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 { 0, TOKENTYPE_NONE, NULL }
374 /* dump the returned token in rv, plus any optional arg in pl_yylval */
377 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
381 PERL_ARGS_ASSERT_TOKEREPORT;
384 const char *name = NULL;
385 enum token_type type = TOKENTYPE_NONE;
386 const struct debug_tokens *p;
387 SV* const report = newSVpvs("<== ");
389 for (p = debug_tokens; p->token; p++) {
390 if (p->token == (int)rv) {
397 Perl_sv_catpv(aTHX_ report, name);
398 else if ((char)rv > ' ' && (char)rv < '~')
399 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
401 sv_catpvs(report, "EOF");
403 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
406 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
409 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
411 case TOKENTYPE_OPNUM:
412 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
413 PL_op_name[lvalp->ival]);
416 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
418 case TOKENTYPE_OPVAL:
420 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
421 PL_op_name[lvalp->opval->op_type]);
422 if (lvalp->opval->op_type == OP_CONST) {
423 Perl_sv_catpvf(aTHX_ report, " %s",
424 SvPEEK(cSVOPx_sv(lvalp->opval)));
429 sv_catpvs(report, "(opval=null)");
432 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
438 /* print the buffer with suitable escapes */
441 S_printbuf(pTHX_ const char *const fmt, const char *const s)
443 SV* const tmp = newSVpvs("");
445 PERL_ARGS_ASSERT_PRINTBUF;
447 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
456 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
457 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
461 S_ao(pTHX_ int toketype)
464 if (*PL_bufptr == '=') {
466 if (toketype == ANDAND)
467 pl_yylval.ival = OP_ANDASSIGN;
468 else if (toketype == OROR)
469 pl_yylval.ival = OP_ORASSIGN;
470 else if (toketype == DORDOR)
471 pl_yylval.ival = OP_DORASSIGN;
479 * When Perl expects an operator and finds something else, no_op
480 * prints the warning. It always prints "<something> found where
481 * operator expected. It prints "Missing semicolon on previous line?"
482 * if the surprise occurs at the start of the line. "do you need to
483 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
484 * where the compiler doesn't know if foo is a method call or a function.
485 * It prints "Missing operator before end of line" if there's nothing
486 * after the missing operator, or "... before <...>" if there is something
487 * after the missing operator.
491 S_no_op(pTHX_ const char *const what, char *s)
494 char * const oldbp = PL_bufptr;
495 const bool is_first = (PL_oldbufptr == PL_linestart);
497 PERL_ARGS_ASSERT_NO_OP;
503 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
504 if (ckWARN_d(WARN_SYNTAX)) {
506 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
507 "\t(Missing semicolon on previous line?)\n");
508 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
510 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
512 if (t < PL_bufptr && isSPACE(*t))
513 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
514 "\t(Do you need to predeclare %.*s?)\n",
515 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
519 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
520 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
528 * Complain about missing quote/regexp/heredoc terminator.
529 * If it's called with NULL then it cauterizes the line buffer.
530 * If we're in a delimited string and the delimiter is a control
531 * character, it's reformatted into a two-char sequence like ^C.
536 S_missingterm(pTHX_ char *s)
542 char * const nl = strrchr(s,'\n');
548 iscntrl(PL_multi_close)
550 PL_multi_close < 32 || PL_multi_close == 127
554 tmpbuf[1] = (char)toCTRL(PL_multi_close);
559 *tmpbuf = (char)PL_multi_close;
563 q = strchr(s,'"') ? '\'' : '"';
564 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
567 #define FEATURE_IS_ENABLED(name) \
568 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
569 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
570 /* The longest string we pass in. */
571 #define MAX_FEATURE_LEN (sizeof("switch")-1)
574 * S_feature_is_enabled
575 * Check whether the named feature is enabled.
578 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
581 HV * const hinthv = GvHV(PL_hintgv);
582 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
584 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
586 assert(namelen <= MAX_FEATURE_LEN);
587 memcpy(&he_name[8], name, namelen);
589 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
597 Perl_deprecate(pTHX_ const char *const s)
599 PERL_ARGS_ASSERT_DEPRECATE;
601 if (ckWARN(WARN_DEPRECATED))
602 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
606 Perl_deprecate_old(pTHX_ const char *const s)
608 /* This function should NOT be called for any new deprecated warnings */
609 /* Use Perl_deprecate instead */
611 /* It is here to maintain backward compatibility with the pre-5.8 */
612 /* warnings category hierarchy. The "deprecated" category used to */
613 /* live under the "syntax" category. It is now a top-level category */
614 /* in its own right. */
616 PERL_ARGS_ASSERT_DEPRECATE_OLD;
618 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
619 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
620 "Use of %s is deprecated", s);
624 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
625 * utf16-to-utf8-reversed.
628 #ifdef PERL_CR_FILTER
632 register const char *s = SvPVX_const(sv);
633 register const char * const e = s + SvCUR(sv);
635 PERL_ARGS_ASSERT_STRIP_RETURN;
637 /* outer loop optimized to do nothing if there are no CR-LFs */
639 if (*s++ == '\r' && *s == '\n') {
640 /* hit a CR-LF, need to copy the rest */
641 register char *d = s - 1;
644 if (*s == '\r' && s[1] == '\n')
655 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
657 const I32 count = FILTER_READ(idx+1, sv, maxlen);
658 if (count > 0 && !maxlen)
669 * Create a parser object and initialise its parser and lexer fields
671 * rsfp is the opened file handle to read from (if any),
673 * line holds any initial content already read from the file (or in
674 * the case of no file, such as an eval, the whole contents);
676 * new_filter indicates that this is a new file and it shouldn't inherit
677 * the filters from the current parser (ie require).
681 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
684 const char *s = NULL;
686 yy_parser *parser, *oparser;
688 /* create and initialise a parser */
690 Newxz(parser, 1, yy_parser);
691 parser->old_parser = oparser = PL_parser;
694 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
695 parser->ps = parser->stack;
696 parser->stack_size = YYINITDEPTH;
698 parser->stack->state = 0;
699 parser->yyerrstatus = 0;
700 parser->yychar = YYEMPTY; /* Cause a token to be read. */
702 /* on scope exit, free this parser and restore any outer one */
704 parser->saved_curcop = PL_curcop;
706 /* initialise lexer state */
709 parser->curforce = -1;
711 parser->nexttoke = 0;
713 parser->error_count = oparser ? oparser->error_count : 0;
714 parser->copline = NOLINE;
715 parser->lex_state = LEX_NORMAL;
716 parser->expect = XSTATE;
718 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
719 : (AV*)SvREFCNT_inc(oparser->rsfp_filters);
721 Newx(parser->lex_brackstack, 120, char);
722 Newx(parser->lex_casestack, 12, char);
723 *parser->lex_casestack = '\0';
726 s = SvPV_const(line, len);
732 parser->linestr = newSVpvs("\n;");
733 } else if (SvREADONLY(line) || s[len-1] != ';') {
734 parser->linestr = newSVsv(line);
736 sv_catpvs(parser->linestr, "\n;");
739 SvREFCNT_inc_simple_void_NN(line);
740 parser->linestr = line;
742 parser->oldoldbufptr =
745 parser->linestart = SvPVX(parser->linestr);
746 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
747 parser->last_lop = parser->last_uni = NULL;
751 /* delete a parser object */
754 Perl_parser_free(pTHX_ const yy_parser *parser)
756 PERL_ARGS_ASSERT_PARSER_FREE;
758 PL_curcop = parser->saved_curcop;
759 SvREFCNT_dec(parser->linestr);
761 if (parser->rsfp == PerlIO_stdin())
762 PerlIO_clearerr(parser->rsfp);
763 else if (parser->rsfp && parser->old_parser
764 && parser->rsfp != parser->old_parser->rsfp)
765 PerlIO_close(parser->rsfp);
766 SvREFCNT_dec(parser->rsfp_filters);
768 Safefree(parser->stack);
769 Safefree(parser->lex_brackstack);
770 Safefree(parser->lex_casestack);
771 PL_parser = parser->old_parser;
778 * Finalizer for lexing operations. Must be called when the parser is
779 * done with the lexer.
786 PL_doextract = FALSE;
791 * This subroutine has nothing to do with tilting, whether at windmills
792 * or pinball tables. Its name is short for "increment line". It
793 * increments the current line number in CopLINE(PL_curcop) and checks
794 * to see whether the line starts with a comment of the form
795 * # line 500 "foo.pm"
796 * If so, it sets the current line number and file to the values in the comment.
800 S_incline(pTHX_ const char *s)
807 PERL_ARGS_ASSERT_INCLINE;
809 CopLINE_inc(PL_curcop);
812 while (SPACE_OR_TAB(*s))
814 if (strnEQ(s, "line", 4))
818 if (SPACE_OR_TAB(*s))
822 while (SPACE_OR_TAB(*s))
830 while (SPACE_OR_TAB(*s))
832 if (*s == '"' && (t = strchr(s+1, '"'))) {
842 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
844 if (*e != '\n' && *e != '\0')
845 return; /* false alarm */
848 const STRLEN len = t - s;
850 SV *const temp_sv = CopFILESV(PL_curcop);
856 tmplen = SvCUR(temp_sv);
862 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
863 /* must copy *{"::_<(eval N)[oldfilename:L]"}
864 * to *{"::_<newfilename"} */
865 /* However, the long form of evals is only turned on by the
866 debugger - usually they're "(eval %lu)" */
870 STRLEN tmplen2 = len;
871 if (tmplen + 2 <= sizeof smallbuf)
874 Newx(tmpbuf, tmplen + 2, char);
877 memcpy(tmpbuf + 2, cf, tmplen);
879 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
884 if (tmplen2 + 2 <= sizeof smallbuf)
887 Newx(tmpbuf2, tmplen2 + 2, char);
889 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
890 /* Either they malloc'd it, or we malloc'd it,
891 so no prefix is present in ours. */
896 memcpy(tmpbuf2 + 2, s, tmplen2);
899 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
901 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
902 /* adjust ${"::_<newfilename"} to store the new file name */
903 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
904 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
905 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
908 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
910 if (tmpbuf != smallbuf) Safefree(tmpbuf);
913 CopFILE_free(PL_curcop);
914 CopFILE_setn(PL_curcop, s, len);
916 CopLINE_set(PL_curcop, atoi(n)-1);
920 /* skip space before PL_thistoken */
923 S_skipspace0(pTHX_ register char *s)
925 PERL_ARGS_ASSERT_SKIPSPACE0;
932 PL_thiswhite = newSVpvs("");
933 sv_catsv(PL_thiswhite, PL_skipwhite);
934 sv_free(PL_skipwhite);
937 PL_realtokenstart = s - SvPVX(PL_linestr);
941 /* skip space after PL_thistoken */
944 S_skipspace1(pTHX_ register char *s)
946 const char *start = s;
947 I32 startoff = start - SvPVX(PL_linestr);
949 PERL_ARGS_ASSERT_SKIPSPACE1;
954 start = SvPVX(PL_linestr) + startoff;
955 if (!PL_thistoken && PL_realtokenstart >= 0) {
956 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
957 PL_thistoken = newSVpvn(tstart, start - tstart);
959 PL_realtokenstart = -1;
962 PL_nextwhite = newSVpvs("");
963 sv_catsv(PL_nextwhite, PL_skipwhite);
964 sv_free(PL_skipwhite);
971 S_skipspace2(pTHX_ register char *s, SV **svp)
974 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
975 const I32 startoff = s - SvPVX(PL_linestr);
977 PERL_ARGS_ASSERT_SKIPSPACE2;
980 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
981 if (!PL_madskills || !svp)
983 start = SvPVX(PL_linestr) + startoff;
984 if (!PL_thistoken && PL_realtokenstart >= 0) {
985 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
986 PL_thistoken = newSVpvn(tstart, start - tstart);
987 PL_realtokenstart = -1;
992 sv_setsv(*svp, PL_skipwhite);
993 sv_free(PL_skipwhite);
1002 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1004 AV *av = CopFILEAVx(PL_curcop);
1006 SV * const sv = newSV_type(SVt_PVMG);
1008 sv_setsv(sv, orig_sv);
1010 sv_setpvn(sv, buf, len);
1013 av_store(av, (I32)CopLINE(PL_curcop), sv);
1019 * Called to gobble the appropriate amount and type of whitespace.
1020 * Skips comments as well.
1024 S_skipspace(pTHX_ register char *s)
1029 int startoff = s - SvPVX(PL_linestr);
1031 PERL_ARGS_ASSERT_SKIPSPACE;
1034 sv_free(PL_skipwhite);
1038 PERL_ARGS_ASSERT_SKIPSPACE;
1040 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1041 while (s < PL_bufend && SPACE_OR_TAB(*s))
1051 SSize_t oldprevlen, oldoldprevlen;
1052 SSize_t oldloplen = 0, oldunilen = 0;
1053 while (s < PL_bufend && isSPACE(*s)) {
1054 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1059 if (s < PL_bufend && *s == '#') {
1060 while (s < PL_bufend && *s != '\n')
1062 if (s < PL_bufend) {
1064 if (PL_in_eval && !PL_rsfp) {
1071 /* only continue to recharge the buffer if we're at the end
1072 * of the buffer, we're not reading from a source filter, and
1073 * we're in normal lexing mode
1075 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1076 PL_lex_state == LEX_FORMLINE)
1083 /* try to recharge the buffer */
1085 curoff = s - SvPVX(PL_linestr);
1088 if ((s = filter_gets(PL_linestr, PL_rsfp,
1089 (prevlen = SvCUR(PL_linestr)))) == NULL)
1092 if (PL_madskills && curoff != startoff) {
1094 PL_skipwhite = newSVpvs("");
1095 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1099 /* mustn't throw out old stuff yet if madpropping */
1100 SvCUR(PL_linestr) = curoff;
1101 s = SvPVX(PL_linestr) + curoff;
1103 if (curoff && s[-1] == '\n')
1107 /* end of file. Add on the -p or -n magic */
1108 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1111 sv_catpvs(PL_linestr,
1112 ";}continue{print or die qq(-p destination: $!\\n);}");
1114 sv_setpvs(PL_linestr,
1115 ";}continue{print or die qq(-p destination: $!\\n);}");
1117 PL_minus_n = PL_minus_p = 0;
1119 else if (PL_minus_n) {
1121 sv_catpvn(PL_linestr, ";}", 2);
1123 sv_setpvn(PL_linestr, ";}", 2);
1129 sv_catpvn(PL_linestr,";", 1);
1131 sv_setpvn(PL_linestr,";", 1);
1134 /* reset variables for next time we lex */
1135 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1141 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1142 PL_last_lop = PL_last_uni = NULL;
1144 /* Close the filehandle. Could be from
1145 * STDIN, or a regular file. If we were reading code from
1146 * STDIN (because the commandline held no -e or filename)
1147 * then we don't close it, we reset it so the code can
1148 * read from STDIN too.
1151 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1152 PerlIO_clearerr(PL_rsfp);
1154 (void)PerlIO_close(PL_rsfp);
1159 /* not at end of file, so we only read another line */
1160 /* make corresponding updates to old pointers, for yyerror() */
1161 oldprevlen = PL_oldbufptr - PL_bufend;
1162 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1164 oldunilen = PL_last_uni - PL_bufend;
1166 oldloplen = PL_last_lop - PL_bufend;
1167 PL_linestart = PL_bufptr = s + prevlen;
1168 PL_bufend = s + SvCUR(PL_linestr);
1170 PL_oldbufptr = s + oldprevlen;
1171 PL_oldoldbufptr = s + oldoldprevlen;
1173 PL_last_uni = s + oldunilen;
1175 PL_last_lop = s + oldloplen;
1178 /* debugger active and we're not compiling the debugger code,
1179 * so store the line into the debugger's array of lines
1181 if (PERLDB_LINE && PL_curstash != PL_debstash)
1182 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1189 PL_skipwhite = newSVpvs("");
1190 curoff = s - SvPVX(PL_linestr);
1191 if (curoff - startoff)
1192 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1201 * Check the unary operators to ensure there's no ambiguity in how they're
1202 * used. An ambiguous piece of code would be:
1204 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1205 * the +5 is its argument.
1215 if (PL_oldoldbufptr != PL_last_uni)
1217 while (isSPACE(*PL_last_uni))
1220 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1222 if ((t = strchr(s, '(')) && t < PL_bufptr)
1225 if (ckWARN_d(WARN_AMBIGUOUS)){
1226 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1227 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1228 (int)(s - PL_last_uni), PL_last_uni);
1233 * LOP : macro to build a list operator. Its behaviour has been replaced
1234 * with a subroutine, S_lop() for which LOP is just another name.
1237 #define LOP(f,x) return lop(f,x,s)
1241 * Build a list operator (or something that might be one). The rules:
1242 * - if we have a next token, then it's a list operator [why?]
1243 * - if the next thing is an opening paren, then it's a function
1244 * - else it's a list operator
1248 S_lop(pTHX_ I32 f, int x, char *s)
1252 PERL_ARGS_ASSERT_LOP;
1258 PL_last_lop = PL_oldbufptr;
1259 PL_last_lop_op = (OPCODE)f;
1262 return REPORT(LSTOP);
1265 return REPORT(LSTOP);
1268 return REPORT(FUNC);
1271 return REPORT(FUNC);
1273 return REPORT(LSTOP);
1279 * Sets up for an eventual force_next(). start_force(0) basically does
1280 * an unshift, while start_force(-1) does a push. yylex removes items
1285 S_start_force(pTHX_ int where)
1289 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1290 where = PL_lasttoke;
1291 assert(PL_curforce < 0 || PL_curforce == where);
1292 if (PL_curforce != where) {
1293 for (i = PL_lasttoke; i > where; --i) {
1294 PL_nexttoke[i] = PL_nexttoke[i-1];
1298 if (PL_curforce < 0) /* in case of duplicate start_force() */
1299 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1300 PL_curforce = where;
1303 curmad('^', newSVpvs(""));
1304 CURMAD('_', PL_nextwhite);
1309 S_curmad(pTHX_ char slot, SV *sv)
1315 if (PL_curforce < 0)
1316 where = &PL_thismad;
1318 where = &PL_nexttoke[PL_curforce].next_mad;
1321 sv_setpvn(sv, "", 0);
1324 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1326 else if (PL_encoding) {
1327 sv_recode_to_utf8(sv, PL_encoding);
1332 /* keep a slot open for the head of the list? */
1333 if (slot != '_' && *where && (*where)->mad_key == '^') {
1334 (*where)->mad_key = slot;
1335 sv_free((SV*)((*where)->mad_val));
1336 (*where)->mad_val = (void*)sv;
1339 addmad(newMADsv(slot, sv), where, 0);
1342 # define start_force(where) NOOP
1343 # define curmad(slot, sv) NOOP
1348 * When the lexer realizes it knows the next token (for instance,
1349 * it is reordering tokens for the parser) then it can call S_force_next
1350 * to know what token to return the next time the lexer is called. Caller
1351 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1352 * and possibly PL_expect to ensure the lexer handles the token correctly.
1356 S_force_next(pTHX_ I32 type)
1361 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1362 tokereport(THING, &NEXTVAL_NEXTTOKE);
1366 if (PL_curforce < 0)
1367 start_force(PL_lasttoke);
1368 PL_nexttoke[PL_curforce].next_type = type;
1369 if (PL_lex_state != LEX_KNOWNEXT)
1370 PL_lex_defer = PL_lex_state;
1371 PL_lex_state = LEX_KNOWNEXT;
1372 PL_lex_expect = PL_expect;
1375 PL_nexttype[PL_nexttoke] = type;
1377 if (PL_lex_state != LEX_KNOWNEXT) {
1378 PL_lex_defer = PL_lex_state;
1379 PL_lex_expect = PL_expect;
1380 PL_lex_state = LEX_KNOWNEXT;
1386 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1389 SV * const sv = newSVpvn_utf8(start, len,
1391 && is_utf8_string((const U8*)start, len));
1397 * When the lexer knows the next thing is a word (for instance, it has
1398 * just seen -> and it knows that the next char is a word char, then
1399 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1403 * char *start : buffer position (must be within PL_linestr)
1404 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1405 * int check_keyword : if true, Perl checks to make sure the word isn't
1406 * a keyword (do this if the word is a label, e.g. goto FOO)
1407 * int allow_pack : if true, : characters will also be allowed (require,
1408 * use, etc. do this)
1409 * int allow_initial_tick : used by the "sub" lexer only.
1413 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1419 PERL_ARGS_ASSERT_FORCE_WORD;
1421 start = SKIPSPACE1(start);
1423 if (isIDFIRST_lazy_if(s,UTF) ||
1424 (allow_pack && *s == ':') ||
1425 (allow_initial_tick && *s == '\'') )
1427 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1428 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1430 start_force(PL_curforce);
1432 curmad('X', newSVpvn(start,s-start));
1433 if (token == METHOD) {
1438 PL_expect = XOPERATOR;
1442 curmad('g', newSVpvs( "forced" ));
1443 NEXTVAL_NEXTTOKE.opval
1444 = (OP*)newSVOP(OP_CONST,0,
1445 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1446 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1454 * Called when the lexer wants $foo *foo &foo etc, but the program
1455 * text only contains the "foo" portion. The first argument is a pointer
1456 * to the "foo", and the second argument is the type symbol to prefix.
1457 * Forces the next token to be a "WORD".
1458 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1462 S_force_ident(pTHX_ register const char *s, int kind)
1466 PERL_ARGS_ASSERT_FORCE_IDENT;
1469 const STRLEN len = strlen(s);
1470 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1471 start_force(PL_curforce);
1472 NEXTVAL_NEXTTOKE.opval = o;
1475 o->op_private = OPpCONST_ENTERED;
1476 /* XXX see note in pp_entereval() for why we forgo typo
1477 warnings if the symbol must be introduced in an eval.
1479 gv_fetchpvn_flags(s, len,
1480 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1482 kind == '$' ? SVt_PV :
1483 kind == '@' ? SVt_PVAV :
1484 kind == '%' ? SVt_PVHV :
1492 Perl_str_to_version(pTHX_ SV *sv)
1497 const char *start = SvPV_const(sv,len);
1498 const char * const end = start + len;
1499 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1501 PERL_ARGS_ASSERT_STR_TO_VERSION;
1503 while (start < end) {
1507 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1512 retval += ((NV)n)/nshift;
1521 * Forces the next token to be a version number.
1522 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1523 * and if "guessing" is TRUE, then no new token is created (and the caller
1524 * must use an alternative parsing method).
1528 S_force_version(pTHX_ char *s, int guessing)
1534 I32 startoff = s - SvPVX(PL_linestr);
1537 PERL_ARGS_ASSERT_FORCE_VERSION;
1545 while (isDIGIT(*d) || *d == '_' || *d == '.')
1549 start_force(PL_curforce);
1550 curmad('X', newSVpvn(s,d-s));
1553 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1555 s = scan_num(s, &pl_yylval);
1556 version = pl_yylval.opval;
1557 ver = cSVOPx(version)->op_sv;
1558 if (SvPOK(ver) && !SvNIOK(ver)) {
1559 SvUPGRADE(ver, SVt_PVNV);
1560 SvNV_set(ver, str_to_version(ver));
1561 SvNOK_on(ver); /* hint that it is a version */
1564 else if (guessing) {
1567 sv_free(PL_nextwhite); /* let next token collect whitespace */
1569 s = SvPVX(PL_linestr) + startoff;
1577 if (PL_madskills && !version) {
1578 sv_free(PL_nextwhite); /* let next token collect whitespace */
1580 s = SvPVX(PL_linestr) + startoff;
1583 /* NOTE: The parser sees the package name and the VERSION swapped */
1584 start_force(PL_curforce);
1585 NEXTVAL_NEXTTOKE.opval = version;
1593 * Tokenize a quoted string passed in as an SV. It finds the next
1594 * chunk, up to end of string or a backslash. It may make a new
1595 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1600 S_tokeq(pTHX_ SV *sv)
1604 register char *send;
1609 PERL_ARGS_ASSERT_TOKEQ;
1614 s = SvPV_force(sv, len);
1615 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1618 while (s < send && *s != '\\')
1623 if ( PL_hints & HINT_NEW_STRING ) {
1624 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
1628 if (s + 1 < send && (s[1] == '\\'))
1629 s++; /* all that, just for this */
1634 SvCUR_set(sv, d - SvPVX_const(sv));
1636 if ( PL_hints & HINT_NEW_STRING )
1637 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
1642 * Now come three functions related to double-quote context,
1643 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1644 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1645 * interact with PL_lex_state, and create fake ( ... ) argument lists
1646 * to handle functions and concatenation.
1647 * They assume that whoever calls them will be setting up a fake
1648 * join call, because each subthing puts a ',' after it. This lets
1651 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1653 * (I'm not sure whether the spurious commas at the end of lcfirst's
1654 * arguments and join's arguments are created or not).
1659 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1661 * Pattern matching will set PL_lex_op to the pattern-matching op to
1662 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
1664 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1666 * Everything else becomes a FUNC.
1668 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1669 * had an OP_CONST or OP_READLINE). This just sets us up for a
1670 * call to S_sublex_push().
1674 S_sublex_start(pTHX)
1677 register const I32 op_type = pl_yylval.ival;
1679 if (op_type == OP_NULL) {
1680 pl_yylval.opval = PL_lex_op;
1684 if (op_type == OP_CONST || op_type == OP_READLINE) {
1685 SV *sv = tokeq(PL_lex_stuff);
1687 if (SvTYPE(sv) == SVt_PVIV) {
1688 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1690 const char * const p = SvPV_const(sv, len);
1691 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
1695 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1696 PL_lex_stuff = NULL;
1697 /* Allow <FH> // "foo" */
1698 if (op_type == OP_READLINE)
1699 PL_expect = XTERMORDORDOR;
1702 else if (op_type == OP_BACKTICK && PL_lex_op) {
1703 /* readpipe() vas overriden */
1704 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
1705 pl_yylval.opval = PL_lex_op;
1707 PL_lex_stuff = NULL;
1711 PL_sublex_info.super_state = PL_lex_state;
1712 PL_sublex_info.sub_inwhat = (U16)op_type;
1713 PL_sublex_info.sub_op = PL_lex_op;
1714 PL_lex_state = LEX_INTERPPUSH;
1718 pl_yylval.opval = PL_lex_op;
1728 * Create a new scope to save the lexing state. The scope will be
1729 * ended in S_sublex_done. Returns a '(', starting the function arguments
1730 * to the uc, lc, etc. found before.
1731 * Sets PL_lex_state to LEX_INTERPCONCAT.
1740 PL_lex_state = PL_sublex_info.super_state;
1741 SAVEBOOL(PL_lex_dojoin);
1742 SAVEI32(PL_lex_brackets);
1743 SAVEI32(PL_lex_casemods);
1744 SAVEI32(PL_lex_starts);
1745 SAVEI8(PL_lex_state);
1746 SAVEVPTR(PL_lex_inpat);
1747 SAVEI16(PL_lex_inwhat);
1748 SAVECOPLINE(PL_curcop);
1749 SAVEPPTR(PL_bufptr);
1750 SAVEPPTR(PL_bufend);
1751 SAVEPPTR(PL_oldbufptr);
1752 SAVEPPTR(PL_oldoldbufptr);
1753 SAVEPPTR(PL_last_lop);
1754 SAVEPPTR(PL_last_uni);
1755 SAVEPPTR(PL_linestart);
1756 SAVESPTR(PL_linestr);
1757 SAVEGENERICPV(PL_lex_brackstack);
1758 SAVEGENERICPV(PL_lex_casestack);
1760 PL_linestr = PL_lex_stuff;
1761 PL_lex_stuff = NULL;
1763 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1764 = SvPVX(PL_linestr);
1765 PL_bufend += SvCUR(PL_linestr);
1766 PL_last_lop = PL_last_uni = NULL;
1767 SAVEFREESV(PL_linestr);
1769 PL_lex_dojoin = FALSE;
1770 PL_lex_brackets = 0;
1771 Newx(PL_lex_brackstack, 120, char);
1772 Newx(PL_lex_casestack, 12, char);
1773 PL_lex_casemods = 0;
1774 *PL_lex_casestack = '\0';
1776 PL_lex_state = LEX_INTERPCONCAT;
1777 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1779 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1780 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1781 PL_lex_inpat = PL_sublex_info.sub_op;
1783 PL_lex_inpat = NULL;
1790 * Restores lexer state after a S_sublex_push.
1797 if (!PL_lex_starts++) {
1798 SV * const sv = newSVpvs("");
1799 if (SvUTF8(PL_linestr))
1801 PL_expect = XOPERATOR;
1802 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1806 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1807 PL_lex_state = LEX_INTERPCASEMOD;
1811 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1812 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1813 PL_linestr = PL_lex_repl;
1815 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1816 PL_bufend += SvCUR(PL_linestr);
1817 PL_last_lop = PL_last_uni = NULL;
1818 SAVEFREESV(PL_linestr);
1819 PL_lex_dojoin = FALSE;
1820 PL_lex_brackets = 0;
1821 PL_lex_casemods = 0;
1822 *PL_lex_casestack = '\0';
1824 if (SvEVALED(PL_lex_repl)) {
1825 PL_lex_state = LEX_INTERPNORMAL;
1827 /* we don't clear PL_lex_repl here, so that we can check later
1828 whether this is an evalled subst; that means we rely on the
1829 logic to ensure sublex_done() is called again only via the
1830 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1833 PL_lex_state = LEX_INTERPCONCAT;
1843 PL_endwhite = newSVpvs("");
1844 sv_catsv(PL_endwhite, PL_thiswhite);
1848 sv_setpvn(PL_thistoken,"",0);
1850 PL_realtokenstart = -1;
1854 PL_bufend = SvPVX(PL_linestr);
1855 PL_bufend += SvCUR(PL_linestr);
1856 PL_expect = XOPERATOR;
1857 PL_sublex_info.sub_inwhat = 0;
1865 Extracts a pattern, double-quoted string, or transliteration. This
1868 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1869 processing a pattern (PL_lex_inpat is true), a transliteration
1870 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1872 Returns a pointer to the character scanned up to. If this is
1873 advanced from the start pointer supplied (i.e. if anything was
1874 successfully parsed), will leave an OP for the substring scanned
1875 in pl_yylval. Caller must intuit reason for not parsing further
1876 by looking at the next characters herself.
1880 double-quoted style: \r and \n
1881 regexp special ones: \D \s
1884 case and quoting: \U \Q \E
1885 stops on @ and $, but not for $ as tail anchor
1887 In transliterations:
1888 characters are VERY literal, except for - not at the start or end
1889 of the string, which indicates a range. If the range is in bytes,
1890 scan_const expands the range to the full set of intermediate
1891 characters. If the range is in utf8, the hyphen is replaced with
1892 a certain range mark which will be handled by pmtrans() in op.c.
1894 In double-quoted strings:
1896 double-quoted style: \r and \n
1898 deprecated backrefs: \1 (in substitution replacements)
1899 case and quoting: \U \Q \E
1902 scan_const does *not* construct ops to handle interpolated strings.
1903 It stops processing as soon as it finds an embedded $ or @ variable
1904 and leaves it to the caller to work out what's going on.
1906 embedded arrays (whether in pattern or not) could be:
1907 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1909 $ in double-quoted strings must be the symbol of an embedded scalar.
1911 $ in pattern could be $foo or could be tail anchor. Assumption:
1912 it's a tail anchor if $ is the last thing in the string, or if it's
1913 followed by one of "()| \r\n\t"
1915 \1 (backreferences) are turned into $1
1917 The structure of the code is
1918 while (there's a character to process) {
1919 handle transliteration ranges
1920 skip regexp comments /(?#comment)/ and codes /(?{code})/
1921 skip #-initiated comments in //x patterns
1922 check for embedded arrays
1923 check for embedded scalars
1925 leave intact backslashes from leaveit (below)
1926 deprecate \1 in substitution replacements
1927 handle string-changing backslashes \l \U \Q \E, etc.
1928 switch (what was escaped) {
1929 handle \- in a transliteration (becomes a literal -)
1930 handle \132 (octal characters)
1931 handle \x15 and \x{1234} (hex characters)
1932 handle \N{name} (named characters)
1933 handle \cV (control characters)
1934 handle printf-style backslashes (\f, \r, \n, etc)
1936 } (end if backslash)
1937 } (end while character to read)
1942 S_scan_const(pTHX_ char *start)
1945 register char *send = PL_bufend; /* end of the constant */
1946 SV *sv = newSV(send - start); /* sv for the constant */
1947 register char *s = start; /* start of the constant */
1948 register char *d = SvPVX(sv); /* destination for copies */
1949 bool dorange = FALSE; /* are we in a translit range? */
1950 bool didrange = FALSE; /* did we just finish a range? */
1951 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1952 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1955 UV literal_endpoint = 0;
1956 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1959 PERL_ARGS_ASSERT_SCAN_CONST;
1961 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1962 /* If we are doing a trans and we know we want UTF8 set expectation */
1963 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1964 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1968 while (s < send || dorange) {
1969 /* get transliterations out of the way (they're most literal) */
1970 if (PL_lex_inwhat == OP_TRANS) {
1971 /* expand a range A-Z to the full set of characters. AIE! */
1973 I32 i; /* current expanded character */
1974 I32 min; /* first character in range */
1975 I32 max; /* last character in range */
1986 char * const c = (char*)utf8_hop((U8*)d, -1);
1990 *c = (char)UTF_TO_NATIVE(0xff);
1991 /* mark the range as done, and continue */
1997 i = d - SvPVX_const(sv); /* remember current offset */
2000 SvLEN(sv) + (has_utf8 ?
2001 (512 - UTF_CONTINUATION_MARK +
2004 /* How many two-byte within 0..255: 128 in UTF-8,
2005 * 96 in UTF-8-mod. */
2007 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
2009 d = SvPVX(sv) + i; /* refresh d after realloc */
2013 for (j = 0; j <= 1; j++) {
2014 char * const c = (char*)utf8_hop((U8*)d, -1);
2015 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2021 max = (U8)0xff; /* only to \xff */
2022 uvmax = uv; /* \x{100} to uvmax */
2024 d = c; /* eat endpoint chars */
2029 d -= 2; /* eat the first char and the - */
2030 min = (U8)*d; /* first char in range */
2031 max = (U8)d[1]; /* last char in range */
2038 "Invalid range \"%c-%c\" in transliteration operator",
2039 (char)min, (char)max);
2043 if (literal_endpoint == 2 &&
2044 ((isLOWER(min) && isLOWER(max)) ||
2045 (isUPPER(min) && isUPPER(max)))) {
2047 for (i = min; i <= max; i++)
2049 *d++ = NATIVE_TO_NEED(has_utf8,i);
2051 for (i = min; i <= max; i++)
2053 *d++ = NATIVE_TO_NEED(has_utf8,i);
2058 for (i = min; i <= max; i++)
2061 const U8 ch = (U8)NATIVE_TO_UTF(i);
2062 if (UNI_IS_INVARIANT(ch))
2065 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2066 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2075 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2077 *d++ = (char)UTF_TO_NATIVE(0xff);
2079 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2083 /* mark the range as done, and continue */
2087 literal_endpoint = 0;
2092 /* range begins (ignore - as first or last char) */
2093 else if (*s == '-' && s+1 < send && s != start) {
2095 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2102 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2112 literal_endpoint = 0;
2113 native_range = TRUE;
2118 /* if we get here, we're not doing a transliteration */
2120 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2121 except for the last char, which will be done separately. */
2122 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2124 while (s+1 < send && *s != ')')
2125 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2127 else if (s[2] == '{' /* This should match regcomp.c */
2128 || (s[2] == '?' && s[3] == '{'))
2131 char *regparse = s + (s[2] == '{' ? 3 : 4);
2134 while (count && (c = *regparse)) {
2135 if (c == '\\' && regparse[1])
2143 if (*regparse != ')')
2144 regparse--; /* Leave one char for continuation. */
2145 while (s < regparse)
2146 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2150 /* likewise skip #-initiated comments in //x patterns */
2151 else if (*s == '#' && PL_lex_inpat &&
2152 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2153 while (s+1 < send && *s != '\n')
2154 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2157 /* check for embedded arrays
2158 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2160 else if (*s == '@' && s[1]) {
2161 if (isALNUM_lazy_if(s+1,UTF))
2163 if (strchr(":'{$", s[1]))
2165 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2166 break; /* in regexp, neither @+ nor @- are interpolated */
2169 /* check for embedded scalars. only stop if we're sure it's a
2172 else if (*s == '$') {
2173 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2175 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
2176 break; /* in regexp, $ might be tail anchor */
2179 /* End of else if chain - OP_TRANS rejoin rest */
2182 if (*s == '\\' && s+1 < send) {
2185 /* deprecate \1 in strings and substitution replacements */
2186 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2187 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2189 if (ckWARN(WARN_SYNTAX))
2190 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2195 /* string-change backslash escapes */
2196 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2200 /* skip any other backslash escapes in a pattern */
2201 else if (PL_lex_inpat) {
2202 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2203 goto default_action;
2206 /* if we get here, it's either a quoted -, or a digit */
2209 /* quoted - in transliterations */
2211 if (PL_lex_inwhat == OP_TRANS) {
2218 if ((isALPHA(*s) || isDIGIT(*s)) &&
2220 Perl_warner(aTHX_ packWARN(WARN_MISC),
2221 "Unrecognized escape \\%c passed through",
2223 /* default action is to copy the quoted character */
2224 goto default_action;
2227 /* \132 indicates an octal constant */
2228 case '0': case '1': case '2': case '3':
2229 case '4': case '5': case '6': case '7':
2233 uv = grok_oct(s, &len, &flags, NULL);
2236 goto NUM_ESCAPE_INSERT;
2238 /* \x24 indicates a hex constant */
2242 char* const e = strchr(s, '}');
2243 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2244 PERL_SCAN_DISALLOW_PREFIX;
2249 yyerror("Missing right brace on \\x{}");
2253 uv = grok_hex(s, &len, &flags, NULL);
2259 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2260 uv = grok_hex(s, &len, &flags, NULL);
2266 /* Insert oct or hex escaped character.
2267 * There will always enough room in sv since such
2268 * escapes will be longer than any UTF-8 sequence
2269 * they can end up as. */
2271 /* We need to map to chars to ASCII before doing the tests
2274 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2275 if (!has_utf8 && uv > 255) {
2276 /* Might need to recode whatever we have
2277 * accumulated so far if it contains any
2280 * (Can't we keep track of that and avoid
2281 * this rescan? --jhi)
2285 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2286 if (!NATIVE_IS_INVARIANT(*c)) {
2291 const STRLEN offset = d - SvPVX_const(sv);
2293 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2297 while (src >= (const U8 *)SvPVX_const(sv)) {
2298 if (!NATIVE_IS_INVARIANT(*src)) {
2299 const U8 ch = NATIVE_TO_ASCII(*src);
2300 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2301 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2311 if (has_utf8 || uv > 255) {
2312 d = (char*)uvchr_to_utf8((U8*)d, uv);
2314 if (PL_lex_inwhat == OP_TRANS &&
2315 PL_sublex_info.sub_op) {
2316 PL_sublex_info.sub_op->op_private |=
2317 (PL_lex_repl ? OPpTRANS_FROM_UTF
2321 if (uv > 255 && !dorange)
2322 native_range = FALSE;
2334 /* \N{LATIN SMALL LETTER A} is a named character */
2338 char* e = strchr(s, '}');
2344 yyerror("Missing right brace on \\N{}");
2348 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2350 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2351 PERL_SCAN_DISALLOW_PREFIX;
2354 uv = grok_hex(s, &len, &flags, NULL);
2355 if ( e > s && len != (STRLEN)(e - s) ) {
2359 goto NUM_ESCAPE_INSERT;
2361 res = newSVpvn(s + 1, e - s - 1);
2362 res = new_constant( NULL, 0, "charnames",
2363 res, NULL, s - 2, e - s + 3 );
2365 sv_utf8_upgrade(res);
2366 str = SvPV_const(res,len);
2367 #ifdef EBCDIC_NEVER_MIND
2368 /* charnames uses pack U and that has been
2369 * recently changed to do the below uni->native
2370 * mapping, so this would be redundant (and wrong,
2371 * the code point would be doubly converted).
2372 * But leave this in just in case the pack U change
2373 * gets revoked, but the semantics is still
2374 * desireable for charnames. --jhi */
2376 UV uv = utf8_to_uvchr((const U8*)str, 0);
2379 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2381 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2382 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2383 str = SvPV_const(res, len);
2387 if (!has_utf8 && SvUTF8(res)) {
2388 const char * const ostart = SvPVX_const(sv);
2389 SvCUR_set(sv, d - ostart);
2392 sv_utf8_upgrade(sv);
2393 /* this just broke our allocation above... */
2394 SvGROW(sv, (STRLEN)(send - start));
2395 d = SvPVX(sv) + SvCUR(sv);
2398 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2399 const char * const odest = SvPVX_const(sv);
2401 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2402 d = SvPVX(sv) + (d - odest);
2406 native_range = FALSE; /* \N{} is guessed to be Unicode */
2408 Copy(str, d, len, char);
2415 yyerror("Missing braces on \\N{}");
2418 /* \c is a control character */
2427 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2430 yyerror("Missing control char name in \\c");
2434 /* printf-style backslashes, formfeeds, newlines, etc */
2436 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2439 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2442 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2445 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2448 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2451 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2454 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2460 } /* end if (backslash) */
2467 /* If we started with encoded form, or already know we want it
2468 and then encode the next character */
2469 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2471 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2472 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2475 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2476 const STRLEN off = d - SvPVX_const(sv);
2477 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2479 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2482 if (uv > 255 && !dorange)
2483 native_range = FALSE;
2487 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2489 } /* while loop to process each character */
2491 /* terminate the string and set up the sv */
2493 SvCUR_set(sv, d - SvPVX_const(sv));
2494 if (SvCUR(sv) >= SvLEN(sv))
2495 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2498 if (PL_encoding && !has_utf8) {
2499 sv_recode_to_utf8(sv, PL_encoding);
2505 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2506 PL_sublex_info.sub_op->op_private |=
2507 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2511 /* shrink the sv if we allocated more than we used */
2512 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2513 SvPV_shrink_to_cur(sv);
2516 /* return the substring (via pl_yylval) only if we parsed anything */
2517 if (s > PL_bufptr) {
2518 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2519 const char *const key = PL_lex_inpat ? "qr" : "q";
2520 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2524 if (PL_lex_inwhat == OP_TRANS) {
2527 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2535 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2538 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2545 * Returns TRUE if there's more to the expression (e.g., a subscript),
2548 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2550 * ->[ and ->{ return TRUE
2551 * { and [ outside a pattern are always subscripts, so return TRUE
2552 * if we're outside a pattern and it's not { or [, then return FALSE
2553 * if we're in a pattern and the first char is a {
2554 * {4,5} (any digits around the comma) returns FALSE
2555 * if we're in a pattern and the first char is a [
2557 * [SOMETHING] has a funky algorithm to decide whether it's a
2558 * character class or not. It has to deal with things like
2559 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2560 * anything else returns TRUE
2563 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2566 S_intuit_more(pTHX_ register char *s)
2570 PERL_ARGS_ASSERT_INTUIT_MORE;
2572 if (PL_lex_brackets)
2574 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2576 if (*s != '{' && *s != '[')
2581 /* In a pattern, so maybe we have {n,m}. */
2598 /* On the other hand, maybe we have a character class */
2601 if (*s == ']' || *s == '^')
2604 /* this is terrifying, and it works */
2605 int weight = 2; /* let's weigh the evidence */
2607 unsigned char un_char = 255, last_un_char;
2608 const char * const send = strchr(s,']');
2609 char tmpbuf[sizeof PL_tokenbuf * 4];
2611 if (!send) /* has to be an expression */
2614 Zero(seen,256,char);
2617 else if (isDIGIT(*s)) {
2619 if (isDIGIT(s[1]) && s[2] == ']')
2625 for (; s < send; s++) {
2626 last_un_char = un_char;
2627 un_char = (unsigned char)*s;
2632 weight -= seen[un_char] * 10;
2633 if (isALNUM_lazy_if(s+1,UTF)) {
2635 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2636 len = (int)strlen(tmpbuf);
2637 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2642 else if (*s == '$' && s[1] &&
2643 strchr("[#!%*<>()-=",s[1])) {
2644 if (/*{*/ strchr("])} =",s[2]))
2653 if (strchr("wds]",s[1]))
2655 else if (seen[(U8)'\''] || seen[(U8)'"'])
2657 else if (strchr("rnftbxcav",s[1]))
2659 else if (isDIGIT(s[1])) {
2661 while (s[1] && isDIGIT(s[1]))
2671 if (strchr("aA01! ",last_un_char))
2673 if (strchr("zZ79~",s[1]))
2675 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2676 weight -= 5; /* cope with negative subscript */
2679 if (!isALNUM(last_un_char)
2680 && !(last_un_char == '$' || last_un_char == '@'
2681 || last_un_char == '&')
2682 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2687 if (keyword(tmpbuf, d - tmpbuf, 0))
2690 if (un_char == last_un_char + 1)
2692 weight -= seen[un_char];
2697 if (weight >= 0) /* probably a character class */
2707 * Does all the checking to disambiguate
2709 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2710 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2712 * First argument is the stuff after the first token, e.g. "bar".
2714 * Not a method if bar is a filehandle.
2715 * Not a method if foo is a subroutine prototyped to take a filehandle.
2716 * Not a method if it's really "Foo $bar"
2717 * Method if it's "foo $bar"
2718 * Not a method if it's really "print foo $bar"
2719 * Method if it's really "foo package::" (interpreted as package->foo)
2720 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2721 * Not a method if bar is a filehandle or package, but is quoted with
2726 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2729 char *s = start + (*start == '$');
2730 char tmpbuf[sizeof PL_tokenbuf];
2737 PERL_ARGS_ASSERT_INTUIT_METHOD;
2740 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2744 const char *proto = SvPVX_const(cv);
2755 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2756 /* start is the beginning of the possible filehandle/object,
2757 * and s is the end of it
2758 * tmpbuf is a copy of it
2761 if (*start == '$') {
2762 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2763 isUPPER(*PL_tokenbuf))
2766 len = start - SvPVX(PL_linestr);
2770 start = SvPVX(PL_linestr) + len;
2774 return *s == '(' ? FUNCMETH : METHOD;
2776 if (!keyword(tmpbuf, len, 0)) {
2777 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2781 soff = s - SvPVX(PL_linestr);
2785 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2786 if (indirgv && GvCVu(indirgv))
2788 /* filehandle or package name makes it a method */
2789 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2791 soff = s - SvPVX(PL_linestr);
2794 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2795 return 0; /* no assumptions -- "=>" quotes bearword */
2797 start_force(PL_curforce);
2798 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2799 newSVpvn(tmpbuf,len));
2800 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2802 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2807 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2809 return *s == '(' ? FUNCMETH : METHOD;
2815 /* Encoded script support. filter_add() effectively inserts a
2816 * 'pre-processing' function into the current source input stream.
2817 * Note that the filter function only applies to the current source file
2818 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2820 * The datasv parameter (which may be NULL) can be used to pass
2821 * private data to this instance of the filter. The filter function
2822 * can recover the SV using the FILTER_DATA macro and use it to
2823 * store private buffers and state information.
2825 * The supplied datasv parameter is upgraded to a PVIO type
2826 * and the IoDIRP/IoANY field is used to store the function pointer,
2827 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2828 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2829 * private use must be set using malloc'd pointers.
2833 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2842 if (!PL_rsfp_filters)
2843 PL_rsfp_filters = newAV();
2846 SvUPGRADE(datasv, SVt_PVIO);
2847 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2848 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2849 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2850 FPTR2DPTR(void *, IoANY(datasv)),
2851 SvPV_nolen(datasv)));
2852 av_unshift(PL_rsfp_filters, 1);
2853 av_store(PL_rsfp_filters, 0, datasv) ;
2858 /* Delete most recently added instance of this filter function. */
2860 Perl_filter_del(pTHX_ filter_t funcp)
2865 PERL_ARGS_ASSERT_FILTER_DEL;
2868 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2869 FPTR2DPTR(void*, funcp)));
2871 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2873 /* if filter is on top of stack (usual case) just pop it off */
2874 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2875 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2876 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2877 IoANY(datasv) = (void *)NULL;
2878 sv_free(av_pop(PL_rsfp_filters));
2882 /* we need to search for the correct entry and clear it */
2883 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2887 /* Invoke the idxth filter function for the current rsfp. */
2888 /* maxlen 0 = read one text line */
2890 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2895 /* This API is bad. It should have been using unsigned int for maxlen.
2896 Not sure if we want to change the API, but if not we should sanity
2897 check the value here. */
2898 const unsigned int correct_length
2907 PERL_ARGS_ASSERT_FILTER_READ;
2909 if (!PL_parser || !PL_rsfp_filters)
2911 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2912 /* Provide a default input filter to make life easy. */
2913 /* Note that we append to the line. This is handy. */
2914 DEBUG_P(PerlIO_printf(Perl_debug_log,
2915 "filter_read %d: from rsfp\n", idx));
2916 if (correct_length) {
2919 const int old_len = SvCUR(buf_sv);
2921 /* ensure buf_sv is large enough */
2922 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2923 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2924 correct_length)) <= 0) {
2925 if (PerlIO_error(PL_rsfp))
2926 return -1; /* error */
2928 return 0 ; /* end of file */
2930 SvCUR_set(buf_sv, old_len + len) ;
2933 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2934 if (PerlIO_error(PL_rsfp))
2935 return -1; /* error */
2937 return 0 ; /* end of file */
2940 return SvCUR(buf_sv);
2942 /* Skip this filter slot if filter has been deleted */
2943 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2944 DEBUG_P(PerlIO_printf(Perl_debug_log,
2945 "filter_read %d: skipped (filter deleted)\n",
2947 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2949 /* Get function pointer hidden within datasv */
2950 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2951 DEBUG_P(PerlIO_printf(Perl_debug_log,
2952 "filter_read %d: via function %p (%s)\n",
2953 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2954 /* Call function. The function is expected to */
2955 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2956 /* Return: <0:error, =0:eof, >0:not eof */
2957 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2961 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2965 PERL_ARGS_ASSERT_FILTER_GETS;
2967 #ifdef PERL_CR_FILTER
2968 if (!PL_rsfp_filters) {
2969 filter_add(S_cr_textfilter,NULL);
2972 if (PL_rsfp_filters) {
2974 SvCUR_set(sv, 0); /* start with empty line */
2975 if (FILTER_READ(0, sv, 0) > 0)
2976 return ( SvPVX(sv) ) ;
2981 return (sv_gets(sv, fp, append));
2985 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
2990 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
2992 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2996 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2997 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
2999 return GvHV(gv); /* Foo:: */
3002 /* use constant CLASS => 'MyClass' */
3003 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3004 if (gv && GvCV(gv)) {
3005 SV * const sv = cv_const_sv(GvCV(gv));
3007 pkgname = SvPV_const(sv, len);
3010 return gv_stashpvn(pkgname, len, 0);
3014 * S_readpipe_override
3015 * Check whether readpipe() is overriden, and generates the appropriate
3016 * optree, provided sublex_start() is called afterwards.
3019 S_readpipe_override(pTHX)
3022 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3023 pl_yylval.ival = OP_BACKTICK;
3025 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3027 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3028 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3029 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3031 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3032 append_elem(OP_LIST,
3033 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3034 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3041 * The intent of this yylex wrapper is to minimize the changes to the
3042 * tokener when we aren't interested in collecting madprops. It remains
3043 * to be seen how successful this strategy will be...
3050 char *s = PL_bufptr;
3052 /* make sure PL_thiswhite is initialized */
3056 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3057 if (PL_pending_ident)
3058 return S_pending_ident(aTHX);
3060 /* previous token ate up our whitespace? */
3061 if (!PL_lasttoke && PL_nextwhite) {
3062 PL_thiswhite = PL_nextwhite;
3066 /* isolate the token, and figure out where it is without whitespace */
3067 PL_realtokenstart = -1;
3071 assert(PL_curforce < 0);
3073 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3074 if (!PL_thistoken) {
3075 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3076 PL_thistoken = newSVpvs("");
3078 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3079 PL_thistoken = newSVpvn(tstart, s - tstart);
3082 if (PL_thismad) /* install head */
3083 CURMAD('X', PL_thistoken);
3086 /* last whitespace of a sublex? */
3087 if (optype == ')' && PL_endwhite) {
3088 CURMAD('X', PL_endwhite);
3093 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3094 if (!PL_thiswhite && !PL_endwhite && !optype) {
3095 sv_free(PL_thistoken);
3100 /* put off final whitespace till peg */
3101 if (optype == ';' && !PL_rsfp) {
3102 PL_nextwhite = PL_thiswhite;
3105 else if (PL_thisopen) {
3106 CURMAD('q', PL_thisopen);
3108 sv_free(PL_thistoken);
3112 /* Store actual token text as madprop X */
3113 CURMAD('X', PL_thistoken);
3117 /* add preceding whitespace as madprop _ */
3118 CURMAD('_', PL_thiswhite);
3122 /* add quoted material as madprop = */
3123 CURMAD('=', PL_thisstuff);
3127 /* add terminating quote as madprop Q */
3128 CURMAD('Q', PL_thisclose);
3132 /* special processing based on optype */
3136 /* opval doesn't need a TOKEN since it can already store mp */
3146 if (pl_yylval.opval)
3147 append_madprops(PL_thismad, pl_yylval.opval, 0);
3155 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3164 /* remember any fake bracket that lexer is about to discard */
3165 if (PL_lex_brackets == 1 &&
3166 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3169 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3172 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3173 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3176 break; /* don't bother looking for trailing comment */
3185 /* attach a trailing comment to its statement instead of next token */
3189 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3191 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3193 if (*s == '\n' || *s == '#') {
3194 while (s < PL_bufend && *s != '\n')
3198 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3199 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3216 /* Create new token struct. Note: opvals return early above. */
3217 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3224 S_tokenize_use(pTHX_ int is_use, char *s) {
3227 PERL_ARGS_ASSERT_TOKENIZE_USE;
3229 if (PL_expect != XSTATE)
3230 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3231 is_use ? "use" : "no"));
3233 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3234 s = force_version(s, TRUE);
3235 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3236 start_force(PL_curforce);
3237 NEXTVAL_NEXTTOKE.opval = NULL;
3240 else if (*s == 'v') {
3241 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3242 s = force_version(s, FALSE);
3246 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3247 s = force_version(s, FALSE);
3249 pl_yylval.ival = is_use;
3253 static const char* const exp_name[] =
3254 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3255 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3262 Works out what to call the token just pulled out of the input
3263 stream. The yacc parser takes care of taking the ops we return and
3264 stitching them into a tree.
3270 if read an identifier
3271 if we're in a my declaration
3272 croak if they tried to say my($foo::bar)
3273 build the ops for a my() declaration
3274 if it's an access to a my() variable
3275 are we in a sort block?
3276 croak if my($a); $a <=> $b
3277 build ops for access to a my() variable
3278 if in a dq string, and they've said @foo and we can't find @foo
3280 build ops for a bareword
3281 if we already built the token before, use it.
3286 #pragma segment Perl_yylex
3292 register char *s = PL_bufptr;
3297 /* orig_keyword, gvp, and gv are initialized here because
3298 * jump to the label just_a_word_zero can bypass their
3299 * initialization later. */
3300 I32 orig_keyword = 0;
3305 SV* tmp = newSVpvs("");
3306 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3307 (IV)CopLINE(PL_curcop),
3308 lex_state_names[PL_lex_state],
3309 exp_name[PL_expect],
3310 pv_display(tmp, s, strlen(s), 0, 60));
3313 /* check if there's an identifier for us to look at */
3314 if (PL_pending_ident)
3315 return REPORT(S_pending_ident(aTHX));
3317 /* no identifier pending identification */
3319 switch (PL_lex_state) {
3321 case LEX_NORMAL: /* Some compilers will produce faster */
3322 case LEX_INTERPNORMAL: /* code if we comment these out. */
3326 /* when we've already built the next token, just pull it out of the queue */
3330 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3332 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3333 PL_nexttoke[PL_lasttoke].next_mad = 0;
3334 if (PL_thismad && PL_thismad->mad_key == '_') {
3335 PL_thiswhite = (SV*)PL_thismad->mad_val;
3336 PL_thismad->mad_val = 0;
3337 mad_free(PL_thismad);
3342 PL_lex_state = PL_lex_defer;
3343 PL_expect = PL_lex_expect;
3344 PL_lex_defer = LEX_NORMAL;
3345 if (!PL_nexttoke[PL_lasttoke].next_type)
3350 pl_yylval = PL_nextval[PL_nexttoke];
3352 PL_lex_state = PL_lex_defer;
3353 PL_expect = PL_lex_expect;
3354 PL_lex_defer = LEX_NORMAL;
3358 /* FIXME - can these be merged? */
3359 return(PL_nexttoke[PL_lasttoke].next_type);
3361 return REPORT(PL_nexttype[PL_nexttoke]);
3364 /* interpolated case modifiers like \L \U, including \Q and \E.
3365 when we get here, PL_bufptr is at the \
3367 case LEX_INTERPCASEMOD:
3369 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3370 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3372 /* handle \E or end of string */
3373 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3375 if (PL_lex_casemods) {
3376 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3377 PL_lex_casestack[PL_lex_casemods] = '\0';
3379 if (PL_bufptr != PL_bufend
3380 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3382 PL_lex_state = LEX_INTERPCONCAT;
3385 PL_thistoken = newSVpvs("\\E");
3391 while (PL_bufptr != PL_bufend &&
3392 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3394 PL_thiswhite = newSVpvs("");
3395 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3399 if (PL_bufptr != PL_bufend)
3402 PL_lex_state = LEX_INTERPCONCAT;
3406 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3407 "### Saw case modifier\n"); });
3409 if (s[1] == '\\' && s[2] == 'E') {
3412 PL_thiswhite = newSVpvs("");
3413 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3416 PL_lex_state = LEX_INTERPCONCAT;
3421 if (!PL_madskills) /* when just compiling don't need correct */
3422 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3423 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3424 if ((*s == 'L' || *s == 'U') &&
3425 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3426 PL_lex_casestack[--PL_lex_casemods] = '\0';
3429 if (PL_lex_casemods > 10)
3430 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3431 PL_lex_casestack[PL_lex_casemods++] = *s;
3432 PL_lex_casestack[PL_lex_casemods] = '\0';
3433 PL_lex_state = LEX_INTERPCONCAT;
3434 start_force(PL_curforce);
3435 NEXTVAL_NEXTTOKE.ival = 0;
3437 start_force(PL_curforce);
3439 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3441 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3443 NEXTVAL_NEXTTOKE.ival = OP_LC;
3445 NEXTVAL_NEXTTOKE.ival = OP_UC;
3447 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3449 Perl_croak(aTHX_ "panic: yylex");
3451 SV* const tmpsv = newSVpvs("\\ ");
3452 /* replace the space with the character we want to escape
3454 SvPVX(tmpsv)[1] = *s;
3460 if (PL_lex_starts) {
3466 sv_free(PL_thistoken);
3467 PL_thistoken = newSVpvs("");
3470 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3471 if (PL_lex_casemods == 1 && PL_lex_inpat)
3480 case LEX_INTERPPUSH:
3481 return REPORT(sublex_push());
3483 case LEX_INTERPSTART:
3484 if (PL_bufptr == PL_bufend)
3485 return REPORT(sublex_done());
3486 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3487 "### Interpolated variable\n"); });
3489 PL_lex_dojoin = (*PL_bufptr == '@');
3490 PL_lex_state = LEX_INTERPNORMAL;
3491 if (PL_lex_dojoin) {
3492 start_force(PL_curforce);
3493 NEXTVAL_NEXTTOKE.ival = 0;
3495 start_force(PL_curforce);
3496 force_ident("\"", '$');
3497 start_force(PL_curforce);
3498 NEXTVAL_NEXTTOKE.ival = 0;
3500 start_force(PL_curforce);
3501 NEXTVAL_NEXTTOKE.ival = 0;
3503 start_force(PL_curforce);
3504 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3507 if (PL_lex_starts++) {
3512 sv_free(PL_thistoken);
3513 PL_thistoken = newSVpvs("");
3516 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3517 if (!PL_lex_casemods && PL_lex_inpat)
3524 case LEX_INTERPENDMAYBE:
3525 if (intuit_more(PL_bufptr)) {
3526 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3532 if (PL_lex_dojoin) {
3533 PL_lex_dojoin = FALSE;
3534 PL_lex_state = LEX_INTERPCONCAT;
3538 sv_free(PL_thistoken);
3539 PL_thistoken = newSVpvs("");
3544 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3545 && SvEVALED(PL_lex_repl))
3547 if (PL_bufptr != PL_bufend)
3548 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3552 case LEX_INTERPCONCAT:
3554 if (PL_lex_brackets)
3555 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3557 if (PL_bufptr == PL_bufend)
3558 return REPORT(sublex_done());
3560 if (SvIVX(PL_linestr) == '\'') {
3561 SV *sv = newSVsv(PL_linestr);
3564 else if ( PL_hints & HINT_NEW_RE )
3565 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3566 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3570 s = scan_const(PL_bufptr);
3572 PL_lex_state = LEX_INTERPCASEMOD;
3574 PL_lex_state = LEX_INTERPSTART;
3577 if (s != PL_bufptr) {
3578 start_force(PL_curforce);
3580 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3582 NEXTVAL_NEXTTOKE = pl_yylval;
3585 if (PL_lex_starts++) {
3589 sv_free(PL_thistoken);
3590 PL_thistoken = newSVpvs("");
3593 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3594 if (!PL_lex_casemods && PL_lex_inpat)
3607 PL_lex_state = LEX_NORMAL;
3608 s = scan_formline(PL_bufptr);
3609 if (!PL_lex_formbrack)
3615 PL_oldoldbufptr = PL_oldbufptr;
3621 sv_free(PL_thistoken);
3624 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3628 if (isIDFIRST_lazy_if(s,UTF))
3630 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3631 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3634 goto fake_eof; /* emulate EOF on ^D or ^Z */
3643 if (PL_lex_brackets) {
3644 yyerror((const char *)
3646 ? "Format not terminated"
3647 : "Missing right curly or square bracket"));
3649 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3650 "### Tokener got EOF\n");
3654 if (s++ < PL_bufend)
3655 goto retry; /* ignore stray nulls */
3658 if (!PL_in_eval && !PL_preambled) {
3659 PL_preambled = TRUE;
3665 /* Generate a string of Perl code to load the debugger.
3666 * If PERL5DB is set, it will return the contents of that,
3667 * otherwise a compile-time require of perl5db.pl. */
3669 const char * const pdb = PerlEnv_getenv("PERL5DB");
3672 sv_setpv(PL_linestr, pdb);
3673 sv_catpvs(PL_linestr,";");
3675 SETERRNO(0,SS_NORMAL);
3676 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3679 sv_setpvs(PL_linestr,"");
3680 if (PL_preambleav) {
3681 SV **svp = AvARRAY(PL_preambleav);
3682 SV **const end = svp + AvFILLp(PL_preambleav);
3684 sv_catsv(PL_linestr, *svp);
3686 sv_catpvs(PL_linestr, ";");
3688 sv_free((SV*)PL_preambleav);
3689 PL_preambleav = NULL;
3691 if (PL_minus_n || PL_minus_p) {
3692 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3694 sv_catpvs(PL_linestr,"chomp;");
3697 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3698 || *PL_splitstr == '"')
3699 && strchr(PL_splitstr + 1, *PL_splitstr))
3700 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3702 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3703 bytes can be used as quoting characters. :-) */
3704 const char *splits = PL_splitstr;
3705 sv_catpvs(PL_linestr, "our @F=split(q\0");
3708 if (*splits == '\\')
3709 sv_catpvn(PL_linestr, splits, 1);
3710 sv_catpvn(PL_linestr, splits, 1);
3711 } while (*splits++);
3712 /* This loop will embed the trailing NUL of
3713 PL_linestr as the last thing it does before
3715 sv_catpvs(PL_linestr, ");");
3719 sv_catpvs(PL_linestr,"our @F=split(' ');");
3723 sv_catpvs(PL_linestr,
3724 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3725 sv_catpvs(PL_linestr, "\n");
3726 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3727 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3728 PL_last_lop = PL_last_uni = NULL;
3729 if (PERLDB_LINE && PL_curstash != PL_debstash)
3730 update_debugger_info(PL_linestr, NULL, 0);
3734 bof = PL_rsfp ? TRUE : FALSE;
3735 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3738 PL_realtokenstart = -1;
3741 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3742 PerlIO_clearerr(PL_rsfp);
3744 (void)PerlIO_close(PL_rsfp);
3746 PL_doextract = FALSE;
3748 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3754 sv_setpvs(PL_linestr, ";}continue{print;}");
3756 sv_setpvs(PL_linestr, ";}");
3757 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3758 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3759 PL_last_lop = PL_last_uni = NULL;
3760 PL_minus_n = PL_minus_p = 0;
3763 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3764 PL_last_lop = PL_last_uni = NULL;
3765 sv_setpvn(PL_linestr,"",0);
3766 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3768 /* If it looks like the start of a BOM or raw UTF-16,
3769 * check if it in fact is. */
3775 #ifdef PERLIO_IS_STDIO
3776 # ifdef __GNU_LIBRARY__
3777 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3778 # define FTELL_FOR_PIPE_IS_BROKEN
3782 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3783 # define FTELL_FOR_PIPE_IS_BROKEN
3788 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3790 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3791 s = swallow_bom((U8*)s);
3795 /* Incest with pod. */
3798 sv_catsv(PL_thiswhite, PL_linestr);
3800 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3801 sv_setpvn(PL_linestr, "", 0);
3802 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3803 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3804 PL_last_lop = PL_last_uni = NULL;
3805 PL_doextract = FALSE;
3809 } while (PL_doextract);
3810 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3811 if (PERLDB_LINE && PL_curstash != PL_debstash)
3812 update_debugger_info(PL_linestr, NULL, 0);
3813 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3814 PL_last_lop = PL_last_uni = NULL;
3815 if (CopLINE(PL_curcop) == 1) {
3816 while (s < PL_bufend && isSPACE(*s))
3818 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3822 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3826 if (*s == '#' && *(s+1) == '!')
3828 #ifdef ALTERNATE_SHEBANG
3830 static char const as[] = ALTERNATE_SHEBANG;
3831 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3832 d = s + (sizeof(as) - 1);
3834 #endif /* ALTERNATE_SHEBANG */
3843 while (*d && !isSPACE(*d))
3847 #ifdef ARG_ZERO_IS_SCRIPT
3848 if (ipathend > ipath) {
3850 * HP-UX (at least) sets argv[0] to the script name,
3851 * which makes $^X incorrect. And Digital UNIX and Linux,
3852 * at least, set argv[0] to the basename of the Perl
3853 * interpreter. So, having found "#!", we'll set it right.
3855 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3857 assert(SvPOK(x) || SvGMAGICAL(x));
3858 if (sv_eq(x, CopFILESV(PL_curcop))) {
3859 sv_setpvn(x, ipath, ipathend - ipath);
3865 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3866 const char * const lstart = SvPV_const(x,llen);
3868 bstart += blen - llen;
3869 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3870 sv_setpvn(x, ipath, ipathend - ipath);
3875 TAINT_NOT; /* $^X is always tainted, but that's OK */
3877 #endif /* ARG_ZERO_IS_SCRIPT */
3882 d = instr(s,"perl -");
3884 d = instr(s,"perl");
3886 /* avoid getting into infinite loops when shebang
3887 * line contains "Perl" rather than "perl" */
3889 for (d = ipathend-4; d >= ipath; --d) {
3890 if ((*d == 'p' || *d == 'P')
3891 && !ibcmp(d, "perl", 4))
3901 #ifdef ALTERNATE_SHEBANG
3903 * If the ALTERNATE_SHEBANG on this system starts with a
3904 * character that can be part of a Perl expression, then if
3905 * we see it but not "perl", we're probably looking at the
3906 * start of Perl code, not a request to hand off to some
3907 * other interpreter. Similarly, if "perl" is there, but
3908 * not in the first 'word' of the line, we assume the line
3909 * contains the start of the Perl program.
3911 if (d && *s != '#') {
3912 const char *c = ipath;
3913 while (*c && !strchr("; \t\r\n\f\v#", *c))
3916 d = NULL; /* "perl" not in first word; ignore */
3918 *s = '#'; /* Don't try to parse shebang line */
3920 #endif /* ALTERNATE_SHEBANG */
3921 #ifndef MACOS_TRADITIONAL
3926 !instr(s,"indir") &&
3927 instr(PL_origargv[0],"perl"))
3934 while (s < PL_bufend && isSPACE(*s))
3936 if (s < PL_bufend) {
3937 Newxz(newargv,PL_origargc+3,char*);
3939 while (s < PL_bufend && !isSPACE(*s))
3942 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3945 newargv = PL_origargv;
3948 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3950 Perl_croak(aTHX_ "Can't exec %s", ipath);
3954 while (*d && !isSPACE(*d))
3956 while (SPACE_OR_TAB(*d))
3960 const bool switches_done = PL_doswitches;
3961 const U32 oldpdb = PL_perldb;
3962 const bool oldn = PL_minus_n;
3963 const bool oldp = PL_minus_p;
3967 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3968 const char * const m = d1;
3969 while (*d1 && !isSPACE(*d1))
3971 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3974 d1 = moreswitches(d1);
3976 if (PL_doswitches && !switches_done) {
3977 int argc = PL_origargc;
3978 char **argv = PL_origargv;
3981 } while (argc && argv[0][0] == '-' && argv[0][1]);
3982 init_argv_symbols(argc,argv);
3984 if ((PERLDB_LINE && !oldpdb) ||
3985 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3986 /* if we have already added "LINE: while (<>) {",
3987 we must not do it again */
3989 sv_setpvn(PL_linestr, "", 0);
3990 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3991 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3992 PL_last_lop = PL_last_uni = NULL;
3993 PL_preambled = FALSE;
3995 (void)gv_fetchfile(PL_origfilename);
4002 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4004 PL_lex_state = LEX_FORMLINE;
4009 #ifdef PERL_STRICT_CR
4010 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4012 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4014 case ' ': case '\t': case '\f': case 013:
4015 #ifdef MACOS_TRADITIONAL
4019 PL_realtokenstart = -1;
4021 PL_thiswhite = newSVpvs("");
4022 sv_catpvn(PL_thiswhite, s, 1);
4029 PL_realtokenstart = -1;
4033 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4034 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4035 /* handle eval qq[#line 1 "foo"\n ...] */
4036 CopLINE_dec(PL_curcop);
4039 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4041 if (!PL_in_eval || PL_rsfp)
4046 while (d < PL_bufend && *d != '\n')
4050 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4051 Perl_croak(aTHX_ "panic: input overflow");
4054 PL_thiswhite = newSVpvn(s, d - s);
4059 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4061 PL_lex_state = LEX_FORMLINE;
4067 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4068 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4071 TOKEN(PEG); /* make sure any #! line is accessible */
4076 /* if (PL_madskills && PL_lex_formbrack) { */
4078 while (d < PL_bufend && *d != '\n')
4082 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4083 Perl_croak(aTHX_ "panic: input overflow");
4084 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4086 PL_thiswhite = newSVpvs("");
4087 if (CopLINE(PL_curcop) == 1) {
4088 sv_setpvn(PL_thiswhite, "", 0);
4091 sv_catpvn(PL_thiswhite, s, d - s);
4105 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4113 while (s < PL_bufend && SPACE_OR_TAB(*s))
4116 if (strnEQ(s,"=>",2)) {
4117 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4118 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4119 OPERATOR('-'); /* unary minus */
4121 PL_last_uni = PL_oldbufptr;
4123 case 'r': ftst = OP_FTEREAD; break;
4124 case 'w': ftst = OP_FTEWRITE; break;
4125 case 'x': ftst = OP_FTEEXEC; break;
4126 case 'o': ftst = OP_FTEOWNED; break;
4127 case 'R': ftst = OP_FTRREAD; break;
4128 case 'W': ftst = OP_FTRWRITE; break;
4129 case 'X': ftst = OP_FTREXEC; break;
4130 case 'O': ftst = OP_FTROWNED; break;
4131 case 'e': ftst = OP_FTIS; break;
4132 case 'z': ftst = OP_FTZERO; break;
4133 case 's': ftst = OP_FTSIZE; break;
4134 case 'f': ftst = OP_FTFILE; break;
4135 case 'd': ftst = OP_FTDIR; break;
4136 case 'l': ftst = OP_FTLINK; break;
4137 case 'p': ftst = OP_FTPIPE; break;
4138 case 'S': ftst = OP_FTSOCK; break;
4139 case 'u': ftst = OP_FTSUID; break;
4140 case 'g': ftst = OP_FTSGID; break;
4141 case 'k': ftst = OP_FTSVTX; break;
4142 case 'b': ftst = OP_FTBLK; break;
4143 case 'c': ftst = OP_FTCHR; break;
4144 case 't': ftst = OP_FTTTY; break;
4145 case 'T': ftst = OP_FTTEXT; break;
4146 case 'B': ftst = OP_FTBINARY; break;
4147 case 'M': case 'A': case 'C':
4148 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4150 case 'M': ftst = OP_FTMTIME; break;
4151 case 'A': ftst = OP_FTATIME; break;
4152 case 'C': ftst = OP_FTCTIME; break;
4160 PL_last_lop_op = (OPCODE)ftst;
4161 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4162 "### Saw file test %c\n", (int)tmp);
4167 /* Assume it was a minus followed by a one-letter named
4168 * subroutine call (or a -bareword), then. */
4169 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4170 "### '-%c' looked like a file test but was not\n",
4177 const char tmp = *s++;
4180 if (PL_expect == XOPERATOR)
4185 else if (*s == '>') {
4188 if (isIDFIRST_lazy_if(s,UTF)) {
4189 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4197 if (PL_expect == XOPERATOR)
4200 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4202 OPERATOR('-'); /* unary minus */
4208 const char tmp = *s++;
4211 if (PL_expect == XOPERATOR)
4216 if (PL_expect == XOPERATOR)
4219 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4226 if (PL_expect != XOPERATOR) {
4227 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4228 PL_expect = XOPERATOR;
4229 force_ident(PL_tokenbuf, '*');
4242 if (PL_expect == XOPERATOR) {
4246 PL_tokenbuf[0] = '%';
4247 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4248 sizeof PL_tokenbuf - 1, FALSE);
4249 if (!PL_tokenbuf[1]) {
4252 PL_pending_ident = '%';
4263 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4270 const char tmp = *s++;
4276 goto just_a_word_zero_gv;
4279 switch (PL_expect) {
4285 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4287 PL_bufptr = s; /* update in case we back off */
4293 PL_expect = XTERMBLOCK;
4296 stuffstart = s - SvPVX(PL_linestr) - 1;
4300 while (isIDFIRST_lazy_if(s,UTF)) {
4303 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4304 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4305 if (tmp < 0) tmp = -tmp;
4319 sv = newSVpvn(s, len);
4321 d = scan_str(d,TRUE,TRUE);
4323 /* MUST advance bufptr here to avoid bogus
4324 "at end of line" context messages from yyerror().
4326 PL_bufptr = s + len;
4327 yyerror("Unterminated attribute parameter in attribute list");
4331 return REPORT(0); /* EOF indicator */
4335 sv_catsv(sv, PL_lex_stuff);
4336 attrs = append_elem(OP_LIST, attrs,
4337 newSVOP(OP_CONST, 0, sv));
4338 SvREFCNT_dec(PL_lex_stuff);
4339 PL_lex_stuff = NULL;
4342 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4344 if (PL_in_my == KEY_our) {
4346 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4348 /* skip to avoid loading attributes.pm */
4350 deprecate(":unique");
4353 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4356 /* NOTE: any CV attrs applied here need to be part of
4357 the CVf_BUILTIN_ATTRS define in cv.h! */
4358 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4360 CvLVALUE_on(PL_compcv);
4362 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4364 CvLOCKED_on(PL_compcv);
4366 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4368 CvMETHOD_on(PL_compcv);
4370 /* After we've set the flags, it could be argued that
4371 we don't need to do the attributes.pm-based setting
4372 process, and shouldn't bother appending recognized
4373 flags. To experiment with that, uncomment the
4374 following "else". (Note that's already been
4375 uncommented. That keeps the above-applied built-in
4376 attributes from being intercepted (and possibly
4377 rejected) by a package's attribute routines, but is
4378 justified by the performance win for the common case
4379 of applying only built-in attributes.) */
4381 attrs = append_elem(OP_LIST, attrs,
4382 newSVOP(OP_CONST, 0,
4386 if (*s == ':' && s[1] != ':')
4389 break; /* require real whitespace or :'s */
4390 /* XXX losing whitespace on sequential attributes here */
4394 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4395 if (*s != ';' && *s != '}' && *s != tmp
4396 && (tmp != '=' || *s != ')')) {
4397 const char q = ((*s == '\'') ? '"' : '\'');
4398 /* If here for an expression, and parsed no attrs, back
4400 if (tmp == '=' && !attrs) {
4404 /* MUST advance bufptr here to avoid bogus "at end of line"
4405 context messages from yyerror().
4408 yyerror( (const char *)
4410 ? Perl_form(aTHX_ "Invalid separator character "
4411 "%c%c%c in attribute list", q, *s, q)
4412 : "Unterminated attribute list" ) );
4420 start_force(PL_curforce);
4421 NEXTVAL_NEXTTOKE.opval = attrs;
4422 CURMAD('_', PL_nextwhite);
4427 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4428 (s - SvPVX(PL_linestr)) - stuffstart);
4436 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4437 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4445 const char tmp = *s++;
4450 const char tmp = *s++;
4458 if (PL_lex_brackets <= 0)
4459 yyerror("Unmatched right square bracket");
4462 if (PL_lex_state == LEX_INTERPNORMAL) {
4463 if (PL_lex_brackets == 0) {
4464 if (*s == '-' && s[1] == '>')
4465 PL_lex_state = LEX_INTERPENDMAYBE;
4466 else if (*s != '[' && *s != '{')
4467 PL_lex_state = LEX_INTERPEND;
4474 if (PL_lex_brackets > 100) {
4475 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4477 switch (PL_expect) {
4479 if (PL_lex_formbrack) {
4483 if (PL_oldoldbufptr == PL_last_lop)
4484 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4486 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4487 OPERATOR(HASHBRACK);
4489 while (s < PL_bufend && SPACE_OR_TAB(*s))
4492 PL_tokenbuf[0] = '\0';
4493 if (d < PL_bufend && *d == '-') {
4494 PL_tokenbuf[0] = '-';
4496 while (d < PL_bufend && SPACE_OR_TAB(*d))
4499 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4500 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4502 while (d < PL_bufend && SPACE_OR_TAB(*d))
4505 const char minus = (PL_tokenbuf[0] == '-');
4506 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4514 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4519 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4524 if (PL_oldoldbufptr == PL_last_lop)
4525 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4527 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4530 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4532 /* This hack is to get the ${} in the message. */
4534 yyerror("syntax error");
4537 OPERATOR(HASHBRACK);
4539 /* This hack serves to disambiguate a pair of curlies
4540 * as being a block or an anon hash. Normally, expectation
4541 * determines that, but in cases where we're not in a
4542 * position to expect anything in particular (like inside
4543 * eval"") we have to resolve the ambiguity. This code
4544 * covers the case where the first term in the curlies is a
4545 * quoted string. Most other cases need to be explicitly
4546 * disambiguated by prepending a "+" before the opening
4547 * curly in order to force resolution as an anon hash.
4549 * XXX should probably propagate the outer expectation
4550 * into eval"" to rely less on this hack, but that could
4551 * potentially break current behavior of eval"".
4555 if (*s == '\'' || *s == '"' || *s == '`') {
4556 /* common case: get past first string, handling escapes */
4557 for (t++; t < PL_bufend && *t != *s;)
4558 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4562 else if (*s == 'q') {
4565 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4568 /* skip q//-like construct */
4570 char open, close, term;
4573 while (t < PL_bufend && isSPACE(*t))
4575 /* check for q => */
4576 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4577 OPERATOR(HASHBRACK);
4581 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4585 for (t++; t < PL_bufend; t++) {
4586 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4588 else if (*t == open)
4592 for (t++; t < PL_bufend; t++) {
4593 if (*t == '\\' && t+1 < PL_bufend)
4595 else if (*t == close && --brackets <= 0)
4597 else if (*t == open)
4604 /* skip plain q word */
4605 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4608 else if (isALNUM_lazy_if(t,UTF)) {
4610 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4613 while (t < PL_bufend && isSPACE(*t))
4615 /* if comma follows first term, call it an anon hash */
4616 /* XXX it could be a comma expression with loop modifiers */
4617 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4618 || (*t == '=' && t[1] == '>')))
4619 OPERATOR(HASHBRACK);
4620 if (PL_expect == XREF)
4623 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4629 pl_yylval.ival = CopLINE(PL_curcop);
4630 if (isSPACE(*s) || *s == '#')
4631 PL_copline = NOLINE; /* invalidate current command line number */
4636 if (PL_lex_brackets <= 0)
4637 yyerror("Unmatched right curly bracket");
4639 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4640 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4641 PL_lex_formbrack = 0;
4642 if (PL_lex_state == LEX_INTERPNORMAL) {
4643 if (PL_lex_brackets == 0) {
4644 if (PL_expect & XFAKEBRACK) {
4645 PL_expect &= XENUMMASK;
4646 PL_lex_state = LEX_INTERPEND;
4651 PL_thiswhite = newSVpvs("");
4652 sv_catpvn(PL_thiswhite,"}",1);
4655 return yylex(); /* ignore fake brackets */
4657 if (*s == '-' && s[1] == '>')
4658 PL_lex_state = LEX_INTERPENDMAYBE;
4659 else if (*s != '[' && *s != '{')
4660 PL_lex_state = LEX_INTERPEND;
4663 if (PL_expect & XFAKEBRACK) {
4664 PL_expect &= XENUMMASK;
4666 return yylex(); /* ignore fake brackets */
4668 start_force(PL_curforce);
4670 curmad('X', newSVpvn(s-1,1));
4671 CURMAD('_', PL_thiswhite);
4676 PL_thistoken = newSVpvs("");
4684 if (PL_expect == XOPERATOR) {
4685 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4686 && isIDFIRST_lazy_if(s,UTF))
4688 CopLINE_dec(PL_curcop);
4689 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4690 CopLINE_inc(PL_curcop);
4695 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4697 PL_expect = XOPERATOR;
4698 force_ident(PL_tokenbuf, '&');
4702 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4714 const char tmp = *s++;
4721 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4722 && strchr("+-*/%.^&|<",tmp))
4723 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4724 "Reversed %c= operator",(int)tmp);
4726 if (PL_expect == XSTATE && isALPHA(tmp) &&
4727 (s == PL_linestart+1 || s[-2] == '\n') )
4729 if (PL_in_eval && !PL_rsfp) {
4734 if (strnEQ(s,"=cut",4)) {
4750 PL_thiswhite = newSVpvs("");
4751 sv_catpvn(PL_thiswhite, PL_linestart,
4752 PL_bufend - PL_linestart);
4756 PL_doextract = TRUE;
4760 if (PL_lex_brackets < PL_lex_formbrack) {
4762 #ifdef PERL_STRICT_CR
4763 while (SPACE_OR_TAB(*t))
4765 while (SPACE_OR_TAB(*t) || *t == '\r')
4768 if (*t == '\n' || *t == '#') {
4779 const char tmp = *s++;
4781 /* was this !=~ where !~ was meant?
4782 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4784 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4785 const char *t = s+1;
4787 while (t < PL_bufend && isSPACE(*t))
4790 if (*t == '/' || *t == '?' ||
4791 ((*t == 'm' || *t == 's' || *t == 'y')
4792 && !isALNUM(t[1])) ||
4793 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4794 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4795 "!=~ should be !~");
4805 if (PL_expect != XOPERATOR) {
4806 if (s[1] != '<' && !strchr(s,'>'))
4809 s = scan_heredoc(s);
4811 s = scan_inputsymbol(s);
4812 TERM(sublex_start());
4818 SHop(OP_LEFT_SHIFT);
4832 const char tmp = *s++;
4834 SHop(OP_RIGHT_SHIFT);
4835 else if (tmp == '=')
4844 if (PL_expect == XOPERATOR) {
4845 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4847 deprecate_old(commaless_variable_list);
4848 return REPORT(','); /* grandfather non-comma-format format */
4852 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4853 PL_tokenbuf[0] = '@';
4854 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4855 sizeof PL_tokenbuf - 1, FALSE);
4856 if (PL_expect == XOPERATOR)
4857 no_op("Array length", s);
4858 if (!PL_tokenbuf[1])
4860 PL_expect = XOPERATOR;
4861 PL_pending_ident = '#';
4865 PL_tokenbuf[0] = '$';
4866 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4867 sizeof PL_tokenbuf - 1, FALSE);
4868 if (PL_expect == XOPERATOR)
4870 if (!PL_tokenbuf[1]) {
4872 yyerror("Final $ should be \\$ or $name");
4876 /* This kludge not intended to be bulletproof. */
4877 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4878 pl_yylval.opval = newSVOP(OP_CONST, 0,
4879 newSViv(CopARYBASE_get(&PL_compiling)));
4880 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4886 const char tmp = *s;
4887 if (PL_lex_state == LEX_NORMAL)
4890 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4891 && intuit_more(s)) {
4893 PL_tokenbuf[0] = '@';
4894 if (ckWARN(WARN_SYNTAX)) {
4897 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4900 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4901 while (t < PL_bufend && *t != ']')
4903 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4904 "Multidimensional syntax %.*s not supported",
4905 (int)((t - PL_bufptr) + 1), PL_bufptr);
4909 else if (*s == '{') {
4911 PL_tokenbuf[0] = '%';
4912 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4913 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4915 char tmpbuf[sizeof PL_tokenbuf];
4918 } while (isSPACE(*t));
4919 if (isIDFIRST_lazy_if(t,UTF)) {
4921 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4925 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4926 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4927 "You need to quote \"%s\"",
4934 PL_expect = XOPERATOR;
4935 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4936 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4937 if (!islop || PL_last_lop_op == OP_GREPSTART)
4938 PL_expect = XOPERATOR;
4939 else if (strchr("$@\"'`q", *s))
4940 PL_expect = XTERM; /* e.g. print $fh "foo" */
4941 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4942 PL_expect = XTERM; /* e.g. print $fh &sub */
4943 else if (isIDFIRST_lazy_if(s,UTF)) {
4944 char tmpbuf[sizeof PL_tokenbuf];
4946 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4947 if ((t2 = keyword(tmpbuf, len, 0))) {
4948 /* binary operators exclude handle interpretations */
4960 PL_expect = XTERM; /* e.g. print $fh length() */
4965 PL_expect = XTERM; /* e.g. print $fh subr() */
4968 else if (isDIGIT(*s))
4969 PL_expect = XTERM; /* e.g. print $fh 3 */
4970 else if (*s == '.' && isDIGIT(s[1]))
4971 PL_expect = XTERM; /* e.g. print $fh .3 */
4972 else if ((*s == '?' || *s == '-' || *s == '+')
4973 && !isSPACE(s[1]) && s[1] != '=')
4974 PL_expect = XTERM; /* e.g. print $fh -1 */
4975 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4977 PL_expect = XTERM; /* e.g. print $fh /.../
4978 XXX except DORDOR operator
4980 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4982 PL_expect = XTERM; /* print $fh <<"EOF" */
4985 PL_pending_ident = '$';
4989 if (PL_expect == XOPERATOR)
4991 PL_tokenbuf[0] = '@';
4992 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
4993 if (!PL_tokenbuf[1]) {
4996 if (PL_lex_state == LEX_NORMAL)
4998 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5000 PL_tokenbuf[0] = '%';
5002 /* Warn about @ where they meant $. */
5003 if (*s == '[' || *s == '{') {
5004 if (ckWARN(WARN_SYNTAX)) {
5005 const char *t = s + 1;
5006 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5008 if (*t == '}' || *t == ']') {
5010 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5011 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5012 "Scalar value %.*s better written as $%.*s",
5013 (int)(t-PL_bufptr), PL_bufptr,
5014 (int)(t-PL_bufptr-1), PL_bufptr+1);
5019 PL_pending_ident = '@';
5022 case '/': /* may be division, defined-or, or pattern */
5023 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5027 case '?': /* may either be conditional or pattern */
5028 if(PL_expect == XOPERATOR) {
5036 /* A // operator. */
5046 /* Disable warning on "study /blah/" */
5047 if (PL_oldoldbufptr == PL_last_uni
5048 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5049 || memNE(PL_last_uni, "study", 5)
5050 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5053 s = scan_pat(s,OP_MATCH);
5054 TERM(sublex_start());
5058 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5059 #ifdef PERL_STRICT_CR
5062 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5064 && (s == PL_linestart || s[-1] == '\n') )
5066 PL_lex_formbrack = 0;
5070 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5076 pl_yylval.ival = OPf_SPECIAL;
5082 if (PL_expect != XOPERATOR)
5087 case '0': case '1': case '2': case '3': case '4':
5088 case '5': case '6': case '7': case '8': case '9':
5089 s = scan_num(s, &pl_yylval);
5090 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5091 if (PL_expect == XOPERATOR)
5096 s = scan_str(s,!!PL_madskills,FALSE);
5097 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5098 if (PL_expect == XOPERATOR) {
5099 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5101 deprecate_old(commaless_variable_list);
5102 return REPORT(','); /* grandfather non-comma-format format */
5109 pl_yylval.ival = OP_CONST;
5110 TERM(sublex_start());
5113 s = scan_str(s,!!PL_madskills,FALSE);
5114 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5115 if (PL_expect == XOPERATOR) {
5116 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5118 deprecate_old(commaless_variable_list);
5119 return REPORT(','); /* grandfather non-comma-format format */
5126 pl_yylval.ival = OP_CONST;
5127 /* FIXME. I think that this can be const if char *d is replaced by
5128 more localised variables. */
5129 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5130 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5131 pl_yylval.ival = OP_STRINGIFY;
5135 TERM(sublex_start());
5138 s = scan_str(s,!!PL_madskills,FALSE);
5139 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5140 if (PL_expect == XOPERATOR)
5141 no_op("Backticks",s);
5144 readpipe_override();
5145 TERM(sublex_start());
5149 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5150 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5152 if (PL_expect == XOPERATOR)
5153 no_op("Backslash",s);
5157 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5158 char *start = s + 2;
5159 while (isDIGIT(*start) || *start == '_')
5161 if (*start == '.' && isDIGIT(start[1])) {
5162 s = scan_num(s, &pl_yylval);
5165 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5166 else if (!isALPHA(*start) && (PL_expect == XTERM
5167 || PL_expect == XREF || PL_expect == XSTATE
5168 || PL_expect == XTERMORDORDOR)) {
5169 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5171 s = scan_num(s, &pl_yylval);
5178 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5220 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5222 /* Some keywords can be followed by any delimiter, including ':' */
5223 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5224 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5225 (PL_tokenbuf[0] == 'q' &&
5226 strchr("qwxr", PL_tokenbuf[1])))));
5228 /* x::* is just a word, unless x is "CORE" */
5229 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5233 while (d < PL_bufend && isSPACE(*d))
5234 d++; /* no comments skipped here, or s### is misparsed */
5236 /* Is this a label? */
5237 if (!tmp && PL_expect == XSTATE
5238 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5240 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5245 /* Check for keywords */
5246 tmp = keyword(PL_tokenbuf, len, 0);
5248 /* Is this a word before a => operator? */
5249 if (*d == '=' && d[1] == '>') {
5252 = (OP*)newSVOP(OP_CONST, 0,
5253 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5254 pl_yylval.opval->op_private = OPpCONST_BARE;
5258 if (tmp < 0) { /* second-class keyword? */
5259 GV *ogv = NULL; /* override (winner) */
5260 GV *hgv = NULL; /* hidden (loser) */
5261 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5263 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5266 if (GvIMPORTED_CV(gv))
5268 else if (! CvMETHOD(cv))
5272 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5273 (gv = *gvp) && isGV_with_GP(gv) &&
5274 GvCVu(gv) && GvIMPORTED_CV(gv))
5281 tmp = 0; /* overridden by import or by GLOBAL */
5284 && -tmp==KEY_lock /* XXX generalizable kludge */
5287 tmp = 0; /* any sub overrides "weak" keyword */
5289 else { /* no override */
5291 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5292 Perl_warner(aTHX_ packWARN(WARN_MISC),
5293 "dump() better written as CORE::dump()");
5297 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5298 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5299 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5300 "Ambiguous call resolved as CORE::%s(), %s",
5301 GvENAME(hgv), "qualify as such or use &");
5308 default: /* not a keyword */
5309 /* Trade off - by using this evil construction we can pull the
5310 variable gv into the block labelled keylookup. If not, then
5311 we have to give it function scope so that the goto from the
5312 earlier ':' case doesn't bypass the initialisation. */
5314 just_a_word_zero_gv:
5322 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5325 SV *nextPL_nextwhite = 0;
5329 /* Get the rest if it looks like a package qualifier */
5331 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5333 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5336 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5337 *s == '\'' ? "'" : "::");
5342 if (PL_expect == XOPERATOR) {
5343 if (PL_bufptr == PL_linestart) {
5344 CopLINE_dec(PL_curcop);
5345 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
5346 CopLINE_inc(PL_curcop);
5349 no_op("Bareword",s);
5352 /* Look for a subroutine with this name in current package,
5353 unless name is "Foo::", in which case Foo is a bearword
5354 (and a package name). */
5356 if (len > 2 && !PL_madskills &&
5357 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5359 if (ckWARN(WARN_BAREWORD)
5360 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5361 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5362 "Bareword \"%s\" refers to nonexistent package",
5365 PL_tokenbuf[len] = '\0';
5371 /* Mustn't actually add anything to a symbol table.
5372 But also don't want to "initialise" any placeholder
5373 constants that might already be there into full
5374 blown PVGVs with attached PVCV. */
5375 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5376 GV_NOADD_NOINIT, SVt_PVCV);
5381 /* if we saw a global override before, get the right name */
5384 sv = newSVpvs("CORE::GLOBAL::");
5385 sv_catpv(sv,PL_tokenbuf);
5388 /* If len is 0, newSVpv does strlen(), which is correct.
5389 If len is non-zero, then it will be the true length,
5390 and so the scalar will be created correctly. */
5391 sv = newSVpv(PL_tokenbuf,len);
5394 if (PL_madskills && !PL_thistoken) {
5395 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5396 PL_thistoken = newSVpvn(start,s - start);
5397 PL_realtokenstart = s - SvPVX(PL_linestr);
5401 /* Presume this is going to be a bareword of some sort. */
5404 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5405 pl_yylval.opval->op_private = OPpCONST_BARE;
5406 /* UTF-8 package name? */
5407 if (UTF && !IN_BYTES &&
5408 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5411 /* And if "Foo::", then that's what it certainly is. */
5416 /* Do the explicit type check so that we don't need to force
5417 the initialisation of the symbol table to have a real GV.
5418 Beware - gv may not really be a PVGV, cv may not really be
5419 a PVCV, (because of the space optimisations that gv_init
5420 understands) But they're true if for this symbol there is
5421 respectively a typeglob and a subroutine.
5423 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5424 /* Real typeglob, so get the real subroutine: */
5426 /* A proxy for a subroutine in this package? */
5427 : SvOK(gv) ? (CV *) gv : NULL)
5430 /* See if it's the indirect object for a list operator. */
5432 if (PL_oldoldbufptr &&
5433 PL_oldoldbufptr < PL_bufptr &&
5434 (PL_oldoldbufptr == PL_last_lop
5435 || PL_oldoldbufptr == PL_last_uni) &&
5436 /* NO SKIPSPACE BEFORE HERE! */
5437 (PL_expect == XREF ||
5438 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5440 bool immediate_paren = *s == '(';
5442 /* (Now we can afford to cross potential line boundary.) */
5443 s = SKIPSPACE2(s,nextPL_nextwhite);
5445 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5448 /* Two barewords in a row may indicate method call. */
5450 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5451 (tmp = intuit_method(s, gv, cv)))
5454 /* If not a declared subroutine, it's an indirect object. */
5455 /* (But it's an indir obj regardless for sort.) */
5456 /* Also, if "_" follows a filetest operator, it's a bareword */
5459 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5461 (PL_last_lop_op != OP_MAPSTART &&
5462 PL_last_lop_op != OP_GREPSTART))))
5463 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5464 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5467 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5472 PL_expect = XOPERATOR;
5475 s = SKIPSPACE2(s,nextPL_nextwhite);
5476 PL_nextwhite = nextPL_nextwhite;
5481 /* Is this a word before a => operator? */
5482 if (*s == '=' && s[1] == '>' && !pkgname) {
5484 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5485 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5486 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5490 /* If followed by a paren, it's certainly a subroutine. */
5495 while (SPACE_OR_TAB(*d))
5497 if (*d == ')' && (sv = gv_const_sv(gv))) {
5504 PL_nextwhite = PL_thiswhite;
5507 start_force(PL_curforce);
5509 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5510 PL_expect = XOPERATOR;
5513 PL_nextwhite = nextPL_nextwhite;
5514 curmad('X', PL_thistoken);
5515 PL_thistoken = newSVpvs("");
5523 /* If followed by var or block, call it a method (unless sub) */
5525 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5526 PL_last_lop = PL_oldbufptr;
5527 PL_last_lop_op = OP_METHOD;
5531 /* If followed by a bareword, see if it looks like indir obj. */
5534 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5535 && (tmp = intuit_method(s, gv, cv)))
5538 /* Not a method, so call it a subroutine (if defined) */
5541 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5542 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5543 "Ambiguous use of -%s resolved as -&%s()",
5544 PL_tokenbuf, PL_tokenbuf);
5545 /* Check for a constant sub */
5546 if ((sv = gv_const_sv(gv))) {
5548 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5549 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5550 pl_yylval.opval->op_private = 0;
5554 /* Resolve to GV now. */
5555 if (SvTYPE(gv) != SVt_PVGV) {
5556 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5557 assert (SvTYPE(gv) == SVt_PVGV);
5558 /* cv must have been some sort of placeholder, so
5559 now needs replacing with a real code reference. */
5563 op_free(pl_yylval.opval);
5564 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5565 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5566 PL_last_lop = PL_oldbufptr;
5567 PL_last_lop_op = OP_ENTERSUB;
5568 /* Is there a prototype? */
5576 const char *proto = SvPV_const((SV*)cv, protolen);
5579 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5581 while (*proto == ';')
5583 if (*proto == '&' && *s == '{') {
5585 sv_setpvs(PL_subname, "__ANON__");
5587 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5594 PL_nextwhite = PL_thiswhite;
5597 start_force(PL_curforce);
5598 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5601 PL_nextwhite = nextPL_nextwhite;
5602 curmad('X', PL_thistoken);
5603 PL_thistoken = newSVpvs("");
5610 /* Guess harder when madskills require "best effort". */
5611 if (PL_madskills && (!gv || !GvCVu(gv))) {
5612 int probable_sub = 0;
5613 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5615 else if (isALPHA(*s)) {
5619 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5620 if (!keyword(tmpbuf, tmplen, 0))
5623 while (d < PL_bufend && isSPACE(*d))
5625 if (*d == '=' && d[1] == '>')
5630 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5631 op_free(pl_yylval.opval);
5632 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5633 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5634 PL_last_lop = PL_oldbufptr;
5635 PL_last_lop_op = OP_ENTERSUB;
5636 PL_nextwhite = PL_thiswhite;
5638 start_force(PL_curforce);
5639 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5641 PL_nextwhite = nextPL_nextwhite;
5642 curmad('X', PL_thistoken);
5643 PL_thistoken = newSVpvs("");
5648 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5655 /* Call it a bare word */
5657 if (PL_hints & HINT_STRICT_SUBS)
5658 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5661 if (lastchar != '-') {
5662 if (ckWARN(WARN_RESERVED)) {
5666 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5667 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5674 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5675 && ckWARN_d(WARN_AMBIGUOUS)) {
5676 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5677 "Operator or semicolon missing before %c%s",
5678 lastchar, PL_tokenbuf);
5679 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5680 "Ambiguous use of %c resolved as operator %c",
5681 lastchar, lastchar);
5687 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5688 newSVpv(CopFILE(PL_curcop),0));
5692 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5693 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5696 case KEY___PACKAGE__:
5697 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5699 ? newSVhek(HvNAME_HEK(PL_curstash))
5706 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5707 const char *pname = "main";
5708 if (PL_tokenbuf[2] == 'D')
5709 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5710 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5714 GvIOp(gv) = newIO();
5715 IoIFP(GvIOp(gv)) = PL_rsfp;
5716 #if defined(HAS_FCNTL) && defined(F_SETFD)
5718 const int fd = PerlIO_fileno(PL_rsfp);
5719 fcntl(fd,F_SETFD,fd >= 3);
5722 /* Mark this internal pseudo-handle as clean */
5723 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5724 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5725 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5727 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5728 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5729 /* if the script was opened in binmode, we need to revert
5730 * it to text mode for compatibility; but only iff it has CRs
5731 * XXX this is a questionable hack at best. */
5732 if (PL_bufend-PL_bufptr > 2
5733 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5736 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5737 loc = PerlIO_tell(PL_rsfp);
5738 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5741 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5743 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5744 #endif /* NETWARE */
5745 #ifdef PERLIO_IS_STDIO /* really? */
5746 # if defined(__BORLANDC__)
5747 /* XXX see note in do_binmode() */
5748 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5752 PerlIO_seek(PL_rsfp, loc, 0);
5756 #ifdef PERLIO_LAYERS
5759 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5760 else if (PL_encoding) {
5767 XPUSHs(PL_encoding);
5769 call_method("name", G_SCALAR);
5773 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5774 Perl_form(aTHX_ ":encoding(%"SVf")",
5783 if (PL_realtokenstart >= 0) {
5784 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5786 PL_endwhite = newSVpvs("");
5787 sv_catsv(PL_endwhite, PL_thiswhite);
5789 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5790 PL_realtokenstart = -1;
5792 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5793 SvCUR(PL_endwhite))) != NULL) ;
5808 if (PL_expect == XSTATE) {
5815 if (*s == ':' && s[1] == ':') {
5818 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5819 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5820 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5823 else if (tmp == KEY_require || tmp == KEY_do)
5824 /* that's a way to remember we saw "CORE::" */
5837 LOP(OP_ACCEPT,XTERM);
5843 LOP(OP_ATAN2,XTERM);
5849 LOP(OP_BINMODE,XTERM);
5852 LOP(OP_BLESS,XTERM);
5861 /* When 'use switch' is in effect, continue has a dual
5862 life as a control operator. */
5864 if (!FEATURE_IS_ENABLED("switch"))
5867 /* We have to disambiguate the two senses of
5868 "continue". If the next token is a '{' then
5869 treat it as the start of a continue block;
5870 otherwise treat it as a control operator.
5882 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5899 if (!PL_cryptseen) {
5900 PL_cryptseen = TRUE;
5904 LOP(OP_CRYPT,XTERM);
5907 LOP(OP_CHMOD,XTERM);
5910 LOP(OP_CHOWN,XTERM);
5913 LOP(OP_CONNECT,XTERM);
5932 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5933 if (orig_keyword == KEY_do) {
5942 PL_hints |= HINT_BLOCK_SCOPE;
5952 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5953 LOP(OP_DBMOPEN,XTERM);
5959 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5966 pl_yylval.ival = CopLINE(PL_curcop);
5982 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
5983 UNIBRACK(OP_ENTEREVAL);
5997 case KEY_endhostent:
6003 case KEY_endservent:
6006 case KEY_endprotoent:
6017 pl_yylval.ival = CopLINE(PL_curcop);
6019 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6022 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6025 if ((PL_bufend - p) >= 3 &&
6026 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6028 else if ((PL_bufend - p) >= 4 &&
6029 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6032 if (isIDFIRST_lazy_if(p,UTF)) {
6033 p = scan_ident(p, PL_bufend,
6034 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6038 Perl_croak(aTHX_ "Missing $ on loop variable");
6040 s = SvPVX(PL_linestr) + soff;
6046 LOP(OP_FORMLINE,XTERM);
6052 LOP(OP_FCNTL,XTERM);
6058 LOP(OP_FLOCK,XTERM);
6067 LOP(OP_GREPSTART, XREF);
6070 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6085 case KEY_getpriority:
6086 LOP(OP_GETPRIORITY,XTERM);
6088 case KEY_getprotobyname:
6091 case KEY_getprotobynumber:
6092 LOP(OP_GPBYNUMBER,XTERM);
6094 case KEY_getprotoent:
6106 case KEY_getpeername:
6107 UNI(OP_GETPEERNAME);
6109 case KEY_gethostbyname:
6112 case KEY_gethostbyaddr:
6113 LOP(OP_GHBYADDR,XTERM);
6115 case KEY_gethostent:
6118 case KEY_getnetbyname:
6121 case KEY_getnetbyaddr:
6122 LOP(OP_GNBYADDR,XTERM);
6127 case KEY_getservbyname:
6128 LOP(OP_GSBYNAME,XTERM);
6130 case KEY_getservbyport:
6131 LOP(OP_GSBYPORT,XTERM);
6133 case KEY_getservent:
6136 case KEY_getsockname:
6137 UNI(OP_GETSOCKNAME);
6139 case KEY_getsockopt:
6140 LOP(OP_GSOCKOPT,XTERM);
6155 pl_yylval.ival = CopLINE(PL_curcop);
6165 pl_yylval.ival = CopLINE(PL_curcop);
6169 LOP(OP_INDEX,XTERM);
6175 LOP(OP_IOCTL,XTERM);
6187 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6219 LOP(OP_LISTEN,XTERM);
6228 s = scan_pat(s,OP_MATCH);
6229 TERM(sublex_start());
6232 LOP(OP_MAPSTART, XREF);
6235 LOP(OP_MKDIR,XTERM);
6238 LOP(OP_MSGCTL,XTERM);
6241 LOP(OP_MSGGET,XTERM);
6244 LOP(OP_MSGRCV,XTERM);
6247 LOP(OP_MSGSND,XTERM);
6252 PL_in_my = (U16)tmp;
6254 if (isIDFIRST_lazy_if(s,UTF)) {
6258 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6259 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6261 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6262 if (!PL_in_my_stash) {
6265 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6269 if (PL_madskills) { /* just add type to declarator token */
6270 sv_catsv(PL_thistoken, PL_nextwhite);
6272 sv_catpvn(PL_thistoken, start, s - start);
6280 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6287 s = tokenize_use(0, s);
6291 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6298 if (isIDFIRST_lazy_if(s,UTF)) {
6300 for (d = s; isALNUM_lazy_if(d,UTF);)
6302 for (t=d; isSPACE(*t);)
6304 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6306 && !(t[0] == '=' && t[1] == '>')
6308 int parms_len = (int)(d-s);
6309 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6310 "Precedence problem: open %.*s should be open(%.*s)",
6311 parms_len, s, parms_len, s);
6317 pl_yylval.ival = OP_OR;
6327 LOP(OP_OPEN_DIR,XTERM);
6330 checkcomma(s,PL_tokenbuf,"filehandle");
6334 checkcomma(s,PL_tokenbuf,"filehandle");
6353 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6357 LOP(OP_PIPE_OP,XTERM);
6360 s = scan_str(s,!!PL_madskills,FALSE);
6363 pl_yylval.ival = OP_CONST;
6364 TERM(sublex_start());
6370 s = scan_str(s,!!PL_madskills,FALSE);
6373 PL_expect = XOPERATOR;
6375 if (SvCUR(PL_lex_stuff)) {
6378 d = SvPV_force(PL_lex_stuff, len);
6380 for (; isSPACE(*d) && len; --len, ++d)
6385 if (!warned && ckWARN(WARN_QW)) {
6386 for (; !isSPACE(*d) && len; --len, ++d) {
6388 Perl_warner(aTHX_ packWARN(WARN_QW),
6389 "Possible attempt to separate words with commas");
6392 else if (*d == '#') {
6393 Perl_warner(aTHX_ packWARN(WARN_QW),
6394 "Possible attempt to put comments in qw() list");
6400 for (; !isSPACE(*d) && len; --len, ++d)
6403 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6404 words = append_elem(OP_LIST, words,
6405 newSVOP(OP_CONST, 0, tokeq(sv)));
6409 start_force(PL_curforce);
6410 NEXTVAL_NEXTTOKE.opval = words;
6415 SvREFCNT_dec(PL_lex_stuff);
6416 PL_lex_stuff = NULL;
6422 s = scan_str(s,!!PL_madskills,FALSE);
6425 pl_yylval.ival = OP_STRINGIFY;
6426 if (SvIVX(PL_lex_stuff) == '\'')
6427 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6428 TERM(sublex_start());
6431 s = scan_pat(s,OP_QR);
6432 TERM(sublex_start());
6435 s = scan_str(s,!!PL_madskills,FALSE);
6438 readpipe_override();
6439 TERM(sublex_start());
6447 s = force_version(s, FALSE);
6449 else if (*s != 'v' || !isDIGIT(s[1])
6450 || (s = force_version(s, TRUE), *s == 'v'))
6452 *PL_tokenbuf = '\0';
6453 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6454 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6455 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6457 yyerror("<> should be quotes");
6459 if (orig_keyword == KEY_require) {
6467 PL_last_uni = PL_oldbufptr;
6468 PL_last_lop_op = OP_REQUIRE;
6470 return REPORT( (int)REQUIRE );
6476 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6480 LOP(OP_RENAME,XTERM);
6489 LOP(OP_RINDEX,XTERM);
6498 UNIDOR(OP_READLINE);
6501 UNIDOR(OP_BACKTICK);
6510 LOP(OP_REVERSE,XTERM);
6513 UNIDOR(OP_READLINK);
6520 if (pl_yylval.opval)
6521 TERM(sublex_start());
6523 TOKEN(1); /* force error */
6526 checkcomma(s,PL_tokenbuf,"filehandle");
6536 LOP(OP_SELECT,XTERM);
6542 LOP(OP_SEMCTL,XTERM);
6545 LOP(OP_SEMGET,XTERM);
6548 LOP(OP_SEMOP,XTERM);
6554 LOP(OP_SETPGRP,XTERM);
6556 case KEY_setpriority:
6557 LOP(OP_SETPRIORITY,XTERM);
6559 case KEY_sethostent:
6565 case KEY_setservent:
6568 case KEY_setprotoent:
6578 LOP(OP_SEEKDIR,XTERM);
6580 case KEY_setsockopt:
6581 LOP(OP_SSOCKOPT,XTERM);
6587 LOP(OP_SHMCTL,XTERM);
6590 LOP(OP_SHMGET,XTERM);
6593 LOP(OP_SHMREAD,XTERM);
6596 LOP(OP_SHMWRITE,XTERM);
6599 LOP(OP_SHUTDOWN,XTERM);
6608 LOP(OP_SOCKET,XTERM);
6610 case KEY_socketpair:
6611 LOP(OP_SOCKPAIR,XTERM);
6614 checkcomma(s,PL_tokenbuf,"subroutine name");
6616 if (*s == ';' || *s == ')') /* probably a close */
6617 Perl_croak(aTHX_ "sort is now a reserved word");
6619 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6623 LOP(OP_SPLIT,XTERM);
6626 LOP(OP_SPRINTF,XTERM);
6629 LOP(OP_SPLICE,XTERM);
6644 LOP(OP_SUBSTR,XTERM);
6650 char tmpbuf[sizeof PL_tokenbuf];
6651 SSize_t tboffset = 0;
6652 expectation attrful;
6653 bool have_name, have_proto;
6654 const int key = tmp;
6659 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6660 SV *subtoken = newSVpvn(tstart, s - tstart);
6664 s = SKIPSPACE2(s,tmpwhite);
6669 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6670 (*s == ':' && s[1] == ':'))
6673 SV *nametoke = NULL;
6677 attrful = XATTRBLOCK;
6678 /* remember buffer pos'n for later force_word */
6679 tboffset = s - PL_oldbufptr;
6680 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6683 nametoke = newSVpvn(s, d - s);
6685 if (memchr(tmpbuf, ':', len))
6686 sv_setpvn(PL_subname, tmpbuf, len);
6688 sv_setsv(PL_subname,PL_curstname);
6689 sv_catpvs(PL_subname,"::");
6690 sv_catpvn(PL_subname,tmpbuf,len);
6697 CURMAD('X', nametoke);
6698 CURMAD('_', tmpwhite);
6699 (void) force_word(PL_oldbufptr + tboffset, WORD,
6702 s = SKIPSPACE2(d,tmpwhite);
6709 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6710 PL_expect = XTERMBLOCK;
6711 attrful = XATTRTERM;
6712 sv_setpvn(PL_subname,"?",1);
6716 if (key == KEY_format) {
6718 PL_lex_formbrack = PL_lex_brackets + 1;
6720 PL_thistoken = subtoken;
6724 (void) force_word(PL_oldbufptr + tboffset, WORD,
6730 /* Look for a prototype */
6733 bool bad_proto = FALSE;
6734 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6736 s = scan_str(s,!!PL_madskills,FALSE);
6738 Perl_croak(aTHX_ "Prototype not terminated");
6739 /* strip spaces and check for bad characters */
6740 d = SvPVX(PL_lex_stuff);
6742 for (p = d; *p; ++p) {
6745 if (warnsyntax && !strchr("$@%*;[]&\\_", *p))
6751 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6752 "Illegal character in prototype for %"SVf" : %s",
6753 SVfARG(PL_subname), d);
6754 SvCUR_set(PL_lex_stuff, tmp);
6759 CURMAD('q', PL_thisopen);
6760 CURMAD('_', tmpwhite);
6761 CURMAD('=', PL_thisstuff);
6762 CURMAD('Q', PL_thisclose);
6763 NEXTVAL_NEXTTOKE.opval =
6764 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6765 PL_lex_stuff = NULL;
6768 s = SKIPSPACE2(s,tmpwhite);
6776 if (*s == ':' && s[1] != ':')
6777 PL_expect = attrful;
6778 else if (*s != '{' && key == KEY_sub) {
6780 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6782 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6789 curmad('^', newSVpvs(""));
6790 CURMAD('_', tmpwhite);
6794 PL_thistoken = subtoken;
6797 NEXTVAL_NEXTTOKE.opval =
6798 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6799 PL_lex_stuff = NULL;
6805 sv_setpvs(PL_subname, "__ANON__");
6807 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6811 (void) force_word(PL_oldbufptr + tboffset, WORD,
6820 LOP(OP_SYSTEM,XREF);
6823 LOP(OP_SYMLINK,XTERM);
6826 LOP(OP_SYSCALL,XTERM);
6829 LOP(OP_SYSOPEN,XTERM);
6832 LOP(OP_SYSSEEK,XTERM);
6835 LOP(OP_SYSREAD,XTERM);
6838 LOP(OP_SYSWRITE,XTERM);
6842 TERM(sublex_start());
6863 LOP(OP_TRUNCATE,XTERM);
6875 pl_yylval.ival = CopLINE(PL_curcop);
6879 pl_yylval.ival = CopLINE(PL_curcop);
6883 LOP(OP_UNLINK,XTERM);
6889 LOP(OP_UNPACK,XTERM);
6892 LOP(OP_UTIME,XTERM);
6898 LOP(OP_UNSHIFT,XTERM);
6901 s = tokenize_use(1, s);
6911 pl_yylval.ival = CopLINE(PL_curcop);
6915 pl_yylval.ival = CopLINE(PL_curcop);
6919 PL_hints |= HINT_BLOCK_SCOPE;
6926 LOP(OP_WAITPID,XTERM);
6935 ctl_l[0] = toCTRL('L');
6937 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6940 /* Make sure $^L is defined */
6941 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
6946 if (PL_expect == XOPERATOR)
6952 pl_yylval.ival = OP_XOR;
6957 TERM(sublex_start());
6962 #pragma segment Main
6966 S_pending_ident(pTHX)
6971 /* pit holds the identifier we read and pending_ident is reset */
6972 char pit = PL_pending_ident;
6973 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
6974 /* All routes through this function want to know if there is a colon. */
6975 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
6976 PL_pending_ident = 0;
6978 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
6979 DEBUG_T({ PerlIO_printf(Perl_debug_log,
6980 "### Pending identifier '%s'\n", PL_tokenbuf); });
6982 /* if we're in a my(), we can't allow dynamics here.
6983 $foo'bar has already been turned into $foo::bar, so
6984 just check for colons.
6986 if it's a legal name, the OP is a PADANY.
6989 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
6991 yyerror(Perl_form(aTHX_ "No package name allowed for "
6992 "variable %s in \"our\"",
6994 tmp = allocmy(PL_tokenbuf);
6998 yyerror(Perl_form(aTHX_ PL_no_myglob,
6999 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7001 pl_yylval.opval = newOP(OP_PADANY, 0);
7002 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7008 build the ops for accesses to a my() variable.
7010 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7011 then used in a comparison. This catches most, but not
7012 all cases. For instance, it catches
7013 sort { my($a); $a <=> $b }
7015 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7016 (although why you'd do that is anyone's guess).
7021 tmp = pad_findmy(PL_tokenbuf);
7022 if (tmp != NOT_IN_PAD) {
7023 /* might be an "our" variable" */
7024 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7025 /* build ops for a bareword */
7026 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7027 HEK * const stashname = HvNAME_HEK(stash);
7028 SV * const sym = newSVhek(stashname);
7029 sv_catpvs(sym, "::");
7030 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7031 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7032 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7035 ? (GV_ADDMULTI | GV_ADDINEVAL)
7038 ((PL_tokenbuf[0] == '$') ? SVt_PV
7039 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7044 /* if it's a sort block and they're naming $a or $b */
7045 if (PL_last_lop_op == OP_SORT &&
7046 PL_tokenbuf[0] == '$' &&
7047 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7050 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7051 d < PL_bufend && *d != '\n';
7054 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7055 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7061 pl_yylval.opval = newOP(OP_PADANY, 0);
7062 pl_yylval.opval->op_targ = tmp;
7068 Whine if they've said @foo in a doublequoted string,
7069 and @foo isn't a variable we can find in the symbol
7072 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7073 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7075 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7076 && ckWARN(WARN_AMBIGUOUS)
7077 /* DO NOT warn for @- and @+ */
7078 && !( PL_tokenbuf[2] == '\0' &&
7079 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7082 /* Downgraded from fatal to warning 20000522 mjd */
7083 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7084 "Possible unintended interpolation of %s in string",
7089 /* build ops for a bareword */
7090 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7092 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7094 PL_tokenbuf + 1, tokenbuf_len - 1,
7095 /* If the identifier refers to a stash, don't autovivify it.
7096 * Change 24660 had the side effect of causing symbol table
7097 * hashes to always be defined, even if they were freshly
7098 * created and the only reference in the entire program was
7099 * the single statement with the defined %foo::bar:: test.
7100 * It appears that all code in the wild doing this actually
7101 * wants to know whether sub-packages have been loaded, so
7102 * by avoiding auto-vivifying symbol tables, we ensure that
7103 * defined %foo::bar:: continues to be false, and the existing
7104 * tests still give the expected answers, even though what
7105 * they're actually testing has now changed subtly.
7107 (*PL_tokenbuf == '%'
7108 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7111 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7112 ((PL_tokenbuf[0] == '$') ? SVt_PV
7113 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7119 * The following code was generated by perl_keyword.pl.
7123 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7127 PERL_ARGS_ASSERT_KEYWORD;
7131 case 1: /* 5 tokens of length 1 */
7163 case 2: /* 18 tokens of length 2 */
7309 case 3: /* 29 tokens of length 3 */
7313 if (name[1] == 'N' &&
7376 if (name[1] == 'i' &&
7408 if (name[1] == 'o' &&
7417 if (name[1] == 'e' &&
7426 if (name[1] == 'n' &&
7435 if (name[1] == 'o' &&
7444 if (name[1] == 'a' &&
7453 if (name[1] == 'o' &&
7515 if (name[1] == 'e' &&
7529 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7555 if (name[1] == 'i' &&
7564 if (name[1] == 's' &&
7573 if (name[1] == 'e' &&
7582 if (name[1] == 'o' &&
7594 case 4: /* 41 tokens of length 4 */
7598 if (name[1] == 'O' &&
7608 if (name[1] == 'N' &&
7618 if (name[1] == 'i' &&
7628 if (name[1] == 'h' &&
7638 if (name[1] == 'u' &&
7651 if (name[2] == 'c' &&
7660 if (name[2] == 's' &&
7669 if (name[2] == 'a' &&
7705 if (name[1] == 'o' &&
7718 if (name[2] == 't' &&
7727 if (name[2] == 'o' &&
7736 if (name[2] == 't' &&
7745 if (name[2] == 'e' &&
7758 if (name[1] == 'o' &&
7771 if (name[2] == 'y' &&
7780 if (name[2] == 'l' &&
7796 if (name[2] == 's' &&
7805 if (name[2] == 'n' &&
7814 if (name[2] == 'c' &&
7827 if (name[1] == 'e' &&
7837 if (name[1] == 'p' &&
7850 if (name[2] == 'c' &&
7859 if (name[2] == 'p' &&
7868 if (name[2] == 's' &&
7884 if (name[2] == 'n' &&
7954 if (name[2] == 'r' &&
7963 if (name[2] == 'r' &&
7972 if (name[2] == 'a' &&
7988 if (name[2] == 'l' &&
8050 if (name[2] == 'e' &&
8053 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8066 case 5: /* 39 tokens of length 5 */
8070 if (name[1] == 'E' &&
8081 if (name[1] == 'H' &&
8095 if (name[2] == 'a' &&
8105 if (name[2] == 'a' &&
8122 if (name[2] == 'e' &&
8132 if (name[2] == 'e' &&
8136 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8152 if (name[3] == 'i' &&
8161 if (name[3] == 'o' &&
8197 if (name[2] == 'o' &&
8207 if (name[2] == 'y' &&
8221 if (name[1] == 'l' &&
8235 if (name[2] == 'n' &&
8245 if (name[2] == 'o' &&
8259 if (name[1] == 'i' &&
8264 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8273 if (name[2] == 'd' &&
8283 if (name[2] == 'c' &&
8300 if (name[2] == 'c' &&
8310 if (name[2] == 't' &&
8324 if (name[1] == 'k' &&
8335 if (name[1] == 'r' &&
8349 if (name[2] == 's' &&
8359 if (name[2] == 'd' &&
8376 if (name[2] == 'm' &&
8386 if (name[2] == 'i' &&
8396 if (name[2] == 'e' &&
8406 if (name[2] == 'l' &&
8416 if (name[2] == 'a' &&
8429 if (name[3] == 't' &&
8432 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8438 if (name[3] == 'd' &&
8455 if (name[1] == 'i' &&
8469 if (name[2] == 'a' &&
8482 if (name[3] == 'e' &&
8517 if (name[2] == 'i' &&
8534 if (name[2] == 'i' &&
8544 if (name[2] == 'i' &&
8561 case 6: /* 33 tokens of length 6 */
8565 if (name[1] == 'c' &&
8580 if (name[2] == 'l' &&
8591 if (name[2] == 'r' &&
8606 if (name[1] == 'e' &&
8621 if (name[2] == 's' &&
8626 if(ckWARN_d(WARN_SYNTAX))
8627 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8633 if (name[2] == 'i' &&
8651 if (name[2] == 'l' &&
8662 if (name[2] == 'r' &&
8677 if (name[1] == 'm' &&
8692 if (name[2] == 'n' &&
8703 if (name[2] == 's' &&
8718 if (name[1] == 's' &&
8724 if (name[4] == 't' &&
8733 if (name[4] == 'e' &&
8742 if (name[4] == 'c' &&
8751 if (name[4] == 'n' &&
8767 if (name[1] == 'r' &&
8785 if (name[3] == 'a' &&
8795 if (name[3] == 'u' &&
8809 if (name[2] == 'n' &&
8827 if (name[2] == 'a' &&
8841 if (name[3] == 'e' &&
8854 if (name[4] == 't' &&
8863 if (name[4] == 'e' &&
8885 if (name[4] == 't' &&
8894 if (name[4] == 'e' &&
8910 if (name[2] == 'c' &&
8921 if (name[2] == 'l' &&
8932 if (name[2] == 'b' &&
8943 if (name[2] == 's' &&
8966 if (name[4] == 's' &&
8975 if (name[4] == 'n' &&
8988 if (name[3] == 'a' &&
9005 if (name[1] == 'a' &&
9020 case 7: /* 29 tokens of length 7 */
9024 if (name[1] == 'E' &&
9037 if (name[1] == '_' &&
9050 if (name[1] == 'i' &&
9057 return -KEY_binmode;
9063 if (name[1] == 'o' &&
9070 return -KEY_connect;
9079 if (name[2] == 'm' &&
9085 return -KEY_dbmopen;
9096 if (name[4] == 'u' &&
9100 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9106 if (name[4] == 'n' &&
9127 if (name[1] == 'o' &&
9140 if (name[1] == 'e' &&
9147 if (name[5] == 'r' &&
9150 return -KEY_getpgrp;
9156 if (name[5] == 'i' &&
9159 return -KEY_getppid;
9172 if (name[1] == 'c' &&
9179 return -KEY_lcfirst;
9185 if (name[1] == 'p' &&
9192 return -KEY_opendir;
9198 if (name[1] == 'a' &&
9216 if (name[3] == 'd' &&
9221 return -KEY_readdir;
9227 if (name[3] == 'u' &&
9238 if (name[3] == 'e' &&
9243 return -KEY_reverse;
9262 if (name[3] == 'k' &&
9267 return -KEY_seekdir;
9273 if (name[3] == 'p' &&
9278 return -KEY_setpgrp;
9288 if (name[2] == 'm' &&
9294 return -KEY_shmread;
9300 if (name[2] == 'r' &&
9306 return -KEY_sprintf;
9315 if (name[3] == 'l' &&
9320 return -KEY_symlink;
9329 if (name[4] == 'a' &&
9333 return -KEY_syscall;
9339 if (name[4] == 'p' &&
9343 return -KEY_sysopen;
9349 if (name[4] == 'e' &&
9353 return -KEY_sysread;
9359 if (name[4] == 'e' &&
9363 return -KEY_sysseek;
9381 if (name[1] == 'e' &&
9388 return -KEY_telldir;
9397 if (name[2] == 'f' &&
9403 return -KEY_ucfirst;
9409 if (name[2] == 's' &&
9415 return -KEY_unshift;
9425 if (name[1] == 'a' &&
9432 return -KEY_waitpid;
9441 case 8: /* 26 tokens of length 8 */
9445 if (name[1] == 'U' &&
9453 return KEY_AUTOLOAD;
9464 if (name[3] == 'A' &&
9470 return KEY___DATA__;
9476 if (name[3] == 'I' &&
9482 return -KEY___FILE__;
9488 if (name[3] == 'I' &&
9494 return -KEY___LINE__;
9510 if (name[2] == 'o' &&
9517 return -KEY_closedir;
9523 if (name[2] == 'n' &&
9530 return -KEY_continue;
9540 if (name[1] == 'b' &&
9548 return -KEY_dbmclose;
9554 if (name[1] == 'n' &&
9560 if (name[4] == 'r' &&
9565 return -KEY_endgrent;
9571 if (name[4] == 'w' &&
9576 return -KEY_endpwent;
9589 if (name[1] == 'o' &&
9597 return -KEY_formline;
9603 if (name[1] == 'e' &&
9614 if (name[6] == 'n' &&
9617 return -KEY_getgrent;
9623 if (name[6] == 'i' &&
9626 return -KEY_getgrgid;
9632 if (name[6] == 'a' &&
9635 return -KEY_getgrnam;
9648 if (name[4] == 'o' &&
9653 return -KEY_getlogin;
9664 if (name[6] == 'n' &&
9667 return -KEY_getpwent;
9673 if (name[6] == 'a' &&
9676 return -KEY_getpwnam;
9682 if (name[6] == 'i' &&
9685 return -KEY_getpwuid;
9705 if (name[1] == 'e' &&
9712 if (name[5] == 'i' &&
9719 return -KEY_readline;
9724 return -KEY_readlink;
9735 if (name[5] == 'i' &&
9739 return -KEY_readpipe;
9760 if (name[4] == 'r' &&
9765 return -KEY_setgrent;
9771 if (name[4] == 'w' &&
9776 return -KEY_setpwent;
9792 if (name[3] == 'w' &&
9798 return -KEY_shmwrite;
9804 if (name[3] == 't' &&
9810 return -KEY_shutdown;
9820 if (name[2] == 's' &&
9827 return -KEY_syswrite;
9837 if (name[1] == 'r' &&
9845 return -KEY_truncate;
9854 case 9: /* 9 tokens of length 9 */
9858 if (name[1] == 'N' &&
9867 return KEY_UNITCHECK;
9873 if (name[1] == 'n' &&
9882 return -KEY_endnetent;
9888 if (name[1] == 'e' &&
9897 return -KEY_getnetent;
9903 if (name[1] == 'o' &&
9912 return -KEY_localtime;
9918 if (name[1] == 'r' &&
9927 return KEY_prototype;
9933 if (name[1] == 'u' &&
9942 return -KEY_quotemeta;
9948 if (name[1] == 'e' &&
9957 return -KEY_rewinddir;
9963 if (name[1] == 'e' &&
9972 return -KEY_setnetent;
9978 if (name[1] == 'a' &&
9987 return -KEY_wantarray;
9996 case 10: /* 9 tokens of length 10 */
10000 if (name[1] == 'n' &&
10006 if (name[4] == 'o' &&
10013 return -KEY_endhostent;
10019 if (name[4] == 'e' &&
10026 return -KEY_endservent;
10039 if (name[1] == 'e' &&
10045 if (name[4] == 'o' &&
10052 return -KEY_gethostent;
10061 if (name[5] == 'r' &&
10067 return -KEY_getservent;
10073 if (name[5] == 'c' &&
10079 return -KEY_getsockopt;
10099 if (name[2] == 't')
10104 if (name[4] == 'o' &&
10111 return -KEY_sethostent;
10120 if (name[5] == 'r' &&
10126 return -KEY_setservent;
10132 if (name[5] == 'c' &&
10138 return -KEY_setsockopt;
10155 if (name[2] == 'c' &&
10164 return -KEY_socketpair;
10177 case 11: /* 8 tokens of length 11 */
10181 if (name[1] == '_' &&
10191 { /* __PACKAGE__ */
10192 return -KEY___PACKAGE__;
10198 if (name[1] == 'n' &&
10208 { /* endprotoent */
10209 return -KEY_endprotoent;
10215 if (name[1] == 'e' &&
10224 if (name[5] == 'e' &&
10230 { /* getpeername */
10231 return -KEY_getpeername;
10240 if (name[6] == 'o' &&
10245 { /* getpriority */
10246 return -KEY_getpriority;
10252 if (name[6] == 't' &&
10257 { /* getprotoent */
10258 return -KEY_getprotoent;
10272 if (name[4] == 'o' &&
10279 { /* getsockname */
10280 return -KEY_getsockname;
10293 if (name[1] == 'e' &&
10301 if (name[6] == 'o' &&
10306 { /* setpriority */
10307 return -KEY_setpriority;
10313 if (name[6] == 't' &&
10318 { /* setprotoent */
10319 return -KEY_setprotoent;
10335 case 12: /* 2 tokens of length 12 */
10336 if (name[0] == 'g' &&
10348 if (name[9] == 'd' &&
10351 { /* getnetbyaddr */
10352 return -KEY_getnetbyaddr;
10358 if (name[9] == 'a' &&
10361 { /* getnetbyname */
10362 return -KEY_getnetbyname;
10374 case 13: /* 4 tokens of length 13 */
10375 if (name[0] == 'g' &&
10382 if (name[4] == 'o' &&
10391 if (name[10] == 'd' &&
10394 { /* gethostbyaddr */
10395 return -KEY_gethostbyaddr;
10401 if (name[10] == 'a' &&
10404 { /* gethostbyname */
10405 return -KEY_gethostbyname;
10418 if (name[4] == 'e' &&
10427 if (name[10] == 'a' &&
10430 { /* getservbyname */
10431 return -KEY_getservbyname;
10437 if (name[10] == 'o' &&
10440 { /* getservbyport */
10441 return -KEY_getservbyport;
10460 case 14: /* 1 tokens of length 14 */
10461 if (name[0] == 'g' &&
10475 { /* getprotobyname */
10476 return -KEY_getprotobyname;
10481 case 16: /* 1 tokens of length 16 */
10482 if (name[0] == 'g' &&
10498 { /* getprotobynumber */
10499 return -KEY_getprotobynumber;
10513 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10517 PERL_ARGS_ASSERT_CHECKCOMMA;
10519 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10520 if (ckWARN(WARN_SYNTAX)) {
10523 for (w = s+2; *w && level; w++) {
10526 else if (*w == ')')
10529 while (isSPACE(*w))
10531 /* the list of chars below is for end of statements or
10532 * block / parens, boolean operators (&&, ||, //) and branch
10533 * constructs (or, and, if, until, unless, while, err, for).
10534 * Not a very solid hack... */
10535 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10536 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10537 "%s (...) interpreted as function",name);
10540 while (s < PL_bufend && isSPACE(*s))
10544 while (s < PL_bufend && isSPACE(*s))
10546 if (isIDFIRST_lazy_if(s,UTF)) {
10547 const char * const w = s++;
10548 while (isALNUM_lazy_if(s,UTF))
10550 while (s < PL_bufend && isSPACE(*s))
10554 if (keyword(w, s - w, 0))
10557 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10558 if (gv && GvCVu(gv))
10560 Perl_croak(aTHX_ "No comma allowed after %s", what);
10565 /* Either returns sv, or mortalizes sv and returns a new SV*.
10566 Best used as sv=new_constant(..., sv, ...).
10567 If s, pv are NULL, calls subroutine with one argument,
10568 and type is used with error messages only. */
10571 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10572 SV *sv, SV *pv, const char *type, STRLEN typelen)
10575 HV * const table = GvHV(PL_hintgv); /* ^H */
10579 const char *why1 = "", *why2 = "", *why3 = "";
10581 PERL_ARGS_ASSERT_NEW_CONSTANT;
10583 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10586 why2 = (const char *)
10587 (strEQ(key,"charnames")
10588 ? "(possibly a missing \"use charnames ...\")"
10590 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10591 (type ? type: "undef"), why2);
10593 /* This is convoluted and evil ("goto considered harmful")
10594 * but I do not understand the intricacies of all the different
10595 * failure modes of %^H in here. The goal here is to make
10596 * the most probable error message user-friendly. --jhi */
10601 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10602 (type ? type: "undef"), why1, why2, why3);
10604 yyerror(SvPVX_const(msg));
10608 cvp = hv_fetch(table, key, keylen, FALSE);
10609 if (!cvp || !SvOK(*cvp)) {
10612 why3 = "} is not defined";
10615 sv_2mortal(sv); /* Parent created it permanently */
10618 pv = newSVpvn_flags(s, len, SVs_TEMP);
10620 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10622 typesv = &PL_sv_undef;
10624 PUSHSTACKi(PERLSI_OVERLOAD);
10636 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10640 /* Check the eval first */
10641 if (!PL_in_eval && SvTRUE(ERRSV)) {
10642 sv_catpvs(ERRSV, "Propagated");
10643 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10645 res = SvREFCNT_inc_simple(sv);
10649 SvREFCNT_inc_simple_void(res);
10658 why1 = "Call to &{$^H{";
10660 why3 = "}} did not return a defined value";
10668 /* Returns a NUL terminated string, with the length of the string written to
10672 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10675 register char *d = dest;
10676 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10678 PERL_ARGS_ASSERT_SCAN_WORD;
10682 Perl_croak(aTHX_ ident_too_long);
10683 if (isALNUM(*s)) /* UTF handled below */
10685 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10690 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10694 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10695 char *t = s + UTF8SKIP(s);
10697 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10701 Perl_croak(aTHX_ ident_too_long);
10702 Copy(s, d, len, char);
10715 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10718 char *bracket = NULL;
10720 register char *d = dest;
10721 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10723 PERL_ARGS_ASSERT_SCAN_IDENT;
10728 while (isDIGIT(*s)) {
10730 Perl_croak(aTHX_ ident_too_long);
10737 Perl_croak(aTHX_ ident_too_long);
10738 if (isALNUM(*s)) /* UTF handled below */
10740 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10745 else if (*s == ':' && s[1] == ':') {
10749 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10750 char *t = s + UTF8SKIP(s);
10751 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10753 if (d + (t - s) > e)
10754 Perl_croak(aTHX_ ident_too_long);
10755 Copy(s, d, t - s, char);
10766 if (PL_lex_state != LEX_NORMAL)
10767 PL_lex_state = LEX_INTERPENDMAYBE;
10770 if (*s == '$' && s[1] &&
10771 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10784 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10789 if (isSPACE(s[-1])) {
10791 const char ch = *s++;
10792 if (!SPACE_OR_TAB(ch)) {
10798 if (isIDFIRST_lazy_if(d,UTF)) {
10802 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10803 end += UTF8SKIP(end);
10804 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10805 end += UTF8SKIP(end);
10807 Copy(s, d, end - s, char);
10812 while ((isALNUM(*s) || *s == ':') && d < e)
10815 Perl_croak(aTHX_ ident_too_long);
10818 while (s < send && SPACE_OR_TAB(*s))
10820 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10821 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10822 const char * const brack =
10824 ((*s == '[') ? "[...]" : "{...}");
10825 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10826 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10827 funny, dest, brack, funny, dest, brack);
10830 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10834 /* Handle extended ${^Foo} variables
10835 * 1999-02-27 mjd-perl-patch@plover.com */
10836 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10840 while (isALNUM(*s) && d < e) {
10844 Perl_croak(aTHX_ ident_too_long);
10849 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10850 PL_lex_state = LEX_INTERPEND;
10853 if (PL_lex_state == LEX_NORMAL) {
10854 if (ckWARN(WARN_AMBIGUOUS) &&
10855 (keyword(dest, d - dest, 0)
10856 || get_cvn_flags(dest, d - dest, 0)))
10860 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10861 "Ambiguous use of %c{%s} resolved to %c%s",
10862 funny, dest, funny, dest);
10867 s = bracket; /* let the parser handle it */
10871 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10872 PL_lex_state = LEX_INTERPEND;
10877 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10879 PERL_ARGS_ASSERT_PMFLAG;
10881 PERL_UNUSED_CONTEXT;
10883 const char c = (char)ch;
10885 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10886 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10887 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10888 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10889 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10895 S_scan_pat(pTHX_ char *start, I32 type)
10899 char *s = scan_str(start,!!PL_madskills,FALSE);
10900 const char * const valid_flags =
10901 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10906 PERL_ARGS_ASSERT_SCAN_PAT;
10909 const char * const delimiter = skipspace(start);
10913 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10914 : "Search pattern not terminated" ));
10917 pm = (PMOP*)newPMOP(type, 0);
10918 if (PL_multi_open == '?') {
10919 /* This is the only point in the code that sets PMf_ONCE: */
10920 pm->op_pmflags |= PMf_ONCE;
10922 /* Hence it's safe to do this bit of PMOP book-keeping here, which
10923 allows us to restrict the list needed by reset to just the ??
10925 assert(type != OP_TRANS);
10927 MAGIC *mg = mg_find((SV*)PL_curstash, PERL_MAGIC_symtab);
10930 mg = sv_magicext((SV*)PL_curstash, 0, PERL_MAGIC_symtab, 0, 0,
10933 elements = mg->mg_len / sizeof(PMOP**);
10934 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
10935 ((PMOP**)mg->mg_ptr) [elements++] = pm;
10936 mg->mg_len = elements * sizeof(PMOP**);
10937 PmopSTASH_set(pm,PL_curstash);
10943 while (*s && strchr(valid_flags, *s))
10944 pmflag(&pm->op_pmflags,*s++);
10946 if (PL_madskills && modstart != s) {
10947 SV* tmptoken = newSVpvn(modstart, s - modstart);
10948 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
10951 /* issue a warning if /c is specified,but /g is not */
10952 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
10953 && ckWARN(WARN_REGEXP))
10955 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
10956 "Use of /c modifier is meaningless without /g" );
10959 PL_lex_op = (OP*)pm;
10960 pl_yylval.ival = OP_MATCH;
10965 S_scan_subst(pTHX_ char *start)
10976 PERL_ARGS_ASSERT_SCAN_SUBST;
10978 pl_yylval.ival = OP_NULL;
10980 s = scan_str(start,!!PL_madskills,FALSE);
10983 Perl_croak(aTHX_ "Substitution pattern not terminated");
10985 if (s[-1] == PL_multi_open)
10988 if (PL_madskills) {
10989 CURMAD('q', PL_thisopen);
10990 CURMAD('_', PL_thiswhite);
10991 CURMAD('E', PL_thisstuff);
10992 CURMAD('Q', PL_thisclose);
10993 PL_realtokenstart = s - SvPVX(PL_linestr);
10997 first_start = PL_multi_start;
10998 s = scan_str(s,!!PL_madskills,FALSE);
11000 if (PL_lex_stuff) {
11001 SvREFCNT_dec(PL_lex_stuff);
11002 PL_lex_stuff = NULL;
11004 Perl_croak(aTHX_ "Substitution replacement not terminated");
11006 PL_multi_start = first_start; /* so whole substitution is taken together */
11008 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11011 if (PL_madskills) {
11012 CURMAD('z', PL_thisopen);
11013 CURMAD('R', PL_thisstuff);
11014 CURMAD('Z', PL_thisclose);
11020 if (*s == EXEC_PAT_MOD) {
11024 else if (strchr(S_PAT_MODS, *s))
11025 pmflag(&pm->op_pmflags,*s++);
11031 if (PL_madskills) {
11033 curmad('m', newSVpvn(modstart, s - modstart));
11034 append_madprops(PL_thismad, (OP*)pm, 0);
11038 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11039 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11043 SV * const repl = newSVpvs("");
11045 PL_sublex_info.super_bufptr = s;
11046 PL_sublex_info.super_bufend = PL_bufend;
11048 pm->op_pmflags |= PMf_EVAL;
11051 sv_catpvs(repl, "eval ");
11053 sv_catpvs(repl, "do ");
11055 sv_catpvs(repl, "{");
11056 sv_catsv(repl, PL_lex_repl);
11057 if (strchr(SvPVX(PL_lex_repl), '#'))
11058 sv_catpvs(repl, "\n");
11059 sv_catpvs(repl, "}");
11061 SvREFCNT_dec(PL_lex_repl);
11062 PL_lex_repl = repl;
11065 PL_lex_op = (OP*)pm;
11066 pl_yylval.ival = OP_SUBST;
11071 S_scan_trans(pTHX_ char *start)
11084 PERL_ARGS_ASSERT_SCAN_TRANS;
11086 pl_yylval.ival = OP_NULL;
11088 s = scan_str(start,!!PL_madskills,FALSE);
11090 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11092 if (s[-1] == PL_multi_open)
11095 if (PL_madskills) {
11096 CURMAD('q', PL_thisopen);
11097 CURMAD('_', PL_thiswhite);
11098 CURMAD('E', PL_thisstuff);
11099 CURMAD('Q', PL_thisclose);
11100 PL_realtokenstart = s - SvPVX(PL_linestr);
11104 s = scan_str(s,!!PL_madskills,FALSE);
11106 if (PL_lex_stuff) {
11107 SvREFCNT_dec(PL_lex_stuff);
11108 PL_lex_stuff = NULL;
11110 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11112 if (PL_madskills) {
11113 CURMAD('z', PL_thisopen);
11114 CURMAD('R', PL_thisstuff);
11115 CURMAD('Z', PL_thisclose);
11118 complement = del = squash = 0;
11125 complement = OPpTRANS_COMPLEMENT;
11128 del = OPpTRANS_DELETE;
11131 squash = OPpTRANS_SQUASH;
11140 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11141 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11142 o->op_private &= ~OPpTRANS_ALL;
11143 o->op_private |= del|squash|complement|
11144 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11145 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11148 pl_yylval.ival = OP_TRANS;
11151 if (PL_madskills) {
11153 curmad('m', newSVpvn(modstart, s - modstart));
11154 append_madprops(PL_thismad, o, 0);
11163 S_scan_heredoc(pTHX_ register char *s)
11167 I32 op_type = OP_SCALAR;
11171 const char *found_newline;
11175 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11177 I32 stuffstart = s - SvPVX(PL_linestr);
11180 PL_realtokenstart = -1;
11183 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11187 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11191 while (SPACE_OR_TAB(*peek))
11193 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11196 s = delimcpy(d, e, s, PL_bufend, term, &len);
11206 if (!isALNUM_lazy_if(s,UTF))
11207 deprecate_old("bare << to mean <<\"\"");
11208 for (; isALNUM_lazy_if(s,UTF); s++) {
11213 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11214 Perl_croak(aTHX_ "Delimiter for here document is too long");
11217 len = d - PL_tokenbuf;
11220 if (PL_madskills) {
11221 tstart = PL_tokenbuf + !outer;
11222 PL_thisclose = newSVpvn(tstart, len - !outer);
11223 tstart = SvPVX(PL_linestr) + stuffstart;
11224 PL_thisopen = newSVpvn(tstart, s - tstart);
11225 stuffstart = s - SvPVX(PL_linestr);
11228 #ifndef PERL_STRICT_CR
11229 d = strchr(s, '\r');
11231 char * const olds = s;
11233 while (s < PL_bufend) {
11239 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11248 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11255 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11256 herewas = newSVpvn(s,PL_bufend-s);
11260 herewas = newSVpvn(s-1,found_newline-s+1);
11263 herewas = newSVpvn(s,found_newline-s);
11267 if (PL_madskills) {
11268 tstart = SvPVX(PL_linestr) + stuffstart;
11270 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11272 PL_thisstuff = newSVpvn(tstart, s - tstart);
11275 s += SvCUR(herewas);
11278 stuffstart = s - SvPVX(PL_linestr);
11284 tmpstr = newSV_type(SVt_PVIV);
11285 SvGROW(tmpstr, 80);
11286 if (term == '\'') {
11287 op_type = OP_CONST;
11288 SvIV_set(tmpstr, -1);
11290 else if (term == '`') {
11291 op_type = OP_BACKTICK;
11292 SvIV_set(tmpstr, '\\');
11296 PL_multi_start = CopLINE(PL_curcop);
11297 PL_multi_open = PL_multi_close = '<';
11298 term = *PL_tokenbuf;
11299 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11300 char * const bufptr = PL_sublex_info.super_bufptr;
11301 char * const bufend = PL_sublex_info.super_bufend;
11302 char * const olds = s - SvCUR(herewas);
11303 s = strchr(bufptr, '\n');
11307 while (s < bufend &&
11308 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11310 CopLINE_inc(PL_curcop);
11313 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11314 missingterm(PL_tokenbuf);
11316 sv_setpvn(herewas,bufptr,d-bufptr+1);
11317 sv_setpvn(tmpstr,d+1,s-d);
11319 sv_catpvn(herewas,s,bufend-s);
11320 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11327 while (s < PL_bufend &&
11328 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11330 CopLINE_inc(PL_curcop);
11332 if (s >= PL_bufend) {
11333 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11334 missingterm(PL_tokenbuf);
11336 sv_setpvn(tmpstr,d+1,s-d);
11338 if (PL_madskills) {
11340 sv_catpvn(PL_thisstuff, d + 1, s - d);
11342 PL_thisstuff = newSVpvn(d + 1, s - d);
11343 stuffstart = s - SvPVX(PL_linestr);
11347 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11349 sv_catpvn(herewas,s,PL_bufend-s);
11350 sv_setsv(PL_linestr,herewas);
11351 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11352 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11353 PL_last_lop = PL_last_uni = NULL;
11356 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
11357 while (s >= PL_bufend) { /* multiple line string? */
11359 if (PL_madskills) {
11360 tstart = SvPVX(PL_linestr) + stuffstart;
11362 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11364 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11368 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11369 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11370 missingterm(PL_tokenbuf);
11373 stuffstart = s - SvPVX(PL_linestr);
11375 CopLINE_inc(PL_curcop);
11376 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11377 PL_last_lop = PL_last_uni = NULL;
11378 #ifndef PERL_STRICT_CR
11379 if (PL_bufend - PL_linestart >= 2) {
11380 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11381 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11383 PL_bufend[-2] = '\n';
11385 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11387 else if (PL_bufend[-1] == '\r')
11388 PL_bufend[-1] = '\n';
11390 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11391 PL_bufend[-1] = '\n';
11393 if (PERLDB_LINE && PL_curstash != PL_debstash)
11394 update_debugger_info(PL_linestr, NULL, 0);
11395 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11396 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11397 *(SvPVX(PL_linestr) + off ) = ' ';
11398 sv_catsv(PL_linestr,herewas);
11399 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11400 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11404 sv_catsv(tmpstr,PL_linestr);
11409 PL_multi_end = CopLINE(PL_curcop);
11410 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11411 SvPV_shrink_to_cur(tmpstr);
11413 SvREFCNT_dec(herewas);
11415 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11417 else if (PL_encoding)
11418 sv_recode_to_utf8(tmpstr, PL_encoding);
11420 PL_lex_stuff = tmpstr;
11421 pl_yylval.ival = op_type;
11425 /* scan_inputsymbol
11426 takes: current position in input buffer
11427 returns: new position in input buffer
11428 side-effects: pl_yylval and lex_op are set.
11433 <FH> read from filehandle
11434 <pkg::FH> read from package qualified filehandle
11435 <pkg'FH> read from package qualified filehandle
11436 <$fh> read from filehandle in $fh
11437 <*.h> filename glob
11442 S_scan_inputsymbol(pTHX_ char *start)
11445 register char *s = start; /* current position in buffer */
11448 char *d = PL_tokenbuf; /* start of temp holding space */
11449 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11451 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11453 end = strchr(s, '\n');
11456 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11458 /* die if we didn't have space for the contents of the <>,
11459 or if it didn't end, or if we see a newline
11462 if (len >= (I32)sizeof PL_tokenbuf)
11463 Perl_croak(aTHX_ "Excessively long <> operator");
11465 Perl_croak(aTHX_ "Unterminated <> operator");
11470 Remember, only scalar variables are interpreted as filehandles by
11471 this code. Anything more complex (e.g., <$fh{$num}>) will be
11472 treated as a glob() call.
11473 This code makes use of the fact that except for the $ at the front,
11474 a scalar variable and a filehandle look the same.
11476 if (*d == '$' && d[1]) d++;
11478 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11479 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11482 /* If we've tried to read what we allow filehandles to look like, and
11483 there's still text left, then it must be a glob() and not a getline.
11484 Use scan_str to pull out the stuff between the <> and treat it
11485 as nothing more than a string.
11488 if (d - PL_tokenbuf != len) {
11489 pl_yylval.ival = OP_GLOB;
11490 s = scan_str(start,!!PL_madskills,FALSE);
11492 Perl_croak(aTHX_ "Glob not terminated");
11496 bool readline_overriden = FALSE;
11499 /* we're in a filehandle read situation */
11502 /* turn <> into <ARGV> */
11504 Copy("ARGV",d,5,char);
11506 /* Check whether readline() is overriden */
11507 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11509 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11511 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11512 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11513 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11514 readline_overriden = TRUE;
11516 /* if <$fh>, create the ops to turn the variable into a
11520 /* try to find it in the pad for this block, otherwise find
11521 add symbol table ops
11523 const PADOFFSET tmp = pad_findmy(d);
11524 if (tmp != NOT_IN_PAD) {
11525 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11526 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11527 HEK * const stashname = HvNAME_HEK(stash);
11528 SV * const sym = sv_2mortal(newSVhek(stashname));
11529 sv_catpvs(sym, "::");
11530 sv_catpv(sym, d+1);
11535 OP * const o = newOP(OP_PADSV, 0);
11537 PL_lex_op = readline_overriden
11538 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11539 append_elem(OP_LIST, o,
11540 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11541 : (OP*)newUNOP(OP_READLINE, 0, o);
11550 ? (GV_ADDMULTI | GV_ADDINEVAL)
11553 PL_lex_op = readline_overriden
11554 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11555 append_elem(OP_LIST,
11556 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11557 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11558 : (OP*)newUNOP(OP_READLINE, 0,
11559 newUNOP(OP_RV2SV, 0,
11560 newGVOP(OP_GV, 0, gv)));
11562 if (!readline_overriden)
11563 PL_lex_op->op_flags |= OPf_SPECIAL;
11564 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11565 pl_yylval.ival = OP_NULL;
11568 /* If it's none of the above, it must be a literal filehandle
11569 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11571 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11572 PL_lex_op = readline_overriden
11573 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11574 append_elem(OP_LIST,
11575 newGVOP(OP_GV, 0, gv),
11576 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11577 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11578 pl_yylval.ival = OP_NULL;
11587 takes: start position in buffer
11588 keep_quoted preserve \ on the embedded delimiter(s)
11589 keep_delims preserve the delimiters around the string
11590 returns: position to continue reading from buffer
11591 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11592 updates the read buffer.
11594 This subroutine pulls a string out of the input. It is called for:
11595 q single quotes q(literal text)
11596 ' single quotes 'literal text'
11597 qq double quotes qq(interpolate $here please)
11598 " double quotes "interpolate $here please"
11599 qx backticks qx(/bin/ls -l)
11600 ` backticks `/bin/ls -l`
11601 qw quote words @EXPORT_OK = qw( func() $spam )
11602 m// regexp match m/this/
11603 s/// regexp substitute s/this/that/
11604 tr/// string transliterate tr/this/that/
11605 y/// string transliterate y/this/that/
11606 ($*@) sub prototypes sub foo ($)
11607 (stuff) sub attr parameters sub foo : attr(stuff)
11608 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11610 In most of these cases (all but <>, patterns and transliterate)
11611 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11612 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11613 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11616 It skips whitespace before the string starts, and treats the first
11617 character as the delimiter. If the delimiter is one of ([{< then
11618 the corresponding "close" character )]}> is used as the closing
11619 delimiter. It allows quoting of delimiters, and if the string has
11620 balanced delimiters ([{<>}]) it allows nesting.
11622 On success, the SV with the resulting string is put into lex_stuff or,
11623 if that is already non-NULL, into lex_repl. The second case occurs only
11624 when parsing the RHS of the special constructs s/// and tr/// (y///).
11625 For convenience, the terminating delimiter character is stuffed into
11630 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11633 SV *sv; /* scalar value: string */
11634 const char *tmps; /* temp string, used for delimiter matching */
11635 register char *s = start; /* current position in the buffer */
11636 register char term; /* terminating character */
11637 register char *to; /* current position in the sv's data */
11638 I32 brackets = 1; /* bracket nesting level */
11639 bool has_utf8 = FALSE; /* is there any utf8 content? */
11640 I32 termcode; /* terminating char. code */
11641 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11642 STRLEN termlen; /* length of terminating string */
11643 int last_off = 0; /* last position for nesting bracket */
11649 PERL_ARGS_ASSERT_SCAN_STR;
11651 /* skip space before the delimiter */
11657 if (PL_realtokenstart >= 0) {
11658 stuffstart = PL_realtokenstart;
11659 PL_realtokenstart = -1;
11662 stuffstart = start - SvPVX(PL_linestr);
11664 /* mark where we are, in case we need to report errors */
11667 /* after skipping whitespace, the next character is the terminator */
11670 termcode = termstr[0] = term;
11674 termcode = utf8_to_uvchr((U8*)s, &termlen);
11675 Copy(s, termstr, termlen, U8);
11676 if (!UTF8_IS_INVARIANT(term))
11680 /* mark where we are */
11681 PL_multi_start = CopLINE(PL_curcop);
11682 PL_multi_open = term;
11684 /* find corresponding closing delimiter */
11685 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11686 termcode = termstr[0] = term = tmps[5];
11688 PL_multi_close = term;
11690 /* create a new SV to hold the contents. 79 is the SV's initial length.
11691 What a random number. */
11692 sv = newSV_type(SVt_PVIV);
11694 SvIV_set(sv, termcode);
11695 (void)SvPOK_only(sv); /* validate pointer */
11697 /* move past delimiter and try to read a complete string */
11699 sv_catpvn(sv, s, termlen);
11702 tstart = SvPVX(PL_linestr) + stuffstart;
11703 if (!PL_thisopen && !keep_delims) {
11704 PL_thisopen = newSVpvn(tstart, s - tstart);
11705 stuffstart = s - SvPVX(PL_linestr);
11709 if (PL_encoding && !UTF) {
11713 int offset = s - SvPVX_const(PL_linestr);
11714 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11715 &offset, (char*)termstr, termlen);
11716 const char * const ns = SvPVX_const(PL_linestr) + offset;
11717 char * const svlast = SvEND(sv) - 1;
11719 for (; s < ns; s++) {
11720 if (*s == '\n' && !PL_rsfp)
11721 CopLINE_inc(PL_curcop);
11724 goto read_more_line;
11726 /* handle quoted delimiters */
11727 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11729 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11731 if ((svlast-1 - t) % 2) {
11732 if (!keep_quoted) {
11733 *(svlast-1) = term;
11735 SvCUR_set(sv, SvCUR(sv) - 1);
11740 if (PL_multi_open == PL_multi_close) {
11746 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11747 /* At here, all closes are "was quoted" one,
11748 so we don't check PL_multi_close. */
11750 if (!keep_quoted && *(t+1) == PL_multi_open)
11755 else if (*t == PL_multi_open)
11763 SvCUR_set(sv, w - SvPVX_const(sv));
11765 last_off = w - SvPVX(sv);
11766 if (--brackets <= 0)
11771 if (!keep_delims) {
11772 SvCUR_set(sv, SvCUR(sv) - 1);
11778 /* extend sv if need be */
11779 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11780 /* set 'to' to the next character in the sv's string */
11781 to = SvPVX(sv)+SvCUR(sv);
11783 /* if open delimiter is the close delimiter read unbridle */
11784 if (PL_multi_open == PL_multi_close) {
11785 for (; s < PL_bufend; s++,to++) {
11786 /* embedded newlines increment the current line number */
11787 if (*s == '\n' && !PL_rsfp)
11788 CopLINE_inc(PL_curcop);
11789 /* handle quoted delimiters */
11790 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11791 if (!keep_quoted && s[1] == term)
11793 /* any other quotes are simply copied straight through */
11797 /* terminate when run out of buffer (the for() condition), or
11798 have found the terminator */
11799 else if (*s == term) {
11802 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11805 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11811 /* if the terminator isn't the same as the start character (e.g.,
11812 matched brackets), we have to allow more in the quoting, and
11813 be prepared for nested brackets.
11816 /* read until we run out of string, or we find the terminator */
11817 for (; s < PL_bufend; s++,to++) {
11818 /* embedded newlines increment the line count */
11819 if (*s == '\n' && !PL_rsfp)
11820 CopLINE_inc(PL_curcop);
11821 /* backslashes can escape the open or closing characters */
11822 if (*s == '\\' && s+1 < PL_bufend) {
11823 if (!keep_quoted &&
11824 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11829 /* allow nested opens and closes */
11830 else if (*s == PL_multi_close && --brackets <= 0)
11832 else if (*s == PL_multi_open)
11834 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11839 /* terminate the copied string and update the sv's end-of-string */
11841 SvCUR_set(sv, to - SvPVX_const(sv));
11844 * this next chunk reads more into the buffer if we're not done yet
11848 break; /* handle case where we are done yet :-) */
11850 #ifndef PERL_STRICT_CR
11851 if (to - SvPVX_const(sv) >= 2) {
11852 if ((to[-2] == '\r' && to[-1] == '\n') ||
11853 (to[-2] == '\n' && to[-1] == '\r'))
11857 SvCUR_set(sv, to - SvPVX_const(sv));
11859 else if (to[-1] == '\r')
11862 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11867 /* if we're out of file, or a read fails, bail and reset the current
11868 line marker so we can report where the unterminated string began
11871 if (PL_madskills) {
11872 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11874 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11876 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11880 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11882 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11888 /* we read a line, so increment our line counter */
11889 CopLINE_inc(PL_curcop);
11891 /* update debugger info */
11892 if (PERLDB_LINE && PL_curstash != PL_debstash)
11893 update_debugger_info(PL_linestr, NULL, 0);
11895 /* having changed the buffer, we must update PL_bufend */
11896 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11897 PL_last_lop = PL_last_uni = NULL;
11900 /* at this point, we have successfully read the delimited string */
11902 if (!PL_encoding || UTF) {
11904 if (PL_madskills) {
11905 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11906 const int len = s - tstart;
11908 sv_catpvn(PL_thisstuff, tstart, len);
11910 PL_thisstuff = newSVpvn(tstart, len);
11911 if (!PL_thisclose && !keep_delims)
11912 PL_thisclose = newSVpvn(s,termlen);
11917 sv_catpvn(sv, s, termlen);
11922 if (PL_madskills) {
11923 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11924 const int len = s - tstart - termlen;
11926 sv_catpvn(PL_thisstuff, tstart, len);
11928 PL_thisstuff = newSVpvn(tstart, len);
11929 if (!PL_thisclose && !keep_delims)
11930 PL_thisclose = newSVpvn(s - termlen,termlen);
11934 if (has_utf8 || PL_encoding)
11937 PL_multi_end = CopLINE(PL_curcop);
11939 /* if we allocated too much space, give some back */
11940 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11941 SvLEN_set(sv, SvCUR(sv) + 1);
11942 SvPV_renew(sv, SvLEN(sv));
11945 /* decide whether this is the first or second quoted string we've read
11958 takes: pointer to position in buffer
11959 returns: pointer to new position in buffer
11960 side-effects: builds ops for the constant in pl_yylval.op
11962 Read a number in any of the formats that Perl accepts:
11964 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
11965 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
11968 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
11970 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
11973 If it reads a number without a decimal point or an exponent, it will
11974 try converting the number to an integer and see if it can do so
11975 without loss of precision.
11979 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
11982 register const char *s = start; /* current position in buffer */
11983 register char *d; /* destination in temp buffer */
11984 register char *e; /* end of temp buffer */
11985 NV nv; /* number read, as a double */
11986 SV *sv = NULL; /* place to put the converted number */
11987 bool floatit; /* boolean: int or float? */
11988 const char *lastub = NULL; /* position of last underbar */
11989 static char const number_too_long[] = "Number too long";
11991 PERL_ARGS_ASSERT_SCAN_NUM;
11993 /* We use the first character to decide what type of number this is */
11997 Perl_croak(aTHX_ "panic: scan_num");
11999 /* if it starts with a 0, it could be an octal number, a decimal in
12000 0.13 disguise, or a hexadecimal number, or a binary number. */
12004 u holds the "number so far"
12005 shift the power of 2 of the base
12006 (hex == 4, octal == 3, binary == 1)
12007 overflowed was the number more than we can hold?
12009 Shift is used when we add a digit. It also serves as an "are
12010 we in octal/hex/binary?" indicator to disallow hex characters
12011 when in octal mode.
12016 bool overflowed = FALSE;
12017 bool just_zero = TRUE; /* just plain 0 or binary number? */
12018 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12019 static const char* const bases[5] =
12020 { "", "binary", "", "octal", "hexadecimal" };
12021 static const char* const Bases[5] =
12022 { "", "Binary", "", "Octal", "Hexadecimal" };
12023 static const char* const maxima[5] =
12025 "0b11111111111111111111111111111111",
12029 const char *base, *Base, *max;
12031 /* check for hex */
12036 } else if (s[1] == 'b') {
12041 /* check for a decimal in disguise */
12042 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12044 /* so it must be octal */
12051 if (ckWARN(WARN_SYNTAX))
12052 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12053 "Misplaced _ in number");
12057 base = bases[shift];
12058 Base = Bases[shift];
12059 max = maxima[shift];
12061 /* read the rest of the number */
12063 /* x is used in the overflow test,
12064 b is the digit we're adding on. */
12069 /* if we don't mention it, we're done */
12073 /* _ are ignored -- but warned about if consecutive */
12075 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12076 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12077 "Misplaced _ in number");
12081 /* 8 and 9 are not octal */
12082 case '8': case '9':
12084 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12088 case '2': case '3': case '4':
12089 case '5': case '6': case '7':
12091 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12094 case '0': case '1':
12095 b = *s++ & 15; /* ASCII digit -> value of digit */
12099 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12100 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12101 /* make sure they said 0x */
12104 b = (*s++ & 7) + 9;
12106 /* Prepare to put the digit we have onto the end
12107 of the number so far. We check for overflows.
12113 x = u << shift; /* make room for the digit */
12115 if ((x >> shift) != u
12116 && !(PL_hints & HINT_NEW_BINARY)) {
12119 if (ckWARN_d(WARN_OVERFLOW))
12120 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12121 "Integer overflow in %s number",
12124 u = x | b; /* add the digit to the end */
12127 n *= nvshift[shift];
12128 /* If an NV has not enough bits in its
12129 * mantissa to represent an UV this summing of
12130 * small low-order numbers is a waste of time
12131 * (because the NV cannot preserve the
12132 * low-order bits anyway): we could just
12133 * remember when did we overflow and in the
12134 * end just multiply n by the right
12142 /* if we get here, we had success: make a scalar value from
12147 /* final misplaced underbar check */
12148 if (s[-1] == '_') {
12149 if (ckWARN(WARN_SYNTAX))
12150 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12155 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12156 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12157 "%s number > %s non-portable",
12163 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12164 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12165 "%s number > %s non-portable",
12170 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12171 sv = new_constant(start, s - start, "integer",
12172 sv, NULL, NULL, 0);
12173 else if (PL_hints & HINT_NEW_BINARY)
12174 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12179 handle decimal numbers.
12180 we're also sent here when we read a 0 as the first digit
12182 case '1': case '2': case '3': case '4': case '5':
12183 case '6': case '7': case '8': case '9': case '.':
12186 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12189 /* read next group of digits and _ and copy into d */
12190 while (isDIGIT(*s) || *s == '_') {
12191 /* skip underscores, checking for misplaced ones
12195 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12196 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12197 "Misplaced _ in number");
12201 /* check for end of fixed-length buffer */
12203 Perl_croak(aTHX_ number_too_long);
12204 /* if we're ok, copy the character */
12209 /* final misplaced underbar check */
12210 if (lastub && s == lastub + 1) {
12211 if (ckWARN(WARN_SYNTAX))
12212 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12215 /* read a decimal portion if there is one. avoid
12216 3..5 being interpreted as the number 3. followed
12219 if (*s == '.' && s[1] != '.') {
12224 if (ckWARN(WARN_SYNTAX))
12225 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12226 "Misplaced _ in number");
12230 /* copy, ignoring underbars, until we run out of digits.
12232 for (; isDIGIT(*s) || *s == '_'; s++) {
12233 /* fixed length buffer check */
12235 Perl_croak(aTHX_ number_too_long);
12237 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12238 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12239 "Misplaced _ in number");
12245 /* fractional part ending in underbar? */
12246 if (s[-1] == '_') {
12247 if (ckWARN(WARN_SYNTAX))
12248 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12249 "Misplaced _ in number");
12251 if (*s == '.' && isDIGIT(s[1])) {
12252 /* oops, it's really a v-string, but without the "v" */
12258 /* read exponent part, if present */
12259 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12263 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12264 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12266 /* stray preinitial _ */
12268 if (ckWARN(WARN_SYNTAX))
12269 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12270 "Misplaced _ in number");
12274 /* allow positive or negative exponent */
12275 if (*s == '+' || *s == '-')
12278 /* stray initial _ */
12280 if (ckWARN(WARN_SYNTAX))
12281 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12282 "Misplaced _ in number");
12286 /* read digits of exponent */
12287 while (isDIGIT(*s) || *s == '_') {
12290 Perl_croak(aTHX_ number_too_long);
12294 if (((lastub && s == lastub + 1) ||
12295 (!isDIGIT(s[1]) && s[1] != '_'))
12296 && ckWARN(WARN_SYNTAX))
12297 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12298 "Misplaced _ in number");
12305 /* make an sv from the string */
12309 We try to do an integer conversion first if no characters
12310 indicating "float" have been found.
12315 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12317 if (flags == IS_NUMBER_IN_UV) {
12319 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12322 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12323 if (uv <= (UV) IV_MIN)
12324 sv_setiv(sv, -(IV)uv);
12331 /* terminate the string */
12333 nv = Atof(PL_tokenbuf);
12338 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12339 const char *const key = floatit ? "float" : "integer";
12340 const STRLEN keylen = floatit ? 5 : 7;
12341 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12342 key, keylen, sv, NULL, NULL, 0);
12346 /* if it starts with a v, it could be a v-string */
12349 sv = newSV(5); /* preallocate storage space */
12350 s = scan_vstring(s, PL_bufend, sv);
12354 /* make the op for the constant and return */
12357 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12359 lvalp->opval = NULL;
12365 S_scan_formline(pTHX_ register char *s)
12368 register char *eol;
12370 SV * const stuff = newSVpvs("");
12371 bool needargs = FALSE;
12372 bool eofmt = FALSE;
12374 char *tokenstart = s;
12375 SV* savewhite = NULL;
12377 if (PL_madskills) {
12378 savewhite = PL_thiswhite;
12383 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12385 while (!needargs) {
12388 #ifdef PERL_STRICT_CR
12389 while (SPACE_OR_TAB(*t))
12392 while (SPACE_OR_TAB(*t) || *t == '\r')
12395 if (*t == '\n' || t == PL_bufend) {
12400 if (PL_in_eval && !PL_rsfp) {
12401 eol = (char *) memchr(s,'\n',PL_bufend-s);
12406 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12408 for (t = s; t < eol; t++) {
12409 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12411 goto enough; /* ~~ must be first line in formline */
12413 if (*t == '@' || *t == '^')
12417 sv_catpvn(stuff, s, eol-s);
12418 #ifndef PERL_STRICT_CR
12419 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12420 char *end = SvPVX(stuff) + SvCUR(stuff);
12423 SvCUR_set(stuff, SvCUR(stuff) - 1);
12433 if (PL_madskills) {
12435 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12437 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12440 s = filter_gets(PL_linestr, PL_rsfp, 0);
12442 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12444 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12446 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12447 PL_last_lop = PL_last_uni = NULL;
12456 if (SvCUR(stuff)) {
12459 PL_lex_state = LEX_NORMAL;
12460 start_force(PL_curforce);
12461 NEXTVAL_NEXTTOKE.ival = 0;
12465 PL_lex_state = LEX_FORMLINE;
12467 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12469 else if (PL_encoding)
12470 sv_recode_to_utf8(stuff, PL_encoding);
12472 start_force(PL_curforce);
12473 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12475 start_force(PL_curforce);
12476 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12480 SvREFCNT_dec(stuff);
12482 PL_lex_formbrack = 0;
12486 if (PL_madskills) {
12488 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12490 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12491 PL_thiswhite = savewhite;
12498 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12501 const I32 oldsavestack_ix = PL_savestack_ix;
12502 CV* const outsidecv = PL_compcv;
12505 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12507 SAVEI32(PL_subline);
12508 save_item(PL_subname);
12509 SAVESPTR(PL_compcv);
12511 PL_compcv = (CV*)newSV_type(is_format ? SVt_PVFM : SVt_PVCV);
12512 CvFLAGS(PL_compcv) |= flags;
12514 PL_subline = CopLINE(PL_curcop);
12515 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12516 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc_simple(outsidecv);
12517 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12519 return oldsavestack_ix;
12523 #pragma segment Perl_yylex
12526 Perl_yywarn(pTHX_ const char *const s)
12530 PERL_ARGS_ASSERT_YYWARN;
12532 PL_in_eval |= EVAL_WARNONLY;
12534 PL_in_eval &= ~EVAL_WARNONLY;
12539 Perl_yyerror(pTHX_ const char *const s)
12542 const char *where = NULL;
12543 const char *context = NULL;
12546 int yychar = PL_parser->yychar;
12548 PERL_ARGS_ASSERT_YYERROR;
12550 if (!yychar || (yychar == ';' && !PL_rsfp))
12552 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12553 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12554 PL_oldbufptr != PL_bufptr) {
12557 The code below is removed for NetWare because it abends/crashes on NetWare
12558 when the script has error such as not having the closing quotes like:
12559 if ($var eq "value)
12560 Checking of white spaces is anyway done in NetWare code.
12563 while (isSPACE(*PL_oldoldbufptr))
12566 context = PL_oldoldbufptr;
12567 contlen = PL_bufptr - PL_oldoldbufptr;
12569 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12570 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12573 The code below is removed for NetWare because it abends/crashes on NetWare
12574 when the script has error such as not having the closing quotes like:
12575 if ($var eq "value)
12576 Checking of white spaces is anyway done in NetWare code.
12579 while (isSPACE(*PL_oldbufptr))
12582 context = PL_oldbufptr;
12583 contlen = PL_bufptr - PL_oldbufptr;
12585 else if (yychar > 255)
12586 where = "next token ???";
12587 else if (yychar == -2) { /* YYEMPTY */
12588 if (PL_lex_state == LEX_NORMAL ||
12589 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12590 where = "at end of line";
12591 else if (PL_lex_inpat)
12592 where = "within pattern";
12594 where = "within string";
12597 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12599 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12600 else if (isPRINT_LC(yychar)) {
12601 const char string = yychar;
12602 sv_catpvn(where_sv, &string, 1);
12605 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12606 where = SvPVX_const(where_sv);
12608 msg = sv_2mortal(newSVpv(s, 0));
12609 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12610 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12612 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12614 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12615 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12616 Perl_sv_catpvf(aTHX_ msg,
12617 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12618 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12621 if (PL_in_eval & EVAL_WARNONLY) {
12622 if (ckWARN_d(WARN_SYNTAX))
12623 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12627 if (PL_error_count >= 10) {
12628 if (PL_in_eval && SvCUR(ERRSV))
12629 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12630 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12632 Perl_croak(aTHX_ "%s has too many errors.\n",
12633 OutCopFILE(PL_curcop));
12636 PL_in_my_stash = NULL;
12640 #pragma segment Main
12644 S_swallow_bom(pTHX_ U8 *s)
12647 const STRLEN slen = SvCUR(PL_linestr);
12649 PERL_ARGS_ASSERT_SWALLOW_BOM;
12653 if (s[1] == 0xFE) {
12654 /* UTF-16 little-endian? (or UTF32-LE?) */
12655 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12656 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12657 #ifndef PERL_NO_UTF16_FILTER
12658 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12661 if (PL_bufend > (char*)s) {
12665 filter_add(utf16rev_textfilter, NULL);
12666 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12667 utf16_to_utf8_reversed(s, news,
12668 PL_bufend - (char*)s - 1,
12670 sv_setpvn(PL_linestr, (const char*)news, newlen);
12672 s = (U8*)SvPVX(PL_linestr);
12673 Copy(news, s, newlen, U8);
12677 SvUTF8_on(PL_linestr);
12678 s = (U8*)SvPVX(PL_linestr);
12680 /* FIXME - is this a general bug fix? */
12683 PL_bufend = SvPVX(PL_linestr) + newlen;
12686 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12691 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12692 #ifndef PERL_NO_UTF16_FILTER
12693 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12696 if (PL_bufend > (char *)s) {
12700 filter_add(utf16_textfilter, NULL);
12701 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12702 utf16_to_utf8(s, news,
12703 PL_bufend - (char*)s,
12705 sv_setpvn(PL_linestr, (const char*)news, newlen);
12707 SvUTF8_on(PL_linestr);
12708 s = (U8*)SvPVX(PL_linestr);
12709 PL_bufend = SvPVX(PL_linestr) + newlen;
12712 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12717 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12718 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12719 s += 3; /* UTF-8 */
12725 if (s[2] == 0xFE && s[3] == 0xFF) {
12726 /* UTF-32 big-endian */
12727 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12730 else if (s[2] == 0 && s[3] != 0) {
12733 * are a good indicator of UTF-16BE. */
12734 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12740 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12741 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12742 s += 4; /* UTF-8 */
12748 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12751 * are a good indicator of UTF-16LE. */
12752 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12760 #ifndef PERL_NO_UTF16_FILTER
12762 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12765 const STRLEN old = SvCUR(sv);
12766 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12767 DEBUG_P(PerlIO_printf(Perl_debug_log,
12768 "utf16_textfilter(%p): %d %d (%d)\n",
12769 FPTR2DPTR(void *, utf16_textfilter),
12770 idx, maxlen, (int) count));
12774 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12775 Copy(SvPVX_const(sv), tmps, old, char);
12776 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12777 SvCUR(sv) - old, &newlen);
12778 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12780 DEBUG_P({sv_dump(sv);});
12785 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12788 const STRLEN old = SvCUR(sv);
12789 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12790 DEBUG_P(PerlIO_printf(Perl_debug_log,
12791 "utf16rev_textfilter(%p): %d %d (%d)\n",
12792 FPTR2DPTR(void *, utf16rev_textfilter),
12793 idx, maxlen, (int) count));
12797 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12798 Copy(SvPVX_const(sv), tmps, old, char);
12799 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12800 SvCUR(sv) - old, &newlen);
12801 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12803 DEBUG_P({ sv_dump(sv); });
12809 Returns a pointer to the next character after the parsed
12810 vstring, as well as updating the passed in sv.
12812 Function must be called like
12815 s = scan_vstring(s,e,sv);
12817 where s and e are the start and end of the string.
12818 The sv should already be large enough to store the vstring
12819 passed in, for performance reasons.
12824 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12827 const char *pos = s;
12828 const char *start = s;
12830 PERL_ARGS_ASSERT_SCAN_VSTRING;
12832 if (*pos == 'v') pos++; /* get past 'v' */
12833 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12835 if ( *pos != '.') {
12836 /* this may not be a v-string if followed by => */
12837 const char *next = pos;
12838 while (next < e && isSPACE(*next))
12840 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12841 /* return string not v-string */
12842 sv_setpvn(sv,(char *)s,pos-s);
12843 return (char *)pos;
12847 if (!isALPHA(*pos)) {
12848 U8 tmpbuf[UTF8_MAXBYTES+1];
12851 s++; /* get past 'v' */
12853 sv_setpvn(sv, "", 0);
12856 /* this is atoi() that tolerates underscores */
12859 const char *end = pos;
12861 while (--end >= s) {
12863 const UV orev = rev;
12864 rev += (*end - '0') * mult;
12866 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12867 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12868 "Integer overflow in decimal number");
12872 if (rev > 0x7FFFFFFF)
12873 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12875 /* Append native character for the rev point */
12876 tmpend = uvchr_to_utf8(tmpbuf, rev);
12877 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12878 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12880 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12886 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12890 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12898 * c-indentation-style: bsd
12899 * c-basic-offset: 4
12900 * indent-tabs-mode: t
12903 * ex: set ts=8 sts=4 sw=4 noet: