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 #ifdef USE_LOCALE_NUMERIC
2097 char *loc = setlocale(LC_NUMERIC, "C");
2099 s = scan_num(s, &pl_yylval);
2100 #ifdef USE_LOCALE_NUMERIC
2101 setlocale(LC_NUMERIC, loc);
2103 version = pl_yylval.opval;
2104 ver = cSVOPx(version)->op_sv;
2105 if (SvPOK(ver) && !SvNIOK(ver)) {
2106 SvUPGRADE(ver, SVt_PVNV);
2107 SvNV_set(ver, str_to_version(ver));
2108 SvNOK_on(ver); /* hint that it is a version */
2111 else if (guessing) {
2114 sv_free(PL_nextwhite); /* let next token collect whitespace */
2116 s = SvPVX(PL_linestr) + startoff;
2124 if (PL_madskills && !version) {
2125 sv_free(PL_nextwhite); /* let next token collect whitespace */
2127 s = SvPVX(PL_linestr) + startoff;
2130 /* NOTE: The parser sees the package name and the VERSION swapped */
2131 start_force(PL_curforce);
2132 NEXTVAL_NEXTTOKE.opval = version;
2139 * S_force_strict_version
2140 * Forces the next token to be a version number using strict syntax rules.
2144 S_force_strict_version(pTHX_ char *s)
2149 I32 startoff = s - SvPVX(PL_linestr);
2151 const char *errstr = NULL;
2153 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2155 while (isSPACE(*s)) /* leading whitespace */
2158 if (is_STRICT_VERSION(s,&errstr)) {
2160 s = (char *)scan_version(s, ver, 0);
2161 version = newSVOP(OP_CONST, 0, ver);
2163 else if ( (*s != ';' && *s != '}' ) && (s = SKIPSPACE1(s), (*s != ';' && *s !='}' ))) {
2166 yyerror(errstr); /* version required */
2171 if (PL_madskills && !version) {
2172 sv_free(PL_nextwhite); /* let next token collect whitespace */
2174 s = SvPVX(PL_linestr) + startoff;
2177 /* NOTE: The parser sees the package name and the VERSION swapped */
2178 start_force(PL_curforce);
2179 NEXTVAL_NEXTTOKE.opval = version;
2187 * Tokenize a quoted string passed in as an SV. It finds the next
2188 * chunk, up to end of string or a backslash. It may make a new
2189 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2194 S_tokeq(pTHX_ SV *sv)
2198 register char *send;
2203 PERL_ARGS_ASSERT_TOKEQ;
2208 s = SvPV_force(sv, len);
2209 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
2212 while (s < send && *s != '\\')
2217 if ( PL_hints & HINT_NEW_STRING ) {
2218 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
2222 if (s + 1 < send && (s[1] == '\\'))
2223 s++; /* all that, just for this */
2228 SvCUR_set(sv, d - SvPVX_const(sv));
2230 if ( PL_hints & HINT_NEW_STRING )
2231 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
2236 * Now come three functions related to double-quote context,
2237 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2238 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2239 * interact with PL_lex_state, and create fake ( ... ) argument lists
2240 * to handle functions and concatenation.
2241 * They assume that whoever calls them will be setting up a fake
2242 * join call, because each subthing puts a ',' after it. This lets
2245 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
2247 * (I'm not sure whether the spurious commas at the end of lcfirst's
2248 * arguments and join's arguments are created or not).
2253 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
2255 * Pattern matching will set PL_lex_op to the pattern-matching op to
2256 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
2258 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2260 * Everything else becomes a FUNC.
2262 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2263 * had an OP_CONST or OP_READLINE). This just sets us up for a
2264 * call to S_sublex_push().
2268 S_sublex_start(pTHX)
2271 register const I32 op_type = pl_yylval.ival;
2273 if (op_type == OP_NULL) {
2274 pl_yylval.opval = PL_lex_op;
2278 if (op_type == OP_CONST || op_type == OP_READLINE) {
2279 SV *sv = tokeq(PL_lex_stuff);
2281 if (SvTYPE(sv) == SVt_PVIV) {
2282 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2284 const char * const p = SvPV_const(sv, len);
2285 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
2289 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
2290 PL_lex_stuff = NULL;
2291 /* Allow <FH> // "foo" */
2292 if (op_type == OP_READLINE)
2293 PL_expect = XTERMORDORDOR;
2296 else if (op_type == OP_BACKTICK && PL_lex_op) {
2297 /* readpipe() vas overriden */
2298 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
2299 pl_yylval.opval = PL_lex_op;
2301 PL_lex_stuff = NULL;
2305 PL_sublex_info.super_state = PL_lex_state;
2306 PL_sublex_info.sub_inwhat = (U16)op_type;
2307 PL_sublex_info.sub_op = PL_lex_op;
2308 PL_lex_state = LEX_INTERPPUSH;
2312 pl_yylval.opval = PL_lex_op;
2322 * Create a new scope to save the lexing state. The scope will be
2323 * ended in S_sublex_done. Returns a '(', starting the function arguments
2324 * to the uc, lc, etc. found before.
2325 * Sets PL_lex_state to LEX_INTERPCONCAT.
2334 PL_lex_state = PL_sublex_info.super_state;
2335 SAVEBOOL(PL_lex_dojoin);
2336 SAVEI32(PL_lex_brackets);
2337 SAVEI32(PL_lex_casemods);
2338 SAVEI32(PL_lex_starts);
2339 SAVEI8(PL_lex_state);
2340 SAVEVPTR(PL_lex_inpat);
2341 SAVEI16(PL_lex_inwhat);
2342 SAVECOPLINE(PL_curcop);
2343 SAVEPPTR(PL_bufptr);
2344 SAVEPPTR(PL_bufend);
2345 SAVEPPTR(PL_oldbufptr);
2346 SAVEPPTR(PL_oldoldbufptr);
2347 SAVEPPTR(PL_last_lop);
2348 SAVEPPTR(PL_last_uni);
2349 SAVEPPTR(PL_linestart);
2350 SAVESPTR(PL_linestr);
2351 SAVEGENERICPV(PL_lex_brackstack);
2352 SAVEGENERICPV(PL_lex_casestack);
2354 PL_linestr = PL_lex_stuff;
2355 PL_lex_stuff = NULL;
2357 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2358 = SvPVX(PL_linestr);
2359 PL_bufend += SvCUR(PL_linestr);
2360 PL_last_lop = PL_last_uni = NULL;
2361 SAVEFREESV(PL_linestr);
2363 PL_lex_dojoin = FALSE;
2364 PL_lex_brackets = 0;
2365 Newx(PL_lex_brackstack, 120, char);
2366 Newx(PL_lex_casestack, 12, char);
2367 PL_lex_casemods = 0;
2368 *PL_lex_casestack = '\0';
2370 PL_lex_state = LEX_INTERPCONCAT;
2371 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
2373 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
2374 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2375 PL_lex_inpat = PL_sublex_info.sub_op;
2377 PL_lex_inpat = NULL;
2384 * Restores lexer state after a S_sublex_push.
2391 if (!PL_lex_starts++) {
2392 SV * const sv = newSVpvs("");
2393 if (SvUTF8(PL_linestr))
2395 PL_expect = XOPERATOR;
2396 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2400 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2401 PL_lex_state = LEX_INTERPCASEMOD;
2405 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
2406 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2407 PL_linestr = PL_lex_repl;
2409 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2410 PL_bufend += SvCUR(PL_linestr);
2411 PL_last_lop = PL_last_uni = NULL;
2412 SAVEFREESV(PL_linestr);
2413 PL_lex_dojoin = FALSE;
2414 PL_lex_brackets = 0;
2415 PL_lex_casemods = 0;
2416 *PL_lex_casestack = '\0';
2418 if (SvEVALED(PL_lex_repl)) {
2419 PL_lex_state = LEX_INTERPNORMAL;
2421 /* we don't clear PL_lex_repl here, so that we can check later
2422 whether this is an evalled subst; that means we rely on the
2423 logic to ensure sublex_done() is called again only via the
2424 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
2427 PL_lex_state = LEX_INTERPCONCAT;
2437 PL_endwhite = newSVpvs("");
2438 sv_catsv(PL_endwhite, PL_thiswhite);
2442 sv_setpvs(PL_thistoken,"");
2444 PL_realtokenstart = -1;
2448 PL_bufend = SvPVX(PL_linestr);
2449 PL_bufend += SvCUR(PL_linestr);
2450 PL_expect = XOPERATOR;
2451 PL_sublex_info.sub_inwhat = 0;
2459 Extracts a pattern, double-quoted string, or transliteration. This
2462 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
2463 processing a pattern (PL_lex_inpat is true), a transliteration
2464 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
2466 Returns a pointer to the character scanned up to. If this is
2467 advanced from the start pointer supplied (i.e. if anything was
2468 successfully parsed), will leave an OP for the substring scanned
2469 in pl_yylval. Caller must intuit reason for not parsing further
2470 by looking at the next characters herself.
2474 double-quoted style: \r and \n
2475 regexp special ones: \D \s
2478 case and quoting: \U \Q \E
2479 stops on @ and $, but not for $ as tail anchor
2481 In transliterations:
2482 characters are VERY literal, except for - not at the start or end
2483 of the string, which indicates a range. If the range is in bytes,
2484 scan_const expands the range to the full set of intermediate
2485 characters. If the range is in utf8, the hyphen is replaced with
2486 a certain range mark which will be handled by pmtrans() in op.c.
2488 In double-quoted strings:
2490 double-quoted style: \r and \n
2492 deprecated backrefs: \1 (in substitution replacements)
2493 case and quoting: \U \Q \E
2496 scan_const does *not* construct ops to handle interpolated strings.
2497 It stops processing as soon as it finds an embedded $ or @ variable
2498 and leaves it to the caller to work out what's going on.
2500 embedded arrays (whether in pattern or not) could be:
2501 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2503 $ in double-quoted strings must be the symbol of an embedded scalar.
2505 $ in pattern could be $foo or could be tail anchor. Assumption:
2506 it's a tail anchor if $ is the last thing in the string, or if it's
2507 followed by one of "()| \r\n\t"
2509 \1 (backreferences) are turned into $1
2511 The structure of the code is
2512 while (there's a character to process) {
2513 handle transliteration ranges
2514 skip regexp comments /(?#comment)/ and codes /(?{code})/
2515 skip #-initiated comments in //x patterns
2516 check for embedded arrays
2517 check for embedded scalars
2519 leave intact backslashes from leaveit (below)
2520 deprecate \1 in substitution replacements
2521 handle string-changing backslashes \l \U \Q \E, etc.
2522 switch (what was escaped) {
2523 handle \- in a transliteration (becomes a literal -)
2524 handle \132 (octal characters)
2525 handle \x15 and \x{1234} (hex characters)
2526 handle \N{name} (named characters)
2527 handle \cV (control characters)
2528 handle printf-style backslashes (\f, \r, \n, etc)
2531 } (end if backslash)
2532 handle regular character
2533 } (end while character to read)
2538 S_scan_const(pTHX_ char *start)
2541 register char *send = PL_bufend; /* end of the constant */
2542 SV *sv = newSV(send - start); /* sv for the constant. See
2543 note below on sizing. */
2544 register char *s = start; /* start of the constant */
2545 register char *d = SvPVX(sv); /* destination for copies */
2546 bool dorange = FALSE; /* are we in a translit range? */
2547 bool didrange = FALSE; /* did we just finish a range? */
2548 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
2549 I32 this_utf8 = UTF; /* Is the source string assumed
2550 to be UTF8? But, this can
2551 show as true when the source
2552 isn't utf8, as for example
2553 when it is entirely composed
2556 /* Note on sizing: The scanned constant is placed into sv, which is
2557 * initialized by newSV() assuming one byte of output for every byte of
2558 * input. This routine expects newSV() to allocate an extra byte for a
2559 * trailing NUL, which this routine will append if it gets to the end of
2560 * the input. There may be more bytes of input than output (eg., \N{LATIN
2561 * CAPITAL LETTER A}), or more output than input if the constant ends up
2562 * recoded to utf8, but each time a construct is found that might increase
2563 * the needed size, SvGROW() is called. Its size parameter each time is
2564 * based on the best guess estimate at the time, namely the length used so
2565 * far, plus the length the current construct will occupy, plus room for
2566 * the trailing NUL, plus one byte for every input byte still unscanned */
2570 UV literal_endpoint = 0;
2571 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
2574 PERL_ARGS_ASSERT_SCAN_CONST;
2576 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2577 /* If we are doing a trans and we know we want UTF8 set expectation */
2578 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
2579 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2583 while (s < send || dorange) {
2584 /* get transliterations out of the way (they're most literal) */
2585 if (PL_lex_inwhat == OP_TRANS) {
2586 /* expand a range A-Z to the full set of characters. AIE! */
2588 I32 i; /* current expanded character */
2589 I32 min; /* first character in range */
2590 I32 max; /* last character in range */
2601 char * const c = (char*)utf8_hop((U8*)d, -1);
2605 *c = (char)UTF_TO_NATIVE(0xff);
2606 /* mark the range as done, and continue */
2612 i = d - SvPVX_const(sv); /* remember current offset */
2615 SvLEN(sv) + (has_utf8 ?
2616 (512 - UTF_CONTINUATION_MARK +
2619 /* How many two-byte within 0..255: 128 in UTF-8,
2620 * 96 in UTF-8-mod. */
2622 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
2624 d = SvPVX(sv) + i; /* refresh d after realloc */
2628 for (j = 0; j <= 1; j++) {
2629 char * const c = (char*)utf8_hop((U8*)d, -1);
2630 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2636 max = (U8)0xff; /* only to \xff */
2637 uvmax = uv; /* \x{100} to uvmax */
2639 d = c; /* eat endpoint chars */
2644 d -= 2; /* eat the first char and the - */
2645 min = (U8)*d; /* first char in range */
2646 max = (U8)d[1]; /* last char in range */
2653 "Invalid range \"%c-%c\" in transliteration operator",
2654 (char)min, (char)max);
2658 if (literal_endpoint == 2 &&
2659 ((isLOWER(min) && isLOWER(max)) ||
2660 (isUPPER(min) && isUPPER(max)))) {
2662 for (i = min; i <= max; i++)
2664 *d++ = NATIVE_TO_NEED(has_utf8,i);
2666 for (i = min; i <= max; i++)
2668 *d++ = NATIVE_TO_NEED(has_utf8,i);
2673 for (i = min; i <= max; i++)
2676 const U8 ch = (U8)NATIVE_TO_UTF(i);
2677 if (UNI_IS_INVARIANT(ch))
2680 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2681 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2690 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2692 *d++ = (char)UTF_TO_NATIVE(0xff);
2694 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2698 /* mark the range as done, and continue */
2702 literal_endpoint = 0;
2707 /* range begins (ignore - as first or last char) */
2708 else if (*s == '-' && s+1 < send && s != start) {
2710 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2717 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2727 literal_endpoint = 0;
2728 native_range = TRUE;
2733 /* if we get here, we're not doing a transliteration */
2735 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2736 except for the last char, which will be done separately. */
2737 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2739 while (s+1 < send && *s != ')')
2740 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2742 else if (s[2] == '{' /* This should match regcomp.c */
2743 || (s[2] == '?' && s[3] == '{'))
2746 char *regparse = s + (s[2] == '{' ? 3 : 4);
2749 while (count && (c = *regparse)) {
2750 if (c == '\\' && regparse[1])
2758 if (*regparse != ')')
2759 regparse--; /* Leave one char for continuation. */
2760 while (s < regparse)
2761 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2765 /* likewise skip #-initiated comments in //x patterns */
2766 else if (*s == '#' && PL_lex_inpat &&
2767 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2768 while (s+1 < send && *s != '\n')
2769 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2772 /* check for embedded arrays
2773 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2775 else if (*s == '@' && s[1]) {
2776 if (isALNUM_lazy_if(s+1,UTF))
2778 if (strchr(":'{$", s[1]))
2780 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2781 break; /* in regexp, neither @+ nor @- are interpolated */
2784 /* check for embedded scalars. only stop if we're sure it's a
2787 else if (*s == '$') {
2788 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2790 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
2792 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2793 "Possible unintended interpolation of $\\ in regex");
2795 break; /* in regexp, $ might be tail anchor */
2799 /* End of else if chain - OP_TRANS rejoin rest */
2802 if (*s == '\\' && s+1 < send) {
2805 /* deprecate \1 in strings and substitution replacements */
2806 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2807 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2809 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2814 /* string-change backslash escapes */
2815 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2819 /* skip any other backslash escapes in a pattern */
2820 else if (PL_lex_inpat) {
2821 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2822 goto default_action;
2825 /* if we get here, it's either a quoted -, or a digit */
2828 /* quoted - in transliterations */
2830 if (PL_lex_inwhat == OP_TRANS) {
2837 if ((isALPHA(*s) || isDIGIT(*s)))
2838 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
2839 "Unrecognized escape \\%c passed through",
2841 /* default action is to copy the quoted character */
2842 goto default_action;
2845 /* eg. \132 indicates the octal constant 0x132 */
2846 case '0': case '1': case '2': case '3':
2847 case '4': case '5': case '6': case '7':
2851 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
2854 goto NUM_ESCAPE_INSERT;
2856 /* eg. \x24 indicates the hex constant 0x24 */
2860 char* const e = strchr(s, '}');
2861 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2862 PERL_SCAN_DISALLOW_PREFIX;
2867 yyerror("Missing right brace on \\x{}");
2871 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
2877 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2878 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
2884 /* Insert oct, hex, or \N{U+...} escaped character. There will
2885 * always be enough room in sv since such escapes will be
2886 * longer than any UTF-8 sequence they can end up as, except if
2887 * they force us to recode the rest of the string into utf8 */
2889 /* Here uv is the ordinal of the next character being added in
2890 * unicode (converted from native). (It has to be done before
2891 * here because \N is interpreted as unicode, and oct and hex
2893 if (!UNI_IS_INVARIANT(uv)) {
2894 if (!has_utf8 && uv > 255) {
2895 /* Might need to recode whatever we have accumulated so
2896 * far if it contains any chars variant in utf8 or
2899 SvCUR_set(sv, d - SvPVX_const(sv));
2902 /* See Note on sizing above. */
2903 sv_utf8_upgrade_flags_grow(sv,
2904 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2905 UNISKIP(uv) + (STRLEN)(send - s) + 1);
2906 d = SvPVX(sv) + SvCUR(sv);
2911 d = (char*)uvuni_to_utf8((U8*)d, uv);
2912 if (PL_lex_inwhat == OP_TRANS &&
2913 PL_sublex_info.sub_op) {
2914 PL_sublex_info.sub_op->op_private |=
2915 (PL_lex_repl ? OPpTRANS_FROM_UTF
2919 if (uv > 255 && !dorange)
2920 native_range = FALSE;
2932 /* \N{LATIN SMALL LETTER A} is a named character, and so is
2937 char* e = strchr(s, '}');
2943 yyerror("Missing right brace on \\N{}");
2947 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2948 /* \N{U+...} The ... is a unicode value even on EBCDIC
2950 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2951 PERL_SCAN_DISALLOW_PREFIX;
2954 uv = grok_hex(s, &len, &flags, NULL);
2955 if ( e > s && len != (STRLEN)(e - s) ) {
2959 goto NUM_ESCAPE_INSERT;
2961 res = newSVpvn(s + 1, e - s - 1);
2962 res = new_constant( NULL, 0, "charnames",
2963 res, NULL, s - 2, e - s + 3 );
2965 sv_utf8_upgrade(res);
2966 str = SvPV_const(res,len);
2967 #ifdef EBCDIC_NEVER_MIND
2968 /* charnames uses pack U and that has been
2969 * recently changed to do the below uni->native
2970 * mapping, so this would be redundant (and wrong,
2971 * the code point would be doubly converted).
2972 * But leave this in just in case the pack U change
2973 * gets revoked, but the semantics is still
2974 * desireable for charnames. --jhi */
2976 UV uv = utf8_to_uvchr((const U8*)str, 0);
2979 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2981 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2982 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2983 str = SvPV_const(res, len);
2987 /* If destination is not in utf8 but this new character is,
2988 * recode the dest to utf8 */
2989 if (!has_utf8 && SvUTF8(res)) {
2990 SvCUR_set(sv, d - SvPVX_const(sv));
2993 /* See Note on sizing above. */
2994 sv_utf8_upgrade_flags_grow(sv,
2995 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
2996 len + (STRLEN)(send - s) + 1);
2997 d = SvPVX(sv) + SvCUR(sv);
2999 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
3001 /* See Note on sizing above. (NOTE: SvCUR() is not set
3002 * correctly here). */
3003 const STRLEN off = d - SvPVX_const(sv);
3004 d = SvGROW(sv, off + len + (STRLEN)(send - s) + 1) + off;
3008 native_range = FALSE; /* \N{} is guessed to be Unicode */
3010 Copy(str, d, len, char);
3017 yyerror("Missing braces on \\N{}");
3020 /* \c is a control character */
3029 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
3032 yyerror("Missing control char name in \\c");
3036 /* printf-style backslashes, formfeeds, newlines, etc */
3038 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
3041 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
3044 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
3047 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
3050 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
3053 *d++ = ASCII_TO_NEED(has_utf8,'\033');
3056 *d++ = ASCII_TO_NEED(has_utf8,'\007');
3062 } /* end if (backslash) */
3069 /* If we started with encoded form, or already know we want it,
3070 then encode the next character */
3071 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
3075 /* One might think that it is wasted effort in the case of the
3076 * source being utf8 (this_utf8 == TRUE) to take the next character
3077 * in the source, convert it to an unsigned value, and then convert
3078 * it back again. But the source has not been validated here. The
3079 * routine that does the conversion checks for errors like
3082 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
3083 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
3085 SvCUR_set(sv, d - SvPVX_const(sv));
3088 /* See Note on sizing above. */
3089 sv_utf8_upgrade_flags_grow(sv,
3090 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3091 need + (STRLEN)(send - s) + 1);
3092 d = SvPVX(sv) + SvCUR(sv);
3094 } else if (need > len) {
3095 /* encoded value larger than old, may need extra space (NOTE:
3096 * SvCUR() is not set correctly here). See Note on sizing
3098 const STRLEN off = d - SvPVX_const(sv);
3099 d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
3103 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
3105 if (uv > 255 && !dorange)
3106 native_range = FALSE;
3110 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
3112 } /* while loop to process each character */
3114 /* terminate the string and set up the sv */
3116 SvCUR_set(sv, d - SvPVX_const(sv));
3117 if (SvCUR(sv) >= SvLEN(sv))
3118 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
3121 if (PL_encoding && !has_utf8) {
3122 sv_recode_to_utf8(sv, PL_encoding);
3128 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
3129 PL_sublex_info.sub_op->op_private |=
3130 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3134 /* shrink the sv if we allocated more than we used */
3135 if (SvCUR(sv) + 5 < SvLEN(sv)) {
3136 SvPV_shrink_to_cur(sv);
3139 /* return the substring (via pl_yylval) only if we parsed anything */
3140 if (s > PL_bufptr) {
3141 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
3142 const char *const key = PL_lex_inpat ? "qr" : "q";
3143 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
3147 if (PL_lex_inwhat == OP_TRANS) {
3150 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
3158 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
3161 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3168 * Returns TRUE if there's more to the expression (e.g., a subscript),
3171 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
3173 * ->[ and ->{ return TRUE
3174 * { and [ outside a pattern are always subscripts, so return TRUE
3175 * if we're outside a pattern and it's not { or [, then return FALSE
3176 * if we're in a pattern and the first char is a {
3177 * {4,5} (any digits around the comma) returns FALSE
3178 * if we're in a pattern and the first char is a [
3180 * [SOMETHING] has a funky algorithm to decide whether it's a
3181 * character class or not. It has to deal with things like
3182 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
3183 * anything else returns TRUE
3186 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
3189 S_intuit_more(pTHX_ register char *s)
3193 PERL_ARGS_ASSERT_INTUIT_MORE;
3195 if (PL_lex_brackets)
3197 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
3199 if (*s != '{' && *s != '[')
3204 /* In a pattern, so maybe we have {n,m}. */
3221 /* On the other hand, maybe we have a character class */
3224 if (*s == ']' || *s == '^')
3227 /* this is terrifying, and it works */
3228 int weight = 2; /* let's weigh the evidence */
3230 unsigned char un_char = 255, last_un_char;
3231 const char * const send = strchr(s,']');
3232 char tmpbuf[sizeof PL_tokenbuf * 4];
3234 if (!send) /* has to be an expression */
3237 Zero(seen,256,char);
3240 else if (isDIGIT(*s)) {
3242 if (isDIGIT(s[1]) && s[2] == ']')
3248 for (; s < send; s++) {
3249 last_un_char = un_char;
3250 un_char = (unsigned char)*s;
3255 weight -= seen[un_char] * 10;
3256 if (isALNUM_lazy_if(s+1,UTF)) {
3258 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
3259 len = (int)strlen(tmpbuf);
3260 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
3265 else if (*s == '$' && s[1] &&
3266 strchr("[#!%*<>()-=",s[1])) {
3267 if (/*{*/ strchr("])} =",s[2]))
3276 if (strchr("wds]",s[1]))
3278 else if (seen[(U8)'\''] || seen[(U8)'"'])
3280 else if (strchr("rnftbxcav",s[1]))
3282 else if (isDIGIT(s[1])) {
3284 while (s[1] && isDIGIT(s[1]))
3294 if (strchr("aA01! ",last_un_char))
3296 if (strchr("zZ79~",s[1]))
3298 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
3299 weight -= 5; /* cope with negative subscript */
3302 if (!isALNUM(last_un_char)
3303 && !(last_un_char == '$' || last_un_char == '@'
3304 || last_un_char == '&')
3305 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
3310 if (keyword(tmpbuf, d - tmpbuf, 0))
3313 if (un_char == last_un_char + 1)
3315 weight -= seen[un_char];
3320 if (weight >= 0) /* probably a character class */
3330 * Does all the checking to disambiguate
3332 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
3333 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
3335 * First argument is the stuff after the first token, e.g. "bar".
3337 * Not a method if bar is a filehandle.
3338 * Not a method if foo is a subroutine prototyped to take a filehandle.
3339 * Not a method if it's really "Foo $bar"
3340 * Method if it's "foo $bar"
3341 * Not a method if it's really "print foo $bar"
3342 * Method if it's really "foo package::" (interpreted as package->foo)
3343 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
3344 * Not a method if bar is a filehandle or package, but is quoted with
3349 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
3352 char *s = start + (*start == '$');
3353 char tmpbuf[sizeof PL_tokenbuf];
3360 PERL_ARGS_ASSERT_INTUIT_METHOD;
3363 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
3367 const char *proto = SvPVX_const(cv);
3378 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
3379 /* start is the beginning of the possible filehandle/object,
3380 * and s is the end of it
3381 * tmpbuf is a copy of it
3384 if (*start == '$') {
3385 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
3386 isUPPER(*PL_tokenbuf))
3389 len = start - SvPVX(PL_linestr);
3393 start = SvPVX(PL_linestr) + len;
3397 return *s == '(' ? FUNCMETH : METHOD;
3399 if (!keyword(tmpbuf, len, 0)) {
3400 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
3404 soff = s - SvPVX(PL_linestr);
3408 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
3409 if (indirgv && GvCVu(indirgv))
3411 /* filehandle or package name makes it a method */
3412 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
3414 soff = s - SvPVX(PL_linestr);
3417 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
3418 return 0; /* no assumptions -- "=>" quotes bearword */
3420 start_force(PL_curforce);
3421 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
3422 S_newSV_maybe_utf8(aTHX_ tmpbuf, len));
3423 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
3425 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
3430 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
3432 return *s == '(' ? FUNCMETH : METHOD;
3438 /* Encoded script support. filter_add() effectively inserts a
3439 * 'pre-processing' function into the current source input stream.
3440 * Note that the filter function only applies to the current source file
3441 * (e.g., it will not affect files 'require'd or 'use'd by this one).
3443 * The datasv parameter (which may be NULL) can be used to pass
3444 * private data to this instance of the filter. The filter function
3445 * can recover the SV using the FILTER_DATA macro and use it to
3446 * store private buffers and state information.
3448 * The supplied datasv parameter is upgraded to a PVIO type
3449 * and the IoDIRP/IoANY field is used to store the function pointer,
3450 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
3451 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
3452 * private use must be set using malloc'd pointers.
3456 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
3465 if (!PL_rsfp_filters)
3466 PL_rsfp_filters = newAV();
3469 SvUPGRADE(datasv, SVt_PVIO);
3470 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
3471 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
3472 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
3473 FPTR2DPTR(void *, IoANY(datasv)),
3474 SvPV_nolen(datasv)));
3475 av_unshift(PL_rsfp_filters, 1);
3476 av_store(PL_rsfp_filters, 0, datasv) ;
3481 /* Delete most recently added instance of this filter function. */
3483 Perl_filter_del(pTHX_ filter_t funcp)
3488 PERL_ARGS_ASSERT_FILTER_DEL;
3491 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
3492 FPTR2DPTR(void*, funcp)));
3494 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
3496 /* if filter is on top of stack (usual case) just pop it off */
3497 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
3498 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
3499 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
3500 IoANY(datasv) = (void *)NULL;
3501 sv_free(av_pop(PL_rsfp_filters));
3505 /* we need to search for the correct entry and clear it */
3506 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
3510 /* Invoke the idxth filter function for the current rsfp. */
3511 /* maxlen 0 = read one text line */
3513 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
3518 /* This API is bad. It should have been using unsigned int for maxlen.
3519 Not sure if we want to change the API, but if not we should sanity
3520 check the value here. */
3521 const unsigned int correct_length
3530 PERL_ARGS_ASSERT_FILTER_READ;
3532 if (!PL_parser || !PL_rsfp_filters)
3534 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
3535 /* Provide a default input filter to make life easy. */
3536 /* Note that we append to the line. This is handy. */
3537 DEBUG_P(PerlIO_printf(Perl_debug_log,
3538 "filter_read %d: from rsfp\n", idx));
3539 if (correct_length) {
3542 const int old_len = SvCUR(buf_sv);
3544 /* ensure buf_sv is large enough */
3545 SvGROW(buf_sv, (STRLEN)(old_len + correct_length + 1)) ;
3546 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
3547 correct_length)) <= 0) {
3548 if (PerlIO_error(PL_rsfp))
3549 return -1; /* error */
3551 return 0 ; /* end of file */
3553 SvCUR_set(buf_sv, old_len + len) ;
3554 SvPVX(buf_sv)[old_len + len] = '\0';
3557 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
3558 if (PerlIO_error(PL_rsfp))
3559 return -1; /* error */
3561 return 0 ; /* end of file */
3564 return SvCUR(buf_sv);
3566 /* Skip this filter slot if filter has been deleted */
3567 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
3568 DEBUG_P(PerlIO_printf(Perl_debug_log,
3569 "filter_read %d: skipped (filter deleted)\n",
3571 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
3573 /* Get function pointer hidden within datasv */
3574 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
3575 DEBUG_P(PerlIO_printf(Perl_debug_log,
3576 "filter_read %d: via function %p (%s)\n",
3577 idx, (void*)datasv, SvPV_nolen_const(datasv)));
3578 /* Call function. The function is expected to */
3579 /* call "FILTER_READ(idx+1, buf_sv)" first. */
3580 /* Return: <0:error, =0:eof, >0:not eof */
3581 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
3585 S_filter_gets(pTHX_ register SV *sv, STRLEN append)
3589 PERL_ARGS_ASSERT_FILTER_GETS;
3591 #ifdef PERL_CR_FILTER
3592 if (!PL_rsfp_filters) {
3593 filter_add(S_cr_textfilter,NULL);
3596 if (PL_rsfp_filters) {
3598 SvCUR_set(sv, 0); /* start with empty line */
3599 if (FILTER_READ(0, sv, 0) > 0)
3600 return ( SvPVX(sv) ) ;
3605 return (sv_gets(sv, PL_rsfp, append));
3609 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
3614 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
3616 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
3620 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
3621 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3623 return GvHV(gv); /* Foo:: */
3626 /* use constant CLASS => 'MyClass' */
3627 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3628 if (gv && GvCV(gv)) {
3629 SV * const sv = cv_const_sv(GvCV(gv));
3631 pkgname = SvPV_const(sv, len);
3634 return gv_stashpvn(pkgname, len, 0);
3638 * S_readpipe_override
3639 * Check whether readpipe() is overriden, and generates the appropriate
3640 * optree, provided sublex_start() is called afterwards.
3643 S_readpipe_override(pTHX)
3646 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3647 pl_yylval.ival = OP_BACKTICK;
3649 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3651 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3652 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3653 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3655 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3656 append_elem(OP_LIST,
3657 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3658 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3665 * The intent of this yylex wrapper is to minimize the changes to the
3666 * tokener when we aren't interested in collecting madprops. It remains
3667 * to be seen how successful this strategy will be...
3674 char *s = PL_bufptr;
3676 /* make sure PL_thiswhite is initialized */
3680 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3681 if (PL_pending_ident)
3682 return S_pending_ident(aTHX);
3684 /* previous token ate up our whitespace? */
3685 if (!PL_lasttoke && PL_nextwhite) {
3686 PL_thiswhite = PL_nextwhite;
3690 /* isolate the token, and figure out where it is without whitespace */
3691 PL_realtokenstart = -1;
3695 assert(PL_curforce < 0);
3697 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3698 if (!PL_thistoken) {
3699 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3700 PL_thistoken = newSVpvs("");
3702 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3703 PL_thistoken = newSVpvn(tstart, s - tstart);
3706 if (PL_thismad) /* install head */
3707 CURMAD('X', PL_thistoken);
3710 /* last whitespace of a sublex? */
3711 if (optype == ')' && PL_endwhite) {
3712 CURMAD('X', PL_endwhite);
3717 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3718 if (!PL_thiswhite && !PL_endwhite && !optype) {
3719 sv_free(PL_thistoken);
3724 /* put off final whitespace till peg */
3725 if (optype == ';' && !PL_rsfp) {
3726 PL_nextwhite = PL_thiswhite;
3729 else if (PL_thisopen) {
3730 CURMAD('q', PL_thisopen);
3732 sv_free(PL_thistoken);
3736 /* Store actual token text as madprop X */
3737 CURMAD('X', PL_thistoken);
3741 /* add preceding whitespace as madprop _ */
3742 CURMAD('_', PL_thiswhite);
3746 /* add quoted material as madprop = */
3747 CURMAD('=', PL_thisstuff);
3751 /* add terminating quote as madprop Q */
3752 CURMAD('Q', PL_thisclose);
3756 /* special processing based on optype */
3760 /* opval doesn't need a TOKEN since it can already store mp */
3770 if (pl_yylval.opval)
3771 append_madprops(PL_thismad, pl_yylval.opval, 0);
3779 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3788 /* remember any fake bracket that lexer is about to discard */
3789 if (PL_lex_brackets == 1 &&
3790 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3793 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3796 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3797 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3800 break; /* don't bother looking for trailing comment */
3809 /* attach a trailing comment to its statement instead of next token */
3813 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3815 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3817 if (*s == '\n' || *s == '#') {
3818 while (s < PL_bufend && *s != '\n')
3822 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3823 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3840 /* Create new token struct. Note: opvals return early above. */
3841 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3848 S_tokenize_use(pTHX_ int is_use, char *s) {
3851 PERL_ARGS_ASSERT_TOKENIZE_USE;
3853 if (PL_expect != XSTATE)
3854 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3855 is_use ? "use" : "no"));
3857 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3858 s = force_version(s, TRUE);
3859 if (*s == ';' || *s == '}'
3860 || (s = SKIPSPACE1(s), (*s == ';' || *s == '}'))) {
3861 start_force(PL_curforce);
3862 NEXTVAL_NEXTTOKE.opval = NULL;
3865 else if (*s == 'v') {
3866 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3867 s = force_version(s, FALSE);
3871 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3872 s = force_version(s, FALSE);
3874 pl_yylval.ival = is_use;
3878 static const char* const exp_name[] =
3879 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3880 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3887 Works out what to call the token just pulled out of the input
3888 stream. The yacc parser takes care of taking the ops we return and
3889 stitching them into a tree.
3895 if read an identifier
3896 if we're in a my declaration
3897 croak if they tried to say my($foo::bar)
3898 build the ops for a my() declaration
3899 if it's an access to a my() variable
3900 are we in a sort block?
3901 croak if my($a); $a <=> $b
3902 build ops for access to a my() variable
3903 if in a dq string, and they've said @foo and we can't find @foo
3905 build ops for a bareword
3906 if we already built the token before, use it.
3911 #pragma segment Perl_yylex
3917 register char *s = PL_bufptr;
3923 /* orig_keyword, gvp, and gv are initialized here because
3924 * jump to the label just_a_word_zero can bypass their
3925 * initialization later. */
3926 I32 orig_keyword = 0;
3931 SV* tmp = newSVpvs("");
3932 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3933 (IV)CopLINE(PL_curcop),
3934 lex_state_names[PL_lex_state],
3935 exp_name[PL_expect],
3936 pv_display(tmp, s, strlen(s), 0, 60));
3939 /* check if there's an identifier for us to look at */
3940 if (PL_pending_ident)
3941 return REPORT(S_pending_ident(aTHX));
3943 /* no identifier pending identification */
3945 switch (PL_lex_state) {
3947 case LEX_NORMAL: /* Some compilers will produce faster */
3948 case LEX_INTERPNORMAL: /* code if we comment these out. */
3952 /* when we've already built the next token, just pull it out of the queue */
3956 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3958 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3959 PL_nexttoke[PL_lasttoke].next_mad = 0;
3960 if (PL_thismad && PL_thismad->mad_key == '_') {
3961 PL_thiswhite = MUTABLE_SV(PL_thismad->mad_val);
3962 PL_thismad->mad_val = 0;
3963 mad_free(PL_thismad);
3968 PL_lex_state = PL_lex_defer;
3969 PL_expect = PL_lex_expect;
3970 PL_lex_defer = LEX_NORMAL;
3971 if (!PL_nexttoke[PL_lasttoke].next_type)
3976 pl_yylval = PL_nextval[PL_nexttoke];
3978 PL_lex_state = PL_lex_defer;
3979 PL_expect = PL_lex_expect;
3980 PL_lex_defer = LEX_NORMAL;
3984 /* FIXME - can these be merged? */
3985 return(PL_nexttoke[PL_lasttoke].next_type);
3987 return REPORT(PL_nexttype[PL_nexttoke]);
3990 /* interpolated case modifiers like \L \U, including \Q and \E.
3991 when we get here, PL_bufptr is at the \
3993 case LEX_INTERPCASEMOD:
3995 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3996 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3998 /* handle \E or end of string */
3999 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
4001 if (PL_lex_casemods) {
4002 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
4003 PL_lex_casestack[PL_lex_casemods] = '\0';
4005 if (PL_bufptr != PL_bufend
4006 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
4008 PL_lex_state = LEX_INTERPCONCAT;
4011 PL_thistoken = newSVpvs("\\E");
4017 while (PL_bufptr != PL_bufend &&
4018 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
4020 PL_thiswhite = newSVpvs("");
4021 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
4025 if (PL_bufptr != PL_bufend)
4028 PL_lex_state = LEX_INTERPCONCAT;
4032 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4033 "### Saw case modifier\n"); });
4035 if (s[1] == '\\' && s[2] == 'E') {
4038 PL_thiswhite = newSVpvs("");
4039 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
4042 PL_lex_state = LEX_INTERPCONCAT;
4047 if (!PL_madskills) /* when just compiling don't need correct */
4048 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
4049 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
4050 if ((*s == 'L' || *s == 'U') &&
4051 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
4052 PL_lex_casestack[--PL_lex_casemods] = '\0';
4055 if (PL_lex_casemods > 10)
4056 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
4057 PL_lex_casestack[PL_lex_casemods++] = *s;
4058 PL_lex_casestack[PL_lex_casemods] = '\0';
4059 PL_lex_state = LEX_INTERPCONCAT;
4060 start_force(PL_curforce);
4061 NEXTVAL_NEXTTOKE.ival = 0;
4063 start_force(PL_curforce);
4065 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
4067 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
4069 NEXTVAL_NEXTTOKE.ival = OP_LC;
4071 NEXTVAL_NEXTTOKE.ival = OP_UC;
4073 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
4075 Perl_croak(aTHX_ "panic: yylex");
4077 SV* const tmpsv = newSVpvs("\\ ");
4078 /* replace the space with the character we want to escape
4080 SvPVX(tmpsv)[1] = *s;
4086 if (PL_lex_starts) {
4092 sv_free(PL_thistoken);
4093 PL_thistoken = newSVpvs("");
4096 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4097 if (PL_lex_casemods == 1 && PL_lex_inpat)
4106 case LEX_INTERPPUSH:
4107 return REPORT(sublex_push());
4109 case LEX_INTERPSTART:
4110 if (PL_bufptr == PL_bufend)
4111 return REPORT(sublex_done());
4112 DEBUG_T({ PerlIO_printf(Perl_debug_log,
4113 "### Interpolated variable\n"); });
4115 PL_lex_dojoin = (*PL_bufptr == '@');
4116 PL_lex_state = LEX_INTERPNORMAL;
4117 if (PL_lex_dojoin) {
4118 start_force(PL_curforce);
4119 NEXTVAL_NEXTTOKE.ival = 0;
4121 start_force(PL_curforce);
4122 force_ident("\"", '$');
4123 start_force(PL_curforce);
4124 NEXTVAL_NEXTTOKE.ival = 0;
4126 start_force(PL_curforce);
4127 NEXTVAL_NEXTTOKE.ival = 0;
4129 start_force(PL_curforce);
4130 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
4133 if (PL_lex_starts++) {
4138 sv_free(PL_thistoken);
4139 PL_thistoken = newSVpvs("");
4142 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4143 if (!PL_lex_casemods && PL_lex_inpat)
4150 case LEX_INTERPENDMAYBE:
4151 if (intuit_more(PL_bufptr)) {
4152 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
4158 if (PL_lex_dojoin) {
4159 PL_lex_dojoin = FALSE;
4160 PL_lex_state = LEX_INTERPCONCAT;
4164 sv_free(PL_thistoken);
4165 PL_thistoken = newSVpvs("");
4170 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
4171 && SvEVALED(PL_lex_repl))
4173 if (PL_bufptr != PL_bufend)
4174 Perl_croak(aTHX_ "Bad evalled substitution pattern");
4178 case LEX_INTERPCONCAT:
4180 if (PL_lex_brackets)
4181 Perl_croak(aTHX_ "panic: INTERPCONCAT");
4183 if (PL_bufptr == PL_bufend)
4184 return REPORT(sublex_done());
4186 if (SvIVX(PL_linestr) == '\'') {
4187 SV *sv = newSVsv(PL_linestr);
4190 else if ( PL_hints & HINT_NEW_RE )
4191 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
4192 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
4196 s = scan_const(PL_bufptr);
4198 PL_lex_state = LEX_INTERPCASEMOD;
4200 PL_lex_state = LEX_INTERPSTART;
4203 if (s != PL_bufptr) {
4204 start_force(PL_curforce);
4206 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
4208 NEXTVAL_NEXTTOKE = pl_yylval;
4211 if (PL_lex_starts++) {
4215 sv_free(PL_thistoken);
4216 PL_thistoken = newSVpvs("");
4219 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
4220 if (!PL_lex_casemods && PL_lex_inpat)
4233 PL_lex_state = LEX_NORMAL;
4234 s = scan_formline(PL_bufptr);
4235 if (!PL_lex_formbrack)
4241 PL_oldoldbufptr = PL_oldbufptr;
4247 sv_free(PL_thistoken);
4250 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
4254 if (isIDFIRST_lazy_if(s,UTF))
4257 unsigned char c = *s;
4258 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
4259 if (len > UNRECOGNIZED_PRECEDE_COUNT) {
4260 d = UTF ? (char *) Perl_utf8_hop(aTHX_ (U8 *) s, -UNRECOGNIZED_PRECEDE_COUNT) : s - UNRECOGNIZED_PRECEDE_COUNT;
4265 Perl_croak(aTHX_ "Unrecognized character \\x%02X; marked by <-- HERE after %s<-- HERE near column %d", c, d, (int) len + 1);
4269 goto fake_eof; /* emulate EOF on ^D or ^Z */
4278 if (PL_lex_brackets) {
4279 yyerror((const char *)
4281 ? "Format not terminated"
4282 : "Missing right curly or square bracket"));
4284 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4285 "### Tokener got EOF\n");
4289 if (s++ < PL_bufend)
4290 goto retry; /* ignore stray nulls */
4293 if (!PL_in_eval && !PL_preambled) {
4294 PL_preambled = TRUE;
4300 /* Generate a string of Perl code to load the debugger.
4301 * If PERL5DB is set, it will return the contents of that,
4302 * otherwise a compile-time require of perl5db.pl. */
4304 const char * const pdb = PerlEnv_getenv("PERL5DB");
4307 sv_setpv(PL_linestr, pdb);
4308 sv_catpvs(PL_linestr,";");
4310 SETERRNO(0,SS_NORMAL);
4311 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
4314 sv_setpvs(PL_linestr,"");
4315 if (PL_preambleav) {
4316 SV **svp = AvARRAY(PL_preambleav);
4317 SV **const end = svp + AvFILLp(PL_preambleav);
4319 sv_catsv(PL_linestr, *svp);
4321 sv_catpvs(PL_linestr, ";");
4323 sv_free(MUTABLE_SV(PL_preambleav));
4324 PL_preambleav = NULL;
4327 sv_catpvs(PL_linestr,
4328 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
4329 if (PL_minus_n || PL_minus_p) {
4330 sv_catpvs(PL_linestr, "LINE: while (<>) {"/*}*/);
4332 sv_catpvs(PL_linestr,"chomp;");
4335 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
4336 || *PL_splitstr == '"')
4337 && strchr(PL_splitstr + 1, *PL_splitstr))
4338 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
4340 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
4341 bytes can be used as quoting characters. :-) */
4342 const char *splits = PL_splitstr;
4343 sv_catpvs(PL_linestr, "our @F=split(q\0");
4346 if (*splits == '\\')
4347 sv_catpvn(PL_linestr, splits, 1);
4348 sv_catpvn(PL_linestr, splits, 1);
4349 } while (*splits++);
4350 /* This loop will embed the trailing NUL of
4351 PL_linestr as the last thing it does before
4353 sv_catpvs(PL_linestr, ");");
4357 sv_catpvs(PL_linestr,"our @F=split(' ');");
4360 sv_catpvs(PL_linestr, "\n");
4361 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4362 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4363 PL_last_lop = PL_last_uni = NULL;
4364 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
4365 update_debugger_info(PL_linestr, NULL, 0);
4370 bof = PL_rsfp ? TRUE : FALSE;
4373 fake_eof = LEX_FAKE_EOF;
4375 PL_bufptr = PL_bufend;
4376 CopLINE_inc(PL_curcop);
4377 if (!lex_next_chunk(fake_eof)) {
4378 CopLINE_dec(PL_curcop);
4380 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
4382 CopLINE_dec(PL_curcop);
4385 PL_realtokenstart = -1;
4388 /* If it looks like the start of a BOM or raw UTF-16,
4389 * check if it in fact is. */
4390 if (bof && PL_rsfp &&
4395 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
4397 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4398 s = swallow_bom((U8*)s);
4402 /* Incest with pod. */
4405 sv_catsv(PL_thiswhite, PL_linestr);
4407 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
4408 sv_setpvs(PL_linestr, "");
4409 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4410 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4411 PL_last_lop = PL_last_uni = NULL;
4412 PL_doextract = FALSE;
4417 } while (PL_doextract);
4418 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
4419 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4420 PL_last_lop = PL_last_uni = NULL;
4421 if (CopLINE(PL_curcop) == 1) {
4422 while (s < PL_bufend && isSPACE(*s))
4424 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
4428 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
4432 if (*s == '#' && *(s+1) == '!')
4434 #ifdef ALTERNATE_SHEBANG
4436 static char const as[] = ALTERNATE_SHEBANG;
4437 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
4438 d = s + (sizeof(as) - 1);
4440 #endif /* ALTERNATE_SHEBANG */
4449 while (*d && !isSPACE(*d))
4453 #ifdef ARG_ZERO_IS_SCRIPT
4454 if (ipathend > ipath) {
4456 * HP-UX (at least) sets argv[0] to the script name,
4457 * which makes $^X incorrect. And Digital UNIX and Linux,
4458 * at least, set argv[0] to the basename of the Perl
4459 * interpreter. So, having found "#!", we'll set it right.
4461 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
4463 assert(SvPOK(x) || SvGMAGICAL(x));
4464 if (sv_eq(x, CopFILESV(PL_curcop))) {
4465 sv_setpvn(x, ipath, ipathend - ipath);
4471 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
4472 const char * const lstart = SvPV_const(x,llen);
4474 bstart += blen - llen;
4475 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
4476 sv_setpvn(x, ipath, ipathend - ipath);
4481 TAINT_NOT; /* $^X is always tainted, but that's OK */
4483 #endif /* ARG_ZERO_IS_SCRIPT */
4488 d = instr(s,"perl -");
4490 d = instr(s,"perl");
4492 /* avoid getting into infinite loops when shebang
4493 * line contains "Perl" rather than "perl" */
4495 for (d = ipathend-4; d >= ipath; --d) {
4496 if ((*d == 'p' || *d == 'P')
4497 && !ibcmp(d, "perl", 4))
4507 #ifdef ALTERNATE_SHEBANG
4509 * If the ALTERNATE_SHEBANG on this system starts with a
4510 * character that can be part of a Perl expression, then if
4511 * we see it but not "perl", we're probably looking at the
4512 * start of Perl code, not a request to hand off to some
4513 * other interpreter. Similarly, if "perl" is there, but
4514 * not in the first 'word' of the line, we assume the line
4515 * contains the start of the Perl program.
4517 if (d && *s != '#') {
4518 const char *c = ipath;
4519 while (*c && !strchr("; \t\r\n\f\v#", *c))
4522 d = NULL; /* "perl" not in first word; ignore */
4524 *s = '#'; /* Don't try to parse shebang line */
4526 #endif /* ALTERNATE_SHEBANG */
4531 !instr(s,"indir") &&
4532 instr(PL_origargv[0],"perl"))
4539 while (s < PL_bufend && isSPACE(*s))
4541 if (s < PL_bufend) {
4542 Newx(newargv,PL_origargc+3,char*);
4544 while (s < PL_bufend && !isSPACE(*s))
4547 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
4550 newargv = PL_origargv;
4553 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
4555 Perl_croak(aTHX_ "Can't exec %s", ipath);
4558 while (*d && !isSPACE(*d))
4560 while (SPACE_OR_TAB(*d))
4564 const bool switches_done = PL_doswitches;
4565 const U32 oldpdb = PL_perldb;
4566 const bool oldn = PL_minus_n;
4567 const bool oldp = PL_minus_p;
4571 bool baduni = FALSE;
4573 const char *d2 = d1 + 1;
4574 if (parse_unicode_opts((const char **)&d2)
4578 if (baduni || *d1 == 'M' || *d1 == 'm') {
4579 const char * const m = d1;
4580 while (*d1 && !isSPACE(*d1))
4582 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
4585 d1 = moreswitches(d1);
4587 if (PL_doswitches && !switches_done) {
4588 int argc = PL_origargc;
4589 char **argv = PL_origargv;
4592 } while (argc && argv[0][0] == '-' && argv[0][1]);
4593 init_argv_symbols(argc,argv);
4595 if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
4596 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
4597 /* if we have already added "LINE: while (<>) {",
4598 we must not do it again */
4600 sv_setpvs(PL_linestr, "");
4601 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
4602 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
4603 PL_last_lop = PL_last_uni = NULL;
4604 PL_preambled = FALSE;
4605 if (PERLDB_LINE || PERLDB_SAVESRC)
4606 (void)gv_fetchfile(PL_origfilename);
4613 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4615 PL_lex_state = LEX_FORMLINE;
4620 #ifdef PERL_STRICT_CR
4621 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4623 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4625 case ' ': case '\t': case '\f': case 013:
4627 PL_realtokenstart = -1;
4629 PL_thiswhite = newSVpvs("");
4630 sv_catpvn(PL_thiswhite, s, 1);
4637 PL_realtokenstart = -1;
4641 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4642 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4643 /* handle eval qq[#line 1 "foo"\n ...] */
4644 CopLINE_dec(PL_curcop);
4647 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4649 if (!PL_in_eval || PL_rsfp)
4654 while (d < PL_bufend && *d != '\n')
4658 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4659 Perl_croak(aTHX_ "panic: input overflow");
4662 PL_thiswhite = newSVpvn(s, d - s);
4667 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4669 PL_lex_state = LEX_FORMLINE;
4675 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4676 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4679 TOKEN(PEG); /* make sure any #! line is accessible */
4684 /* if (PL_madskills && PL_lex_formbrack) { */
4686 while (d < PL_bufend && *d != '\n')
4690 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4691 Perl_croak(aTHX_ "panic: input overflow");
4692 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4694 PL_thiswhite = newSVpvs("");
4695 if (CopLINE(PL_curcop) == 1) {
4696 sv_setpvs(PL_thiswhite, "");
4699 sv_catpvn(PL_thiswhite, s, d - s);
4713 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4721 while (s < PL_bufend && SPACE_OR_TAB(*s))
4724 if (strnEQ(s,"=>",2)) {
4725 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4726 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4727 OPERATOR('-'); /* unary minus */
4729 PL_last_uni = PL_oldbufptr;
4731 case 'r': ftst = OP_FTEREAD; break;
4732 case 'w': ftst = OP_FTEWRITE; break;
4733 case 'x': ftst = OP_FTEEXEC; break;
4734 case 'o': ftst = OP_FTEOWNED; break;
4735 case 'R': ftst = OP_FTRREAD; break;
4736 case 'W': ftst = OP_FTRWRITE; break;
4737 case 'X': ftst = OP_FTREXEC; break;
4738 case 'O': ftst = OP_FTROWNED; break;
4739 case 'e': ftst = OP_FTIS; break;
4740 case 'z': ftst = OP_FTZERO; break;
4741 case 's': ftst = OP_FTSIZE; break;
4742 case 'f': ftst = OP_FTFILE; break;
4743 case 'd': ftst = OP_FTDIR; break;
4744 case 'l': ftst = OP_FTLINK; break;
4745 case 'p': ftst = OP_FTPIPE; break;
4746 case 'S': ftst = OP_FTSOCK; break;
4747 case 'u': ftst = OP_FTSUID; break;
4748 case 'g': ftst = OP_FTSGID; break;
4749 case 'k': ftst = OP_FTSVTX; break;
4750 case 'b': ftst = OP_FTBLK; break;
4751 case 'c': ftst = OP_FTCHR; break;
4752 case 't': ftst = OP_FTTTY; break;
4753 case 'T': ftst = OP_FTTEXT; break;
4754 case 'B': ftst = OP_FTBINARY; break;
4755 case 'M': case 'A': case 'C':
4756 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4758 case 'M': ftst = OP_FTMTIME; break;
4759 case 'A': ftst = OP_FTATIME; break;
4760 case 'C': ftst = OP_FTCTIME; break;
4768 PL_last_lop_op = (OPCODE)ftst;
4769 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4770 "### Saw file test %c\n", (int)tmp);
4775 /* Assume it was a minus followed by a one-letter named
4776 * subroutine call (or a -bareword), then. */
4777 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4778 "### '-%c' looked like a file test but was not\n",
4785 const char tmp = *s++;
4788 if (PL_expect == XOPERATOR)
4793 else if (*s == '>') {
4796 if (isIDFIRST_lazy_if(s,UTF)) {
4797 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4805 if (PL_expect == XOPERATOR)
4808 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4810 OPERATOR('-'); /* unary minus */
4816 const char tmp = *s++;
4819 if (PL_expect == XOPERATOR)
4824 if (PL_expect == XOPERATOR)
4827 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4834 if (PL_expect != XOPERATOR) {
4835 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4836 PL_expect = XOPERATOR;
4837 force_ident(PL_tokenbuf, '*');
4850 if (PL_expect == XOPERATOR) {
4854 PL_tokenbuf[0] = '%';
4855 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4856 sizeof PL_tokenbuf - 1, FALSE);
4857 if (!PL_tokenbuf[1]) {
4860 PL_pending_ident = '%';
4869 const char tmp = *s++;
4874 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4881 const char tmp = *s++;
4887 goto just_a_word_zero_gv;
4890 switch (PL_expect) {
4896 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4898 PL_bufptr = s; /* update in case we back off */
4900 deprecate(":= for an empty attribute list");
4907 PL_expect = XTERMBLOCK;
4910 stuffstart = s - SvPVX(PL_linestr) - 1;
4914 while (isIDFIRST_lazy_if(s,UTF)) {
4917 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4918 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4919 if (tmp < 0) tmp = -tmp;
4934 sv = newSVpvn(s, len);
4936 d = scan_str(d,TRUE,TRUE);
4938 /* MUST advance bufptr here to avoid bogus
4939 "at end of line" context messages from yyerror().
4941 PL_bufptr = s + len;
4942 yyerror("Unterminated attribute parameter in attribute list");
4946 return REPORT(0); /* EOF indicator */
4950 sv_catsv(sv, PL_lex_stuff);
4951 attrs = append_elem(OP_LIST, attrs,
4952 newSVOP(OP_CONST, 0, sv));
4953 SvREFCNT_dec(PL_lex_stuff);
4954 PL_lex_stuff = NULL;
4957 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4959 if (PL_in_my == KEY_our) {
4960 deprecate(":unique");
4963 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4966 /* NOTE: any CV attrs applied here need to be part of
4967 the CVf_BUILTIN_ATTRS define in cv.h! */
4968 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4970 CvLVALUE_on(PL_compcv);
4972 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4974 deprecate(":locked");
4976 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4978 CvMETHOD_on(PL_compcv);
4980 /* After we've set the flags, it could be argued that
4981 we don't need to do the attributes.pm-based setting
4982 process, and shouldn't bother appending recognized
4983 flags. To experiment with that, uncomment the
4984 following "else". (Note that's already been
4985 uncommented. That keeps the above-applied built-in
4986 attributes from being intercepted (and possibly
4987 rejected) by a package's attribute routines, but is
4988 justified by the performance win for the common case
4989 of applying only built-in attributes.) */
4991 attrs = append_elem(OP_LIST, attrs,
4992 newSVOP(OP_CONST, 0,
4996 if (*s == ':' && s[1] != ':')
4999 break; /* require real whitespace or :'s */
5000 /* XXX losing whitespace on sequential attributes here */
5004 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
5005 if (*s != ';' && *s != '}' && *s != tmp
5006 && (tmp != '=' || *s != ')')) {
5007 const char q = ((*s == '\'') ? '"' : '\'');
5008 /* If here for an expression, and parsed no attrs, back
5010 if (tmp == '=' && !attrs) {
5014 /* MUST advance bufptr here to avoid bogus "at end of line"
5015 context messages from yyerror().
5018 yyerror( (const char *)
5020 ? Perl_form(aTHX_ "Invalid separator character "
5021 "%c%c%c in attribute list", q, *s, q)
5022 : "Unterminated attribute list" ) );
5030 start_force(PL_curforce);
5031 NEXTVAL_NEXTTOKE.opval = attrs;
5032 CURMAD('_', PL_nextwhite);
5037 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
5038 (s - SvPVX(PL_linestr)) - stuffstart);
5046 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
5047 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
5055 const char tmp = *s++;
5060 const char tmp = *s++;
5068 if (PL_lex_brackets <= 0)
5069 yyerror("Unmatched right square bracket");
5072 if (PL_lex_state == LEX_INTERPNORMAL) {
5073 if (PL_lex_brackets == 0) {
5074 if (*s == '-' && s[1] == '>')
5075 PL_lex_state = LEX_INTERPENDMAYBE;
5076 else if (*s != '[' && *s != '{')
5077 PL_lex_state = LEX_INTERPEND;
5084 if (PL_lex_brackets > 100) {
5085 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
5087 switch (PL_expect) {
5089 if (PL_lex_formbrack) {
5093 if (PL_oldoldbufptr == PL_last_lop)
5094 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
5096 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5097 OPERATOR(HASHBRACK);
5099 while (s < PL_bufend && SPACE_OR_TAB(*s))
5102 PL_tokenbuf[0] = '\0';
5103 if (d < PL_bufend && *d == '-') {
5104 PL_tokenbuf[0] = '-';
5106 while (d < PL_bufend && SPACE_OR_TAB(*d))
5109 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
5110 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
5112 while (d < PL_bufend && SPACE_OR_TAB(*d))
5115 const char minus = (PL_tokenbuf[0] == '-');
5116 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
5124 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
5129 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5134 if (PL_oldoldbufptr == PL_last_lop)
5135 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
5137 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
5140 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
5142 /* This hack is to get the ${} in the message. */
5144 yyerror("syntax error");
5147 OPERATOR(HASHBRACK);
5149 /* This hack serves to disambiguate a pair of curlies
5150 * as being a block or an anon hash. Normally, expectation
5151 * determines that, but in cases where we're not in a
5152 * position to expect anything in particular (like inside
5153 * eval"") we have to resolve the ambiguity. This code
5154 * covers the case where the first term in the curlies is a
5155 * quoted string. Most other cases need to be explicitly
5156 * disambiguated by prepending a "+" before the opening
5157 * curly in order to force resolution as an anon hash.
5159 * XXX should probably propagate the outer expectation
5160 * into eval"" to rely less on this hack, but that could
5161 * potentially break current behavior of eval"".
5165 if (*s == '\'' || *s == '"' || *s == '`') {
5166 /* common case: get past first string, handling escapes */
5167 for (t++; t < PL_bufend && *t != *s;)
5168 if (*t++ == '\\' && (*t == '\\' || *t == *s))
5172 else if (*s == 'q') {
5175 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
5178 /* skip q//-like construct */
5180 char open, close, term;
5183 while (t < PL_bufend && isSPACE(*t))
5185 /* check for q => */
5186 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
5187 OPERATOR(HASHBRACK);
5191 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
5195 for (t++; t < PL_bufend; t++) {
5196 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
5198 else if (*t == open)
5202 for (t++; t < PL_bufend; t++) {
5203 if (*t == '\\' && t+1 < PL_bufend)
5205 else if (*t == close && --brackets <= 0)
5207 else if (*t == open)
5214 /* skip plain q word */
5215 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
5218 else if (isALNUM_lazy_if(t,UTF)) {
5220 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
5223 while (t < PL_bufend && isSPACE(*t))
5225 /* if comma follows first term, call it an anon hash */
5226 /* XXX it could be a comma expression with loop modifiers */
5227 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
5228 || (*t == '=' && t[1] == '>')))
5229 OPERATOR(HASHBRACK);
5230 if (PL_expect == XREF)
5233 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
5239 pl_yylval.ival = CopLINE(PL_curcop);
5240 if (isSPACE(*s) || *s == '#')
5241 PL_copline = NOLINE; /* invalidate current command line number */
5246 if (PL_lex_brackets <= 0)
5247 yyerror("Unmatched right curly bracket");
5249 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
5250 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
5251 PL_lex_formbrack = 0;
5252 if (PL_lex_state == LEX_INTERPNORMAL) {
5253 if (PL_lex_brackets == 0) {
5254 if (PL_expect & XFAKEBRACK) {
5255 PL_expect &= XENUMMASK;
5256 PL_lex_state = LEX_INTERPEND;
5261 PL_thiswhite = newSVpvs("");
5262 sv_catpvs(PL_thiswhite,"}");
5265 return yylex(); /* ignore fake brackets */
5267 if (*s == '-' && s[1] == '>')
5268 PL_lex_state = LEX_INTERPENDMAYBE;
5269 else if (*s != '[' && *s != '{')
5270 PL_lex_state = LEX_INTERPEND;
5273 if (PL_expect & XFAKEBRACK) {
5274 PL_expect &= XENUMMASK;
5276 return yylex(); /* ignore fake brackets */
5278 start_force(PL_curforce);
5280 curmad('X', newSVpvn(s-1,1));
5281 CURMAD('_', PL_thiswhite);
5286 PL_thistoken = newSVpvs("");
5294 if (PL_expect == XOPERATOR) {
5295 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
5296 && isIDFIRST_lazy_if(s,UTF))
5298 CopLINE_dec(PL_curcop);
5299 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5300 CopLINE_inc(PL_curcop);
5305 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
5307 PL_expect = XOPERATOR;
5308 force_ident(PL_tokenbuf, '&');
5312 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
5324 const char tmp = *s++;
5331 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
5332 && strchr("+-*/%.^&|<",tmp))
5333 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5334 "Reversed %c= operator",(int)tmp);
5336 if (PL_expect == XSTATE && isALPHA(tmp) &&
5337 (s == PL_linestart+1 || s[-2] == '\n') )
5339 if (PL_in_eval && !PL_rsfp) {
5344 if (strnEQ(s,"=cut",4)) {
5360 PL_thiswhite = newSVpvs("");
5361 sv_catpvn(PL_thiswhite, PL_linestart,
5362 PL_bufend - PL_linestart);
5366 PL_doextract = TRUE;
5370 if (PL_lex_brackets < PL_lex_formbrack) {
5372 #ifdef PERL_STRICT_CR
5373 while (SPACE_OR_TAB(*t))
5375 while (SPACE_OR_TAB(*t) || *t == '\r')
5378 if (*t == '\n' || *t == '#') {
5389 const char tmp = *s++;
5391 /* was this !=~ where !~ was meant?
5392 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
5394 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
5395 const char *t = s+1;
5397 while (t < PL_bufend && isSPACE(*t))
5400 if (*t == '/' || *t == '?' ||
5401 ((*t == 'm' || *t == 's' || *t == 'y')
5402 && !isALNUM(t[1])) ||
5403 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
5404 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5405 "!=~ should be !~");
5415 if (PL_expect != XOPERATOR) {
5416 if (s[1] != '<' && !strchr(s,'>'))
5419 s = scan_heredoc(s);
5421 s = scan_inputsymbol(s);
5422 TERM(sublex_start());
5428 SHop(OP_LEFT_SHIFT);
5442 const char tmp = *s++;
5444 SHop(OP_RIGHT_SHIFT);
5445 else if (tmp == '=')
5454 if (PL_expect == XOPERATOR) {
5455 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5456 return deprecate_commaless_var_list();
5460 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
5461 PL_tokenbuf[0] = '@';
5462 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
5463 sizeof PL_tokenbuf - 1, FALSE);
5464 if (PL_expect == XOPERATOR)
5465 no_op("Array length", s);
5466 if (!PL_tokenbuf[1])
5468 PL_expect = XOPERATOR;
5469 PL_pending_ident = '#';
5473 PL_tokenbuf[0] = '$';
5474 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
5475 sizeof PL_tokenbuf - 1, FALSE);
5476 if (PL_expect == XOPERATOR)
5478 if (!PL_tokenbuf[1]) {
5480 yyerror("Final $ should be \\$ or $name");
5484 /* This kludge not intended to be bulletproof. */
5485 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
5486 pl_yylval.opval = newSVOP(OP_CONST, 0,
5487 newSViv(CopARYBASE_get(&PL_compiling)));
5488 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
5494 const char tmp = *s;
5495 if (PL_lex_state == LEX_NORMAL || PL_lex_brackets)
5498 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
5499 && intuit_more(s)) {
5501 PL_tokenbuf[0] = '@';
5502 if (ckWARN(WARN_SYNTAX)) {
5505 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
5508 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5509 while (t < PL_bufend && *t != ']')
5511 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5512 "Multidimensional syntax %.*s not supported",
5513 (int)((t - PL_bufptr) + 1), PL_bufptr);
5517 else if (*s == '{') {
5519 PL_tokenbuf[0] = '%';
5520 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
5521 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
5523 char tmpbuf[sizeof PL_tokenbuf];
5526 } while (isSPACE(*t));
5527 if (isIDFIRST_lazy_if(t,UTF)) {
5529 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
5533 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
5534 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5535 "You need to quote \"%s\"",
5542 PL_expect = XOPERATOR;
5543 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
5544 const bool islop = (PL_last_lop == PL_oldoldbufptr);
5545 if (!islop || PL_last_lop_op == OP_GREPSTART)
5546 PL_expect = XOPERATOR;
5547 else if (strchr("$@\"'`q", *s))
5548 PL_expect = XTERM; /* e.g. print $fh "foo" */
5549 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
5550 PL_expect = XTERM; /* e.g. print $fh &sub */
5551 else if (isIDFIRST_lazy_if(s,UTF)) {
5552 char tmpbuf[sizeof PL_tokenbuf];
5554 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
5555 if ((t2 = keyword(tmpbuf, len, 0))) {
5556 /* binary operators exclude handle interpretations */
5568 PL_expect = XTERM; /* e.g. print $fh length() */
5573 PL_expect = XTERM; /* e.g. print $fh subr() */
5576 else if (isDIGIT(*s))
5577 PL_expect = XTERM; /* e.g. print $fh 3 */
5578 else if (*s == '.' && isDIGIT(s[1]))
5579 PL_expect = XTERM; /* e.g. print $fh .3 */
5580 else if ((*s == '?' || *s == '-' || *s == '+')
5581 && !isSPACE(s[1]) && s[1] != '=')
5582 PL_expect = XTERM; /* e.g. print $fh -1 */
5583 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
5585 PL_expect = XTERM; /* e.g. print $fh /.../
5586 XXX except DORDOR operator
5588 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
5590 PL_expect = XTERM; /* print $fh <<"EOF" */
5593 PL_pending_ident = '$';
5597 if (PL_expect == XOPERATOR)
5599 PL_tokenbuf[0] = '@';
5600 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
5601 if (!PL_tokenbuf[1]) {
5604 if (PL_lex_state == LEX_NORMAL)
5606 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5608 PL_tokenbuf[0] = '%';
5610 /* Warn about @ where they meant $. */
5611 if (*s == '[' || *s == '{') {
5612 if (ckWARN(WARN_SYNTAX)) {
5613 const char *t = s + 1;
5614 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5616 if (*t == '}' || *t == ']') {
5618 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5619 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5620 "Scalar value %.*s better written as $%.*s",
5621 (int)(t-PL_bufptr), PL_bufptr,
5622 (int)(t-PL_bufptr-1), PL_bufptr+1);
5627 PL_pending_ident = '@';
5630 case '/': /* may be division, defined-or, or pattern */
5631 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5635 case '?': /* may either be conditional or pattern */
5636 if (PL_expect == XOPERATOR) {
5644 /* A // operator. */
5654 /* Disable warning on "study /blah/" */
5655 if (PL_oldoldbufptr == PL_last_uni
5656 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5657 || memNE(PL_last_uni, "study", 5)
5658 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5661 s = scan_pat(s,OP_MATCH);
5662 TERM(sublex_start());
5666 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5667 #ifdef PERL_STRICT_CR
5670 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5672 && (s == PL_linestart || s[-1] == '\n') )
5674 PL_lex_formbrack = 0;
5678 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5682 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5688 pl_yylval.ival = OPf_SPECIAL;
5697 case '0': case '1': case '2': case '3': case '4':
5698 case '5': case '6': case '7': case '8': case '9':
5699 s = scan_num(s, &pl_yylval);
5700 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5701 if (PL_expect == XOPERATOR)
5706 s = scan_str(s,!!PL_madskills,FALSE);
5707 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5708 if (PL_expect == XOPERATOR) {
5709 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5710 return deprecate_commaless_var_list();
5717 pl_yylval.ival = OP_CONST;
5718 TERM(sublex_start());
5721 s = scan_str(s,!!PL_madskills,FALSE);
5722 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5723 if (PL_expect == XOPERATOR) {
5724 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5725 return deprecate_commaless_var_list();
5732 pl_yylval.ival = OP_CONST;
5733 /* FIXME. I think that this can be const if char *d is replaced by
5734 more localised variables. */
5735 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5736 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5737 pl_yylval.ival = OP_STRINGIFY;
5741 TERM(sublex_start());
5744 s = scan_str(s,!!PL_madskills,FALSE);
5745 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5746 if (PL_expect == XOPERATOR)
5747 no_op("Backticks",s);
5750 readpipe_override();
5751 TERM(sublex_start());
5755 if (PL_lex_inwhat && isDIGIT(*s))
5756 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5758 if (PL_expect == XOPERATOR)
5759 no_op("Backslash",s);
5763 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5764 char *start = s + 2;
5765 while (isDIGIT(*start) || *start == '_')
5767 if (*start == '.' && isDIGIT(start[1])) {
5768 s = scan_num(s, &pl_yylval);
5771 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5772 else if (!isALPHA(*start) && (PL_expect == XTERM
5773 || PL_expect == XREF || PL_expect == XSTATE
5774 || PL_expect == XTERMORDORDOR)) {
5775 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5777 s = scan_num(s, &pl_yylval);
5784 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5827 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5829 /* Some keywords can be followed by any delimiter, including ':' */
5830 anydelim = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5831 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5832 (PL_tokenbuf[0] == 'q' &&
5833 strchr("qwxr", PL_tokenbuf[1])))));
5835 /* x::* is just a word, unless x is "CORE" */
5836 if (!anydelim && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5840 while (d < PL_bufend && isSPACE(*d))
5841 d++; /* no comments skipped here, or s### is misparsed */
5843 /* Is this a word before a => operator? */
5844 if (*d == '=' && d[1] == '>') {
5847 = (OP*)newSVOP(OP_CONST, 0,
5848 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5849 pl_yylval.opval->op_private = OPpCONST_BARE;
5853 /* Check for plugged-in keyword */
5857 char *saved_bufptr = PL_bufptr;
5859 result = CALL_FPTR(PL_keyword_plugin)(aTHX_ PL_tokenbuf, len, &o);
5861 if (result == KEYWORD_PLUGIN_DECLINE) {
5862 /* not a plugged-in keyword */
5863 PL_bufptr = saved_bufptr;
5864 } else if (result == KEYWORD_PLUGIN_STMT) {
5865 pl_yylval.opval = o;
5868 return REPORT(PLUGSTMT);
5869 } else if (result == KEYWORD_PLUGIN_EXPR) {
5870 pl_yylval.opval = o;
5872 PL_expect = XOPERATOR;
5873 return REPORT(PLUGEXPR);
5875 Perl_croak(aTHX_ "Bad plugin affecting keyword '%s'",
5880 /* Check for built-in keyword */
5881 tmp = keyword(PL_tokenbuf, len, 0);
5883 /* Is this a label? */
5884 if (!anydelim && PL_expect == XSTATE
5885 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5887 Perl_croak(aTHX_ "Can't use keyword '%s' as a label", PL_tokenbuf);
5889 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5894 if (tmp < 0) { /* second-class keyword? */
5895 GV *ogv = NULL; /* override (winner) */
5896 GV *hgv = NULL; /* hidden (loser) */
5897 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5899 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5902 if (GvIMPORTED_CV(gv))
5904 else if (! CvMETHOD(cv))
5908 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5909 (gv = *gvp) && isGV_with_GP(gv) &&
5910 GvCVu(gv) && GvIMPORTED_CV(gv))
5917 tmp = 0; /* overridden by import or by GLOBAL */
5920 && -tmp==KEY_lock /* XXX generalizable kludge */
5923 tmp = 0; /* any sub overrides "weak" keyword */
5925 else { /* no override */
5927 if (tmp == KEY_dump) {
5928 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
5929 "dump() better written as CORE::dump()");
5933 if (hgv && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */
5934 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5935 "Ambiguous call resolved as CORE::%s(), %s",
5936 GvENAME(hgv), "qualify as such or use &");
5943 default: /* not a keyword */
5944 /* Trade off - by using this evil construction we can pull the
5945 variable gv into the block labelled keylookup. If not, then
5946 we have to give it function scope so that the goto from the
5947 earlier ':' case doesn't bypass the initialisation. */
5949 just_a_word_zero_gv:
5957 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5961 SV *nextPL_nextwhite = 0;
5965 /* Get the rest if it looks like a package qualifier */
5967 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5969 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5972 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5973 *s == '\'' ? "'" : "::");
5978 if (PL_expect == XOPERATOR) {
5979 if (PL_bufptr == PL_linestart) {
5980 CopLINE_dec(PL_curcop);
5981 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5982 CopLINE_inc(PL_curcop);
5985 no_op("Bareword",s);
5988 /* Look for a subroutine with this name in current package,
5989 unless name is "Foo::", in which case Foo is a bearword
5990 (and a package name). */
5992 if (len > 2 && !PL_madskills &&
5993 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5995 if (ckWARN(WARN_BAREWORD)
5996 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5997 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5998 "Bareword \"%s\" refers to nonexistent package",
6001 PL_tokenbuf[len] = '\0';
6007 /* Mustn't actually add anything to a symbol table.
6008 But also don't want to "initialise" any placeholder
6009 constants that might already be there into full
6010 blown PVGVs with attached PVCV. */
6011 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
6012 GV_NOADD_NOINIT, SVt_PVCV);
6017 /* if we saw a global override before, get the right name */
6020 sv = newSVpvs("CORE::GLOBAL::");
6021 sv_catpv(sv,PL_tokenbuf);
6024 /* If len is 0, newSVpv does strlen(), which is correct.
6025 If len is non-zero, then it will be the true length,
6026 and so the scalar will be created correctly. */
6027 sv = newSVpv(PL_tokenbuf,len);
6030 if (PL_madskills && !PL_thistoken) {
6031 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
6032 PL_thistoken = newSVpvn(start,s - start);
6033 PL_realtokenstart = s - SvPVX(PL_linestr);
6037 /* Presume this is going to be a bareword of some sort. */
6040 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
6041 pl_yylval.opval->op_private = OPpCONST_BARE;
6042 /* UTF-8 package name? */
6043 if (UTF && !IN_BYTES &&
6044 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
6047 /* And if "Foo::", then that's what it certainly is. */
6054 OP *const_op = newSVOP(OP_CONST, 0, SvREFCNT_inc(sv));
6055 const_op->op_private = OPpCONST_BARE;
6056 rv2cv_op = newCVREF(0, const_op);
6058 if (rv2cv_op->op_type == OP_RV2CV &&
6059 (rv2cv_op->op_flags & OPf_KIDS)) {
6060 OP *rv_op = cUNOPx(rv2cv_op)->op_first;
6061 switch (rv_op->op_type) {
6063 SV *sv = cSVOPx_sv(rv_op);
6064 if (SvROK(sv) && SvTYPE(SvRV(sv)) == SVt_PVCV)
6068 GV *gv = cGVOPx_gv(rv_op);
6069 CV *maybe_cv = GvCVu(gv);
6070 if (maybe_cv && SvTYPE((SV*)maybe_cv) == SVt_PVCV)
6076 /* See if it's the indirect object for a list operator. */
6078 if (PL_oldoldbufptr &&
6079 PL_oldoldbufptr < PL_bufptr &&
6080 (PL_oldoldbufptr == PL_last_lop
6081 || PL_oldoldbufptr == PL_last_uni) &&
6082 /* NO SKIPSPACE BEFORE HERE! */
6083 (PL_expect == XREF ||
6084 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
6086 bool immediate_paren = *s == '(';
6088 /* (Now we can afford to cross potential line boundary.) */
6089 s = SKIPSPACE2(s,nextPL_nextwhite);
6091 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
6094 /* Two barewords in a row may indicate method call. */
6096 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
6097 (tmp = intuit_method(s, gv, cv))) {
6102 /* If not a declared subroutine, it's an indirect object. */
6103 /* (But it's an indir obj regardless for sort.) */
6104 /* Also, if "_" follows a filetest operator, it's a bareword */
6107 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
6109 (PL_last_lop_op != OP_MAPSTART &&
6110 PL_last_lop_op != OP_GREPSTART))))
6111 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
6112 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
6115 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
6120 PL_expect = XOPERATOR;
6123 s = SKIPSPACE2(s,nextPL_nextwhite);
6124 PL_nextwhite = nextPL_nextwhite;
6129 /* Is this a word before a => operator? */
6130 if (*s == '=' && s[1] == '>' && !pkgname) {
6133 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
6134 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
6135 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
6139 /* If followed by a paren, it's certainly a subroutine. */
6144 while (SPACE_OR_TAB(*d))
6146 if (*d == ')' && (sv = cv_const_sv(cv))) {
6153 PL_nextwhite = PL_thiswhite;
6156 start_force(PL_curforce);
6158 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6159 PL_expect = XOPERATOR;
6162 PL_nextwhite = nextPL_nextwhite;
6163 curmad('X', PL_thistoken);
6164 PL_thistoken = newSVpvs("");
6173 /* If followed by var or block, call it a method (unless sub) */
6175 if ((*s == '$' || *s == '{') && !cv) {
6177 PL_last_lop = PL_oldbufptr;
6178 PL_last_lop_op = OP_METHOD;
6182 /* If followed by a bareword, see if it looks like indir obj. */
6185 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
6186 && (tmp = intuit_method(s, gv, cv))) {
6191 /* Not a method, so call it a subroutine (if defined) */
6194 if (lastchar == '-')
6195 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
6196 "Ambiguous use of -%s resolved as -&%s()",
6197 PL_tokenbuf, PL_tokenbuf);
6198 /* Check for a constant sub */
6199 if ((sv = cv_const_sv(cv))) {
6202 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
6203 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
6204 pl_yylval.opval->op_private = 0;
6208 op_free(pl_yylval.opval);
6209 pl_yylval.opval = rv2cv_op;
6210 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
6211 PL_last_lop = PL_oldbufptr;
6212 PL_last_lop_op = OP_ENTERSUB;
6213 /* Is there a prototype? */
6221 const char *proto = SvPV_const(MUTABLE_SV(cv), protolen);
6224 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
6226 while (*proto == ';')
6228 if (*proto == '&' && *s == '{') {
6230 sv_setpvs(PL_subname, "__ANON__");
6232 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6239 PL_nextwhite = PL_thiswhite;
6242 start_force(PL_curforce);
6243 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6246 PL_nextwhite = nextPL_nextwhite;
6247 curmad('X', PL_thistoken);
6248 PL_thistoken = newSVpvs("");
6255 /* Guess harder when madskills require "best effort". */
6256 if (PL_madskills && (!gv || !GvCVu(gv))) {
6257 int probable_sub = 0;
6258 if (strchr("\"'`$@%0123456789!*+{[<", *s))
6260 else if (isALPHA(*s)) {
6264 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
6265 if (!keyword(tmpbuf, tmplen, 0))
6268 while (d < PL_bufend && isSPACE(*d))
6270 if (*d == '=' && d[1] == '>')
6275 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
6276 op_free(pl_yylval.opval);
6277 pl_yylval.opval = rv2cv_op;
6278 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
6279 PL_last_lop = PL_oldbufptr;
6280 PL_last_lop_op = OP_ENTERSUB;
6281 PL_nextwhite = PL_thiswhite;
6283 start_force(PL_curforce);
6284 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6286 PL_nextwhite = nextPL_nextwhite;
6287 curmad('X', PL_thistoken);
6288 PL_thistoken = newSVpvs("");
6293 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
6300 /* Call it a bare word */
6302 if (PL_hints & HINT_STRICT_SUBS)
6303 pl_yylval.opval->op_private |= OPpCONST_STRICT;
6306 /* after "print" and similar functions (corresponding to
6307 * "F? L" in opcode.pl), whatever wasn't already parsed as
6308 * a filehandle should be subject to "strict subs".
6309 * Likewise for the optional indirect-object argument to system
6310 * or exec, which can't be a bareword */
6311 if ((PL_last_lop_op == OP_PRINT
6312 || PL_last_lop_op == OP_PRTF
6313 || PL_last_lop_op == OP_SAY
6314 || PL_last_lop_op == OP_SYSTEM
6315 || PL_last_lop_op == OP_EXEC)
6316 && (PL_hints & HINT_STRICT_SUBS))
6317 pl_yylval.opval->op_private |= OPpCONST_STRICT;
6318 if (lastchar != '-') {
6319 if (ckWARN(WARN_RESERVED)) {
6323 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
6324 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
6332 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')) {
6333 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
6334 "Operator or semicolon missing before %c%s",
6335 lastchar, PL_tokenbuf);
6336 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
6337 "Ambiguous use of %c resolved as operator %c",
6338 lastchar, lastchar);
6344 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
6345 newSVpv(CopFILE(PL_curcop),0));
6349 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
6350 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
6353 case KEY___PACKAGE__:
6354 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
6356 ? newSVhek(HvNAME_HEK(PL_curstash))
6363 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
6364 const char *pname = "main";
6365 if (PL_tokenbuf[2] == 'D')
6366 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
6367 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
6371 GvIOp(gv) = newIO();
6372 IoIFP(GvIOp(gv)) = PL_rsfp;
6373 #if defined(HAS_FCNTL) && defined(F_SETFD)
6375 const int fd = PerlIO_fileno(PL_rsfp);
6376 fcntl(fd,F_SETFD,fd >= 3);
6379 /* Mark this internal pseudo-handle as clean */
6380 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
6381 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
6382 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
6384 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
6385 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
6386 /* if the script was opened in binmode, we need to revert
6387 * it to text mode for compatibility; but only iff it has CRs
6388 * XXX this is a questionable hack at best. */
6389 if (PL_bufend-PL_bufptr > 2
6390 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
6393 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
6394 loc = PerlIO_tell(PL_rsfp);
6395 (void)PerlIO_seek(PL_rsfp, 0L, 0);
6398 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
6400 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
6401 #endif /* NETWARE */
6402 #ifdef PERLIO_IS_STDIO /* really? */
6403 # if defined(__BORLANDC__)
6404 /* XXX see note in do_binmode() */
6405 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
6409 PerlIO_seek(PL_rsfp, loc, 0);
6413 #ifdef PERLIO_LAYERS
6416 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
6417 else if (PL_encoding) {
6424 XPUSHs(PL_encoding);
6426 call_method("name", G_SCALAR);
6430 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
6431 Perl_form(aTHX_ ":encoding(%"SVf")",
6440 if (PL_realtokenstart >= 0) {
6441 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6443 PL_endwhite = newSVpvs("");
6444 sv_catsv(PL_endwhite, PL_thiswhite);
6446 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
6447 PL_realtokenstart = -1;
6449 while ((s = filter_gets(PL_endwhite, SvCUR(PL_endwhite)))
6465 if (PL_expect == XSTATE) {
6472 if (*s == ':' && s[1] == ':') {
6475 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
6476 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
6477 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
6480 else if (tmp == KEY_require || tmp == KEY_do)
6481 /* that's a way to remember we saw "CORE::" */
6494 LOP(OP_ACCEPT,XTERM);
6500 LOP(OP_ATAN2,XTERM);
6506 LOP(OP_BINMODE,XTERM);
6509 LOP(OP_BLESS,XTERM);
6518 /* When 'use switch' is in effect, continue has a dual
6519 life as a control operator. */
6521 if (!FEATURE_IS_ENABLED("switch"))
6524 /* We have to disambiguate the two senses of
6525 "continue". If the next token is a '{' then
6526 treat it as the start of a continue block;
6527 otherwise treat it as a control operator.
6539 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
6556 if (!PL_cryptseen) {
6557 PL_cryptseen = TRUE;
6561 LOP(OP_CRYPT,XTERM);
6564 LOP(OP_CHMOD,XTERM);
6567 LOP(OP_CHOWN,XTERM);
6570 LOP(OP_CONNECT,XTERM);
6589 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6590 if (orig_keyword == KEY_do) {
6599 PL_hints |= HINT_BLOCK_SCOPE;
6609 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
6610 LOP(OP_DBMOPEN,XTERM);
6616 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6623 pl_yylval.ival = CopLINE(PL_curcop);
6639 if (*s == '{') { /* block eval */
6640 PL_expect = XTERMBLOCK;
6641 UNIBRACK(OP_ENTERTRY);
6643 else { /* string eval */
6645 UNIBRACK(OP_ENTEREVAL);
6660 case KEY_endhostent:
6666 case KEY_endservent:
6669 case KEY_endprotoent:
6680 pl_yylval.ival = CopLINE(PL_curcop);
6682 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6685 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6688 if ((PL_bufend - p) >= 3 &&
6689 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6691 else if ((PL_bufend - p) >= 4 &&
6692 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6695 if (isIDFIRST_lazy_if(p,UTF)) {
6696 p = scan_ident(p, PL_bufend,
6697 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6701 Perl_croak(aTHX_ "Missing $ on loop variable");
6703 s = SvPVX(PL_linestr) + soff;
6709 LOP(OP_FORMLINE,XTERM);
6715 LOP(OP_FCNTL,XTERM);
6721 LOP(OP_FLOCK,XTERM);
6730 LOP(OP_GREPSTART, XREF);
6733 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6748 case KEY_getpriority:
6749 LOP(OP_GETPRIORITY,XTERM);
6751 case KEY_getprotobyname:
6754 case KEY_getprotobynumber:
6755 LOP(OP_GPBYNUMBER,XTERM);
6757 case KEY_getprotoent:
6769 case KEY_getpeername:
6770 UNI(OP_GETPEERNAME);
6772 case KEY_gethostbyname:
6775 case KEY_gethostbyaddr:
6776 LOP(OP_GHBYADDR,XTERM);
6778 case KEY_gethostent:
6781 case KEY_getnetbyname:
6784 case KEY_getnetbyaddr:
6785 LOP(OP_GNBYADDR,XTERM);
6790 case KEY_getservbyname:
6791 LOP(OP_GSBYNAME,XTERM);
6793 case KEY_getservbyport:
6794 LOP(OP_GSBYPORT,XTERM);
6796 case KEY_getservent:
6799 case KEY_getsockname:
6800 UNI(OP_GETSOCKNAME);
6802 case KEY_getsockopt:
6803 LOP(OP_GSOCKOPT,XTERM);
6818 pl_yylval.ival = CopLINE(PL_curcop);
6828 pl_yylval.ival = CopLINE(PL_curcop);
6832 LOP(OP_INDEX,XTERM);
6838 LOP(OP_IOCTL,XTERM);
6850 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6882 LOP(OP_LISTEN,XTERM);
6891 s = scan_pat(s,OP_MATCH);
6892 TERM(sublex_start());
6895 LOP(OP_MAPSTART, XREF);
6898 LOP(OP_MKDIR,XTERM);
6901 LOP(OP_MSGCTL,XTERM);
6904 LOP(OP_MSGGET,XTERM);
6907 LOP(OP_MSGRCV,XTERM);
6910 LOP(OP_MSGSND,XTERM);
6915 PL_in_my = (U16)tmp;
6917 if (isIDFIRST_lazy_if(s,UTF)) {
6921 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6922 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6924 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6925 if (!PL_in_my_stash) {
6928 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6932 if (PL_madskills) { /* just add type to declarator token */
6933 sv_catsv(PL_thistoken, PL_nextwhite);
6935 sv_catpvn(PL_thistoken, start, s - start);
6943 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6950 s = tokenize_use(0, s);
6954 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6961 if (isIDFIRST_lazy_if(s,UTF)) {
6963 for (d = s; isALNUM_lazy_if(d,UTF);)
6965 for (t=d; isSPACE(*t);)
6967 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6969 && !(t[0] == '=' && t[1] == '>')
6971 int parms_len = (int)(d-s);
6972 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6973 "Precedence problem: open %.*s should be open(%.*s)",
6974 parms_len, s, parms_len, s);
6980 pl_yylval.ival = OP_OR;
6990 LOP(OP_OPEN_DIR,XTERM);
6993 checkcomma(s,PL_tokenbuf,"filehandle");
6997 checkcomma(s,PL_tokenbuf,"filehandle");
7016 s = force_word(s,WORD,FALSE,TRUE,FALSE);
7018 s = force_strict_version(s);
7022 LOP(OP_PIPE_OP,XTERM);
7025 s = scan_str(s,!!PL_madskills,FALSE);
7028 pl_yylval.ival = OP_CONST;
7029 TERM(sublex_start());
7035 s = scan_str(s,!!PL_madskills,FALSE);
7038 PL_expect = XOPERATOR;
7040 if (SvCUR(PL_lex_stuff)) {
7043 d = SvPV_force(PL_lex_stuff, len);
7045 for (; isSPACE(*d) && len; --len, ++d)
7050 if (!warned && ckWARN(WARN_QW)) {
7051 for (; !isSPACE(*d) && len; --len, ++d) {
7053 Perl_warner(aTHX_ packWARN(WARN_QW),
7054 "Possible attempt to separate words with commas");
7057 else if (*d == '#') {
7058 Perl_warner(aTHX_ packWARN(WARN_QW),
7059 "Possible attempt to put comments in qw() list");
7065 for (; !isSPACE(*d) && len; --len, ++d)
7068 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
7069 words = append_elem(OP_LIST, words,
7070 newSVOP(OP_CONST, 0, tokeq(sv)));
7074 start_force(PL_curforce);
7075 NEXTVAL_NEXTTOKE.opval = words;
7080 SvREFCNT_dec(PL_lex_stuff);
7081 PL_lex_stuff = NULL;
7087 s = scan_str(s,!!PL_madskills,FALSE);
7090 pl_yylval.ival = OP_STRINGIFY;
7091 if (SvIVX(PL_lex_stuff) == '\'')
7092 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
7093 TERM(sublex_start());
7096 s = scan_pat(s,OP_QR);
7097 TERM(sublex_start());
7100 s = scan_str(s,!!PL_madskills,FALSE);
7103 readpipe_override();
7104 TERM(sublex_start());
7112 s = force_version(s, FALSE);
7114 else if (*s != 'v' || !isDIGIT(s[1])
7115 || (s = force_version(s, TRUE), *s == 'v'))
7117 *PL_tokenbuf = '\0';
7118 s = force_word(s,WORD,TRUE,TRUE,FALSE);
7119 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
7120 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
7122 yyerror("<> should be quotes");
7124 if (orig_keyword == KEY_require) {
7132 PL_last_uni = PL_oldbufptr;
7133 PL_last_lop_op = OP_REQUIRE;
7135 return REPORT( (int)REQUIRE );
7141 s = force_word(s,WORD,TRUE,FALSE,FALSE);
7145 LOP(OP_RENAME,XTERM);
7154 LOP(OP_RINDEX,XTERM);
7163 UNIDOR(OP_READLINE);
7166 UNIDOR(OP_BACKTICK);
7175 LOP(OP_REVERSE,XTERM);
7178 UNIDOR(OP_READLINK);
7185 if (pl_yylval.opval)
7186 TERM(sublex_start());
7188 TOKEN(1); /* force error */
7191 checkcomma(s,PL_tokenbuf,"filehandle");
7201 LOP(OP_SELECT,XTERM);
7207 LOP(OP_SEMCTL,XTERM);
7210 LOP(OP_SEMGET,XTERM);
7213 LOP(OP_SEMOP,XTERM);
7219 LOP(OP_SETPGRP,XTERM);
7221 case KEY_setpriority:
7222 LOP(OP_SETPRIORITY,XTERM);
7224 case KEY_sethostent:
7230 case KEY_setservent:
7233 case KEY_setprotoent:
7243 LOP(OP_SEEKDIR,XTERM);
7245 case KEY_setsockopt:
7246 LOP(OP_SSOCKOPT,XTERM);
7252 LOP(OP_SHMCTL,XTERM);
7255 LOP(OP_SHMGET,XTERM);
7258 LOP(OP_SHMREAD,XTERM);
7261 LOP(OP_SHMWRITE,XTERM);
7264 LOP(OP_SHUTDOWN,XTERM);
7273 LOP(OP_SOCKET,XTERM);
7275 case KEY_socketpair:
7276 LOP(OP_SOCKPAIR,XTERM);
7279 checkcomma(s,PL_tokenbuf,"subroutine name");
7281 if (*s == ';' || *s == ')') /* probably a close */
7282 Perl_croak(aTHX_ "sort is now a reserved word");
7284 s = force_word(s,WORD,TRUE,TRUE,FALSE);
7288 LOP(OP_SPLIT,XTERM);
7291 LOP(OP_SPRINTF,XTERM);
7294 LOP(OP_SPLICE,XTERM);
7309 LOP(OP_SUBSTR,XTERM);
7315 char tmpbuf[sizeof PL_tokenbuf];
7316 SSize_t tboffset = 0;
7317 expectation attrful;
7318 bool have_name, have_proto;
7319 const int key = tmp;
7324 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
7325 SV *subtoken = newSVpvn(tstart, s - tstart);
7329 s = SKIPSPACE2(s,tmpwhite);
7334 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
7335 (*s == ':' && s[1] == ':'))
7338 SV *nametoke = NULL;
7342 attrful = XATTRBLOCK;
7343 /* remember buffer pos'n for later force_word */
7344 tboffset = s - PL_oldbufptr;
7345 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
7348 nametoke = newSVpvn(s, d - s);
7350 if (memchr(tmpbuf, ':', len))
7351 sv_setpvn(PL_subname, tmpbuf, len);
7353 sv_setsv(PL_subname,PL_curstname);
7354 sv_catpvs(PL_subname,"::");
7355 sv_catpvn(PL_subname,tmpbuf,len);
7362 CURMAD('X', nametoke);
7363 CURMAD('_', tmpwhite);
7364 (void) force_word(PL_oldbufptr + tboffset, WORD,
7367 s = SKIPSPACE2(d,tmpwhite);
7374 Perl_croak(aTHX_ "Missing name in \"my sub\"");
7375 PL_expect = XTERMBLOCK;
7376 attrful = XATTRTERM;
7377 sv_setpvs(PL_subname,"?");
7381 if (key == KEY_format) {
7383 PL_lex_formbrack = PL_lex_brackets + 1;
7385 PL_thistoken = subtoken;
7389 (void) force_word(PL_oldbufptr + tboffset, WORD,
7395 /* Look for a prototype */
7398 bool bad_proto = FALSE;
7399 bool in_brackets = FALSE;
7400 char greedy_proto = ' ';
7401 bool proto_after_greedy_proto = FALSE;
7402 bool must_be_last = FALSE;
7403 bool underscore = FALSE;
7404 bool seen_underscore = FALSE;
7405 const bool warnillegalproto = ckWARN(WARN_ILLEGALPROTO);
7407 s = scan_str(s,!!PL_madskills,FALSE);
7409 Perl_croak(aTHX_ "Prototype not terminated");
7410 /* strip spaces and check for bad characters */
7411 d = SvPVX(PL_lex_stuff);
7413 for (p = d; *p; ++p) {
7417 if (warnillegalproto) {
7419 proto_after_greedy_proto = TRUE;
7420 if (!strchr("$@%*;[]&\\_", *p)) {
7432 else if ( *p == ']' ) {
7433 in_brackets = FALSE;
7435 else if ( (*p == '@' || *p == '%') &&
7436 ( tmp < 2 || d[tmp-2] != '\\' ) &&
7438 must_be_last = TRUE;
7441 else if ( *p == '_' ) {
7442 underscore = seen_underscore = TRUE;
7449 if (proto_after_greedy_proto)
7450 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
7451 "Prototype after '%c' for %"SVf" : %s",
7452 greedy_proto, SVfARG(PL_subname), d);
7454 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
7455 "Illegal character %sin prototype for %"SVf" : %s",
7456 seen_underscore ? "after '_' " : "",
7457 SVfARG(PL_subname), d);
7458 SvCUR_set(PL_lex_stuff, tmp);
7463 CURMAD('q', PL_thisopen);
7464 CURMAD('_', tmpwhite);
7465 CURMAD('=', PL_thisstuff);
7466 CURMAD('Q', PL_thisclose);
7467 NEXTVAL_NEXTTOKE.opval =
7468 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
7469 PL_lex_stuff = NULL;
7472 s = SKIPSPACE2(s,tmpwhite);
7480 if (*s == ':' && s[1] != ':')
7481 PL_expect = attrful;
7482 else if (*s != '{' && key == KEY_sub) {
7484 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
7485 else if (*s != ';' && *s != '}')
7486 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
7493 curmad('^', newSVpvs(""));
7494 CURMAD('_', tmpwhite);
7498 PL_thistoken = subtoken;
7501 NEXTVAL_NEXTTOKE.opval =
7502 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
7503 PL_lex_stuff = NULL;
7509 sv_setpvs(PL_subname, "__ANON__");
7511 sv_setpvs(PL_subname, "__ANON__::__ANON__");
7515 (void) force_word(PL_oldbufptr + tboffset, WORD,
7524 LOP(OP_SYSTEM,XREF);
7527 LOP(OP_SYMLINK,XTERM);
7530 LOP(OP_SYSCALL,XTERM);
7533 LOP(OP_SYSOPEN,XTERM);
7536 LOP(OP_SYSSEEK,XTERM);
7539 LOP(OP_SYSREAD,XTERM);
7542 LOP(OP_SYSWRITE,XTERM);
7546 TERM(sublex_start());
7567 LOP(OP_TRUNCATE,XTERM);
7579 pl_yylval.ival = CopLINE(PL_curcop);
7583 pl_yylval.ival = CopLINE(PL_curcop);
7587 LOP(OP_UNLINK,XTERM);
7593 LOP(OP_UNPACK,XTERM);
7596 LOP(OP_UTIME,XTERM);
7602 LOP(OP_UNSHIFT,XTERM);
7605 s = tokenize_use(1, s);
7615 pl_yylval.ival = CopLINE(PL_curcop);
7619 pl_yylval.ival = CopLINE(PL_curcop);
7623 PL_hints |= HINT_BLOCK_SCOPE;
7630 LOP(OP_WAITPID,XTERM);
7639 ctl_l[0] = toCTRL('L');
7641 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
7644 /* Make sure $^L is defined */
7645 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
7650 if (PL_expect == XOPERATOR)
7656 pl_yylval.ival = OP_XOR;
7661 TERM(sublex_start());
7666 #pragma segment Main
7670 S_pending_ident(pTHX)
7675 /* pit holds the identifier we read and pending_ident is reset */
7676 char pit = PL_pending_ident;
7677 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7678 /* All routes through this function want to know if there is a colon. */
7679 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7680 PL_pending_ident = 0;
7682 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7683 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7684 "### Pending identifier '%s'\n", PL_tokenbuf); });
7686 /* if we're in a my(), we can't allow dynamics here.
7687 $foo'bar has already been turned into $foo::bar, so
7688 just check for colons.
7690 if it's a legal name, the OP is a PADANY.
7693 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7695 yyerror(Perl_form(aTHX_ "No package name allowed for "
7696 "variable %s in \"our\"",
7698 tmp = allocmy(PL_tokenbuf, tokenbuf_len, 0);
7702 yyerror(Perl_form(aTHX_ PL_no_myglob,
7703 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7705 pl_yylval.opval = newOP(OP_PADANY, 0);
7706 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf, tokenbuf_len, 0);
7712 build the ops for accesses to a my() variable.
7714 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7715 then used in a comparison. This catches most, but not
7716 all cases. For instance, it catches
7717 sort { my($a); $a <=> $b }
7719 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7720 (although why you'd do that is anyone's guess).
7725 tmp = pad_findmy(PL_tokenbuf, tokenbuf_len, 0);
7726 if (tmp != NOT_IN_PAD) {
7727 /* might be an "our" variable" */
7728 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7729 /* build ops for a bareword */
7730 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7731 HEK * const stashname = HvNAME_HEK(stash);
7732 SV * const sym = newSVhek(stashname);
7733 sv_catpvs(sym, "::");
7734 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7735 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7736 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7739 ? (GV_ADDMULTI | GV_ADDINEVAL)
7742 ((PL_tokenbuf[0] == '$') ? SVt_PV
7743 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7748 /* if it's a sort block and they're naming $a or $b */
7749 if (PL_last_lop_op == OP_SORT &&
7750 PL_tokenbuf[0] == '$' &&
7751 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7754 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7755 d < PL_bufend && *d != '\n';
7758 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7759 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7765 pl_yylval.opval = newOP(OP_PADANY, 0);
7766 pl_yylval.opval->op_targ = tmp;
7772 Whine if they've said @foo in a doublequoted string,
7773 and @foo isn't a variable we can find in the symbol
7776 if (ckWARN(WARN_AMBIGUOUS) &&
7777 pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7778 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7780 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7781 /* DO NOT warn for @- and @+ */
7782 && !( PL_tokenbuf[2] == '\0' &&
7783 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7786 /* Downgraded from fatal to warning 20000522 mjd */
7787 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7788 "Possible unintended interpolation of %s in string",
7793 /* build ops for a bareword */
7794 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7796 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7798 PL_tokenbuf + 1, tokenbuf_len - 1,
7799 /* If the identifier refers to a stash, don't autovivify it.
7800 * Change 24660 had the side effect of causing symbol table
7801 * hashes to always be defined, even if they were freshly
7802 * created and the only reference in the entire program was
7803 * the single statement with the defined %foo::bar:: test.
7804 * It appears that all code in the wild doing this actually
7805 * wants to know whether sub-packages have been loaded, so
7806 * by avoiding auto-vivifying symbol tables, we ensure that
7807 * defined %foo::bar:: continues to be false, and the existing
7808 * tests still give the expected answers, even though what
7809 * they're actually testing has now changed subtly.
7811 (*PL_tokenbuf == '%'
7812 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7815 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7816 ((PL_tokenbuf[0] == '$') ? SVt_PV
7817 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7823 * The following code was generated by perl_keyword.pl.
7827 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7831 PERL_ARGS_ASSERT_KEYWORD;
7835 case 1: /* 5 tokens of length 1 */
7867 case 2: /* 18 tokens of length 2 */
8013 case 3: /* 29 tokens of length 3 */
8017 if (name[1] == 'N' &&
8080 if (name[1] == 'i' &&
8112 if (name[1] == 'o' &&
8121 if (name[1] == 'e' &&
8130 if (name[1] == 'n' &&
8139 if (name[1] == 'o' &&
8148 if (name[1] == 'a' &&
8157 if (name[1] == 'o' &&
8219 if (name[1] == 'e' &&
8233 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
8259 if (name[1] == 'i' &&
8268 if (name[1] == 's' &&
8277 if (name[1] == 'e' &&
8286 if (name[1] == 'o' &&
8298 case 4: /* 41 tokens of length 4 */
8302 if (name[1] == 'O' &&
8312 if (name[1] == 'N' &&
8322 if (name[1] == 'i' &&
8332 if (name[1] == 'h' &&
8342 if (name[1] == 'u' &&
8355 if (name[2] == 'c' &&
8364 if (name[2] == 's' &&
8373 if (name[2] == 'a' &&
8409 if (name[1] == 'o' &&
8422 if (name[2] == 't' &&
8431 if (name[2] == 'o' &&
8440 if (name[2] == 't' &&
8449 if (name[2] == 'e' &&
8462 if (name[1] == 'o' &&
8475 if (name[2] == 'y' &&
8484 if (name[2] == 'l' &&
8500 if (name[2] == 's' &&
8509 if (name[2] == 'n' &&
8518 if (name[2] == 'c' &&
8531 if (name[1] == 'e' &&
8541 if (name[1] == 'p' &&
8554 if (name[2] == 'c' &&
8563 if (name[2] == 'p' &&
8572 if (name[2] == 's' &&
8588 if (name[2] == 'n' &&
8658 if (name[2] == 'r' &&
8667 if (name[2] == 'r' &&
8676 if (name[2] == 'a' &&
8692 if (name[2] == 'l' &&
8754 if (name[2] == 'e' &&
8757 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8770 case 5: /* 39 tokens of length 5 */
8774 if (name[1] == 'E' &&
8785 if (name[1] == 'H' &&
8799 if (name[2] == 'a' &&
8809 if (name[2] == 'a' &&
8826 if (name[2] == 'e' &&
8836 if (name[2] == 'e' &&
8840 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8856 if (name[3] == 'i' &&
8865 if (name[3] == 'o' &&
8901 if (name[2] == 'o' &&
8911 if (name[2] == 'y' &&
8925 if (name[1] == 'l' &&
8939 if (name[2] == 'n' &&
8949 if (name[2] == 'o' &&
8963 if (name[1] == 'i' &&
8968 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8977 if (name[2] == 'd' &&
8987 if (name[2] == 'c' &&
9004 if (name[2] == 'c' &&
9014 if (name[2] == 't' &&
9028 if (name[1] == 'k' &&
9039 if (name[1] == 'r' &&
9053 if (name[2] == 's' &&
9063 if (name[2] == 'd' &&
9080 if (name[2] == 'm' &&
9090 if (name[2] == 'i' &&
9100 if (name[2] == 'e' &&
9110 if (name[2] == 'l' &&
9120 if (name[2] == 'a' &&
9133 if (name[3] == 't' &&
9136 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
9142 if (name[3] == 'd' &&
9159 if (name[1] == 'i' &&
9173 if (name[2] == 'a' &&
9186 if (name[3] == 'e' &&
9221 if (name[2] == 'i' &&
9238 if (name[2] == 'i' &&
9248 if (name[2] == 'i' &&
9265 case 6: /* 33 tokens of length 6 */
9269 if (name[1] == 'c' &&
9284 if (name[2] == 'l' &&
9295 if (name[2] == 'r' &&
9310 if (name[1] == 'e' &&
9325 if (name[2] == 's' &&
9330 Perl_ck_warner_d(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
9336 if (name[2] == 'i' &&
9354 if (name[2] == 'l' &&
9365 if (name[2] == 'r' &&
9380 if (name[1] == 'm' &&
9395 if (name[2] == 'n' &&
9406 if (name[2] == 's' &&
9421 if (name[1] == 's' &&
9427 if (name[4] == 't' &&
9436 if (name[4] == 'e' &&
9445 if (name[4] == 'c' &&
9454 if (name[4] == 'n' &&
9470 if (name[1] == 'r' &&
9488 if (name[3] == 'a' &&
9498 if (name[3] == 'u' &&
9512 if (name[2] == 'n' &&
9530 if (name[2] == 'a' &&
9544 if (name[3] == 'e' &&
9557 if (name[4] == 't' &&
9566 if (name[4] == 'e' &&
9588 if (name[4] == 't' &&
9597 if (name[4] == 'e' &&
9613 if (name[2] == 'c' &&
9624 if (name[2] == 'l' &&
9635 if (name[2] == 'b' &&
9646 if (name[2] == 's' &&
9669 if (name[4] == 's' &&
9678 if (name[4] == 'n' &&
9691 if (name[3] == 'a' &&
9708 if (name[1] == 'a' &&
9723 case 7: /* 29 tokens of length 7 */
9727 if (name[1] == 'E' &&
9740 if (name[1] == '_' &&
9753 if (name[1] == 'i' &&
9760 return -KEY_binmode;
9766 if (name[1] == 'o' &&
9773 return -KEY_connect;
9782 if (name[2] == 'm' &&
9788 return -KEY_dbmopen;
9799 if (name[4] == 'u' &&
9803 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9809 if (name[4] == 'n' &&
9830 if (name[1] == 'o' &&
9843 if (name[1] == 'e' &&
9850 if (name[5] == 'r' &&
9853 return -KEY_getpgrp;
9859 if (name[5] == 'i' &&
9862 return -KEY_getppid;
9875 if (name[1] == 'c' &&
9882 return -KEY_lcfirst;
9888 if (name[1] == 'p' &&
9895 return -KEY_opendir;
9901 if (name[1] == 'a' &&
9919 if (name[3] == 'd' &&
9924 return -KEY_readdir;
9930 if (name[3] == 'u' &&
9941 if (name[3] == 'e' &&
9946 return -KEY_reverse;
9965 if (name[3] == 'k' &&
9970 return -KEY_seekdir;
9976 if (name[3] == 'p' &&
9981 return -KEY_setpgrp;
9991 if (name[2] == 'm' &&
9997 return -KEY_shmread;
10003 if (name[2] == 'r' &&
10009 return -KEY_sprintf;
10018 if (name[3] == 'l' &&
10023 return -KEY_symlink;
10032 if (name[4] == 'a' &&
10036 return -KEY_syscall;
10042 if (name[4] == 'p' &&
10046 return -KEY_sysopen;
10052 if (name[4] == 'e' &&
10056 return -KEY_sysread;
10062 if (name[4] == 'e' &&
10066 return -KEY_sysseek;
10084 if (name[1] == 'e' &&
10091 return -KEY_telldir;
10100 if (name[2] == 'f' &&
10106 return -KEY_ucfirst;
10112 if (name[2] == 's' &&
10118 return -KEY_unshift;
10128 if (name[1] == 'a' &&
10135 return -KEY_waitpid;
10144 case 8: /* 26 tokens of length 8 */
10148 if (name[1] == 'U' &&
10156 return KEY_AUTOLOAD;
10162 if (name[1] == '_')
10167 if (name[3] == 'A' &&
10173 return KEY___DATA__;
10179 if (name[3] == 'I' &&
10185 return -KEY___FILE__;
10191 if (name[3] == 'I' &&
10197 return -KEY___LINE__;
10213 if (name[2] == 'o' &&
10220 return -KEY_closedir;
10226 if (name[2] == 'n' &&
10233 return -KEY_continue;
10243 if (name[1] == 'b' &&
10251 return -KEY_dbmclose;
10257 if (name[1] == 'n' &&
10263 if (name[4] == 'r' &&
10268 return -KEY_endgrent;
10274 if (name[4] == 'w' &&
10279 return -KEY_endpwent;
10292 if (name[1] == 'o' &&
10300 return -KEY_formline;
10306 if (name[1] == 'e' &&
10312 if (name[4] == 'r')
10317 if (name[6] == 'n' &&
10320 return -KEY_getgrent;
10326 if (name[6] == 'i' &&
10329 return -KEY_getgrgid;
10335 if (name[6] == 'a' &&
10338 return -KEY_getgrnam;
10351 if (name[4] == 'o' &&
10356 return -KEY_getlogin;
10362 if (name[4] == 'w')
10367 if (name[6] == 'n' &&
10370 return -KEY_getpwent;
10376 if (name[6] == 'a' &&
10379 return -KEY_getpwnam;
10385 if (name[6] == 'i' &&
10388 return -KEY_getpwuid;
10408 if (name[1] == 'e' &&
10415 if (name[5] == 'i' &&
10422 return -KEY_readline;
10427 return -KEY_readlink;
10438 if (name[5] == 'i' &&
10442 return -KEY_readpipe;
10458 if (name[2] == 't')
10463 if (name[4] == 'r' &&
10468 return -KEY_setgrent;
10474 if (name[4] == 'w' &&
10479 return -KEY_setpwent;
10495 if (name[3] == 'w' &&
10501 return -KEY_shmwrite;
10507 if (name[3] == 't' &&
10513 return -KEY_shutdown;
10523 if (name[2] == 's' &&
10530 return -KEY_syswrite;
10540 if (name[1] == 'r' &&
10548 return -KEY_truncate;
10557 case 9: /* 9 tokens of length 9 */
10561 if (name[1] == 'N' &&
10570 return KEY_UNITCHECK;
10576 if (name[1] == 'n' &&
10585 return -KEY_endnetent;
10591 if (name[1] == 'e' &&
10600 return -KEY_getnetent;
10606 if (name[1] == 'o' &&
10615 return -KEY_localtime;
10621 if (name[1] == 'r' &&
10630 return KEY_prototype;
10636 if (name[1] == 'u' &&
10645 return -KEY_quotemeta;
10651 if (name[1] == 'e' &&
10660 return -KEY_rewinddir;
10666 if (name[1] == 'e' &&
10675 return -KEY_setnetent;
10681 if (name[1] == 'a' &&
10690 return -KEY_wantarray;
10699 case 10: /* 9 tokens of length 10 */
10703 if (name[1] == 'n' &&
10709 if (name[4] == 'o' &&
10716 return -KEY_endhostent;
10722 if (name[4] == 'e' &&
10729 return -KEY_endservent;
10742 if (name[1] == 'e' &&
10748 if (name[4] == 'o' &&
10755 return -KEY_gethostent;
10764 if (name[5] == 'r' &&
10770 return -KEY_getservent;
10776 if (name[5] == 'c' &&
10782 return -KEY_getsockopt;
10802 if (name[2] == 't')
10807 if (name[4] == 'o' &&
10814 return -KEY_sethostent;
10823 if (name[5] == 'r' &&
10829 return -KEY_setservent;
10835 if (name[5] == 'c' &&
10841 return -KEY_setsockopt;
10858 if (name[2] == 'c' &&
10867 return -KEY_socketpair;
10880 case 11: /* 8 tokens of length 11 */
10884 if (name[1] == '_' &&
10894 { /* __PACKAGE__ */
10895 return -KEY___PACKAGE__;
10901 if (name[1] == 'n' &&
10911 { /* endprotoent */
10912 return -KEY_endprotoent;
10918 if (name[1] == 'e' &&
10927 if (name[5] == 'e' &&
10933 { /* getpeername */
10934 return -KEY_getpeername;
10943 if (name[6] == 'o' &&
10948 { /* getpriority */
10949 return -KEY_getpriority;
10955 if (name[6] == 't' &&
10960 { /* getprotoent */
10961 return -KEY_getprotoent;
10975 if (name[4] == 'o' &&
10982 { /* getsockname */
10983 return -KEY_getsockname;
10996 if (name[1] == 'e' &&
11004 if (name[6] == 'o' &&
11009 { /* setpriority */
11010 return -KEY_setpriority;
11016 if (name[6] == 't' &&
11021 { /* setprotoent */
11022 return -KEY_setprotoent;
11038 case 12: /* 2 tokens of length 12 */
11039 if (name[0] == 'g' &&
11051 if (name[9] == 'd' &&
11054 { /* getnetbyaddr */
11055 return -KEY_getnetbyaddr;
11061 if (name[9] == 'a' &&
11064 { /* getnetbyname */
11065 return -KEY_getnetbyname;
11077 case 13: /* 4 tokens of length 13 */
11078 if (name[0] == 'g' &&
11085 if (name[4] == 'o' &&
11094 if (name[10] == 'd' &&
11097 { /* gethostbyaddr */
11098 return -KEY_gethostbyaddr;
11104 if (name[10] == 'a' &&
11107 { /* gethostbyname */
11108 return -KEY_gethostbyname;
11121 if (name[4] == 'e' &&
11130 if (name[10] == 'a' &&
11133 { /* getservbyname */
11134 return -KEY_getservbyname;
11140 if (name[10] == 'o' &&
11143 { /* getservbyport */
11144 return -KEY_getservbyport;
11163 case 14: /* 1 tokens of length 14 */
11164 if (name[0] == 'g' &&
11178 { /* getprotobyname */
11179 return -KEY_getprotobyname;
11184 case 16: /* 1 tokens of length 16 */
11185 if (name[0] == 'g' &&
11201 { /* getprotobynumber */
11202 return -KEY_getprotobynumber;
11216 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
11220 PERL_ARGS_ASSERT_CHECKCOMMA;
11222 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
11223 if (ckWARN(WARN_SYNTAX)) {
11226 for (w = s+2; *w && level; w++) {
11229 else if (*w == ')')
11232 while (isSPACE(*w))
11234 /* the list of chars below is for end of statements or
11235 * block / parens, boolean operators (&&, ||, //) and branch
11236 * constructs (or, and, if, until, unless, while, err, for).
11237 * Not a very solid hack... */
11238 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
11239 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11240 "%s (...) interpreted as function",name);
11243 while (s < PL_bufend && isSPACE(*s))
11247 while (s < PL_bufend && isSPACE(*s))
11249 if (isIDFIRST_lazy_if(s,UTF)) {
11250 const char * const w = s++;
11251 while (isALNUM_lazy_if(s,UTF))
11253 while (s < PL_bufend && isSPACE(*s))
11257 if (keyword(w, s - w, 0))
11260 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
11261 if (gv && GvCVu(gv))
11263 Perl_croak(aTHX_ "No comma allowed after %s", what);
11268 /* Either returns sv, or mortalizes sv and returns a new SV*.
11269 Best used as sv=new_constant(..., sv, ...).
11270 If s, pv are NULL, calls subroutine with one argument,
11271 and type is used with error messages only. */
11274 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
11275 SV *sv, SV *pv, const char *type, STRLEN typelen)
11278 HV * const table = GvHV(PL_hintgv); /* ^H */
11282 const char *why1 = "", *why2 = "", *why3 = "";
11284 PERL_ARGS_ASSERT_NEW_CONSTANT;
11286 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
11289 why2 = (const char *)
11290 (strEQ(key,"charnames")
11291 ? "(possibly a missing \"use charnames ...\")"
11293 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
11294 (type ? type: "undef"), why2);
11296 /* This is convoluted and evil ("goto considered harmful")
11297 * but I do not understand the intricacies of all the different
11298 * failure modes of %^H in here. The goal here is to make
11299 * the most probable error message user-friendly. --jhi */
11304 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
11305 (type ? type: "undef"), why1, why2, why3);
11307 yyerror(SvPVX_const(msg));
11311 cvp = hv_fetch(table, key, keylen, FALSE);
11312 if (!cvp || !SvOK(*cvp)) {
11315 why3 = "} is not defined";
11318 sv_2mortal(sv); /* Parent created it permanently */
11321 pv = newSVpvn_flags(s, len, SVs_TEMP);
11323 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
11325 typesv = &PL_sv_undef;
11327 PUSHSTACKi(PERLSI_OVERLOAD);
11339 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
11343 /* Check the eval first */
11344 if (!PL_in_eval && SvTRUE(ERRSV)) {
11345 sv_catpvs(ERRSV, "Propagated");
11346 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
11348 res = SvREFCNT_inc_simple(sv);
11352 SvREFCNT_inc_simple_void(res);
11361 why1 = "Call to &{$^H{";
11363 why3 = "}} did not return a defined value";
11371 /* Returns a NUL terminated string, with the length of the string written to
11375 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
11378 register char *d = dest;
11379 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
11381 PERL_ARGS_ASSERT_SCAN_WORD;
11385 Perl_croak(aTHX_ ident_too_long);
11386 if (isALNUM(*s)) /* UTF handled below */
11388 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
11393 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
11397 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
11398 char *t = s + UTF8SKIP(s);
11400 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
11404 Perl_croak(aTHX_ ident_too_long);
11405 Copy(s, d, len, char);
11418 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
11421 char *bracket = NULL;
11423 register char *d = dest;
11424 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
11426 PERL_ARGS_ASSERT_SCAN_IDENT;
11431 while (isDIGIT(*s)) {
11433 Perl_croak(aTHX_ ident_too_long);
11440 Perl_croak(aTHX_ ident_too_long);
11441 if (isALNUM(*s)) /* UTF handled below */
11443 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
11448 else if (*s == ':' && s[1] == ':') {
11452 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
11453 char *t = s + UTF8SKIP(s);
11454 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
11456 if (d + (t - s) > e)
11457 Perl_croak(aTHX_ ident_too_long);
11458 Copy(s, d, t - s, char);
11469 if (PL_lex_state != LEX_NORMAL)
11470 PL_lex_state = LEX_INTERPENDMAYBE;
11473 if (*s == '$' && s[1] &&
11474 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
11487 if (*d == '^' && *s && isCONTROLVAR(*s)) {
11492 if (isSPACE(s[-1])) {
11494 const char ch = *s++;
11495 if (!SPACE_OR_TAB(ch)) {
11501 if (isIDFIRST_lazy_if(d,UTF)) {
11505 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
11506 end += UTF8SKIP(end);
11507 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
11508 end += UTF8SKIP(end);
11510 Copy(s, d, end - s, char);
11515 while ((isALNUM(*s) || *s == ':') && d < e)
11518 Perl_croak(aTHX_ ident_too_long);
11521 while (s < send && SPACE_OR_TAB(*s))
11523 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
11524 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
11525 const char * const brack =
11527 ((*s == '[') ? "[...]" : "{...}");
11528 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
11529 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
11530 funny, dest, brack, funny, dest, brack);
11533 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
11537 /* Handle extended ${^Foo} variables
11538 * 1999-02-27 mjd-perl-patch@plover.com */
11539 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
11543 while (isALNUM(*s) && d < e) {
11547 Perl_croak(aTHX_ ident_too_long);
11552 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
11553 PL_lex_state = LEX_INTERPEND;
11556 if (PL_lex_state == LEX_NORMAL) {
11557 if (ckWARN(WARN_AMBIGUOUS) &&
11558 (keyword(dest, d - dest, 0)
11559 || get_cvn_flags(dest, d - dest, 0)))
11563 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
11564 "Ambiguous use of %c{%s} resolved to %c%s",
11565 funny, dest, funny, dest);
11570 s = bracket; /* let the parser handle it */
11574 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
11575 PL_lex_state = LEX_INTERPEND;
11580 S_pmflag(U32 pmfl, const char ch) {
11582 CASE_STD_PMMOD_FLAGS_PARSE_SET(&pmfl);
11583 case GLOBAL_PAT_MOD: pmfl |= PMf_GLOBAL; break;
11584 case CONTINUE_PAT_MOD: pmfl |= PMf_CONTINUE; break;
11585 case ONCE_PAT_MOD: pmfl |= PMf_KEEP; break;
11586 case KEEPCOPY_PAT_MOD: pmfl |= PMf_KEEPCOPY; break;
11592 Perl_pmflag(pTHX_ U32* pmfl, int ch)
11594 PERL_ARGS_ASSERT_PMFLAG;
11596 Perl_ck_warner_d(aTHX_ packWARN(WARN_DEPRECATED),
11597 "Perl_pmflag() is deprecated, and will be removed from the XS API");
11600 *pmfl = S_pmflag(*pmfl, (char)ch);
11605 S_scan_pat(pTHX_ char *start, I32 type)
11609 char *s = scan_str(start,!!PL_madskills,FALSE);
11610 const char * const valid_flags =
11611 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
11616 PERL_ARGS_ASSERT_SCAN_PAT;
11619 const char * const delimiter = skipspace(start);
11623 ? "Search pattern not terminated or ternary operator parsed as search pattern"
11624 : "Search pattern not terminated" ));
11627 pm = (PMOP*)newPMOP(type, 0);
11628 if (PL_multi_open == '?') {
11629 /* This is the only point in the code that sets PMf_ONCE: */
11630 pm->op_pmflags |= PMf_ONCE;
11632 /* Hence it's safe to do this bit of PMOP book-keeping here, which
11633 allows us to restrict the list needed by reset to just the ??
11635 assert(type != OP_TRANS);
11637 MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
11640 mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
11643 elements = mg->mg_len / sizeof(PMOP**);
11644 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
11645 ((PMOP**)mg->mg_ptr) [elements++] = pm;
11646 mg->mg_len = elements * sizeof(PMOP**);
11647 PmopSTASH_set(pm,PL_curstash);
11653 while (*s && strchr(valid_flags, *s))
11654 pm->op_pmflags = S_pmflag(pm->op_pmflags, *s++);
11656 if (PL_madskills && modstart != s) {
11657 SV* tmptoken = newSVpvn(modstart, s - modstart);
11658 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11661 /* issue a warning if /c is specified,but /g is not */
11662 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL))
11664 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP),
11665 "Use of /c modifier is meaningless without /g" );
11668 PL_lex_op = (OP*)pm;
11669 pl_yylval.ival = OP_MATCH;
11674 S_scan_subst(pTHX_ char *start)
11685 PERL_ARGS_ASSERT_SCAN_SUBST;
11687 pl_yylval.ival = OP_NULL;
11689 s = scan_str(start,!!PL_madskills,FALSE);
11692 Perl_croak(aTHX_ "Substitution pattern not terminated");
11694 if (s[-1] == PL_multi_open)
11697 if (PL_madskills) {
11698 CURMAD('q', PL_thisopen);
11699 CURMAD('_', PL_thiswhite);
11700 CURMAD('E', PL_thisstuff);
11701 CURMAD('Q', PL_thisclose);
11702 PL_realtokenstart = s - SvPVX(PL_linestr);
11706 first_start = PL_multi_start;
11707 s = scan_str(s,!!PL_madskills,FALSE);
11709 if (PL_lex_stuff) {
11710 SvREFCNT_dec(PL_lex_stuff);
11711 PL_lex_stuff = NULL;
11713 Perl_croak(aTHX_ "Substitution replacement not terminated");
11715 PL_multi_start = first_start; /* so whole substitution is taken together */
11717 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11720 if (PL_madskills) {
11721 CURMAD('z', PL_thisopen);
11722 CURMAD('R', PL_thisstuff);
11723 CURMAD('Z', PL_thisclose);
11729 if (*s == EXEC_PAT_MOD) {
11733 else if (strchr(S_PAT_MODS, *s))
11734 pm->op_pmflags = S_pmflag(pm->op_pmflags, *s++);
11740 if (PL_madskills) {
11742 curmad('m', newSVpvn(modstart, s - modstart));
11743 append_madprops(PL_thismad, (OP*)pm, 0);
11747 if ((pm->op_pmflags & PMf_CONTINUE)) {
11748 Perl_ck_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11752 SV * const repl = newSVpvs("");
11754 PL_sublex_info.super_bufptr = s;
11755 PL_sublex_info.super_bufend = PL_bufend;
11757 pm->op_pmflags |= PMf_EVAL;
11760 sv_catpvs(repl, "eval ");
11762 sv_catpvs(repl, "do ");
11764 sv_catpvs(repl, "{");
11765 sv_catsv(repl, PL_lex_repl);
11766 if (strchr(SvPVX(PL_lex_repl), '#'))
11767 sv_catpvs(repl, "\n");
11768 sv_catpvs(repl, "}");
11770 SvREFCNT_dec(PL_lex_repl);
11771 PL_lex_repl = repl;
11774 PL_lex_op = (OP*)pm;
11775 pl_yylval.ival = OP_SUBST;
11780 S_scan_trans(pTHX_ char *start)
11793 PERL_ARGS_ASSERT_SCAN_TRANS;
11795 pl_yylval.ival = OP_NULL;
11797 s = scan_str(start,!!PL_madskills,FALSE);
11799 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11801 if (s[-1] == PL_multi_open)
11804 if (PL_madskills) {
11805 CURMAD('q', PL_thisopen);
11806 CURMAD('_', PL_thiswhite);
11807 CURMAD('E', PL_thisstuff);
11808 CURMAD('Q', PL_thisclose);
11809 PL_realtokenstart = s - SvPVX(PL_linestr);
11813 s = scan_str(s,!!PL_madskills,FALSE);
11815 if (PL_lex_stuff) {
11816 SvREFCNT_dec(PL_lex_stuff);
11817 PL_lex_stuff = NULL;
11819 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11821 if (PL_madskills) {
11822 CURMAD('z', PL_thisopen);
11823 CURMAD('R', PL_thisstuff);
11824 CURMAD('Z', PL_thisclose);
11827 complement = del = squash = 0;
11834 complement = OPpTRANS_COMPLEMENT;
11837 del = OPpTRANS_DELETE;
11840 squash = OPpTRANS_SQUASH;
11849 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11850 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11851 o->op_private &= ~OPpTRANS_ALL;
11852 o->op_private |= del|squash|complement|
11853 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11854 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11857 pl_yylval.ival = OP_TRANS;
11860 if (PL_madskills) {
11862 curmad('m', newSVpvn(modstart, s - modstart));
11863 append_madprops(PL_thismad, o, 0);
11872 S_scan_heredoc(pTHX_ register char *s)
11876 I32 op_type = OP_SCALAR;
11880 const char *found_newline;
11884 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11886 I32 stuffstart = s - SvPVX(PL_linestr);
11889 PL_realtokenstart = -1;
11892 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11896 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11900 while (SPACE_OR_TAB(*peek))
11902 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11905 s = delimcpy(d, e, s, PL_bufend, term, &len);
11915 if (!isALNUM_lazy_if(s,UTF))
11916 deprecate("bare << to mean <<\"\"");
11917 for (; isALNUM_lazy_if(s,UTF); s++) {
11922 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11923 Perl_croak(aTHX_ "Delimiter for here document is too long");
11926 len = d - PL_tokenbuf;
11929 if (PL_madskills) {
11930 tstart = PL_tokenbuf + !outer;
11931 PL_thisclose = newSVpvn(tstart, len - !outer);
11932 tstart = SvPVX(PL_linestr) + stuffstart;
11933 PL_thisopen = newSVpvn(tstart, s - tstart);
11934 stuffstart = s - SvPVX(PL_linestr);
11937 #ifndef PERL_STRICT_CR
11938 d = strchr(s, '\r');
11940 char * const olds = s;
11942 while (s < PL_bufend) {
11948 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11957 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11964 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11965 herewas = newSVpvn(s,PL_bufend-s);
11969 herewas = newSVpvn(s-1,found_newline-s+1);
11972 herewas = newSVpvn(s,found_newline-s);
11976 if (PL_madskills) {
11977 tstart = SvPVX(PL_linestr) + stuffstart;
11979 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11981 PL_thisstuff = newSVpvn(tstart, s - tstart);
11984 s += SvCUR(herewas);
11987 stuffstart = s - SvPVX(PL_linestr);
11993 tmpstr = newSV_type(SVt_PVIV);
11994 SvGROW(tmpstr, 80);
11995 if (term == '\'') {
11996 op_type = OP_CONST;
11997 SvIV_set(tmpstr, -1);
11999 else if (term == '`') {
12000 op_type = OP_BACKTICK;
12001 SvIV_set(tmpstr, '\\');
12005 PL_multi_start = CopLINE(PL_curcop);
12006 PL_multi_open = PL_multi_close = '<';
12007 term = *PL_tokenbuf;
12008 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
12009 char * const bufptr = PL_sublex_info.super_bufptr;
12010 char * const bufend = PL_sublex_info.super_bufend;
12011 char * const olds = s - SvCUR(herewas);
12012 s = strchr(bufptr, '\n');
12016 while (s < bufend &&
12017 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
12019 CopLINE_inc(PL_curcop);
12022 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12023 missingterm(PL_tokenbuf);
12025 sv_setpvn(herewas,bufptr,d-bufptr+1);
12026 sv_setpvn(tmpstr,d+1,s-d);
12028 sv_catpvn(herewas,s,bufend-s);
12029 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
12036 while (s < PL_bufend &&
12037 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
12039 CopLINE_inc(PL_curcop);
12041 if (s >= PL_bufend) {
12042 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12043 missingterm(PL_tokenbuf);
12045 sv_setpvn(tmpstr,d+1,s-d);
12047 if (PL_madskills) {
12049 sv_catpvn(PL_thisstuff, d + 1, s - d);
12051 PL_thisstuff = newSVpvn(d + 1, s - d);
12052 stuffstart = s - SvPVX(PL_linestr);
12056 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
12058 sv_catpvn(herewas,s,PL_bufend-s);
12059 sv_setsv(PL_linestr,herewas);
12060 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
12061 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12062 PL_last_lop = PL_last_uni = NULL;
12065 sv_setpvs(tmpstr,""); /* avoid "uninitialized" warning */
12066 while (s >= PL_bufend) { /* multiple line string? */
12068 if (PL_madskills) {
12069 tstart = SvPVX(PL_linestr) + stuffstart;
12071 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
12073 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
12077 CopLINE_inc(PL_curcop);
12078 if (!outer || !lex_next_chunk(0)) {
12079 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12080 missingterm(PL_tokenbuf);
12082 CopLINE_dec(PL_curcop);
12085 stuffstart = s - SvPVX(PL_linestr);
12087 CopLINE_inc(PL_curcop);
12088 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12089 PL_last_lop = PL_last_uni = NULL;
12090 #ifndef PERL_STRICT_CR
12091 if (PL_bufend - PL_linestart >= 2) {
12092 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
12093 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
12095 PL_bufend[-2] = '\n';
12097 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
12099 else if (PL_bufend[-1] == '\r')
12100 PL_bufend[-1] = '\n';
12102 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
12103 PL_bufend[-1] = '\n';
12105 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
12106 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
12107 *(SvPVX(PL_linestr) + off ) = ' ';
12108 sv_catsv(PL_linestr,herewas);
12109 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12110 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
12114 sv_catsv(tmpstr,PL_linestr);
12119 PL_multi_end = CopLINE(PL_curcop);
12120 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
12121 SvPV_shrink_to_cur(tmpstr);
12123 SvREFCNT_dec(herewas);
12125 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
12127 else if (PL_encoding)
12128 sv_recode_to_utf8(tmpstr, PL_encoding);
12130 PL_lex_stuff = tmpstr;
12131 pl_yylval.ival = op_type;
12135 /* scan_inputsymbol
12136 takes: current position in input buffer
12137 returns: new position in input buffer
12138 side-effects: pl_yylval and lex_op are set.
12143 <FH> read from filehandle
12144 <pkg::FH> read from package qualified filehandle
12145 <pkg'FH> read from package qualified filehandle
12146 <$fh> read from filehandle in $fh
12147 <*.h> filename glob
12152 S_scan_inputsymbol(pTHX_ char *start)
12155 register char *s = start; /* current position in buffer */
12158 char *d = PL_tokenbuf; /* start of temp holding space */
12159 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
12161 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
12163 end = strchr(s, '\n');
12166 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
12168 /* die if we didn't have space for the contents of the <>,
12169 or if it didn't end, or if we see a newline
12172 if (len >= (I32)sizeof PL_tokenbuf)
12173 Perl_croak(aTHX_ "Excessively long <> operator");
12175 Perl_croak(aTHX_ "Unterminated <> operator");
12180 Remember, only scalar variables are interpreted as filehandles by
12181 this code. Anything more complex (e.g., <$fh{$num}>) will be
12182 treated as a glob() call.
12183 This code makes use of the fact that except for the $ at the front,
12184 a scalar variable and a filehandle look the same.
12186 if (*d == '$' && d[1]) d++;
12188 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
12189 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
12192 /* If we've tried to read what we allow filehandles to look like, and
12193 there's still text left, then it must be a glob() and not a getline.
12194 Use scan_str to pull out the stuff between the <> and treat it
12195 as nothing more than a string.
12198 if (d - PL_tokenbuf != len) {
12199 pl_yylval.ival = OP_GLOB;
12200 s = scan_str(start,!!PL_madskills,FALSE);
12202 Perl_croak(aTHX_ "Glob not terminated");
12206 bool readline_overriden = FALSE;
12209 /* we're in a filehandle read situation */
12212 /* turn <> into <ARGV> */
12214 Copy("ARGV",d,5,char);
12216 /* Check whether readline() is overriden */
12217 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
12219 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
12221 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
12222 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
12223 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
12224 readline_overriden = TRUE;
12226 /* if <$fh>, create the ops to turn the variable into a
12230 /* try to find it in the pad for this block, otherwise find
12231 add symbol table ops
12233 const PADOFFSET tmp = pad_findmy(d, len, 0);
12234 if (tmp != NOT_IN_PAD) {
12235 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
12236 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
12237 HEK * const stashname = HvNAME_HEK(stash);
12238 SV * const sym = sv_2mortal(newSVhek(stashname));
12239 sv_catpvs(sym, "::");
12240 sv_catpv(sym, d+1);
12245 OP * const o = newOP(OP_PADSV, 0);
12247 PL_lex_op = readline_overriden
12248 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
12249 append_elem(OP_LIST, o,
12250 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
12251 : (OP*)newUNOP(OP_READLINE, 0, o);
12260 ? (GV_ADDMULTI | GV_ADDINEVAL)
12263 PL_lex_op = readline_overriden
12264 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
12265 append_elem(OP_LIST,
12266 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
12267 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
12268 : (OP*)newUNOP(OP_READLINE, 0,
12269 newUNOP(OP_RV2SV, 0,
12270 newGVOP(OP_GV, 0, gv)));
12272 if (!readline_overriden)
12273 PL_lex_op->op_flags |= OPf_SPECIAL;
12274 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
12275 pl_yylval.ival = OP_NULL;
12278 /* If it's none of the above, it must be a literal filehandle
12279 (<Foo::BAR> or <FOO>) so build a simple readline OP */
12281 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
12282 PL_lex_op = readline_overriden
12283 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
12284 append_elem(OP_LIST,
12285 newGVOP(OP_GV, 0, gv),
12286 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
12287 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
12288 pl_yylval.ival = OP_NULL;
12297 takes: start position in buffer
12298 keep_quoted preserve \ on the embedded delimiter(s)
12299 keep_delims preserve the delimiters around the string
12300 returns: position to continue reading from buffer
12301 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
12302 updates the read buffer.
12304 This subroutine pulls a string out of the input. It is called for:
12305 q single quotes q(literal text)
12306 ' single quotes 'literal text'
12307 qq double quotes qq(interpolate $here please)
12308 " double quotes "interpolate $here please"
12309 qx backticks qx(/bin/ls -l)
12310 ` backticks `/bin/ls -l`
12311 qw quote words @EXPORT_OK = qw( func() $spam )
12312 m// regexp match m/this/
12313 s/// regexp substitute s/this/that/
12314 tr/// string transliterate tr/this/that/
12315 y/// string transliterate y/this/that/
12316 ($*@) sub prototypes sub foo ($)
12317 (stuff) sub attr parameters sub foo : attr(stuff)
12318 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
12320 In most of these cases (all but <>, patterns and transliterate)
12321 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
12322 calls scan_str(). s/// makes yylex() call scan_subst() which calls
12323 scan_str(). tr/// and y/// make yylex() call scan_trans() which
12326 It skips whitespace before the string starts, and treats the first
12327 character as the delimiter. If the delimiter is one of ([{< then
12328 the corresponding "close" character )]}> is used as the closing
12329 delimiter. It allows quoting of delimiters, and if the string has
12330 balanced delimiters ([{<>}]) it allows nesting.
12332 On success, the SV with the resulting string is put into lex_stuff or,
12333 if that is already non-NULL, into lex_repl. The second case occurs only
12334 when parsing the RHS of the special constructs s/// and tr/// (y///).
12335 For convenience, the terminating delimiter character is stuffed into
12340 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
12343 SV *sv; /* scalar value: string */
12344 const char *tmps; /* temp string, used for delimiter matching */
12345 register char *s = start; /* current position in the buffer */
12346 register char term; /* terminating character */
12347 register char *to; /* current position in the sv's data */
12348 I32 brackets = 1; /* bracket nesting level */
12349 bool has_utf8 = FALSE; /* is there any utf8 content? */
12350 I32 termcode; /* terminating char. code */
12351 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
12352 STRLEN termlen; /* length of terminating string */
12353 int last_off = 0; /* last position for nesting bracket */
12359 PERL_ARGS_ASSERT_SCAN_STR;
12361 /* skip space before the delimiter */
12367 if (PL_realtokenstart >= 0) {
12368 stuffstart = PL_realtokenstart;
12369 PL_realtokenstart = -1;
12372 stuffstart = start - SvPVX(PL_linestr);
12374 /* mark where we are, in case we need to report errors */
12377 /* after skipping whitespace, the next character is the terminator */
12380 termcode = termstr[0] = term;
12384 termcode = utf8_to_uvchr((U8*)s, &termlen);
12385 Copy(s, termstr, termlen, U8);
12386 if (!UTF8_IS_INVARIANT(term))
12390 /* mark where we are */
12391 PL_multi_start = CopLINE(PL_curcop);
12392 PL_multi_open = term;
12394 /* find corresponding closing delimiter */
12395 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
12396 termcode = termstr[0] = term = tmps[5];
12398 PL_multi_close = term;
12400 /* create a new SV to hold the contents. 79 is the SV's initial length.
12401 What a random number. */
12402 sv = newSV_type(SVt_PVIV);
12404 SvIV_set(sv, termcode);
12405 (void)SvPOK_only(sv); /* validate pointer */
12407 /* move past delimiter and try to read a complete string */
12409 sv_catpvn(sv, s, termlen);
12412 tstart = SvPVX(PL_linestr) + stuffstart;
12413 if (!PL_thisopen && !keep_delims) {
12414 PL_thisopen = newSVpvn(tstart, s - tstart);
12415 stuffstart = s - SvPVX(PL_linestr);
12419 if (PL_encoding && !UTF) {
12423 int offset = s - SvPVX_const(PL_linestr);
12424 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
12425 &offset, (char*)termstr, termlen);
12426 const char * const ns = SvPVX_const(PL_linestr) + offset;
12427 char * const svlast = SvEND(sv) - 1;
12429 for (; s < ns; s++) {
12430 if (*s == '\n' && !PL_rsfp)
12431 CopLINE_inc(PL_curcop);
12434 goto read_more_line;
12436 /* handle quoted delimiters */
12437 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
12439 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
12441 if ((svlast-1 - t) % 2) {
12442 if (!keep_quoted) {
12443 *(svlast-1) = term;
12445 SvCUR_set(sv, SvCUR(sv) - 1);
12450 if (PL_multi_open == PL_multi_close) {
12456 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
12457 /* At here, all closes are "was quoted" one,
12458 so we don't check PL_multi_close. */
12460 if (!keep_quoted && *(t+1) == PL_multi_open)
12465 else if (*t == PL_multi_open)
12473 SvCUR_set(sv, w - SvPVX_const(sv));
12475 last_off = w - SvPVX(sv);
12476 if (--brackets <= 0)
12481 if (!keep_delims) {
12482 SvCUR_set(sv, SvCUR(sv) - 1);
12488 /* extend sv if need be */
12489 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
12490 /* set 'to' to the next character in the sv's string */
12491 to = SvPVX(sv)+SvCUR(sv);
12493 /* if open delimiter is the close delimiter read unbridle */
12494 if (PL_multi_open == PL_multi_close) {
12495 for (; s < PL_bufend; s++,to++) {
12496 /* embedded newlines increment the current line number */
12497 if (*s == '\n' && !PL_rsfp)
12498 CopLINE_inc(PL_curcop);
12499 /* handle quoted delimiters */
12500 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
12501 if (!keep_quoted && s[1] == term)
12503 /* any other quotes are simply copied straight through */
12507 /* terminate when run out of buffer (the for() condition), or
12508 have found the terminator */
12509 else if (*s == term) {
12512 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
12515 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
12521 /* if the terminator isn't the same as the start character (e.g.,
12522 matched brackets), we have to allow more in the quoting, and
12523 be prepared for nested brackets.
12526 /* read until we run out of string, or we find the terminator */
12527 for (; s < PL_bufend; s++,to++) {
12528 /* embedded newlines increment the line count */
12529 if (*s == '\n' && !PL_rsfp)
12530 CopLINE_inc(PL_curcop);
12531 /* backslashes can escape the open or closing characters */
12532 if (*s == '\\' && s+1 < PL_bufend) {
12533 if (!keep_quoted &&
12534 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
12539 /* allow nested opens and closes */
12540 else if (*s == PL_multi_close && --brackets <= 0)
12542 else if (*s == PL_multi_open)
12544 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
12549 /* terminate the copied string and update the sv's end-of-string */
12551 SvCUR_set(sv, to - SvPVX_const(sv));
12554 * this next chunk reads more into the buffer if we're not done yet
12558 break; /* handle case where we are done yet :-) */
12560 #ifndef PERL_STRICT_CR
12561 if (to - SvPVX_const(sv) >= 2) {
12562 if ((to[-2] == '\r' && to[-1] == '\n') ||
12563 (to[-2] == '\n' && to[-1] == '\r'))
12567 SvCUR_set(sv, to - SvPVX_const(sv));
12569 else if (to[-1] == '\r')
12572 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
12577 /* if we're out of file, or a read fails, bail and reset the current
12578 line marker so we can report where the unterminated string began
12581 if (PL_madskills) {
12582 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12584 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
12586 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
12589 CopLINE_inc(PL_curcop);
12590 PL_bufptr = PL_bufend;
12591 if (!lex_next_chunk(0)) {
12593 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
12602 /* at this point, we have successfully read the delimited string */
12604 if (!PL_encoding || UTF) {
12606 if (PL_madskills) {
12607 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12608 const int len = s - tstart;
12610 sv_catpvn(PL_thisstuff, tstart, len);
12612 PL_thisstuff = newSVpvn(tstart, len);
12613 if (!PL_thisclose && !keep_delims)
12614 PL_thisclose = newSVpvn(s,termlen);
12619 sv_catpvn(sv, s, termlen);
12624 if (PL_madskills) {
12625 char * const tstart = SvPVX(PL_linestr) + stuffstart;
12626 const int len = s - tstart - termlen;
12628 sv_catpvn(PL_thisstuff, tstart, len);
12630 PL_thisstuff = newSVpvn(tstart, len);
12631 if (!PL_thisclose && !keep_delims)
12632 PL_thisclose = newSVpvn(s - termlen,termlen);
12636 if (has_utf8 || PL_encoding)
12639 PL_multi_end = CopLINE(PL_curcop);
12641 /* if we allocated too much space, give some back */
12642 if (SvCUR(sv) + 5 < SvLEN(sv)) {
12643 SvLEN_set(sv, SvCUR(sv) + 1);
12644 SvPV_renew(sv, SvLEN(sv));
12647 /* decide whether this is the first or second quoted string we've read
12660 takes: pointer to position in buffer
12661 returns: pointer to new position in buffer
12662 side-effects: builds ops for the constant in pl_yylval.op
12664 Read a number in any of the formats that Perl accepts:
12666 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12667 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12670 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12672 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12675 If it reads a number without a decimal point or an exponent, it will
12676 try converting the number to an integer and see if it can do so
12677 without loss of precision.
12681 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12684 register const char *s = start; /* current position in buffer */
12685 register char *d; /* destination in temp buffer */
12686 register char *e; /* end of temp buffer */
12687 NV nv; /* number read, as a double */
12688 SV *sv = NULL; /* place to put the converted number */
12689 bool floatit; /* boolean: int or float? */
12690 const char *lastub = NULL; /* position of last underbar */
12691 static char const number_too_long[] = "Number too long";
12693 PERL_ARGS_ASSERT_SCAN_NUM;
12695 /* We use the first character to decide what type of number this is */
12699 Perl_croak(aTHX_ "panic: scan_num");
12701 /* if it starts with a 0, it could be an octal number, a decimal in
12702 0.13 disguise, or a hexadecimal number, or a binary number. */
12706 u holds the "number so far"
12707 shift the power of 2 of the base
12708 (hex == 4, octal == 3, binary == 1)
12709 overflowed was the number more than we can hold?
12711 Shift is used when we add a digit. It also serves as an "are
12712 we in octal/hex/binary?" indicator to disallow hex characters
12713 when in octal mode.
12718 bool overflowed = FALSE;
12719 bool just_zero = TRUE; /* just plain 0 or binary number? */
12720 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12721 static const char* const bases[5] =
12722 { "", "binary", "", "octal", "hexadecimal" };
12723 static const char* const Bases[5] =
12724 { "", "Binary", "", "Octal", "Hexadecimal" };
12725 static const char* const maxima[5] =
12727 "0b11111111111111111111111111111111",
12731 const char *base, *Base, *max;
12733 /* check for hex */
12738 } else if (s[1] == 'b') {
12743 /* check for a decimal in disguise */
12744 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12746 /* so it must be octal */
12753 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12754 "Misplaced _ in number");
12758 base = bases[shift];
12759 Base = Bases[shift];
12760 max = maxima[shift];
12762 /* read the rest of the number */
12764 /* x is used in the overflow test,
12765 b is the digit we're adding on. */
12770 /* if we don't mention it, we're done */
12774 /* _ are ignored -- but warned about if consecutive */
12776 if (lastub && s == lastub + 1)
12777 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12778 "Misplaced _ in number");
12782 /* 8 and 9 are not octal */
12783 case '8': case '9':
12785 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12789 case '2': case '3': case '4':
12790 case '5': case '6': case '7':
12792 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12795 case '0': case '1':
12796 b = *s++ & 15; /* ASCII digit -> value of digit */
12800 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12801 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12802 /* make sure they said 0x */
12805 b = (*s++ & 7) + 9;
12807 /* Prepare to put the digit we have onto the end
12808 of the number so far. We check for overflows.
12814 x = u << shift; /* make room for the digit */
12816 if ((x >> shift) != u
12817 && !(PL_hints & HINT_NEW_BINARY)) {
12820 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
12821 "Integer overflow in %s number",
12824 u = x | b; /* add the digit to the end */
12827 n *= nvshift[shift];
12828 /* If an NV has not enough bits in its
12829 * mantissa to represent an UV this summing of
12830 * small low-order numbers is a waste of time
12831 * (because the NV cannot preserve the
12832 * low-order bits anyway): we could just
12833 * remember when did we overflow and in the
12834 * end just multiply n by the right
12842 /* if we get here, we had success: make a scalar value from
12847 /* final misplaced underbar check */
12848 if (s[-1] == '_') {
12849 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12854 if (n > 4294967295.0)
12855 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
12856 "%s number > %s non-portable",
12862 if (u > 0xffffffff)
12863 Perl_ck_warner(aTHX_ packWARN(WARN_PORTABLE),
12864 "%s number > %s non-portable",
12869 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12870 sv = new_constant(start, s - start, "integer",
12871 sv, NULL, NULL, 0);
12872 else if (PL_hints & HINT_NEW_BINARY)
12873 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12878 handle decimal numbers.
12879 we're also sent here when we read a 0 as the first digit
12881 case '1': case '2': case '3': case '4': case '5':
12882 case '6': case '7': case '8': case '9': case '.':
12885 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12888 /* read next group of digits and _ and copy into d */
12889 while (isDIGIT(*s) || *s == '_') {
12890 /* skip underscores, checking for misplaced ones
12894 if (lastub && s == lastub + 1)
12895 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12896 "Misplaced _ in number");
12900 /* check for end of fixed-length buffer */
12902 Perl_croak(aTHX_ number_too_long);
12903 /* if we're ok, copy the character */
12908 /* final misplaced underbar check */
12909 if (lastub && s == lastub + 1) {
12910 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12913 /* read a decimal portion if there is one. avoid
12914 3..5 being interpreted as the number 3. followed
12917 if (*s == '.' && s[1] != '.') {
12922 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12923 "Misplaced _ in number");
12927 /* copy, ignoring underbars, until we run out of digits.
12929 for (; isDIGIT(*s) || *s == '_'; s++) {
12930 /* fixed length buffer check */
12932 Perl_croak(aTHX_ number_too_long);
12934 if (lastub && s == lastub + 1)
12935 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12936 "Misplaced _ in number");
12942 /* fractional part ending in underbar? */
12943 if (s[-1] == '_') {
12944 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12945 "Misplaced _ in number");
12947 if (*s == '.' && isDIGIT(s[1])) {
12948 /* oops, it's really a v-string, but without the "v" */
12954 /* read exponent part, if present */
12955 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12959 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12960 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12962 /* stray preinitial _ */
12964 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12965 "Misplaced _ in number");
12969 /* allow positive or negative exponent */
12970 if (*s == '+' || *s == '-')
12973 /* stray initial _ */
12975 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12976 "Misplaced _ in number");
12980 /* read digits of exponent */
12981 while (isDIGIT(*s) || *s == '_') {
12984 Perl_croak(aTHX_ number_too_long);
12988 if (((lastub && s == lastub + 1) ||
12989 (!isDIGIT(s[1]) && s[1] != '_')))
12990 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX),
12991 "Misplaced _ in number");
12998 /* make an sv from the string */
13002 We try to do an integer conversion first if no characters
13003 indicating "float" have been found.
13008 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
13010 if (flags == IS_NUMBER_IN_UV) {
13012 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
13015 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
13016 if (uv <= (UV) IV_MIN)
13017 sv_setiv(sv, -(IV)uv);
13024 /* terminate the string */
13026 nv = Atof(PL_tokenbuf);
13031 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
13032 const char *const key = floatit ? "float" : "integer";
13033 const STRLEN keylen = floatit ? 5 : 7;
13034 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
13035 key, keylen, sv, NULL, NULL, 0);
13039 /* if it starts with a v, it could be a v-string */
13042 sv = newSV(5); /* preallocate storage space */
13043 s = scan_vstring(s, PL_bufend, sv);
13047 /* make the op for the constant and return */
13050 lvalp->opval = newSVOP(OP_CONST, 0, sv);
13052 lvalp->opval = NULL;
13058 S_scan_formline(pTHX_ register char *s)
13061 register char *eol;
13063 SV * const stuff = newSVpvs("");
13064 bool needargs = FALSE;
13065 bool eofmt = FALSE;
13067 char *tokenstart = s;
13068 SV* savewhite = NULL;
13070 if (PL_madskills) {
13071 savewhite = PL_thiswhite;
13076 PERL_ARGS_ASSERT_SCAN_FORMLINE;
13078 while (!needargs) {
13081 #ifdef PERL_STRICT_CR
13082 while (SPACE_OR_TAB(*t))
13085 while (SPACE_OR_TAB(*t) || *t == '\r')
13088 if (*t == '\n' || t == PL_bufend) {
13093 if (PL_in_eval && !PL_rsfp) {
13094 eol = (char *) memchr(s,'\n',PL_bufend-s);
13099 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
13101 for (t = s; t < eol; t++) {
13102 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
13104 goto enough; /* ~~ must be first line in formline */
13106 if (*t == '@' || *t == '^')
13110 sv_catpvn(stuff, s, eol-s);
13111 #ifndef PERL_STRICT_CR
13112 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
13113 char *end = SvPVX(stuff) + SvCUR(stuff);
13116 SvCUR_set(stuff, SvCUR(stuff) - 1);
13127 if (PL_madskills) {
13129 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
13131 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
13134 PL_bufptr = PL_bufend;
13135 CopLINE_inc(PL_curcop);
13136 got_some = lex_next_chunk(0);
13137 CopLINE_dec(PL_curcop);
13140 tokenstart = PL_bufptr;
13148 if (SvCUR(stuff)) {
13151 PL_lex_state = LEX_NORMAL;
13152 start_force(PL_curforce);
13153 NEXTVAL_NEXTTOKE.ival = 0;
13157 PL_lex_state = LEX_FORMLINE;
13159 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
13161 else if (PL_encoding)
13162 sv_recode_to_utf8(stuff, PL_encoding);
13164 start_force(PL_curforce);
13165 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
13167 start_force(PL_curforce);
13168 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
13172 SvREFCNT_dec(stuff);
13174 PL_lex_formbrack = 0;
13178 if (PL_madskills) {
13180 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
13182 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
13183 PL_thiswhite = savewhite;
13190 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
13193 const I32 oldsavestack_ix = PL_savestack_ix;
13194 CV* const outsidecv = PL_compcv;
13197 assert(SvTYPE(PL_compcv) == SVt_PVCV);
13199 SAVEI32(PL_subline);
13200 save_item(PL_subname);
13201 SAVESPTR(PL_compcv);
13203 PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
13204 CvFLAGS(PL_compcv) |= flags;
13206 PL_subline = CopLINE(PL_curcop);
13207 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
13208 CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
13209 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
13211 return oldsavestack_ix;
13215 #pragma segment Perl_yylex
13218 S_yywarn(pTHX_ const char *const s)
13222 PERL_ARGS_ASSERT_YYWARN;
13224 PL_in_eval |= EVAL_WARNONLY;
13226 PL_in_eval &= ~EVAL_WARNONLY;
13231 Perl_yyerror(pTHX_ const char *const s)
13234 const char *where = NULL;
13235 const char *context = NULL;
13238 int yychar = PL_parser->yychar;
13240 PERL_ARGS_ASSERT_YYERROR;
13242 if (!yychar || (yychar == ';' && !PL_rsfp))
13244 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
13245 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
13246 PL_oldbufptr != PL_bufptr) {
13249 The code below is removed for NetWare because it abends/crashes on NetWare
13250 when the script has error such as not having the closing quotes like:
13251 if ($var eq "value)
13252 Checking of white spaces is anyway done in NetWare code.
13255 while (isSPACE(*PL_oldoldbufptr))
13258 context = PL_oldoldbufptr;
13259 contlen = PL_bufptr - PL_oldoldbufptr;
13261 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
13262 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
13265 The code below is removed for NetWare because it abends/crashes on NetWare
13266 when the script has error such as not having the closing quotes like:
13267 if ($var eq "value)
13268 Checking of white spaces is anyway done in NetWare code.
13271 while (isSPACE(*PL_oldbufptr))
13274 context = PL_oldbufptr;
13275 contlen = PL_bufptr - PL_oldbufptr;
13277 else if (yychar > 255)
13278 where = "next token ???";
13279 else if (yychar == -2) { /* YYEMPTY */
13280 if (PL_lex_state == LEX_NORMAL ||
13281 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
13282 where = "at end of line";
13283 else if (PL_lex_inpat)
13284 where = "within pattern";
13286 where = "within string";
13289 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
13291 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
13292 else if (isPRINT_LC(yychar)) {
13293 const char string = yychar;
13294 sv_catpvn(where_sv, &string, 1);
13297 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
13298 where = SvPVX_const(where_sv);
13300 msg = sv_2mortal(newSVpv(s, 0));
13301 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
13302 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
13304 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
13306 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
13307 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
13308 Perl_sv_catpvf(aTHX_ msg,
13309 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
13310 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
13313 if (PL_in_eval & EVAL_WARNONLY) {
13314 Perl_ck_warner_d(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
13318 if (PL_error_count >= 10) {
13319 if (PL_in_eval && SvCUR(ERRSV))
13320 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
13321 SVfARG(ERRSV), OutCopFILE(PL_curcop));
13323 Perl_croak(aTHX_ "%s has too many errors.\n",
13324 OutCopFILE(PL_curcop));
13327 PL_in_my_stash = NULL;
13331 #pragma segment Main
13335 S_swallow_bom(pTHX_ U8 *s)
13338 const STRLEN slen = SvCUR(PL_linestr);
13340 PERL_ARGS_ASSERT_SWALLOW_BOM;
13344 if (s[1] == 0xFE) {
13345 /* UTF-16 little-endian? (or UTF-32LE?) */
13346 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
13347 Perl_croak(aTHX_ "Unsupported script encoding UTF-32LE");
13348 #ifndef PERL_NO_UTF16_FILTER
13349 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (BOM)\n");
13351 if (PL_bufend > (char*)s) {
13352 s = add_utf16_textfilter(s, TRUE);
13355 Perl_croak(aTHX_ "Unsupported script encoding UTF-16LE");
13360 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
13361 #ifndef PERL_NO_UTF16_FILTER
13362 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
13364 if (PL_bufend > (char *)s) {
13365 s = add_utf16_textfilter(s, FALSE);
13368 Perl_croak(aTHX_ "Unsupported script encoding UTF-16BE");
13373 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
13374 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
13375 s += 3; /* UTF-8 */
13381 if (s[2] == 0xFE && s[3] == 0xFF) {
13382 /* UTF-32 big-endian */
13383 Perl_croak(aTHX_ "Unsupported script encoding UTF-32BE");
13386 else if (s[2] == 0 && s[3] != 0) {
13389 * are a good indicator of UTF-16BE. */
13390 #ifndef PERL_NO_UTF16_FILTER
13391 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
13392 s = add_utf16_textfilter(s, FALSE);
13394 Perl_croak(aTHX_ "Unsupported script encoding UTF-16BE");
13400 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
13401 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
13402 s += 4; /* UTF-8 */
13408 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
13411 * are a good indicator of UTF-16LE. */
13412 #ifndef PERL_NO_UTF16_FILTER
13413 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
13414 s = add_utf16_textfilter(s, TRUE);
13416 Perl_croak(aTHX_ "Unsupported script encoding UTF-16LE");
13424 #ifndef PERL_NO_UTF16_FILTER
13426 S_utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
13429 SV *const filter = FILTER_DATA(idx);
13430 /* We re-use this each time round, throwing the contents away before we
13432 SV *const utf16_buffer = MUTABLE_SV(IoTOP_GV(filter));
13433 SV *const utf8_buffer = filter;
13434 IV status = IoPAGE(filter);
13435 const bool reverse = (bool) IoLINES(filter);
13438 /* As we're automatically added, at the lowest level, and hence only called
13439 from this file, we can be sure that we're not called in block mode. Hence
13440 don't bother writing code to deal with block mode. */
13442 Perl_croak(aTHX_ "panic: utf16_textfilter called in block mode (for %d characters)", maxlen);
13445 Perl_croak(aTHX_ "panic: utf16_textfilter called after error (status=%"IVdf")", status);
13447 DEBUG_P(PerlIO_printf(Perl_debug_log,
13448 "utf16_textfilter(%p,%ce): idx=%d maxlen=%d status=%"IVdf" utf16=%"UVuf" utf8=%"UVuf"\n",
13449 FPTR2DPTR(void *, S_utf16_textfilter),
13450 reverse ? 'l' : 'b', idx, maxlen, status,
13451 (UV)SvCUR(utf16_buffer), (UV)SvCUR(utf8_buffer)));
13458 /* First, look in our buffer of existing UTF-8 data: */
13459 char *nl = (char *)memchr(SvPVX(utf8_buffer), '\n', SvCUR(utf8_buffer));
13463 } else if (status == 0) {
13465 IoPAGE(filter) = 0;
13466 nl = SvEND(utf8_buffer);
13469 STRLEN got = nl - SvPVX(utf8_buffer);
13470 /* Did we have anything to append? */
13472 sv_catpvn(sv, SvPVX(utf8_buffer), got);
13473 /* Everything else in this code works just fine if SVp_POK isn't
13474 set. This, however, needs it, and we need it to work, else
13475 we loop infinitely because the buffer is never consumed. */
13476 sv_chop(utf8_buffer, nl);
13480 /* OK, not a complete line there, so need to read some more UTF-16.
13481 Read an extra octect if the buffer currently has an odd number. */
13485 if (SvCUR(utf16_buffer) >= 2) {
13486 /* Location of the high octet of the last complete code point.
13487 Gosh, UTF-16 is a pain. All the benefits of variable length,
13488 *coupled* with all the benefits of partial reads and
13490 const U8 *const last_hi = (U8*)SvPVX(utf16_buffer)
13491 + ((SvCUR(utf16_buffer) & ~1) - (reverse ? 1 : 2));
13493 if (*last_hi < 0xd8 || *last_hi > 0xdb) {
13497 /* We have the first half of a surrogate. Read more. */
13498 DEBUG_P(PerlIO_printf(Perl_debug_log, "utf16_textfilter partial surrogate detected at %p\n", last_hi));
13501 status = FILTER_READ(idx + 1, utf16_buffer,
13502 160 + (SvCUR(utf16_buffer) & 1));
13503 DEBUG_P(PerlIO_printf(Perl_debug_log, "utf16_textfilter status=%"IVdf" SvCUR(sv)=%"UVuf"\n", status, (UV)SvCUR(utf16_buffer)));
13504 DEBUG_P({ sv_dump(utf16_buffer); sv_dump(utf8_buffer);});
13507 IoPAGE(filter) = status;
13512 chars = SvCUR(utf16_buffer) >> 1;
13513 have = SvCUR(utf8_buffer);
13514 SvGROW(utf8_buffer, have + chars * 3 + 1);
13517 end = utf16_to_utf8_reversed((U8*)SvPVX(utf16_buffer),
13518 (U8*)SvPVX_const(utf8_buffer) + have,
13519 chars * 2, &newlen);
13521 end = utf16_to_utf8((U8*)SvPVX(utf16_buffer),
13522 (U8*)SvPVX_const(utf8_buffer) + have,
13523 chars * 2, &newlen);
13525 SvCUR_set(utf8_buffer, have + newlen);
13528 /* No need to keep this SV "well-formed" with a '\0' after the end, as
13529 it's private to us, and utf16_to_utf8{,reversed} take a
13530 (pointer,length) pair, rather than a NUL-terminated string. */
13531 if(SvCUR(utf16_buffer) & 1) {
13532 *SvPVX(utf16_buffer) = SvEND(utf16_buffer)[-1];
13533 SvCUR_set(utf16_buffer, 1);
13535 SvCUR_set(utf16_buffer, 0);
13538 DEBUG_P(PerlIO_printf(Perl_debug_log,
13539 "utf16_textfilter: returns, status=%"IVdf" utf16=%"UVuf" utf8=%"UVuf"\n",
13541 (UV)SvCUR(utf16_buffer), (UV)SvCUR(utf8_buffer)));
13542 DEBUG_P({ sv_dump(utf8_buffer); sv_dump(sv);});
13547 S_add_utf16_textfilter(pTHX_ U8 *const s, bool reversed)
13549 SV *filter = filter_add(S_utf16_textfilter, NULL);
13551 IoTOP_GV(filter) = MUTABLE_GV(newSVpvn((char *)s, PL_bufend - (char*)s));
13552 sv_setpvs(filter, "");
13553 IoLINES(filter) = reversed;
13554 IoPAGE(filter) = 1; /* Not EOF */
13556 /* Sadly, we have to return a valid pointer, come what may, so we have to
13557 ignore any error return from this. */
13558 SvCUR_set(PL_linestr, 0);
13559 if (FILTER_READ(0, PL_linestr, 0)) {
13560 SvUTF8_on(PL_linestr);
13562 SvUTF8_on(PL_linestr);
13564 PL_bufend = SvEND(PL_linestr);
13565 return (U8*)SvPVX(PL_linestr);
13570 Returns a pointer to the next character after the parsed
13571 vstring, as well as updating the passed in sv.
13573 Function must be called like
13576 s = scan_vstring(s,e,sv);
13578 where s and e are the start and end of the string.
13579 The sv should already be large enough to store the vstring
13580 passed in, for performance reasons.
13585 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
13588 const char *pos = s;
13589 const char *start = s;
13591 PERL_ARGS_ASSERT_SCAN_VSTRING;
13593 if (*pos == 'v') pos++; /* get past 'v' */
13594 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
13596 if ( *pos != '.') {
13597 /* this may not be a v-string if followed by => */
13598 const char *next = pos;
13599 while (next < e && isSPACE(*next))
13601 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
13602 /* return string not v-string */
13603 sv_setpvn(sv,(char *)s,pos-s);
13604 return (char *)pos;
13608 if (!isALPHA(*pos)) {
13609 U8 tmpbuf[UTF8_MAXBYTES+1];
13612 s++; /* get past 'v' */
13617 /* this is atoi() that tolerates underscores */
13620 const char *end = pos;
13622 while (--end >= s) {
13624 const UV orev = rev;
13625 rev += (*end - '0') * mult;
13628 Perl_ck_warner_d(aTHX_ packWARN(WARN_OVERFLOW),
13629 "Integer overflow in decimal number");
13633 if (rev > 0x7FFFFFFF)
13634 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
13636 /* Append native character for the rev point */
13637 tmpend = uvchr_to_utf8(tmpbuf, rev);
13638 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
13639 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
13641 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
13647 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
13651 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
13658 Perl_keyword_plugin_standard(pTHX_
13659 char *keyword_ptr, STRLEN keyword_len, OP **op_ptr)
13661 PERL_ARGS_ASSERT_KEYWORD_PLUGIN_STANDARD;
13662 PERL_UNUSED_CONTEXT;
13663 PERL_UNUSED_ARG(keyword_ptr);
13664 PERL_UNUSED_ARG(keyword_len);
13665 PERL_UNUSED_ARG(op_ptr);
13666 return KEYWORD_PLUGIN_DECLINE;
13671 * c-indentation-style: bsd
13672 * c-basic-offset: 4
13673 * indent-tabs-mode: t
13676 * ex: set ts=8 sts=4 sw=4 noet: