3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * 'It all comes from here, the stench and the peril.' --Frodo
14 * [p.719 of _The Lord of the Rings_, IV/ix: "Shelob's Lair"]
18 * This file is the lexer for Perl. It's closely linked to the
21 * The main routine is yylex(), which returns the next token.
25 =head1 Lexer interface
27 This is the lower layer of the Perl parser, managing characters and tokens.
29 =for apidoc AmU|yy_parser *|PL_parser
31 Pointer to a structure encapsulating the state of the parsing operation
32 currently in progress. The pointer can be locally changed to perform
33 a nested parse without interfering with the state of an outer parse.
34 Individual members of C<PL_parser> have their own documentation.
40 #define PERL_IN_TOKE_C
43 #define new_constant(a,b,c,d,e,f,g) \
44 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
46 #define pl_yylval (PL_parser->yylval)
48 /* YYINITDEPTH -- initial size of the parser's stacks. */
49 #define YYINITDEPTH 200
51 /* XXX temporary backwards compatibility */
52 #define PL_lex_brackets (PL_parser->lex_brackets)
53 #define PL_lex_brackstack (PL_parser->lex_brackstack)
54 #define PL_lex_casemods (PL_parser->lex_casemods)
55 #define PL_lex_casestack (PL_parser->lex_casestack)
56 #define PL_lex_defer (PL_parser->lex_defer)
57 #define PL_lex_dojoin (PL_parser->lex_dojoin)
58 #define PL_lex_expect (PL_parser->lex_expect)
59 #define PL_lex_formbrack (PL_parser->lex_formbrack)
60 #define PL_lex_inpat (PL_parser->lex_inpat)
61 #define PL_lex_inwhat (PL_parser->lex_inwhat)
62 #define PL_lex_op (PL_parser->lex_op)
63 #define PL_lex_repl (PL_parser->lex_repl)
64 #define PL_lex_starts (PL_parser->lex_starts)
65 #define PL_lex_stuff (PL_parser->lex_stuff)
66 #define PL_multi_start (PL_parser->multi_start)
67 #define PL_multi_open (PL_parser->multi_open)
68 #define PL_multi_close (PL_parser->multi_close)
69 #define PL_pending_ident (PL_parser->pending_ident)
70 #define PL_preambled (PL_parser->preambled)
71 #define PL_sublex_info (PL_parser->sublex_info)
72 #define PL_linestr (PL_parser->linestr)
73 #define PL_expect (PL_parser->expect)
74 #define PL_copline (PL_parser->copline)
75 #define PL_bufptr (PL_parser->bufptr)
76 #define PL_oldbufptr (PL_parser->oldbufptr)
77 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
78 #define PL_linestart (PL_parser->linestart)
79 #define PL_bufend (PL_parser->bufend)
80 #define PL_last_uni (PL_parser->last_uni)
81 #define PL_last_lop (PL_parser->last_lop)
82 #define PL_last_lop_op (PL_parser->last_lop_op)
83 #define PL_lex_state (PL_parser->lex_state)
84 #define PL_rsfp (PL_parser->rsfp)
85 #define PL_rsfp_filters (PL_parser->rsfp_filters)
86 #define PL_in_my (PL_parser->in_my)
87 #define PL_in_my_stash (PL_parser->in_my_stash)
88 #define PL_tokenbuf (PL_parser->tokenbuf)
89 #define PL_multi_end (PL_parser->multi_end)
90 #define PL_error_count (PL_parser->error_count)
93 # define PL_endwhite (PL_parser->endwhite)
94 # define PL_faketokens (PL_parser->faketokens)
95 # define PL_lasttoke (PL_parser->lasttoke)
96 # define PL_nextwhite (PL_parser->nextwhite)
97 # define PL_realtokenstart (PL_parser->realtokenstart)
98 # define PL_skipwhite (PL_parser->skipwhite)
99 # define PL_thisclose (PL_parser->thisclose)
100 # define PL_thismad (PL_parser->thismad)
101 # define PL_thisopen (PL_parser->thisopen)
102 # define PL_thisstuff (PL_parser->thisstuff)
103 # define PL_thistoken (PL_parser->thistoken)
104 # define PL_thiswhite (PL_parser->thiswhite)
105 # define PL_thiswhite (PL_parser->thiswhite)
106 # define PL_nexttoke (PL_parser->nexttoke)
107 # define PL_curforce (PL_parser->curforce)
109 # define PL_nexttoke (PL_parser->nexttoke)
110 # define PL_nexttype (PL_parser->nexttype)
111 # define PL_nextval (PL_parser->nextval)
114 /* This can't be done with embed.fnc, because struct yy_parser contains a
115 member named pending_ident, which clashes with the generated #define */
117 S_pending_ident(pTHX);
119 static const char ident_too_long[] = "Identifier too long";
122 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
123 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
125 # define CURMAD(slot,sv)
126 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
129 #define XFAKEBRACK 128
130 #define XENUMMASK 127
132 #ifdef USE_UTF8_SCRIPTS
133 # define UTF (!IN_BYTES)
135 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
138 /* The maximum number of characters preceding the unrecognized one to display */
139 #define UNRECOGNIZED_PRECEDE_COUNT 10
141 /* In variables named $^X, these are the legal values for X.
142 * 1999-02-27 mjd-perl-patch@plover.com */
143 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
145 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
147 /* LEX_* are values for PL_lex_state, the state of the lexer.
148 * They are arranged oddly so that the guard on the switch statement
149 * can get by with a single comparison (if the compiler is smart enough).
152 /* #define LEX_NOTPARSING 11 is done in perl.h. */
154 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
155 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
156 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
157 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
158 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
160 /* at end of code, eg "$x" followed by: */
161 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
162 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
164 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
165 string or after \E, $foo, etc */
166 #define LEX_INTERPCONST 2 /* NOT USED */
167 #define LEX_FORMLINE 1 /* expecting a format line */
168 #define LEX_KNOWNEXT 0 /* next token known; just return it */
172 static const char* const lex_state_names[] = {
191 #include "keywords.h"
193 /* CLINE is a macro that ensures PL_copline has a sane value */
198 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
201 # define SKIPSPACE0(s) skipspace0(s)
202 # define SKIPSPACE1(s) skipspace1(s)
203 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
204 # define PEEKSPACE(s) skipspace2(s,0)
206 # define SKIPSPACE0(s) skipspace(s)
207 # define SKIPSPACE1(s) skipspace(s)
208 # define SKIPSPACE2(s,tsv) skipspace(s)
209 # define PEEKSPACE(s) skipspace(s)
213 * Convenience functions to return different tokens and prime the
214 * lexer for the next token. They all take an argument.
216 * TOKEN : generic token (used for '(', DOLSHARP, etc)
217 * OPERATOR : generic operator
218 * AOPERATOR : assignment operator
219 * PREBLOCK : beginning the block after an if, while, foreach, ...
220 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
221 * PREREF : *EXPR where EXPR is not a simple identifier
222 * TERM : expression term
223 * LOOPX : loop exiting command (goto, last, dump, etc)
224 * FTST : file test operator
225 * FUN0 : zero-argument function
226 * FUN1 : not used, except for not, which isn't a UNIOP
227 * BOop : bitwise or or xor
229 * SHop : shift operator
230 * PWop : power operator
231 * PMop : pattern-matching operator
232 * Aop : addition-level operator
233 * Mop : multiplication-level operator
234 * Eop : equality-testing operator
235 * Rop : relational operator <= != gt
237 * Also see LOP and lop() below.
240 #ifdef DEBUGGING /* Serve -DT. */
241 # define REPORT(retval) tokereport((I32)retval, &pl_yylval)
243 # define REPORT(retval) (retval)
246 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
247 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
248 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
249 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
250 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
251 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
252 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
253 #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
254 #define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
255 #define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
256 #define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
257 #define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
258 #define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
259 #define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
260 #define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
261 #define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
262 #define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
263 #define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
264 #define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
265 #define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
267 /* This bit of chicanery makes a unary function followed by
268 * a parenthesis into a function with one argument, highest precedence.
269 * The UNIDOR macro is for unary functions that can be followed by the //
270 * operator (such as C<shift // 0>).
272 #define UNI2(f,x) { \
273 pl_yylval.ival = f; \
276 PL_last_uni = PL_oldbufptr; \
277 PL_last_lop_op = f; \
279 return REPORT( (int)FUNC1 ); \
281 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
283 #define UNI(f) UNI2(f,XTERM)
284 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
286 #define UNIBRACK(f) { \
287 pl_yylval.ival = f; \
289 PL_last_uni = PL_oldbufptr; \
291 return REPORT( (int)FUNC1 ); \
293 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
296 /* grandfather return to old style */
297 #define OLDLOP(f) return(pl_yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
301 /* how to interpret the pl_yylval associated with the token */
305 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
311 static struct debug_tokens {
313 enum token_type type;
315 } const debug_tokens[] =
317 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
318 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
319 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
320 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
321 { ARROW, TOKENTYPE_NONE, "ARROW" },
322 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
323 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
324 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
325 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
326 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
327 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
328 { DO, TOKENTYPE_NONE, "DO" },
329 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
330 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
331 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
332 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
333 { ELSE, TOKENTYPE_NONE, "ELSE" },
334 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
335 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
336 { FOR, TOKENTYPE_IVAL, "FOR" },
337 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
338 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
339 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
340 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
341 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
342 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
343 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
344 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
345 { IF, TOKENTYPE_IVAL, "IF" },
346 { LABEL, TOKENTYPE_PVAL, "LABEL" },
347 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
348 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
349 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
350 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
351 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
352 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
353 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
354 { MY, TOKENTYPE_IVAL, "MY" },
355 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
356 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
357 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
358 { OROP, TOKENTYPE_IVAL, "OROP" },
359 { OROR, TOKENTYPE_NONE, "OROR" },
360 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
361 { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" },
362 { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" },
363 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
364 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
365 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
366 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
367 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
368 { PREINC, TOKENTYPE_NONE, "PREINC" },
369 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
370 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
371 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
372 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
373 { SUB, TOKENTYPE_NONE, "SUB" },
374 { THING, TOKENTYPE_OPVAL, "THING" },
375 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
376 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
377 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
378 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
379 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
380 { USE, TOKENTYPE_IVAL, "USE" },
381 { WHEN, TOKENTYPE_IVAL, "WHEN" },
382 { WHILE, TOKENTYPE_IVAL, "WHILE" },
383 { WORD, TOKENTYPE_OPVAL, "WORD" },
384 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
385 { 0, TOKENTYPE_NONE, NULL }
388 /* dump the returned token in rv, plus any optional arg in pl_yylval */
391 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
395 PERL_ARGS_ASSERT_TOKEREPORT;
398 const char *name = NULL;
399 enum token_type type = TOKENTYPE_NONE;
400 const struct debug_tokens *p;
401 SV* const report = newSVpvs("<== ");
403 for (p = debug_tokens; p->token; p++) {
404 if (p->token == (int)rv) {
411 Perl_sv_catpv(aTHX_ report, name);
412 else if ((char)rv > ' ' && (char)rv < '~')
413 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
415 sv_catpvs(report, "EOF");
417 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
420 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
423 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
425 case TOKENTYPE_OPNUM:
426 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
427 PL_op_name[lvalp->ival]);
430 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
432 case TOKENTYPE_OPVAL:
434 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
435 PL_op_name[lvalp->opval->op_type]);
436 if (lvalp->opval->op_type == OP_CONST) {
437 Perl_sv_catpvf(aTHX_ report, " %s",
438 SvPEEK(cSVOPx_sv(lvalp->opval)));
443 sv_catpvs(report, "(opval=null)");
446 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
452 /* print the buffer with suitable escapes */
455 S_printbuf(pTHX_ const char *const fmt, const char *const s)
457 SV* const tmp = newSVpvs("");
459 PERL_ARGS_ASSERT_PRINTBUF;
461 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
468 S_deprecate_commaless_var_list(pTHX) {
470 deprecate("comma-less variable list");
471 return REPORT(','); /* grandfather non-comma-format format */
477 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
478 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
482 S_ao(pTHX_ int toketype)
485 if (*PL_bufptr == '=') {
487 if (toketype == ANDAND)
488 pl_yylval.ival = OP_ANDASSIGN;
489 else if (toketype == OROR)
490 pl_yylval.ival = OP_ORASSIGN;
491 else if (toketype == DORDOR)
492 pl_yylval.ival = OP_DORASSIGN;
500 * When Perl expects an operator and finds something else, no_op
501 * prints the warning. It always prints "<something> found where
502 * operator expected. It prints "Missing semicolon on previous line?"
503 * if the surprise occurs at the start of the line. "do you need to
504 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
505 * where the compiler doesn't know if foo is a method call or a function.
506 * It prints "Missing operator before end of line" if there's nothing
507 * after the missing operator, or "... before <...>" if there is something
508 * after the missing operator.
512 S_no_op(pTHX_ const char *const what, char *s)
515 char * const oldbp = PL_bufptr;
516 const bool is_first = (PL_oldbufptr == PL_linestart);
518 PERL_ARGS_ASSERT_NO_OP;
524 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
525 if (ckWARN_d(WARN_SYNTAX)) {
527 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
528 "\t(Missing semicolon on previous line?)\n");
529 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
531 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
533 if (t < PL_bufptr && isSPACE(*t))
534 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
535 "\t(Do you need to predeclare %.*s?)\n",
536 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
540 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
541 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
549 * Complain about missing quote/regexp/heredoc terminator.
550 * If it's called with NULL then it cauterizes the line buffer.
551 * If we're in a delimited string and the delimiter is a control
552 * character, it's reformatted into a two-char sequence like ^C.
557 S_missingterm(pTHX_ char *s)
563 char * const nl = strrchr(s,'\n');
567 else if (isCNTRL(PL_multi_close)) {
569 tmpbuf[1] = (char)toCTRL(PL_multi_close);
574 *tmpbuf = (char)PL_multi_close;
578 q = strchr(s,'"') ? '\'' : '"';
579 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
582 #define FEATURE_IS_ENABLED(name) \
583 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
584 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
585 /* The longest string we pass in. */
586 #define MAX_FEATURE_LEN (sizeof("unicode_strings")-1)
589 * S_feature_is_enabled
590 * Check whether the named feature is enabled.
593 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
596 HV * const hinthv = GvHV(PL_hintgv);
597 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
599 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
601 assert(namelen <= MAX_FEATURE_LEN);
602 memcpy(&he_name[8], name, namelen);
604 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
608 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
609 * utf16-to-utf8-reversed.
612 #ifdef PERL_CR_FILTER
616 register const char *s = SvPVX_const(sv);
617 register const char * const e = s + SvCUR(sv);
619 PERL_ARGS_ASSERT_STRIP_RETURN;
621 /* outer loop optimized to do nothing if there are no CR-LFs */
623 if (*s++ == '\r' && *s == '\n') {
624 /* hit a CR-LF, need to copy the rest */
625 register char *d = s - 1;
628 if (*s == '\r' && s[1] == '\n')
639 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
641 const I32 count = FILTER_READ(idx+1, sv, maxlen);
642 if (count > 0 && !maxlen)
653 * Create a parser object and initialise its parser and lexer fields
655 * rsfp is the opened file handle to read from (if any),
657 * line holds any initial content already read from the file (or in
658 * the case of no file, such as an eval, the whole contents);
660 * new_filter indicates that this is a new file and it shouldn't inherit
661 * the filters from the current parser (ie require).
665 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
668 const char *s = NULL;
670 yy_parser *parser, *oparser;
672 /* create and initialise a parser */
674 Newxz(parser, 1, yy_parser);
675 parser->old_parser = oparser = PL_parser;
678 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
679 parser->ps = parser->stack;
680 parser->stack_size = YYINITDEPTH;
682 parser->stack->state = 0;
683 parser->yyerrstatus = 0;
684 parser->yychar = YYEMPTY; /* Cause a token to be read. */
686 /* on scope exit, free this parser and restore any outer one */
688 parser->saved_curcop = PL_curcop;
690 /* initialise lexer state */
693 parser->curforce = -1;
695 parser->nexttoke = 0;
697 parser->error_count = oparser ? oparser->error_count : 0;
698 parser->copline = NOLINE;
699 parser->lex_state = LEX_NORMAL;
700 parser->expect = XSTATE;
702 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
703 : MUTABLE_AV(SvREFCNT_inc(oparser->rsfp_filters));
705 Newx(parser->lex_brackstack, 120, char);
706 Newx(parser->lex_casestack, 12, char);
707 *parser->lex_casestack = '\0';
710 s = SvPV_const(line, len);
716 parser->linestr = newSVpvs("\n;");
717 } else if (SvREADONLY(line) || s[len-1] != ';') {
718 parser->linestr = newSVsv(line);
720 sv_catpvs(parser->linestr, "\n;");
723 SvREFCNT_inc_simple_void_NN(line);
724 parser->linestr = line;
726 parser->oldoldbufptr =
729 parser->linestart = SvPVX(parser->linestr);
730 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
731 parser->last_lop = parser->last_uni = NULL;
735 /* delete a parser object */
738 Perl_parser_free(pTHX_ const yy_parser *parser)
740 PERL_ARGS_ASSERT_PARSER_FREE;
742 PL_curcop = parser->saved_curcop;
743 SvREFCNT_dec(parser->linestr);
745 if (parser->rsfp == PerlIO_stdin())
746 PerlIO_clearerr(parser->rsfp);
747 else if (parser->rsfp && (!parser->old_parser ||
748 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
749 PerlIO_close(parser->rsfp);
750 SvREFCNT_dec(parser->rsfp_filters);
752 Safefree(parser->stack);
753 Safefree(parser->lex_brackstack);
754 Safefree(parser->lex_casestack);
755 PL_parser = parser->old_parser;
762 * Finalizer for lexing operations. Must be called when the parser is
763 * done with the lexer.
770 PL_doextract = FALSE;
774 =for apidoc AmxU|SV *|PL_parser-E<gt>linestr
776 Buffer scalar containing the chunk currently under consideration of the
777 text currently being lexed. This is always a plain string scalar (for
778 which C<SvPOK> is true). It is not intended to be used as a scalar by
779 normal scalar means; instead refer to the buffer directly by the pointer
780 variables described below.
782 The lexer maintains various C<char*> pointers to things in the
783 C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever
784 reallocated, all of these pointers must be updated. Don't attempt to
785 do this manually, but rather use L</lex_grow_linestr> if you need to
786 reallocate the buffer.
788 The content of the text chunk in the buffer is commonly exactly one
789 complete line of input, up to and including a newline terminator,
790 but there are situations where it is otherwise. The octets of the
791 buffer may be intended to be interpreted as either UTF-8 or Latin-1.
792 The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8>
793 flag on this scalar, which may disagree with it.
795 For direct examination of the buffer, the variable
796 L</PL_parser-E<gt>bufend> points to the end of the buffer. The current
797 lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use
798 of these pointers is usually preferable to examination of the scalar
799 through normal scalar means.
801 =for apidoc AmxU|char *|PL_parser-E<gt>bufend
803 Direct pointer to the end of the chunk of text currently being lexed, the
804 end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr)
805 + SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is
806 always located at the end of the buffer, and does not count as part of
807 the buffer's contents.
809 =for apidoc AmxU|char *|PL_parser-E<gt>bufptr
811 Points to the current position of lexing inside the lexer buffer.
812 Characters around this point may be freely examined, within
813 the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
814 L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be
815 interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
817 Lexing code (whether in the Perl core or not) moves this pointer past
818 the characters that it consumes. It is also expected to perform some
819 bookkeeping whenever a newline character is consumed. This movement
820 can be more conveniently performed by the function L</lex_read_to>,
821 which handles newlines appropriately.
823 Interpretation of the buffer's octets can be abstracted out by
824 using the slightly higher-level functions L</lex_peek_unichar> and
825 L</lex_read_unichar>.
827 =for apidoc AmxU|char *|PL_parser-E<gt>linestart
829 Points to the start of the current line inside the lexer buffer.
830 This is useful for indicating at which column an error occurred, and
831 not much else. This must be updated by any lexing code that consumes
832 a newline; the function L</lex_read_to> handles this detail.
838 =for apidoc Amx|bool|lex_bufutf8
840 Indicates whether the octets in the lexer buffer
841 (L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
842 of Unicode characters. If not, they should be interpreted as Latin-1
843 characters. This is analogous to the C<SvUTF8> flag for scalars.
845 In UTF-8 mode, it is not guaranteed that the lexer buffer actually
846 contains valid UTF-8. Lexing code must be robust in the face of invalid
849 The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
850 is significant, but not the whole story regarding the input character
851 encoding. Normally, when a file is being read, the scalar contains octets
852 and its C<SvUTF8> flag is off, but the octets should be interpreted as
853 UTF-8 if the C<use utf8> pragma is in effect. During a string eval,
854 however, the scalar may have the C<SvUTF8> flag on, and in this case its
855 octets should be interpreted as UTF-8 unless the C<use bytes> pragma
856 is in effect. This logic may change in the future; use this function
857 instead of implementing the logic yourself.
863 Perl_lex_bufutf8(pTHX)
869 =for apidoc Amx|char *|lex_grow_linestr|STRLEN len
871 Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
872 at least I<len> octets (including terminating NUL). Returns a
873 pointer to the reallocated buffer. This is necessary before making
874 any direct modification of the buffer that would increase its length.
875 L</lex_stuff_pvn> provides a more convenient way to insert text into
878 Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
879 this function updates all of the lexer's variables that point directly
886 Perl_lex_grow_linestr(pTHX_ STRLEN len)
890 STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
891 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
892 linestr = PL_parser->linestr;
893 buf = SvPVX(linestr);
894 if (len <= SvLEN(linestr))
896 bufend_pos = PL_parser->bufend - buf;
897 bufptr_pos = PL_parser->bufptr - buf;
898 oldbufptr_pos = PL_parser->oldbufptr - buf;
899 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
900 linestart_pos = PL_parser->linestart - buf;
901 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
902 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
903 buf = sv_grow(linestr, len);
904 PL_parser->bufend = buf + bufend_pos;
905 PL_parser->bufptr = buf + bufptr_pos;
906 PL_parser->oldbufptr = buf + oldbufptr_pos;
907 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
908 PL_parser->linestart = buf + linestart_pos;
909 if (PL_parser->last_uni)
910 PL_parser->last_uni = buf + last_uni_pos;
911 if (PL_parser->last_lop)
912 PL_parser->last_lop = buf + last_lop_pos;
917 =for apidoc Amx|void|lex_stuff_pvn|char *pv|STRLEN len|U32 flags
919 Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
920 immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
921 reallocating the buffer if necessary. This means that lexing code that
922 runs later will see the characters as if they had appeared in the input.
923 It is not recommended to do this as part of normal parsing, and most
924 uses of this facility run the risk of the inserted characters being
925 interpreted in an unintended manner.
927 The string to be inserted is represented by I<len> octets starting
928 at I<pv>. These octets are interpreted as either UTF-8 or Latin-1,
929 according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
930 The characters are recoded for the lexer buffer, according to how the
931 buffer is currently being interpreted (L</lex_bufutf8>). If a string
932 to be interpreted is available as a Perl scalar, the L</lex_stuff_sv>
933 function is more convenient.
939 Perl_lex_stuff_pvn(pTHX_ char *pv, STRLEN len, U32 flags)
942 PERL_ARGS_ASSERT_LEX_STUFF_PVN;
943 if (flags & ~(LEX_STUFF_UTF8))
944 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
946 if (flags & LEX_STUFF_UTF8) {
950 char *p, *e = pv+len;
951 for (p = pv; p != e; p++)
952 highhalf += !!(((U8)*p) & 0x80);
955 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
956 bufptr = PL_parser->bufptr;
957 Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
958 PL_parser->bufend += len+highhalf;
959 for (p = pv; p != e; p++) {
962 *bufptr++ = (char)(0xc0 | (c >> 6));
963 *bufptr++ = (char)(0x80 | (c & 0x3f));
970 if (flags & LEX_STUFF_UTF8) {
972 char *p, *e = pv+len;
973 for (p = pv; p != e; p++) {
976 Perl_croak(aTHX_ "Lexing code attempted to stuff "
977 "non-Latin-1 character into Latin-1 input");
978 } else if (c >= 0xc2 && p+1 != e &&
979 (((U8)p[1]) & 0xc0) == 0x80) {
982 } else if (c >= 0x80) {
983 /* malformed UTF-8 */
985 SAVESPTR(PL_warnhook);
986 PL_warnhook = PERL_WARNHOOK_FATAL;
987 utf8n_to_uvuni((U8*)p, e-p, NULL, 0);
993 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
994 bufptr = PL_parser->bufptr;
995 Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
996 PL_parser->bufend += len-highhalf;
997 for (p = pv; p != e; p++) {
1000 *bufptr++ = (char)(((c & 0x3) << 6) | (p[1] & 0x3f));
1003 *bufptr++ = (char)c;
1008 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1009 bufptr = PL_parser->bufptr;
1010 Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
1011 PL_parser->bufend += len;
1012 Copy(pv, bufptr, len, char);
1018 =for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1020 Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1021 immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1022 reallocating the buffer if necessary. This means that lexing code that
1023 runs later will see the characters as if they had appeared in the input.
1024 It is not recommended to do this as part of normal parsing, and most
1025 uses of this facility run the risk of the inserted characters being
1026 interpreted in an unintended manner.
1028 The string to be inserted is the string value of I<sv>. The characters
1029 are recoded for the lexer buffer, according to how the buffer is currently
1030 being interpreted (L</lex_bufutf8>). If a string to be interpreted is
1031 not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1032 need to construct a scalar.
1038 Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1042 PERL_ARGS_ASSERT_LEX_STUFF_SV;
1044 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1046 lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1050 =for apidoc Amx|void|lex_unstuff|char *ptr
1052 Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1053 I<ptr>. Text following I<ptr> will be moved, and the buffer shortened.
1054 This hides the discarded text from any lexing code that runs later,
1055 as if the text had never appeared.
1057 This is not the normal way to consume lexed text. For that, use
1064 Perl_lex_unstuff(pTHX_ char *ptr)
1068 PERL_ARGS_ASSERT_LEX_UNSTUFF;
1069 buf = PL_parser->bufptr;
1071 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1074 bufend = PL_parser->bufend;
1076 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1077 unstuff_len = ptr - buf;
1078 Move(ptr, buf, bufend+1-ptr, char);
1079 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1080 PL_parser->bufend = bufend - unstuff_len;
1084 =for apidoc Amx|void|lex_read_to|char *ptr
1086 Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1087 to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1088 performing the correct bookkeeping whenever a newline character is passed.
1089 This is the normal way to consume lexed text.
1091 Interpretation of the buffer's octets can be abstracted out by
1092 using the slightly higher-level functions L</lex_peek_unichar> and
1093 L</lex_read_unichar>.
1099 Perl_lex_read_to(pTHX_ char *ptr)
1102 PERL_ARGS_ASSERT_LEX_READ_TO;
1103 s = PL_parser->bufptr;
1104 if (ptr < s || ptr > PL_parser->bufend)
1105 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1106 for (; s != ptr; s++)
1108 CopLINE_inc(PL_curcop);
1109 PL_parser->linestart = s+1;
1111 PL_parser->bufptr = ptr;
1115 =for apidoc Amx|void|lex_discard_to|char *ptr
1117 Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1118 up to I<ptr>. The remaining content of the buffer will be moved, and
1119 all pointers into the buffer updated appropriately. I<ptr> must not
1120 be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1121 it is not permitted to discard text that has yet to be lexed.
1123 Normally it is not necessarily to do this directly, because it suffices to
1124 use the implicit discarding behaviour of L</lex_next_chunk> and things
1125 based on it. However, if a token stretches across multiple lines,
1126 and the lexing code has kept multiple lines of text in the buffer fof
1127 that purpose, then after completion of the token it would be wise to
1128 explicitly discard the now-unneeded earlier lines, to avoid future
1129 multi-line tokens growing the buffer without bound.
1135 Perl_lex_discard_to(pTHX_ char *ptr)
1139 PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1140 buf = SvPVX(PL_parser->linestr);
1142 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1145 if (ptr > PL_parser->bufptr)
1146 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1147 discard_len = ptr - buf;
1148 if (PL_parser->oldbufptr < ptr)
1149 PL_parser->oldbufptr = ptr;
1150 if (PL_parser->oldoldbufptr < ptr)
1151 PL_parser->oldoldbufptr = ptr;
1152 if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1153 PL_parser->last_uni = NULL;
1154 if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1155 PL_parser->last_lop = NULL;
1156 Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1157 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1158 PL_parser->bufend -= discard_len;
1159 PL_parser->bufptr -= discard_len;
1160 PL_parser->oldbufptr -= discard_len;
1161 PL_parser->oldoldbufptr -= discard_len;
1162 if (PL_parser->last_uni)
1163 PL_parser->last_uni -= discard_len;
1164 if (PL_parser->last_lop)
1165 PL_parser->last_lop -= discard_len;
1169 =for apidoc Amx|bool|lex_next_chunk|U32 flags
1171 Reads in the next chunk of text to be lexed, appending it to
1172 L</PL_parser-E<gt>linestr>. This should be called when lexing code has
1173 looked to the end of the current chunk and wants to know more. It is
1174 usual, but not necessary, for lexing to have consumed the entirety of
1175 the current chunk at this time.
1177 If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1178 chunk (i.e., the current chunk has been entirely consumed), normally the
1179 current chunk will be discarded at the same time that the new chunk is
1180 read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1181 will not be discarded. If the current chunk has not been entirely
1182 consumed, then it will not be discarded regardless of the flag.
1184 Returns true if some new text was added to the buffer, or false if the
1185 buffer has reached the end of the input text.
1190 #define LEX_FAKE_EOF 0x80000000
1193 Perl_lex_next_chunk(pTHX_ U32 flags)
1197 STRLEN old_bufend_pos, new_bufend_pos;
1198 STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1199 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
1200 bool got_some_for_debugger = 0;
1202 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF))
1203 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
1204 linestr = PL_parser->linestr;
1205 buf = SvPVX(linestr);
1206 if (!(flags & LEX_KEEP_PREVIOUS) &&
1207 PL_parser->bufptr == PL_parser->bufend) {
1208 old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1210 if (PL_parser->last_uni != PL_parser->bufend)
1211 PL_parser->last_uni = NULL;
1212 if (PL_parser->last_lop != PL_parser->bufend)
1213 PL_parser->last_lop = NULL;
1214 last_uni_pos = last_lop_pos = 0;
1218 old_bufend_pos = PL_parser->bufend - buf;
1219 bufptr_pos = PL_parser->bufptr - buf;
1220 oldbufptr_pos = PL_parser->oldbufptr - buf;
1221 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1222 linestart_pos = PL_parser->linestart - buf;
1223 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1224 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1226 if (flags & LEX_FAKE_EOF) {
1228 } else if (!PL_parser->rsfp) {
1230 } else if (filter_gets(linestr, old_bufend_pos)) {
1232 got_some_for_debugger = 1;
1234 if (!SvPOK(linestr)) /* can get undefined by filter_gets */
1235 sv_setpvs(linestr, "");
1237 /* End of real input. Close filehandle (unless it was STDIN),
1238 * then add implicit termination.
1240 if ((PerlIO*)PL_parser->rsfp == PerlIO_stdin())
1241 PerlIO_clearerr(PL_parser->rsfp);
1242 else if (PL_parser->rsfp)
1243 (void)PerlIO_close(PL_parser->rsfp);
1244 PL_parser->rsfp = NULL;
1245 PL_doextract = FALSE;
1247 if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1250 if (!PL_in_eval && PL_minus_p) {
1252 /*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1253 PL_minus_n = PL_minus_p = 0;
1254 } else if (!PL_in_eval && PL_minus_n) {
1255 sv_catpvs(linestr, /*{*/";}");
1258 sv_catpvs(linestr, ";");
1261 buf = SvPVX(linestr);
1262 new_bufend_pos = SvCUR(linestr);
1263 PL_parser->bufend = buf + new_bufend_pos;
1264 PL_parser->bufptr = buf + bufptr_pos;
1265 PL_parser->oldbufptr = buf + oldbufptr_pos;
1266 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1267 PL_parser->linestart = buf + linestart_pos;
1268 if (PL_parser->last_uni)
1269 PL_parser->last_uni = buf + last_uni_pos;
1270 if (PL_parser->last_lop)
1271 PL_parser->last_lop = buf + last_lop_pos;
1272 if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
1273 PL_curstash != PL_debstash) {
1274 /* debugger active and we're not compiling the debugger code,
1275 * so store the line into the debugger's array of lines
1277 update_debugger_info(NULL, buf+old_bufend_pos,
1278 new_bufend_pos-old_bufend_pos);
1284 =for apidoc Amx|I32|lex_peek_unichar|U32 flags
1286 Looks ahead one (Unicode) character in the text currently being lexed.
1287 Returns the codepoint (unsigned integer value) of the next character,
1288 or -1 if lexing has reached the end of the input text. To consume the
1289 peeked character, use L</lex_read_unichar>.
1291 If the next character is in (or extends into) the next chunk of input
1292 text, the next chunk will be read in. Normally the current chunk will be
1293 discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1294 then the current chunk will not be discarded.
1296 If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1297 is encountered, an exception is generated.
1303 Perl_lex_peek_unichar(pTHX_ U32 flags)
1306 if (flags & ~(LEX_KEEP_PREVIOUS))
1307 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1308 s = PL_parser->bufptr;
1309 bufend = PL_parser->bufend;
1315 if (!lex_next_chunk(flags))
1317 s = PL_parser->bufptr;
1318 bufend = PL_parser->bufend;
1324 len = PL_utf8skip[head];
1325 while ((STRLEN)(bufend-s) < len) {
1326 if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1328 s = PL_parser->bufptr;
1329 bufend = PL_parser->bufend;
1332 unichar = utf8n_to_uvuni((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
1333 if (retlen == (STRLEN)-1) {
1334 /* malformed UTF-8 */
1336 SAVESPTR(PL_warnhook);
1337 PL_warnhook = PERL_WARNHOOK_FATAL;
1338 utf8n_to_uvuni((U8*)s, bufend-s, NULL, 0);
1344 if (!lex_next_chunk(flags))
1346 s = PL_parser->bufptr;
1353 =for apidoc Amx|I32|lex_read_unichar|U32 flags
1355 Reads the next (Unicode) character in the text currently being lexed.
1356 Returns the codepoint (unsigned integer value) of the character read,
1357 and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1358 if lexing has reached the end of the input text. To non-destructively
1359 examine the next character, use L</lex_peek_unichar> instead.
1361 If the next character is in (or extends into) the next chunk of input
1362 text, the next chunk will be read in. Normally the current chunk will be
1363 discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1364 then the current chunk will not be discarded.
1366 If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1367 is encountered, an exception is generated.
1373 Perl_lex_read_unichar(pTHX_ U32 flags)
1376 if (flags & ~(LEX_KEEP_PREVIOUS))
1377 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1378 c = lex_peek_unichar(flags);
1381 CopLINE_inc(PL_curcop);
1382 PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1388 =for apidoc Amx|void|lex_read_space|U32 flags
1390 Reads optional spaces, in Perl style, in the text currently being
1391 lexed. The spaces may include ordinary whitespace characters and
1392 Perl-style comments. C<#line> directives are processed if encountered.
1393 L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1394 at a non-space character (or the end of the input text).
1396 If spaces extend into the next chunk of input text, the next chunk will
1397 be read in. Normally the current chunk will be discarded at the same
1398 time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1399 chunk will not be discarded.
1404 #define LEX_NO_NEXT_CHUNK 0x80000000
1407 Perl_lex_read_space(pTHX_ U32 flags)
1410 bool need_incline = 0;
1411 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK))
1412 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1415 sv_free(PL_skipwhite);
1416 PL_skipwhite = NULL;
1419 PL_skipwhite = newSVpvs("");
1420 #endif /* PERL_MAD */
1421 s = PL_parser->bufptr;
1422 bufend = PL_parser->bufend;
1428 } while (!(c == '\n' || (c == 0 && s == bufend)));
1429 } else if (c == '\n') {
1431 PL_parser->linestart = s;
1436 } else if (isSPACE(c)) {
1438 } else if (c == 0 && s == bufend) {
1442 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1443 #endif /* PERL_MAD */
1444 if (flags & LEX_NO_NEXT_CHUNK)
1446 PL_parser->bufptr = s;
1447 CopLINE_inc(PL_curcop);
1448 got_more = lex_next_chunk(flags);
1449 CopLINE_dec(PL_curcop);
1450 s = PL_parser->bufptr;
1451 bufend = PL_parser->bufend;
1454 if (need_incline && PL_parser->rsfp) {
1464 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1465 #endif /* PERL_MAD */
1466 PL_parser->bufptr = s;
1471 * This subroutine has nothing to do with tilting, whether at windmills
1472 * or pinball tables. Its name is short for "increment line". It
1473 * increments the current line number in CopLINE(PL_curcop) and checks
1474 * to see whether the line starts with a comment of the form
1475 * # line 500 "foo.pm"
1476 * If so, it sets the current line number and file to the values in the comment.
1480 S_incline(pTHX_ const char *s)
1487 PERL_ARGS_ASSERT_INCLINE;
1489 CopLINE_inc(PL_curcop);
1492 while (SPACE_OR_TAB(*s))
1494 if (strnEQ(s, "line", 4))
1498 if (SPACE_OR_TAB(*s))
1502 while (SPACE_OR_TAB(*s))
1510 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
1512 while (SPACE_OR_TAB(*s))
1514 if (*s == '"' && (t = strchr(s+1, '"'))) {
1520 while (!isSPACE(*t))
1524 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
1526 if (*e != '\n' && *e != '\0')
1527 return; /* false alarm */
1530 const STRLEN len = t - s;
1531 #ifndef USE_ITHREADS
1532 SV *const temp_sv = CopFILESV(PL_curcop);
1537 cf = SvPVX(temp_sv);
1538 tmplen = SvCUR(temp_sv);
1544 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
1545 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1546 * to *{"::_<newfilename"} */
1547 /* However, the long form of evals is only turned on by the
1548 debugger - usually they're "(eval %lu)" */
1552 STRLEN tmplen2 = len;
1553 if (tmplen + 2 <= sizeof smallbuf)
1556 Newx(tmpbuf, tmplen + 2, char);
1559 memcpy(tmpbuf + 2, cf, tmplen);
1561 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
1566 if (tmplen2 + 2 <= sizeof smallbuf)
1569 Newx(tmpbuf2, tmplen2 + 2, char);
1571 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
1572 /* Either they malloc'd it, or we malloc'd it,
1573 so no prefix is present in ours. */
1578 memcpy(tmpbuf2 + 2, s, tmplen2);
1581 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
1583 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
1584 /* adjust ${"::_<newfilename"} to store the new file name */
1585 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
1586 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
1587 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
1590 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
1592 if (tmpbuf != smallbuf) Safefree(tmpbuf);
1595 CopFILE_free(PL_curcop);
1596 CopFILE_setn(PL_curcop, s, len);
1598 CopLINE_set(PL_curcop, atoi(n)-1);
1602 /* skip space before PL_thistoken */
1605 S_skipspace0(pTHX_ register char *s)
1607 PERL_ARGS_ASSERT_SKIPSPACE0;
1614 PL_thiswhite = newSVpvs("");
1615 sv_catsv(PL_thiswhite, PL_skipwhite);
1616 sv_free(PL_skipwhite);
1619 PL_realtokenstart = s - SvPVX(PL_linestr);
1623 /* skip space after PL_thistoken */
1626 S_skipspace1(pTHX_ register char *s)
1628 const char *start = s;
1629 I32 startoff = start - SvPVX(PL_linestr);
1631 PERL_ARGS_ASSERT_SKIPSPACE1;
1636 start = SvPVX(PL_linestr) + startoff;
1637 if (!PL_thistoken && PL_realtokenstart >= 0) {
1638 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
1639 PL_thistoken = newSVpvn(tstart, start - tstart);
1641 PL_realtokenstart = -1;
1644 PL_nextwhite = newSVpvs("");
1645 sv_catsv(PL_nextwhite, PL_skipwhite);
1646 sv_free(PL_skipwhite);
1653 S_skipspace2(pTHX_ register char *s, SV **svp)
1656 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
1657 const I32 startoff = s - SvPVX(PL_linestr);
1659 PERL_ARGS_ASSERT_SKIPSPACE2;
1662 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
1663 if (!PL_madskills || !svp)
1665 start = SvPVX(PL_linestr) + startoff;
1666 if (!PL_thistoken && PL_realtokenstart >= 0) {
1667 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
1668 PL_thistoken = newSVpvn(tstart, start - tstart);
1669 PL_realtokenstart = -1;
1673 *svp = newSVpvs("");
1674 sv_setsv(*svp, PL_skipwhite);
1675 sv_free(PL_skipwhite);
1684 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1686 AV *av = CopFILEAVx(PL_curcop);
1688 SV * const sv = newSV_type(SVt_PVMG);
1690 sv_setsv(sv, orig_sv);
1692 sv_setpvn(sv, buf, len);
1695 av_store(av, (I32)CopLINE(PL_curcop), sv);
1701 * Called to gobble the appropriate amount and type of whitespace.
1702 * Skips comments as well.
1706 S_skipspace(pTHX_ register char *s)
1710 #endif /* PERL_MAD */
1711 PERL_ARGS_ASSERT_SKIPSPACE;
1714 sv_free(PL_skipwhite);
1715 PL_skipwhite = NULL;
1717 #endif /* PERL_MAD */
1718 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1719 while (s < PL_bufend && SPACE_OR_TAB(*s))
1722 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1724 lex_read_space(LEX_KEEP_PREVIOUS |
1725 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1726 LEX_NO_NEXT_CHUNK : 0));
1728 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1729 if (PL_linestart > PL_bufptr)
1730 PL_bufptr = PL_linestart;
1735 PL_skipwhite = newSVpvn(start, s-start);
1736 #endif /* PERL_MAD */
1742 * Check the unary operators to ensure there's no ambiguity in how they're
1743 * used. An ambiguous piece of code would be:
1745 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1746 * the +5 is its argument.
1756 if (PL_oldoldbufptr != PL_last_uni)
1758 while (isSPACE(*PL_last_uni))
1761 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1763 if ((t = strchr(s, '(')) && t < PL_bufptr)
1766 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1767 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1768 (int)(s - PL_last_uni), PL_last_uni);
1772 * LOP : macro to build a list operator. Its behaviour has been replaced
1773 * with a subroutine, S_lop() for which LOP is just another name.
1776 #define LOP(f,x) return lop(f,x,s)
1780 * Build a list operator (or something that might be one). The rules:
1781 * - if we have a next token, then it's a list operator [why?]
1782 * - if the next thing is an opening paren, then it's a function
1783 * - else it's a list operator
1787 S_lop(pTHX_ I32 f, int x, char *s)
1791 PERL_ARGS_ASSERT_LOP;
1797 PL_last_lop = PL_oldbufptr;
1798 PL_last_lop_op = (OPCODE)f;
1801 return REPORT(LSTOP);
1804 return REPORT(LSTOP);
1807 return REPORT(FUNC);
1810 return REPORT(FUNC);
1812 return REPORT(LSTOP);
1818 * Sets up for an eventual force_next(). start_force(0) basically does
1819 * an unshift, while start_force(-1) does a push. yylex removes items
1824 S_start_force(pTHX_ int where)
1828 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1829 where = PL_lasttoke;
1830 assert(PL_curforce < 0 || PL_curforce == where);
1831 if (PL_curforce != where) {
1832 for (i = PL_lasttoke; i > where; --i) {
1833 PL_nexttoke[i] = PL_nexttoke[i-1];
1837 if (PL_curforce < 0) /* in case of duplicate start_force() */
1838 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1839 PL_curforce = where;
1842 curmad('^', newSVpvs(""));
1843 CURMAD('_', PL_nextwhite);
1848 S_curmad(pTHX_ char slot, SV *sv)
1854 if (PL_curforce < 0)
1855 where = &PL_thismad;
1857 where = &PL_nexttoke[PL_curforce].next_mad;
1863 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1865 else if (PL_encoding) {
1866 sv_recode_to_utf8(sv, PL_encoding);
1871 /* keep a slot open for the head of the list? */
1872 if (slot != '_' && *where && (*where)->mad_key == '^') {
1873 (*where)->mad_key = slot;
1874 sv_free(MUTABLE_SV(((*where)->mad_val)));
1875 (*where)->mad_val = (void*)sv;
1878 addmad(newMADsv(slot, sv), where, 0);
1881 # define start_force(where) NOOP
1882 # define curmad(slot, sv) NOOP
1887 * When the lexer realizes it knows the next token (for instance,
1888 * it is reordering tokens for the parser) then it can call S_force_next
1889 * to know what token to return the next time the lexer is called. Caller
1890 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1891 * and possibly PL_expect to ensure the lexer handles the token correctly.
1895 S_force_next(pTHX_ I32 type)
1900 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1901 tokereport(type, &NEXTVAL_NEXTTOKE);
1905 if (PL_curforce < 0)
1906 start_force(PL_lasttoke);
1907 PL_nexttoke[PL_curforce].next_type = type;
1908 if (PL_lex_state != LEX_KNOWNEXT)
1909 PL_lex_defer = PL_lex_state;
1910 PL_lex_state = LEX_KNOWNEXT;
1911 PL_lex_expect = PL_expect;
1914 PL_nexttype[PL_nexttoke] = type;
1916 if (PL_lex_state != LEX_KNOWNEXT) {
1917 PL_lex_defer = PL_lex_state;
1918 PL_lex_expect = PL_expect;
1919 PL_lex_state = LEX_KNOWNEXT;
1925 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1928 SV * const sv = newSVpvn_utf8(start, len,
1931 && !is_ascii_string((const U8*)start, len)
1932 && is_utf8_string((const U8*)start, len));
1938 * When the lexer knows the next thing is a word (for instance, it has
1939 * just seen -> and it knows that the next char is a word char, then
1940 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1944 * char *start : buffer position (must be within PL_linestr)
1945 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1946 * int check_keyword : if true, Perl checks to make sure the word isn't
1947 * a keyword (do this if the word is a label, e.g. goto FOO)
1948 * int allow_pack : if true, : characters will also be allowed (require,
1949 * use, etc. do this)
1950 * int allow_initial_tick : used by the "sub" lexer only.
1954 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1960 PERL_ARGS_ASSERT_FORCE_WORD;
1962 start = SKIPSPACE1(start);
1964 if (isIDFIRST_lazy_if(s,UTF) ||
1965 (allow_pack && *s == ':') ||
1966 (allow_initial_tick && *s == '\'') )
1968 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1969 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1971 start_force(PL_curforce);
1973 curmad('X', newSVpvn(start,s-start));
1974 if (token == METHOD) {
1979 PL_expect = XOPERATOR;
1983 curmad('g', newSVpvs( "forced" ));
1984 NEXTVAL_NEXTTOKE.opval
1985 = (OP*)newSVOP(OP_CONST,0,
1986 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1987 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1995 * Called when the lexer wants $foo *foo &foo etc, but the program
1996 * text only contains the "foo" portion. The first argument is a pointer
1997 * to the "foo", and the second argument is the type symbol to prefix.
1998 * Forces the next token to be a "WORD".
1999 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
2003 S_force_ident(pTHX_ register const char *s, int kind)
2007 PERL_ARGS_ASSERT_FORCE_IDENT;
2010 const STRLEN len = strlen(s);
2011 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
2012 start_force(PL_curforce);
2013 NEXTVAL_NEXTTOKE.opval = o;
2016 o->op_private = OPpCONST_ENTERED;
2017 /* XXX see note in pp_entereval() for why we forgo typo
2018 warnings if the symbol must be introduced in an eval.
2020 gv_fetchpvn_flags(s, len,
2021 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2023 kind == '$' ? SVt_PV :
2024 kind == '@' ? SVt_PVAV :
2025 kind == '%' ? SVt_PVHV :
2033 Perl_str_to_version(pTHX_ SV *sv)
2038 const char *start = SvPV_const(sv,len);
2039 const char * const end = start + len;
2040 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
2042 PERL_ARGS_ASSERT_STR_TO_VERSION;
2044 while (start < end) {
2048 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
2053 retval += ((NV)n)/nshift;
2062 * Forces the next token to be a version number.
2063 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2064 * and if "guessing" is TRUE, then no new token is created (and the caller
2065 * must use an alternative parsing method).
2069 S_force_version(pTHX_ char *s, int guessing)
2075 I32 startoff = s - SvPVX(PL_linestr);
2078 PERL_ARGS_ASSERT_FORCE_VERSION;
2086 while (isDIGIT(*d) || *d == '_' || *d == '.')
2090 start_force(PL_curforce);
2091 curmad('X', newSVpvn(s,d-s));
2094 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
2096 s = scan_num(s, &pl_yylval);
2097 version = pl_yylval.opval;
2098 ver = cSVOPx(version)->op_sv;
2099 if (SvPOK(ver) && !SvNIOK(ver)) {
2100 SvUPGRADE(ver, SVt_PVNV);
2101 SvNV_set(ver, str_to_version(ver));
2102 SvNOK_on(ver); /* hint that it is a version */
2105 else if (guessing) {
2108 sv_free(PL_nextwhite); /* let next token collect whitespace */
2110 s = SvPVX(PL_linestr) + startoff;
2118 if (PL_madskills && !version) {
2119 sv_free(PL_nextwhite); /* let next token collect whitespace */
2121 s = SvPVX(PL_linestr) + startoff;
2124 /* NOTE: The parser sees the package name and the VERSION swapped */
2125 start_force(PL_curforce);
2126 NEXTVAL_NEXTTOKE.opval = version;
2133 * S_force_strict_version
2134 * Forces the next token to be a version number using strict syntax rules.
2138 S_force_strict_version(pTHX_ char *s)
2143 I32 startoff = s - SvPVX(PL_linestr);
2145 const char *errstr = NULL;
2147 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2149 while (isSPACE(*s)) /* leading whitespace */
2152 if (is_STRICT_VERSION(s,&errstr)) {
2154 s = (char *)scan_version(s, ver, 0);
2155 version = newSVOP(OP_CONST, 0, ver);
2157 else if ( (*s != ';' && *s != '}' ) && (s = SKIPSPACE1(s), (*s != ';' && *s !='}' ))) {
2160 yyerror(errstr); /* version required */
2165 if (PL_madskills && !version) {
2166 sv_free(PL_nextwhite); /* let next token collect whitespace */
2168 s = SvPVX(PL_linestr) + startoff;
2171 /* NOTE: The parser sees the package name and the VERSION swapped */
2172 start_force(PL_curforce);
2173 NEXTVAL_NEXTTOKE.opval = version;
2181 * Tokenize a quoted string passed in as an SV. It finds the next
2182 * chunk, up to end of string or a backslash. It may make a new
2183 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2188 S_tokeq(pTHX_ SV *sv)
2192 register char *send;
2197 PERL_ARGS_ASSERT_TOKEQ;
2202 s = SvPV_force(sv, len);
2203 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
2206 while (s < send && *s != '\\')
2211 if ( PL_hints & HINT_NEW_STRING ) {
2212 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
2216 if (s + 1 < send && (s[1] == '\\'))
2217 s++; /* all that, just for this */
2222 SvCUR_set(sv, d - SvPVX_const(sv));
2224 if ( PL_hints & HINT_NEW_STRING )
2225 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
2230 * Now come three functions related to double-quote context,
2231 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2232 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2233 * interact with PL_lex_state, and create fake ( ... ) argument lists
2234 * to handle functions and concatenation.
2235 * They assume that whoever calls them will be setting up a fake
2236 * join call, because each subthing puts a ',' after it. This lets
2239 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
2241 * (I'm not sure whether the spurious commas at the end of lcfirst's
2242 * arguments and join's arguments are created or not).
2247 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
2249 * Pattern matching will set PL_lex_op to the pattern-matching op to
2250 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
2252 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2254 * Everything else becomes a FUNC.
2256 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2257 * had an OP_CONST or OP_READLINE). This just sets us up for a
2258 * call to S_sublex_push().
2262 S_sublex_start(pTHX)
2265 register const I32 op_type = pl_yylval.ival;
2267 if (op_type == OP_NULL) {
2268 pl_yylval.opval = PL_lex_op;
2272 if (op_type == OP_CONST || op_type == OP_READLINE) {
2273 SV *sv = tokeq(PL_lex_stuff);
2275 if (SvTYPE(sv) == SVt_PVIV) {
2276 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2278 const char * const p = SvPV_const(sv, len);
2279 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
2283 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
2284 PL_lex_stuff = NULL;
2285 /* Allow <FH> // "foo" */
2286 if (op_type == OP_READLINE)
2287 PL_expect = XTERMORDORDOR;
2290 else if (op_type == OP_BACKTICK && PL_lex_op) {
2291 /* readpipe() vas overriden */
2292 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
2293 pl_yylval.opval = PL_lex_op;
2295 PL_lex_stuff = NULL;
2299 PL_sublex_info.super_state = PL_lex_state;
2300 PL_sublex_info.sub_inwhat = (U16)op_type;
2301 PL_sublex_info.sub_op = PL_lex_op;
2302 PL_lex_state = LEX_INTERPPUSH;
2306 pl_yylval.opval = PL_lex_op;
2316 * Create a new scope to save the lexing state. The scope will be
2317 * ended in S_sublex_done. Returns a '(', starting the function arguments
2318 * to the uc, lc, etc. found before.
2319 * Sets PL_lex_state to LEX_INTERPCONCAT.
2328 PL_lex_state = PL_sublex_info.super_state;
2329 SAVEBOOL(PL_lex_dojoin);
2330 SAVEI32(PL_lex_brackets);
2331 SAVEI32(PL_lex_casemods);
2332 SAVEI32(PL_lex_starts);
2333 SAVEI8(PL_lex_state);
2334 SAVEVPTR(PL_lex_inpat);
2335 SAVEI16(PL_lex_inwhat);
2336 SAVECOPLINE(PL_curcop);
2337 SAVEPPTR(PL_bufptr);
2338 SAVEPPTR(PL_bufend);
2339 SAVEPPTR(PL_oldbufptr);
2340 SAVEPPTR(PL_oldoldbufptr);
2341 SAVEPPTR(PL_last_lop);
2342 SAVEPPTR(PL_last_uni);
2343 SAVEPPTR(PL_linestart);
2344 SAVESPTR(PL_linestr);
2345 SAVEGENERICPV(PL_lex_brackstack);
2346 SAVEGENERICPV(PL_lex_casestack);
2348 PL_linestr = PL_lex_stuff;
2349 PL_lex_stuff = NULL;
2351 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2352 = SvPVX(PL_linestr);
2353 PL_bufend += SvCUR(PL_linestr);
2354 PL_last_lop = PL_last_uni = NULL;
2355 SAVEFREESV(PL_linestr);
2357 PL_lex_dojoin = FALSE;
2358 PL_lex_brackets = 0;
2359 Newx(PL_lex_brackstack, 120, char);
2360 Newx(PL_lex_casestack, 12, char);
2361 PL_lex_casemods = 0;
2362 *PL_lex_casestack = '\0';
2364 PL_lex_state = LEX_INTERPCONCAT;
2365 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
2367 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
2368 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2369 PL_lex_inpat = PL_sublex_info.sub_op;
2371 PL_lex_inpat = NULL;
2378 * Restores lexer state after a S_sublex_push.
2385 if (!PL_lex_starts++) {
2386 SV * const sv = newSVpvs("");
2387 if (SvUTF8(PL_linestr))
2389 PL_expect = XOPERATOR;
2390 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2394 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2395 PL_lex_state = LEX_INTERPCASEMOD;
2399 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
2400 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2401 PL_linestr = PL_lex_repl;
2403 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2404 PL_bufend += SvCUR(PL_linestr);
2405 PL_last_lop = PL_last_uni = NULL;
2406 SAVEFREESV(PL_linestr);
2407 PL_lex_dojoin = FALSE;
2408 PL_lex_brackets = 0;
2409 PL_lex_casemods = 0;
2410 *PL_lex_casestack = '\0';
2412 if (SvEVALED(PL_lex_repl)) {
2413 PL_lex_state = LEX_INTERPNORMAL;
2415 /* we don't clear PL_lex_repl here, so that we can check later
2416 whether this is an evalled subst; that means we rely on the
2417 logic to ensure sublex_done() is called again only via the
2418 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
2421 PL_lex_state = LEX_INTERPCONCAT;
2431 PL_endwhite = newSVpvs("");
2432 sv_catsv(PL_endwhite, PL_thiswhite);
2436 sv_setpvs(PL_thistoken,"");
2438 PL_realtokenstart = -1;
2442 PL_bufend = SvPVX(PL_linestr);
2443 PL_bufend += SvCUR(PL_linestr);
2444 PL_expect = XOPERATOR;
2445 PL_sublex_info.sub_inwhat = 0;
2453 Extracts a pattern, double-quoted string, or transliteration. This
2456 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
2457 processing a pattern (PL_lex_inpat is true), a transliteration
2458 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
2460 Returns a pointer to the character scanned up to. If this is
2461 advanced from the start pointer supplied (i.e. if anything was
2462 successfully parsed), will leave an OP for the substring scanned
2463 in pl_yylval. Caller must intuit reason for not parsing further
2464 by looking at the next characters herself.
2468 double-quoted style: \r and \n
2469 regexp special ones: \D \s
2472 case and quoting: \U \Q \E
2473 stops on @ and $, but not for $ as tail anchor
2475 In transliterations:
2476 characters are VERY literal, except for - not at the start or end
2477 of the string, which indicates a range. If the range is in bytes,
2478 scan_const expands the range to the full set of intermediate
2479 characters. If the range is in utf8, the hyphen is replaced with
2480 a certain range mark which will be handled by pmtrans() in op.c.
2482 In double-quoted strings:
2484 double-quoted style: \r and \n
2486 deprecated backrefs: \1 (in substitution replacements)
2487 case and quoting: \U \Q \E
2490 scan_const does *not* construct ops to handle interpolated strings.
2491 It stops processing as soon as it finds an embedded $ or @ variable
2492 and leaves it to the caller to work out what's going on.
2494 embedded arrays (whether in pattern or not) could be:
2495 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2497 $ in double-quoted strings must be the symbol of an embedded scalar.
2499 $ in pattern could be $foo or could be tail anchor. Assumption:
2500 it's a tail anchor if $ is the last thing in the string, or if it's
2501 followed by one of "()| \r\n\t"
2503 \1 (backreferences) are turned into $1
2505 The structure of the code is
2506 while (there's a character to process) {
2507 handle transliteration ranges
2508 skip regexp comments /(?#comment)/ and codes /(?{code})/
2509 skip #-initiated comments in //x patterns
2510 check for embedded arrays
2511 check for embedded scalars
2513 leave intact backslashes from leaveit (below)
2514 deprecate \1 in substitution replacements
2515 handle string-changing backslashes \l \U \Q \E, etc.
2516 switch (what was escaped) {
2517 handle \- in a transliteration (becomes a literal -)
2518 handle \132 (octal characters)
2519 handle \x15 and \x{1234} (hex characters)
2520 handle \N{name} (named characters)
2521 handle \cV (control characters)
2522 handle printf-style backslashes (\f, \r, \n, etc)
2525 } (end if backslash)
2526 handle regular character
2527 } (end while character to read)
2532 S_scan_const(pTHX_ char *start)
2535 register char *send = PL_bufend; /* end of the constant */
2536 SV *sv = newSV(send - start); /* sv for the constant. See
2537 note below on sizing. */
2538 register char *s = start; /* start of the constant */
2539 register char *d = SvPVX(sv); /* destination for copies */
2540 bool dorange = FALSE; /* are we in a translit range? */
2541 bool didrange = FALSE; /* did we just finish a range? */
2542 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
2543 I32 this_utf8 = UTF; /* Is the source string assumed
2544 to be UTF8? But, this can
2545 show as true when the source
2546 isn't utf8, as for example
2547 when it is entirely composed
2550 /* Note on sizing: The scanned constant is placed into sv, which is
2551 * initialized by newSV() assuming one byte of output for every byte of
2552 * input. This routine expects newSV() to allocate an extra byte for a
2553 * trailing NUL, which this routine will append if it gets to the end of
2554 * the input. There may be more bytes of input than output (eg., \N{LATIN
2555 * CAPITAL LETTER A}), or more output than input if the constant ends up
2556 * recoded to utf8, but each time a construct is found that might increase
2557 * the needed size, SvGROW() is called. Its size parameter each time is
2558 * based on the best guess estimate at the time, namely the length used so
2559 * far, plus the length the current construct will occupy, plus room for
2560 * the trailing NUL, plus one byte for every input byte still unscanned */
2564 UV literal_endpoint = 0;
2565 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
2568 PERL_ARGS_ASSERT_SCAN_CONST;
2570 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2571 /* If we are doing a trans and we know we want UTF8 set expectation */
2572 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
2573 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2577 while (s < send || dorange) {
2578 /* get transliterations out of the way (they're most literal) */
2579 if (PL_lex_inwhat == OP_TRANS) {
2580 /* expand a range A-Z to the full set of characters. AIE! */
2582 I32 i; /* current expanded character */
2583 I32 min; /* first character in range */
2584 I32 max; /* last character in range */
2595 char * const c = (char*)utf8_hop((U8*)d, -1);
2599 *c = (char)UTF_TO_NATIVE(0xff);
2600 /* mark the range as done, and continue */
2606 i = d - SvPVX_const(sv); /* remember current offset */
2609 SvLEN(sv) + (has_utf8 ?
2610 (512 - UTF_CONTINUATION_MARK +
2613 /* How many two-byte within 0..255: 128 in UTF-8,
2614 * 96 in UTF-8-mod. */
2616 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
2618 d = SvPVX(sv) + i; /* refresh d after realloc */
2622 for (j = 0; j <= 1; j++) {
2623 char * const c = (char*)utf8_hop((U8*)d, -1);
2624 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2630 max = (U8)0xff; /* only to \xff */
2631 uvmax = uv; /* \x{100} to uvmax */
2633 d = c; /* eat endpoint chars */
2638 d -= 2; /* eat the first char and the - */
2639 min = (U8)*d; /* first char in range */
2640 max = (U8)d[1]; /* last char in range */
2647 "Invalid range \"%c-%c\" in transliteration operator",
2648 (char)min, (char)max);
2652 if (literal_endpoint == 2 &&
2653 ((isLOWER(min) && isLOWER(max)) ||
2654 (isUPPER(min) && isUPPER(max)))) {
2656 for (i = min; i <= max; i++)
2658 *d++ = NATIVE_TO_NEED(has_utf8,i);
2660 for (i = min; i <= max; i++)
2662 *d++ = NATIVE_TO_NEED(has_utf8,i);
2667 for (i = min; i <= max; i++)
2670 const U8 ch = (U8)NATIVE_TO_UTF(i);
2671 if (UNI_IS_INVARIANT(ch))
2674 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2675 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2684 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2686 *d++ = (char)UTF_TO_NATIVE(0xff);
2688 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2692 /* mark the range as done, and continue */
2696 literal_endpoint = 0;
2701 /* range begins (ignore - as first or last char) */
2702 else if (*s == '-' && s+1 < send && s != start) {
2704 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2711 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2721 literal_endpoint = 0;
2722 native_range = TRUE;
2727 /* if we get here, we're not doing a transliteration */
2729 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2730 except for the last char, which will be done separately. */
2731 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2733 while (s+1 < send && *s != ')')
2734 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2736 else if (s[2] == '{' /* This should match regcomp.c */
2737 || (s[2] == '?' && s[3] == '{'))
2740 char *regparse = s + (s[2] == '{' ? 3 : 4);
2743 while (count && (c = *regparse)) {
2744 if (c == '\\' && regparse[1])
2752 if (*regparse != ')')
2753 regparse--; /* Leave one char for continuation. */
2754 while (s < regparse)
2755 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2759 /* likewise skip #-initiated comments in //x patterns */
2760 else if (*s == '#' && PL_lex_inpat &&
2761 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2762 while (s+1 < send && *s != '\n')
2763 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2766 /* check for embedded arrays
2767 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2769 else if (*s == '@' && s[1]) {
2770 if (isALNUM_lazy_if(s+1,UTF))
2772 if (strchr(":'{$", s[1]))
2774 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2775 break; /* in regexp, neither @+ nor @- are interpolated */
2778 /* check for embedded scalars. only stop if we're sure it's a
2781 else if (*s == '$') {
2782 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2784 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
2786 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2787 "Possible unintended interpolation of $\\ in regex");
2789 break; /* in regexp, $ might be tail anchor */
2793 /* End of else if chain - OP_TRANS rejoin rest */
2796 if (*s == '\\' && s+1 < send) {
2799 /* deprecate \1 in strings and substitution replacements */
2800 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2801 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2803 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2808 /* string-change backslash escapes */
2809 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2813 /* skip any other backslash escapes in a pattern */
2814 else if (PL_lex_inpat) {
2815 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2816 goto default_action;
2819 /* if we get here, it's either a quoted -, or a digit */
2822 /* quoted - in transliterations */
2824 if (PL_lex_inwhat == OP_TRANS) {
2831 if ((isALPHA(*s) || isDIGIT(*s)))
2832 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
2833 "Unrecognized escape \\%c passed through",
2835 /* default action is to copy the quoted character */
2836 goto default_action;
2839 /* eg. \132 indicates the octal constant 0x132 */
2840 case '0': case '1': case '2': case '3':
2841 case '4': case '5': case '6': case '7':
2845 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
2848 goto NUM_ESCAPE_INSERT;
2850 /* eg. \x24 indicates the hex constant 0x24 */
2854 char* const e = strchr(s, '}');
2855 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2856 PERL_SCAN_DISALLOW_PREFIX;
2861 yyerror("Missing right brace on \\x{}");
2865 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
2871 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2872 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
2878 /* Insert oct, hex, or \N{U+...} escaped character. There will
2879 * always be enough room in sv since such escapes will be
2880 * longer than any UTF-8 sequence they can end up as, except if
2881 * they force us to recode the rest of the string into utf8 */
2883 /* Here uv is the ordinal of the next character being added in
2884 * unicode (converted from native). (It has to be done before
2885 * here because \N is interpreted as unicode, and oct and hex
2887 if (!UNI_IS_INVARIANT(uv)) {
2888 if (!has_utf8 && uv > 255) {
2889 /* Might need to recode whatever we have accumulated so
2890 * far if it contains any chars variant in utf8 or
2893 SvCUR_set(sv, d - SvPVX_const(sv));
2896 /* See Note on sizing above. */
2897 sv_utf8_upgrade_flags_grow(sv,
2898 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2899 UNISKIP(uv) + (STRLEN)(send - s) + 1);
2900 d = SvPVX(sv) + SvCUR(sv);
2905 d = (char*)uvuni_to_utf8((U8*)d, uv);
2906 if (PL_lex_inwhat == OP_TRANS &&
2907 PL_sublex_info.sub_op) {
2908 PL_sublex_info.sub_op->op_private |=
2909 (PL_lex_repl ? OPpTRANS_FROM_UTF
2913 if (uv > 255 && !dorange)
2914 native_range = FALSE;
2926 /* \N{LATIN SMALL LETTER A} is a named character, and so is
2931 char* e = strchr(s, '}');
2937 yyerror("Missing right brace on \\N{}");
2941 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2942 /* \N{U+...} The ... is a unicode value even on EBCDIC
2944 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2945 PERL_SCAN_DISALLOW_PREFIX;
2948 uv = grok_hex(s, &len, &flags, NULL);
2949 if ( e > s && len != (STRLEN)(e - s) ) {
2953 goto NUM_ESCAPE_INSERT;
2955 res = newSVpvn(s + 1, e - s - 1);
2956 res = new_constant( NULL, 0, "charnames",
2957 res, NULL, s - 2, e - s + 3 );
2959 sv_utf8_upgrade(res);
2960 str = SvPV_const(res,len);
2961 #ifdef EBCDIC_NEVER_MIND
2962 /* charnames uses pack U and that has been
2963 * recently changed to do the below uni->native
2964 * mapping, so this would be redundant (and wrong,
2965 * the code point would be doubly converted).
2966 * But leave this in just in case the pack U change
2967 * gets revoked, but the semantics is still
2968 * desireable for charnames. --jhi */
2970 UV uv = utf8_to_uvchr((const U8*)str, 0);
2973 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2975 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2976 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2977 str = SvPV_const(res, len);
2981 /* If destination is not in utf8 but this new character is,
2982 * recode the dest to utf8 */
2983 if (!has_utf8 && SvUTF8(res)) {
2984 SvCUR_set(sv, d - SvPVX_const(sv));
2987 /* See Note on sizing above. */
2988 sv_utf8_upgrade_flags_grow(sv,
2989 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2990 len + (STRLEN)(send - s) + 1);
2991 d = SvPVX(sv) + SvCUR(sv);
2993 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2995 /* See Note on sizing above. (NOTE: SvCUR() is not set
2996 * correctly here). */
2997 const STRLEN off = d - SvPVX_const(sv);
2998 d = SvGROW(sv, off + len + (STRLEN)(send - s) + 1) + off;
3002 native_range = FALSE; /* \N{} is guessed to be Unicode */
3004 Copy(str, d, len, char);
3011 yyerror("Missing braces on \\N{}");
3014 /* \c is a control character */
3023 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
3026 yyerror("Missing control char name in \\c");
3030 /* printf-style backslashes, formfeeds, newlines, etc */
3032 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
3035 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
3038 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
3041 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
3044 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
3047 *d++ = ASCII_TO_NEED(has_utf8,'\033');
3050 *d++ = ASCII_TO_NEED(has_utf8,'\007');
3056 } /* end if (backslash) */
3063 /* If we started with encoded form, or already know we want it,
3064 then encode the next character */
3065 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
3069 /* One might think that it is wasted effort in the case of the
3070 * source being utf8 (this_utf8 == TRUE) to take the next character
3071 * in the source, convert it to an unsigned value, and then convert
3072 * it back again. But the source has not been validated here. The
3073 * routine that does the conversion checks for errors like
3076 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
3077 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
3079 SvCUR_set(sv, d - SvPVX_const(sv));
3082 /* See Note on sizing above. */
3083 sv_utf8_upgrade_flags_grow(sv,
3084 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3085 need + (STRLEN)(send - s) + 1);
3086 d = SvPVX(sv) + SvCUR(sv);
3088 } else if (need > len) {
3089 /* encoded value larger than old, may need extra space (NOTE:
3090 * SvCUR() is not set correctly here). See Note on sizing
3092 const STRLEN off = d - SvPVX_const(sv);
3093 d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
3097 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
3099 if (uv > 255 && !dorange)
3100 native_range = FALSE;
3104 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
3106 } /* while loop to process each character */
3108 /* terminate the string and set up the sv */
3110 SvCUR_set(sv, d - SvPVX_const(sv));
3111 if (SvCUR(sv) >= SvLEN(sv))
3112 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
3115 if (PL_encoding && !has_utf8) {
3116 sv_recode_to_utf8(sv, PL_encoding);
3122 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
3123 PL_sublex_info.sub_op->op_private |=
3124 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3128 /* shrink the sv if we allocated more than we used */
3129 if (SvCUR(sv) + 5 < SvLEN(sv)) {
3130 SvPV_shrink_to_cur(sv);
3133 /* return the substring (via pl_yylval) only if we parsed anything */
3134 if (s > PL_bufptr) {
3135 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
3136 const char *const key = PL_lex_inpat ? "qr" : "q";
3137 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
3141 if (PL_lex_inwhat == OP_TRANS) {
3144 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
3152 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
3155 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3162 * Returns TRUE if there's more to the expression (e.g., a subscript),
3165 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
3167 * ->[ and ->{ return TRUE
3168 * { and [ outside a pattern are always subscripts, so return TRUE
3169 * if we're outside a pattern and it's not { or [, then return FALSE
3170 * if we're in a pattern and the first char is a {
3171 * {4,5} (any digits around the comma) returns FALSE
3172 * if we're in a pattern and the first char is a [
3174 * [SOMETHING] has a funky algorithm to decide whether it's a
3175 * character class or not. It has to deal with things like
3176 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
3177 * anything else returns TRUE
3180 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
3183 S_intuit_more(pTHX_ register char *s)
3187 PERL_ARGS_ASSERT_INTUIT_MORE;
3189 if (PL_lex_brackets)
3191 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
3193 if (*s != '{' && *s != '[')
3198 /* In a pattern, so maybe we have {n,m}. */
3215 /* On the other hand, maybe we have a character class */
3218 if (*s == ']' || *s == '^')
3221 /* this is terrifying, and it works */
3222 int weight = 2; /* let's weigh the evidence */
3224 unsigned char un_char = 255, last_un_char;
3225 const char * const send = strchr(s,']');
3226 char tmpbuf[sizeof PL_tokenbuf * 4];
3228 if (!send) /* has to be an expression */
3231 Zero(seen,256,char);
3234 else if (isDIGIT(*s)) {
3236 if (isDIGIT(s[1]) && s[2] == ']')
3242 for (; s < send; s++) {
3243 last_un_char = un_char;
3244 un_char = (unsigned char)*s;
3249 weight -= seen[un_char] * 10;
3250 if (isALNUM_lazy_if(s+1,UTF)) {
3252 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
3253 len = (int)strlen(tmpbuf);
3254 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
3259 else if (*s == '$' && s[1] &&
3260 strchr("[#!%*<>()-=",s[1])) {
3261 if (/*{*/ strchr("])} =",s[2]))
3270 if (strchr("wds]",s[1]))
3272 else if (seen[(U8)'\''] || seen[(U8)'"'])
3274 else if (strchr("rnftbxcav",s[1]))
3276 else if (isDIGIT(s[1])) {
3278 while (s[1] && isDIGIT(s[1]))
3288 if (strchr("aA01! ",last_un_char))
3290 if (strchr("zZ79~",s[1]))
3292 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
3293 weight -= 5; /* cope with negative subscript */
3296 if (!isALNUM(last_un_char)
3297 && !(last_un_char == '$' || last_un_char == '@'
3298 || last_un_char == '&')
3299 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
3304 if (keyword(tmpbuf, d - tmpbuf, 0))
3307 if (un_char == last_un_char + 1)
3309 weight -= seen[un_char];
3314 if (weight >= 0) /* probably a character class */
3324 * Does all the checking to disambiguate
3326 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
3327 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
3329 * First argument is the stuff after the first token, e.g. "bar".
3331 * Not a method if bar is a filehandle.
3332 * Not a method if foo is a subroutine prototyped to take a filehandle.
3333 * Not a method if it's really "Foo $bar"
3334 * Method if it's "foo $bar"
3335 * Not a method if it's really "print foo $bar"
3336 * Method if it's really "foo package::" (interpreted as package->foo)
3337 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
3338 * Not a method if bar is a filehandle or package, but is quoted with
3343 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
3346 char *s = start + (*start == '$');
3347 char tmpbuf[sizeof PL_tokenbuf];
3354 PERL_ARGS_ASSERT_INTUIT_METHOD;
3357 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
3361 const char *proto = SvPVX_const(cv);
3372 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3373 /* start is the beginning of the possible filehandle/object,
3374 * and s is the end of it
3375 * tmpbuf is a copy of it
3378 if (*start == '$') {
3379 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
3380 isUPPER(*PL_tokenbuf))
3383 len = start - SvPVX(PL_linestr);
3387 start = SvPVX(PL_linestr) + len;
3391 return *s == '(' ? FUNCMETH : METHOD;
3393 if (!keyword(tmpbuf, len, 0)) {
3394 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
3398 soff = s - SvPVX(PL_linestr);
3402 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
3403 if (indirgv && GvCVu(indirgv))
3405 /* filehandle or package name makes it a method */
3406 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
3408 soff = s - SvPVX(PL_linestr);
3411 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
3412 return 0; /* no assumptions -- "=>" quotes bearword */
3414 start_force(PL_curforce);
3415 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
3416 S_newSV_maybe_utf8(aTHX_ tmpbuf, len));
3417 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
3419 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
3424 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
3426 return *s == '(' ? FUNCMETH : METHOD;
3432 /* Encoded script support. filter_add() effectively inserts a
3433 * 'pre-processing' function into the current source input stream.
3434 * Note that the filter function only applies to the current source file
3435 * (e.g., it will not affect files 'require'd or 'use'd by this one).
3437 * The datasv parameter (which may be NULL) can be used to pass
3438 * private data to this instance of the filter. The filter function
3439 * can recover the SV using the FILTER_DATA macro and use it to
3440 * store private buffers and state information.
3442 * The supplied datasv parameter is upgraded to a PVIO type
3443 * and the IoDIRP/IoANY field is used to store the function pointer,
3444 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
3445 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
3446 * private use must be set using malloc'd pointers.
3450 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
3459 if (!PL_rsfp_filters)
3460 PL_rsfp_filters = newAV();
3463 SvUPGRADE(datasv, SVt_PVIO);
3464 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
3465 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
3466 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
3467 FPTR2DPTR(void *, IoANY(datasv)),
3468 SvPV_nolen(datasv)));
3469 av_unshift(PL_rsfp_filters, 1);
3470 av_store(PL_rsfp_filters, 0, datasv) ;
3475 /* Delete most recently added instance of this filter function. */
3477 Perl_filter_del(pTHX_ filter_t funcp)
3482 PERL_ARGS_ASSERT_FILTER_DEL;
3485 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
3486 FPTR2DPTR(void*, funcp)));
3488 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
3490 /* if filter is on top of stack (usual case) just pop it off */
3491 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
3492 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
3493 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
3494 IoANY(datasv) = (void *)NULL;
3495 sv_free(av_pop(PL_rsfp_filters));
3499 /* we need to search for the correct entry and clear it */
3500 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
3504 /* Invoke the idxth filter function for the current rsfp. */
3505 /* maxlen 0 = read one text line */
3507 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
3512 /* This API is bad. It should have been using unsigned int for maxlen.
3513 Not sure if we want to change the API, but if not we should sanity
3514 check the value here. */
3515 const unsigned int correct_length
3524 PERL_ARGS_ASSERT_FILTER_READ;
3526 if (!PL_parser || !PL_rsfp_filters)
3528 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
3529 /* Provide a default input filter to make life easy. */
3530 /* Note that we append to the line. This is handy. */
3531 DEBUG_P(PerlIO_printf(Perl_debug_log,
3532 "filter_read %d: from rsfp\n", idx));
3533 if (correct_length) {
3536 const int old_len = SvCUR(buf_sv);
3538 /* ensure buf_sv is large enough */
3539 SvGROW(buf_sv, (STRLEN)(old_len + correct_length + 1)) ;
3540 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
3541 correct_length)) <= 0) {
3542 if (PerlIO_error(PL_rsfp))
3543 return -1; /* error */
3545 return 0 ; /* end of file */
3547 SvCUR_set(buf_sv, old_len + len) ;
3548 SvPVX(buf_sv)[old_len + len] = '\0';
3551 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
3552 if (PerlIO_error(PL_rsfp))
3553 return -1; /* error */
3555 return 0 ; /* end of file */
3558 return SvCUR(buf_sv);
3560 /* Skip this filter slot if filter has been deleted */
3561 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
3562 DEBUG_P(PerlIO_printf(Perl_debug_log,
3563 "filter_read %d: skipped (filter deleted)\n",
3565 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
3567 /* Get function pointer hidden within datasv */
3568 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
3569 DEBUG_P(PerlIO_printf(Perl_debug_log,
3570 "filter_read %d: via function %p (%s)\n",
3571 idx, (void*)datasv, SvPV_nolen_const(datasv)));
3572 /* Call function. The function is expected to */
3573 /* call "FILTER_READ(idx+1, buf_sv)" first. */
3574 /* Return: <0:error, =0:eof, >0:not eof */
3575 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
3579 S_filter_gets(pTHX_ register SV *sv, STRLEN append)
3583 PERL_ARGS_ASSERT_FILTER_GETS;
3585 #ifdef PERL_CR_FILTER
3586 if (!PL_rsfp_filters) {
3587 filter_add(S_cr_textfilter,NULL);
3590 if (PL_rsfp_filters) {
3592 SvCUR_set(sv, 0); /* start with empty line */
3593 if (FILTER_READ(0, sv, 0) > 0)
3594 return ( SvPVX(sv) ) ;
3599 return (sv_gets(sv, PL_rsfp, append));
3603 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
3608 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
3610 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
3614 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
3615 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3617 return GvHV(gv); /* Foo:: */
3620 /* use constant CLASS => 'MyClass' */
3621 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3622 if (gv && GvCV(gv)) {
3623 SV * const sv = cv_const_sv(GvCV(gv));
3625 pkgname = SvPV_const(sv, len);
3628 return gv_stashpvn(pkgname, len, 0);
3632 * S_readpipe_override
3633 * Check whether readpipe() is overriden, and generates the appropriate
3634 * optree, provided sublex_start() is called afterwards.
3637 S_readpipe_override(pTHX)
3640 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3641 pl_yylval.ival = OP_BACKTICK;
3643 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3645 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3646 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3647 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3649 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3650 append_elem(OP_LIST,
3651 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3652 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3659 * The intent of this yylex wrapper is to minimize the changes to the
3660 * tokener when we aren't interested in collecting madprops. It remains
3661 * to be seen how successful this strategy will be...
3668 char *s = PL_bufptr;
3670 /* make sure PL_thiswhite is initialized */
3674 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3675 if (PL_pending_ident)
3676 return S_pending_ident(aTHX);
3678 /* previous token ate up our whitespace? */
3679 if (!PL_lasttoke && PL_nextwhite) {
3680 PL_thiswhite = PL_nextwhite;
3684 /* isolate the token, and figure out where it is without whitespace */
3685 PL_realtokenstart = -1;
3689 assert(PL_curforce < 0);
3691 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3692 if (!PL_thistoken) {
3693 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3694 PL_thistoken = newSVpvs("");
3696 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3697 PL_thistoken = newSVpvn(tstart, s - tstart);
3700 if (PL_thismad) /* install head */
3701 CURMAD('X', PL_thistoken);
3704 /* last whitespace of a sublex? */
3705 if (optype == ')' && PL_endwhite) {
3706 CURMAD('X', PL_endwhite);
3711 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3712 if (!PL_thiswhite && !PL_endwhite && !optype) {
3713 sv_free(PL_thistoken);
3718 /* put off final whitespace till peg */
3719 if (optype == ';' && !PL_rsfp) {
3720 PL_nextwhite = PL_thiswhite;
3723 else if (PL_thisopen) {
3724 CURMAD('q', PL_thisopen);
3726 sv_free(PL_thistoken);
3730 /* Store actual token text as madprop X */
3731 CURMAD('X', PL_thistoken);
3735 /* add preceding whitespace as madprop _ */
3736 CURMAD('_', PL_thiswhite);
3740 /* add quoted material as madprop = */
3741 CURMAD('=', PL_thisstuff);
3745 /* add terminating quote as madprop Q */
3746 CURMAD('Q', PL_thisclose);
3750 /* special processing based on optype */
3754 /* opval doesn't need a TOKEN since it can already store mp */
3764 if (pl_yylval.opval)
3765 append_madprops(PL_thismad, pl_yylval.opval, 0);
3773 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3782 /* remember any fake bracket that lexer is about to discard */
3783 if (PL_lex_brackets == 1 &&
3784 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3787 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3790 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3791 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3794 break; /* don't bother looking for trailing comment */
3803 /* attach a trailing comment to its statement instead of next token */
3807 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3809 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3811 if (*s == '\n' || *s == '#') {
3812 while (s < PL_bufend && *s != '\n')
3816 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3817 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3834 /* Create new token struct. Note: opvals return early above. */
3835 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3842 S_tokenize_use(pTHX_ int is_use, char *s) {
3845 PERL_ARGS_ASSERT_TOKENIZE_USE;
3847 if (PL_expect != XSTATE)
3848 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3849 is_use ? "use" : "no"));
3851 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3852 s = force_version(s, TRUE);
3853 if (*s == ';' || *s == '}'
3854 || (s = SKIPSPACE1(s), (*s == ';' || *s == '}'))) {
3855 start_force(PL_curforce);
3856 NEXTVAL_NEXTTOKE.opval = NULL;
3859 else if (*s == 'v') {
3860 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3861 s = force_version(s, FALSE);
3865 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3866 s = force_version(s, FALSE);
3868 pl_yylval.ival = is_use;
3872 static const char* const exp_name[] =
3873 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3874 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3881 Works out what to call the token just pulled out of the input
3882 stream. The yacc parser takes care of taking the ops we return and
3883 stitching them into a tree.
3889 if read an identifier
3890 if we're in a my declaration
3891 croak if they tried to say my($foo::bar)
3892 build the ops for a my() declaration
3893 if it's an access to a my() variable
3894 are we in a sort block?
3895 croak if my($a); $a <=> $b
3896 build ops for access to a my() variable
3897 if in a dq string, and they've said @foo and we can't find @foo
3899 build ops for a bareword
3900 if we already built the token before, use it.
3905 #pragma segment Perl_yylex
3911 register char *s = PL_bufptr;
3917 /* orig_keyword, gvp, and gv are initialized here because
3918 * jump to the label just_a_word_zero can bypass their
3919 * initialization later. */
3920 I32 orig_keyword = 0;
3925 SV* tmp = newSVpvs("");
3926 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3927 (IV)CopLINE(PL_curcop),
3928 lex_state_names[PL_lex_state],
3929 exp_name[PL_expect],
3930 pv_display(tmp, s, strlen(s), 0, 60));
3933 /* check if there's an identifier for us to look at */
3934 if (PL_pending_ident)
3935 return REPORT(S_pending_ident(aTHX));
3937 /* no identifier pending identification */
3939 switch (PL_lex_state) {
3941 case LEX_NORMAL: /* Some compilers will produce faster */
3942 case LEX_INTERPNORMAL: /* code if we comment these out. */
3946 /* when we've already built the next token, just pull it out of the queue */
3950 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3952 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3953 PL_nexttoke[PL_lasttoke].next_mad = 0;
3954 if (PL_thismad && PL_thismad->mad_key == '_') {
3955 PL_thiswhite = MUTABLE_SV(PL_thismad->mad_val);
3956 PL_thismad->mad_val = 0;
3957 mad_free(PL_thismad);
3962 PL_lex_state = PL_lex_defer;
3963 PL_expect = PL_lex_expect;
3964 PL_lex_defer = LEX_NORMAL;
3965 if (!PL_nexttoke[PL_lasttoke].next_type)
3970 pl_yylval = PL_nextval[PL_nexttoke];
3972 PL_lex_state = PL_lex_defer;
3973 PL_expect = PL_lex_expect;
3974 PL_lex_defer = LEX_NORMAL;
3978 /* FIXME - can these be merged? */
3979 return(PL_nexttoke[PL_lasttoke].next_type);
3981 return REPORT(PL_nexttype[PL_nexttoke]);
3984 /* interpolated case modifiers like \L \U, including \Q and \E.
3985 when we get here, PL_bufptr is at the \
3987 case LEX_INTERPCASEMOD:
3989 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3990 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3992 /* handle \E or end of string */
3993 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3995 if (PL_lex_casemods) {
3996 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3997 PL_lex_casestack[PL_lex_casemods] = '\0';
3999 if (PL_bufptr != PL_bufend
4000 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
4002 PL_lex_state = LEX_INTERPCONCAT;
4005 PL_thistoken = newSVpvs("\\E");
4011 while (PL_bufptr != PL_bufend &&
4012 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
4014 PL_thiswhite = newSVpvs("");
4015 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
4019 if (PL_bufptr != PL_bufend)
4022 PL_lex_state = LEX_INTERPCONCAT;
4026 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4027 "### Saw case modifier\n"); });
4029 if (s[1] == '\\' && s[2] == 'E') {
4032 PL_thiswhite = newSVpvs("");
4033 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
4036 PL_lex_state = LEX_INTERPCONCAT;
4041 if (!PL_madskills) /* when just compiling don't need correct */
4042 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
4043 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
4044 if ((*s == 'L' || *s == 'U') &&
4045 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
4046 PL_lex_casestack[--PL_lex_casemods] = '\0';
4049 if (PL_lex_casemods > 10)
4050 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
4051 PL_lex_casestack[PL_lex_casemods++] = *s;
4052 PL_lex_casestack[PL_lex_casemods] = '\0';
4053 PL_lex_state = LEX_INTERPCONCAT;
4054 start_force(PL_curforce);
4055 NEXTVAL_NEXTTOKE.ival = 0;
4057 start_force(PL_curforce);
4059 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
4061 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
4063 NEXTVAL_NEXTTOKE.ival = OP_LC;
4065 NEXTVAL_NEXTTOKE.ival = OP_UC;
4067 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
4069 Perl_croak(aTHX_ "panic: yylex");
4071 SV* const tmpsv = newSVpvs("\\ ");
4072 /* replace the space with the character we want to escape
4074 SvPVX(tmpsv)[1] = *s;
4080 if (PL_lex_starts) {
4086 sv_free(PL_thistoken);
4087 PL_thistoken = newSVpvs("");
4090 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4091 if (PL_lex_casemods == 1 && PL_lex_inpat)
4100 case LEX_INTERPPUSH:
4101 return REPORT(sublex_push());
4103 case LEX_INTERPSTART:
4104 if (PL_bufptr == PL_bufend)
4105 return REPORT(sublex_done());
4106 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4107 "### Interpolated variable\n"); });
4109 PL_lex_dojoin = (*PL_bufptr == '@');
4110 PL_lex_state = LEX_INTERPNORMAL;
4111 if (PL_lex_dojoin) {
4112 start_force(PL_curforce);
4113 NEXTVAL_NEXTTOKE.ival = 0;
4115 start_force(PL_curforce);
4116 force_ident("\"", '$');
4117 start_force(PL_curforce);
4118 NEXTVAL_NEXTTOKE.ival = 0;
4120 start_force(PL_curforce);
4121 NEXTVAL_NEXTTOKE.ival = 0;
4123 start_force(PL_curforce);
4124 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
4127 if (PL_lex_starts++) {
4132 sv_free(PL_thistoken);
4133 PL_thistoken = newSVpvs("");
4136 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4137 if (!PL_lex_casemods && PL_lex_inpat)
4144 case LEX_INTERPENDMAYBE:
4145 if (intuit_more(PL_bufptr)) {
4146 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
4152 if (PL_lex_dojoin) {
4153 PL_lex_dojoin = FALSE;
4154 PL_lex_state = LEX_INTERPCONCAT;
4158 sv_free(PL_thistoken);
4159 PL_thistoken = newSVpvs("");
4164 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
4165 && SvEVALED(PL_lex_repl))
4167 if (PL_bufptr != PL_bufend)
4168 Perl_croak(aTHX_ "Bad evalled substitution pattern");
4172 case LEX_INTERPCONCAT:
4174 if (PL_lex_brackets)
4175 Perl_croak(aTHX_ "panic: INTERPCONCAT");
4177 if (PL_bufptr == PL_bufend)
4178 return REPORT(sublex_done());
4180 if (SvIVX(PL_linestr) == '\'') {
4181 SV *sv = newSVsv(PL_linestr);
4184 else if ( PL_hints & HINT_NEW_RE )
4185 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
4186 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4190 s = scan_const(PL_bufptr);
4192 PL_lex_state = LEX_INTERPCASEMOD;
4194 PL_lex_state = LEX_INTERPSTART;
4197 if (s != PL_bufptr) {
4198 start_force(PL_curforce);
4200 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
4202 NEXTVAL_NEXTTOKE = pl_yylval;
4205 if (PL_lex_starts++) {
4209 sv_free(PL_thistoken);
4210 PL_thistoken = newSVpvs("");
4213 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4214 if (!PL_lex_casemods && PL_lex_inpat)
4227 PL_lex_state = LEX_NORMAL;
4228 s = scan_formline(PL_bufptr);
4229 if (!PL_lex_formbrack)
4235 PL_oldoldbufptr = PL_oldbufptr;
4241 sv_free(PL_thistoken);
4244 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
4248 if (isIDFIRST_lazy_if(s,UTF))
4251 unsigned char c = *s;
4252 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
4253 if (len > UNRECOGNIZED_PRECEDE_COUNT) {
4254 d = UTF ? (char *) Perl_utf8_hop(aTHX_ (U8 *) s, -UNRECOGNIZED_PRECEDE_COUNT) : s - UNRECOGNIZED_PRECEDE_COUNT;
4259 Perl_croak(aTHX_ "Unrecognized character \\x%02X; marked by <-- HERE after %s<-- HERE near column %d", c, d, (int) len + 1);
4263 goto fake_eof; /* emulate EOF on ^D or ^Z */
4272 if (PL_lex_brackets) {
4273 yyerror((const char *)
4275 ? "Format not terminated"
4276 : "Missing right curly or square bracket"));
4278 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4279 "### Tokener got EOF\n");
4283 if (s++ < PL_bufend)
4284 goto retry; /* ignore stray nulls */
4287 if (!PL_in_eval && !PL_preambled) {
4288 PL_preambled = TRUE;
4294 /* Generate a string of Perl code to load the debugger.
4295 * If PERL5DB is set, it will return the contents of that,
4296 * otherwise a compile-time require of perl5db.pl. */
4298 const char * const pdb = PerlEnv_getenv("PERL5DB");
4301 sv_setpv(PL_linestr, pdb);
4302 sv_catpvs(PL_linestr,";");
4304 SETERRNO(0,SS_NORMAL);
4305 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
4308 sv_setpvs(PL_linestr,"");
4309 if (PL_preambleav) {
4310 SV **svp = AvARRAY(PL_preambleav);
4311 SV **const end = svp + AvFILLp(PL_preambleav);
4313 sv_catsv(PL_linestr, *svp);
4315 sv_catpvs(PL_linestr, ";");
4317 sv_free(MUTABLE_SV(PL_preambleav));
4318 PL_preambleav = NULL;
4321 sv_catpvs(PL_linestr,
4322 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
4323 if (PL_minus_n || PL_minus_p) {
4324 sv_catpvs(PL_linestr, "LINE: while (<>) {"/*}*/);
4326 sv_catpvs(PL_linestr,"chomp;");
4329 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
4330 || *PL_splitstr == '"')
4331 && strchr(PL_splitstr + 1, *PL_splitstr))
4332 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
4334 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
4335 bytes can be used as quoting characters. :-) */
4336 const char *splits = PL_splitstr;
4337 sv_catpvs(PL_linestr, "our @F=split(q\0");
4340 if (*splits == '\\')
4341 sv_catpvn(PL_linestr, splits, 1);
4342 sv_catpvn(PL_linestr, splits, 1);
4343 } while (*splits++);
4344 /* This loop will embed the trailing NUL of
4345 PL_linestr as the last thing it does before
4347 sv_catpvs(PL_linestr, ");");
4351 sv_catpvs(PL_linestr,"our @F=split(' ');");
4354 sv_catpvs(PL_linestr, "\n");
4355 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4356 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4357 PL_last_lop = PL_last_uni = NULL;
4358 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
4359 update_debugger_info(PL_linestr, NULL, 0);
4364 bof = PL_rsfp ? TRUE : FALSE;
4367 fake_eof = LEX_FAKE_EOF;
4369 PL_bufptr = PL_bufend;
4370 CopLINE_inc(PL_curcop);
4371 if (!lex_next_chunk(fake_eof)) {
4372 CopLINE_dec(PL_curcop);
4374 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
4376 CopLINE_dec(PL_curcop);
4379 PL_realtokenstart = -1;
4382 /* If it looks like the start of a BOM or raw UTF-16,
4383 * check if it in fact is. */
4384 if (bof && PL_rsfp &&
4389 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
4391 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4392 s = swallow_bom((U8*)s);
4396 /* Incest with pod. */
4399 sv_catsv(PL_thiswhite, PL_linestr);
4401 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
4402 sv_setpvs(PL_linestr, "");
4403 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4404 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4405 PL_last_lop = PL_last_uni = NULL;
4406 PL_doextract = FALSE;
4411 } while (PL_doextract);
4412 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
4413 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4414 PL_last_lop = PL_last_uni = NULL;
4415 if (CopLINE(PL_curcop) == 1) {
4416 while (s < PL_bufend && isSPACE(*s))
4418 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
4422 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
4426 if (*s == '#' && *(s+1) == '!')
4428 #ifdef ALTERNATE_SHEBANG
4430 static char const as[] = ALTERNATE_SHEBANG;
4431 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
4432 d = s + (sizeof(as) - 1);
4434 #endif /* ALTERNATE_SHEBANG */
4443 while (*d && !isSPACE(*d))
4447 #ifdef ARG_ZERO_IS_SCRIPT
4448 if (ipathend > ipath) {
4450 * HP-UX (at least) sets argv[0] to the script name,
4451 * which makes $^X incorrect. And Digital UNIX and Linux,
4452 * at least, set argv[0] to the basename of the Perl
4453 * interpreter. So, having found "#!", we'll set it right.
4455 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
4457 assert(SvPOK(x) || SvGMAGICAL(x));
4458 if (sv_eq(x, CopFILESV(PL_curcop))) {
4459 sv_setpvn(x, ipath, ipathend - ipath);
4465 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
4466 const char * const lstart = SvPV_const(x,llen);
4468 bstart += blen - llen;
4469 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
4470 sv_setpvn(x, ipath, ipathend - ipath);
4475 TAINT_NOT; /* $^X is always tainted, but that's OK */
4477 #endif /* ARG_ZERO_IS_SCRIPT */
4482 d = instr(s,"perl -");
4484 d = instr(s,"perl");
4486 /* avoid getting into infinite loops when shebang
4487 * line contains "Perl" rather than "perl" */
4489 for (d = ipathend-4; d >= ipath; --d) {
4490 if ((*d == 'p' || *d == 'P')
4491 && !ibcmp(d, "perl", 4))
4501 #ifdef ALTERNATE_SHEBANG
4503 * If the ALTERNATE_SHEBANG on this system starts with a
4504 * character that can be part of a Perl expression, then if
4505 * we see it but not "perl", we're probably looking at the
4506 * start of Perl code, not a request to hand off to some
4507 * other interpreter. Similarly, if "perl" is there, but
4508 * not in the first 'word' of the line, we assume the line
4509 * contains the start of the Perl program.
4511 if (d && *s != '#') {
4512 const char *c = ipath;
4513 while (*c && !strchr("; \t\r\n\f\v#", *c))
4516 d = NULL; /* "perl" not in first word; ignore */
4518 *s = '#'; /* Don't try to parse shebang line */
4520 #endif /* ALTERNATE_SHEBANG */
4525 !instr(s,"indir") &&
4526 instr(PL_origargv[0],"perl"))
4533 while (s < PL_bufend && isSPACE(*s))
4535 if (s < PL_bufend) {
4536 Newx(newargv,PL_origargc+3,char*);
4538 while (s < PL_bufend && !isSPACE(*s))
4541 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
4544 newargv = PL_origargv;
4547 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
4549 Perl_croak(aTHX_ "Can't exec %s", ipath);
4552 while (*d && !isSPACE(*d))
4554 while (SPACE_OR_TAB(*d))
4558 const bool switches_done = PL_doswitches;
4559 const U32 oldpdb = PL_perldb;
4560 const bool oldn = PL_minus_n;
4561 const bool oldp = PL_minus_p;
4565 bool baduni = FALSE;
4567 const char *d2 = d1 + 1;
4568 if (parse_unicode_opts((const char **)&d2)
4572 if (baduni || *d1 == 'M' || *d1 == 'm') {
4573 const char * const m = d1;
4574 while (*d1 && !isSPACE(*d1))
4576 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
4579 d1 = moreswitches(d1);
4581 if (PL_doswitches && !switches_done) {
4582 int argc = PL_origargc;
4583 char **argv = PL_origargv;
4586 } while (argc && argv[0][0] == '-' && argv[0][1]);
4587 init_argv_symbols(argc,argv);
4589 if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
4590 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
4591 /* if we have already added "LINE: while (<>) {",
4592 we must not do it again */
4594 sv_setpvs(PL_linestr, "");
4595 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4596 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4597 PL_last_lop = PL_last_uni = NULL;
4598 PL_preambled = FALSE;
4599 if (PERLDB_LINE || PERLDB_SAVESRC)
4600 (void)gv_fetchfile(PL_origfilename);
4607 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4609 PL_lex_state = LEX_FORMLINE;
4614 #ifdef PERL_STRICT_CR
4615 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4617 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4619 case ' ': case '\t': case '\f': case 013:
4621 PL_realtokenstart = -1;
4623 PL_thiswhite = newSVpvs("");
4624 sv_catpvn(PL_thiswhite, s, 1);
4631 PL_realtokenstart = -1;
4635 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4636 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4637 /* handle eval qq[#line 1 "foo"\n ...] */
4638 CopLINE_dec(PL_curcop);
4641 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4643 if (!PL_in_eval || PL_rsfp)
4648 while (d < PL_bufend && *d != '\n')
4652 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4653 Perl_croak(aTHX_ "panic: input overflow");
4656 PL_thiswhite = newSVpvn(s, d - s);
4661 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4663 PL_lex_state = LEX_FORMLINE;
4669 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4670 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4673 TOKEN(PEG); /* make sure any #! line is accessible */
4678 /* if (PL_madskills && PL_lex_formbrack) { */
4680 while (d < PL_bufend && *d != '\n')
4684 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4685 Perl_croak(aTHX_ "panic: input overflow");
4686 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4688 PL_thiswhite = newSVpvs("");
4689 if (CopLINE(PL_curcop) == 1) {
4690 sv_setpvs(PL_thiswhite, "");
4693 sv_catpvn(PL_thiswhite, s, d - s);
4707 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4715 while (s < PL_bufend && SPACE_OR_TAB(*s))
4718 if (strnEQ(s,"=>",2)) {
4719 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4720 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4721 OPERATOR('-'); /* unary minus */
4723 PL_last_uni = PL_oldbufptr;
4725 case 'r': ftst = OP_FTEREAD; break;
4726 case 'w': ftst = OP_FTEWRITE; break;
4727 case 'x': ftst = OP_FTEEXEC; break;
4728 case 'o': ftst = OP_FTEOWNED; break;
4729 case 'R': ftst = OP_FTRREAD; break;
4730 case 'W': ftst = OP_FTRWRITE; break;
4731 case 'X': ftst = OP_FTREXEC; break;
4732 case 'O': ftst = OP_FTROWNED; break;
4733 case 'e': ftst = OP_FTIS; break;
4734 case 'z': ftst = OP_FTZERO; break;
4735 case 's': ftst = OP_FTSIZE; break;
4736 case 'f': ftst = OP_FTFILE; break;
4737 case 'd': ftst = OP_FTDIR; break;
4738 case 'l': ftst = OP_FTLINK; break;
4739 case 'p': ftst = OP_FTPIPE; break;
4740 case 'S': ftst = OP_FTSOCK; break;
4741 case 'u': ftst = OP_FTSUID; break;
4742 case 'g': ftst = OP_FTSGID; break;
4743 case 'k': ftst = OP_FTSVTX; break;
4744 case 'b': ftst = OP_FTBLK; break;
4745 case 'c': ftst = OP_FTCHR; break;
4746 case 't': ftst = OP_FTTTY; break;
4747 case 'T': ftst = OP_FTTEXT; break;
4748 case 'B': ftst = OP_FTBINARY; break;
4749 case 'M': case 'A': case 'C':
4750 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4752 case 'M': ftst = OP_FTMTIME; break;
4753 case 'A': ftst = OP_FTATIME; break;
4754 case 'C': ftst = OP_FTCTIME; break;
4762 PL_last_lop_op = (OPCODE)ftst;
4763 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4764 "### Saw file test %c\n", (int)tmp);
4769 /* Assume it was a minus followed by a one-letter named
4770 * subroutine call (or a -bareword), then. */
4771 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4772 "### '-%c' looked like a file test but was not\n",
4779 const char tmp = *s++;
4782 if (PL_expect == XOPERATOR)
4787 else if (*s == '>') {
4790 if (isIDFIRST_lazy_if(s,UTF)) {
4791 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4799 if (PL_expect == XOPERATOR)
4802 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4804 OPERATOR('-'); /* unary minus */
4810 const char tmp = *s++;
4813 if (PL_expect == XOPERATOR)
4818 if (PL_expect == XOPERATOR)
4821 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4828 if (PL_expect != XOPERATOR) {
4829 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4830 PL_expect = XOPERATOR;
4831 force_ident(PL_tokenbuf, '*');
4844 if (PL_expect == XOPERATOR) {
4848 PL_tokenbuf[0] = '%';
4849 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4850 sizeof PL_tokenbuf - 1, FALSE);
4851 if (!PL_tokenbuf[1]) {
4854 PL_pending_ident = '%';
4863 const char tmp = *s++;
4868 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4875 const char tmp = *s++;
4881 goto just_a_word_zero_gv;
4884 switch (PL_expect) {
4890 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4892 PL_bufptr = s; /* update in case we back off */
4894 deprecate(":= for an empty attribute list");
4901 PL_expect = XTERMBLOCK;
4904 stuffstart = s - SvPVX(PL_linestr) - 1;
4908 while (isIDFIRST_lazy_if(s,UTF)) {
4911 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4912 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4913 if (tmp < 0) tmp = -tmp;
4928 sv = newSVpvn(s, len);
4930 d = scan_str(d,TRUE,TRUE);
4932 /* MUST advance bufptr here to avoid bogus
4933 "at end of line" context messages from yyerror().
4935 PL_bufptr = s + len;
4936 yyerror("Unterminated attribute parameter in attribute list");
4940 return REPORT(0); /* EOF indicator */
4944 sv_catsv(sv, PL_lex_stuff);
4945 attrs = append_elem(OP_LIST, attrs,
4946 newSVOP(OP_CONST, 0, sv));
4947 SvREFCNT_dec(PL_lex_stuff);
4948 PL_lex_stuff = NULL;
4951 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4953 if (PL_in_my == KEY_our) {
4954 deprecate(":unique");
4957 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4960 /* NOTE: any CV attrs applied here need to be part of
4961 the CVf_BUILTIN_ATTRS define in cv.h! */
4962 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4964 CvLVALUE_on(PL_compcv);
4966 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4968 deprecate(":locked");
4970 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4972 CvMETHOD_on(PL_compcv);
4974 /* After we've set the flags, it could be argued that
4975 we don't need to do the attributes.pm-based setting
4976 process, and shouldn't bother appending recognized
4977 flags. To experiment with that, uncomment the
4978 following "else". (Note that's already been
4979 uncommented. That keeps the above-applied built-in
4980 attributes from being intercepted (and possibly
4981 rejected) by a package's attribute routines, but is
4982 justified by the performance win for the common case
4983 of applying only built-in attributes.) */
4985 attrs = append_elem(OP_LIST, attrs,
4986 newSVOP(OP_CONST, 0,
4990 if (*s == ':' && s[1] != ':')
4993 break; /* require real whitespace or :'s */
4994 /* XXX losing whitespace on sequential attributes here */
4998 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4999 if (*s != ';' && *s != '}' && *s != tmp
5000 && (tmp != '=' || *s != ')')) {
5001 const char q = ((*s == '\'') ? '"' : '\'');
5002 /* If here for an expression, and parsed no attrs, back
5004 if (tmp == '=' && !attrs) {
5008 /* MUST advance bufptr here to avoid bogus "at end of line"
5009 context messages from yyerror().
5012 yyerror( (const char *)
5014 ? Perl_form(aTHX_ "Invalid separator character "
5015 "%c%c%c in attribute list", q, *s, q)
5016 : "Unterminated attribute list" ) );
5024 start_force(PL_curforce);
5025 NEXTVAL_NEXTTOKE.opval = attrs;
5026 CURMAD('_', PL_nextwhite);
5031 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
5032 (s - SvPVX(PL_linestr)) - stuffstart);
5040 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
5041 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
5049 const char tmp = *s++;
5054 const char tmp = *s++;
5062 if (PL_lex_brackets <= 0)
5063 yyerror("Unmatched right square bracket");
5066 if (PL_lex_state == LEX_INTERPNORMAL) {
5067 if (PL_lex_brackets == 0) {
5068 if (*s == '-' && s[1] == '>')
5069 PL_lex_state = LEX_INTERPENDMAYBE;
5070 else if (*s != '[' && *s != '{')
5071 PL_lex_state = LEX_INTERPEND;
5078 if (PL_lex_brackets > 100) {
5079 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
5081 switch (PL_expect) {
5083 if (PL_lex_formbrack) {
5087 if (PL_oldoldbufptr == PL_last_lop)
5088 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
5090 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5091 OPERATOR(HASHBRACK);
5093 while (s < PL_bufend && SPACE_OR_TAB(*s))
5096 PL_tokenbuf[0] = '\0';
5097 if (d < PL_bufend && *d == '-') {
5098 PL_tokenbuf[0] = '-';
5100 while (d < PL_bufend && SPACE_OR_TAB(*d))
5103 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
5104 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
5106 while (d < PL_bufend && SPACE_OR_TAB(*d))
5109 const char minus = (PL_tokenbuf[0] == '-');
5110 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
5118 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
5123 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5128 if (PL_oldoldbufptr == PL_last_lop)
5129 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
5131 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5134 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
5136 /* This hack is to get the ${} in the message. */
5138 yyerror("syntax error");
5141 OPERATOR(HASHBRACK);
5143 /* This hack serves to disambiguate a pair of curlies
5144 * as being a block or an anon hash. Normally, expectation
5145 * determines that, but in cases where we're not in a
5146 * position to expect anything in particular (like inside
5147 * eval"") we have to resolve the ambiguity. This code
5148 * covers the case where the first term in the curlies is a
5149 * quoted string. Most other cases need to be explicitly
5150 * disambiguated by prepending a "+" before the opening
5151 * curly in order to force resolution as an anon hash.
5153 * XXX should probably propagate the outer expectation
5154 * into eval"" to rely less on this hack, but that could
5155 * potentially break current behavior of eval"".
5159 if (*s == '\'' || *s == '"' || *s == '`') {
5160 /* common case: get past first string, handling escapes */
5161 for (t++; t < PL_bufend && *t != *s;)
5162 if (*t++ == '\\' && (*t == '\\' || *t == *s))
5166 else if (*s == 'q') {
5169 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
5172 /* skip q//-like construct */
5174 char open, close, term;
5177 while (t < PL_bufend && isSPACE(*t))
5179 /* check for q => */
5180 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
5181 OPERATOR(HASHBRACK);
5185 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
5189 for (t++; t < PL_bufend; t++) {
5190 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
5192 else if (*t == open)
5196 for (t++; t < PL_bufend; t++) {
5197 if (*t == '\\' && t+1 < PL_bufend)
5199 else if (*t == close && --brackets <= 0)
5201 else if (*t == open)
5208 /* skip plain q word */
5209 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
5212 else if (isALNUM_lazy_if(t,UTF)) {
5214 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
5217 while (t < PL_bufend && isSPACE(*t))
5219 /* if comma follows first term, call it an anon hash */
5220 /* XXX it could be a comma expression with loop modifiers */
5221 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
5222 || (*t == '=' && t[1] == '>')))
5223 OPERATOR(HASHBRACK);
5224 if (PL_expect == XREF)
5227 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
5233 pl_yylval.ival = CopLINE(PL_curcop);
5234 if (isSPACE(*s) || *s == '#')
5235 PL_copline = NOLINE; /* invalidate current command line number */
5240 if (PL_lex_brackets <= 0)
5241 yyerror("Unmatched right curly bracket");
5243 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
5244 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
5245 PL_lex_formbrack = 0;
5246 if (PL_lex_state == LEX_INTERPNORMAL) {
5247 if (PL_lex_brackets == 0) {
5248 if (PL_expect & XFAKEBRACK) {
5249 PL_expect &= XENUMMASK;
5250 PL_lex_state = LEX_INTERPEND;
5255 PL_thiswhite = newSVpvs("");
5256 sv_catpvs(PL_thiswhite,"}");
5259 return yylex(); /* ignore fake brackets */
5261 if (*s == '-' && s[1] == '>')
5262 PL_lex_state = LEX_INTERPENDMAYBE;
5263 else if (*s != '[' && *s != '{')
5264 PL_lex_state = LEX_INTERPEND;
5267 if (PL_expect & XFAKEBRACK) {
5268 PL_expect &= XENUMMASK;
5270 return yylex(); /* ignore fake brackets */
5272 start_force(PL_curforce);
5274 curmad('X', newSVpvn(s-1,1));
5275 CURMAD('_', PL_thiswhite);
5280 PL_thistoken = newSVpvs("");
5288 if (PL_expect == XOPERATOR) {
5289 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
5290 && isIDFIRST_lazy_if(s,UTF))
5292 CopLINE_dec(PL_curcop);
5293 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5294 CopLINE_inc(PL_curcop);
5299 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
5301 PL_expect = XOPERATOR;
5302 force_ident(PL_tokenbuf, '&');
5306 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
5318 const char tmp = *s++;
5325 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
5326 && strchr("+-*/%.^&|<",tmp))
5327 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5328 "Reversed %c= operator",(int)tmp);
5330 if (PL_expect == XSTATE && isALPHA(tmp) &&
5331 (s == PL_linestart+1 || s[-2] == '\n') )
5333 if (PL_in_eval && !PL_rsfp) {
5338 if (strnEQ(s,"=cut",4)) {
5354 PL_thiswhite = newSVpvs("");
5355 sv_catpvn(PL_thiswhite, PL_linestart,
5356 PL_bufend - PL_linestart);
5360 PL_doextract = TRUE;
5364 if (PL_lex_brackets < PL_lex_formbrack) {
5366 #ifdef PERL_STRICT_CR
5367 while (SPACE_OR_TAB(*t))
5369 while (SPACE_OR_TAB(*t) || *t == '\r')
5372 if (*t == '\n' || *t == '#') {
5383 const char tmp = *s++;
5385 /* was this !=~ where !~ was meant?
5386 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
5388 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
5389 const char *t = s+1;
5391 while (t < PL_bufend && isSPACE(*t))
5394 if (*t == '/' || *t == '?' ||
5395 ((*t == 'm' || *t == 's' || *t == 'y')
5396 && !isALNUM(t[1])) ||
5397 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
5398 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5399 "!=~ should be !~");
5409 if (PL_expect != XOPERATOR) {
5410 if (s[1] != '<' && !strchr(s,'>'))
5413 s = scan_heredoc(s);
5415 s = scan_inputsymbol(s);
5416 TERM(sublex_start());
5422 SHop(OP_LEFT_SHIFT);
5436 const char tmp = *s++;
5438 SHop(OP_RIGHT_SHIFT);
5439 else if (tmp == '=')
5448 if (PL_expect == XOPERATOR) {
5449 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5450 return deprecate_commaless_var_list();
5454 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
5455 PL_tokenbuf[0] = '@';
5456 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
5457 sizeof PL_tokenbuf - 1, FALSE);
5458 if (PL_expect == XOPERATOR)
5459 no_op("Array length", s);
5460 if (!PL_tokenbuf[1])
5462 PL_expect = XOPERATOR;
5463 PL_pending_ident = '#';
5467 PL_tokenbuf[0] = '$';
5468 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
5469 sizeof PL_tokenbuf - 1, FALSE);
5470 if (PL_expect == XOPERATOR)
5472 if (!PL_tokenbuf[1]) {
5474 yyerror("Final $ should be \\$ or $name");
5478 /* This kludge not intended to be bulletproof. */
5479 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
5480 pl_yylval.opval = newSVOP(OP_CONST, 0,
5481 newSViv(CopARYBASE_get(&PL_compiling)));
5482 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
5488 const char tmp = *s;
5489 if (PL_lex_state == LEX_NORMAL || PL_lex_brackets)
5492 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
5493 && intuit_more(s)) {
5495 PL_tokenbuf[0] = '@';
5496 if (ckWARN(WARN_SYNTAX)) {
5499 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
5502 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5503 while (t < PL_bufend && *t != ']')
5505 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5506 "Multidimensional syntax %.*s not supported",
5507 (int)((t - PL_bufptr) + 1), PL_bufptr);
5511 else if (*s == '{') {
5513 PL_tokenbuf[0] = '%';
5514 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
5515 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
5517 char tmpbuf[sizeof PL_tokenbuf];
5520 } while (isSPACE(*t));
5521 if (isIDFIRST_lazy_if(t,UTF)) {
5523 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
5527 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
5528 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5529 "You need to quote \"%s\"",
5536 PL_expect = XOPERATOR;
5537 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
5538 const bool islop = (PL_last_lop == PL_oldoldbufptr);
5539 if (!islop || PL_last_lop_op == OP_GREPSTART)
5540 PL_expect = XOPERATOR;
5541 else if (strchr("$@\"'`q", *s))
5542 PL_expect = XTERM; /* e.g. print $fh "foo" */
5543 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
5544 PL_expect = XTERM; /* e.g. print $fh &sub */
5545 else if (isIDFIRST_lazy_if(s,UTF)) {
5546 char tmpbuf[sizeof PL_tokenbuf];
5548 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5549 if ((t2 = keyword(tmpbuf, len, 0))) {
5550 /* binary operators exclude handle interpretations */
5562 PL_expect = XTERM; /* e.g. print $fh length() */
5567 PL_expect = XTERM; /* e.g. print $fh subr() */
5570 else if (isDIGIT(*s))
5571 PL_expect = XTERM; /* e.g. print $fh 3 */
5572 else if (*s == '.' && isDIGIT(s[1]))
5573 PL_expect = XTERM; /* e.g. print $fh .3 */
5574 else if ((*s == '?' || *s == '-' || *s == '+')
5575 && !isSPACE(s[1]) && s[1] != '=')
5576 PL_expect = XTERM; /* e.g. print $fh -1 */
5577 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
5579 PL_expect = XTERM; /* e.g. print $fh /.../
5580 XXX except DORDOR operator
5582 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
5584 PL_expect = XTERM; /* print $fh <<"EOF" */
5587 PL_pending_ident = '$';
5591 if (PL_expect == XOPERATOR)
5593 PL_tokenbuf[0] = '@';
5594 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
5595 if (!PL_tokenbuf[1]) {
5598 if (PL_lex_state == LEX_NORMAL)
5600 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5602 PL_tokenbuf[0] = '%';
5604 /* Warn about @ where they meant $. */
5605 if (*s == '[' || *s == '{') {
5606 if (ckWARN(WARN_SYNTAX)) {
5607 const char *t = s + 1;
5608 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5610 if (*t == '}' || *t == ']') {
5612 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5613 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5614 "Scalar value %.*s better written as $%.*s",
5615 (int)(t-PL_bufptr), PL_bufptr,
5616 (int)(t-PL_bufptr-1), PL_bufptr+1);
5621 PL_pending_ident = '@';
5624 case '/': /* may be division, defined-or, or pattern */
5625 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5629 case '?': /* may either be conditional or pattern */
5630 if (PL_expect == XOPERATOR) {
5638 /* A // operator. */
5648 /* Disable warning on "study /blah/" */
5649 if (PL_oldoldbufptr == PL_last_uni
5650 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5651 || memNE(PL_last_uni, "study", 5)
5652 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5655 s = scan_pat(s,OP_MATCH);
5656 TERM(sublex_start());
5660 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5661 #ifdef PERL_STRICT_CR
5664 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5666 && (s == PL_linestart || s[-1] == '\n') )
5668 PL_lex_formbrack = 0;
5672 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5676 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5682 pl_yylval.ival = OPf_SPECIAL;
5691 case '0': case '1': case '2': case '3': case '4':
5692 case '5': case '6': case '7': case '8': case '9':
5693 s = scan_num(s, &pl_yylval);
5694 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5695 if (PL_expect == XOPERATOR)
5700 s = scan_str(s,!!PL_madskills,FALSE);
5701 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5702 if (PL_expect == XOPERATOR) {
5703 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5704 return deprecate_commaless_var_list();
5711 pl_yylval.ival = OP_CONST;
5712 TERM(sublex_start());
5715 s = scan_str(s,!!PL_madskills,FALSE);
5716 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5717 if (PL_expect == XOPERATOR) {
5718 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5719 return deprecate_commaless_var_list();
5726 pl_yylval.ival = OP_CONST;
5727 /* FIXME. I think that this can be const if char *d is replaced by
5728 more localised variables. */
5729 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5730 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5731 pl_yylval.ival = OP_STRINGIFY;
5735 TERM(sublex_start());
5738 s = scan_str(s,!!PL_madskills,FALSE);
5739 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5740 if (PL_expect == XOPERATOR)
5741 no_op("Backticks",s);
5744 readpipe_override();
5745 TERM(sublex_start());
5749 if (PL_lex_inwhat && isDIGIT(*s))
5750 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5752 if (PL_expect == XOPERATOR)
5753 no_op("Backslash",s);
5757 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5758 char *start = s + 2;
5759 while (isDIGIT(*start) || *start == '_')
5761 if (*start == '.' && isDIGIT(start[1])) {
5762 s = scan_num(s, &pl_yylval);
5765 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5766 else if (!isALPHA(*start) && (PL_expect == XTERM
5767 || PL_expect == XREF || PL_expect == XSTATE
5768 || PL_expect == XTERMORDORDOR)) {
5769 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5771 s = scan_num(s, &pl_yylval);
5778 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5821 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5823 /* Some keywords can be followed by any delimiter, including ':' */
5824 anydelim = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5825 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5826 (PL_tokenbuf[0] == 'q' &&
5827 strchr("qwxr", PL_tokenbuf[1])))));
5829 /* x::* is just a word, unless x is "CORE" */
5830 if (!anydelim && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5834 while (d < PL_bufend && isSPACE(*d))
5835 d++; /* no comments skipped here, or s### is misparsed */
5837 /* Is this a word before a => operator? */
5838 if (*d == '=' && d[1] == '>') {
5841 = (OP*)newSVOP(OP_CONST, 0,
5842 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5843 pl_yylval.opval->op_private = OPpCONST_BARE;
5847 /* Check for plugged-in keyword */
5851 char *saved_bufptr = PL_bufptr;
5853 result = CALL_FPTR(PL_keyword_plugin)(aTHX_ PL_tokenbuf, len, &o);
5855 if (result == KEYWORD_PLUGIN_DECLINE) {
5856 /* not a plugged-in keyword */
5857 PL_bufptr = saved_bufptr;
5858 } else if (result == KEYWORD_PLUGIN_STMT) {
5859 pl_yylval.opval = o;
5862 return REPORT(PLUGSTMT);
5863 } else if (result == KEYWORD_PLUGIN_EXPR) {
5864 pl_yylval.opval = o;
5866 PL_expect = XOPERATOR;
5867 return REPORT(PLUGEXPR);
5869 Perl_croak(aTHX_ "Bad plugin affecting keyword '%s'",
5874 /* Check for built-in keyword */
5875 tmp = keyword(PL_tokenbuf, len, 0);
5877 /* Is this a label? */
5878 if (!anydelim && PL_expect == XSTATE
5879 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5881 Perl_croak(aTHX_ "Can't use keyword '%s' as a label", PL_tokenbuf);
5883 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5888 if (tmp < 0) { /* second-class keyword? */
5889 GV *ogv = NULL; /* override (winner) */
5890 GV *hgv = NULL; /* hidden (loser) */
5891 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5893 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5896 if (GvIMPORTED_CV(gv))
5898 else if (! CvMETHOD(cv))
5902 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5903 (gv = *gvp) && isGV_with_GP(gv) &&
5904 GvCVu(gv) && GvIMPORTED_CV(gv))
5911 tmp = 0; /* overridden by import or by GLOBAL */
5914 && -tmp==KEY_lock /* XXX generalizable kludge */
5917 tmp = 0; /* any sub overrides "weak" keyword */
5919 else { /* no override */
5921 if (tmp == KEY_dump) {
5922 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
5923 "dump() better written as CORE::dump()");
5927 if (hgv && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
5928 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5929 "Ambiguous call resolved as CORE::%s(), %s",
5930 GvENAME(hgv), "qualify as such or use &");
5937 default: /* not a keyword */
5938 /* Trade off - by using this evil construction we can pull the
5939 variable gv into the block labelled keylookup. If not, then
5940 we have to give it function scope so that the goto from the
5941 earlier ':' case doesn't bypass the initialisation. */
5943 just_a_word_zero_gv:
5951 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5955 SV *nextPL_nextwhite = 0;
5959 /* Get the rest if it looks like a package qualifier */
5961 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5963 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5966 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5967 *s == '\'' ? "'" : "::");
5972 if (PL_expect == XOPERATOR) {
5973 if (PL_bufptr == PL_linestart) {
5974 CopLINE_dec(PL_curcop);
5975 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5976 CopLINE_inc(PL_curcop);
5979 no_op("Bareword",s);
5982 /* Look for a subroutine with this name in current package,
5983 unless name is "Foo::", in which case Foo is a bearword
5984 (and a package name). */
5986 if (len > 2 && !PL_madskills &&
5987 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5989 if (ckWARN(WARN_BAREWORD)
5990 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5991 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5992 "Bareword \"%s\" refers to nonexistent package",
5995 PL_tokenbuf[len] = '\0';
6001 /* Mustn't actually add anything to a symbol table.
6002 But also don't want to "initialise" any placeholder
6003 constants that might already be there into full
6004 blown PVGVs with attached PVCV. */
6005 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
6006 GV_NOADD_NOINIT, SVt_PVCV);
6011 /* if we saw a global override before, get the right name */
6014 sv = newSVpvs("CORE::GLOBAL::");
6015 sv_catpv(sv,PL_tokenbuf);
6018 /* If len is 0, newSVpv does strlen(), which is correct.
6019 If len is non-zero, then it will be the true length,
6020 and so the scalar will be created correctly. */
6021 sv = newSVpv(PL_tokenbuf,len);
6024 if (PL_madskills && !PL_thistoken) {
6025 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
6026 PL_thistoken = newSVpvn(start,s - start);
6027 PL_realtokenstart = s - SvPVX(PL_linestr);
6031 /* Presume this is going to be a bareword of some sort. */
6034 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
6035 pl_yylval.opval->op_private = OPpCONST_BARE;
6036 /* UTF-8 package name? */
6037 if (UTF && !IN_BYTES &&
6038 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
6041 /* And if "Foo::", then that's what it certainly is. */
6048 OP *const_op = newSVOP(OP_CONST, 0, SvREFCNT_inc(sv));
6049 const_op->op_private = OPpCONST_BARE;
6050 rv2cv_op = newCVREF(0, const_op);
6052 if (rv2cv_op->op_type == OP_RV2CV &&
6053 (rv2cv_op->op_flags & OPf_KIDS)) {
6054 OP *rv_op = cUNOPx(rv2cv_op)->op_first;
6055 switch (rv_op->op_type) {
6057 SV *sv = cSVOPx_sv(rv_op);
6058 if (SvROK(sv) && SvTYPE(SvRV(sv)) == SVt_PVCV)
6062 GV *gv = cGVOPx_gv(rv_op);
6063 CV *maybe_cv = GvCVu(gv);
6064 if (maybe_cv && SvTYPE((SV*)maybe_cv) == SVt_PVCV)
6070 /* See if it's the indirect object for a list operator. */
6072 if (PL_oldoldbufptr &&
6073 PL_oldoldbufptr < PL_bufptr &&
6074 (PL_oldoldbufptr == PL_last_lop
6075 || PL_oldoldbufptr == PL_last_uni) &&
6076 /* NO SKIPSPACE BEFORE HERE! */
6077 (PL_expect == XREF ||
6078 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
6080 bool immediate_paren = *s == '(';
6082 /* (Now we can afford to cross potential line boundary.) */
6083 s = SKIPSPACE2(s,nextPL_nextwhite);
6085 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
6088 /* Two barewords in a row may indicate method call. */
6090 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
6091 (tmp = intuit_method(s, gv, cv))) {
6096 /* If not a declared subroutine, it's an indirect object. */
6097 /* (But it's an indir obj regardless for sort.) */
6098 /* Also, if "_" follows a filetest operator, it's a bareword */
6101 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
6103 (PL_last_lop_op != OP_MAPSTART &&
6104 PL_last_lop_op != OP_GREPSTART))))
6105 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
6106 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
6109 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
6114 PL_expect = XOPERATOR;
6117 s = SKIPSPACE2(s,nextPL_nextwhite);
6118 PL_nextwhite = nextPL_nextwhite;
6123 /* Is this a word before a => operator? */
6124 if (*s == '=' && s[1] == '>' && !pkgname) {
6127 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
6128 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
6129 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
6133 /* If followed by a paren, it's certainly a subroutine. */
6138 while (SPACE_OR_TAB(*d))
6140 if (*d == ')' && (sv = cv_const_sv(cv))) {
6147 PL_nextwhite = PL_thiswhite;
6150 start_force(PL_curforce);
6152 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6153 PL_expect = XOPERATOR;
6156 PL_nextwhite = nextPL_nextwhite;
6157 curmad('X', PL_thistoken);
6158 PL_thistoken = newSVpvs("");
6167 /* If followed by var or block, call it a method (unless sub) */
6169 if ((*s == '$' || *s == '{') && !cv) {
6171 PL_last_lop = PL_oldbufptr;
6172 PL_last_lop_op = OP_METHOD;
6176 /* If followed by a bareword, see if it looks like indir obj. */
6179 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
6180 && (tmp = intuit_method(s, gv, cv))) {
6185 /* Not a method, so call it a subroutine (if defined) */
6188 if (lastchar == '-')
6189 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
6190 "Ambiguous use of -%s resolved as -&%s()",
6191 PL_tokenbuf, PL_tokenbuf);
6192 /* Check for a constant sub */
6193 if ((sv = cv_const_sv(cv))) {
6196 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
6197 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
6198 pl_yylval.opval->op_private = 0;
6202 op_free(pl_yylval.opval);
6203 pl_yylval.opval = rv2cv_op;
6204 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
6205 PL_last_lop = PL_oldbufptr;
6206 PL_last_lop_op = OP_ENTERSUB;
6207 /* Is there a prototype? */
6215 const char *proto = SvPV_const(MUTABLE_SV(cv), protolen);
6218 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
6220 while (*proto == ';')
6222 if (*proto == '&' && *s == '{') {
6224 sv_setpvs(PL_subname, "__ANON__");
6226 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6233 PL_nextwhite = PL_thiswhite;
6236 start_force(PL_curforce);
6237 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6240 PL_nextwhite = nextPL_nextwhite;
6241 curmad('X', PL_thistoken);
6242 PL_thistoken = newSVpvs("");
6249 /* Guess harder when madskills require "best effort". */
6250 if (PL_madskills && (!gv || !GvCVu(gv))) {
6251 int probable_sub = 0;
6252 if (strchr("\"'`$@%0123456789!*+{[<", *s))
6254 else if (isALPHA(*s)) {
6258 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
6259 if (!keyword(tmpbuf, tmplen, 0))
6262 while (d < PL_bufend && isSPACE(*d))
6264 if (*d == '=' && d[1] == '>')
6269 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
6270 op_free(pl_yylval.opval);
6271 pl_yylval.opval = rv2cv_op;
6272 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
6273 PL_last_lop = PL_oldbufptr;
6274 PL_last_lop_op = OP_ENTERSUB;
6275 PL_nextwhite = PL_thiswhite;
6277 start_force(PL_curforce);
6278 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6280 PL_nextwhite = nextPL_nextwhite;
6281 curmad('X', PL_thistoken);
6282 PL_thistoken = newSVpvs("");
6287 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6294 /* Call it a bare word */
6296 if (PL_hints & HINT_STRICT_SUBS)
6297 pl_yylval.opval->op_private |= OPpCONST_STRICT;
6300 /* after "print" and similar functions (corresponding to
6301 * "F? L" in opcode.pl), whatever wasn't already parsed as
6302 * a filehandle should be subject to "strict subs".
6303 * Likewise for the optional indirect-object argument to system
6304 * or exec, which can't be a bareword */
6305 if ((PL_last_lop_op == OP_PRINT
6306 || PL_last_lop_op == OP_PRTF
6307 || PL_last_lop_op == OP_SAY
6308 || PL_last_lop_op == OP_SYSTEM
6309 || PL_last_lop_op == OP_EXEC)
6310 && (PL_hints & HINT_STRICT_SUBS))
6311 pl_yylval.opval->op_private |= OPpCONST_STRICT;
6312 if (lastchar != '-') {
6313 if (ckWARN(WARN_RESERVED)) {
6317 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
6318 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
6326 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')) {
6327 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
6328 "Operator or semicolon missing before %c%s",
6329 lastchar, PL_tokenbuf);
6330 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
6331 "Ambiguous use of %c resolved as operator %c",
6332 lastchar, lastchar);
6338 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
6339 newSVpv(CopFILE(PL_curcop),0));
6343 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
6344 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
6347 case KEY___PACKAGE__:
6348 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
6350 ? newSVhek(HvNAME_HEK(PL_curstash))
6357 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
6358 const char *pname = "main";
6359 if (PL_tokenbuf[2] == 'D')
6360 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
6361 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
6365 GvIOp(gv) = newIO();
6366 IoIFP(GvIOp(gv)) = PL_rsfp;
6367 #if defined(HAS_FCNTL) && defined(F_SETFD)
6369 const int fd = PerlIO_fileno(PL_rsfp);
6370 fcntl(fd,F_SETFD,fd >= 3);
6373 /* Mark this internal pseudo-handle as clean */
6374 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
6375 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
6376 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
6378 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
6379 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
6380 /* if the script was opened in binmode, we need to revert
6381 * it to text mode for compatibility; but only iff it has CRs
6382 * XXX this is a questionable hack at best. */
6383 if (PL_bufend-PL_bufptr > 2
6384 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
6387 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
6388 loc = PerlIO_tell(PL_rsfp);
6389 (void)PerlIO_seek(PL_rsfp, 0L, 0);
6392 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
6394 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
6395 #endif /* NETWARE */
6396 #ifdef PERLIO_IS_STDIO /* really? */
6397 # if defined(__BORLANDC__)
6398 /* XXX see note in do_binmode() */
6399 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
6403 PerlIO_seek(PL_rsfp, loc, 0);
6407 #ifdef PERLIO_LAYERS
6410 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
6411 else if (PL_encoding) {
6418 XPUSHs(PL_encoding);
6420 call_method("name", G_SCALAR);
6424 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
6425 Perl_form(aTHX_ ":encoding(%"SVf")",
6434 if (PL_realtokenstart >= 0) {
6435 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6437 PL_endwhite = newSVpvs("");
6438 sv_catsv(PL_endwhite, PL_thiswhite);
6440 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
6441 PL_realtokenstart = -1;
6443 while ((s = filter_gets(PL_endwhite, SvCUR(PL_endwhite)))
6459 if (PL_expect == XSTATE) {
6466 if (*s == ':' && s[1] == ':') {
6469 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
6470 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
6471 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
6474 else if (tmp == KEY_require || tmp == KEY_do)
6475 /* that's a way to remember we saw "CORE::" */
6488 LOP(OP_ACCEPT,XTERM);
6494 LOP(OP_ATAN2,XTERM);
6500 LOP(OP_BINMODE,XTERM);
6503 LOP(OP_BLESS,XTERM);
6512 /* When 'use switch' is in effect, continue has a dual
6513 life as a control operator. */
6515 if (!FEATURE_IS_ENABLED("switch"))
6518 /* We have to disambiguate the two senses of
6519 "continue". If the next token is a '{' then
6520 treat it as the start of a continue block;
6521 otherwise treat it as a control operator.
6533 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
6550 if (!PL_cryptseen) {
6551 PL_cryptseen = TRUE;
6555 LOP(OP_CRYPT,XTERM);
6558 LOP(OP_CHMOD,XTERM);
6561 LOP(OP_CHOWN,XTERM);
6564 LOP(OP_CONNECT,XTERM);
6583 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6584 if (orig_keyword == KEY_do) {
6593 PL_hints |= HINT_BLOCK_SCOPE;
6603 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
6604 LOP(OP_DBMOPEN,XTERM);
6610 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6617 pl_yylval.ival = CopLINE(PL_curcop);
6633 if (*s == '{') { /* block eval */
6634 PL_expect = XTERMBLOCK;
6635 UNIBRACK(OP_ENTERTRY);
6637 else { /* string eval */
6639 UNIBRACK(OP_ENTEREVAL);
6654 case KEY_endhostent:
6660 case KEY_endservent:
6663 case KEY_endprotoent:
6674 pl_yylval.ival = CopLINE(PL_curcop);
6676 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6679 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6682 if ((PL_bufend - p) >= 3 &&
6683 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6685 else if ((PL_bufend - p) >= 4 &&
6686 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6689 if (isIDFIRST_lazy_if(p,UTF)) {
6690 p = scan_ident(p, PL_bufend,
6691 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6695 Perl_croak(aTHX_ "Missing $ on loop variable");
6697 s = SvPVX(PL_linestr) + soff;
6703 LOP(OP_FORMLINE,XTERM);
6709 LOP(OP_FCNTL,XTERM);
6715 LOP(OP_FLOCK,XTERM);
6724 LOP(OP_GREPSTART, XREF);
6727 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6742 case KEY_getpriority:
6743 LOP(OP_GETPRIORITY,XTERM);
6745 case KEY_getprotobyname:
6748 case KEY_getprotobynumber:
6749 LOP(OP_GPBYNUMBER,XTERM);
6751 case KEY_getprotoent:
6763 case KEY_getpeername:
6764 UNI(OP_GETPEERNAME);
6766 case KEY_gethostbyname:
6769 case KEY_gethostbyaddr:
6770 LOP(OP_GHBYADDR,XTERM);
6772 case KEY_gethostent:
6775 case KEY_getnetbyname:
6778 case KEY_getnetbyaddr:
6779 LOP(OP_GNBYADDR,XTERM);
6784 case KEY_getservbyname:
6785 LOP(OP_GSBYNAME,XTERM);
6787 case KEY_getservbyport:
6788 LOP(OP_GSBYPORT,XTERM);
6790 case KEY_getservent:
6793 case KEY_getsockname:
6794 UNI(OP_GETSOCKNAME);
6796 case KEY_getsockopt:
6797 LOP(OP_GSOCKOPT,XTERM);
6812 pl_yylval.ival = CopLINE(PL_curcop);
6822 pl_yylval.ival = CopLINE(PL_curcop);
6826 LOP(OP_INDEX,XTERM);
6832 LOP(OP_IOCTL,XTERM);
6844 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6876 LOP(OP_LISTEN,XTERM);
6885 s = scan_pat(s,OP_MATCH);
6886 TERM(sublex_start());
6889 LOP(OP_MAPSTART, XREF);
6892 LOP(OP_MKDIR,XTERM);
6895 LOP(OP_MSGCTL,XTERM);
6898 LOP(OP_MSGGET,XTERM);
6901 LOP(OP_MSGRCV,XTERM);
6904 LOP(OP_MSGSND,XTERM);
6909 PL_in_my = (U16)tmp;
6911 if (isIDFIRST_lazy_if(s,UTF)) {
6915 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6916 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6918 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6919 if (!PL_in_my_stash) {
6922 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6926 if (PL_madskills) { /* just add type to declarator token */
6927 sv_catsv(PL_thistoken, PL_nextwhite);
6929 sv_catpvn(PL_thistoken, start, s - start);
6937 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6944 s = tokenize_use(0, s);
6948 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6955 if (isIDFIRST_lazy_if(s,UTF)) {
6957 for (d = s; isALNUM_lazy_if(d,UTF);)
6959 for (t=d; isSPACE(*t);)
6961 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6963 && !(t[0] == '=' && t[1] == '>')
6965 int parms_len = (int)(d-s);
6966 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6967 "Precedence problem: open %.*s should be open(%.*s)",
6968 parms_len, s, parms_len, s);
6974 pl_yylval.ival = OP_OR;
6984 LOP(OP_OPEN_DIR,XTERM);
6987 checkcomma(s,PL_tokenbuf,"filehandle");
6991 checkcomma(s,PL_tokenbuf,"filehandle");
7010 s = force_word(s,WORD,FALSE,TRUE,FALSE);
7011 s = force_strict_version(s);
7015 LOP(OP_PIPE_OP,XTERM);
7018 s = scan_str(s,!!PL_madskills,FALSE);
7021 pl_yylval.ival = OP_CONST;
7022 TERM(sublex_start());
7028 s = scan_str(s,!!PL_madskills,FALSE);
7031 PL_expect = XOPERATOR;
7033 if (SvCUR(PL_lex_stuff)) {
7036 d = SvPV_force(PL_lex_stuff, len);
7038 for (; isSPACE(*d) && len; --len, ++d)
7043 if (!warned && ckWARN(WARN_QW)) {
7044 for (; !isSPACE(*d) && len; --len, ++d) {
7046 Perl_warner(aTHX_ packWARN(WARN_QW),
7047 "Possible attempt to separate words with commas");
7050 else if (*d == '#') {
7051 Perl_warner(aTHX_ packWARN(WARN_QW),
7052 "Possible attempt to put comments in qw() list");
7058 for (; !isSPACE(*d) && len; --len, ++d)
7061 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
7062 words = append_elem(OP_LIST, words,
7063 newSVOP(OP_CONST, 0, tokeq(sv)));
7067 start_force(PL_curforce);
7068 NEXTVAL_NEXTTOKE.opval = words;
7073 SvREFCNT_dec(PL_lex_stuff);
7074 PL_lex_stuff = NULL;
7080 s = scan_str(s,!!PL_madskills,FALSE);
7083 pl_yylval.ival = OP_STRINGIFY;
7084 if (SvIVX(PL_lex_stuff) == '\'')
7085 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
7086 TERM(sublex_start());
7089 s = scan_pat(s,OP_QR);
7090 TERM(sublex_start());
7093 s = scan_str(s,!!PL_madskills,FALSE);
7096 readpipe_override();
7097 TERM(sublex_start());
7105 s = force_version(s, FALSE);
7107 else if (*s != 'v' || !isDIGIT(s[1])
7108 || (s = force_version(s, TRUE), *s == 'v'))
7110 *PL_tokenbuf = '\0';
7111 s = force_word(s,WORD,TRUE,TRUE,FALSE);
7112 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
7113 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
7115 yyerror("<> should be quotes");
7117 if (orig_keyword == KEY_require) {
7125 PL_last_uni = PL_oldbufptr;
7126 PL_last_lop_op = OP_REQUIRE;
7128 return REPORT( (int)REQUIRE );
7134 s = force_word(s,WORD,TRUE,FALSE,FALSE);
7138 LOP(OP_RENAME,XTERM);
7147 LOP(OP_RINDEX,XTERM);
7156 UNIDOR(OP_READLINE);
7159 UNIDOR(OP_BACKTICK);
7168 LOP(OP_REVERSE,XTERM);
7171 UNIDOR(OP_READLINK);
7178 if (pl_yylval.opval)
7179 TERM(sublex_start());
7181 TOKEN(1); /* force error */
7184 checkcomma(s,PL_tokenbuf,"filehandle");
7194 LOP(OP_SELECT,XTERM);
7200 LOP(OP_SEMCTL,XTERM);
7203 LOP(OP_SEMGET,XTERM);
7206 LOP(OP_SEMOP,XTERM);
7212 LOP(OP_SETPGRP,XTERM);
7214 case KEY_setpriority:
7215 LOP(OP_SETPRIORITY,XTERM);
7217 case KEY_sethostent:
7223 case KEY_setservent:
7226 case KEY_setprotoent:
7236 LOP(OP_SEEKDIR,XTERM);
7238 case KEY_setsockopt:
7239 LOP(OP_SSOCKOPT,XTERM);
7245 LOP(OP_SHMCTL,XTERM);
7248 LOP(OP_SHMGET,XTERM);
7251 LOP(OP_SHMREAD,XTERM);
7254 LOP(OP_SHMWRITE,XTERM);
7257 LOP(OP_SHUTDOWN,XTERM);
7266 LOP(OP_SOCKET,XTERM);
7268 case KEY_socketpair:
7269 LOP(OP_SOCKPAIR,XTERM);
7272 checkcomma(s,PL_tokenbuf,"subroutine name");
7274 if (*s == ';' || *s == ')') /* probably a close */
7275 Perl_croak(aTHX_ "sort is now a reserved word");
7277 s = force_word(s,WORD,TRUE,TRUE,FALSE);
7281 LOP(OP_SPLIT,XTERM);
7284 LOP(OP_SPRINTF,XTERM);
7287 LOP(OP_SPLICE,XTERM);
7302 LOP(OP_SUBSTR,XTERM);
7308 char tmpbuf[sizeof PL_tokenbuf];
7309 SSize_t tboffset = 0;
7310 expectation attrful;
7311 bool have_name, have_proto;
7312 const int key = tmp;
7317 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
7318 SV *subtoken = newSVpvn(tstart, s - tstart);
7322 s = SKIPSPACE2(s,tmpwhite);
7327 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
7328 (*s == ':' && s[1] == ':'))
7331 SV *nametoke = NULL;
7335 attrful = XATTRBLOCK;
7336 /* remember buffer pos'n for later force_word */
7337 tboffset = s - PL_oldbufptr;
7338 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
7341 nametoke = newSVpvn(s, d - s);
7343 if (memchr(tmpbuf, ':', len))
7344 sv_setpvn(PL_subname, tmpbuf, len);
7346 sv_setsv(PL_subname,PL_curstname);
7347 sv_catpvs(PL_subname,"::");
7348 sv_catpvn(PL_subname,tmpbuf,len);
7355 CURMAD('X', nametoke);
7356 CURMAD('_', tmpwhite);
7357 (void) force_word(PL_oldbufptr + tboffset, WORD,
7360 s = SKIPSPACE2(d,tmpwhite);
7367 Perl_croak(aTHX_ "Missing name in \"my sub\"");
7368 PL_expect = XTERMBLOCK;
7369 attrful = XATTRTERM;
7370 sv_setpvs(PL_subname,"?");
7374 if (key == KEY_format) {
7376 PL_lex_formbrack = PL_lex_brackets + 1;
7378 PL_thistoken = subtoken;
7382 (void) force_word(PL_oldbufptr + tboffset, WORD,
7388 /* Look for a prototype */
7391 bool bad_proto = FALSE;
7392 bool in_brackets = FALSE;
7393 char greedy_proto = ' ';
7394 bool proto_after_greedy_proto = FALSE;
7395 bool must_be_last = FALSE;
7396 bool underscore = FALSE;
7397 bool seen_underscore = FALSE;
7398 const bool warnillegalproto = ckWARN(WARN_ILLEGALPROTO);
7400 s = scan_str(s,!!PL_madskills,FALSE);
7402 Perl_croak(aTHX_ "Prototype not terminated");
7403 /* strip spaces and check for bad characters */
7404 d = SvPVX(PL_lex_stuff);
7406 for (p = d; *p; ++p) {
7410 if (warnillegalproto) {
7412 proto_after_greedy_proto = TRUE;
7413 if (!strchr("$@%*;[]&\\_", *p)) {
7425 else if ( *p == ']' ) {
7426 in_brackets = FALSE;
7428 else if ( (*p == '@' || *p == '%') &&
7429 ( tmp < 2 || d[tmp-2] != '\\' ) &&
7431 must_be_last = TRUE;
7434 else if ( *p == '_' ) {
7435 underscore = seen_underscore = TRUE;
7442 if (proto_after_greedy_proto)
7443 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
7444 "Prototype after '%c' for %"SVf" : %s",
7445 greedy_proto, SVfARG(PL_subname), d);
7447 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
7448 "Illegal character %sin prototype for %"SVf" : %s",
7449 seen_underscore ? "after '_' " : "",
7450 SVfARG(PL_subname), d);
7451 SvCUR_set(PL_lex_stuff, tmp);
7456 CURMAD('q', PL_thisopen);
7457 CURMAD('_', tmpwhite);
7458 CURMAD('=', PL_thisstuff);
7459 CURMAD('Q', PL_thisclose);
7460 NEXTVAL_NEXTTOKE.opval =
7461 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
7462 PL_lex_stuff = NULL;
7465 s = SKIPSPACE2(s,tmpwhite);
7473 if (*s == ':' && s[1] != ':')
7474 PL_expect = attrful;
7475 else if (*s != '{' && key == KEY_sub) {
7477 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
7478 else if (*s != ';' && *s != '}')
7479 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
7486 curmad('^', newSVpvs(""));
7487 CURMAD('_', tmpwhite);
7491 PL_thistoken = subtoken;
7494 NEXTVAL_NEXTTOKE.opval =
7495 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
7496 PL_lex_stuff = NULL;
7502 sv_setpvs(PL_subname, "__ANON__");
7504 sv_setpvs(PL_subname, "__ANON__::__ANON__");
7508 (void) force_word(PL_oldbufptr + tboffset, WORD,
7517 LOP(OP_SYSTEM,XREF);
7520 LOP(OP_SYMLINK,XTERM);
7523 LOP(OP_SYSCALL,XTERM);
7526 LOP(OP_SYSOPEN,XTERM);
7529 LOP(OP_SYSSEEK,XTERM);
7532 LOP(OP_SYSREAD,XTERM);
7535 LOP(OP_SYSWRITE,XTERM);
7539 TERM(sublex_start());
7560 LOP(OP_TRUNCATE,XTERM);
7572 pl_yylval.ival = CopLINE(PL_curcop);
7576 pl_yylval.ival = CopLINE(PL_curcop);
7580 LOP(OP_UNLINK,XTERM);
7586 LOP(OP_UNPACK,XTERM);
7589 LOP(OP_UTIME,XTERM);
7595 LOP(OP_UNSHIFT,XTERM);
7598 s = tokenize_use(1, s);
7608 pl_yylval.ival = CopLINE(PL_curcop);
7612 pl_yylval.ival = CopLINE(PL_curcop);
7616 PL_hints |= HINT_BLOCK_SCOPE;
7623 LOP(OP_WAITPID,XTERM);
7632 ctl_l[0] = toCTRL('L');
7634 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
7637 /* Make sure $^L is defined */
7638 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
7643 if (PL_expect == XOPERATOR)
7649 pl_yylval.ival = OP_XOR;
7654 TERM(sublex_start());
7659 #pragma segment Main
7663 S_pending_ident(pTHX)
7668 /* pit holds the identifier we read and pending_ident is reset */
7669 char pit = PL_pending_ident;
7670 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7671 /* All routes through this function want to know if there is a colon. */
7672 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7673 PL_pending_ident = 0;
7675 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7676 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7677 "### Pending identifier '%s'\n", PL_tokenbuf); });
7679 /* if we're in a my(), we can't allow dynamics here.
7680 $foo'bar has already been turned into $foo::bar, so
7681 just check for colons.
7683 if it's a legal name, the OP is a PADANY.
7686 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7688 yyerror(Perl_form(aTHX_ "No package name allowed for "
7689 "variable %s in \"our\"",
7691 tmp = allocmy(PL_tokenbuf, tokenbuf_len, 0);
7695 yyerror(Perl_form(aTHX_ PL_no_myglob,
7696 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7698 pl_yylval.opval = newOP(OP_PADANY, 0);
7699 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf, tokenbuf_len, 0);
7705 build the ops for accesses to a my() variable.
7707 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7708 then used in a comparison. This catches most, but not
7709 all cases. For instance, it catches
7710 sort { my($a); $a <=> $b }
7712 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7713 (although why you'd do that is anyone's guess).
7718 tmp = pad_findmy(PL_tokenbuf, tokenbuf_len, 0);
7719 if (tmp != NOT_IN_PAD) {
7720 /* might be an "our" variable" */
7721 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7722 /* build ops for a bareword */
7723 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7724 HEK * const stashname = HvNAME_HEK(stash);
7725 SV * const sym = newSVhek(stashname);
7726 sv_catpvs(sym, "::");
7727 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7728 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7729 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7732 ? (GV_ADDMULTI | GV_ADDINEVAL)
7735 ((PL_tokenbuf[0] == '$') ? SVt_PV
7736 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7741 /* if it's a sort block and they're naming $a or $b */
7742 if (PL_last_lop_op == OP_SORT &&
7743 PL_tokenbuf[0] == '$' &&
7744 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7747 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7748 d < PL_bufend && *d != '\n';
7751 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7752 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7758 pl_yylval.opval = newOP(OP_PADANY, 0);
7759 pl_yylval.opval->op_targ = tmp;
7765 Whine if they've said @foo in a doublequoted string,
7766 and @foo isn't a variable we can find in the symbol
7769 if (ckWARN(WARN_AMBIGUOUS) &&
7770 pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7771 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7773 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7774 /* DO NOT warn for @- and @+ */
7775 && !( PL_tokenbuf[2] == '\0' &&
7776 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7779 /* Downgraded from fatal to warning 20000522 mjd */
7780 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7781 "Possible unintended interpolation of %s in string",
7786 /* build ops for a bareword */
7787 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7789 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7791 PL_tokenbuf + 1, tokenbuf_len - 1,
7792 /* If the identifier refers to a stash, don't autovivify it.
7793 * Change 24660 had the side effect of causing symbol table
7794 * hashes to always be defined, even if they were freshly
7795 * created and the only reference in the entire program was
7796 * the single statement with the defined %foo::bar:: test.
7797 * It appears that all code in the wild doing this actually
7798 * wants to know whether sub-packages have been loaded, so
7799 * by avoiding auto-vivifying symbol tables, we ensure that
7800 * defined %foo::bar:: continues to be false, and the existing
7801 * tests still give the expected answers, even though what
7802 * they're actually testing has now changed subtly.
7804 (*PL_tokenbuf == '%'
7805 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7808 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7809 ((PL_tokenbuf[0] == '$') ? SVt_PV
7810 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7816 * The following code was generated by perl_keyword.pl.
7820 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7824 PERL_ARGS_ASSERT_KEYWORD;
7828 case 1: /* 5 tokens of length 1 */
7860 case 2: /* 18 tokens of length 2 */
8006 case 3: /* 29 tokens of length 3 */
8010 if (name[1] == 'N' &&
8073 if (name[1] == 'i' &&
8105 if (name[1] == 'o' &&
8114 if (name[1] == 'e' &&
8123 if (name[1] == 'n' &&
8132 if (name[1] == 'o' &&
8141 if (name[1] == 'a' &&
8150 if (name[1] == 'o' &&
8212 if (name[1] == 'e' &&
8226 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
8252 if (name[1] == 'i' &&
8261 if (name[1] == 's' &&
8270 if (name[1] == 'e' &&
8279 if (name[1] == 'o' &&
8291 case 4: /* 41 tokens of length 4 */
8295 if (name[1] == 'O' &&
8305 if (name[1] == 'N' &&
8315 if (name[1] == 'i' &&
8325 if (name[1] == 'h' &&
8335 if (name[1] == 'u' &&
8348 if (name[2] == 'c' &&
8357 if (name[2] == 's' &&
8366 if (name[2] == 'a' &&
8402 if (name[1] == 'o' &&
8415 if (name[2] == 't' &&
8424 if (name[2] == 'o' &&
8433 if (name[2] == 't' &&
8442 if (name[2] == 'e' &&
8455 if (name[1] == 'o' &&
8468 if (name[2] == 'y' &&
8477 if (name[2] == 'l' &&
8493 if (name[2] == 's' &&
8502 if (name[2] == 'n' &&
8511 if (name[2] == 'c' &&
8524 if (name[1] == 'e' &&
8534 if (name[1] == 'p' &&
8547 if (name[2] == 'c' &&
8556 if (name[2] == 'p' &&
8565 if (name[2] == 's' &&
8581 if (name[2] == 'n' &&
8651 if (name[2] == 'r' &&
8660 if (name[2] == 'r' &&
8669 if (name[2] == 'a' &&
8685 if (name[2] == 'l' &&
8747 if (name[2] == 'e' &&
8750 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8763 case 5: /* 39 tokens of length 5 */
8767 if (name[1] == 'E' &&
8778 if (name[1] == 'H' &&
8792 if (name[2] == 'a' &&
8802 if (name[2] == 'a' &&
8819 if (name[2] == 'e' &&
8829 if (name[2] == 'e' &&
8833 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8849 if (name[3] == 'i' &&
8858 if (name[3] == 'o' &&
8894 if (name[2] == 'o' &&
8904 if (name[2] == 'y' &&
8918 if (name[1] == 'l' &&
8932 if (name[2] == 'n' &&
8942 if (name[2] == 'o' &&
8956 if (name[1] == 'i' &&
8961 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8970 if (name[2] == 'd' &&
8980 if (name[2] == 'c' &&
8997 if (name[2] == 'c' &&
9007 if (name[2] == 't' &&
9021 if (name[1] == 'k' &&
9032 if (name[1] == 'r' &&
9046 if (name[2] == 's' &&
9056 if (name[2] == 'd' &&
9073 if (name[2] == 'm' &&
9083 if (name[2] == 'i' &&
9093 if (name[2] == 'e' &&
9103 if (name[2] == 'l' &&
9113 if (name[2] == 'a' &&
9126 if (name[3] == 't' &&
9129 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
9135 if (name[3] == 'd' &&
9152 if (name[1] == 'i' &&
9166 if (name[2] == 'a' &&
9179 if (name[3] == 'e' &&
9214 if (name[2] == 'i' &&
9231 if (name[2] == 'i' &&
9241 if (name[2] == 'i' &&
9258 case 6: /* 33 tokens of length 6 */
9262 if (name[1] == 'c' &&
9277 if (name[2] == 'l' &&
9288 if (name[2] == 'r' &&
9303 if (name[1] == 'e' &&
9318 if (name[2] == 's' &&
9323 Perl_ck_warner_d(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
9329 if (name[2] == 'i' &&
9347 if (name[2] == 'l' &&
9358 if (name[2] == 'r' &&
9373 if (name[1] == 'm' &&
9388 if (name[2] == 'n' &&
9399 if (name[2] == 's' &&
9414 if (name[1] == 's' &&
9420 if (name[4] == 't' &&
9429 if (name[4] == 'e' &&
9438 if (name[4] == 'c' &&
9447 if (name[4] == 'n' &&
9463 if (name[1] == 'r' &&
9481 if (name[3] == 'a' &&
9491 if (name[3] == 'u' &&
9505 if (name[2] == 'n' &&
9523 if (name[2] == 'a' &&
9537 if (name[3] == 'e' &&
9550 if (name[4] == 't' &&
9559 if (name[4] == 'e' &&
9581 if (name[4] == 't' &&
9590 if (name[4] == 'e' &&
9606 if (name[2] == 'c' &&
9617 if (name[2] == 'l' &&
9628 if (name[2] == 'b' &&
9639 if (name[2] == 's' &&
9662 if (name[4] == 's' &&
9671 if (name[4] == 'n' &&
9684 if (name[3] == 'a' &&
9701 if (name[1] == 'a' &&
9716 case 7: /* 29 tokens of length 7 */
9720 if (name[1] == 'E' &&
9733 if (name[1] == '_' &&
9746 if (name[1] == 'i' &&
9753 return -KEY_binmode;
9759 if (name[1] == 'o' &&
9766 return -KEY_connect;
9775 if (name[2] == 'm' &&
9781 return -KEY_dbmopen;
9792 if (name[4] == 'u' &&
9796 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9802 if (name[4] == 'n' &&
9823 if (name[1] == 'o' &&
9836 if (name[1] == 'e' &&
9843 if (name[5] == 'r' &&
9846 return -KEY_getpgrp;
9852 if (name[5] == 'i' &&
9855 return -KEY_getppid;
9868 if (name[1] == 'c' &&
9875 return -KEY_lcfirst;
9881 if (name[1] == 'p' &&
9888 return -KEY_opendir;
9894 if (name[1] == 'a' &&
9912 if (name[3] == 'd' &&
9917 return -KEY_readdir;
9923 if (name[3] == 'u' &&
9934 if (name[3] == 'e' &&
9939 return -KEY_reverse;
9958 if (name[3] == 'k' &&
9963 return -KEY_seekdir;
9969 if (name[3] == 'p' &&
9974 return -KEY_setpgrp;
9984 if (name[2] == 'm' &&
9990 return -KEY_shmread;
9996 if (name[2] == 'r' &&
10002 return -KEY_sprintf;
10011 if (name[3] == 'l' &&
10016 return -KEY_symlink;
10025 if (name[4] == 'a' &&
10029 return -KEY_syscall;
10035 if (name[4] == 'p' &&
10039 return -KEY_sysopen;
10045 if (name[4] == 'e' &&
10049 return -KEY_sysread;
10055 if (name[4] == 'e' &&
10059 return -KEY_sysseek;
10077 if (name[1] == 'e' &&
10084 return -KEY_telldir;
10093 if (name[2] == 'f' &&
10099 return -KEY_ucfirst;
10105 if (name[2] == 's' &&
10111 return -KEY_unshift;
10121 if (name[1] == 'a' &&
10128 return -KEY_waitpid;
10137 case 8: /* 26 tokens of length 8 */
10141 if (name[1] == 'U' &&
10149 return KEY_AUTOLOAD;
10155 if (name[1] == '_')
10160 if (name[3] == 'A' &&
10166 return KEY___DATA__;
10172 if (name[3] == 'I' &&
10178 return -KEY___FILE__;
10184 if (name[3] == 'I' &&
10190 return -KEY___LINE__;
10206 if (name[2] == 'o' &&
10213 return -KEY_closedir;
10219 if (name[2] == 'n' &&
10226 return -KEY_continue;
10236 if (name[1] == 'b' &&
10244 return -KEY_dbmclose;
10250 if (name[1] == 'n' &&
10256 if (name[4] == 'r' &&
10261 return -KEY_endgrent;
10267 if (name[4] == 'w' &&
10272 return -KEY_endpwent;
10285 if (name[1] == 'o' &&
10293 return -KEY_formline;
10299 if (name[1] == 'e' &&
10305 if (name[4] == 'r')
10310 if (name[6] == 'n' &&
10313 return -KEY_getgrent;
10319 if (name[6] == 'i' &&
10322 return -KEY_getgrgid;
10328 if (name[6] == 'a' &&
10331 return -KEY_getgrnam;
10344 if (name[4] == 'o' &&
10349 return -KEY_getlogin;
10355 if (name[4] == 'w')
10360 if (name[6] == 'n' &&
10363 return -KEY_getpwent;
10369 if (name[6] == 'a' &&
10372 return -KEY_getpwnam;
10378 if (name[6] == 'i' &&
10381 return -KEY_getpwuid;
10401 if (name[1] == 'e' &&
10408 if (name[5] == 'i' &&
10415 return -KEY_readline;
10420 return -KEY_readlink;
10431 if (name[5] == 'i' &&
10435 return -KEY_readpipe;
10451 if (name[2] == 't')
10456 if (name[4] == 'r' &&
10461 return -KEY_setgrent;
10467 if (name[4] == 'w' &&
10472 return -KEY_setpwent;
10488 if (name[3] == 'w' &&
10494 return -KEY_shmwrite;
10500 if (name[3] == 't' &&
10506 return -KEY_shutdown;
10516 if (name[2] == 's' &&
10523 return -KEY_syswrite;
10533 if (name[1] == 'r' &&
10541 return -KEY_truncate;
10550 case 9: /* 9 tokens of length 9 */
10554 if (name[1] == 'N' &&
10563 return KEY_UNITCHECK;
10569 if (name[1] == 'n' &&
10578 return -KEY_endnetent;
10584 if (name[1] == 'e' &&
10593 return -KEY_getnetent;
10599 if (name[1] == 'o' &&
10608 return -KEY_localtime;
10614 if (name[1] == 'r' &&
10623 return KEY_prototype;
10629 if (name[1] == 'u' &&
10638 return -KEY_quotemeta;
10644 if (name[1] == 'e' &&
10653 return -KEY_rewinddir;
10659 if (name[1] == 'e' &&
10668 return -KEY_setnetent;
10674 if (name[1] == 'a' &&
10683 return -KEY_wantarray;
10692 case 10: /* 9 tokens of length 10 */
10696 if (name[1] == 'n' &&
10702 if (name[4] == 'o' &&
10709 return -KEY_endhostent;
10715 if (name[4] == 'e' &&
10722 return -KEY_endservent;
10735 if (name[1] == 'e' &&
10741 if (name[4] == 'o' &&
10748 return -KEY_gethostent;
10757 if (name[5] == 'r' &&
10763 return -KEY_getservent;
10769 if (name[5] == 'c' &&
10775 return -KEY_getsockopt;
10795 if (name[2] == 't')
10800 if (name[4] == 'o' &&
10807 return -KEY_sethostent;
10816 if (name[5] == 'r' &&
10822 return -KEY_setservent;
10828 if (name[5] == 'c' &&
10834 return -KEY_setsockopt;
10851 if (name[2] == 'c' &&
10860 return -KEY_socketpair;
10873 case 11: /* 8 tokens of length 11 */
10877 if (name[1] == '_' &&
10887 { /* __PACKAGE__ */
10888 return -KEY___PACKAGE__;
10894 if (name[1] == 'n' &&
10904 { /* endprotoent */
10905 return -KEY_endprotoent;
10911 if (name[1] == 'e' &&
10920 if (name[5] == 'e' &&
10926 { /* getpeername */
10927 return -KEY_getpeername;
10936 if (name[6] == 'o' &&
10941 { /* getpriority */
10942 return -KEY_getpriority;
10948 if (name[6] == 't' &&
10953 { /* getprotoent */
10954 return -KEY_getprotoent;
10968 if (name[4] == 'o' &&
10975 { /* getsockname */
10976 return -KEY_getsockname;
10989 if (name[1] == 'e' &&
10997 if (name[6] == 'o' &&
11002 { /* setpriority */
11003 return -KEY_setpriority;
11009 if (name[6] == 't' &&
11014 { /* setprotoent */
11015 return -KEY_setprotoent;
11031 case 12: /* 2 tokens of length 12 */
11032 if (name[0] == 'g' &&
11044 if (name[9] == 'd' &&
11047 { /* getnetbyaddr */
11048 return -KEY_getnetbyaddr;
11054 if (name[9] == 'a' &&
11057 { /* getnetbyname */
11058 return -KEY_getnetbyname;
11070 case 13: /* 4 tokens of length 13 */
11071 if (name[0] == 'g' &&
11078 if (name[4] == 'o' &&
11087 if (name[10] == 'd' &&
11090 { /* gethostbyaddr */
11091 return -KEY_gethostbyaddr;
11097 if (name[10] == 'a' &&
11100 { /* gethostbyname */
11101 return -KEY_gethostbyname;
11114 if (name[4] == 'e' &&
11123 if (name[10] == 'a' &&
11126 { /* getservbyname */
11127 return -KEY_getservbyname;
11133 if (name[10] == 'o' &&
11136 { /* getservbyport */
11137 return -KEY_getservbyport;
11156 case 14: /* 1 tokens of length 14 */
11157 if (name[0] == 'g' &&
11171 { /* getprotobyname */
11172 return -KEY_getprotobyname;
11177 case 16: /* 1 tokens of length 16 */
11178 if (name[0] == 'g' &&
11194 { /* getprotobynumber */
11195 return -KEY_getprotobynumber;
11209 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
11213 PERL_ARGS_ASSERT_CHECKCOMMA;
11215 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
11216 if (ckWARN(WARN_SYNTAX)) {
11219 for (w = s+2; *w && level; w++) {
11222 else if (*w == ')')
11225 while (isSPACE(*w))
11227 /* the list of chars below is for end of statements or
11228 * block / parens, boolean operators (&&, ||, //) and branch
11229 * constructs (or, and, if, until, unless, while, err, for).
11230 * Not a very solid hack... */
11231 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
11232 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11233 "%s (...) interpreted as function",name);
11236 while (s < PL_bufend && isSPACE(*s))
11240 while (s < PL_bufend && isSPACE(*s))
11242 if (isIDFIRST_lazy_if(s,UTF)) {
11243 const char * const w = s++;
11244 while (isALNUM_lazy_if(s,UTF))
11246 while (s < PL_bufend && isSPACE(*s))
11250 if (keyword(w, s - w, 0))
11253 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
11254 if (gv && GvCVu(gv))
11256 Perl_croak(aTHX_ "No comma allowed after %s", what);
11261 /* Either returns sv, or mortalizes sv and returns a new SV*.
11262 Best used as sv=new_constant(..., sv, ...).
11263 If s, pv are NULL, calls subroutine with one argument,
11264 and type is used with error messages only. */
11267 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
11268 SV *sv, SV *pv, const char *type, STRLEN typelen)
11271 HV * const table = GvHV(PL_hintgv); /* ^H */
11275 const char *why1 = "", *why2 = "", *why3 = "";
11277 PERL_ARGS_ASSERT_NEW_CONSTANT;
11279 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
11282 why2 = (const char *)
11283 (strEQ(key,"charnames")
11284 ? "(possibly a missing \"use charnames ...\")"
11286 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
11287 (type ? type: "undef"), why2);
11289 /* This is convoluted and evil ("goto considered harmful")
11290 * but I do not understand the intricacies of all the different
11291 * failure modes of %^H in here. The goal here is to make
11292 * the most probable error message user-friendly. --jhi */
11297 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
11298 (type ? type: "undef"), why1, why2, why3);
11300 yyerror(SvPVX_const(msg));
11304 cvp = hv_fetch(table, key, keylen, FALSE);
11305 if (!cvp || !SvOK(*cvp)) {
11308 why3 = "} is not defined";
11311 sv_2mortal(sv); /* Parent created it permanently */
11314 pv = newSVpvn_flags(s, len, SVs_TEMP);
11316 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
11318 typesv = &PL_sv_undef;
11320 PUSHSTACKi(PERLSI_OVERLOAD);
11332 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
11336 /* Check the eval first */
11337 if (!PL_in_eval && SvTRUE(ERRSV)) {
11338 sv_catpvs(ERRSV, "Propagated");
11339 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
11341 res = SvREFCNT_inc_simple(sv);
11345 SvREFCNT_inc_simple_void(res);
11354 why1 = "Call to &{$^H{";
11356 why3 = "}} did not return a defined value";
11364 /* Returns a NUL terminated string, with the length of the string written to
11368 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
11371 register char *d = dest;
11372 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
11374 PERL_ARGS_ASSERT_SCAN_WORD;
11378 Perl_croak(aTHX_ ident_too_long);
11379 if (isALNUM(*s)) /* UTF handled below */
11381 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
11386 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
11390 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
11391 char *t = s + UTF8SKIP(s);
11393 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
11397 Perl_croak(aTHX_ ident_too_long);
11398 Copy(s, d, len, char);
11411 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
11414 char *bracket = NULL;
11416 register char *d = dest;
11417 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
11419 PERL_ARGS_ASSERT_SCAN_IDENT;
11424 while (isDIGIT(*s)) {
11426 Perl_croak(aTHX_ ident_too_long);
11433 Perl_croak(aTHX_ ident_too_long);
11434 if (isALNUM(*s)) /* UTF handled below */
11436 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
11441 else if (*s == ':' && s[1] == ':') {
11445 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
11446 char *t = s + UTF8SKIP(s);
11447 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
11449 if (d + (t - s) > e)
11450 Perl_croak(aTHX_ ident_too_long);
11451 Copy(s, d, t - s, char);
11462 if (PL_lex_state != LEX_NORMAL)
11463 PL_lex_state = LEX_INTERPENDMAYBE;
11466 if (*s == '$' && s[1] &&
11467 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
11480 if (*d == '^' && *s && isCONTROLVAR(*s)) {
11485 if (isSPACE(s[-1])) {
11487 const char ch = *s++;
11488 if (!SPACE_OR_TAB(ch)) {
11494 if (isIDFIRST_lazy_if(d,UTF)) {
11498 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
11499 end += UTF8SKIP(end);
11500 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
11501 end += UTF8SKIP(end);
11503 Copy(s, d, end - s, char);
11508 while ((isALNUM(*s) || *s == ':') && d < e)
11511 Perl_croak(aTHX_ ident_too_long);
11514 while (s < send && SPACE_OR_TAB(*s))
11516 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
11517 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
11518 const char * const brack =
11520 ((*s == '[') ? "[...]" : "{...}");
11521 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
11522 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
11523 funny, dest, brack, funny, dest, brack);
11526 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
11530 /* Handle extended ${^Foo} variables
11531 * 1999-02-27 mjd-perl-patch@plover.com */
11532 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
11536 while (isALNUM(*s) && d < e) {
11540 Perl_croak(aTHX_ ident_too_long);
11545 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
11546 PL_lex_state = LEX_INTERPEND;
11549 if (PL_lex_state == LEX_NORMAL) {
11550 if (ckWARN(WARN_AMBIGUOUS) &&
11551 (keyword(dest, d - dest, 0)
11552 || get_cvn_flags(dest, d - dest, 0)))
11556 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
11557 "Ambiguous use of %c{%s} resolved to %c%s",
11558 funny, dest, funny, dest);
11563 s = bracket; /* let the parser handle it */
11567 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
11568 PL_lex_state = LEX_INTERPEND;
11573 S_pmflag(U32 pmfl, const char ch) {
11575 CASE_STD_PMMOD_FLAGS_PARSE_SET(&pmfl);
11576 case GLOBAL_PAT_MOD: pmfl |= PMf_GLOBAL; break;
11577 case CONTINUE_PAT_MOD: pmfl |= PMf_CONTINUE; break;
11578 case ONCE_PAT_MOD: pmfl |= PMf_KEEP; break;
11579 case KEEPCOPY_PAT_MOD: pmfl |= PMf_KEEPCOPY; break;
11585 Perl_pmflag(pTHX_ U32* pmfl, int ch)
11587 PERL_ARGS_ASSERT_PMFLAG;
11589 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED),
11590 "Perl_pmflag() is deprecated, and will be removed from the XS API");
11593 *pmfl = S_pmflag(*pmfl, (char)ch);
11598 S_scan_pat(pTHX_ char *start, I32 type)
11602 char *s = scan_str(start,!!PL_madskills,FALSE);
11603 const char * const valid_flags =
11604 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
11609 PERL_ARGS_ASSERT_SCAN_PAT;
11612 const char * const delimiter = skipspace(start);
11616 ? "Search pattern not terminated or ternary operator parsed as search pattern"
11617 : "Search pattern not terminated" ));
11620 pm = (PMOP*)newPMOP(type, 0);
11621 if (PL_multi_open == '?') {
11622 /* This is the only point in the code that sets PMf_ONCE: */
11623 pm->op_pmflags |= PMf_ONCE;
11625 /* Hence it's safe to do this bit of PMOP book-keeping here, which
11626 allows us to restrict the list needed by reset to just the ??
11628 assert(type != OP_TRANS);
11630 MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
11633 mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
11636 elements = mg->mg_len / sizeof(PMOP**);
11637 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
11638 ((PMOP**)mg->mg_ptr) [elements++] = pm;
11639 mg->mg_len = elements * sizeof(PMOP**);
11640 PmopSTASH_set(pm,PL_curstash);
11646 while (*s && strchr(valid_flags, *s))
11647 pm->op_pmflags = S_pmflag(pm->op_pmflags, *s++);
11649 if (PL_madskills && modstart != s) {
11650 SV* tmptoken = newSVpvn(modstart, s - modstart);
11651 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11654 /* issue a warning if /c is specified,but /g is not */
11655 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
11657 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP),
11658 "Use of /c modifier is meaningless without /g" );
11661 PL_lex_op = (OP*)pm;
11662 pl_yylval.ival = OP_MATCH;
11667 S_scan_subst(pTHX_ char *start)
11678 PERL_ARGS_ASSERT_SCAN_SUBST;
11680 pl_yylval.ival = OP_NULL;
11682 s = scan_str(start,!!PL_madskills,FALSE);
11685 Perl_croak(aTHX_ "Substitution pattern not terminated");
11687 if (s[-1] == PL_multi_open)
11690 if (PL_madskills) {
11691 CURMAD('q', PL_thisopen);
11692 CURMAD('_', PL_thiswhite);
11693 CURMAD('E', PL_thisstuff);
11694 CURMAD('Q', PL_thisclose);
11695 PL_realtokenstart = s - SvPVX(PL_linestr);
11699 first_start = PL_multi_start;
11700 s = scan_str(s,!!PL_madskills,FALSE);
11702 if (PL_lex_stuff) {
11703 SvREFCNT_dec(PL_lex_stuff);
11704 PL_lex_stuff = NULL;
11706 Perl_croak(aTHX_ "Substitution replacement not terminated");
11708 PL_multi_start = first_start; /* so whole substitution is taken together */
11710 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11713 if (PL_madskills) {
11714 CURMAD('z', PL_thisopen);
11715 CURMAD('R', PL_thisstuff);
11716 CURMAD('Z', PL_thisclose);
11722 if (*s == EXEC_PAT_MOD) {
11726 else if (strchr(S_PAT_MODS, *s))
11727 pm->op_pmflags = S_pmflag(pm->op_pmflags, *s++);
11733 if (PL_madskills) {
11735 curmad('m', newSVpvn(modstart, s - modstart));
11736 append_madprops(PL_thismad, (OP*)pm, 0);
11740 if ((pm->op_pmflags & PMf_CONTINUE)) {
11741 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11745 SV * const repl = newSVpvs("");
11747 PL_sublex_info.super_bufptr = s;
11748 PL_sublex_info.super_bufend = PL_bufend;
11750 pm->op_pmflags |= PMf_EVAL;
11753 sv_catpvs(repl, "eval ");
11755 sv_catpvs(repl, "do ");
11757 sv_catpvs(repl, "{");
11758 sv_catsv(repl, PL_lex_repl);
11759 if (strchr(SvPVX(PL_lex_repl), '#'))
11760 sv_catpvs(repl, "\n");
11761 sv_catpvs(repl, "}");
11763 SvREFCNT_dec(PL_lex_repl);
11764 PL_lex_repl = repl;
11767 PL_lex_op = (OP*)pm;
11768 pl_yylval.ival = OP_SUBST;
11773 S_scan_trans(pTHX_ char *start)
11786 PERL_ARGS_ASSERT_SCAN_TRANS;
11788 pl_yylval.ival = OP_NULL;
11790 s = scan_str(start,!!PL_madskills,FALSE);
11792 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11794 if (s[-1] == PL_multi_open)
11797 if (PL_madskills) {
11798 CURMAD('q', PL_thisopen);
11799 CURMAD('_', PL_thiswhite);
11800 CURMAD('E', PL_thisstuff);
11801 CURMAD('Q', PL_thisclose);
11802 PL_realtokenstart = s - SvPVX(PL_linestr);
11806 s = scan_str(s,!!PL_madskills,FALSE);
11808 if (PL_lex_stuff) {
11809 SvREFCNT_dec(PL_lex_stuff);
11810 PL_lex_stuff = NULL;
11812 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11814 if (PL_madskills) {
11815 CURMAD('z', PL_thisopen);
11816 CURMAD('R', PL_thisstuff);
11817 CURMAD('Z', PL_thisclose);
11820 complement = del = squash = 0;
11827 complement = OPpTRANS_COMPLEMENT;
11830 del = OPpTRANS_DELETE;
11833 squash = OPpTRANS_SQUASH;
11842 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11843 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11844 o->op_private &= ~OPpTRANS_ALL;
11845 o->op_private |= del|squash|complement|
11846 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11847 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11850 pl_yylval.ival = OP_TRANS;
11853 if (PL_madskills) {
11855 curmad('m', newSVpvn(modstart, s - modstart));
11856 append_madprops(PL_thismad, o, 0);
11865 S_scan_heredoc(pTHX_ register char *s)
11869 I32 op_type = OP_SCALAR;
11873 const char *found_newline;
11877 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11879 I32 stuffstart = s - SvPVX(PL_linestr);
11882 PL_realtokenstart = -1;
11885 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11889 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11893 while (SPACE_OR_TAB(*peek))
11895 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11898 s = delimcpy(d, e, s, PL_bufend, term, &len);
11908 if (!isALNUM_lazy_if(s,UTF))
11909 deprecate("bare << to mean <<\"\"");
11910 for (; isALNUM_lazy_if(s,UTF); s++) {
11915 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11916 Perl_croak(aTHX_ "Delimiter for here document is too long");
11919 len = d - PL_tokenbuf;
11922 if (PL_madskills) {
11923 tstart = PL_tokenbuf + !outer;
11924 PL_thisclose = newSVpvn(tstart, len - !outer);
11925 tstart = SvPVX(PL_linestr) + stuffstart;
11926 PL_thisopen = newSVpvn(tstart, s - tstart);
11927 stuffstart = s - SvPVX(PL_linestr);
11930 #ifndef PERL_STRICT_CR
11931 d = strchr(s, '\r');
11933 char * const olds = s;
11935 while (s < PL_bufend) {
11941 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11950 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11957 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11958 herewas = newSVpvn(s,PL_bufend-s);
11962 herewas = newSVpvn(s-1,found_newline-s+1);
11965 herewas = newSVpvn(s,found_newline-s);
11969 if (PL_madskills) {
11970 tstart = SvPVX(PL_linestr) + stuffstart;
11972 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11974 PL_thisstuff = newSVpvn(tstart, s - tstart);
11977 s += SvCUR(herewas);
11980 stuffstart = s - SvPVX(PL_linestr);
11986 tmpstr = newSV_type(SVt_PVIV);
11987 SvGROW(tmpstr, 80);
11988 if (term == '\'') {
11989 op_type = OP_CONST;
11990 SvIV_set(tmpstr, -1);
11992 else if (term == '`') {
11993 op_type = OP_BACKTICK;
11994 SvIV_set(tmpstr, '\\');
11998 PL_multi_start = CopLINE(PL_curcop);
11999 PL_multi_open = PL_multi_close = '<';
12000 term = *PL_tokenbuf;
12001 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
12002 char * const bufptr = PL_sublex_info.super_bufptr;
12003 char * const bufend = PL_sublex_info.super_bufend;
12004 char * const olds = s - SvCUR(herewas);
12005 s = strchr(bufptr, '\n');
12009 while (s < bufend &&
12010 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
12012 CopLINE_inc(PL_curcop);
12015 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12016 missingterm(PL_tokenbuf);
12018 sv_setpvn(herewas,bufptr,d-bufptr+1);
12019 sv_setpvn(tmpstr,d+1,s-d);
12021 sv_catpvn(herewas,s,bufend-s);
12022 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
12029 while (s < PL_bufend &&
12030 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
12032 CopLINE_inc(PL_curcop);
12034 if (s >= PL_bufend) {
12035 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12036 missingterm(PL_tokenbuf);
12038 sv_setpvn(tmpstr,d+1,s-d);
12040 if (PL_madskills) {
12042 sv_catpvn(PL_thisstuff, d + 1, s - d);
12044 PL_thisstuff = newSVpvn(d + 1, s - d);
12045 stuffstart = s - SvPVX(PL_linestr);
12049 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
12051 sv_catpvn(herewas,s,PL_bufend-s);
12052 sv_setsv(PL_linestr,herewas);
12053 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
12054 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12055 PL_last_lop = PL_last_uni = NULL;
12058 sv_setpvs(tmpstr,""); /* avoid "uninitialized" warning */
12059 while (s >= PL_bufend) { /* multiple line string? */
12061 if (PL_madskills) {
12062 tstart = SvPVX(PL_linestr) + stuffstart;
12064 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
12066 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
12070 CopLINE_inc(PL_curcop);
12071 if (!outer || !lex_next_chunk(0)) {
12072 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12073 missingterm(PL_tokenbuf);
12075 CopLINE_dec(PL_curcop);
12078 stuffstart = s - SvPVX(PL_linestr);
12080 CopLINE_inc(PL_curcop);
12081 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12082 PL_last_lop = PL_last_uni = NULL;
12083 #ifndef PERL_STRICT_CR
12084 if (PL_bufend - PL_linestart >= 2) {
12085 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
12086 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
12088 PL_bufend[-2] = '\n';
12090 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
12092 else if (PL_bufend[-1] == '\r')
12093 PL_bufend[-1] = '\n';
12095 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
12096 PL_bufend[-1] = '\n';
12098 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
12099 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
12100 *(SvPVX(PL_linestr) + off ) = ' ';
12101 sv_catsv(PL_linestr,herewas);
12102 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12103 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
12107 sv_catsv(tmpstr,PL_linestr);
12112 PL_multi_end = CopLINE(PL_curcop);
12113 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
12114 SvPV_shrink_to_cur(tmpstr);
12116 SvREFCNT_dec(herewas);
12118 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
12120 else if (PL_encoding)
12121 sv_recode_to_utf8(tmpstr, PL_encoding);
12123 PL_lex_stuff = tmpstr;
12124 pl_yylval.ival = op_type;
12128 /* scan_inputsymbol
12129 takes: current position in input buffer
12130 returns: new position in input buffer
12131 side-effects: pl_yylval and lex_op are set.
12136 <FH> read from filehandle
12137 <pkg::FH> read from package qualified filehandle
12138 <pkg'FH> read from package qualified filehandle
12139 <$fh> read from filehandle in $fh
12140 <*.h> filename glob
12145 S_scan_inputsymbol(pTHX_ char *start)
12148 register char *s = start; /* current position in buffer */
12151 char *d = PL_tokenbuf; /* start of temp holding space */
12152 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
12154 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
12156 end = strchr(s, '\n');
12159 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
12161 /* die if we didn't have space for the contents of the <>,
12162 or if it didn't end, or if we see a newline
12165 if (len >= (I32)sizeof PL_tokenbuf)
12166 Perl_croak(aTHX_ "Excessively long <> operator");
12168 Perl_croak(aTHX_ "Unterminated <> operator");
12173 Remember, only scalar variables are interpreted as filehandles by
12174 this code. Anything more complex (e.g., <$fh{$num}>) will be
12175 treated as a glob() call.
12176 This code makes use of the fact that except for the $ at the front,
12177 a scalar variable and a filehandle look the same.
12179 if (*d == '$' && d[1]) d++;
12181 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
12182 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
12185 /* If we've tried to read what we allow filehandles to look like, and
12186 there's still text left, then it must be a glob() and not a getline.
12187 Use scan_str to pull out the stuff between the <> and treat it
12188 as nothing more than a string.
12191 if (d - PL_tokenbuf != len) {
12192 pl_yylval.ival = OP_GLOB;
12193 s = scan_str(start,!!PL_madskills,FALSE);
12195 Perl_croak(aTHX_ "Glob not terminated");
12199 bool readline_overriden = FALSE;
12202 /* we're in a filehandle read situation */
12205 /* turn <> into <ARGV> */
12207 Copy("ARGV",d,5,char);
12209 /* Check whether readline() is overriden */
12210 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
12212 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
12214 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
12215 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
12216 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
12217 readline_overriden = TRUE;
12219 /* if <$fh>, create the ops to turn the variable into a
12223 /* try to find it in the pad for this block, otherwise find
12224 add symbol table ops
12226 const PADOFFSET tmp = pad_findmy(d, len, 0);
12227 if (tmp != NOT_IN_PAD) {
12228 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
12229 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
12230 HEK * const stashname = HvNAME_HEK(stash);
12231 SV * const sym = sv_2mortal(newSVhek(stashname));
12232 sv_catpvs(sym, "::");
12233 sv_catpv(sym, d+1);
12238 OP * const o = newOP(OP_PADSV, 0);
12240 PL_lex_op = readline_overriden
12241 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
12242 append_elem(OP_LIST, o,
12243 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
12244 : (OP*)newUNOP(OP_READLINE, 0, o);
12253 ? (GV_ADDMULTI | GV_ADDINEVAL)
12256 PL_lex_op = readline_overriden
12257 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
12258 append_elem(OP_LIST,
12259 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
12260 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
12261 : (OP*)newUNOP(OP_READLINE, 0,
12262 newUNOP(OP_RV2SV, 0,
12263 newGVOP(OP_GV, 0, gv)));
12265 if (!readline_overriden)
12266 PL_lex_op->op_flags |= OPf_SPECIAL;
12267 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
12268 pl_yylval.ival = OP_NULL;
12271 /* If it's none of the above, it must be a literal filehandle
12272 (<Foo::BAR> or <FOO>) so build a simple readline OP */
12274 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
12275 PL_lex_op = readline_overriden
12276 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
12277 append_elem(OP_LIST,
12278 newGVOP(OP_GV, 0, gv),
12279 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
12280 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
12281 pl_yylval.ival = OP_NULL;
12290 takes: start position in buffer
12291 keep_quoted preserve \ on the embedded delimiter(s)
12292 keep_delims preserve the delimiters around the string
12293 returns: position to continue reading from buffer
12294 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
12295 updates the read buffer.
12297 This subroutine pulls a string out of the input. It is called for:
12298 q single quotes q(literal text)
12299 ' single quotes 'literal text'
12300 qq double quotes qq(interpolate $here please)
12301 " double quotes "interpolate $here please"
12302 qx backticks qx(/bin/ls -l)
12303 ` backticks `/bin/ls -l`
12304 qw quote words @EXPORT_OK = qw( func() $spam )
12305 m// regexp match m/this/
12306 s/// regexp substitute s/this/that/
12307 tr/// string transliterate tr/this/that/
12308 y/// string transliterate y/this/that/
12309 ($*@) sub prototypes sub foo ($)
12310 (stuff) sub attr parameters sub foo : attr(stuff)
12311 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
12313 In most of these cases (all but <>, patterns and transliterate)
12314 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
12315 calls scan_str(). s/// makes yylex() call scan_subst() which calls
12316 scan_str(). tr/// and y/// make yylex() call scan_trans() which
12319 It skips whitespace before the string starts, and treats the first
12320 character as the delimiter. If the delimiter is one of ([{< then
12321 the corresponding "close" character )]}> is used as the closing
12322 delimiter. It allows quoting of delimiters, and if the string has
12323 balanced delimiters ([{<>}]) it allows nesting.
12325 On success, the SV with the resulting string is put into lex_stuff or,
12326 if that is already non-NULL, into lex_repl. The second case occurs only
12327 when parsing the RHS of the special constructs s/// and tr/// (y///).
12328 For convenience, the terminating delimiter character is stuffed into
12333 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
12336 SV *sv; /* scalar value: string */
12337 const char *tmps; /* temp string, used for delimiter matching */
12338 register char *s = start; /* current position in the buffer */
12339 register char term; /* terminating character */
12340 register char *to; /* current position in the sv's data */
12341 I32 brackets = 1; /* bracket nesting level */
12342 bool has_utf8 = FALSE; /* is there any utf8 content? */
12343 I32 termcode; /* terminating char. code */
12344 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
12345 STRLEN termlen; /* length of terminating string */
12346 int last_off = 0; /* last position for nesting bracket */
12352 PERL_ARGS_ASSERT_SCAN_STR;
12354 /* skip space before the delimiter */
12360 if (PL_realtokenstart >= 0) {
12361 stuffstart = PL_realtokenstart;
12362 PL_realtokenstart = -1;
12365 stuffstart = start - SvPVX(PL_linestr);
12367 /* mark where we are, in case we need to report errors */
12370 /* after skipping whitespace, the next character is the terminator */
12373 termcode = termstr[0] = term;
12377 termcode = utf8_to_uvchr((U8*)s, &termlen);
12378 Copy(s, termstr, termlen, U8);
12379 if (!UTF8_IS_INVARIANT(term))
12383 /* mark where we are */
12384 PL_multi_start = CopLINE(PL_curcop);
12385 PL_multi_open = term;
12387 /* find corresponding closing delimiter */
12388 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
12389 termcode = termstr[0] = term = tmps[5];
12391 PL_multi_close = term;
12393 /* create a new SV to hold the contents. 79 is the SV's initial length.
12394 What a random number. */
12395 sv = newSV_type(SVt_PVIV);
12397 SvIV_set(sv, termcode);
12398 (void)SvPOK_only(sv); /* validate pointer */
12400 /* move past delimiter and try to read a complete string */
12402 sv_catpvn(sv, s, termlen);
12405 tstart = SvPVX(PL_linestr) + stuffstart;
12406 if (!PL_thisopen && !keep_delims) {
12407 PL_thisopen = newSVpvn(tstart, s - tstart);
12408 stuffstart = s - SvPVX(PL_linestr);
12412 if (PL_encoding && !UTF) {
12416 int offset = s - SvPVX_const(PL_linestr);
12417 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
12418 &offset, (char*)termstr, termlen);
12419 const char * const ns = SvPVX_const(PL_linestr) + offset;
12420 char * const svlast = SvEND(sv) - 1;
12422 for (; s < ns; s++) {
12423 if (*s == '\n' && !PL_rsfp)
12424 CopLINE_inc(PL_curcop);
12427 goto read_more_line;
12429 /* handle quoted delimiters */
12430 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
12432 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
12434 if ((svlast-1 - t) % 2) {
12435 if (!keep_quoted) {
12436 *(svlast-1) = term;
12438 SvCUR_set(sv, SvCUR(sv) - 1);
12443 if (PL_multi_open == PL_multi_close) {
12449 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
12450 /* At here, all closes are "was quoted" one,
12451 so we don't check PL_multi_close. */
12453 if (!keep_quoted && *(t+1) == PL_multi_open)
12458 else if (*t == PL_multi_open)
12466 SvCUR_set(sv, w - SvPVX_const(sv));
12468 last_off = w - SvPVX(sv);
12469 if (--brackets <= 0)
12474 if (!keep_delims) {
12475 SvCUR_set(sv, SvCUR(sv) - 1);
12481 /* extend sv if need be */
12482 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
12483 /* set 'to' to the next character in the sv's string */
12484 to = SvPVX(sv)+SvCUR(sv);
12486 /* if open delimiter is the close delimiter read unbridle */
12487 if (PL_multi_open == PL_multi_close) {
12488 for (; s < PL_bufend; s++,to++) {
12489 /* embedded newlines increment the current line number */
12490 if (*s == '\n' && !PL_rsfp)
12491 CopLINE_inc(PL_curcop);
12492 /* handle quoted delimiters */
12493 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
12494 if (!keep_quoted && s[1] == term)
12496 /* any other quotes are simply copied straight through */
12500 /* terminate when run out of buffer (the for() condition), or
12501 have found the terminator */
12502 else if (*s == term) {
12505 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
12508 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
12514 /* if the terminator isn't the same as the start character (e.g.,
12515 matched brackets), we have to allow more in the quoting, and
12516 be prepared for nested brackets.
12519 /* read until we run out of string, or we find the terminator */
12520 for (; s < PL_bufend; s++,to++) {
12521 /* embedded newlines increment the line count */
12522 if (*s == '\n' && !PL_rsfp)
12523 CopLINE_inc(PL_curcop);
12524 /* backslashes can escape the open or closing characters */
12525 if (*s == '\\' && s+1 < PL_bufend) {
12526 if (!keep_quoted &&
12527 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
12532 /* allow nested opens and closes */
12533 else if (*s == PL_multi_close && --brackets <= 0)
12535 else if (*s == PL_multi_open)
12537 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
12542 /* terminate the copied string and update the sv's end-of-string */
12544 SvCUR_set(sv, to - SvPVX_const(sv));
12547 * this next chunk reads more into the buffer if we're not done yet
12551 break; /* handle case where we are done yet :-) */
12553 #ifndef PERL_STRICT_CR
12554 if (to - SvPVX_const(sv) >= 2) {
12555 if ((to[-2] == '\r' && to[-1] == '\n') ||
12556 (to[-2] == '\n' && to[-1] == '\r'))
12560 SvCUR_set(sv, to - SvPVX_const(sv));
12562 else if (to[-1] == '\r')
12565 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
12570 /* if we're out of file, or a read fails, bail and reset the current
12571 line marker so we can report where the unterminated string began
12574 if (PL_madskills) {
12575 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12577 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
12579 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
12582 CopLINE_inc(PL_curcop);
12583 PL_bufptr = PL_bufend;
12584 if (!lex_next_chunk(0)) {
12586 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12595 /* at this point, we have successfully read the delimited string */
12597 if (!PL_encoding || UTF) {
12599 if (PL_madskills) {
12600 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12601 const int len = s - tstart;
12603 sv_catpvn(PL_thisstuff, tstart, len);
12605 PL_thisstuff = newSVpvn(tstart, len);
12606 if (!PL_thisclose && !keep_delims)
12607 PL_thisclose = newSVpvn(s,termlen);
12612 sv_catpvn(sv, s, termlen);
12617 if (PL_madskills) {
12618 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12619 const int len = s - tstart - termlen;
12621 sv_catpvn(PL_thisstuff, tstart, len);
12623 PL_thisstuff = newSVpvn(tstart, len);
12624 if (!PL_thisclose && !keep_delims)
12625 PL_thisclose = newSVpvn(s - termlen,termlen);
12629 if (has_utf8 || PL_encoding)
12632 PL_multi_end = CopLINE(PL_curcop);
12634 /* if we allocated too much space, give some back */
12635 if (SvCUR(sv) + 5 < SvLEN(sv)) {
12636 SvLEN_set(sv, SvCUR(sv) + 1);
12637 SvPV_renew(sv, SvLEN(sv));
12640 /* decide whether this is the first or second quoted string we've read
12653 takes: pointer to position in buffer
12654 returns: pointer to new position in buffer
12655 side-effects: builds ops for the constant in pl_yylval.op
12657 Read a number in any of the formats that Perl accepts:
12659 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12660 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12663 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12665 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12668 If it reads a number without a decimal point or an exponent, it will
12669 try converting the number to an integer and see if it can do so
12670 without loss of precision.
12674 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12677 register const char *s = start; /* current position in buffer */
12678 register char *d; /* destination in temp buffer */
12679 register char *e; /* end of temp buffer */
12680 NV nv; /* number read, as a double */
12681 SV *sv = NULL; /* place to put the converted number */
12682 bool floatit; /* boolean: int or float? */
12683 const char *lastub = NULL; /* position of last underbar */
12684 static char const number_too_long[] = "Number too long";
12686 PERL_ARGS_ASSERT_SCAN_NUM;
12688 /* We use the first character to decide what type of number this is */
12692 Perl_croak(aTHX_ "panic: scan_num");
12694 /* if it starts with a 0, it could be an octal number, a decimal in
12695 0.13 disguise, or a hexadecimal number, or a binary number. */
12699 u holds the "number so far"
12700 shift the power of 2 of the base
12701 (hex == 4, octal == 3, binary == 1)
12702 overflowed was the number more than we can hold?
12704 Shift is used when we add a digit. It also serves as an "are
12705 we in octal/hex/binary?" indicator to disallow hex characters
12706 when in octal mode.
12711 bool overflowed = FALSE;
12712 bool just_zero = TRUE; /* just plain 0 or binary number? */
12713 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12714 static const char* const bases[5] =
12715 { "", "binary", "", "octal", "hexadecimal" };
12716 static const char* const Bases[5] =
12717 { "", "Binary", "", "Octal", "Hexadecimal" };
12718 static const char* const maxima[5] =
12720 "0b11111111111111111111111111111111",
12724 const char *base, *Base, *max;
12726 /* check for hex */
12731 } else if (s[1] == 'b') {
12736 /* check for a decimal in disguise */
12737 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12739 /* so it must be octal */
12746 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12747 "Misplaced _ in number");
12751 base = bases[shift];
12752 Base = Bases[shift];
12753 max = maxima[shift];
12755 /* read the rest of the number */
12757 /* x is used in the overflow test,
12758 b is the digit we're adding on. */
12763 /* if we don't mention it, we're done */
12767 /* _ are ignored -- but warned about if consecutive */
12769 if (lastub && s == lastub + 1)
12770 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12771 "Misplaced _ in number");
12775 /* 8 and 9 are not octal */
12776 case '8': case '9':
12778 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12782 case '2': case '3': case '4':
12783 case '5': case '6': case '7':
12785 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12788 case '0': case '1':
12789 b = *s++ & 15; /* ASCII digit -> value of digit */
12793 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12794 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12795 /* make sure they said 0x */
12798 b = (*s++ & 7) + 9;
12800 /* Prepare to put the digit we have onto the end
12801 of the number so far. We check for overflows.
12807 x = u << shift; /* make room for the digit */
12809 if ((x >> shift) != u
12810 && !(PL_hints & HINT_NEW_BINARY)) {
12813 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
12814 "Integer overflow in %s number",
12817 u = x | b; /* add the digit to the end */
12820 n *= nvshift[shift];
12821 /* If an NV has not enough bits in its
12822 * mantissa to represent an UV this summing of
12823 * small low-order numbers is a waste of time
12824 * (because the NV cannot preserve the
12825 * low-order bits anyway): we could just
12826 * remember when did we overflow and in the
12827 * end just multiply n by the right
12835 /* if we get here, we had success: make a scalar value from
12840 /* final misplaced underbar check */
12841 if (s[-1] == '_') {
12842 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12847 if (n > 4294967295.0)
12848 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
12849 "%s number > %s non-portable",
12855 if (u > 0xffffffff)
12856 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
12857 "%s number > %s non-portable",
12862 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12863 sv = new_constant(start, s - start, "integer",
12864 sv, NULL, NULL, 0);
12865 else if (PL_hints & HINT_NEW_BINARY)
12866 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12871 handle decimal numbers.
12872 we're also sent here when we read a 0 as the first digit
12874 case '1': case '2': case '3': case '4': case '5':
12875 case '6': case '7': case '8': case '9': case '.':
12878 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12881 /* read next group of digits and _ and copy into d */
12882 while (isDIGIT(*s) || *s == '_') {
12883 /* skip underscores, checking for misplaced ones
12887 if (lastub && s == lastub + 1)
12888 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12889 "Misplaced _ in number");
12893 /* check for end of fixed-length buffer */
12895 Perl_croak(aTHX_ number_too_long);
12896 /* if we're ok, copy the character */
12901 /* final misplaced underbar check */
12902 if (lastub && s == lastub + 1) {
12903 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12906 /* read a decimal portion if there is one. avoid
12907 3..5 being interpreted as the number 3. followed
12910 if (*s == '.' && s[1] != '.') {
12915 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12916 "Misplaced _ in number");
12920 /* copy, ignoring underbars, until we run out of digits.
12922 for (; isDIGIT(*s) || *s == '_'; s++) {
12923 /* fixed length buffer check */
12925 Perl_croak(aTHX_ number_too_long);
12927 if (lastub && s == lastub + 1)
12928 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12929 "Misplaced _ in number");
12935 /* fractional part ending in underbar? */
12936 if (s[-1] == '_') {
12937 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12938 "Misplaced _ in number");
12940 if (*s == '.' && isDIGIT(s[1])) {
12941 /* oops, it's really a v-string, but without the "v" */
12947 /* read exponent part, if present */
12948 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12952 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12953 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12955 /* stray preinitial _ */
12957 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12958 "Misplaced _ in number");
12962 /* allow positive or negative exponent */
12963 if (*s == '+' || *s == '-')
12966 /* stray initial _ */
12968 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12969 "Misplaced _ in number");
12973 /* read digits of exponent */
12974 while (isDIGIT(*s) || *s == '_') {
12977 Perl_croak(aTHX_ number_too_long);
12981 if (((lastub && s == lastub + 1) ||
12982 (!isDIGIT(s[1]) && s[1] != '_')))
12983 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12984 "Misplaced _ in number");
12991 /* make an sv from the string */
12995 We try to do an integer conversion first if no characters
12996 indicating "float" have been found.
13001 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
13003 if (flags == IS_NUMBER_IN_UV) {
13005 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
13008 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
13009 if (uv <= (UV) IV_MIN)
13010 sv_setiv(sv, -(IV)uv);
13017 /* terminate the string */
13019 nv = Atof(PL_tokenbuf);
13024 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
13025 const char *const key = floatit ? "float" : "integer";
13026 const STRLEN keylen = floatit ? 5 : 7;
13027 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
13028 key, keylen, sv, NULL, NULL, 0);
13032 /* if it starts with a v, it could be a v-string */
13035 sv = newSV(5); /* preallocate storage space */
13036 s = scan_vstring(s, PL_bufend, sv);
13040 /* make the op for the constant and return */
13043 lvalp->opval = newSVOP(OP_CONST, 0, sv);
13045 lvalp->opval = NULL;
13051 S_scan_formline(pTHX_ register char *s)
13054 register char *eol;
13056 SV * const stuff = newSVpvs("");
13057 bool needargs = FALSE;
13058 bool eofmt = FALSE;
13060 char *tokenstart = s;
13061 SV* savewhite = NULL;
13063 if (PL_madskills) {
13064 savewhite = PL_thiswhite;
13069 PERL_ARGS_ASSERT_SCAN_FORMLINE;
13071 while (!needargs) {
13074 #ifdef PERL_STRICT_CR
13075 while (SPACE_OR_TAB(*t))
13078 while (SPACE_OR_TAB(*t) || *t == '\r')
13081 if (*t == '\n' || t == PL_bufend) {
13086 if (PL_in_eval && !PL_rsfp) {
13087 eol = (char *) memchr(s,'\n',PL_bufend-s);
13092 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
13094 for (t = s; t < eol; t++) {
13095 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
13097 goto enough; /* ~~ must be first line in formline */
13099 if (*t == '@' || *t == '^')
13103 sv_catpvn(stuff, s, eol-s);
13104 #ifndef PERL_STRICT_CR
13105 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
13106 char *end = SvPVX(stuff) + SvCUR(stuff);
13109 SvCUR_set(stuff, SvCUR(stuff) - 1);
13120 if (PL_madskills) {
13122 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
13124 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
13127 PL_bufptr = PL_bufend;
13128 CopLINE_inc(PL_curcop);
13129 got_some = lex_next_chunk(0);
13130 CopLINE_dec(PL_curcop);
13133 tokenstart = PL_bufptr;
13141 if (SvCUR(stuff)) {
13144 PL_lex_state = LEX_NORMAL;
13145 start_force(PL_curforce);
13146 NEXTVAL_NEXTTOKE.ival = 0;
13150 PL_lex_state = LEX_FORMLINE;
13152 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
13154 else if (PL_encoding)
13155 sv_recode_to_utf8(stuff, PL_encoding);
13157 start_force(PL_curforce);
13158 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
13160 start_force(PL_curforce);
13161 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
13165 SvREFCNT_dec(stuff);
13167 PL_lex_formbrack = 0;
13171 if (PL_madskills) {
13173 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
13175 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
13176 PL_thiswhite = savewhite;
13183 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
13186 const I32 oldsavestack_ix = PL_savestack_ix;
13187 CV* const outsidecv = PL_compcv;
13190 assert(SvTYPE(PL_compcv) == SVt_PVCV);
13192 SAVEI32(PL_subline);
13193 save_item(PL_subname);
13194 SAVESPTR(PL_compcv);
13196 PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
13197 CvFLAGS(PL_compcv) |= flags;
13199 PL_subline = CopLINE(PL_curcop);
13200 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
13201 CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
13202 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
13204 return oldsavestack_ix;
13208 #pragma segment Perl_yylex
13211 S_yywarn(pTHX_ const char *const s)
13215 PERL_ARGS_ASSERT_YYWARN;
13217 PL_in_eval |= EVAL_WARNONLY;
13219 PL_in_eval &= ~EVAL_WARNONLY;
13224 Perl_yyerror(pTHX_ const char *const s)
13227 const char *where = NULL;
13228 const char *context = NULL;
13231 int yychar = PL_parser->yychar;
13233 PERL_ARGS_ASSERT_YYERROR;
13235 if (!yychar || (yychar == ';' && !PL_rsfp))
13237 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
13238 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
13239 PL_oldbufptr != PL_bufptr) {
13242 The code below is removed for NetWare because it abends/crashes on NetWare
13243 when the script has error such as not having the closing quotes like:
13244 if ($var eq "value)
13245 Checking of white spaces is anyway done in NetWare code.
13248 while (isSPACE(*PL_oldoldbufptr))
13251 context = PL_oldoldbufptr;
13252 contlen = PL_bufptr - PL_oldoldbufptr;
13254 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
13255 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
13258 The code below is removed for NetWare because it abends/crashes on NetWare
13259 when the script has error such as not having the closing quotes like:
13260 if ($var eq "value)
13261 Checking of white spaces is anyway done in NetWare code.
13264 while (isSPACE(*PL_oldbufptr))
13267 context = PL_oldbufptr;
13268 contlen = PL_bufptr - PL_oldbufptr;
13270 else if (yychar > 255)
13271 where = "next token ???";
13272 else if (yychar == -2) { /* YYEMPTY */
13273 if (PL_lex_state == LEX_NORMAL ||
13274 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
13275 where = "at end of line";
13276 else if (PL_lex_inpat)
13277 where = "within pattern";
13279 where = "within string";
13282 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
13284 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
13285 else if (isPRINT_LC(yychar)) {
13286 const char string = yychar;
13287 sv_catpvn(where_sv, &string, 1);
13290 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
13291 where = SvPVX_const(where_sv);
13293 msg = sv_2mortal(newSVpv(s, 0));
13294 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
13295 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
13297 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
13299 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
13300 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
13301 Perl_sv_catpvf(aTHX_ msg,
13302 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
13303 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
13306 if (PL_in_eval & EVAL_WARNONLY) {
13307 Perl_ck_warner_d(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
13311 if (PL_error_count >= 10) {
13312 if (PL_in_eval && SvCUR(ERRSV))
13313 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
13314 SVfARG(ERRSV), OutCopFILE(PL_curcop));
13316 Perl_croak(aTHX_ "%s has too many errors.\n",
13317 OutCopFILE(PL_curcop));
13320 PL_in_my_stash = NULL;
13324 #pragma segment Main
13328 S_swallow_bom(pTHX_ U8 *s)
13331 const STRLEN slen = SvCUR(PL_linestr);
13333 PERL_ARGS_ASSERT_SWALLOW_BOM;
13337 if (s[1] == 0xFE) {
13338 /* UTF-16 little-endian? (or UTF-32LE?) */
13339 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
13340 Perl_croak(aTHX_ "Unsupported script encoding UTF-32LE");
13341 #ifndef PERL_NO_UTF16_FILTER
13342 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (BOM)\n");
13344 if (PL_bufend > (char*)s) {
13345 s = add_utf16_textfilter(s, TRUE);
13348 Perl_croak(aTHX_ "Unsupported script encoding UTF-16LE");
13353 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
13354 #ifndef PERL_NO_UTF16_FILTER
13355 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
13357 if (PL_bufend > (char *)s) {
13358 s = add_utf16_textfilter(s, FALSE);
13361 Perl_croak(aTHX_ "Unsupported script encoding UTF-16BE");
13366 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
13367 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
13368 s += 3; /* UTF-8 */
13374 if (s[2] == 0xFE && s[3] == 0xFF) {
13375 /* UTF-32 big-endian */
13376 Perl_croak(aTHX_ "Unsupported script encoding UTF-32BE");
13379 else if (s[2] == 0 && s[3] != 0) {
13382 * are a good indicator of UTF-16BE. */
13383 #ifndef PERL_NO_UTF16_FILTER
13384 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
13385 s = add_utf16_textfilter(s, FALSE);
13387 Perl_croak(aTHX_ "Unsupported script encoding UTF-16BE");
13393 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
13394 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
13395 s += 4; /* UTF-8 */
13401 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
13404 * are a good indicator of UTF-16LE. */
13405 #ifndef PERL_NO_UTF16_FILTER
13406 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
13407 s = add_utf16_textfilter(s, TRUE);
13409 Perl_croak(aTHX_ "Unsupported script encoding UTF-16LE");
13417 #ifndef PERL_NO_UTF16_FILTER
13419 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
13422 SV *const filter = FILTER_DATA(idx);
13423 /* We re-use this each time round, throwing the contents away before we
13425 SV *const utf16_buffer = MUTABLE_SV(IoTOP_GV(filter));
13426 SV *const utf8_buffer = filter;
13427 IV status = IoPAGE(filter);
13428 const bool reverse = (bool) IoLINES(filter);
13431 /* As we're automatically added, at the lowest level, and hence only called
13432 from this file, we can be sure that we're not called in block mode. Hence
13433 don't bother writing code to deal with block mode. */
13435 Perl_croak(aTHX_ "panic: utf16_textfilter called in block mode (for %d characters)", maxlen);
13438 Perl_croak(aTHX_ "panic: utf16_textfilter called after error (status=%"IVdf")", status);
13440 DEBUG_P(PerlIO_printf(Perl_debug_log,
13441 "utf16_textfilter(%p,%ce): idx=%d maxlen=%d status=%"IVdf" utf16=%"UVuf" utf8=%"UVuf"\n",
13442 FPTR2DPTR(void *, S_utf16_textfilter),
13443 reverse ? 'l' : 'b', idx, maxlen, status,
13444 (UV)SvCUR(utf16_buffer), (UV)SvCUR(utf8_buffer)));
13451 /* First, look in our buffer of existing UTF-8 data: */
13452 char *nl = (char *)memchr(SvPVX(utf8_buffer), '\n', SvCUR(utf8_buffer));
13456 } else if (status == 0) {
13458 IoPAGE(filter) = 0;
13459 nl = SvEND(utf8_buffer);
13462 STRLEN got = nl - SvPVX(utf8_buffer);
13463 /* Did we have anything to append? */
13465 sv_catpvn(sv, SvPVX(utf8_buffer), got);
13466 /* Everything else in this code works just fine if SVp_POK isn't
13467 set. This, however, needs it, and we need it to work, else
13468 we loop infinitely because the buffer is never consumed. */
13469 sv_chop(utf8_buffer, nl);
13473 /* OK, not a complete line there, so need to read some more UTF-16.
13474 Read an extra octect if the buffer currently has an odd number. */
13478 if (SvCUR(utf16_buffer) >= 2) {
13479 /* Location of the high octet of the last complete code point.
13480 Gosh, UTF-16 is a pain. All the benefits of variable length,
13481 *coupled* with all the benefits of partial reads and
13483 const U8 *const last_hi = (U8*)SvPVX(utf16_buffer)
13484 + ((SvCUR(utf16_buffer) & ~1) - (reverse ? 1 : 2));
13486 if (*last_hi < 0xd8 || *last_hi > 0xdb) {
13490 /* We have the first half of a surrogate. Read more. */
13491 DEBUG_P(PerlIO_printf(Perl_debug_log, "utf16_textfilter partial surrogate detected at %p\n", last_hi));
13494 status = FILTER_READ(idx + 1, utf16_buffer,
13495 160 + (SvCUR(utf16_buffer) & 1));
13496 DEBUG_P(PerlIO_printf(Perl_debug_log, "utf16_textfilter status=%"IVdf" SvCUR(sv)=%"UVuf"\n", status, (UV)SvCUR(utf16_buffer)));
13497 DEBUG_P({ sv_dump(utf16_buffer); sv_dump(utf8_buffer);});
13500 IoPAGE(filter) = status;
13505 chars = SvCUR(utf16_buffer) >> 1;
13506 have = SvCUR(utf8_buffer);
13507 SvGROW(utf8_buffer, have + chars * 3 + 1);
13510 end = utf16_to_utf8_reversed((U8*)SvPVX(utf16_buffer),
13511 (U8*)SvPVX_const(utf8_buffer) + have,
13512 chars * 2, &newlen);
13514 end = utf16_to_utf8((U8*)SvPVX(utf16_buffer),
13515 (U8*)SvPVX_const(utf8_buffer) + have,
13516 chars * 2, &newlen);
13518 SvCUR_set(utf8_buffer, have + newlen);
13521 /* No need to keep this SV "well-formed" with a '\0' after the end, as
13522 it's private to us, and utf16_to_utf8{,reversed} take a
13523 (pointer,length) pair, rather than a NUL-terminated string. */
13524 if(SvCUR(utf16_buffer) & 1) {
13525 *SvPVX(utf16_buffer) = SvEND(utf16_buffer)[-1];
13526 SvCUR_set(utf16_buffer, 1);
13528 SvCUR_set(utf16_buffer, 0);
13531 DEBUG_P(PerlIO_printf(Perl_debug_log,
13532 "utf16_textfilter: returns, status=%"IVdf" utf16=%"UVuf" utf8=%"UVuf"\n",
13534 (UV)SvCUR(utf16_buffer), (UV)SvCUR(utf8_buffer)));
13535 DEBUG_P({ sv_dump(utf8_buffer); sv_dump(sv);});
13540 S_add_utf16_textfilter(pTHX_ U8 *const s, bool reversed)
13542 SV *filter = filter_add(S_utf16_textfilter, NULL);
13544 IoTOP_GV(filter) = MUTABLE_GV(newSVpvn((char *)s, PL_bufend - (char*)s));
13545 sv_setpvs(filter, "");
13546 IoLINES(filter) = reversed;
13547 IoPAGE(filter) = 1; /* Not EOF */
13549 /* Sadly, we have to return a valid pointer, come what may, so we have to
13550 ignore any error return from this. */
13551 SvCUR_set(PL_linestr, 0);
13552 if (FILTER_READ(0, PL_linestr, 0)) {
13553 SvUTF8_on(PL_linestr);
13555 SvUTF8_on(PL_linestr);
13557 PL_bufend = SvEND(PL_linestr);
13558 return (U8*)SvPVX(PL_linestr);
13563 Returns a pointer to the next character after the parsed
13564 vstring, as well as updating the passed in sv.
13566 Function must be called like
13569 s = scan_vstring(s,e,sv);
13571 where s and e are the start and end of the string.
13572 The sv should already be large enough to store the vstring
13573 passed in, for performance reasons.
13578 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
13581 const char *pos = s;
13582 const char *start = s;
13584 PERL_ARGS_ASSERT_SCAN_VSTRING;
13586 if (*pos == 'v') pos++; /* get past 'v' */
13587 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
13589 if ( *pos != '.') {
13590 /* this may not be a v-string if followed by => */
13591 const char *next = pos;
13592 while (next < e && isSPACE(*next))
13594 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
13595 /* return string not v-string */
13596 sv_setpvn(sv,(char *)s,pos-s);
13597 return (char *)pos;
13601 if (!isALPHA(*pos)) {
13602 U8 tmpbuf[UTF8_MAXBYTES+1];
13605 s++; /* get past 'v' */
13610 /* this is atoi() that tolerates underscores */
13613 const char *end = pos;
13615 while (--end >= s) {
13617 const UV orev = rev;
13618 rev += (*end - '0') * mult;
13621 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
13622 "Integer overflow in decimal number");
13626 if (rev > 0x7FFFFFFF)
13627 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
13629 /* Append native character for the rev point */
13630 tmpend = uvchr_to_utf8(tmpbuf, rev);
13631 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
13632 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
13634 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
13640 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
13644 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
13651 Perl_keyword_plugin_standard(pTHX_
13652 char *keyword_ptr, STRLEN keyword_len, OP **op_ptr)
13654 PERL_ARGS_ASSERT_KEYWORD_PLUGIN_STANDARD;
13655 PERL_UNUSED_CONTEXT;
13656 PERL_UNUSED_ARG(keyword_ptr);
13657 PERL_UNUSED_ARG(keyword_len);
13658 PERL_UNUSED_ARG(op_ptr);
13659 return KEYWORD_PLUGIN_DECLINE;
13664 * c-indentation-style: bsd
13665 * c-basic-offset: 4
13666 * indent-tabs-mode: t
13669 * ex: set ts=8 sts=4 sw=4 noet: