !attrs) {
4405 /* MUST advance bufptr here to avoid bogus "at end of line"
4406 context messages from yyerror().
4409 yyerror( (const char *)
4411 ? Perl_form(aTHX_ "Invalid separator character "
4412 "%c%c%c in attribute list", q, *s, q)
4413 : "Unterminated attribute list" ) );
4421 start_force(PL_curforce);
4422 NEXTVAL_NEXTTOKE.opval = attrs;
4423 CURMAD('_', PL_nextwhite);
4428 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4429 (s - SvPVX(PL_linestr)) - stuffstart);
4437 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4438 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4446 const char tmp = *s++;
4451 const char tmp = *s++;
4459 if (PL_lex_brackets <= 0)
4460 yyerror("Unmatched right square bracket");
4463 if (PL_lex_state == LEX_INTERPNORMAL) {
4464 if (PL_lex_brackets == 0) {
4465 if (*s == '-' && s[1] == '>')
4466 PL_lex_state = LEX_INTERPENDMAYBE;
4467 else if (*s != '[' && *s != '{')
4468 PL_lex_state = LEX_INTERPEND;
4475 if (PL_lex_brackets > 100) {
4476 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4478 switch (PL_expect) {
4480 if (PL_lex_formbrack) {
4484 if (PL_oldoldbufptr == PL_last_lop)
4485 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4487 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4488 OPERATOR(HASHBRACK);
4490 while (s < PL_bufend && SPACE_OR_TAB(*s))
4493 PL_tokenbuf[0] = '\0';
4494 if (d < PL_bufend && *d == '-') {
4495 PL_tokenbuf[0] = '-';
4497 while (d < PL_bufend && SPACE_OR_TAB(*d))
4500 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4501 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4503 while (d < PL_bufend && SPACE_OR_TAB(*d))
4506 const char minus = (PL_tokenbuf[0] == '-');
4507 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4515 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4520 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4525 if (PL_oldoldbufptr == PL_last_lop)
4526 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4528 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4531 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4533 /* This hack is to get the ${} in the message. */
4535 yyerror("syntax error");
4538 OPERATOR(HASHBRACK);
4540 /* This hack serves to disambiguate a pair of curlies
4541 * as being a block or an anon hash. Normally, expectation
4542 * determines that, but in cases where we're not in a
4543 * position to expect anything in particular (like inside
4544 * eval"") we have to resolve the ambiguity. This code
4545 * covers the case where the first term in the curlies is a
4546 * quoted string. Most other cases need to be explicitly
4547 * disambiguated by prepending a "+" before the opening
4548 * curly in order to force resolution as an anon hash.
4550 * XXX should probably propagate the outer expectation
4551 * into eval"" to rely less on this hack, but that could
4552 * potentially break current behavior of eval"".
4556 if (*s == '\'' || *s == '"' || *s == '`') {
4557 /* common case: get past first string, handling escapes */
4558 for (t++; t < PL_bufend && *t != *s;)
4559 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4563 else if (*s == 'q') {
4566 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4569 /* skip q//-like construct */
4571 char open, close, term;
4574 while (t < PL_bufend && isSPACE(*t))
4576 /* check for q => */
4577 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4578 OPERATOR(HASHBRACK);
4582 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4586 for (t++; t < PL_bufend; t++) {
4587 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4589 else if (*t == open)
4593 for (t++; t < PL_bufend; t++) {
4594 if (*t == '\\' && t+1 < PL_bufend)
4596 else if (*t == close && --brackets <= 0)
4598 else if (*t == open)
4605 /* skip plain q word */
4606 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4609 else if (isALNUM_lazy_if(t,UTF)) {
4611 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4614 while (t < PL_bufend && isSPACE(*t))
4616 /* if comma follows first term, call it an anon hash */
4617 /* XXX it could be a comma expression with loop modifiers */
4618 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4619 || (*t == '=' && t[1] == '>')))
4620 OPERATOR(HASHBRACK);
4621 if (PL_expect == XREF)
4624 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4630 pl_yylval.ival = CopLINE(PL_curcop);
4631 if (isSPACE(*s) || *s == '#')
4632 PL_copline = NOLINE; /* invalidate current command line number */
4637 if (PL_lex_brackets <= 0)
4638 yyerror("Unmatched right curly bracket");
4640 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4641 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4642 PL_lex_formbrack = 0;
4643 if (PL_lex_state == LEX_INTERPNORMAL) {
4644 if (PL_lex_brackets == 0) {
4645 if (PL_expect & XFAKEBRACK) {
4646 PL_expect &= XENUMMASK;
4647 PL_lex_state = LEX_INTERPEND;
4652 PL_thiswhite = newSVpvs("");
4653 sv_catpvn(PL_thiswhite,"}",1);
4656 return yylex(); /* ignore fake brackets */
4658 if (*s == '-' && s[1] == '>')
4659 PL_lex_state = LEX_INTERPENDMAYBE;
4660 else if (*s != '[' && *s != '{')
4661 PL_lex_state = LEX_INTERPEND;
4664 if (PL_expect & XFAKEBRACK) {
4665 PL_expect &= XENUMMASK;
4667 return yylex(); /* ignore fake brackets */
4669 start_force(PL_curforce);
4671 curmad('X', newSVpvn(s-1,1));
4672 CURMAD('_', PL_thiswhite);
4677 PL_thistoken = newSVpvs("");
4685 if (PL_expect == XOPERATOR) {
4686 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4687 && isIDFIRST_lazy_if(s,UTF))
4689 CopLINE_dec(PL_curcop);
4690 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4691 CopLINE_inc(PL_curcop);
4696 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4698 PL_expect = XOPERATOR;
4699 force_ident(PL_tokenbuf, '&');
4703 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4715 const char tmp = *s++;
4722 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4723 && strchr("+-*/%.^&|<",tmp))
4724 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4725 "Reversed %c= operator",(int)tmp);
4727 if (PL_expect == XSTATE && isALPHA(tmp) &&
4728 (s == PL_linestart+1 || s[-2] == '\n') )
4730 if (PL_in_eval && !PL_rsfp) {
4735 if (strnEQ(s,"=cut",4)) {
4751 PL_thiswhite = newSVpvs("");
4752 sv_catpvn(PL_thiswhite, PL_linestart,
4753 PL_bufend - PL_linestart);
4757 PL_doextract = TRUE;
4761 if (PL_lex_brackets < PL_lex_formbrack) {
4763 #ifdef PERL_STRICT_CR
4764 while (SPACE_OR_TAB(*t))
4766 while (SPACE_OR_TAB(*t) || *t == '\r')
4769 if (*t == '\n' || *t == '#') {
4778 if (PL_expect == XSTATE && s[1] == '!' && s[2] == '!') {
4784 const char tmp = *s++;
4786 /* was this !=~ where !~ was meant?
4787 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4789 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4790 const char *t = s+1;
4792 while (t < PL_bufend && isSPACE(*t))
4795 if (*t == '/' || *t == '?' ||
4796 ((*t == 'm' || *t == 's' || *t == 'y')
4797 && !isALNUM(t[1])) ||
4798 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4799 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4800 "!=~ should be !~");
4810 if (PL_expect != XOPERATOR) {
4811 if (s[1] != '<' && !strchr(s,'>'))
4814 s = scan_heredoc(s);
4816 s = scan_inputsymbol(s);
4817 TERM(sublex_start());
4823 SHop(OP_LEFT_SHIFT);
4837 const char tmp = *s++;
4839 SHop(OP_RIGHT_SHIFT);
4840 else if (tmp == '=')
4849 if (PL_expect == XOPERATOR) {
4850 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4852 deprecate_old(commaless_variable_list);
4853 return REPORT(','); /* grandfather non-comma-format format */
4857 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4858 PL_tokenbuf[0] = '@';
4859 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4860 sizeof PL_tokenbuf - 1, FALSE);
4861 if (PL_expect == XOPERATOR)
4862 no_op("Array length", s);
4863 if (!PL_tokenbuf[1])
4865 PL_expect = XOPERATOR;
4866 PL_pending_ident = '#';
4870 PL_tokenbuf[0] = '$';
4871 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4872 sizeof PL_tokenbuf - 1, FALSE);
4873 if (PL_expect == XOPERATOR)
4875 if (!PL_tokenbuf[1]) {
4877 yyerror("Final $ should be \\$ or $name");
4881 /* This kludge not intended to be bulletproof. */
4882 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4883 pl_yylval.opval = newSVOP(OP_CONST, 0,
4884 newSViv(CopARYBASE_get(&PL_compiling)));
4885 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4891 const char tmp = *s;
4892 if (PL_lex_state == LEX_NORMAL)
4895 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4896 && intuit_more(s)) {
4898 PL_tokenbuf[0] = '@';
4899 if (ckWARN(WARN_SYNTAX)) {
4902 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4905 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4906 while (t < PL_bufend && *t != ']')
4908 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4909 "Multidimensional syntax %.*s not supported",
4910 (int)((t - PL_bufptr) + 1), PL_bufptr);
4914 else if (*s == '{') {
4916 PL_tokenbuf[0] = '%';
4917 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4918 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4920 char tmpbuf[sizeof PL_tokenbuf];
4923 } while (isSPACE(*t));
4924 if (isIDFIRST_lazy_if(t,UTF)) {
4926 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4930 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4931 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4932 "You need to quote \"%s\"",
4939 PL_expect = XOPERATOR;
4940 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4941 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4942 if (!islop || PL_last_lop_op == OP_GREPSTART)
4943 PL_expect = XOPERATOR;
4944 else if (strchr("$@\"'`q", *s))
4945 PL_expect = XTERM; /* e.g. print $fh "foo" */
4946 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4947 PL_expect = XTERM; /* e.g. print $fh &sub */
4948 else if (isIDFIRST_lazy_if(s,UTF)) {
4949 char tmpbuf[sizeof PL_tokenbuf];
4951 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4952 if ((t2 = keyword(tmpbuf, len, 0))) {
4953 /* binary operators exclude handle interpretations */
4965 PL_expect = XTERM; /* e.g. print $fh length() */
4970 PL_expect = XTERM; /* e.g. print $fh subr() */
4973 else if (isDIGIT(*s))
4974 PL_expect = XTERM; /* e.g. print $fh 3 */
4975 else if (*s == '.' && isDIGIT(s[1]))
4976 PL_expect = XTERM; /* e.g. print $fh .3 */
4977 else if ((*s == '?' || *s == '-' || *s == '+')
4978 && !isSPACE(s[1]) && s[1] != '=')
4979 PL_expect = XTERM; /* e.g. print $fh -1 */
4980 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4982 PL_expect = XTERM; /* e.g. print $fh /.../
4983 XXX except DORDOR operator
4985 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4987 PL_expect = XTERM; /* print $fh <<"EOF" */
4990 PL_pending_ident = '$';
4994 if (PL_expect == XOPERATOR)
4996 PL_tokenbuf[0] = '@';
4997 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
4998 if (!PL_tokenbuf[1]) {
5001 if (PL_lex_state == LEX_NORMAL)
5003 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5005 PL_tokenbuf[0] = '%';
5007 /* Warn about @ where they meant $. */
5008 if (*s == '[' || *s == '{') {
5009 if (ckWARN(WARN_SYNTAX)) {
5010 const char *t = s + 1;
5011 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5013 if (*t == '}' || *t == ']') {
5015 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5016 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5017 "Scalar value %.*s better written as $%.*s",
5018 (int)(t-PL_bufptr), PL_bufptr,
5019 (int)(t-PL_bufptr-1), PL_bufptr+1);
5024 PL_pending_ident = '@';
5027 case '/': /* may be division, defined-or, or pattern */
5028 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5032 case '?': /* may either be conditional or pattern */
5033 if (PL_expect == XSTATE && s[1] == '?' && s[2] == '?') {
5037 if (PL_expect == XOPERATOR) {
5045 /* A // operator. */
5055 /* Disable warning on "study /blah/" */
5056 if (PL_oldoldbufptr == PL_last_uni
5057 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5058 || memNE(PL_last_uni, "study", 5)
5059 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5062 s = scan_pat(s,OP_MATCH);
5063 TERM(sublex_start());
5067 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5068 #ifdef PERL_STRICT_CR
5071 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5073 && (s == PL_linestart || s[-1] == '\n') )
5075 PL_lex_formbrack = 0;
5079 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5083 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5089 pl_yylval.ival = OPf_SPECIAL;
5095 if (PL_expect != XOPERATOR)
5100 case '0': case '1': case '2': case '3': case '4':
5101 case '5': case '6': case '7': case '8': case '9':
5102 s = scan_num(s, &pl_yylval);
5103 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5104 if (PL_expect == XOPERATOR)
5109 s = scan_str(s,!!PL_madskills,FALSE);
5110 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5111 if (PL_expect == XOPERATOR) {
5112 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5114 deprecate_old(commaless_variable_list);
5115 return REPORT(','); /* grandfather non-comma-format format */
5122 pl_yylval.ival = OP_CONST;
5123 TERM(sublex_start());
5126 s = scan_str(s,!!PL_madskills,FALSE);
5127 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5128 if (PL_expect == XOPERATOR) {
5129 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5131 deprecate_old(commaless_variable_list);
5132 return REPORT(','); /* grandfather non-comma-format format */
5139 pl_yylval.ival = OP_CONST;
5140 /* FIXME. I think that this can be const if char *d is replaced by
5141 more localised variables. */
5142 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5143 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5144 pl_yylval.ival = OP_STRINGIFY;
5148 TERM(sublex_start());
5151 s = scan_str(s,!!PL_madskills,FALSE);
5152 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5153 if (PL_expect == XOPERATOR)
5154 no_op("Backticks",s);
5157 readpipe_override();
5158 TERM(sublex_start());
5162 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5163 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5165 if (PL_expect == XOPERATOR)
5166 no_op("Backslash",s);
5170 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5171 char *start = s + 2;
5172 while (isDIGIT(*start) || *start == '_')
5174 if (*start == '.' && isDIGIT(start[1])) {
5175 s = scan_num(s, &pl_yylval);
5178 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5179 else if (!isALPHA(*start) && (PL_expect == XTERM
5180 || PL_expect == XREF || PL_expect == XSTATE
5181 || PL_expect == XTERMORDORDOR)) {
5182 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5184 s = scan_num(s, &pl_yylval);
5191 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5233 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5235 /* Some keywords can be followed by any delimiter, including ':' */
5236 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5237 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5238 (PL_tokenbuf[0] == 'q' &&
5239 strchr("qwxr", PL_tokenbuf[1])))));
5241 /* x::* is just a word, unless x is "CORE" */
5242 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5246 while (d < PL_bufend && isSPACE(*d))
5247 d++; /* no comments skipped here, or s### is misparsed */
5249 /* Is this a label? */
5250 if (!tmp && PL_expect == XSTATE
5251 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5253 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5258 /* Check for keywords */
5259 tmp = keyword(PL_tokenbuf, len, 0);
5261 /* Is this a word before a => operator? */
5262 if (*d == '=' && d[1] == '>') {
5265 = (OP*)newSVOP(OP_CONST, 0,
5266 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5267 pl_yylval.opval->op_private = OPpCONST_BARE;
5271 if (tmp < 0) { /* second-class keyword? */
5272 GV *ogv = NULL; /* override (winner) */
5273 GV *hgv = NULL; /* hidden (loser) */
5274 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5276 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5279 if (GvIMPORTED_CV(gv))
5281 else if (! CvMETHOD(cv))
5285 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5286 (gv = *gvp) && isGV_with_GP(gv) &&
5287 GvCVu(gv) && GvIMPORTED_CV(gv))
5294 tmp = 0; /* overridden by import or by GLOBAL */
5297 && -tmp==KEY_lock /* XXX generalizable kludge */
5300 tmp = 0; /* any sub overrides "weak" keyword */
5302 else { /* no override */
5304 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5305 Perl_warner(aTHX_ packWARN(WARN_MISC),
5306 "dump() better written as CORE::dump()");
5310 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5311 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5312 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5313 "Ambiguous call resolved as CORE::%s(), %s",
5314 GvENAME(hgv), "qualify as such or use &");
5321 default: /* not a keyword */
5322 /* Trade off - by using this evil construction we can pull the
5323 variable gv into the block labelled keylookup. If not, then
5324 we have to give it function scope so that the goto from the
5325 earlier ':' case doesn't bypass the initialisation. */
5327 just_a_word_zero_gv:
5335 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5338 SV *nextPL_nextwhite = 0;
5342 /* Get the rest if it looks like a package qualifier */
5344 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5346 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5349 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5350 *s == '\'' ? "'" : "::");
5355 if (PL_expect == XOPERATOR) {
5356 if (PL_bufptr == PL_linestart) {
5357 CopLINE_dec(PL_curcop);
5358 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
5359 CopLINE_inc(PL_curcop);
5362 no_op("Bareword",s);
5365 /* Look for a subroutine with this name in current package,
5366 unless name is "Foo::", in which case Foo is a bearword
5367 (and a package name). */
5369 if (len > 2 && !PL_madskills &&
5370 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5372 if (ckWARN(WARN_BAREWORD)
5373 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5374 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5375 "Bareword \"%s\" refers to nonexistent package",
5378 PL_tokenbuf[len] = '\0';
5384 /* Mustn't actually add anything to a symbol table.
5385 But also don't want to "initialise" any placeholder
5386 constants that might already be there into full
5387 blown PVGVs with attached PVCV. */
5388 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5389 GV_NOADD_NOINIT, SVt_PVCV);
5394 /* if we saw a global override before, get the right name */
5397 sv = newSVpvs("CORE::GLOBAL::");
5398 sv_catpv(sv,PL_tokenbuf);
5401 /* If len is 0, newSVpv does strlen(), which is correct.
5402 If len is non-zero, then it will be the true length,
5403 and so the scalar will be created correctly. */
5404 sv = newSVpv(PL_tokenbuf,len);
5407 if (PL_madskills && !PL_thistoken) {
5408 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5409 PL_thistoken = newSVpvn(start,s - start);
5410 PL_realtokenstart = s - SvPVX(PL_linestr);
5414 /* Presume this is going to be a bareword of some sort. */
5417 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5418 pl_yylval.opval->op_private = OPpCONST_BARE;
5419 /* UTF-8 package name? */
5420 if (UTF && !IN_BYTES &&
5421 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5424 /* And if "Foo::", then that's what it certainly is. */
5429 /* Do the explicit type check so that we don't need to force
5430 the initialisation of the symbol table to have a real GV.
5431 Beware - gv may not really be a PVGV, cv may not really be
5432 a PVCV, (because of the space optimisations that gv_init
5433 understands) But they're true if for this symbol there is
5434 respectively a typeglob and a subroutine.
5436 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5437 /* Real typeglob, so get the real subroutine: */
5439 /* A proxy for a subroutine in this package? */
5440 : SvOK(gv) ? (CV *) gv : NULL)
5443 /* See if it's the indirect object for a list operator. */
5445 if (PL_oldoldbufptr &&
5446 PL_oldoldbufptr < PL_bufptr &&
5447 (PL_oldoldbufptr == PL_last_lop
5448 || PL_oldoldbufptr == PL_last_uni) &&
5449 /* NO SKIPSPACE BEFORE HERE! */
5450 (PL_expect == XREF ||
5451 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5453 bool immediate_paren = *s == '(';
5455 /* (Now we can afford to cross potential line boundary.) */
5456 s = SKIPSPACE2(s,nextPL_nextwhite);
5458 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5461 /* Two barewords in a row may indicate method call. */
5463 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5464 (tmp = intuit_method(s, gv, cv)))
5467 /* If not a declared subroutine, it's an indirect object. */
5468 /* (But it's an indir obj regardless for sort.) */
5469 /* Also, if "_" follows a filetest operator, it's a bareword */
5472 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5474 (PL_last_lop_op != OP_MAPSTART &&
5475 PL_last_lop_op != OP_GREPSTART))))
5476 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5477 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5480 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5485 PL_expect = XOPERATOR;
5488 s = SKIPSPACE2(s,nextPL_nextwhite);
5489 PL_nextwhite = nextPL_nextwhite;
5494 /* Is this a word before a => operator? */
5495 if (*s == '=' && s[1] == '>' && !pkgname) {
5497 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5498 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5499 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5503 /* If followed by a paren, it's certainly a subroutine. */
5508 while (SPACE_OR_TAB(*d))
5510 if (*d == ')' && (sv = gv_const_sv(gv))) {
5517 PL_nextwhite = PL_thiswhite;
5520 start_force(PL_curforce);
5522 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5523 PL_expect = XOPERATOR;
5526 PL_nextwhite = nextPL_nextwhite;
5527 curmad('X', PL_thistoken);
5528 PL_thistoken = newSVpvs("");
5536 /* If followed by var or block, call it a method (unless sub) */
5538 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5539 PL_last_lop = PL_oldbufptr;
5540 PL_last_lop_op = OP_METHOD;
5544 /* If followed by a bareword, see if it looks like indir obj. */
5547 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5548 && (tmp = intuit_method(s, gv, cv)))
5551 /* Not a method, so call it a subroutine (if defined) */
5554 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5555 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5556 "Ambiguous use of -%s resolved as -&%s()",
5557 PL_tokenbuf, PL_tokenbuf);
5558 /* Check for a constant sub */
5559 if ((sv = gv_const_sv(gv))) {
5561 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5562 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5563 pl_yylval.opval->op_private = 0;
5567 /* Resolve to GV now. */
5568 if (SvTYPE(gv) != SVt_PVGV) {
5569 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5570 assert (SvTYPE(gv) == SVt_PVGV);
5571 /* cv must have been some sort of placeholder, so
5572 now needs replacing with a real code reference. */
5576 op_free(pl_yylval.opval);
5577 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5578 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5579 PL_last_lop = PL_oldbufptr;
5580 PL_last_lop_op = OP_ENTERSUB;
5581 /* Is there a prototype? */
5589 const char *proto = SvPV_const((SV*)cv, protolen);
5592 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5594 while (*proto == ';')
5596 if (*proto == '&' && *s == '{') {
5598 sv_setpvs(PL_subname, "__ANON__");
5600 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5607 PL_nextwhite = PL_thiswhite;
5610 start_force(PL_curforce);
5611 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5614 PL_nextwhite = nextPL_nextwhite;
5615 curmad('X', PL_thistoken);
5616 PL_thistoken = newSVpvs("");
5623 /* Guess harder when madskills require "best effort". */
5624 if (PL_madskills && (!gv || !GvCVu(gv))) {
5625 int probable_sub = 0;
5626 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5628 else if (isALPHA(*s)) {
5632 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5633 if (!keyword(tmpbuf, tmplen, 0))
5636 while (d < PL_bufend && isSPACE(*d))
5638 if (*d == '=' && d[1] == '>')
5643 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5644 op_free(pl_yylval.opval);
5645 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5646 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5647 PL_last_lop = PL_oldbufptr;
5648 PL_last_lop_op = OP_ENTERSUB;
5649 PL_nextwhite = PL_thiswhite;
5651 start_force(PL_curforce);
5652 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5654 PL_nextwhite = nextPL_nextwhite;
5655 curmad('X', PL_thistoken);
5656 PL_thistoken = newSVpvs("");
5661 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5668 /* Call it a bare word */
5671 if (PL_hints & HINT_STRICT_SUBS)
5672 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5674 if (lastchar != '-') {
5675 if (ckWARN(WARN_RESERVED)) {
5679 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5680 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5687 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5688 && ckWARN_d(WARN_AMBIGUOUS)) {
5689 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5690 "Operator or semicolon missing before %c%s",
5691 lastchar, PL_tokenbuf);
5692 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5693 "Ambiguous use of %c resolved as operator %c",
5694 lastchar, lastchar);
5700 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5701 newSVpv(CopFILE(PL_curcop),0));
5705 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5706 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5709 case KEY___PACKAGE__:
5710 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5712 ? newSVhek(HvNAME_HEK(PL_curstash))
5719 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5720 const char *pname = "main";
5721 if (PL_tokenbuf[2] == 'D')
5722 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5723 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5727 GvIOp(gv) = newIO();
5728 IoIFP(GvIOp(gv)) = PL_rsfp;
5729 #if defined(HAS_FCNTL) && defined(F_SETFD)
5731 const int fd = PerlIO_fileno(PL_rsfp);
5732 fcntl(fd,F_SETFD,fd >= 3);
5735 /* Mark this internal pseudo-handle as clean */
5736 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5737 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5738 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5740 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5741 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5742 /* if the script was opened in binmode, we need to revert
5743 * it to text mode for compatibility; but only iff it has CRs
5744 * XXX this is a questionable hack at best. */
5745 if (PL_bufend-PL_bufptr > 2
5746 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5749 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5750 loc = PerlIO_tell(PL_rsfp);
5751 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5754 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5756 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5757 #endif /* NETWARE */
5758 #ifdef PERLIO_IS_STDIO /* really? */
5759 # if defined(__BORLANDC__)
5760 /* XXX see note in do_binmode() */
5761 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5765 PerlIO_seek(PL_rsfp, loc, 0);
5769 #ifdef PERLIO_LAYERS
5772 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5773 else if (PL_encoding) {
5780 XPUSHs(PL_encoding);
5782 call_method("name", G_SCALAR);
5786 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5787 Perl_form(aTHX_ ":encoding(%"SVf")",
5796 if (PL_realtokenstart >= 0) {
5797 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5799 PL_endwhite = newSVpvs("");
5800 sv_catsv(PL_endwhite, PL_thiswhite);
5802 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5803 PL_realtokenstart = -1;
5805 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5806 SvCUR(PL_endwhite))) != NULL) ;
5821 if (PL_expect == XSTATE) {
5828 if (*s == ':' && s[1] == ':') {
5831 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5832 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5833 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5836 else if (tmp == KEY_require || tmp == KEY_do)
5837 /* that's a way to remember we saw "CORE::" */
5850 LOP(OP_ACCEPT,XTERM);
5856 LOP(OP_ATAN2,XTERM);
5862 LOP(OP_BINMODE,XTERM);
5865 LOP(OP_BLESS,XTERM);
5874 /* When 'use switch' is in effect, continue has a dual
5875 life as a control operator. */
5877 if (!FEATURE_IS_ENABLED("switch"))
5880 /* We have to disambiguate the two senses of
5881 "continue". If the next token is a '{' then
5882 treat it as the start of a continue block;
5883 otherwise treat it as a control operator.
5895 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5912 if (!PL_cryptseen) {
5913 PL_cryptseen = TRUE;
5917 LOP(OP_CRYPT,XTERM);
5920 LOP(OP_CHMOD,XTERM);
5923 LOP(OP_CHOWN,XTERM);
5926 LOP(OP_CONNECT,XTERM);
5945 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5946 if (orig_keyword == KEY_do) {
5955 PL_hints |= HINT_BLOCK_SCOPE;
5965 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5966 LOP(OP_DBMOPEN,XTERM);
5972 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5979 pl_yylval.ival = CopLINE(PL_curcop);
5995 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
5996 UNIBRACK(OP_ENTEREVAL);
6010 case KEY_endhostent:
6016 case KEY_endservent:
6019 case KEY_endprotoent:
6030 pl_yylval.ival = CopLINE(PL_curcop);
6032 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6035 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6038 if ((PL_bufend - p) >= 3 &&
6039 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6041 else if ((PL_bufend - p) >= 4 &&
6042 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6045 if (isIDFIRST_lazy_if(p,UTF)) {
6046 p = scan_ident(p, PL_bufend,
6047 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6051 Perl_croak(aTHX_ "Missing $ on loop variable");
6053 s = SvPVX(PL_linestr) + soff;
6059 LOP(OP_FORMLINE,XTERM);
6065 LOP(OP_FCNTL,XTERM);
6071 LOP(OP_FLOCK,XTERM);
6080 LOP(OP_GREPSTART, XREF);
6083 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6098 case KEY_getpriority:
6099 LOP(OP_GETPRIORITY,XTERM);
6101 case KEY_getprotobyname:
6104 case KEY_getprotobynumber:
6105 LOP(OP_GPBYNUMBER,XTERM);
6107 case KEY_getprotoent:
6119 case KEY_getpeername:
6120 UNI(OP_GETPEERNAME);
6122 case KEY_gethostbyname:
6125 case KEY_gethostbyaddr:
6126 LOP(OP_GHBYADDR,XTERM);
6128 case KEY_gethostent:
6131 case KEY_getnetbyname:
6134 case KEY_getnetbyaddr:
6135 LOP(OP_GNBYADDR,XTERM);
6140 case KEY_getservbyname:
6141 LOP(OP_GSBYNAME,XTERM);
6143 case KEY_getservbyport:
6144 LOP(OP_GSBYPORT,XTERM);
6146 case KEY_getservent:
6149 case KEY_getsockname:
6150 UNI(OP_GETSOCKNAME);
6152 case KEY_getsockopt:
6153 LOP(OP_GSOCKOPT,XTERM);
6168 pl_yylval.ival = CopLINE(PL_curcop);
6178 pl_yylval.ival = CopLINE(PL_curcop);
6182 LOP(OP_INDEX,XTERM);
6188 LOP(OP_IOCTL,XTERM);
6200 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6232 LOP(OP_LISTEN,XTERM);
6241 s = scan_pat(s,OP_MATCH);
6242 TERM(sublex_start());
6245 LOP(OP_MAPSTART, XREF);
6248 LOP(OP_MKDIR,XTERM);
6251 LOP(OP_MSGCTL,XTERM);
6254 LOP(OP_MSGGET,XTERM);
6257 LOP(OP_MSGRCV,XTERM);
6260 LOP(OP_MSGSND,XTERM);
6265 PL_in_my = (U16)tmp;
6267 if (isIDFIRST_lazy_if(s,UTF)) {
6271 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6272 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6274 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6275 if (!PL_in_my_stash) {
6278 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6282 if (PL_madskills) { /* just add type to declarator token */
6283 sv_catsv(PL_thistoken, PL_nextwhite);
6285 sv_catpvn(PL_thistoken, start, s - start);
6293 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6300 s = tokenize_use(0, s);
6304 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6311 if (isIDFIRST_lazy_if(s,UTF)) {
6313 for (d = s; isALNUM_lazy_if(d,UTF);)
6315 for (t=d; isSPACE(*t);)
6317 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6319 && !(t[0] == '=' && t[1] == '>')
6321 int parms_len = (int)(d-s);
6322 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6323 "Precedence problem: open %.*s should be open(%.*s)",
6324 parms_len, s, parms_len, s);
6330 pl_yylval.ival = OP_OR;
6340 LOP(OP_OPEN_DIR,XTERM);
6343 checkcomma(s,PL_tokenbuf,"filehandle");
6347 checkcomma(s,PL_tokenbuf,"filehandle");
6366 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6370 LOP(OP_PIPE_OP,XTERM);
6373 s = scan_str(s,!!PL_madskills,FALSE);
6376 pl_yylval.ival = OP_CONST;
6377 TERM(sublex_start());
6383 s = scan_str(s,!!PL_madskills,FALSE);
6386 PL_expect = XOPERATOR;
6388 if (SvCUR(PL_lex_stuff)) {
6391 d = SvPV_force(PL_lex_stuff, len);
6393 for (; isSPACE(*d) && len; --len, ++d)
6398 if (!warned && ckWARN(WARN_QW)) {
6399 for (; !isSPACE(*d) && len; --len, ++d) {
6401 Perl_warner(aTHX_ packWARN(WARN_QW),
6402 "Possible attempt to separate words with commas");
6405 else if (*d == '#') {
6406 Perl_warner(aTHX_ packWARN(WARN_QW),
6407 "Possible attempt to put comments in qw() list");
6413 for (; !isSPACE(*d) && len; --len, ++d)
6416 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6417 words = append_elem(OP_LIST, words,
6418 newSVOP(OP_CONST, 0, tokeq(sv)));
6422 start_force(PL_curforce);
6423 NEXTVAL_NEXTTOKE.opval = words;
6428 SvREFCNT_dec(PL_lex_stuff);
6429 PL_lex_stuff = NULL;
6435 s = scan_str(s,!!PL_madskills,FALSE);
6438 pl_yylval.ival = OP_STRINGIFY;
6439 if (SvIVX(PL_lex_stuff) == '\'')
6440 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6441 TERM(sublex_start());
6444 s = scan_pat(s,OP_QR);
6445 TERM(sublex_start());
6448 s = scan_str(s,!!PL_madskills,FALSE);
6451 readpipe_override();
6452 TERM(sublex_start());
6460 s = force_version(s, FALSE);
6462 else if (*s != 'v' || !isDIGIT(s[1])
6463 || (s = force_version(s, TRUE), *s == 'v'))
6465 *PL_tokenbuf = '\0';
6466 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6467 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6468 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6470 yyerror("<> should be quotes");
6472 if (orig_keyword == KEY_require) {
6480 PL_last_uni = PL_oldbufptr;
6481 PL_last_lop_op = OP_REQUIRE;
6483 return REPORT( (int)REQUIRE );
6489 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6493 LOP(OP_RENAME,XTERM);
6502 LOP(OP_RINDEX,XTERM);
6511 UNIDOR(OP_READLINE);
6514 UNIDOR(OP_BACKTICK);
6523 LOP(OP_REVERSE,XTERM);
6526 UNIDOR(OP_READLINK);
6533 if (pl_yylval.opval)
6534 TERM(sublex_start());
6536 TOKEN(1); /* force error */
6539 checkcomma(s,PL_tokenbuf,"filehandle");
6549 LOP(OP_SELECT,XTERM);
6555 LOP(OP_SEMCTL,XTERM);
6558 LOP(OP_SEMGET,XTERM);
6561 LOP(OP_SEMOP,XTERM);
6567 LOP(OP_SETPGRP,XTERM);
6569 case KEY_setpriority:
6570 LOP(OP_SETPRIORITY,XTERM);
6572 case KEY_sethostent:
6578 case KEY_setservent:
6581 case KEY_setprotoent:
6591 LOP(OP_SEEKDIR,XTERM);
6593 case KEY_setsockopt:
6594 LOP(OP_SSOCKOPT,XTERM);
6600 LOP(OP_SHMCTL,XTERM);
6603 LOP(OP_SHMGET,XTERM);
6606 LOP(OP_SHMREAD,XTERM);
6609 LOP(OP_SHMWRITE,XTERM);
6612 LOP(OP_SHUTDOWN,XTERM);
6621 LOP(OP_SOCKET,XTERM);
6623 case KEY_socketpair:
6624 LOP(OP_SOCKPAIR,XTERM);
6627 checkcomma(s,PL_tokenbuf,"subroutine name");
6629 if (*s == ';' || *s == ')') /* probably a close */
6630 Perl_croak(aTHX_ "sort is now a reserved word");
6632 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6636 LOP(OP_SPLIT,XTERM);
6639 LOP(OP_SPRINTF,XTERM);
6642 LOP(OP_SPLICE,XTERM);
6657 LOP(OP_SUBSTR,XTERM);
6663 char tmpbuf[sizeof PL_tokenbuf];
6664 SSize_t tboffset = 0;
6665 expectation attrful;
6666 bool have_name, have_proto;
6667 const int key = tmp;
6672 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6673 SV *subtoken = newSVpvn(tstart, s - tstart);
6677 s = SKIPSPACE2(s,tmpwhite);
6682 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6683 (*s == ':' && s[1] == ':'))
6686 SV *nametoke = NULL;
6690 attrful = XATTRBLOCK;
6691 /* remember buffer pos'n for later force_word */
6692 tboffset = s - PL_oldbufptr;
6693 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6696 nametoke = newSVpvn(s, d - s);
6698 if (memchr(tmpbuf, ':', len))
6699 sv_setpvn(PL_subname, tmpbuf, len);
6701 sv_setsv(PL_subname,PL_curstname);
6702 sv_catpvs(PL_subname,"::");
6703 sv_catpvn(PL_subname,tmpbuf,len);
6710 CURMAD('X', nametoke);
6711 CURMAD('_', tmpwhite);
6712 (void) force_word(PL_oldbufptr + tboffset, WORD,
6715 s = SKIPSPACE2(d,tmpwhite);
6722 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6723 PL_expect = XTERMBLOCK;
6724 attrful = XATTRTERM;
6725 sv_setpvn(PL_subname,"?",1);
6729 if (key == KEY_format) {
6731 PL_lex_formbrack = PL_lex_brackets + 1;
6733 PL_thistoken = subtoken;
6737 (void) force_word(PL_oldbufptr + tboffset, WORD,
6743 /* Look for a prototype */
6746 bool bad_proto = FALSE;
6747 bool in_brackets = FALSE;
6748 char greedy_proto = ' ';
6749 bool proto_after_greedy_proto = FALSE;
6750 bool must_be_last = FALSE;
6751 bool underscore = FALSE;
6752 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6754 s = scan_str(s,!!PL_madskills,FALSE);
6756 Perl_croak(aTHX_ "Prototype not terminated");
6757 /* strip spaces and check for bad characters */
6758 d = SvPVX(PL_lex_stuff);
6760 for (p = d; *p; ++p) {
6766 proto_after_greedy_proto = TRUE;
6767 if (!strchr("$@%*;[]&\\_", *p)) {
6779 else if ( *p == ']' ) {
6780 in_brackets = FALSE;
6782 else if ( (*p == '@' || *p == '%') &&
6783 ( tmp < 2 || d[tmp-2] != '\\' ) &&
6785 must_be_last = TRUE;
6788 else if ( *p == '_' ) {
6796 if (proto_after_greedy_proto)
6797 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6798 "Prototype after '%c' for %"SVf" : %s",
6799 greedy_proto, SVfARG(PL_subname), d);
6801 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6802 "Illegal character in prototype for %"SVf" : %s",
6803 SVfARG(PL_subname), d);
6804 SvCUR_set(PL_lex_stuff, tmp);
6809 CURMAD('q', PL_thisopen);
6810 CURMAD('_', tmpwhite);
6811 CURMAD('=', PL_thisstuff);
6812 CURMAD('Q', PL_thisclose);
6813 NEXTVAL_NEXTTOKE.opval =
6814 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6815 PL_lex_stuff = NULL;
6818 s = SKIPSPACE2(s,tmpwhite);
6826 if (*s == ':' && s[1] != ':')
6827 PL_expect = attrful;
6828 else if (*s != '{' && key == KEY_sub) {
6830 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6832 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6839 curmad('^', newSVpvs(""));
6840 CURMAD('_', tmpwhite);
6844 PL_thistoken = subtoken;
6847 NEXTVAL_NEXTTOKE.opval =
6848 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6849 PL_lex_stuff = NULL;
6855 sv_setpvs(PL_subname, "__ANON__");
6857 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6861 (void) force_word(PL_oldbufptr + tboffset, WORD,
6870 LOP(OP_SYSTEM,XREF);
6873 LOP(OP_SYMLINK,XTERM);
6876 LOP(OP_SYSCALL,XTERM);
6879 LOP(OP_SYSOPEN,XTERM);
6882 LOP(OP_SYSSEEK,XTERM);
6885 LOP(OP_SYSREAD,XTERM);
6888 LOP(OP_SYSWRITE,XTERM);
6892 TERM(sublex_start());
6913 LOP(OP_TRUNCATE,XTERM);
6925 pl_yylval.ival = CopLINE(PL_curcop);
6929 pl_yylval.ival = CopLINE(PL_curcop);
6933 LOP(OP_UNLINK,XTERM);
6939 LOP(OP_UNPACK,XTERM);
6942 LOP(OP_UTIME,XTERM);
6948 LOP(OP_UNSHIFT,XTERM);
6951 s = tokenize_use(1, s);
6961 pl_yylval.ival = CopLINE(PL_curcop);
6965 pl_yylval.ival = CopLINE(PL_curcop);
6969 PL_hints |= HINT_BLOCK_SCOPE;
6976 LOP(OP_WAITPID,XTERM);
6985 ctl_l[0] = toCTRL('L');
6987 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6990 /* Make sure $^L is defined */
6991 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
6996 if (PL_expect == XOPERATOR)
7002 pl_yylval.ival = OP_XOR;
7007 TERM(sublex_start());
7012 #pragma segment Main
7016 S_pending_ident(pTHX)
7021 /* pit holds the identifier we read and pending_ident is reset */
7022 char pit = PL_pending_ident;
7023 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7024 /* All routes through this function want to know if there is a colon. */
7025 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7026 PL_pending_ident = 0;
7028 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7029 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7030 "### Pending identifier '%s'\n", PL_tokenbuf); });
7032 /* if we're in a my(), we can't allow dynamics here.
7033 $foo'bar has already been turned into $foo::bar, so
7034 just check for colons.
7036 if it's a legal name, the OP is a PADANY.
7039 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7041 yyerror(Perl_form(aTHX_ "No package name allowed for "
7042 "variable %s in \"our\"",
7044 tmp = allocmy(PL_tokenbuf);
7048 yyerror(Perl_form(aTHX_ PL_no_myglob,
7049 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7051 pl_yylval.opval = newOP(OP_PADANY, 0);
7052 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7058 build the ops for accesses to a my() variable.
7060 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7061 then used in a comparison. This catches most, but not
7062 all cases. For instance, it catches
7063 sort { my($a); $a <=> $b }
7065 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7066 (although why you'd do that is anyone's guess).
7071 tmp = pad_findmy(PL_tokenbuf);
7072 if (tmp != NOT_IN_PAD) {
7073 /* might be an "our" variable" */
7074 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7075 /* build ops for a bareword */
7076 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7077 HEK * const stashname = HvNAME_HEK(stash);
7078 SV * const sym = newSVhek(stashname);
7079 sv_catpvs(sym, "::");
7080 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7081 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7082 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7085 ? (GV_ADDMULTI | GV_ADDINEVAL)
7088 ((PL_tokenbuf[0] == '$') ? SVt_PV
7089 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7094 /* if it's a sort block and they're naming $a or $b */
7095 if (PL_last_lop_op == OP_SORT &&
7096 PL_tokenbuf[0] == '$' &&
7097 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7100 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7101 d < PL_bufend && *d != '\n';
7104 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7105 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7111 pl_yylval.opval = newOP(OP_PADANY, 0);
7112 pl_yylval.opval->op_targ = tmp;
7118 Whine if they've said @foo in a doublequoted string,
7119 and @foo isn't a variable we can find in the symbol
7122 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7123 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7125 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7126 && ckWARN(WARN_AMBIGUOUS)
7127 /* DO NOT warn for @- and @+ */
7128 && !( PL_tokenbuf[2] == '\0' &&
7129 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7132 /* Downgraded from fatal to warning 20000522 mjd */
7133 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7134 "Possible unintended interpolation of %s in string",
7139 /* build ops for a bareword */
7140 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7142 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7144 PL_tokenbuf + 1, tokenbuf_len - 1,
7145 /* If the identifier refers to a stash, don't autovivify it.
7146 * Change 24660 had the side effect of causing symbol table
7147 * hashes to always be defined, even if they were freshly
7148 * created and the only reference in the entire program was
7149 * the single statement with the defined %foo::bar:: test.
7150 * It appears that all code in the wild doing this actually
7151 * wants to know whether sub-packages have been loaded, so
7152 * by avoiding auto-vivifying symbol tables, we ensure that
7153 * defined %foo::bar:: continues to be false, and the existing
7154 * tests still give the expected answers, even though what
7155 * they're actually testing has now changed subtly.
7157 (*PL_tokenbuf == '%'
7158 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7161 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7162 ((PL_tokenbuf[0] == '$') ? SVt_PV
7163 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7169 * The following code was generated by perl_keyword.pl.
7173 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7177 PERL_ARGS_ASSERT_KEYWORD;
7181 case 1: /* 5 tokens of length 1 */
7213 case 2: /* 18 tokens of length 2 */
7359 case 3: /* 29 tokens of length 3 */
7363 if (name[1] == 'N' &&
7426 if (name[1] == 'i' &&
7458 if (name[1] == 'o' &&
7467 if (name[1] == 'e' &&
7476 if (name[1] == 'n' &&
7485 if (name[1] == 'o' &&
7494 if (name[1] == 'a' &&
7503 if (name[1] == 'o' &&
7565 if (name[1] == 'e' &&
7579 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7605 if (name[1] == 'i' &&
7614 if (name[1] == 's' &&
7623 if (name[1] == 'e' &&
7632 if (name[1] == 'o' &&
7644 case 4: /* 41 tokens of length 4 */
7648 if (name[1] == 'O' &&
7658 if (name[1] == 'N' &&
7668 if (name[1] == 'i' &&
7678 if (name[1] == 'h' &&
7688 if (name[1] == 'u' &&
7701 if (name[2] == 'c' &&
7710 if (name[2] == 's' &&
7719 if (name[2] == 'a' &&
7755 if (name[1] == 'o' &&
7768 if (name[2] == 't' &&
7777 if (name[2] == 'o' &&
7786 if (name[2] == 't' &&
7795 if (name[2] == 'e' &&
7808 if (name[1] == 'o' &&
7821 if (name[2] == 'y' &&
7830 if (name[2] == 'l' &&
7846 if (name[2] == 's' &&
7855 if (name[2] == 'n' &&
7864 if (name[2] == 'c' &&
7877 if (name[1] == 'e' &&
7887 if (name[1] == 'p' &&
7900 if (name[2] == 'c' &&
7909 if (name[2] == 'p' &&
7918 if (name[2] == 's' &&
7934 if (name[2] == 'n' &&
8004 if (name[2] == 'r' &&
8013 if (name[2] == 'r' &&
8022 if (name[2] == 'a' &&
8038 if (name[2] == 'l' &&
8100 if (name[2] == 'e' &&
8103 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8116 case 5: /* 39 tokens of length 5 */
8120 if (name[1] == 'E' &&
8131 if (name[1] == 'H' &&
8145 if (name[2] == 'a' &&
8155 if (name[2] == 'a' &&
8172 if (name[2] == 'e' &&
8182 if (name[2] == 'e' &&
8186 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8202 if (name[3] == 'i' &&
8211 if (name[3] == 'o' &&
8247 if (name[2] == 'o' &&
8257 if (name[2] == 'y' &&
8271 if (name[1] == 'l' &&
8285 if (name[2] == 'n' &&
8295 if (name[2] == 'o' &&
8309 if (name[1] == 'i' &&
8314 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8323 if (name[2] == 'd' &&
8333 if (name[2] == 'c' &&
8350 if (name[2] == 'c' &&
8360 if (name[2] == 't' &&
8374 if (name[1] == 'k' &&
8385 if (name[1] == 'r' &&
8399 if (name[2] == 's' &&
8409 if (name[2] == 'd' &&
8426 if (name[2] == 'm' &&
8436 if (name[2] == 'i' &&
8446 if (name[2] == 'e' &&
8456 if (name[2] == 'l' &&
8466 if (name[2] == 'a' &&
8479 if (name[3] == 't' &&
8482 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8488 if (name[3] == 'd' &&
8505 if (name[1] == 'i' &&
8519 if (name[2] == 'a' &&
8532 if (name[3] == 'e' &&
8567 if (name[2] == 'i' &&
8584 if (name[2] == 'i' &&
8594 if (name[2] == 'i' &&
8611 case 6: /* 33 tokens of length 6 */
8615 if (name[1] == 'c' &&
8630 if (name[2] == 'l' &&
8641 if (name[2] == 'r' &&
8656 if (name[1] == 'e' &&
8671 if (name[2] == 's' &&
8676 if(ckWARN_d(WARN_SYNTAX))
8677 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8683 if (name[2] == 'i' &&
8701 if (name[2] == 'l' &&
8712 if (name[2] == 'r' &&
8727 if (name[1] == 'm' &&
8742 if (name[2] == 'n' &&
8753 if (name[2] == 's' &&
8768 if (name[1] == 's' &&
8774 if (name[4] == 't' &&
8783 if (name[4] == 'e' &&
8792 if (name[4] == 'c' &&
8801 if (name[4] == 'n' &&
8817 if (name[1] == 'r' &&
8835 if (name[3] == 'a' &&
8845 if (name[3] == 'u' &&
8859 if (name[2] == 'n' &&
8877 if (name[2] == 'a' &&
8891 if (name[3] == 'e' &&
8904 if (name[4] == 't' &&
8913 if (name[4] == 'e' &&
8935 if (name[4] == 't' &&
8944 if (name[4] == 'e' &&
8960 if (name[2] == 'c' &&
8971 if (name[2] == 'l' &&
8982 if (name[2] == 'b' &&
8993 if (name[2] == 's' &&
9016 if (name[4] == 's' &&
9025 if (name[4] == 'n' &&
9038 if (name[3] == 'a' &&
9055 if (name[1] == 'a' &&
9070 case 7: /* 29 tokens of length 7 */
9074 if (name[1] == 'E' &&
9087 if (name[1] == '_' &&
9100 if (name[1] == 'i' &&
9107 return -KEY_binmode;
9113 if (name[1] == 'o' &&
9120 return -KEY_connect;
9129 if (name[2] == 'm' &&
9135 return -KEY_dbmopen;
9146 if (name[4] == 'u' &&
9150 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9156 if (name[4] == 'n' &&
9177 if (name[1] == 'o' &&
9190 if (name[1] == 'e' &&
9197 if (name[5] == 'r' &&
9200 return -KEY_getpgrp;
9206 if (name[5] == 'i' &&
9209 return -KEY_getppid;
9222 if (name[1] == 'c' &&
9229 return -KEY_lcfirst;
9235 if (name[1] == 'p' &&
9242 return -KEY_opendir;
9248 if (name[1] == 'a' &&
9266 if (name[3] == 'd' &&
9271 return -KEY_readdir;
9277 if (name[3] == 'u' &&
9288 if (name[3] == 'e' &&
9293 return -KEY_reverse;
9312 if (name[3] == 'k' &&
9317 return -KEY_seekdir;
9323 if (name[3] == 'p' &&
9328 return -KEY_setpgrp;
9338 if (name[2] == 'm' &&
9344 return -KEY_shmread;
9350 if (name[2] == 'r' &&
9356 return -KEY_sprintf;
9365 if (name[3] == 'l' &&
9370 return -KEY_symlink;
9379 if (name[4] == 'a' &&
9383 return -KEY_syscall;
9389 if (name[4] == 'p' &&
9393 return -KEY_sysopen;
9399 if (name[4] == 'e' &&
9403 return -KEY_sysread;
9409 if (name[4] == 'e' &&
9413 return -KEY_sysseek;
9431 if (name[1] == 'e' &&
9438 return -KEY_telldir;
9447 if (name[2] == 'f' &&
9453 return -KEY_ucfirst;
9459 if (name[2] == 's' &&
9465 return -KEY_unshift;
9475 if (name[1] == 'a' &&
9482 return -KEY_waitpid;
9491 case 8: /* 26 tokens of length 8 */
9495 if (name[1] == 'U' &&
9503 return KEY_AUTOLOAD;
9514 if (name[3] == 'A' &&
9520 return KEY___DATA__;
9526 if (name[3] == 'I' &&
9532 return -KEY___FILE__;
9538 if (name[3] == 'I' &&
9544 return -KEY___LINE__;
9560 if (name[2] == 'o' &&
9567 return -KEY_closedir;
9573 if (name[2] == 'n' &&
9580 return -KEY_continue;
9590 if (name[1] == 'b' &&
9598 return -KEY_dbmclose;
9604 if (name[1] == 'n' &&
9610 if (name[4] == 'r' &&
9615 return -KEY_endgrent;
9621 if (name[4] == 'w' &&
9626 return -KEY_endpwent;
9639 if (name[1] == 'o' &&
9647 return -KEY_formline;
9653 if (name[1] == 'e' &&
9664 if (name[6] == 'n' &&
9667 return -KEY_getgrent;
9673 if (name[6] == 'i' &&
9676 return -KEY_getgrgid;
9682 if (name[6] == 'a' &&
9685 return -KEY_getgrnam;
9698 if (name[4] == 'o' &&
9703 return -KEY_getlogin;
9714 if (name[6] == 'n' &&
9717 return -KEY_getpwent;
9723 if (name[6] == 'a' &&
9726 return -KEY_getpwnam;
9732 if (name[6] == 'i' &&
9735 return -KEY_getpwuid;
9755 if (name[1] == 'e' &&
9762 if (name[5] == 'i' &&
9769 return -KEY_readline;
9774 return -KEY_readlink;
9785 if (name[5] == 'i' &&
9789 return -KEY_readpipe;
9810 if (name[4] == 'r' &&
9815 return -KEY_setgrent;
9821 if (name[4] == 'w' &&
9826 return -KEY_setpwent;
9842 if (name[3] == 'w' &&
9848 return -KEY_shmwrite;
9854 if (name[3] == 't' &&
9860 return -KEY_shutdown;
9870 if (name[2] == 's' &&
9877 return -KEY_syswrite;
9887 if (name[1] == 'r' &&
9895 return -KEY_truncate;
9904 case 9: /* 9 tokens of length 9 */
9908 if (name[1] == 'N' &&
9917 return KEY_UNITCHECK;
9923 if (name[1] == 'n' &&
9932 return -KEY_endnetent;
9938 if (name[1] == 'e' &&
9947 return -KEY_getnetent;
9953 if (name[1] == 'o' &&
9962 return -KEY_localtime;
9968 if (name[1] == 'r' &&
9977 return KEY_prototype;
9983 if (name[1] == 'u' &&
9992 return -KEY_quotemeta;
9998 if (name[1] == 'e' &&
10007 return -KEY_rewinddir;
10013 if (name[1] == 'e' &&
10022 return -KEY_setnetent;
10028 if (name[1] == 'a' &&
10037 return -KEY_wantarray;
10046 case 10: /* 9 tokens of length 10 */
10050 if (name[1] == 'n' &&
10056 if (name[4] == 'o' &&
10063 return -KEY_endhostent;
10069 if (name[4] == 'e' &&
10076 return -KEY_endservent;
10089 if (name[1] == 'e' &&
10095 if (name[4] == 'o' &&
10102 return -KEY_gethostent;
10111 if (name[5] == 'r' &&
10117 return -KEY_getservent;
10123 if (name[5] == 'c' &&
10129 return -KEY_getsockopt;
10149 if (name[2] == 't')
10154 if (name[4] == 'o' &&
10161 return -KEY_sethostent;
10170 if (name[5] == 'r' &&
10176 return -KEY_setservent;
10182 if (name[5] == 'c' &&
10188 return -KEY_setsockopt;
10205 if (name[2] == 'c' &&
10214 return -KEY_socketpair;
10227 case 11: /* 8 tokens of length 11 */
10231 if (name[1] == '_' &&
10241 { /* __PACKAGE__ */
10242 return -KEY___PACKAGE__;
10248 if (name[1] == 'n' &&
10258 { /* endprotoent */
10259 return -KEY_endprotoent;
10265 if (name[1] == 'e' &&
10274 if (name[5] == 'e' &&
10280 { /* getpeername */
10281 return -KEY_getpeername;
10290 if (name[6] == 'o' &&
10295 { /* getpriority */
10296 return -KEY_getpriority;
10302 if (name[6] == 't' &&
10307 { /* getprotoent */
10308 return -KEY_getprotoent;
10322 if (name[4] == 'o' &&
10329 { /* getsockname */
10330 return -KEY_getsockname;
10343 if (name[1] == 'e' &&
10351 if (name[6] == 'o' &&
10356 { /* setpriority */
10357 return -KEY_setpriority;
10363 if (name[6] == 't' &&
10368 { /* setprotoent */
10369 return -KEY_setprotoent;
10385 case 12: /* 2 tokens of length 12 */
10386 if (name[0] == 'g' &&
10398 if (name[9] == 'd' &&
10401 { /* getnetbyaddr */
10402 return -KEY_getnetbyaddr;
10408 if (name[9] == 'a' &&
10411 { /* getnetbyname */
10412 return -KEY_getnetbyname;
10424 case 13: /* 4 tokens of length 13 */
10425 if (name[0] == 'g' &&
10432 if (name[4] == 'o' &&
10441 if (name[10] == 'd' &&
10444 { /* gethostbyaddr */
10445 return -KEY_gethostbyaddr;
10451 if (name[10] == 'a' &&
10454 { /* gethostbyname */
10455 return -KEY_gethostbyname;
10468 if (name[4] == 'e' &&
10477 if (name[10] == 'a' &&
10480 { /* getservbyname */
10481 return -KEY_getservbyname;
10487 if (name[10] == 'o' &&
10490 { /* getservbyport */
10491 return -KEY_getservbyport;
10510 case 14: /* 1 tokens of length 14 */
10511 if (name[0] == 'g' &&
10525 { /* getprotobyname */
10526 return -KEY_getprotobyname;
10531 case 16: /* 1 tokens of length 16 */
10532 if (name[0] == 'g' &&
10548 { /* getprotobynumber */
10549 return -KEY_getprotobynumber;
10563 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10567 PERL_ARGS_ASSERT_CHECKCOMMA;
10569 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10570 if (ckWARN(WARN_SYNTAX)) {
10573 for (w = s+2; *w && level; w++) {
10576 else if (*w == ')')
10579 while (isSPACE(*w))
10581 /* the list of chars below is for end of statements or
10582 * block / parens, boolean operators (&&, ||, //) and branch
10583 * constructs (or, and, if, until, unless, while, err, for).
10584 * Not a very solid hack... */
10585 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10586 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10587 "%s (...) interpreted as function",name);
10590 while (s < PL_bufend && isSPACE(*s))
10594 while (s < PL_bufend && isSPACE(*s))
10596 if (isIDFIRST_lazy_if(s,UTF)) {
10597 const char * const w = s++;
10598 while (isALNUM_lazy_if(s,UTF))
10600 while (s < PL_bufend && isSPACE(*s))
10604 if (keyword(w, s - w, 0))
10607 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10608 if (gv && GvCVu(gv))
10610 Perl_croak(aTHX_ "No comma allowed after %s", what);
10615 /* Either returns sv, or mortalizes sv and returns a new SV*.
10616 Best used as sv=new_constant(..., sv, ...).
10617 If s, pv are NULL, calls subroutine with one argument,
10618 and type is used with error messages only. */
10621 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10622 SV *sv, SV *pv, const char *type, STRLEN typelen)
10625 HV * const table = GvHV(PL_hintgv); /* ^H */
10629 const char *why1 = "", *why2 = "", *why3 = "";
10631 PERL_ARGS_ASSERT_NEW_CONSTANT;
10633 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10636 why2 = (const char *)
10637 (strEQ(key,"charnames")
10638 ? "(possibly a missing \"use charnames ...\")"
10640 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10641 (type ? type: "undef"), why2);
10643 /* This is convoluted and evil ("goto considered harmful")
10644 * but I do not understand the intricacies of all the different
10645 * failure modes of %^H in here. The goal here is to make
10646 * the most probable error message user-friendly. --jhi */
10651 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10652 (type ? type: "undef"), why1, why2, why3);
10654 yyerror(SvPVX_const(msg));
10658 cvp = hv_fetch(table, key, keylen, FALSE);
10659 if (!cvp || !SvOK(*cvp)) {
10662 why3 = "} is not defined";
10665 sv_2mortal(sv); /* Parent created it permanently */
10668 pv = newSVpvn_flags(s, len, SVs_TEMP);
10670 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10672 typesv = &PL_sv_undef;
10674 PUSHSTACKi(PERLSI_OVERLOAD);
10686 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10690 /* Check the eval first */
10691 if (!PL_in_eval && SvTRUE(ERRSV)) {
10692 sv_catpvs(ERRSV, "Propagated");
10693 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10695 res = SvREFCNT_inc_simple(sv);
10699 SvREFCNT_inc_simple_void(res);
10708 why1 = "Call to &{$^H{";
10710 why3 = "}} did not return a defined value";
10718 /* Returns a NUL terminated string, with the length of the string written to
10722 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10725 register char *d = dest;
10726 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10728 PERL_ARGS_ASSERT_SCAN_WORD;
10732 Perl_croak(aTHX_ ident_too_long);
10733 if (isALNUM(*s)) /* UTF handled below */
10735 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10740 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10744 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10745 char *t = s + UTF8SKIP(s);
10747 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10751 Perl_croak(aTHX_ ident_too_long);
10752 Copy(s, d, len, char);
10765 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10768 char *bracket = NULL;
10770 register char *d = dest;
10771 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10773 PERL_ARGS_ASSERT_SCAN_IDENT;
10778 while (isDIGIT(*s)) {
10780 Perl_croak(aTHX_ ident_too_long);
10787 Perl_croak(aTHX_ ident_too_long);
10788 if (isALNUM(*s)) /* UTF handled below */
10790 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10795 else if (*s == ':' && s[1] == ':') {
10799 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10800 char *t = s + UTF8SKIP(s);
10801 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10803 if (d + (t - s) > e)
10804 Perl_croak(aTHX_ ident_too_long);
10805 Copy(s, d, t - s, char);
10816 if (PL_lex_state != LEX_NORMAL)
10817 PL_lex_state = LEX_INTERPENDMAYBE;
10820 if (*s == '$' && s[1] &&
10821 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10834 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10839 if (isSPACE(s[-1])) {
10841 const char ch = *s++;
10842 if (!SPACE_OR_TAB(ch)) {
10848 if (isIDFIRST_lazy_if(d,UTF)) {
10852 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10853 end += UTF8SKIP(end);
10854 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10855 end += UTF8SKIP(end);
10857 Copy(s, d, end - s, char);
10862 while ((isALNUM(*s) || *s == ':') && d < e)
10865 Perl_croak(aTHX_ ident_too_long);
10868 while (s < send && SPACE_OR_TAB(*s))
10870 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10871 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10872 const char * const brack =
10874 ((*s == '[') ? "[...]" : "{...}");
10875 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10876 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10877 funny, dest, brack, funny, dest, brack);
10880 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10884 /* Handle extended ${^Foo} variables
10885 * 1999-02-27 mjd-perl-patch@plover.com */
10886 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10890 while (isALNUM(*s) && d < e) {
10894 Perl_croak(aTHX_ ident_too_long);
10899 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10900 PL_lex_state = LEX_INTERPEND;
10903 if (PL_lex_state == LEX_NORMAL) {
10904 if (ckWARN(WARN_AMBIGUOUS) &&
10905 (keyword(dest, d - dest, 0)
10906 || get_cvn_flags(dest, d - dest, 0)))
10910 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10911 "Ambiguous use of %c{%s} resolved to %c%s",
10912 funny, dest, funny, dest);
10917 s = bracket; /* let the parser handle it */
10921 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10922 PL_lex_state = LEX_INTERPEND;
10927 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10929 PERL_ARGS_ASSERT_PMFLAG;
10931 PERL_UNUSED_CONTEXT;
10933 const char c = (char)ch;
10935 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10936 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10937 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10938 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10939 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10945 S_scan_pat(pTHX_ char *start, I32 type)
10949 char *s = scan_str(start,!!PL_madskills,FALSE);
10950 const char * const valid_flags =
10951 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10956 PERL_ARGS_ASSERT_SCAN_PAT;
10959 const char * const delimiter = skipspace(start);
10963 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10964 : "Search pattern not terminated" ));
10967 pm = (PMOP*)newPMOP(type, 0);
10968 if (PL_multi_open == '?') {
10969 /* This is the only point in the code that sets PMf_ONCE: */
10970 pm->op_pmflags |= PMf_ONCE;
10972 /* Hence it's safe to do this bit of PMOP book-keeping here, which
10973 allows us to restrict the list needed by reset to just the ??
10975 assert(type != OP_TRANS);
10977 MAGIC *mg = mg_find((SV*)PL_curstash, PERL_MAGIC_symtab);
10980 mg = sv_magicext((SV*)PL_curstash, 0, PERL_MAGIC_symtab, 0, 0,
10983 elements = mg->mg_len / sizeof(PMOP**);
10984 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
10985 ((PMOP**)mg->mg_ptr) [elements++] = pm;
10986 mg->mg_len = elements * sizeof(PMOP**);
10987 PmopSTASH_set(pm,PL_curstash);
10993 while (*s && strchr(valid_flags, *s))
10994 pmflag(&pm->op_pmflags,*s++);
10996 if (PL_madskills && modstart != s) {
10997 SV* tmptoken = newSVpvn(modstart, s - modstart);
10998 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11001 /* issue a warning if /c is specified,but /g is not */
11002 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
11003 && ckWARN(WARN_REGEXP))
11005 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
11006 "Use of /c modifier is meaningless without /g" );
11009 PL_lex_op = (OP*)pm;
11010 pl_yylval.ival = OP_MATCH;
11015 S_scan_subst(pTHX_ char *start)
11026 PERL_ARGS_ASSERT_SCAN_SUBST;
11028 pl_yylval.ival = OP_NULL;
11030 s = scan_str(start,!!PL_madskills,FALSE);
11033 Perl_croak(aTHX_ "Substitution pattern not terminated");
11035 if (s[-1] == PL_multi_open)
11038 if (PL_madskills) {
11039 CURMAD('q', PL_thisopen);
11040 CURMAD('_', PL_thiswhite);
11041 CURMAD('E', PL_thisstuff);
11042 CURMAD('Q', PL_thisclose);
11043 PL_realtokenstart = s - SvPVX(PL_linestr);
11047 first_start = PL_multi_start;
11048 s = scan_str(s,!!PL_madskills,FALSE);
11050 if (PL_lex_stuff) {
11051 SvREFCNT_dec(PL_lex_stuff);
11052 PL_lex_stuff = NULL;
11054 Perl_croak(aTHX_ "Substitution replacement not terminated");
11056 PL_multi_start = first_start; /* so whole substitution is taken together */
11058 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11061 if (PL_madskills) {
11062 CURMAD('z', PL_thisopen);
11063 CURMAD('R', PL_thisstuff);
11064 CURMAD('Z', PL_thisclose);
11070 if (*s == EXEC_PAT_MOD) {
11074 else if (strchr(S_PAT_MODS, *s))
11075 pmflag(&pm->op_pmflags,*s++);
11081 if (PL_madskills) {
11083 curmad('m', newSVpvn(modstart, s - modstart));
11084 append_madprops(PL_thismad, (OP*)pm, 0);
11088 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11089 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11093 SV * const repl = newSVpvs("");
11095 PL_sublex_info.super_bufptr = s;
11096 PL_sublex_info.super_bufend = PL_bufend;
11098 pm->op_pmflags |= PMf_EVAL;
11101 sv_catpvs(repl, "eval ");
11103 sv_catpvs(repl, "do ");
11105 sv_catpvs(repl, "{");
11106 sv_catsv(repl, PL_lex_repl);
11107 if (strchr(SvPVX(PL_lex_repl), '#'))
11108 sv_catpvs(repl, "\n");
11109 sv_catpvs(repl, "}");
11111 SvREFCNT_dec(PL_lex_repl);
11112 PL_lex_repl = repl;
11115 PL_lex_op = (OP*)pm;
11116 pl_yylval.ival = OP_SUBST;
11121 S_scan_trans(pTHX_ char *start)
11134 PERL_ARGS_ASSERT_SCAN_TRANS;
11136 pl_yylval.ival = OP_NULL;
11138 s = scan_str(start,!!PL_madskills,FALSE);
11140 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11142 if (s[-1] == PL_multi_open)
11145 if (PL_madskills) {
11146 CURMAD('q', PL_thisopen);
11147 CURMAD('_', PL_thiswhite);
11148 CURMAD('E', PL_thisstuff);
11149 CURMAD('Q', PL_thisclose);
11150 PL_realtokenstart = s - SvPVX(PL_linestr);
11154 s = scan_str(s,!!PL_madskills,FALSE);
11156 if (PL_lex_stuff) {
11157 SvREFCNT_dec(PL_lex_stuff);
11158 PL_lex_stuff = NULL;
11160 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11162 if (PL_madskills) {
11163 CURMAD('z', PL_thisopen);
11164 CURMAD('R', PL_thisstuff);
11165 CURMAD('Z', PL_thisclose);
11168 complement = del = squash = 0;
11175 complement = OPpTRANS_COMPLEMENT;
11178 del = OPpTRANS_DELETE;
11181 squash = OPpTRANS_SQUASH;
11190 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11191 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11192 o->op_private &= ~OPpTRANS_ALL;
11193 o->op_private |= del|squash|complement|
11194 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11195 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11198 pl_yylval.ival = OP_TRANS;
11201 if (PL_madskills) {
11203 curmad('m', newSVpvn(modstart, s - modstart));
11204 append_madprops(PL_thismad, o, 0);
11213 S_scan_heredoc(pTHX_ register char *s)
11217 I32 op_type = OP_SCALAR;
11221 const char *found_newline;
11225 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11227 I32 stuffstart = s - SvPVX(PL_linestr);
11230 PL_realtokenstart = -1;
11233 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11237 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11241 while (SPACE_OR_TAB(*peek))
11243 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11246 s = delimcpy(d, e, s, PL_bufend, term, &len);
11256 if (!isALNUM_lazy_if(s,UTF))
11257 deprecate_old("bare << to mean <<\"\"");
11258 for (; isALNUM_lazy_if(s,UTF); s++) {
11263 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11264 Perl_croak(aTHX_ "Delimiter for here document is too long");
11267 len = d - PL_tokenbuf;
11270 if (PL_madskills) {
11271 tstart = PL_tokenbuf + !outer;
11272 PL_thisclose = newSVpvn(tstart, len - !outer);
11273 tstart = SvPVX(PL_linestr) + stuffstart;
11274 PL_thisopen = newSVpvn(tstart, s - tstart);
11275 stuffstart = s - SvPVX(PL_linestr);
11278 #ifndef PERL_STRICT_CR
11279 d = strchr(s, '\r');
11281 char * const olds = s;
11283 while (s < PL_bufend) {
11289 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11298 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11305 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11306 herewas = newSVpvn(s,PL_bufend-s);
11310 herewas = newSVpvn(s-1,found_newline-s+1);
11313 herewas = newSVpvn(s,found_newline-s);
11317 if (PL_madskills) {
11318 tstart = SvPVX(PL_linestr) + stuffstart;
11320 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11322 PL_thisstuff = newSVpvn(tstart, s - tstart);
11325 s += SvCUR(herewas);
11328 stuffstart = s - SvPVX(PL_linestr);
11334 tmpstr = newSV_type(SVt_PVIV);
11335 SvGROW(tmpstr, 80);
11336 if (term == '\'') {
11337 op_type = OP_CONST;
11338 SvIV_set(tmpstr, -1);
11340 else if (term == '`') {
11341 op_type = OP_BACKTICK;
11342 SvIV_set(tmpstr, '\\');
11346 PL_multi_start = CopLINE(PL_curcop);
11347 PL_multi_open = PL_multi_close = '<';
11348 term = *PL_tokenbuf;
11349 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11350 char * const bufptr = PL_sublex_info.super_bufptr;
11351 char * const bufend = PL_sublex_info.super_bufend;
11352 char * const olds = s - SvCUR(herewas);
11353 s = strchr(bufptr, '\n');
11357 while (s < bufend &&
11358 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11360 CopLINE_inc(PL_curcop);
11363 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11364 missingterm(PL_tokenbuf);
11366 sv_setpvn(herewas,bufptr,d-bufptr+1);
11367 sv_setpvn(tmpstr,d+1,s-d);
11369 sv_catpvn(herewas,s,bufend-s);
11370 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11377 while (s < PL_bufend &&
11378 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11380 CopLINE_inc(PL_curcop);
11382 if (s >= PL_bufend) {
11383 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11384 missingterm(PL_tokenbuf);
11386 sv_setpvn(tmpstr,d+1,s-d);
11388 if (PL_madskills) {
11390 sv_catpvn(PL_thisstuff, d + 1, s - d);
11392 PL_thisstuff = newSVpvn(d + 1, s - d);
11393 stuffstart = s - SvPVX(PL_linestr);
11397 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11399 sv_catpvn(herewas,s,PL_bufend-s);
11400 sv_setsv(PL_linestr,herewas);
11401 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11402 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11403 PL_last_lop = PL_last_uni = NULL;
11406 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
11407 while (s >= PL_bufend) { /* multiple line string? */
11409 if (PL_madskills) {
11410 tstart = SvPVX(PL_linestr) + stuffstart;
11412 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11414 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11418 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11419 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11420 missingterm(PL_tokenbuf);
11423 stuffstart = s - SvPVX(PL_linestr);
11425 CopLINE_inc(PL_curcop);
11426 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11427 PL_last_lop = PL_last_uni = NULL;
11428 #ifndef PERL_STRICT_CR
11429 if (PL_bufend - PL_linestart >= 2) {
11430 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11431 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11433 PL_bufend[-2] = '\n';
11435 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11437 else if (PL_bufend[-1] == '\r')
11438 PL_bufend[-1] = '\n';
11440 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11441 PL_bufend[-1] = '\n';
11443 if (PERLDB_LINE && PL_curstash != PL_debstash)
11444 update_debugger_info(PL_linestr, NULL, 0);
11445 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11446 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11447 *(SvPVX(PL_linestr) + off ) = ' ';
11448 sv_catsv(PL_linestr,herewas);
11449 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11450 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11454 sv_catsv(tmpstr,PL_linestr);
11459 PL_multi_end = CopLINE(PL_curcop);
11460 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11461 SvPV_shrink_to_cur(tmpstr);
11463 SvREFCNT_dec(herewas);
11465 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11467 else if (PL_encoding)
11468 sv_recode_to_utf8(tmpstr, PL_encoding);
11470 PL_lex_stuff = tmpstr;
11471 pl_yylval.ival = op_type;
11475 /* scan_inputsymbol
11476 takes: current position in input buffer
11477 returns: new position in input buffer
11478 side-effects: pl_yylval and lex_op are set.
11483 <FH> read from filehandle
11484 <pkg::FH> read from package qualified filehandle
11485 <pkg'FH> read from package qualified filehandle
11486 <$fh> read from filehandle in $fh
11487 <*.h> filename glob
11492 S_scan_inputsymbol(pTHX_ char *start)
11495 register char *s = start; /* current position in buffer */
11498 char *d = PL_tokenbuf; /* start of temp holding space */
11499 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11501 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11503 end = strchr(s, '\n');
11506 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11508 /* die if we didn't have space for the contents of the <>,
11509 or if it didn't end, or if we see a newline
11512 if (len >= (I32)sizeof PL_tokenbuf)
11513 Perl_croak(aTHX_ "Excessively long <> operator");
11515 Perl_croak(aTHX_ "Unterminated <> operator");
11520 Remember, only scalar variables are interpreted as filehandles by
11521 this code. Anything more complex (e.g., <$fh{$num}>) will be
11522 treated as a glob() call.
11523 This code makes use of the fact that except for the $ at the front,
11524 a scalar variable and a filehandle look the same.
11526 if (*d == '$' && d[1]) d++;
11528 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11529 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11532 /* If we've tried to read what we allow filehandles to look like, and
11533 there's still text left, then it must be a glob() and not a getline.
11534 Use scan_str to pull out the stuff between the <> and treat it
11535 as nothing more than a string.
11538 if (d - PL_tokenbuf != len) {
11539 pl_yylval.ival = OP_GLOB;
11540 s = scan_str(start,!!PL_madskills,FALSE);
11542 Perl_croak(aTHX_ "Glob not terminated");
11546 bool readline_overriden = FALSE;
11549 /* we're in a filehandle read situation */
11552 /* turn <> into <ARGV> */
11554 Copy("ARGV",d,5,char);
11556 /* Check whether readline() is overriden */
11557 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11559 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11561 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11562 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11563 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11564 readline_overriden = TRUE;
11566 /* if <$fh>, create the ops to turn the variable into a
11570 /* try to find it in the pad for this block, otherwise find
11571 add symbol table ops
11573 const PADOFFSET tmp = pad_findmy(d);
11574 if (tmp != NOT_IN_PAD) {
11575 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11576 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11577 HEK * const stashname = HvNAME_HEK(stash);
11578 SV * const sym = sv_2mortal(newSVhek(stashname));
11579 sv_catpvs(sym, "::");
11580 sv_catpv(sym, d+1);
11585 OP * const o = newOP(OP_PADSV, 0);
11587 PL_lex_op = readline_overriden
11588 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11589 append_elem(OP_LIST, o,
11590 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11591 : (OP*)newUNOP(OP_READLINE, 0, o);
11600 ? (GV_ADDMULTI | GV_ADDINEVAL)
11603 PL_lex_op = readline_overriden
11604 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11605 append_elem(OP_LIST,
11606 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11607 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11608 : (OP*)newUNOP(OP_READLINE, 0,
11609 newUNOP(OP_RV2SV, 0,
11610 newGVOP(OP_GV, 0, gv)));
11612 if (!readline_overriden)
11613 PL_lex_op->op_flags |= OPf_SPECIAL;
11614 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11615 pl_yylval.ival = OP_NULL;
11618 /* If it's none of the above, it must be a literal filehandle
11619 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11621 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11622 PL_lex_op = readline_overriden
11623 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11624 append_elem(OP_LIST,
11625 newGVOP(OP_GV, 0, gv),
11626 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11627 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11628 pl_yylval.ival = OP_NULL;
11637 takes: start position in buffer
11638 keep_quoted preserve \ on the embedded delimiter(s)
11639 keep_delims preserve the delimiters around the string
11640 returns: position to continue reading from buffer
11641 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11642 updates the read buffer.
11644 This subroutine pulls a string out of the input. It is called for:
11645 q single quotes q(literal text)
11646 ' single quotes 'literal text'
11647 qq double quotes qq(interpolate $here please)
11648 " double quotes "interpolate $here please"
11649 qx backticks qx(/bin/ls -l)
11650 ` backticks `/bin/ls -l`
11651 qw quote words @EXPORT_OK = qw( func() $spam )
11652 m// regexp match m/this/
11653 s/// regexp substitute s/this/that/
11654 tr/// string transliterate tr/this/that/
11655 y/// string transliterate y/this/that/
11656 ($*@) sub prototypes sub foo ($)
11657 (stuff) sub attr parameters sub foo : attr(stuff)
11658 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11660 In most of these cases (all but <>, patterns and transliterate)
11661 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11662 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11663 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11666 It skips whitespace before the string starts, and treats the first
11667 character as the delimiter. If the delimiter is one of ([{< then
11668 the corresponding "close" character )]}> is used as the closing
11669 delimiter. It allows quoting of delimiters, and if the string has
11670 balanced delimiters ([{<>}]) it allows nesting.
11672 On success, the SV with the resulting string is put into lex_stuff or,
11673 if that is already non-NULL, into lex_repl. The second case occurs only
11674 when parsing the RHS of the special constructs s/// and tr/// (y///).
11675 For convenience, the terminating delimiter character is stuffed into
11680 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11683 SV *sv; /* scalar value: string */
11684 const char *tmps; /* temp string, used for delimiter matching */
11685 register char *s = start; /* current position in the buffer */
11686 register char term; /* terminating character */
11687 register char *to; /* current position in the sv's data */
11688 I32 brackets = 1; /* bracket nesting level */
11689 bool has_utf8 = FALSE; /* is there any utf8 content? */
11690 I32 termcode; /* terminating char. code */
11691 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11692 STRLEN termlen; /* length of terminating string */
11693 int last_off = 0; /* last position for nesting bracket */
11699 PERL_ARGS_ASSERT_SCAN_STR;
11701 /* skip space before the delimiter */
11707 if (PL_realtokenstart >= 0) {
11708 stuffstart = PL_realtokenstart;
11709 PL_realtokenstart = -1;
11712 stuffstart = start - SvPVX(PL_linestr);
11714 /* mark where we are, in case we need to report errors */
11717 /* after skipping whitespace, the next character is the terminator */
11720 termcode = termstr[0] = term;
11724 termcode = utf8_to_uvchr((U8*)s, &termlen);
11725 Copy(s, termstr, termlen, U8);
11726 if (!UTF8_IS_INVARIANT(term))
11730 /* mark where we are */
11731 PL_multi_start = CopLINE(PL_curcop);
11732 PL_multi_open = term;
11734 /* find corresponding closing delimiter */
11735 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11736 termcode = termstr[0] = term = tmps[5];
11738 PL_multi_close = term;
11740 /* create a new SV to hold the contents. 79 is the SV's initial length.
11741 What a random number. */
11742 sv = newSV_type(SVt_PVIV);
11744 SvIV_set(sv, termcode);
11745 (void)SvPOK_only(sv); /* validate pointer */
11747 /* move past delimiter and try to read a complete string */
11749 sv_catpvn(sv, s, termlen);
11752 tstart = SvPVX(PL_linestr) + stuffstart;
11753 if (!PL_thisopen && !keep_delims) {
11754 PL_thisopen = newSVpvn(tstart, s - tstart);
11755 stuffstart = s - SvPVX(PL_linestr);
11759 if (PL_encoding && !UTF) {
11763 int offset = s - SvPVX_const(PL_linestr);
11764 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11765 &offset, (char*)termstr, termlen);
11766 const char * const ns = SvPVX_const(PL_linestr) + offset;
11767 char * const svlast = SvEND(sv) - 1;
11769 for (; s < ns; s++) {
11770 if (*s == '\n' && !PL_rsfp)
11771 CopLINE_inc(PL_curcop);
11774 goto read_more_line;
11776 /* handle quoted delimiters */
11777 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11779 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11781 if ((svlast-1 - t) % 2) {
11782 if (!keep_quoted) {
11783 *(svlast-1) = term;
11785 SvCUR_set(sv, SvCUR(sv) - 1);
11790 if (PL_multi_open == PL_multi_close) {
11796 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11797 /* At here, all closes are "was quoted" one,
11798 so we don't check PL_multi_close. */
11800 if (!keep_quoted && *(t+1) == PL_multi_open)
11805 else if (*t == PL_multi_open)
11813 SvCUR_set(sv, w - SvPVX_const(sv));
11815 last_off = w - SvPVX(sv);
11816 if (--brackets <= 0)
11821 if (!keep_delims) {
11822 SvCUR_set(sv, SvCUR(sv) - 1);
11828 /* extend sv if need be */
11829 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11830 /* set 'to' to the next character in the sv's string */
11831 to = SvPVX(sv)+SvCUR(sv);
11833 /* if open delimiter is the close delimiter read unbridle */
11834 if (PL_multi_open == PL_multi_close) {
11835 for (; s < PL_bufend; s++,to++) {
11836 /* embedded newlines increment the current line number */
11837 if (*s == '\n' && !PL_rsfp)
11838 CopLINE_inc(PL_curcop);
11839 /* handle quoted delimiters */
11840 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11841 if (!keep_quoted && s[1] == term)
11843 /* any other quotes are simply copied straight through */
11847 /* terminate when run out of buffer (the for() condition), or
11848 have found the terminator */
11849 else if (*s == term) {
11852 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11855 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11861 /* if the terminator isn't the same as the start character (e.g.,
11862 matched brackets), we have to allow more in the quoting, and
11863 be prepared for nested brackets.
11866 /* read until we run out of string, or we find the terminator */
11867 for (; s < PL_bufend; s++,to++) {
11868 /* embedded newlines increment the line count */
11869 if (*s == '\n' && !PL_rsfp)
11870 CopLINE_inc(PL_curcop);
11871 /* backslashes can escape the open or closing characters */
11872 if (*s == '\\' && s+1 < PL_bufend) {
11873 if (!keep_quoted &&
11874 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11879 /* allow nested opens and closes */
11880 else if (*s == PL_multi_close && --brackets <= 0)
11882 else if (*s == PL_multi_open)
11884 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11889 /* terminate the copied string and update the sv's end-of-string */
11891 SvCUR_set(sv, to - SvPVX_const(sv));
11894 * this next chunk reads more into the buffer if we're not done yet
11898 break; /* handle case where we are done yet :-) */
11900 #ifndef PERL_STRICT_CR
11901 if (to - SvPVX_const(sv) >= 2) {
11902 if ((to[-2] == '\r' && to[-1] == '\n') ||
11903 (to[-2] == '\n' && to[-1] == '\r'))
11907 SvCUR_set(sv, to - SvPVX_const(sv));
11909 else if (to[-1] == '\r')
11912 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11917 /* if we're out of file, or a read fails, bail and reset the current
11918 line marker so we can report where the unterminated string began
11921 if (PL_madskills) {
11922 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11924 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11926 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11930 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11932 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11938 /* we read a line, so increment our line counter */
11939 CopLINE_inc(PL_curcop);
11941 /* update debugger info */
11942 if (PERLDB_LINE && PL_curstash != PL_debstash)
11943 update_debugger_info(PL_linestr, NULL, 0);
11945 /* having changed the buffer, we must update PL_bufend */
11946 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11947 PL_last_lop = PL_last_uni = NULL;
11950 /* at this point, we have successfully read the delimited string */
11952 if (!PL_encoding || UTF) {
11954 if (PL_madskills) {
11955 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11956 const int len = s - tstart;
11958 sv_catpvn(PL_thisstuff, tstart, len);
11960 PL_thisstuff = newSVpvn(tstart, len);
11961 if (!PL_thisclose && !keep_delims)
11962 PL_thisclose = newSVpvn(s,termlen);
11967 sv_catpvn(sv, s, termlen);
11972 if (PL_madskills) {
11973 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11974 const int len = s - tstart - termlen;
11976 sv_catpvn(PL_thisstuff, tstart, len);
11978 PL_thisstuff = newSVpvn(tstart, len);
11979 if (!PL_thisclose && !keep_delims)
11980 PL_thisclose = newSVpvn(s - termlen,termlen);
11984 if (has_utf8 || PL_encoding)
11987 PL_multi_end = CopLINE(PL_curcop);
11989 /* if we allocated too much space, give some back */
11990 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11991 SvLEN_set(sv, SvCUR(sv) + 1);
11992 SvPV_renew(sv, SvLEN(sv));
11995 /* decide whether this is the first or second quoted string we've read
12008 takes: pointer to position in buffer
12009 returns: pointer to new position in buffer
12010 side-effects: builds ops for the constant in pl_yylval.op
12012 Read a number in any of the formats that Perl accepts:
12014 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12015 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12018 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12020 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12023 If it reads a number without a decimal point or an exponent, it will
12024 try converting the number to an integer and see if it can do so
12025 without loss of precision.
12029 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12032 register const char *s = start; /* current position in buffer */
12033 register char *d; /* destination in temp buffer */
12034 register char *e; /* end of temp buffer */
12035 NV nv; /* number read, as a double */
12036 SV *sv = NULL; /* place to put the converted number */
12037 bool floatit; /* boolean: int or float? */
12038 const char *lastub = NULL; /* position of last underbar */
12039 static char const number_too_long[] = "Number too long";
12041 PERL_ARGS_ASSERT_SCAN_NUM;
12043 /* We use the first character to decide what type of number this is */
12047 Perl_croak(aTHX_ "panic: scan_num");
12049 /* if it starts with a 0, it could be an octal number, a decimal in
12050 0.13 disguise, or a hexadecimal number, or a binary number. */
12054 u holds the "number so far"
12055 shift the power of 2 of the base
12056 (hex == 4, octal == 3, binary == 1)
12057 overflowed was the number more than we can hold?
12059 Shift is used when we add a digit. It also serves as an "are
12060 we in octal/hex/binary?" indicator to disallow hex characters
12061 when in octal mode.
12066 bool overflowed = FALSE;
12067 bool just_zero = TRUE; /* just plain 0 or binary number? */
12068 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12069 static const char* const bases[5] =
12070 { "", "binary", "", "octal", "hexadecimal" };
12071 static const char* const Bases[5] =
12072 { "", "Binary", "", "Octal", "Hexadecimal" };
12073 static const char* const maxima[5] =
12075 "0b11111111111111111111111111111111",
12079 const char *base, *Base, *max;
12081 /* check for hex */
12086 } else if (s[1] == 'b') {
12091 /* check for a decimal in disguise */
12092 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12094 /* so it must be octal */
12101 if (ckWARN(WARN_SYNTAX))
12102 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12103 "Misplaced _ in number");
12107 base = bases[shift];
12108 Base = Bases[shift];
12109 max = maxima[shift];
12111 /* read the rest of the number */
12113 /* x is used in the overflow test,
12114 b is the digit we're adding on. */
12119 /* if we don't mention it, we're done */
12123 /* _ are ignored -- but warned about if consecutive */
12125 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12126 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12127 "Misplaced _ in number");
12131 /* 8 and 9 are not octal */
12132 case '8': case '9':
12134 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12138 case '2': case '3': case '4':
12139 case '5': case '6': case '7':
12141 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12144 case '0': case '1':
12145 b = *s++ & 15; /* ASCII digit -> value of digit */
12149 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12150 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12151 /* make sure they said 0x */
12154 b = (*s++ & 7) + 9;
12156 /* Prepare to put the digit we have onto the end
12157 of the number so far. We check for overflows.
12163 x = u << shift; /* make room for the digit */
12165 if ((x >> shift) != u
12166 && !(PL_hints & HINT_NEW_BINARY)) {
12169 if (ckWARN_d(WARN_OVERFLOW))
12170 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12171 "Integer overflow in %s number",
12174 u = x | b; /* add the digit to the end */
12177 n *= nvshift[shift];
12178 /* If an NV has not enough bits in its
12179 * mantissa to represent an UV this summing of
12180 * small low-order numbers is a waste of time
12181 * (because the NV cannot preserve the
12182 * low-order bits anyway): we could just
12183 * remember when did we overflow and in the
12184 * end just multiply n by the right
12192 /* if we get here, we had success: make a scalar value from
12197 /* final misplaced underbar check */
12198 if (s[-1] == '_') {
12199 if (ckWARN(WARN_SYNTAX))
12200 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12205 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12206 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12207 "%s number > %s non-portable",
12213 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12214 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12215 "%s number > %s non-portable",
12220 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12221 sv = new_constant(start, s - start, "integer",
12222 sv, NULL, NULL, 0);
12223 else if (PL_hints & HINT_NEW_BINARY)
12224 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12229 handle decimal numbers.
12230 we're also sent here when we read a 0 as the first digit
12232 case '1': case '2': case '3': case '4': case '5':
12233 case '6': case '7': case '8': case '9': case '.':
12236 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12239 /* read next group of digits and _ and copy into d */
12240 while (isDIGIT(*s) || *s == '_') {
12241 /* skip underscores, checking for misplaced ones
12245 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12246 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12247 "Misplaced _ in number");
12251 /* check for end of fixed-length buffer */
12253 Perl_croak(aTHX_ number_too_long);
12254 /* if we're ok, copy the character */
12259 /* final misplaced underbar check */
12260 if (lastub && s == lastub + 1) {
12261 if (ckWARN(WARN_SYNTAX))
12262 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12265 /* read a decimal portion if there is one. avoid
12266 3..5 being interpreted as the number 3. followed
12269 if (*s == '.' && s[1] != '.') {
12274 if (ckWARN(WARN_SYNTAX))
12275 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12276 "Misplaced _ in number");
12280 /* copy, ignoring underbars, until we run out of digits.
12282 for (; isDIGIT(*s) || *s == '_'; s++) {
12283 /* fixed length buffer check */
12285 Perl_croak(aTHX_ number_too_long);
12287 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12288 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12289 "Misplaced _ in number");
12295 /* fractional part ending in underbar? */
12296 if (s[-1] == '_') {
12297 if (ckWARN(WARN_SYNTAX))
12298 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12299 "Misplaced _ in number");
12301 if (*s == '.' && isDIGIT(s[1])) {
12302 /* oops, it's really a v-string, but without the "v" */
12308 /* read exponent part, if present */
12309 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12313 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12314 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12316 /* stray preinitial _ */
12318 if (ckWARN(WARN_SYNTAX))
12319 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12320 "Misplaced _ in number");
12324 /* allow positive or negative exponent */
12325 if (*s == '+' || *s == '-')
12328 /* stray initial _ */
12330 if (ckWARN(WARN_SYNTAX))
12331 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12332 "Misplaced _ in number");
12336 /* read digits of exponent */
12337 while (isDIGIT(*s) || *s == '_') {
12340 Perl_croak(aTHX_ number_too_long);
12344 if (((lastub && s == lastub + 1) ||
12345 (!isDIGIT(s[1]) && s[1] != '_'))
12346 && ckWARN(WARN_SYNTAX))
12347 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12348 "Misplaced _ in number");
12355 /* make an sv from the string */
12359 We try to do an integer conversion first if no characters
12360 indicating "float" have been found.
12365 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12367 if (flags == IS_NUMBER_IN_UV) {
12369 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12372 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12373 if (uv <= (UV) IV_MIN)
12374 sv_setiv(sv, -(IV)uv);
12381 /* terminate the string */
12383 nv = Atof(PL_tokenbuf);
12388 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12389 const char *const key = floatit ? "float" : "integer";
12390 const STRLEN keylen = floatit ? 5 : 7;
12391 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12392 key, keylen, sv, NULL, NULL, 0);
12396 /* if it starts with a v, it could be a v-string */
12399 sv = newSV(5); /* preallocate storage space */
12400 s = scan_vstring(s, PL_bufend, sv);
12404 /* make the op for the constant and return */
12407 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12409 lvalp->opval = NULL;
12415 S_scan_formline(pTHX_ register char *s)
12418 register char *eol;
12420 SV * const stuff = newSVpvs("");
12421 bool needargs = FALSE;
12422 bool eofmt = FALSE;
12424 char *tokenstart = s;
12425 SV* savewhite = NULL;
12427 if (PL_madskills) {
12428 savewhite = PL_thiswhite;
12433 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12435 while (!needargs) {
12438 #ifdef PERL_STRICT_CR
12439 while (SPACE_OR_TAB(*t))
12442 while (SPACE_OR_TAB(*t) || *t == '\r')
12445 if (*t == '\n' || t == PL_bufend) {
12450 if (PL_in_eval && !PL_rsfp) {
12451 eol = (char *) memchr(s,'\n',PL_bufend-s);
12456 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12458 for (t = s; t < eol; t++) {
12459 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12461 goto enough; /* ~~ must be first line in formline */
12463 if (*t == '@' || *t == '^')
12467 sv_catpvn(stuff, s, eol-s);
12468 #ifndef PERL_STRICT_CR
12469 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12470 char *end = SvPVX(stuff) + SvCUR(stuff);
12473 SvCUR_set(stuff, SvCUR(stuff) - 1);
12483 if (PL_madskills) {
12485 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12487 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12490 s = filter_gets(PL_linestr, PL_rsfp, 0);
12492 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12494 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12496 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12497 PL_last_lop = PL_last_uni = NULL;
12506 if (SvCUR(stuff)) {
12509 PL_lex_state = LEX_NORMAL;
12510 start_force(PL_curforce);
12511 NEXTVAL_NEXTTOKE.ival = 0;
12515 PL_lex_state = LEX_FORMLINE;
12517 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12519 else if (PL_encoding)
12520 sv_recode_to_utf8(stuff, PL_encoding);
12522 start_force(PL_curforce);
12523 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12525 start_force(PL_curforce);
12526 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12530 SvREFCNT_dec(stuff);
12532 PL_lex_formbrack = 0;
12536 if (PL_madskills) {
12538 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12540 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12541 PL_thiswhite = savewhite;
12548 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12551 const I32 oldsavestack_ix = PL_savestack_ix;
12552 CV* const outsidecv = PL_compcv;
12555 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12557 SAVEI32(PL_subline);
12558 save_item(PL_subname);
12559 SAVESPTR(PL_compcv);
12561 PL_compcv = (CV*)newSV_type(is_format ? SVt_PVFM : SVt_PVCV);
12562 CvFLAGS(PL_compcv) |= flags;
12564 PL_subline = CopLINE(PL_curcop);
12565 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12566 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc_simple(outsidecv);
12567 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12569 return oldsavestack_ix;
12573 #pragma segment Perl_yylex
12576 Perl_yywarn(pTHX_ const char *const s)
12580 PERL_ARGS_ASSERT_YYWARN;
12582 PL_in_eval |= EVAL_WARNONLY;
12584 PL_in_eval &= ~EVAL_WARNONLY;
12589 Perl_yyerror(pTHX_ const char *const s)
12592 const char *where = NULL;
12593 const char *context = NULL;
12596 int yychar = PL_parser->yychar;
12598 PERL_ARGS_ASSERT_YYERROR;
12600 if (!yychar || (yychar == ';' && !PL_rsfp))
12602 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12603 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12604 PL_oldbufptr != PL_bufptr) {
12607 The code below is removed for NetWare because it abends/crashes on NetWare
12608 when the script has error such as not having the closing quotes like:
12609 if ($var eq "value)
12610 Checking of white spaces is anyway done in NetWare code.
12613 while (isSPACE(*PL_oldoldbufptr))
12616 context = PL_oldoldbufptr;
12617 contlen = PL_bufptr - PL_oldoldbufptr;
12619 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12620 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12623 The code below is removed for NetWare because it abends/crashes on NetWare
12624 when the script has error such as not having the closing quotes like:
12625 if ($var eq "value)
12626 Checking of white spaces is anyway done in NetWare code.
12629 while (isSPACE(*PL_oldbufptr))
12632 context = PL_oldbufptr;
12633 contlen = PL_bufptr - PL_oldbufptr;
12635 else if (yychar > 255)
12636 where = "next token ???";
12637 else if (yychar == -2) { /* YYEMPTY */
12638 if (PL_lex_state == LEX_NORMAL ||
12639 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12640 where = "at end of line";
12641 else if (PL_lex_inpat)
12642 where = "within pattern";
12644 where = "within string";
12647 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12649 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12650 else if (isPRINT_LC(yychar)) {
12651 const char string = yychar;
12652 sv_catpvn(where_sv, &string, 1);
12655 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12656 where = SvPVX_const(where_sv);
12658 msg = sv_2mortal(newSVpv(s, 0));
12659 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12660 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12662 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12664 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12665 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12666 Perl_sv_catpvf(aTHX_ msg,
12667 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12668 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12671 if (PL_in_eval & EVAL_WARNONLY) {
12672 if (ckWARN_d(WARN_SYNTAX))
12673 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12677 if (PL_error_count >= 10) {
12678 if (PL_in_eval && SvCUR(ERRSV))
12679 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12680 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12682 Perl_croak(aTHX_ "%s has too many errors.\n",
12683 OutCopFILE(PL_curcop));
12686 PL_in_my_stash = NULL;
12690 #pragma segment Main
12694 S_swallow_bom(pTHX_ U8 *s)
12697 const STRLEN slen = SvCUR(PL_linestr);
12699 PERL_ARGS_ASSERT_SWALLOW_BOM;
12703 if (s[1] == 0xFE) {
12704 /* UTF-16 little-endian? (or UTF32-LE?) */
12705 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12706 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12707 #ifndef PERL_NO_UTF16_FILTER
12708 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12711 if (PL_bufend > (char*)s) {
12715 filter_add(utf16rev_textfilter, NULL);
12716 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12717 utf16_to_utf8_reversed(s, news,
12718 PL_bufend - (char*)s - 1,
12720 sv_setpvn(PL_linestr, (const char*)news, newlen);
12722 s = (U8*)SvPVX(PL_linestr);
12723 Copy(news, s, newlen, U8);
12727 SvUTF8_on(PL_linestr);
12728 s = (U8*)SvPVX(PL_linestr);
12730 /* FIXME - is this a general bug fix? */
12733 PL_bufend = SvPVX(PL_linestr) + newlen;
12736 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12741 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12742 #ifndef PERL_NO_UTF16_FILTER
12743 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12746 if (PL_bufend > (char *)s) {
12750 filter_add(utf16_textfilter, NULL);
12751 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12752 utf16_to_utf8(s, news,
12753 PL_bufend - (char*)s,
12755 sv_setpvn(PL_linestr, (const char*)news, newlen);
12757 SvUTF8_on(PL_linestr);
12758 s = (U8*)SvPVX(PL_linestr);
12759 PL_bufend = SvPVX(PL_linestr) + newlen;
12762 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12767 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12768 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12769 s += 3; /* UTF-8 */
12775 if (s[2] == 0xFE && s[3] == 0xFF) {
12776 /* UTF-32 big-endian */
12777 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12780 else if (s[2] == 0 && s[3] != 0) {
12783 * are a good indicator of UTF-16BE. */
12784 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12790 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12791 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12792 s += 4; /* UTF-8 */
12798 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12801 * are a good indicator of UTF-16LE. */
12802 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12810 #ifndef PERL_NO_UTF16_FILTER
12812 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12815 const STRLEN old = SvCUR(sv);
12816 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12817 DEBUG_P(PerlIO_printf(Perl_debug_log,
12818 "utf16_textfilter(%p): %d %d (%d)\n",
12819 FPTR2DPTR(void *, utf16_textfilter),
12820 idx, maxlen, (int) count));
12824 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12825 Copy(SvPVX_const(sv), tmps, old, char);
12826 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12827 SvCUR(sv) - old, &newlen);
12828 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12830 DEBUG_P({sv_dump(sv);});
12835 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12838 const STRLEN old = SvCUR(sv);
12839 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12840 DEBUG_P(PerlIO_printf(Perl_debug_log,
12841 "utf16rev_textfilter(%p): %d %d (%d)\n",
12842 FPTR2DPTR(void *, utf16rev_textfilter),
12843 idx, maxlen, (int) count));
12847 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12848 Copy(SvPVX_const(sv), tmps, old, char);
12849 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12850 SvCUR(sv) - old, &newlen);
12851 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12853 DEBUG_P({ sv_dump(sv); });
12859 Returns a pointer to the next character after the parsed
12860 vstring, as well as updating the passed in sv.
12862 Function must be called like
12865 s = scan_vstring(s,e,sv);
12867 where s and e are the start and end of the string.
12868 The sv should already be large enough to store the vstring
12869 passed in, for performance reasons.
12874 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12877 const char *pos = s;
12878 const char *start = s;
12880 PERL_ARGS_ASSERT_SCAN_VSTRING;
12882 if (*pos == 'v') pos++; /* get past 'v' */
12883 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12885 if ( *pos != '.') {
12886 /* this may not be a v-string if followed by => */
12887 const char *next = pos;
12888 while (next < e && isSPACE(*next))
12890 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12891 /* return string not v-string */
12892 sv_setpvn(sv,(char *)s,pos-s);
12893 return (char *)pos;
12897 if (!isALPHA(*pos)) {
12898 U8 tmpbuf[UTF8_MAXBYTES+1];
12901 s++; /* get past 'v' */
12903 sv_setpvn(sv, "", 0);
12906 /* this is atoi() that tolerates underscores */
12909 const char *end = pos;
12911 while (--end >= s) {
12913 const UV orev = rev;
12914 rev += (*end - '0') * mult;
12916 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12917 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12918 "Integer overflow in decimal number");
12922 if (rev > 0x7FFFFFFF)
12923 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12925 /* Append native character for the rev point */
12926 tmpend = uvchr_to_utf8(tmpbuf, rev);
12927 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12928 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12930 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12936 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12940 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12948 * c-indentation-style: bsd
12949 * c-basic-offset: 4
12950 * indent-tabs-mode: t
12953 * ex: set ts=8 sts=4 sw=4 noet: