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 #define PERL_IN_TOKE_C
28 #define new_constant(a,b,c,d,e,f,g) \
29 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
31 #define pl_yylval (PL_parser->yylval)
33 /* YYINITDEPTH -- initial size of the parser's stacks. */
34 #define YYINITDEPTH 200
36 /* XXX temporary backwards compatibility */
37 #define PL_lex_brackets (PL_parser->lex_brackets)
38 #define PL_lex_brackstack (PL_parser->lex_brackstack)
39 #define PL_lex_casemods (PL_parser->lex_casemods)
40 #define PL_lex_casestack (PL_parser->lex_casestack)
41 #define PL_lex_defer (PL_parser->lex_defer)
42 #define PL_lex_dojoin (PL_parser->lex_dojoin)
43 #define PL_lex_expect (PL_parser->lex_expect)
44 #define PL_lex_formbrack (PL_parser->lex_formbrack)
45 #define PL_lex_inpat (PL_parser->lex_inpat)
46 #define PL_lex_inwhat (PL_parser->lex_inwhat)
47 #define PL_lex_op (PL_parser->lex_op)
48 #define PL_lex_repl (PL_parser->lex_repl)
49 #define PL_lex_starts (PL_parser->lex_starts)
50 #define PL_lex_stuff (PL_parser->lex_stuff)
51 #define PL_multi_start (PL_parser->multi_start)
52 #define PL_multi_open (PL_parser->multi_open)
53 #define PL_multi_close (PL_parser->multi_close)
54 #define PL_pending_ident (PL_parser->pending_ident)
55 #define PL_preambled (PL_parser->preambled)
56 #define PL_sublex_info (PL_parser->sublex_info)
57 #define PL_linestr (PL_parser->linestr)
58 #define PL_expect (PL_parser->expect)
59 #define PL_copline (PL_parser->copline)
60 #define PL_bufptr (PL_parser->bufptr)
61 #define PL_oldbufptr (PL_parser->oldbufptr)
62 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
63 #define PL_linestart (PL_parser->linestart)
64 #define PL_bufend (PL_parser->bufend)
65 #define PL_last_uni (PL_parser->last_uni)
66 #define PL_last_lop (PL_parser->last_lop)
67 #define PL_last_lop_op (PL_parser->last_lop_op)
68 #define PL_lex_state (PL_parser->lex_state)
69 #define PL_rsfp (PL_parser->rsfp)
70 #define PL_rsfp_filters (PL_parser->rsfp_filters)
71 #define PL_in_my (PL_parser->in_my)
72 #define PL_in_my_stash (PL_parser->in_my_stash)
73 #define PL_tokenbuf (PL_parser->tokenbuf)
74 #define PL_multi_end (PL_parser->multi_end)
75 #define PL_error_count (PL_parser->error_count)
78 # define PL_endwhite (PL_parser->endwhite)
79 # define PL_faketokens (PL_parser->faketokens)
80 # define PL_lasttoke (PL_parser->lasttoke)
81 # define PL_nextwhite (PL_parser->nextwhite)
82 # define PL_realtokenstart (PL_parser->realtokenstart)
83 # define PL_skipwhite (PL_parser->skipwhite)
84 # define PL_thisclose (PL_parser->thisclose)
85 # define PL_thismad (PL_parser->thismad)
86 # define PL_thisopen (PL_parser->thisopen)
87 # define PL_thisstuff (PL_parser->thisstuff)
88 # define PL_thistoken (PL_parser->thistoken)
89 # define PL_thiswhite (PL_parser->thiswhite)
90 # define PL_thiswhite (PL_parser->thiswhite)
91 # define PL_nexttoke (PL_parser->nexttoke)
92 # define PL_curforce (PL_parser->curforce)
94 # define PL_nexttoke (PL_parser->nexttoke)
95 # define PL_nexttype (PL_parser->nexttype)
96 # define PL_nextval (PL_parser->nextval)
100 S_pending_ident(pTHX);
102 static const char ident_too_long[] = "Identifier too long";
103 static const char commaless_variable_list[] = "comma-less variable list";
105 #ifndef PERL_NO_UTF16_FILTER
106 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
107 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
111 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
112 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
114 # define CURMAD(slot,sv)
115 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
118 #define XFAKEBRACK 128
119 #define XENUMMASK 127
121 #ifdef USE_UTF8_SCRIPTS
122 # define UTF (!IN_BYTES)
124 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
127 /* In variables named $^X, these are the legal values for X.
128 * 1999-02-27 mjd-perl-patch@plover.com */
129 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
131 /* On MacOS, respect nonbreaking spaces */
132 #ifdef MACOS_TRADITIONAL
133 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
135 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
138 /* LEX_* are values for PL_lex_state, the state of the lexer.
139 * They are arranged oddly so that the guard on the switch statement
140 * can get by with a single comparison (if the compiler is smart enough).
143 /* #define LEX_NOTPARSING 11 is done in perl.h. */
145 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
146 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
147 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
148 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
149 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
151 /* at end of code, eg "$x" followed by: */
152 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
153 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
155 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
156 string or after \E, $foo, etc */
157 #define LEX_INTERPCONST 2 /* NOT USED */
158 #define LEX_FORMLINE 1 /* expecting a format line */
159 #define LEX_KNOWNEXT 0 /* next token known; just return it */
163 static const char* const lex_state_names[] = {
182 #include "keywords.h"
184 /* CLINE is a macro that ensures PL_copline has a sane value */
189 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
192 # define SKIPSPACE0(s) skipspace0(s)
193 # define SKIPSPACE1(s) skipspace1(s)
194 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
195 # define PEEKSPACE(s) skipspace2(s,0)
197 # define SKIPSPACE0(s) skipspace(s)
198 # define SKIPSPACE1(s) skipspace(s)
199 # define SKIPSPACE2(s,tsv) skipspace(s)
200 # define PEEKSPACE(s) skipspace(s)
204 * Convenience functions to return different tokens and prime the
205 * lexer for the next token. They all take an argument.
207 * TOKEN : generic token (used for '(', DOLSHARP, etc)
208 * OPERATOR : generic operator
209 * AOPERATOR : assignment operator
210 * PREBLOCK : beginning the block after an if, while, foreach, ...
211 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
212 * PREREF : *EXPR where EXPR is not a simple identifier
213 * TERM : expression term
214 * LOOPX : loop exiting command (goto, last, dump, etc)
215 * FTST : file test operator
216 * FUN0 : zero-argument function
217 * FUN1 : not used, except for not, which isn't a UNIOP
218 * BOop : bitwise or or xor
220 * SHop : shift operator
221 * PWop : power operator
222 * PMop : pattern-matching operator
223 * Aop : addition-level operator
224 * Mop : multiplication-level operator
225 * Eop : equality-testing operator
226 * Rop : relational operator <= != gt
228 * Also see LOP and lop() below.
231 #ifdef DEBUGGING /* Serve -DT. */
232 # define REPORT(retval) tokereport((I32)retval, &pl_yylval)
234 # define REPORT(retval) (retval)
237 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
238 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
239 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
240 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
241 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
242 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
243 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
244 #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
245 #define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
246 #define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
247 #define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
248 #define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
249 #define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
250 #define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
251 #define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
252 #define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
253 #define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
254 #define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
255 #define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
256 #define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
258 /* This bit of chicanery makes a unary function followed by
259 * a parenthesis into a function with one argument, highest precedence.
260 * The UNIDOR macro is for unary functions that can be followed by the //
261 * operator (such as C<shift // 0>).
263 #define UNI2(f,x) { \
264 pl_yylval.ival = f; \
267 PL_last_uni = PL_oldbufptr; \
268 PL_last_lop_op = f; \
270 return REPORT( (int)FUNC1 ); \
272 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
274 #define UNI(f) UNI2(f,XTERM)
275 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
277 #define UNIBRACK(f) { \
278 pl_yylval.ival = f; \
280 PL_last_uni = PL_oldbufptr; \
282 return REPORT( (int)FUNC1 ); \
284 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
287 /* grandfather return to old style */
288 #define OLDLOP(f) return(pl_yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
292 /* how to interpret the pl_yylval associated with the token */
296 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
302 static struct debug_tokens {
304 enum token_type type;
306 } const debug_tokens[] =
308 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
309 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
310 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
311 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
312 { ARROW, TOKENTYPE_NONE, "ARROW" },
313 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
314 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
315 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
316 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
317 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
318 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
319 { DO, TOKENTYPE_NONE, "DO" },
320 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
321 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
322 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
323 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
324 { ELSE, TOKENTYPE_NONE, "ELSE" },
325 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
326 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
327 { FOR, TOKENTYPE_IVAL, "FOR" },
328 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
329 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
330 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
331 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
332 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
333 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
334 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
335 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
336 { IF, TOKENTYPE_IVAL, "IF" },
337 { LABEL, TOKENTYPE_PVAL, "LABEL" },
338 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
339 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
340 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
341 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
342 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
343 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
344 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
345 { MY, TOKENTYPE_IVAL, "MY" },
346 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
347 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
348 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
349 { OROP, TOKENTYPE_IVAL, "OROP" },
350 { OROR, TOKENTYPE_NONE, "OROR" },
351 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
352 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
353 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
354 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
355 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
356 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
357 { PREINC, TOKENTYPE_NONE, "PREINC" },
358 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
359 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
360 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
361 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
362 { SUB, TOKENTYPE_NONE, "SUB" },
363 { THING, TOKENTYPE_OPVAL, "THING" },
364 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
365 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
366 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
367 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
368 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
369 { USE, TOKENTYPE_IVAL, "USE" },
370 { WHEN, TOKENTYPE_IVAL, "WHEN" },
371 { WHILE, TOKENTYPE_IVAL, "WHILE" },
372 { WORD, TOKENTYPE_OPVAL, "WORD" },
373 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
374 { 0, TOKENTYPE_NONE, NULL }
377 /* dump the returned token in rv, plus any optional arg in pl_yylval */
380 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
384 PERL_ARGS_ASSERT_TOKEREPORT;
387 const char *name = NULL;
388 enum token_type type = TOKENTYPE_NONE;
389 const struct debug_tokens *p;
390 SV* const report = newSVpvs("<== ");
392 for (p = debug_tokens; p->token; p++) {
393 if (p->token == (int)rv) {
400 Perl_sv_catpv(aTHX_ report, name);
401 else if ((char)rv > ' ' && (char)rv < '~')
402 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
404 sv_catpvs(report, "EOF");
406 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
409 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
412 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
414 case TOKENTYPE_OPNUM:
415 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
416 PL_op_name[lvalp->ival]);
419 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
421 case TOKENTYPE_OPVAL:
423 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
424 PL_op_name[lvalp->opval->op_type]);
425 if (lvalp->opval->op_type == OP_CONST) {
426 Perl_sv_catpvf(aTHX_ report, " %s",
427 SvPEEK(cSVOPx_sv(lvalp->opval)));
432 sv_catpvs(report, "(opval=null)");
435 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
441 /* print the buffer with suitable escapes */
444 S_printbuf(pTHX_ const char *const fmt, const char *const s)
446 SV* const tmp = newSVpvs("");
448 PERL_ARGS_ASSERT_PRINTBUF;
450 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
459 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
460 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
464 S_ao(pTHX_ int toketype)
467 if (*PL_bufptr == '=') {
469 if (toketype == ANDAND)
470 pl_yylval.ival = OP_ANDASSIGN;
471 else if (toketype == OROR)
472 pl_yylval.ival = OP_ORASSIGN;
473 else if (toketype == DORDOR)
474 pl_yylval.ival = OP_DORASSIGN;
482 * When Perl expects an operator and finds something else, no_op
483 * prints the warning. It always prints "<something> found where
484 * operator expected. It prints "Missing semicolon on previous line?"
485 * if the surprise occurs at the start of the line. "do you need to
486 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
487 * where the compiler doesn't know if foo is a method call or a function.
488 * It prints "Missing operator before end of line" if there's nothing
489 * after the missing operator, or "... before <...>" if there is something
490 * after the missing operator.
494 S_no_op(pTHX_ const char *const what, char *s)
497 char * const oldbp = PL_bufptr;
498 const bool is_first = (PL_oldbufptr == PL_linestart);
500 PERL_ARGS_ASSERT_NO_OP;
506 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
507 if (ckWARN_d(WARN_SYNTAX)) {
509 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
510 "\t(Missing semicolon on previous line?)\n");
511 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
513 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
515 if (t < PL_bufptr && isSPACE(*t))
516 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
517 "\t(Do you need to predeclare %.*s?)\n",
518 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
522 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
523 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
531 * Complain about missing quote/regexp/heredoc terminator.
532 * If it's called with NULL then it cauterizes the line buffer.
533 * If we're in a delimited string and the delimiter is a control
534 * character, it's reformatted into a two-char sequence like ^C.
539 S_missingterm(pTHX_ char *s)
545 char * const nl = strrchr(s,'\n');
549 else if (isCNTRL(PL_multi_close)) {
551 tmpbuf[1] = (char)toCTRL(PL_multi_close);
556 *tmpbuf = (char)PL_multi_close;
560 q = strchr(s,'"') ? '\'' : '"';
561 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
564 #define FEATURE_IS_ENABLED(name) \
565 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
566 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
567 /* The longest string we pass in. */
568 #define MAX_FEATURE_LEN (sizeof("switch")-1)
571 * S_feature_is_enabled
572 * Check whether the named feature is enabled.
575 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
578 HV * const hinthv = GvHV(PL_hintgv);
579 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
581 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
583 assert(namelen <= MAX_FEATURE_LEN);
584 memcpy(&he_name[8], name, namelen);
586 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
594 Perl_deprecate(pTHX_ const char *const s)
596 PERL_ARGS_ASSERT_DEPRECATE;
598 if (ckWARN(WARN_DEPRECATED))
599 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
603 Perl_deprecate_old(pTHX_ const char *const s)
605 /* This function should NOT be called for any new deprecated warnings */
606 /* Use Perl_deprecate instead */
608 /* It is here to maintain backward compatibility with the pre-5.8 */
609 /* warnings category hierarchy. The "deprecated" category used to */
610 /* live under the "syntax" category. It is now a top-level category */
611 /* in its own right. */
613 PERL_ARGS_ASSERT_DEPRECATE_OLD;
615 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
616 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
617 "Use of %s is deprecated", s);
621 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
622 * utf16-to-utf8-reversed.
625 #ifdef PERL_CR_FILTER
629 register const char *s = SvPVX_const(sv);
630 register const char * const e = s + SvCUR(sv);
632 PERL_ARGS_ASSERT_STRIP_RETURN;
634 /* outer loop optimized to do nothing if there are no CR-LFs */
636 if (*s++ == '\r' && *s == '\n') {
637 /* hit a CR-LF, need to copy the rest */
638 register char *d = s - 1;
641 if (*s == '\r' && s[1] == '\n')
652 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
654 const I32 count = FILTER_READ(idx+1, sv, maxlen);
655 if (count > 0 && !maxlen)
666 * Create a parser object and initialise its parser and lexer fields
668 * rsfp is the opened file handle to read from (if any),
670 * line holds any initial content already read from the file (or in
671 * the case of no file, such as an eval, the whole contents);
673 * new_filter indicates that this is a new file and it shouldn't inherit
674 * the filters from the current parser (ie require).
678 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
681 const char *s = NULL;
683 yy_parser *parser, *oparser;
685 /* create and initialise a parser */
687 Newxz(parser, 1, yy_parser);
688 parser->old_parser = oparser = PL_parser;
691 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
692 parser->ps = parser->stack;
693 parser->stack_size = YYINITDEPTH;
695 parser->stack->state = 0;
696 parser->yyerrstatus = 0;
697 parser->yychar = YYEMPTY; /* Cause a token to be read. */
699 /* on scope exit, free this parser and restore any outer one */
701 parser->saved_curcop = PL_curcop;
703 /* initialise lexer state */
706 parser->curforce = -1;
708 parser->nexttoke = 0;
710 parser->error_count = oparser ? oparser->error_count : 0;
711 parser->copline = NOLINE;
712 parser->lex_state = LEX_NORMAL;
713 parser->expect = XSTATE;
715 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
716 : MUTABLE_AV(SvREFCNT_inc(oparser->rsfp_filters));
718 Newx(parser->lex_brackstack, 120, char);
719 Newx(parser->lex_casestack, 12, char);
720 *parser->lex_casestack = '\0';
723 s = SvPV_const(line, len);
729 parser->linestr = newSVpvs("\n;");
730 } else if (SvREADONLY(line) || s[len-1] != ';') {
731 parser->linestr = newSVsv(line);
733 sv_catpvs(parser->linestr, "\n;");
736 SvREFCNT_inc_simple_void_NN(line);
737 parser->linestr = line;
739 parser->oldoldbufptr =
742 parser->linestart = SvPVX(parser->linestr);
743 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
744 parser->last_lop = parser->last_uni = NULL;
748 /* delete a parser object */
751 Perl_parser_free(pTHX_ const yy_parser *parser)
753 PERL_ARGS_ASSERT_PARSER_FREE;
755 PL_curcop = parser->saved_curcop;
756 SvREFCNT_dec(parser->linestr);
758 if (parser->rsfp == PerlIO_stdin())
759 PerlIO_clearerr(parser->rsfp);
760 else if (parser->rsfp && (!parser->old_parser ||
761 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
762 PerlIO_close(parser->rsfp);
763 SvREFCNT_dec(parser->rsfp_filters);
765 Safefree(parser->stack);
766 Safefree(parser->lex_brackstack);
767 Safefree(parser->lex_casestack);
768 PL_parser = parser->old_parser;
775 * Finalizer for lexing operations. Must be called when the parser is
776 * done with the lexer.
783 PL_doextract = FALSE;
788 * This subroutine has nothing to do with tilting, whether at windmills
789 * or pinball tables. Its name is short for "increment line". It
790 * increments the current line number in CopLINE(PL_curcop) and checks
791 * to see whether the line starts with a comment of the form
792 * # line 500 "foo.pm"
793 * If so, it sets the current line number and file to the values in the comment.
797 S_incline(pTHX_ const char *s)
804 PERL_ARGS_ASSERT_INCLINE;
806 CopLINE_inc(PL_curcop);
809 while (SPACE_OR_TAB(*s))
811 if (strnEQ(s, "line", 4))
815 if (SPACE_OR_TAB(*s))
819 while (SPACE_OR_TAB(*s))
827 while (SPACE_OR_TAB(*s))
829 if (*s == '"' && (t = strchr(s+1, '"'))) {
839 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
841 if (*e != '\n' && *e != '\0')
842 return; /* false alarm */
845 const STRLEN len = t - s;
847 SV *const temp_sv = CopFILESV(PL_curcop);
853 tmplen = SvCUR(temp_sv);
859 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
860 /* must copy *{"::_<(eval N)[oldfilename:L]"}
861 * to *{"::_<newfilename"} */
862 /* However, the long form of evals is only turned on by the
863 debugger - usually they're "(eval %lu)" */
867 STRLEN tmplen2 = len;
868 if (tmplen + 2 <= sizeof smallbuf)
871 Newx(tmpbuf, tmplen + 2, char);
874 memcpy(tmpbuf + 2, cf, tmplen);
876 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
881 if (tmplen2 + 2 <= sizeof smallbuf)
884 Newx(tmpbuf2, tmplen2 + 2, char);
886 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
887 /* Either they malloc'd it, or we malloc'd it,
888 so no prefix is present in ours. */
893 memcpy(tmpbuf2 + 2, s, tmplen2);
896 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
898 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
899 /* adjust ${"::_<newfilename"} to store the new file name */
900 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
901 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
902 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
905 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
907 if (tmpbuf != smallbuf) Safefree(tmpbuf);
910 CopFILE_free(PL_curcop);
911 CopFILE_setn(PL_curcop, s, len);
913 CopLINE_set(PL_curcop, atoi(n)-1);
917 /* skip space before PL_thistoken */
920 S_skipspace0(pTHX_ register char *s)
922 PERL_ARGS_ASSERT_SKIPSPACE0;
929 PL_thiswhite = newSVpvs("");
930 sv_catsv(PL_thiswhite, PL_skipwhite);
931 sv_free(PL_skipwhite);
934 PL_realtokenstart = s - SvPVX(PL_linestr);
938 /* skip space after PL_thistoken */
941 S_skipspace1(pTHX_ register char *s)
943 const char *start = s;
944 I32 startoff = start - SvPVX(PL_linestr);
946 PERL_ARGS_ASSERT_SKIPSPACE1;
951 start = SvPVX(PL_linestr) + startoff;
952 if (!PL_thistoken && PL_realtokenstart >= 0) {
953 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
954 PL_thistoken = newSVpvn(tstart, start - tstart);
956 PL_realtokenstart = -1;
959 PL_nextwhite = newSVpvs("");
960 sv_catsv(PL_nextwhite, PL_skipwhite);
961 sv_free(PL_skipwhite);
968 S_skipspace2(pTHX_ register char *s, SV **svp)
971 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
972 const I32 startoff = s - SvPVX(PL_linestr);
974 PERL_ARGS_ASSERT_SKIPSPACE2;
977 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
978 if (!PL_madskills || !svp)
980 start = SvPVX(PL_linestr) + startoff;
981 if (!PL_thistoken && PL_realtokenstart >= 0) {
982 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
983 PL_thistoken = newSVpvn(tstart, start - tstart);
984 PL_realtokenstart = -1;
989 sv_setsv(*svp, PL_skipwhite);
990 sv_free(PL_skipwhite);
999 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
1001 AV *av = CopFILEAVx(PL_curcop);
1003 SV * const sv = newSV_type(SVt_PVMG);
1005 sv_setsv(sv, orig_sv);
1007 sv_setpvn(sv, buf, len);
1010 av_store(av, (I32)CopLINE(PL_curcop), sv);
1016 * Called to gobble the appropriate amount and type of whitespace.
1017 * Skips comments as well.
1021 S_skipspace(pTHX_ register char *s)
1026 int startoff = s - SvPVX(PL_linestr);
1028 PERL_ARGS_ASSERT_SKIPSPACE;
1031 sv_free(PL_skipwhite);
1035 PERL_ARGS_ASSERT_SKIPSPACE;
1037 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1038 while (s < PL_bufend && SPACE_OR_TAB(*s))
1048 SSize_t oldprevlen, oldoldprevlen;
1049 SSize_t oldloplen = 0, oldunilen = 0;
1050 while (s < PL_bufend && isSPACE(*s)) {
1051 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1056 if (s < PL_bufend && *s == '#') {
1057 while (s < PL_bufend && *s != '\n')
1059 if (s < PL_bufend) {
1061 if (PL_in_eval && !PL_rsfp) {
1068 /* only continue to recharge the buffer if we're at the end
1069 * of the buffer, we're not reading from a source filter, and
1070 * we're in normal lexing mode
1072 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1073 PL_lex_state == LEX_FORMLINE)
1080 /* try to recharge the buffer */
1082 curoff = s - SvPVX(PL_linestr);
1085 if ((s = filter_gets(PL_linestr, PL_rsfp,
1086 (prevlen = SvCUR(PL_linestr)))) == NULL)
1089 if (PL_madskills && curoff != startoff) {
1091 PL_skipwhite = newSVpvs("");
1092 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1096 /* mustn't throw out old stuff yet if madpropping */
1097 SvCUR(PL_linestr) = curoff;
1098 s = SvPVX(PL_linestr) + curoff;
1100 if (curoff && s[-1] == '\n')
1104 /* end of file. Add on the -p or -n magic */
1105 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1108 sv_catpvs(PL_linestr,
1109 ";}continue{print or die qq(-p destination: $!\\n);}");
1111 sv_setpvs(PL_linestr,
1112 ";}continue{print or die qq(-p destination: $!\\n);}");
1114 PL_minus_n = PL_minus_p = 0;
1116 else if (PL_minus_n) {
1118 sv_catpvs(PL_linestr, ";}");
1120 sv_setpvs(PL_linestr, ";}");
1126 sv_catpvs(PL_linestr,";");
1128 sv_setpvs(PL_linestr,";");
1131 /* reset variables for next time we lex */
1132 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1138 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1139 PL_last_lop = PL_last_uni = NULL;
1141 /* Close the filehandle. Could be from
1142 * STDIN, or a regular file. If we were reading code from
1143 * STDIN (because the commandline held no -e or filename)
1144 * then we don't close it, we reset it so the code can
1145 * read from STDIN too.
1148 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1149 PerlIO_clearerr(PL_rsfp);
1151 (void)PerlIO_close(PL_rsfp);
1156 /* not at end of file, so we only read another line */
1157 /* make corresponding updates to old pointers, for yyerror() */
1158 oldprevlen = PL_oldbufptr - PL_bufend;
1159 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1161 oldunilen = PL_last_uni - PL_bufend;
1163 oldloplen = PL_last_lop - PL_bufend;
1164 PL_linestart = PL_bufptr = s + prevlen;
1165 PL_bufend = s + SvCUR(PL_linestr);
1167 PL_oldbufptr = s + oldprevlen;
1168 PL_oldoldbufptr = s + oldoldprevlen;
1170 PL_last_uni = s + oldunilen;
1172 PL_last_lop = s + oldloplen;
1175 /* debugger active and we're not compiling the debugger code,
1176 * so store the line into the debugger's array of lines
1178 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
1179 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1186 PL_skipwhite = newSVpvs("");
1187 curoff = s - SvPVX(PL_linestr);
1188 if (curoff - startoff)
1189 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1198 * Check the unary operators to ensure there's no ambiguity in how they're
1199 * used. An ambiguous piece of code would be:
1201 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1202 * the +5 is its argument.
1212 if (PL_oldoldbufptr != PL_last_uni)
1214 while (isSPACE(*PL_last_uni))
1217 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1219 if ((t = strchr(s, '(')) && t < PL_bufptr)
1222 if (ckWARN_d(WARN_AMBIGUOUS)){
1223 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1224 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1225 (int)(s - PL_last_uni), PL_last_uni);
1230 * LOP : macro to build a list operator. Its behaviour has been replaced
1231 * with a subroutine, S_lop() for which LOP is just another name.
1234 #define LOP(f,x) return lop(f,x,s)
1238 * Build a list operator (or something that might be one). The rules:
1239 * - if we have a next token, then it's a list operator [why?]
1240 * - if the next thing is an opening paren, then it's a function
1241 * - else it's a list operator
1245 S_lop(pTHX_ I32 f, int x, char *s)
1249 PERL_ARGS_ASSERT_LOP;
1255 PL_last_lop = PL_oldbufptr;
1256 PL_last_lop_op = (OPCODE)f;
1259 return REPORT(LSTOP);
1262 return REPORT(LSTOP);
1265 return REPORT(FUNC);
1268 return REPORT(FUNC);
1270 return REPORT(LSTOP);
1276 * Sets up for an eventual force_next(). start_force(0) basically does
1277 * an unshift, while start_force(-1) does a push. yylex removes items
1282 S_start_force(pTHX_ int where)
1286 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1287 where = PL_lasttoke;
1288 assert(PL_curforce < 0 || PL_curforce == where);
1289 if (PL_curforce != where) {
1290 for (i = PL_lasttoke; i > where; --i) {
1291 PL_nexttoke[i] = PL_nexttoke[i-1];
1295 if (PL_curforce < 0) /* in case of duplicate start_force() */
1296 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1297 PL_curforce = where;
1300 curmad('^', newSVpvs(""));
1301 CURMAD('_', PL_nextwhite);
1306 S_curmad(pTHX_ char slot, SV *sv)
1312 if (PL_curforce < 0)
1313 where = &PL_thismad;
1315 where = &PL_nexttoke[PL_curforce].next_mad;
1321 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1323 else if (PL_encoding) {
1324 sv_recode_to_utf8(sv, PL_encoding);
1329 /* keep a slot open for the head of the list? */
1330 if (slot != '_' && *where && (*where)->mad_key == '^') {
1331 (*where)->mad_key = slot;
1332 sv_free(MUTABLE_SV(((*where)->mad_val)));
1333 (*where)->mad_val = (void*)sv;
1336 addmad(newMADsv(slot, sv), where, 0);
1339 # define start_force(where) NOOP
1340 # define curmad(slot, sv) NOOP
1345 * When the lexer realizes it knows the next token (for instance,
1346 * it is reordering tokens for the parser) then it can call S_force_next
1347 * to know what token to return the next time the lexer is called. Caller
1348 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1349 * and possibly PL_expect to ensure the lexer handles the token correctly.
1353 S_force_next(pTHX_ I32 type)
1358 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1359 tokereport(type, &NEXTVAL_NEXTTOKE);
1363 if (PL_curforce < 0)
1364 start_force(PL_lasttoke);
1365 PL_nexttoke[PL_curforce].next_type = type;
1366 if (PL_lex_state != LEX_KNOWNEXT)
1367 PL_lex_defer = PL_lex_state;
1368 PL_lex_state = LEX_KNOWNEXT;
1369 PL_lex_expect = PL_expect;
1372 PL_nexttype[PL_nexttoke] = type;
1374 if (PL_lex_state != LEX_KNOWNEXT) {
1375 PL_lex_defer = PL_lex_state;
1376 PL_lex_expect = PL_expect;
1377 PL_lex_state = LEX_KNOWNEXT;
1383 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1386 SV * const sv = newSVpvn_utf8(start, len,
1388 && is_utf8_string((const U8*)start, len));
1394 * When the lexer knows the next thing is a word (for instance, it has
1395 * just seen -> and it knows that the next char is a word char, then
1396 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1400 * char *start : buffer position (must be within PL_linestr)
1401 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1402 * int check_keyword : if true, Perl checks to make sure the word isn't
1403 * a keyword (do this if the word is a label, e.g. goto FOO)
1404 * int allow_pack : if true, : characters will also be allowed (require,
1405 * use, etc. do this)
1406 * int allow_initial_tick : used by the "sub" lexer only.
1410 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1416 PERL_ARGS_ASSERT_FORCE_WORD;
1418 start = SKIPSPACE1(start);
1420 if (isIDFIRST_lazy_if(s,UTF) ||
1421 (allow_pack && *s == ':') ||
1422 (allow_initial_tick && *s == '\'') )
1424 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1425 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1427 start_force(PL_curforce);
1429 curmad('X', newSVpvn(start,s-start));
1430 if (token == METHOD) {
1435 PL_expect = XOPERATOR;
1439 curmad('g', newSVpvs( "forced" ));
1440 NEXTVAL_NEXTTOKE.opval
1441 = (OP*)newSVOP(OP_CONST,0,
1442 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1443 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1451 * Called when the lexer wants $foo *foo &foo etc, but the program
1452 * text only contains the "foo" portion. The first argument is a pointer
1453 * to the "foo", and the second argument is the type symbol to prefix.
1454 * Forces the next token to be a "WORD".
1455 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1459 S_force_ident(pTHX_ register const char *s, int kind)
1463 PERL_ARGS_ASSERT_FORCE_IDENT;
1466 const STRLEN len = strlen(s);
1467 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1468 start_force(PL_curforce);
1469 NEXTVAL_NEXTTOKE.opval = o;
1472 o->op_private = OPpCONST_ENTERED;
1473 /* XXX see note in pp_entereval() for why we forgo typo
1474 warnings if the symbol must be introduced in an eval.
1476 gv_fetchpvn_flags(s, len,
1477 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1479 kind == '$' ? SVt_PV :
1480 kind == '@' ? SVt_PVAV :
1481 kind == '%' ? SVt_PVHV :
1489 Perl_str_to_version(pTHX_ SV *sv)
1494 const char *start = SvPV_const(sv,len);
1495 const char * const end = start + len;
1496 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1498 PERL_ARGS_ASSERT_STR_TO_VERSION;
1500 while (start < end) {
1504 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1509 retval += ((NV)n)/nshift;
1518 * Forces the next token to be a version number.
1519 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1520 * and if "guessing" is TRUE, then no new token is created (and the caller
1521 * must use an alternative parsing method).
1525 S_force_version(pTHX_ char *s, int guessing)
1531 I32 startoff = s - SvPVX(PL_linestr);
1534 PERL_ARGS_ASSERT_FORCE_VERSION;
1542 while (isDIGIT(*d) || *d == '_' || *d == '.')
1546 start_force(PL_curforce);
1547 curmad('X', newSVpvn(s,d-s));
1550 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1552 s = scan_num(s, &pl_yylval);
1553 version = pl_yylval.opval;
1554 ver = cSVOPx(version)->op_sv;
1555 if (SvPOK(ver) && !SvNIOK(ver)) {
1556 SvUPGRADE(ver, SVt_PVNV);
1557 SvNV_set(ver, str_to_version(ver));
1558 SvNOK_on(ver); /* hint that it is a version */
1561 else if (guessing) {
1564 sv_free(PL_nextwhite); /* let next token collect whitespace */
1566 s = SvPVX(PL_linestr) + startoff;
1574 if (PL_madskills && !version) {
1575 sv_free(PL_nextwhite); /* let next token collect whitespace */
1577 s = SvPVX(PL_linestr) + startoff;
1580 /* NOTE: The parser sees the package name and the VERSION swapped */
1581 start_force(PL_curforce);
1582 NEXTVAL_NEXTTOKE.opval = version;
1590 * Tokenize a quoted string passed in as an SV. It finds the next
1591 * chunk, up to end of string or a backslash. It may make a new
1592 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1597 S_tokeq(pTHX_ SV *sv)
1601 register char *send;
1606 PERL_ARGS_ASSERT_TOKEQ;
1611 s = SvPV_force(sv, len);
1612 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1615 while (s < send && *s != '\\')
1620 if ( PL_hints & HINT_NEW_STRING ) {
1621 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
1625 if (s + 1 < send && (s[1] == '\\'))
1626 s++; /* all that, just for this */
1631 SvCUR_set(sv, d - SvPVX_const(sv));
1633 if ( PL_hints & HINT_NEW_STRING )
1634 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
1639 * Now come three functions related to double-quote context,
1640 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1641 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1642 * interact with PL_lex_state, and create fake ( ... ) argument lists
1643 * to handle functions and concatenation.
1644 * They assume that whoever calls them will be setting up a fake
1645 * join call, because each subthing puts a ',' after it. This lets
1648 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1650 * (I'm not sure whether the spurious commas at the end of lcfirst's
1651 * arguments and join's arguments are created or not).
1656 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1658 * Pattern matching will set PL_lex_op to the pattern-matching op to
1659 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
1661 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1663 * Everything else becomes a FUNC.
1665 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1666 * had an OP_CONST or OP_READLINE). This just sets us up for a
1667 * call to S_sublex_push().
1671 S_sublex_start(pTHX)
1674 register const I32 op_type = pl_yylval.ival;
1676 if (op_type == OP_NULL) {
1677 pl_yylval.opval = PL_lex_op;
1681 if (op_type == OP_CONST || op_type == OP_READLINE) {
1682 SV *sv = tokeq(PL_lex_stuff);
1684 if (SvTYPE(sv) == SVt_PVIV) {
1685 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1687 const char * const p = SvPV_const(sv, len);
1688 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
1692 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1693 PL_lex_stuff = NULL;
1694 /* Allow <FH> // "foo" */
1695 if (op_type == OP_READLINE)
1696 PL_expect = XTERMORDORDOR;
1699 else if (op_type == OP_BACKTICK && PL_lex_op) {
1700 /* readpipe() vas overriden */
1701 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
1702 pl_yylval.opval = PL_lex_op;
1704 PL_lex_stuff = NULL;
1708 PL_sublex_info.super_state = PL_lex_state;
1709 PL_sublex_info.sub_inwhat = (U16)op_type;
1710 PL_sublex_info.sub_op = PL_lex_op;
1711 PL_lex_state = LEX_INTERPPUSH;
1715 pl_yylval.opval = PL_lex_op;
1725 * Create a new scope to save the lexing state. The scope will be
1726 * ended in S_sublex_done. Returns a '(', starting the function arguments
1727 * to the uc, lc, etc. found before.
1728 * Sets PL_lex_state to LEX_INTERPCONCAT.
1737 PL_lex_state = PL_sublex_info.super_state;
1738 SAVEBOOL(PL_lex_dojoin);
1739 SAVEI32(PL_lex_brackets);
1740 SAVEI32(PL_lex_casemods);
1741 SAVEI32(PL_lex_starts);
1742 SAVEI8(PL_lex_state);
1743 SAVEVPTR(PL_lex_inpat);
1744 SAVEI16(PL_lex_inwhat);
1745 SAVECOPLINE(PL_curcop);
1746 SAVEPPTR(PL_bufptr);
1747 SAVEPPTR(PL_bufend);
1748 SAVEPPTR(PL_oldbufptr);
1749 SAVEPPTR(PL_oldoldbufptr);
1750 SAVEPPTR(PL_last_lop);
1751 SAVEPPTR(PL_last_uni);
1752 SAVEPPTR(PL_linestart);
1753 SAVESPTR(PL_linestr);
1754 SAVEGENERICPV(PL_lex_brackstack);
1755 SAVEGENERICPV(PL_lex_casestack);
1757 PL_linestr = PL_lex_stuff;
1758 PL_lex_stuff = NULL;
1760 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1761 = SvPVX(PL_linestr);
1762 PL_bufend += SvCUR(PL_linestr);
1763 PL_last_lop = PL_last_uni = NULL;
1764 SAVEFREESV(PL_linestr);
1766 PL_lex_dojoin = FALSE;
1767 PL_lex_brackets = 0;
1768 Newx(PL_lex_brackstack, 120, char);
1769 Newx(PL_lex_casestack, 12, char);
1770 PL_lex_casemods = 0;
1771 *PL_lex_casestack = '\0';
1773 PL_lex_state = LEX_INTERPCONCAT;
1774 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1776 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1777 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1778 PL_lex_inpat = PL_sublex_info.sub_op;
1780 PL_lex_inpat = NULL;
1787 * Restores lexer state after a S_sublex_push.
1794 if (!PL_lex_starts++) {
1795 SV * const sv = newSVpvs("");
1796 if (SvUTF8(PL_linestr))
1798 PL_expect = XOPERATOR;
1799 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1803 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1804 PL_lex_state = LEX_INTERPCASEMOD;
1808 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1809 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1810 PL_linestr = PL_lex_repl;
1812 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1813 PL_bufend += SvCUR(PL_linestr);
1814 PL_last_lop = PL_last_uni = NULL;
1815 SAVEFREESV(PL_linestr);
1816 PL_lex_dojoin = FALSE;
1817 PL_lex_brackets = 0;
1818 PL_lex_casemods = 0;
1819 *PL_lex_casestack = '\0';
1821 if (SvEVALED(PL_lex_repl)) {
1822 PL_lex_state = LEX_INTERPNORMAL;
1824 /* we don't clear PL_lex_repl here, so that we can check later
1825 whether this is an evalled subst; that means we rely on the
1826 logic to ensure sublex_done() is called again only via the
1827 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1830 PL_lex_state = LEX_INTERPCONCAT;
1840 PL_endwhite = newSVpvs("");
1841 sv_catsv(PL_endwhite, PL_thiswhite);
1845 sv_setpvs(PL_thistoken,"");
1847 PL_realtokenstart = -1;
1851 PL_bufend = SvPVX(PL_linestr);
1852 PL_bufend += SvCUR(PL_linestr);
1853 PL_expect = XOPERATOR;
1854 PL_sublex_info.sub_inwhat = 0;
1862 Extracts a pattern, double-quoted string, or transliteration. This
1865 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1866 processing a pattern (PL_lex_inpat is true), a transliteration
1867 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1869 Returns a pointer to the character scanned up to. If this is
1870 advanced from the start pointer supplied (i.e. if anything was
1871 successfully parsed), will leave an OP for the substring scanned
1872 in pl_yylval. Caller must intuit reason for not parsing further
1873 by looking at the next characters herself.
1877 double-quoted style: \r and \n
1878 regexp special ones: \D \s
1881 case and quoting: \U \Q \E
1882 stops on @ and $, but not for $ as tail anchor
1884 In transliterations:
1885 characters are VERY literal, except for - not at the start or end
1886 of the string, which indicates a range. If the range is in bytes,
1887 scan_const expands the range to the full set of intermediate
1888 characters. If the range is in utf8, the hyphen is replaced with
1889 a certain range mark which will be handled by pmtrans() in op.c.
1891 In double-quoted strings:
1893 double-quoted style: \r and \n
1895 deprecated backrefs: \1 (in substitution replacements)
1896 case and quoting: \U \Q \E
1899 scan_const does *not* construct ops to handle interpolated strings.
1900 It stops processing as soon as it finds an embedded $ or @ variable
1901 and leaves it to the caller to work out what's going on.
1903 embedded arrays (whether in pattern or not) could be:
1904 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1906 $ in double-quoted strings must be the symbol of an embedded scalar.
1908 $ in pattern could be $foo or could be tail anchor. Assumption:
1909 it's a tail anchor if $ is the last thing in the string, or if it's
1910 followed by one of "()| \r\n\t"
1912 \1 (backreferences) are turned into $1
1914 The structure of the code is
1915 while (there's a character to process) {
1916 handle transliteration ranges
1917 skip regexp comments /(?#comment)/ and codes /(?{code})/
1918 skip #-initiated comments in //x patterns
1919 check for embedded arrays
1920 check for embedded scalars
1922 leave intact backslashes from leaveit (below)
1923 deprecate \1 in substitution replacements
1924 handle string-changing backslashes \l \U \Q \E, etc.
1925 switch (what was escaped) {
1926 handle \- in a transliteration (becomes a literal -)
1927 handle \132 (octal characters)
1928 handle \x15 and \x{1234} (hex characters)
1929 handle \N{name} (named characters)
1930 handle \cV (control characters)
1931 handle printf-style backslashes (\f, \r, \n, etc)
1933 } (end if backslash)
1934 } (end while character to read)
1939 S_scan_const(pTHX_ char *start)
1942 register char *send = PL_bufend; /* end of the constant */
1943 SV *sv = newSV(send - start); /* sv for the constant */
1944 register char *s = start; /* start of the constant */
1945 register char *d = SvPVX(sv); /* destination for copies */
1946 bool dorange = FALSE; /* are we in a translit range? */
1947 bool didrange = FALSE; /* did we just finish a range? */
1948 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1949 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1952 UV literal_endpoint = 0;
1953 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1956 PERL_ARGS_ASSERT_SCAN_CONST;
1958 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1959 /* If we are doing a trans and we know we want UTF8 set expectation */
1960 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1961 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1965 while (s < send || dorange) {
1966 /* get transliterations out of the way (they're most literal) */
1967 if (PL_lex_inwhat == OP_TRANS) {
1968 /* expand a range A-Z to the full set of characters. AIE! */
1970 I32 i; /* current expanded character */
1971 I32 min; /* first character in range */
1972 I32 max; /* last character in range */
1983 char * const c = (char*)utf8_hop((U8*)d, -1);
1987 *c = (char)UTF_TO_NATIVE(0xff);
1988 /* mark the range as done, and continue */
1994 i = d - SvPVX_const(sv); /* remember current offset */
1997 SvLEN(sv) + (has_utf8 ?
1998 (512 - UTF_CONTINUATION_MARK +
2001 /* How many two-byte within 0..255: 128 in UTF-8,
2002 * 96 in UTF-8-mod. */
2004 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
2006 d = SvPVX(sv) + i; /* refresh d after realloc */
2010 for (j = 0; j <= 1; j++) {
2011 char * const c = (char*)utf8_hop((U8*)d, -1);
2012 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2018 max = (U8)0xff; /* only to \xff */
2019 uvmax = uv; /* \x{100} to uvmax */
2021 d = c; /* eat endpoint chars */
2026 d -= 2; /* eat the first char and the - */
2027 min = (U8)*d; /* first char in range */
2028 max = (U8)d[1]; /* last char in range */
2035 "Invalid range \"%c-%c\" in transliteration operator",
2036 (char)min, (char)max);
2040 if (literal_endpoint == 2 &&
2041 ((isLOWER(min) && isLOWER(max)) ||
2042 (isUPPER(min) && isUPPER(max)))) {
2044 for (i = min; i <= max; i++)
2046 *d++ = NATIVE_TO_NEED(has_utf8,i);
2048 for (i = min; i <= max; i++)
2050 *d++ = NATIVE_TO_NEED(has_utf8,i);
2055 for (i = min; i <= max; i++)
2058 const U8 ch = (U8)NATIVE_TO_UTF(i);
2059 if (UNI_IS_INVARIANT(ch))
2062 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2063 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2072 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2074 *d++ = (char)UTF_TO_NATIVE(0xff);
2076 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2080 /* mark the range as done, and continue */
2084 literal_endpoint = 0;
2089 /* range begins (ignore - as first or last char) */
2090 else if (*s == '-' && s+1 < send && s != start) {
2092 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2099 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2109 literal_endpoint = 0;
2110 native_range = TRUE;
2115 /* if we get here, we're not doing a transliteration */
2117 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2118 except for the last char, which will be done separately. */
2119 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2121 while (s+1 < send && *s != ')')
2122 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2124 else if (s[2] == '{' /* This should match regcomp.c */
2125 || (s[2] == '?' && s[3] == '{'))
2128 char *regparse = s + (s[2] == '{' ? 3 : 4);
2131 while (count && (c = *regparse)) {
2132 if (c == '\\' && regparse[1])
2140 if (*regparse != ')')
2141 regparse--; /* Leave one char for continuation. */
2142 while (s < regparse)
2143 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2147 /* likewise skip #-initiated comments in //x patterns */
2148 else if (*s == '#' && PL_lex_inpat &&
2149 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2150 while (s+1 < send && *s != '\n')
2151 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2154 /* check for embedded arrays
2155 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2157 else if (*s == '@' && s[1]) {
2158 if (isALNUM_lazy_if(s+1,UTF))
2160 if (strchr(":'{$", s[1]))
2162 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2163 break; /* in regexp, neither @+ nor @- are interpolated */
2166 /* check for embedded scalars. only stop if we're sure it's a
2169 else if (*s == '$') {
2170 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2172 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
2173 if (s[1] == '\\' && ckWARN(WARN_AMBIGUOUS)) {
2174 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2175 "Possible unintended interpolation of $\\ in regex");
2177 break; /* in regexp, $ might be tail anchor */
2181 /* End of else if chain - OP_TRANS rejoin rest */
2184 if (*s == '\\' && s+1 < send) {
2187 /* deprecate \1 in strings and substitution replacements */
2188 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2189 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2191 if (ckWARN(WARN_SYNTAX))
2192 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2197 /* string-change backslash escapes */
2198 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2202 /* skip any other backslash escapes in a pattern */
2203 else if (PL_lex_inpat) {
2204 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2205 goto default_action;
2208 /* if we get here, it's either a quoted -, or a digit */
2211 /* quoted - in transliterations */
2213 if (PL_lex_inwhat == OP_TRANS) {
2220 if ((isALPHA(*s) || isDIGIT(*s)) &&
2222 Perl_warner(aTHX_ packWARN(WARN_MISC),
2223 "Unrecognized escape \\%c passed through",
2225 /* default action is to copy the quoted character */
2226 goto default_action;
2229 /* \132 indicates an octal constant */
2230 case '0': case '1': case '2': case '3':
2231 case '4': case '5': case '6': case '7':
2235 uv = grok_oct(s, &len, &flags, NULL);
2238 goto NUM_ESCAPE_INSERT;
2240 /* \x24 indicates a hex constant */
2244 char* const e = strchr(s, '}');
2245 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2246 PERL_SCAN_DISALLOW_PREFIX;
2251 yyerror("Missing right brace on \\x{}");
2255 uv = grok_hex(s, &len, &flags, NULL);
2261 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2262 uv = grok_hex(s, &len, &flags, NULL);
2268 /* Insert oct or hex escaped character.
2269 * There will always enough room in sv since such
2270 * escapes will be longer than any UTF-8 sequence
2271 * they can end up as. */
2273 /* We need to map to chars to ASCII before doing the tests
2276 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2277 if (!has_utf8 && uv > 255) {
2278 /* Might need to recode whatever we have
2279 * accumulated so far if it contains any
2282 * (Can't we keep track of that and avoid
2283 * this rescan? --jhi)
2287 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2288 if (!NATIVE_IS_INVARIANT(*c)) {
2293 const STRLEN offset = d - SvPVX_const(sv);
2295 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2299 while (src >= (const U8 *)SvPVX_const(sv)) {
2300 if (!NATIVE_IS_INVARIANT(*src)) {
2301 const U8 ch = NATIVE_TO_ASCII(*src);
2302 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2303 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2313 if (has_utf8 || uv > 255) {
2314 d = (char*)uvchr_to_utf8((U8*)d, uv);
2316 if (PL_lex_inwhat == OP_TRANS &&
2317 PL_sublex_info.sub_op) {
2318 PL_sublex_info.sub_op->op_private |=
2319 (PL_lex_repl ? OPpTRANS_FROM_UTF
2323 if (uv > 255 && !dorange)
2324 native_range = FALSE;
2336 /* \N{LATIN SMALL LETTER A} is a named character */
2340 char* e = strchr(s, '}');
2346 yyerror("Missing right brace on \\N{}");
2350 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2352 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2353 PERL_SCAN_DISALLOW_PREFIX;
2356 uv = grok_hex(s, &len, &flags, NULL);
2357 if ( e > s && len != (STRLEN)(e - s) ) {
2361 goto NUM_ESCAPE_INSERT;
2363 res = newSVpvn(s + 1, e - s - 1);
2364 res = new_constant( NULL, 0, "charnames",
2365 res, NULL, s - 2, e - s + 3 );
2367 sv_utf8_upgrade(res);
2368 str = SvPV_const(res,len);
2369 #ifdef EBCDIC_NEVER_MIND
2370 /* charnames uses pack U and that has been
2371 * recently changed to do the below uni->native
2372 * mapping, so this would be redundant (and wrong,
2373 * the code point would be doubly converted).
2374 * But leave this in just in case the pack U change
2375 * gets revoked, but the semantics is still
2376 * desireable for charnames. --jhi */
2378 UV uv = utf8_to_uvchr((const U8*)str, 0);
2381 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2383 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2384 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2385 str = SvPV_const(res, len);
2389 if (!has_utf8 && SvUTF8(res)) {
2390 const char * const ostart = SvPVX_const(sv);
2391 SvCUR_set(sv, d - ostart);
2394 sv_utf8_upgrade(sv);
2395 /* this just broke our allocation above... */
2396 SvGROW(sv, (STRLEN)(send - start));
2397 d = SvPVX(sv) + SvCUR(sv);
2400 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2401 const char * const odest = SvPVX_const(sv);
2403 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2404 d = SvPVX(sv) + (d - odest);
2408 native_range = FALSE; /* \N{} is guessed to be Unicode */
2410 Copy(str, d, len, char);
2417 yyerror("Missing braces on \\N{}");
2420 /* \c is a control character */
2429 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2432 yyerror("Missing control char name in \\c");
2436 /* printf-style backslashes, formfeeds, newlines, etc */
2438 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2441 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2444 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2447 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2450 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2453 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2456 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2462 } /* end if (backslash) */
2469 /* If we started with encoded form, or already know we want it
2470 and then encode the next character */
2471 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2473 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2474 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2477 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2478 const STRLEN off = d - SvPVX_const(sv);
2479 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2481 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2484 if (uv > 255 && !dorange)
2485 native_range = FALSE;
2489 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2491 } /* while loop to process each character */
2493 /* terminate the string and set up the sv */
2495 SvCUR_set(sv, d - SvPVX_const(sv));
2496 if (SvCUR(sv) >= SvLEN(sv))
2497 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2500 if (PL_encoding && !has_utf8) {
2501 sv_recode_to_utf8(sv, PL_encoding);
2507 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2508 PL_sublex_info.sub_op->op_private |=
2509 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2513 /* shrink the sv if we allocated more than we used */
2514 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2515 SvPV_shrink_to_cur(sv);
2518 /* return the substring (via pl_yylval) only if we parsed anything */
2519 if (s > PL_bufptr) {
2520 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2521 const char *const key = PL_lex_inpat ? "qr" : "q";
2522 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2526 if (PL_lex_inwhat == OP_TRANS) {
2529 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2537 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2540 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2547 * Returns TRUE if there's more to the expression (e.g., a subscript),
2550 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2552 * ->[ and ->{ return TRUE
2553 * { and [ outside a pattern are always subscripts, so return TRUE
2554 * if we're outside a pattern and it's not { or [, then return FALSE
2555 * if we're in a pattern and the first char is a {
2556 * {4,5} (any digits around the comma) returns FALSE
2557 * if we're in a pattern and the first char is a [
2559 * [SOMETHING] has a funky algorithm to decide whether it's a
2560 * character class or not. It has to deal with things like
2561 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2562 * anything else returns TRUE
2565 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2568 S_intuit_more(pTHX_ register char *s)
2572 PERL_ARGS_ASSERT_INTUIT_MORE;
2574 if (PL_lex_brackets)
2576 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2578 if (*s != '{' && *s != '[')
2583 /* In a pattern, so maybe we have {n,m}. */
2600 /* On the other hand, maybe we have a character class */
2603 if (*s == ']' || *s == '^')
2606 /* this is terrifying, and it works */
2607 int weight = 2; /* let's weigh the evidence */
2609 unsigned char un_char = 255, last_un_char;
2610 const char * const send = strchr(s,']');
2611 char tmpbuf[sizeof PL_tokenbuf * 4];
2613 if (!send) /* has to be an expression */
2616 Zero(seen,256,char);
2619 else if (isDIGIT(*s)) {
2621 if (isDIGIT(s[1]) && s[2] == ']')
2627 for (; s < send; s++) {
2628 last_un_char = un_char;
2629 un_char = (unsigned char)*s;
2634 weight -= seen[un_char] * 10;
2635 if (isALNUM_lazy_if(s+1,UTF)) {
2637 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2638 len = (int)strlen(tmpbuf);
2639 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2644 else if (*s == '$' && s[1] &&
2645 strchr("[#!%*<>()-=",s[1])) {
2646 if (/*{*/ strchr("])} =",s[2]))
2655 if (strchr("wds]",s[1]))
2657 else if (seen[(U8)'\''] || seen[(U8)'"'])
2659 else if (strchr("rnftbxcav",s[1]))
2661 else if (isDIGIT(s[1])) {
2663 while (s[1] && isDIGIT(s[1]))
2673 if (strchr("aA01! ",last_un_char))
2675 if (strchr("zZ79~",s[1]))
2677 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2678 weight -= 5; /* cope with negative subscript */
2681 if (!isALNUM(last_un_char)
2682 && !(last_un_char == '$' || last_un_char == '@'
2683 || last_un_char == '&')
2684 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2689 if (keyword(tmpbuf, d - tmpbuf, 0))
2692 if (un_char == last_un_char + 1)
2694 weight -= seen[un_char];
2699 if (weight >= 0) /* probably a character class */
2709 * Does all the checking to disambiguate
2711 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2712 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2714 * First argument is the stuff after the first token, e.g. "bar".
2716 * Not a method if bar is a filehandle.
2717 * Not a method if foo is a subroutine prototyped to take a filehandle.
2718 * Not a method if it's really "Foo $bar"
2719 * Method if it's "foo $bar"
2720 * Not a method if it's really "print foo $bar"
2721 * Method if it's really "foo package::" (interpreted as package->foo)
2722 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2723 * Not a method if bar is a filehandle or package, but is quoted with
2728 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2731 char *s = start + (*start == '$');
2732 char tmpbuf[sizeof PL_tokenbuf];
2739 PERL_ARGS_ASSERT_INTUIT_METHOD;
2742 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2746 const char *proto = SvPVX_const(cv);
2757 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2758 /* start is the beginning of the possible filehandle/object,
2759 * and s is the end of it
2760 * tmpbuf is a copy of it
2763 if (*start == '$') {
2764 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2765 isUPPER(*PL_tokenbuf))
2768 len = start - SvPVX(PL_linestr);
2772 start = SvPVX(PL_linestr) + len;
2776 return *s == '(' ? FUNCMETH : METHOD;
2778 if (!keyword(tmpbuf, len, 0)) {
2779 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2783 soff = s - SvPVX(PL_linestr);
2787 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2788 if (indirgv && GvCVu(indirgv))
2790 /* filehandle or package name makes it a method */
2791 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2793 soff = s - SvPVX(PL_linestr);
2796 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2797 return 0; /* no assumptions -- "=>" quotes bearword */
2799 start_force(PL_curforce);
2800 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2801 newSVpvn(tmpbuf,len));
2802 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2804 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2809 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2811 return *s == '(' ? FUNCMETH : METHOD;
2817 /* Encoded script support. filter_add() effectively inserts a
2818 * 'pre-processing' function into the current source input stream.
2819 * Note that the filter function only applies to the current source file
2820 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2822 * The datasv parameter (which may be NULL) can be used to pass
2823 * private data to this instance of the filter. The filter function
2824 * can recover the SV using the FILTER_DATA macro and use it to
2825 * store private buffers and state information.
2827 * The supplied datasv parameter is upgraded to a PVIO type
2828 * and the IoDIRP/IoANY field is used to store the function pointer,
2829 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2830 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2831 * private use must be set using malloc'd pointers.
2835 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2844 if (!PL_rsfp_filters)
2845 PL_rsfp_filters = newAV();
2848 SvUPGRADE(datasv, SVt_PVIO);
2849 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2850 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2851 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2852 FPTR2DPTR(void *, IoANY(datasv)),
2853 SvPV_nolen(datasv)));
2854 av_unshift(PL_rsfp_filters, 1);
2855 av_store(PL_rsfp_filters, 0, datasv) ;
2860 /* Delete most recently added instance of this filter function. */
2862 Perl_filter_del(pTHX_ filter_t funcp)
2867 PERL_ARGS_ASSERT_FILTER_DEL;
2870 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2871 FPTR2DPTR(void*, funcp)));
2873 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2875 /* if filter is on top of stack (usual case) just pop it off */
2876 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2877 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2878 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2879 IoANY(datasv) = (void *)NULL;
2880 sv_free(av_pop(PL_rsfp_filters));
2884 /* we need to search for the correct entry and clear it */
2885 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2889 /* Invoke the idxth filter function for the current rsfp. */
2890 /* maxlen 0 = read one text line */
2892 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2897 /* This API is bad. It should have been using unsigned int for maxlen.
2898 Not sure if we want to change the API, but if not we should sanity
2899 check the value here. */
2900 const unsigned int correct_length
2909 PERL_ARGS_ASSERT_FILTER_READ;
2911 if (!PL_parser || !PL_rsfp_filters)
2913 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2914 /* Provide a default input filter to make life easy. */
2915 /* Note that we append to the line. This is handy. */
2916 DEBUG_P(PerlIO_printf(Perl_debug_log,
2917 "filter_read %d: from rsfp\n", idx));
2918 if (correct_length) {
2921 const int old_len = SvCUR(buf_sv);
2923 /* ensure buf_sv is large enough */
2924 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2925 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2926 correct_length)) <= 0) {
2927 if (PerlIO_error(PL_rsfp))
2928 return -1; /* error */
2930 return 0 ; /* end of file */
2932 SvCUR_set(buf_sv, old_len + len) ;
2935 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2936 if (PerlIO_error(PL_rsfp))
2937 return -1; /* error */
2939 return 0 ; /* end of file */
2942 return SvCUR(buf_sv);
2944 /* Skip this filter slot if filter has been deleted */
2945 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2946 DEBUG_P(PerlIO_printf(Perl_debug_log,
2947 "filter_read %d: skipped (filter deleted)\n",
2949 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2951 /* Get function pointer hidden within datasv */
2952 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2953 DEBUG_P(PerlIO_printf(Perl_debug_log,
2954 "filter_read %d: via function %p (%s)\n",
2955 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2956 /* Call function. The function is expected to */
2957 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2958 /* Return: <0:error, =0:eof, >0:not eof */
2959 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2963 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2967 PERL_ARGS_ASSERT_FILTER_GETS;
2969 #ifdef PERL_CR_FILTER
2970 if (!PL_rsfp_filters) {
2971 filter_add(S_cr_textfilter,NULL);
2974 if (PL_rsfp_filters) {
2976 SvCUR_set(sv, 0); /* start with empty line */
2977 if (FILTER_READ(0, sv, 0) > 0)
2978 return ( SvPVX(sv) ) ;
2983 return (sv_gets(sv, fp, append));
2987 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
2992 PERL_ARGS_ASSERT_FIND_IN_MY_STASH;
2994 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2998 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2999 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
3001 return GvHV(gv); /* Foo:: */
3004 /* use constant CLASS => 'MyClass' */
3005 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
3006 if (gv && GvCV(gv)) {
3007 SV * const sv = cv_const_sv(GvCV(gv));
3009 pkgname = SvPV_const(sv, len);
3012 return gv_stashpvn(pkgname, len, 0);
3016 * S_readpipe_override
3017 * Check whether readpipe() is overriden, and generates the appropriate
3018 * optree, provided sublex_start() is called afterwards.
3021 S_readpipe_override(pTHX)
3024 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
3025 pl_yylval.ival = OP_BACKTICK;
3027 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
3029 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
3030 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
3031 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
3033 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
3034 append_elem(OP_LIST,
3035 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
3036 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
3043 * The intent of this yylex wrapper is to minimize the changes to the
3044 * tokener when we aren't interested in collecting madprops. It remains
3045 * to be seen how successful this strategy will be...
3052 char *s = PL_bufptr;
3054 /* make sure PL_thiswhite is initialized */
3058 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
3059 if (PL_pending_ident)
3060 return S_pending_ident(aTHX);
3062 /* previous token ate up our whitespace? */
3063 if (!PL_lasttoke && PL_nextwhite) {
3064 PL_thiswhite = PL_nextwhite;
3068 /* isolate the token, and figure out where it is without whitespace */
3069 PL_realtokenstart = -1;
3073 assert(PL_curforce < 0);
3075 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3076 if (!PL_thistoken) {
3077 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3078 PL_thistoken = newSVpvs("");
3080 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3081 PL_thistoken = newSVpvn(tstart, s - tstart);
3084 if (PL_thismad) /* install head */
3085 CURMAD('X', PL_thistoken);
3088 /* last whitespace of a sublex? */
3089 if (optype == ')' && PL_endwhite) {
3090 CURMAD('X', PL_endwhite);
3095 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3096 if (!PL_thiswhite && !PL_endwhite && !optype) {
3097 sv_free(PL_thistoken);
3102 /* put off final whitespace till peg */
3103 if (optype == ';' && !PL_rsfp) {
3104 PL_nextwhite = PL_thiswhite;
3107 else if (PL_thisopen) {
3108 CURMAD('q', PL_thisopen);
3110 sv_free(PL_thistoken);
3114 /* Store actual token text as madprop X */
3115 CURMAD('X', PL_thistoken);
3119 /* add preceding whitespace as madprop _ */
3120 CURMAD('_', PL_thiswhite);
3124 /* add quoted material as madprop = */
3125 CURMAD('=', PL_thisstuff);
3129 /* add terminating quote as madprop Q */
3130 CURMAD('Q', PL_thisclose);
3134 /* special processing based on optype */
3138 /* opval doesn't need a TOKEN since it can already store mp */
3148 if (pl_yylval.opval)
3149 append_madprops(PL_thismad, pl_yylval.opval, 0);
3157 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3166 /* remember any fake bracket that lexer is about to discard */
3167 if (PL_lex_brackets == 1 &&
3168 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3171 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3174 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3175 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3178 break; /* don't bother looking for trailing comment */
3187 /* attach a trailing comment to its statement instead of next token */
3191 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3193 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3195 if (*s == '\n' || *s == '#') {
3196 while (s < PL_bufend && *s != '\n')
3200 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3201 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3218 /* Create new token struct. Note: opvals return early above. */
3219 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3226 S_tokenize_use(pTHX_ int is_use, char *s) {
3229 PERL_ARGS_ASSERT_TOKENIZE_USE;
3231 if (PL_expect != XSTATE)
3232 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3233 is_use ? "use" : "no"));
3235 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3236 s = force_version(s, TRUE);
3237 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3238 start_force(PL_curforce);
3239 NEXTVAL_NEXTTOKE.opval = NULL;
3242 else if (*s == 'v') {
3243 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3244 s = force_version(s, FALSE);
3248 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3249 s = force_version(s, FALSE);
3251 pl_yylval.ival = is_use;
3255 static const char* const exp_name[] =
3256 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3257 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3264 Works out what to call the token just pulled out of the input
3265 stream. The yacc parser takes care of taking the ops we return and
3266 stitching them into a tree.
3272 if read an identifier
3273 if we're in a my declaration
3274 croak if they tried to say my($foo::bar)
3275 build the ops for a my() declaration
3276 if it's an access to a my() variable
3277 are we in a sort block?
3278 croak if my($a); $a <=> $b
3279 build ops for access to a my() variable
3280 if in a dq string, and they've said @foo and we can't find @foo
3282 build ops for a bareword
3283 if we already built the token before, use it.
3288 #pragma segment Perl_yylex
3294 register char *s = PL_bufptr;
3299 /* orig_keyword, gvp, and gv are initialized here because
3300 * jump to the label just_a_word_zero can bypass their
3301 * initialization later. */
3302 I32 orig_keyword = 0;
3307 SV* tmp = newSVpvs("");
3308 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3309 (IV)CopLINE(PL_curcop),
3310 lex_state_names[PL_lex_state],
3311 exp_name[PL_expect],
3312 pv_display(tmp, s, strlen(s), 0, 60));
3315 /* check if there's an identifier for us to look at */
3316 if (PL_pending_ident)
3317 return REPORT(S_pending_ident(aTHX));
3319 /* no identifier pending identification */
3321 switch (PL_lex_state) {
3323 case LEX_NORMAL: /* Some compilers will produce faster */
3324 case LEX_INTERPNORMAL: /* code if we comment these out. */
3328 /* when we've already built the next token, just pull it out of the queue */
3332 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3334 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3335 PL_nexttoke[PL_lasttoke].next_mad = 0;
3336 if (PL_thismad && PL_thismad->mad_key == '_') {
3337 PL_thiswhite = MUTABLE_SV(PL_thismad->mad_val);
3338 PL_thismad->mad_val = 0;
3339 mad_free(PL_thismad);
3344 PL_lex_state = PL_lex_defer;
3345 PL_expect = PL_lex_expect;
3346 PL_lex_defer = LEX_NORMAL;
3347 if (!PL_nexttoke[PL_lasttoke].next_type)
3352 pl_yylval = PL_nextval[PL_nexttoke];
3354 PL_lex_state = PL_lex_defer;
3355 PL_expect = PL_lex_expect;
3356 PL_lex_defer = LEX_NORMAL;
3360 /* FIXME - can these be merged? */
3361 return(PL_nexttoke[PL_lasttoke].next_type);
3363 return REPORT(PL_nexttype[PL_nexttoke]);
3366 /* interpolated case modifiers like \L \U, including \Q and \E.
3367 when we get here, PL_bufptr is at the \
3369 case LEX_INTERPCASEMOD:
3371 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3372 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3374 /* handle \E or end of string */
3375 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3377 if (PL_lex_casemods) {
3378 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3379 PL_lex_casestack[PL_lex_casemods] = '\0';
3381 if (PL_bufptr != PL_bufend
3382 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3384 PL_lex_state = LEX_INTERPCONCAT;
3387 PL_thistoken = newSVpvs("\\E");
3393 while (PL_bufptr != PL_bufend &&
3394 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3396 PL_thiswhite = newSVpvs("");
3397 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3401 if (PL_bufptr != PL_bufend)
3404 PL_lex_state = LEX_INTERPCONCAT;
3408 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3409 "### Saw case modifier\n"); });
3411 if (s[1] == '\\' && s[2] == 'E') {
3414 PL_thiswhite = newSVpvs("");
3415 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3418 PL_lex_state = LEX_INTERPCONCAT;
3423 if (!PL_madskills) /* when just compiling don't need correct */
3424 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3425 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3426 if ((*s == 'L' || *s == 'U') &&
3427 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3428 PL_lex_casestack[--PL_lex_casemods] = '\0';
3431 if (PL_lex_casemods > 10)
3432 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3433 PL_lex_casestack[PL_lex_casemods++] = *s;
3434 PL_lex_casestack[PL_lex_casemods] = '\0';
3435 PL_lex_state = LEX_INTERPCONCAT;
3436 start_force(PL_curforce);
3437 NEXTVAL_NEXTTOKE.ival = 0;
3439 start_force(PL_curforce);
3441 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3443 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3445 NEXTVAL_NEXTTOKE.ival = OP_LC;
3447 NEXTVAL_NEXTTOKE.ival = OP_UC;
3449 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3451 Perl_croak(aTHX_ "panic: yylex");
3453 SV* const tmpsv = newSVpvs("\\ ");
3454 /* replace the space with the character we want to escape
3456 SvPVX(tmpsv)[1] = *s;
3462 if (PL_lex_starts) {
3468 sv_free(PL_thistoken);
3469 PL_thistoken = newSVpvs("");
3472 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3473 if (PL_lex_casemods == 1 && PL_lex_inpat)
3482 case LEX_INTERPPUSH:
3483 return REPORT(sublex_push());
3485 case LEX_INTERPSTART:
3486 if (PL_bufptr == PL_bufend)
3487 return REPORT(sublex_done());
3488 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3489 "### Interpolated variable\n"); });
3491 PL_lex_dojoin = (*PL_bufptr == '@');
3492 PL_lex_state = LEX_INTERPNORMAL;
3493 if (PL_lex_dojoin) {
3494 start_force(PL_curforce);
3495 NEXTVAL_NEXTTOKE.ival = 0;
3497 start_force(PL_curforce);
3498 force_ident("\"", '$');
3499 start_force(PL_curforce);
3500 NEXTVAL_NEXTTOKE.ival = 0;
3502 start_force(PL_curforce);
3503 NEXTVAL_NEXTTOKE.ival = 0;
3505 start_force(PL_curforce);
3506 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3509 if (PL_lex_starts++) {
3514 sv_free(PL_thistoken);
3515 PL_thistoken = newSVpvs("");
3518 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3519 if (!PL_lex_casemods && PL_lex_inpat)
3526 case LEX_INTERPENDMAYBE:
3527 if (intuit_more(PL_bufptr)) {
3528 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3534 if (PL_lex_dojoin) {
3535 PL_lex_dojoin = FALSE;
3536 PL_lex_state = LEX_INTERPCONCAT;
3540 sv_free(PL_thistoken);
3541 PL_thistoken = newSVpvs("");
3546 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3547 && SvEVALED(PL_lex_repl))
3549 if (PL_bufptr != PL_bufend)
3550 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3554 case LEX_INTERPCONCAT:
3556 if (PL_lex_brackets)
3557 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3559 if (PL_bufptr == PL_bufend)
3560 return REPORT(sublex_done());
3562 if (SvIVX(PL_linestr) == '\'') {
3563 SV *sv = newSVsv(PL_linestr);
3566 else if ( PL_hints & HINT_NEW_RE )
3567 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3568 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3572 s = scan_const(PL_bufptr);
3574 PL_lex_state = LEX_INTERPCASEMOD;
3576 PL_lex_state = LEX_INTERPSTART;
3579 if (s != PL_bufptr) {
3580 start_force(PL_curforce);
3582 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3584 NEXTVAL_NEXTTOKE = pl_yylval;
3587 if (PL_lex_starts++) {
3591 sv_free(PL_thistoken);
3592 PL_thistoken = newSVpvs("");
3595 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3596 if (!PL_lex_casemods && PL_lex_inpat)
3609 PL_lex_state = LEX_NORMAL;
3610 s = scan_formline(PL_bufptr);
3611 if (!PL_lex_formbrack)
3617 PL_oldoldbufptr = PL_oldbufptr;
3623 sv_free(PL_thistoken);
3626 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3630 if (isIDFIRST_lazy_if(s,UTF))
3632 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3633 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3636 goto fake_eof; /* emulate EOF on ^D or ^Z */
3645 if (PL_lex_brackets) {
3646 yyerror((const char *)
3648 ? "Format not terminated"
3649 : "Missing right curly or square bracket"));
3651 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3652 "### Tokener got EOF\n");
3656 if (s++ < PL_bufend)
3657 goto retry; /* ignore stray nulls */
3660 if (!PL_in_eval && !PL_preambled) {
3661 PL_preambled = TRUE;
3667 /* Generate a string of Perl code to load the debugger.
3668 * If PERL5DB is set, it will return the contents of that,
3669 * otherwise a compile-time require of perl5db.pl. */
3671 const char * const pdb = PerlEnv_getenv("PERL5DB");
3674 sv_setpv(PL_linestr, pdb);
3675 sv_catpvs(PL_linestr,";");
3677 SETERRNO(0,SS_NORMAL);
3678 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3681 sv_setpvs(PL_linestr,"");
3682 if (PL_preambleav) {
3683 SV **svp = AvARRAY(PL_preambleav);
3684 SV **const end = svp + AvFILLp(PL_preambleav);
3686 sv_catsv(PL_linestr, *svp);
3688 sv_catpvs(PL_linestr, ";");
3690 sv_free(MUTABLE_SV(PL_preambleav));
3691 PL_preambleav = NULL;
3694 sv_catpvs(PL_linestr,
3695 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3696 if (PL_minus_n || PL_minus_p) {
3697 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3699 sv_catpvs(PL_linestr,"chomp;");
3702 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3703 || *PL_splitstr == '"')
3704 && strchr(PL_splitstr + 1, *PL_splitstr))
3705 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3707 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3708 bytes can be used as quoting characters. :-) */
3709 const char *splits = PL_splitstr;
3710 sv_catpvs(PL_linestr, "our @F=split(q\0");
3713 if (*splits == '\\')
3714 sv_catpvn(PL_linestr, splits, 1);
3715 sv_catpvn(PL_linestr, splits, 1);
3716 } while (*splits++);
3717 /* This loop will embed the trailing NUL of
3718 PL_linestr as the last thing it does before
3720 sv_catpvs(PL_linestr, ");");
3724 sv_catpvs(PL_linestr,"our @F=split(' ');");
3727 sv_catpvs(PL_linestr, "\n");
3728 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3729 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3730 PL_last_lop = PL_last_uni = NULL;
3731 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3732 update_debugger_info(PL_linestr, NULL, 0);
3736 bof = PL_rsfp ? TRUE : FALSE;
3737 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3740 PL_realtokenstart = -1;
3743 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3744 PerlIO_clearerr(PL_rsfp);
3746 (void)PerlIO_close(PL_rsfp);
3748 PL_doextract = FALSE;
3750 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3756 sv_setpvs(PL_linestr, ";}continue{print;}");
3758 sv_setpvs(PL_linestr, ";}");
3759 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3760 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3761 PL_last_lop = PL_last_uni = NULL;
3762 PL_minus_n = PL_minus_p = 0;
3765 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3766 PL_last_lop = PL_last_uni = NULL;
3767 sv_setpvs(PL_linestr,"");
3768 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3770 /* If it looks like the start of a BOM or raw UTF-16,
3771 * check if it in fact is. */
3777 #ifdef PERLIO_IS_STDIO
3778 # ifdef __GNU_LIBRARY__
3779 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3780 # define FTELL_FOR_PIPE_IS_BROKEN
3784 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3785 # define FTELL_FOR_PIPE_IS_BROKEN
3790 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3792 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3793 s = swallow_bom((U8*)s);
3797 /* Incest with pod. */
3800 sv_catsv(PL_thiswhite, PL_linestr);
3802 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3803 sv_setpvs(PL_linestr, "");
3804 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3805 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3806 PL_last_lop = PL_last_uni = NULL;
3807 PL_doextract = FALSE;
3811 } while (PL_doextract);
3812 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3813 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
3814 update_debugger_info(PL_linestr, NULL, 0);
3815 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3816 PL_last_lop = PL_last_uni = NULL;
3817 if (CopLINE(PL_curcop) == 1) {
3818 while (s < PL_bufend && isSPACE(*s))
3820 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3824 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3828 if (*s == '#' && *(s+1) == '!')
3830 #ifdef ALTERNATE_SHEBANG
3832 static char const as[] = ALTERNATE_SHEBANG;
3833 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3834 d = s + (sizeof(as) - 1);
3836 #endif /* ALTERNATE_SHEBANG */
3845 while (*d && !isSPACE(*d))
3849 #ifdef ARG_ZERO_IS_SCRIPT
3850 if (ipathend > ipath) {
3852 * HP-UX (at least) sets argv[0] to the script name,
3853 * which makes $^X incorrect. And Digital UNIX and Linux,
3854 * at least, set argv[0] to the basename of the Perl
3855 * interpreter. So, having found "#!", we'll set it right.
3857 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3859 assert(SvPOK(x) || SvGMAGICAL(x));
3860 if (sv_eq(x, CopFILESV(PL_curcop))) {
3861 sv_setpvn(x, ipath, ipathend - ipath);
3867 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3868 const char * const lstart = SvPV_const(x,llen);
3870 bstart += blen - llen;
3871 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3872 sv_setpvn(x, ipath, ipathend - ipath);
3877 TAINT_NOT; /* $^X is always tainted, but that's OK */
3879 #endif /* ARG_ZERO_IS_SCRIPT */
3884 d = instr(s,"perl -");
3886 d = instr(s,"perl");
3888 /* avoid getting into infinite loops when shebang
3889 * line contains "Perl" rather than "perl" */
3891 for (d = ipathend-4; d >= ipath; --d) {
3892 if ((*d == 'p' || *d == 'P')
3893 && !ibcmp(d, "perl", 4))
3903 #ifdef ALTERNATE_SHEBANG
3905 * If the ALTERNATE_SHEBANG on this system starts with a
3906 * character that can be part of a Perl expression, then if
3907 * we see it but not "perl", we're probably looking at the
3908 * start of Perl code, not a request to hand off to some
3909 * other interpreter. Similarly, if "perl" is there, but
3910 * not in the first 'word' of the line, we assume the line
3911 * contains the start of the Perl program.
3913 if (d && *s != '#') {
3914 const char *c = ipath;
3915 while (*c && !strchr("; \t\r\n\f\v#", *c))
3918 d = NULL; /* "perl" not in first word; ignore */
3920 *s = '#'; /* Don't try to parse shebang line */
3922 #endif /* ALTERNATE_SHEBANG */
3923 #ifndef MACOS_TRADITIONAL
3928 !instr(s,"indir") &&
3929 instr(PL_origargv[0],"perl"))
3936 while (s < PL_bufend && isSPACE(*s))
3938 if (s < PL_bufend) {
3939 Newxz(newargv,PL_origargc+3,char*);
3941 while (s < PL_bufend && !isSPACE(*s))
3944 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3947 newargv = PL_origargv;
3950 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3952 Perl_croak(aTHX_ "Can't exec %s", ipath);
3956 while (*d && !isSPACE(*d))
3958 while (SPACE_OR_TAB(*d))
3962 const bool switches_done = PL_doswitches;
3963 const U32 oldpdb = PL_perldb;
3964 const bool oldn = PL_minus_n;
3965 const bool oldp = PL_minus_p;
3969 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3970 const char * const m = d1;
3971 while (*d1 && !isSPACE(*d1))
3973 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3976 d1 = moreswitches(d1);
3978 if (PL_doswitches && !switches_done) {
3979 int argc = PL_origargc;
3980 char **argv = PL_origargv;
3983 } while (argc && argv[0][0] == '-' && argv[0][1]);
3984 init_argv_symbols(argc,argv);
3986 if (((PERLDB_LINE || PERLDB_SAVESRC) && !oldpdb) ||
3987 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3988 /* if we have already added "LINE: while (<>) {",
3989 we must not do it again */
3991 sv_setpvs(PL_linestr, "");
3992 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3993 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3994 PL_last_lop = PL_last_uni = NULL;
3995 PL_preambled = FALSE;
3996 if (PERLDB_LINE || PERLDB_SAVESRC)
3997 (void)gv_fetchfile(PL_origfilename);
4004 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4006 PL_lex_state = LEX_FORMLINE;
4011 #ifdef PERL_STRICT_CR
4012 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
4014 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
4016 case ' ': case '\t': case '\f': case 013:
4017 #ifdef MACOS_TRADITIONAL
4021 PL_realtokenstart = -1;
4023 PL_thiswhite = newSVpvs("");
4024 sv_catpvn(PL_thiswhite, s, 1);
4031 PL_realtokenstart = -1;
4035 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
4036 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
4037 /* handle eval qq[#line 1 "foo"\n ...] */
4038 CopLINE_dec(PL_curcop);
4041 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
4043 if (!PL_in_eval || PL_rsfp)
4048 while (d < PL_bufend && *d != '\n')
4052 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4053 Perl_croak(aTHX_ "panic: input overflow");
4056 PL_thiswhite = newSVpvn(s, d - s);
4061 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
4063 PL_lex_state = LEX_FORMLINE;
4069 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4070 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4073 TOKEN(PEG); /* make sure any #! line is accessible */
4078 /* if (PL_madskills && PL_lex_formbrack) { */
4080 while (d < PL_bufend && *d != '\n')
4084 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4085 Perl_croak(aTHX_ "panic: input overflow");
4086 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4088 PL_thiswhite = newSVpvs("");
4089 if (CopLINE(PL_curcop) == 1) {
4090 sv_setpvs(PL_thiswhite, "");
4093 sv_catpvn(PL_thiswhite, s, d - s);
4107 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4115 while (s < PL_bufend && SPACE_OR_TAB(*s))
4118 if (strnEQ(s,"=>",2)) {
4119 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4120 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4121 OPERATOR('-'); /* unary minus */
4123 PL_last_uni = PL_oldbufptr;
4125 case 'r': ftst = OP_FTEREAD; break;
4126 case 'w': ftst = OP_FTEWRITE; break;
4127 case 'x': ftst = OP_FTEEXEC; break;
4128 case 'o': ftst = OP_FTEOWNED; break;
4129 case 'R': ftst = OP_FTRREAD; break;
4130 case 'W': ftst = OP_FTRWRITE; break;
4131 case 'X': ftst = OP_FTREXEC; break;
4132 case 'O': ftst = OP_FTROWNED; break;
4133 case 'e': ftst = OP_FTIS; break;
4134 case 'z': ftst = OP_FTZERO; break;
4135 case 's': ftst = OP_FTSIZE; break;
4136 case 'f': ftst = OP_FTFILE; break;
4137 case 'd': ftst = OP_FTDIR; break;
4138 case 'l': ftst = OP_FTLINK; break;
4139 case 'p': ftst = OP_FTPIPE; break;
4140 case 'S': ftst = OP_FTSOCK; break;
4141 case 'u': ftst = OP_FTSUID; break;
4142 case 'g': ftst = OP_FTSGID; break;
4143 case 'k': ftst = OP_FTSVTX; break;
4144 case 'b': ftst = OP_FTBLK; break;
4145 case 'c': ftst = OP_FTCHR; break;
4146 case 't': ftst = OP_FTTTY; break;
4147 case 'T': ftst = OP_FTTEXT; break;
4148 case 'B': ftst = OP_FTBINARY; break;
4149 case 'M': case 'A': case 'C':
4150 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4152 case 'M': ftst = OP_FTMTIME; break;
4153 case 'A': ftst = OP_FTATIME; break;
4154 case 'C': ftst = OP_FTCTIME; break;
4162 PL_last_lop_op = (OPCODE)ftst;
4163 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4164 "### Saw file test %c\n", (int)tmp);
4169 /* Assume it was a minus followed by a one-letter named
4170 * subroutine call (or a -bareword), then. */
4171 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4172 "### '-%c' looked like a file test but was not\n",
4179 const char tmp = *s++;
4182 if (PL_expect == XOPERATOR)
4187 else if (*s == '>') {
4190 if (isIDFIRST_lazy_if(s,UTF)) {
4191 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4199 if (PL_expect == XOPERATOR)
4202 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4204 OPERATOR('-'); /* unary minus */
4210 const char tmp = *s++;
4213 if (PL_expect == XOPERATOR)
4218 if (PL_expect == XOPERATOR)
4221 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4228 if (PL_expect != XOPERATOR) {
4229 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4230 PL_expect = XOPERATOR;
4231 force_ident(PL_tokenbuf, '*');
4244 if (PL_expect == XOPERATOR) {
4248 PL_tokenbuf[0] = '%';
4249 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4250 sizeof PL_tokenbuf - 1, FALSE);
4251 if (!PL_tokenbuf[1]) {
4254 PL_pending_ident = '%';
4265 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4272 const char tmp = *s++;
4278 goto just_a_word_zero_gv;
4281 switch (PL_expect) {
4287 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4289 PL_bufptr = s; /* update in case we back off */
4295 PL_expect = XTERMBLOCK;
4298 stuffstart = s - SvPVX(PL_linestr) - 1;
4302 while (isIDFIRST_lazy_if(s,UTF)) {
4305 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4306 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4307 if (tmp < 0) tmp = -tmp;
4321 sv = newSVpvn(s, len);
4323 d = scan_str(d,TRUE,TRUE);
4325 /* MUST advance bufptr here to avoid bogus
4326 "at end of line" context messages from yyerror().
4328 PL_bufptr = s + len;
4329 yyerror("Unterminated attribute parameter in attribute list");
4333 return REPORT(0); /* EOF indicator */
4337 sv_catsv(sv, PL_lex_stuff);
4338 attrs = append_elem(OP_LIST, attrs,
4339 newSVOP(OP_CONST, 0, sv));
4340 SvREFCNT_dec(PL_lex_stuff);
4341 PL_lex_stuff = NULL;
4344 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4346 if (PL_in_my == KEY_our) {
4348 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4350 /* skip to avoid loading attributes.pm */
4352 deprecate(":unique");
4355 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4358 /* NOTE: any CV attrs applied here need to be part of
4359 the CVf_BUILTIN_ATTRS define in cv.h! */
4360 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4362 CvLVALUE_on(PL_compcv);
4364 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4366 CvLOCKED_on(PL_compcv);
4368 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4370 CvMETHOD_on(PL_compcv);
4372 /* After we've set the flags, it could be argued that
4373 we don't need to do the attributes.pm-based setting
4374 process, and shouldn't bother appending recognized
4375 flags. To experiment with that, uncomment the
4376 following "else". (Note that's already been
4377 uncommented. That keeps the above-applied built-in
4378 attributes from being intercepted (and possibly
4379 rejected) by a package's attribute routines, but is
4380 justified by the performance win for the common case
4381 of applying only built-in attributes.) */
4383 attrs = append_elem(OP_LIST, attrs,
4384 newSVOP(OP_CONST, 0,
4388 if (*s == ':' && s[1] != ':')
4391 break; /* require real whitespace or :'s */
4392 /* XXX losing whitespace on sequential attributes here */
4396 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4397 if (*s != ';' && *s != '}' && *s != tmp
4398 && (tmp != '=' || *s != ')')) {
4399 const char q = ((*s == '\'') ? '"' : '\'');
4400 /* If here for an expression, and parsed no attrs, back
4402 if (tmp == '=' && !attrs) {
4406 /* MUST advance bufptr here to avoid bogus "at end of line"
4407 context messages from yyerror().
4410 yyerror( (const char *)
4412 ? Perl_form(aTHX_ "Invalid separator character "
4413 "%c%c%c in attribute list", q, *s, q)
4414 : "Unterminated attribute list" ) );
4422 start_force(PL_curforce);
4423 NEXTVAL_NEXTTOKE.opval = attrs;
4424 CURMAD('_', PL_nextwhite);
4429 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4430 (s - SvPVX(PL_linestr)) - stuffstart);
4438 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4439 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4447 const char tmp = *s++;
4452 const char tmp = *s++;
4460 if (PL_lex_brackets <= 0)
4461 yyerror("Unmatched right square bracket");
4464 if (PL_lex_state == LEX_INTERPNORMAL) {
4465 if (PL_lex_brackets == 0) {
4466 if (*s == '-' && s[1] == '>')
4467 PL_lex_state = LEX_INTERPENDMAYBE;
4468 else if (*s != '[' && *s != '{')
4469 PL_lex_state = LEX_INTERPEND;
4476 if (PL_lex_brackets > 100) {
4477 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4479 switch (PL_expect) {
4481 if (PL_lex_formbrack) {
4485 if (PL_oldoldbufptr == PL_last_lop)
4486 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4488 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4489 OPERATOR(HASHBRACK);
4491 while (s < PL_bufend && SPACE_OR_TAB(*s))
4494 PL_tokenbuf[0] = '\0';
4495 if (d < PL_bufend && *d == '-') {
4496 PL_tokenbuf[0] = '-';
4498 while (d < PL_bufend && SPACE_OR_TAB(*d))
4501 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4502 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4504 while (d < PL_bufend && SPACE_OR_TAB(*d))
4507 const char minus = (PL_tokenbuf[0] == '-');
4508 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4516 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4521 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4526 if (PL_oldoldbufptr == PL_last_lop)
4527 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4529 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4532 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4534 /* This hack is to get the ${} in the message. */
4536 yyerror("syntax error");
4539 OPERATOR(HASHBRACK);
4541 /* This hack serves to disambiguate a pair of curlies
4542 * as being a block or an anon hash. Normally, expectation
4543 * determines that, but in cases where we're not in a
4544 * position to expect anything in particular (like inside
4545 * eval"") we have to resolve the ambiguity. This code
4546 * covers the case where the first term in the curlies is a
4547 * quoted string. Most other cases need to be explicitly
4548 * disambiguated by prepending a "+" before the opening
4549 * curly in order to force resolution as an anon hash.
4551 * XXX should probably propagate the outer expectation
4552 * into eval"" to rely less on this hack, but that could
4553 * potentially break current behavior of eval"".
4557 if (*s == '\'' || *s == '"' || *s == '`') {
4558 /* common case: get past first string, handling escapes */
4559 for (t++; t < PL_bufend && *t != *s;)
4560 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4564 else if (*s == 'q') {
4567 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4570 /* skip q//-like construct */
4572 char open, close, term;
4575 while (t < PL_bufend && isSPACE(*t))
4577 /* check for q => */
4578 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4579 OPERATOR(HASHBRACK);
4583 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4587 for (t++; t < PL_bufend; t++) {
4588 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4590 else if (*t == open)
4594 for (t++; t < PL_bufend; t++) {
4595 if (*t == '\\' && t+1 < PL_bufend)
4597 else if (*t == close && --brackets <= 0)
4599 else if (*t == open)
4606 /* skip plain q word */
4607 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4610 else if (isALNUM_lazy_if(t,UTF)) {
4612 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4615 while (t < PL_bufend && isSPACE(*t))
4617 /* if comma follows first term, call it an anon hash */
4618 /* XXX it could be a comma expression with loop modifiers */
4619 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4620 || (*t == '=' && t[1] == '>')))
4621 OPERATOR(HASHBRACK);
4622 if (PL_expect == XREF)
4625 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4631 pl_yylval.ival = CopLINE(PL_curcop);
4632 if (isSPACE(*s) || *s == '#')
4633 PL_copline = NOLINE; /* invalidate current command line number */
4638 if (PL_lex_brackets <= 0)
4639 yyerror("Unmatched right curly bracket");
4641 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4642 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4643 PL_lex_formbrack = 0;
4644 if (PL_lex_state == LEX_INTERPNORMAL) {
4645 if (PL_lex_brackets == 0) {
4646 if (PL_expect & XFAKEBRACK) {
4647 PL_expect &= XENUMMASK;
4648 PL_lex_state = LEX_INTERPEND;
4653 PL_thiswhite = newSVpvs("");
4654 sv_catpvs(PL_thiswhite,"}");
4657 return yylex(); /* ignore fake brackets */
4659 if (*s == '-' && s[1] == '>')
4660 PL_lex_state = LEX_INTERPENDMAYBE;
4661 else if (*s != '[' && *s != '{')
4662 PL_lex_state = LEX_INTERPEND;
4665 if (PL_expect & XFAKEBRACK) {
4666 PL_expect &= XENUMMASK;
4668 return yylex(); /* ignore fake brackets */
4670 start_force(PL_curforce);
4672 curmad('X', newSVpvn(s-1,1));
4673 CURMAD('_', PL_thiswhite);
4678 PL_thistoken = newSVpvs("");
4686 if (PL_expect == XOPERATOR) {
4687 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4688 && isIDFIRST_lazy_if(s,UTF))
4690 CopLINE_dec(PL_curcop);
4691 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
4692 CopLINE_inc(PL_curcop);
4697 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4699 PL_expect = XOPERATOR;
4700 force_ident(PL_tokenbuf, '&');
4704 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4716 const char tmp = *s++;
4723 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4724 && strchr("+-*/%.^&|<",tmp))
4725 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4726 "Reversed %c= operator",(int)tmp);
4728 if (PL_expect == XSTATE && isALPHA(tmp) &&
4729 (s == PL_linestart+1 || s[-2] == '\n') )
4731 if (PL_in_eval && !PL_rsfp) {
4736 if (strnEQ(s,"=cut",4)) {
4752 PL_thiswhite = newSVpvs("");
4753 sv_catpvn(PL_thiswhite, PL_linestart,
4754 PL_bufend - PL_linestart);
4758 PL_doextract = TRUE;
4762 if (PL_lex_brackets < PL_lex_formbrack) {
4764 #ifdef PERL_STRICT_CR
4765 while (SPACE_OR_TAB(*t))
4767 while (SPACE_OR_TAB(*t) || *t == '\r')
4770 if (*t == '\n' || *t == '#') {
4779 if (PL_expect == XSTATE && s[1] == '!' && s[2] == '!') {
4785 const char tmp = *s++;
4787 /* was this !=~ where !~ was meant?
4788 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4790 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4791 const char *t = s+1;
4793 while (t < PL_bufend && isSPACE(*t))
4796 if (*t == '/' || *t == '?' ||
4797 ((*t == 'm' || *t == 's' || *t == 'y')
4798 && !isALNUM(t[1])) ||
4799 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4800 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4801 "!=~ should be !~");
4811 if (PL_expect != XOPERATOR) {
4812 if (s[1] != '<' && !strchr(s,'>'))
4815 s = scan_heredoc(s);
4817 s = scan_inputsymbol(s);
4818 TERM(sublex_start());
4824 SHop(OP_LEFT_SHIFT);
4838 const char tmp = *s++;
4840 SHop(OP_RIGHT_SHIFT);
4841 else if (tmp == '=')
4850 if (PL_expect == XOPERATOR) {
4851 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4853 deprecate_old(commaless_variable_list);
4854 return REPORT(','); /* grandfather non-comma-format format */
4858 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4859 PL_tokenbuf[0] = '@';
4860 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4861 sizeof PL_tokenbuf - 1, FALSE);
4862 if (PL_expect == XOPERATOR)
4863 no_op("Array length", s);
4864 if (!PL_tokenbuf[1])
4866 PL_expect = XOPERATOR;
4867 PL_pending_ident = '#';
4871 PL_tokenbuf[0] = '$';
4872 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4873 sizeof PL_tokenbuf - 1, FALSE);
4874 if (PL_expect == XOPERATOR)
4876 if (!PL_tokenbuf[1]) {
4878 yyerror("Final $ should be \\$ or $name");
4882 /* This kludge not intended to be bulletproof. */
4883 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4884 pl_yylval.opval = newSVOP(OP_CONST, 0,
4885 newSViv(CopARYBASE_get(&PL_compiling)));
4886 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4892 const char tmp = *s;
4893 if (PL_lex_state == LEX_NORMAL)
4896 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4897 && intuit_more(s)) {
4899 PL_tokenbuf[0] = '@';
4900 if (ckWARN(WARN_SYNTAX)) {
4903 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4906 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4907 while (t < PL_bufend && *t != ']')
4909 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4910 "Multidimensional syntax %.*s not supported",
4911 (int)((t - PL_bufptr) + 1), PL_bufptr);
4915 else if (*s == '{') {
4917 PL_tokenbuf[0] = '%';
4918 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4919 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4921 char tmpbuf[sizeof PL_tokenbuf];
4924 } while (isSPACE(*t));
4925 if (isIDFIRST_lazy_if(t,UTF)) {
4927 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4931 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4932 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4933 "You need to quote \"%s\"",
4940 PL_expect = XOPERATOR;
4941 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4942 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4943 if (!islop || PL_last_lop_op == OP_GREPSTART)
4944 PL_expect = XOPERATOR;
4945 else if (strchr("$@\"'`q", *s))
4946 PL_expect = XTERM; /* e.g. print $fh "foo" */
4947 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4948 PL_expect = XTERM; /* e.g. print $fh &sub */
4949 else if (isIDFIRST_lazy_if(s,UTF)) {
4950 char tmpbuf[sizeof PL_tokenbuf];
4952 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4953 if ((t2 = keyword(tmpbuf, len, 0))) {
4954 /* binary operators exclude handle interpretations */
4966 PL_expect = XTERM; /* e.g. print $fh length() */
4971 PL_expect = XTERM; /* e.g. print $fh subr() */
4974 else if (isDIGIT(*s))
4975 PL_expect = XTERM; /* e.g. print $fh 3 */
4976 else if (*s == '.' && isDIGIT(s[1]))
4977 PL_expect = XTERM; /* e.g. print $fh .3 */
4978 else if ((*s == '?' || *s == '-' || *s == '+')
4979 && !isSPACE(s[1]) && s[1] != '=')
4980 PL_expect = XTERM; /* e.g. print $fh -1 */
4981 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4983 PL_expect = XTERM; /* e.g. print $fh /.../
4984 XXX except DORDOR operator
4986 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4988 PL_expect = XTERM; /* print $fh <<"EOF" */
4991 PL_pending_ident = '$';
4995 if (PL_expect == XOPERATOR)
4997 PL_tokenbuf[0] = '@';
4998 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
4999 if (!PL_tokenbuf[1]) {
5002 if (PL_lex_state == LEX_NORMAL)
5004 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
5006 PL_tokenbuf[0] = '%';
5008 /* Warn about @ where they meant $. */
5009 if (*s == '[' || *s == '{') {
5010 if (ckWARN(WARN_SYNTAX)) {
5011 const char *t = s + 1;
5012 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
5014 if (*t == '}' || *t == ']') {
5016 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
5017 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
5018 "Scalar value %.*s better written as $%.*s",
5019 (int)(t-PL_bufptr), PL_bufptr,
5020 (int)(t-PL_bufptr-1), PL_bufptr+1);
5025 PL_pending_ident = '@';
5028 case '/': /* may be division, defined-or, or pattern */
5029 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
5033 case '?': /* may either be conditional or pattern */
5034 if (PL_expect == XSTATE && s[1] == '?' && s[2] == '?') {
5038 if (PL_expect == XOPERATOR) {
5046 /* A // operator. */
5056 /* Disable warning on "study /blah/" */
5057 if (PL_oldoldbufptr == PL_last_uni
5058 && (*PL_last_uni != 's' || s - PL_last_uni < 5
5059 || memNE(PL_last_uni, "study", 5)
5060 || isALNUM_lazy_if(PL_last_uni+5,UTF)
5063 s = scan_pat(s,OP_MATCH);
5064 TERM(sublex_start());
5068 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
5069 #ifdef PERL_STRICT_CR
5072 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
5074 && (s == PL_linestart || s[-1] == '\n') )
5076 PL_lex_formbrack = 0;
5080 if (PL_expect == XSTATE && s[1] == '.' && s[2] == '.') {
5084 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5090 pl_yylval.ival = OPf_SPECIAL;
5096 if (PL_expect != XOPERATOR)
5101 case '0': case '1': case '2': case '3': case '4':
5102 case '5': case '6': case '7': case '8': case '9':
5103 s = scan_num(s, &pl_yylval);
5104 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5105 if (PL_expect == XOPERATOR)
5110 s = scan_str(s,!!PL_madskills,FALSE);
5111 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5112 if (PL_expect == XOPERATOR) {
5113 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5115 deprecate_old(commaless_variable_list);
5116 return REPORT(','); /* grandfather non-comma-format format */
5123 pl_yylval.ival = OP_CONST;
5124 TERM(sublex_start());
5127 s = scan_str(s,!!PL_madskills,FALSE);
5128 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5129 if (PL_expect == XOPERATOR) {
5130 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5132 deprecate_old(commaless_variable_list);
5133 return REPORT(','); /* grandfather non-comma-format format */
5140 pl_yylval.ival = OP_CONST;
5141 /* FIXME. I think that this can be const if char *d is replaced by
5142 more localised variables. */
5143 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5144 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5145 pl_yylval.ival = OP_STRINGIFY;
5149 TERM(sublex_start());
5152 s = scan_str(s,!!PL_madskills,FALSE);
5153 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5154 if (PL_expect == XOPERATOR)
5155 no_op("Backticks",s);
5158 readpipe_override();
5159 TERM(sublex_start());
5163 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5164 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5166 if (PL_expect == XOPERATOR)
5167 no_op("Backslash",s);
5171 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5172 char *start = s + 2;
5173 while (isDIGIT(*start) || *start == '_')
5175 if (*start == '.' && isDIGIT(start[1])) {
5176 s = scan_num(s, &pl_yylval);
5179 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5180 else if (!isALPHA(*start) && (PL_expect == XTERM
5181 || PL_expect == XREF || PL_expect == XSTATE
5182 || PL_expect == XTERMORDORDOR)) {
5183 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5185 s = scan_num(s, &pl_yylval);
5192 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5234 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5236 /* Some keywords can be followed by any delimiter, including ':' */
5237 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5238 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5239 (PL_tokenbuf[0] == 'q' &&
5240 strchr("qwxr", PL_tokenbuf[1])))));
5242 /* x::* is just a word, unless x is "CORE" */
5243 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5247 while (d < PL_bufend && isSPACE(*d))
5248 d++; /* no comments skipped here, or s### is misparsed */
5250 /* Is this a label? */
5251 if (!tmp && PL_expect == XSTATE
5252 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5254 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5259 /* Check for keywords */
5260 tmp = keyword(PL_tokenbuf, len, 0);
5262 /* Is this a word before a => operator? */
5263 if (*d == '=' && d[1] == '>') {
5266 = (OP*)newSVOP(OP_CONST, 0,
5267 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5268 pl_yylval.opval->op_private = OPpCONST_BARE;
5272 if (tmp < 0) { /* second-class keyword? */
5273 GV *ogv = NULL; /* override (winner) */
5274 GV *hgv = NULL; /* hidden (loser) */
5275 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5277 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5280 if (GvIMPORTED_CV(gv))
5282 else if (! CvMETHOD(cv))
5286 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5287 (gv = *gvp) && isGV_with_GP(gv) &&
5288 GvCVu(gv) && GvIMPORTED_CV(gv))
5295 tmp = 0; /* overridden by import or by GLOBAL */
5298 && -tmp==KEY_lock /* XXX generalizable kludge */
5301 tmp = 0; /* any sub overrides "weak" keyword */
5303 else { /* no override */
5305 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5306 Perl_warner(aTHX_ packWARN(WARN_MISC),
5307 "dump() better written as CORE::dump()");
5311 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5312 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5313 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5314 "Ambiguous call resolved as CORE::%s(), %s",
5315 GvENAME(hgv), "qualify as such or use &");
5322 default: /* not a keyword */
5323 /* Trade off - by using this evil construction we can pull the
5324 variable gv into the block labelled keylookup. If not, then
5325 we have to give it function scope so that the goto from the
5326 earlier ':' case doesn't bypass the initialisation. */
5328 just_a_word_zero_gv:
5336 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5339 SV *nextPL_nextwhite = 0;
5343 /* Get the rest if it looks like a package qualifier */
5345 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5347 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5350 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5351 *s == '\'' ? "'" : "::");
5356 if (PL_expect == XOPERATOR) {
5357 if (PL_bufptr == PL_linestart) {
5358 CopLINE_dec(PL_curcop);
5359 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), "%s", PL_warn_nosemi);
5360 CopLINE_inc(PL_curcop);
5363 no_op("Bareword",s);
5366 /* Look for a subroutine with this name in current package,
5367 unless name is "Foo::", in which case Foo is a bearword
5368 (and a package name). */
5370 if (len > 2 && !PL_madskills &&
5371 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5373 if (ckWARN(WARN_BAREWORD)
5374 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5375 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5376 "Bareword \"%s\" refers to nonexistent package",
5379 PL_tokenbuf[len] = '\0';
5385 /* Mustn't actually add anything to a symbol table.
5386 But also don't want to "initialise" any placeholder
5387 constants that might already be there into full
5388 blown PVGVs with attached PVCV. */
5389 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5390 GV_NOADD_NOINIT, SVt_PVCV);
5395 /* if we saw a global override before, get the right name */
5398 sv = newSVpvs("CORE::GLOBAL::");
5399 sv_catpv(sv,PL_tokenbuf);
5402 /* If len is 0, newSVpv does strlen(), which is correct.
5403 If len is non-zero, then it will be the true length,
5404 and so the scalar will be created correctly. */
5405 sv = newSVpv(PL_tokenbuf,len);
5408 if (PL_madskills && !PL_thistoken) {
5409 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5410 PL_thistoken = newSVpvn(start,s - start);
5411 PL_realtokenstart = s - SvPVX(PL_linestr);
5415 /* Presume this is going to be a bareword of some sort. */
5418 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5419 pl_yylval.opval->op_private = OPpCONST_BARE;
5420 /* UTF-8 package name? */
5421 if (UTF && !IN_BYTES &&
5422 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5425 /* And if "Foo::", then that's what it certainly is. */
5430 /* Do the explicit type check so that we don't need to force
5431 the initialisation of the symbol table to have a real GV.
5432 Beware - gv may not really be a PVGV, cv may not really be
5433 a PVCV, (because of the space optimisations that gv_init
5434 understands) But they're true if for this symbol there is
5435 respectively a typeglob and a subroutine.
5437 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5438 /* Real typeglob, so get the real subroutine: */
5440 /* A proxy for a subroutine in this package? */
5441 : SvOK(gv) ? MUTABLE_CV(gv) : NULL)
5444 /* See if it's the indirect object for a list operator. */
5446 if (PL_oldoldbufptr &&
5447 PL_oldoldbufptr < PL_bufptr &&
5448 (PL_oldoldbufptr == PL_last_lop
5449 || PL_oldoldbufptr == PL_last_uni) &&
5450 /* NO SKIPSPACE BEFORE HERE! */
5451 (PL_expect == XREF ||
5452 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5454 bool immediate_paren = *s == '(';
5456 /* (Now we can afford to cross potential line boundary.) */
5457 s = SKIPSPACE2(s,nextPL_nextwhite);
5459 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5462 /* Two barewords in a row may indicate method call. */
5464 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5465 (tmp = intuit_method(s, gv, cv)))
5468 /* If not a declared subroutine, it's an indirect object. */
5469 /* (But it's an indir obj regardless for sort.) */
5470 /* Also, if "_" follows a filetest operator, it's a bareword */
5473 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5475 (PL_last_lop_op != OP_MAPSTART &&
5476 PL_last_lop_op != OP_GREPSTART))))
5477 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5478 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5481 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5486 PL_expect = XOPERATOR;
5489 s = SKIPSPACE2(s,nextPL_nextwhite);
5490 PL_nextwhite = nextPL_nextwhite;
5495 /* Is this a word before a => operator? */
5496 if (*s == '=' && s[1] == '>' && !pkgname) {
5498 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5499 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5500 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5504 /* If followed by a paren, it's certainly a subroutine. */
5509 while (SPACE_OR_TAB(*d))
5511 if (*d == ')' && (sv = gv_const_sv(gv))) {
5518 PL_nextwhite = PL_thiswhite;
5521 start_force(PL_curforce);
5523 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5524 PL_expect = XOPERATOR;
5527 PL_nextwhite = nextPL_nextwhite;
5528 curmad('X', PL_thistoken);
5529 PL_thistoken = newSVpvs("");
5537 /* If followed by var or block, call it a method (unless sub) */
5539 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5540 PL_last_lop = PL_oldbufptr;
5541 PL_last_lop_op = OP_METHOD;
5545 /* If followed by a bareword, see if it looks like indir obj. */
5548 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5549 && (tmp = intuit_method(s, gv, cv)))
5552 /* Not a method, so call it a subroutine (if defined) */
5555 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5556 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5557 "Ambiguous use of -%s resolved as -&%s()",
5558 PL_tokenbuf, PL_tokenbuf);
5559 /* Check for a constant sub */
5560 if ((sv = gv_const_sv(gv))) {
5562 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5563 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5564 pl_yylval.opval->op_private = 0;
5568 /* Resolve to GV now. */
5569 if (SvTYPE(gv) != SVt_PVGV) {
5570 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5571 assert (SvTYPE(gv) == SVt_PVGV);
5572 /* cv must have been some sort of placeholder, so
5573 now needs replacing with a real code reference. */
5577 op_free(pl_yylval.opval);
5578 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5579 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5580 PL_last_lop = PL_oldbufptr;
5581 PL_last_lop_op = OP_ENTERSUB;
5582 /* Is there a prototype? */
5590 const char *proto = SvPV_const(MUTABLE_SV(cv), protolen);
5593 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5595 while (*proto == ';')
5597 if (*proto == '&' && *s == '{') {
5599 sv_setpvs(PL_subname, "__ANON__");
5601 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5608 PL_nextwhite = PL_thiswhite;
5611 start_force(PL_curforce);
5612 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5615 PL_nextwhite = nextPL_nextwhite;
5616 curmad('X', PL_thistoken);
5617 PL_thistoken = newSVpvs("");
5624 /* Guess harder when madskills require "best effort". */
5625 if (PL_madskills && (!gv || !GvCVu(gv))) {
5626 int probable_sub = 0;
5627 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5629 else if (isALPHA(*s)) {
5633 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5634 if (!keyword(tmpbuf, tmplen, 0))
5637 while (d < PL_bufend && isSPACE(*d))
5639 if (*d == '=' && d[1] == '>')
5644 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5645 op_free(pl_yylval.opval);
5646 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5647 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5648 PL_last_lop = PL_oldbufptr;
5649 PL_last_lop_op = OP_ENTERSUB;
5650 PL_nextwhite = PL_thiswhite;
5652 start_force(PL_curforce);
5653 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5655 PL_nextwhite = nextPL_nextwhite;
5656 curmad('X', PL_thistoken);
5657 PL_thistoken = newSVpvs("");
5662 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5669 /* Call it a bare word */
5672 if (PL_hints & HINT_STRICT_SUBS)
5673 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5675 if (lastchar != '-') {
5676 if (ckWARN(WARN_RESERVED)) {
5680 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5681 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5688 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5689 && ckWARN_d(WARN_AMBIGUOUS)) {
5690 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5691 "Operator or semicolon missing before %c%s",
5692 lastchar, PL_tokenbuf);
5693 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5694 "Ambiguous use of %c resolved as operator %c",
5695 lastchar, lastchar);
5701 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5702 newSVpv(CopFILE(PL_curcop),0));
5706 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5707 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5710 case KEY___PACKAGE__:
5711 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5713 ? newSVhek(HvNAME_HEK(PL_curstash))
5720 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5721 const char *pname = "main";
5722 if (PL_tokenbuf[2] == 'D')
5723 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5724 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5728 GvIOp(gv) = newIO();
5729 IoIFP(GvIOp(gv)) = PL_rsfp;
5730 #if defined(HAS_FCNTL) && defined(F_SETFD)
5732 const int fd = PerlIO_fileno(PL_rsfp);
5733 fcntl(fd,F_SETFD,fd >= 3);
5736 /* Mark this internal pseudo-handle as clean */
5737 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5738 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5739 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5741 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5742 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5743 /* if the script was opened in binmode, we need to revert
5744 * it to text mode for compatibility; but only iff it has CRs
5745 * XXX this is a questionable hack at best. */
5746 if (PL_bufend-PL_bufptr > 2
5747 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5750 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5751 loc = PerlIO_tell(PL_rsfp);
5752 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5755 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5757 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5758 #endif /* NETWARE */
5759 #ifdef PERLIO_IS_STDIO /* really? */
5760 # if defined(__BORLANDC__)
5761 /* XXX see note in do_binmode() */
5762 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5766 PerlIO_seek(PL_rsfp, loc, 0);
5770 #ifdef PERLIO_LAYERS
5773 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5774 else if (PL_encoding) {
5781 XPUSHs(PL_encoding);
5783 call_method("name", G_SCALAR);
5787 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5788 Perl_form(aTHX_ ":encoding(%"SVf")",
5797 if (PL_realtokenstart >= 0) {
5798 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5800 PL_endwhite = newSVpvs("");
5801 sv_catsv(PL_endwhite, PL_thiswhite);
5803 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5804 PL_realtokenstart = -1;
5806 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5807 SvCUR(PL_endwhite))) != NULL) ;
5822 if (PL_expect == XSTATE) {
5829 if (*s == ':' && s[1] == ':') {
5832 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5833 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5834 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5837 else if (tmp == KEY_require || tmp == KEY_do)
5838 /* that's a way to remember we saw "CORE::" */
5851 LOP(OP_ACCEPT,XTERM);
5857 LOP(OP_ATAN2,XTERM);
5863 LOP(OP_BINMODE,XTERM);
5866 LOP(OP_BLESS,XTERM);
5875 /* When 'use switch' is in effect, continue has a dual
5876 life as a control operator. */
5878 if (!FEATURE_IS_ENABLED("switch"))
5881 /* We have to disambiguate the two senses of
5882 "continue". If the next token is a '{' then
5883 treat it as the start of a continue block;
5884 otherwise treat it as a control operator.
5896 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5913 if (!PL_cryptseen) {
5914 PL_cryptseen = TRUE;
5918 LOP(OP_CRYPT,XTERM);
5921 LOP(OP_CHMOD,XTERM);
5924 LOP(OP_CHOWN,XTERM);
5927 LOP(OP_CONNECT,XTERM);
5946 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5947 if (orig_keyword == KEY_do) {
5956 PL_hints |= HINT_BLOCK_SCOPE;
5966 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5967 LOP(OP_DBMOPEN,XTERM);
5973 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5980 pl_yylval.ival = CopLINE(PL_curcop);
5996 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
5997 UNIBRACK(OP_ENTEREVAL);
6011 case KEY_endhostent:
6017 case KEY_endservent:
6020 case KEY_endprotoent:
6031 pl_yylval.ival = CopLINE(PL_curcop);
6033 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
6036 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
6039 if ((PL_bufend - p) >= 3 &&
6040 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
6042 else if ((PL_bufend - p) >= 4 &&
6043 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
6046 if (isIDFIRST_lazy_if(p,UTF)) {
6047 p = scan_ident(p, PL_bufend,
6048 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
6052 Perl_croak(aTHX_ "Missing $ on loop variable");
6054 s = SvPVX(PL_linestr) + soff;
6060 LOP(OP_FORMLINE,XTERM);
6066 LOP(OP_FCNTL,XTERM);
6072 LOP(OP_FLOCK,XTERM);
6081 LOP(OP_GREPSTART, XREF);
6084 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6099 case KEY_getpriority:
6100 LOP(OP_GETPRIORITY,XTERM);
6102 case KEY_getprotobyname:
6105 case KEY_getprotobynumber:
6106 LOP(OP_GPBYNUMBER,XTERM);
6108 case KEY_getprotoent:
6120 case KEY_getpeername:
6121 UNI(OP_GETPEERNAME);
6123 case KEY_gethostbyname:
6126 case KEY_gethostbyaddr:
6127 LOP(OP_GHBYADDR,XTERM);
6129 case KEY_gethostent:
6132 case KEY_getnetbyname:
6135 case KEY_getnetbyaddr:
6136 LOP(OP_GNBYADDR,XTERM);
6141 case KEY_getservbyname:
6142 LOP(OP_GSBYNAME,XTERM);
6144 case KEY_getservbyport:
6145 LOP(OP_GSBYPORT,XTERM);
6147 case KEY_getservent:
6150 case KEY_getsockname:
6151 UNI(OP_GETSOCKNAME);
6153 case KEY_getsockopt:
6154 LOP(OP_GSOCKOPT,XTERM);
6169 pl_yylval.ival = CopLINE(PL_curcop);
6179 pl_yylval.ival = CopLINE(PL_curcop);
6183 LOP(OP_INDEX,XTERM);
6189 LOP(OP_IOCTL,XTERM);
6201 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6233 LOP(OP_LISTEN,XTERM);
6242 s = scan_pat(s,OP_MATCH);
6243 TERM(sublex_start());
6246 LOP(OP_MAPSTART, XREF);
6249 LOP(OP_MKDIR,XTERM);
6252 LOP(OP_MSGCTL,XTERM);
6255 LOP(OP_MSGGET,XTERM);
6258 LOP(OP_MSGRCV,XTERM);
6261 LOP(OP_MSGSND,XTERM);
6266 PL_in_my = (U16)tmp;
6268 if (isIDFIRST_lazy_if(s,UTF)) {
6272 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6273 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6275 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6276 if (!PL_in_my_stash) {
6279 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6283 if (PL_madskills) { /* just add type to declarator token */
6284 sv_catsv(PL_thistoken, PL_nextwhite);
6286 sv_catpvn(PL_thistoken, start, s - start);
6294 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6301 s = tokenize_use(0, s);
6305 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6312 if (isIDFIRST_lazy_if(s,UTF)) {
6314 for (d = s; isALNUM_lazy_if(d,UTF);)
6316 for (t=d; isSPACE(*t);)
6318 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6320 && !(t[0] == '=' && t[1] == '>')
6322 int parms_len = (int)(d-s);
6323 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6324 "Precedence problem: open %.*s should be open(%.*s)",
6325 parms_len, s, parms_len, s);
6331 pl_yylval.ival = OP_OR;
6341 LOP(OP_OPEN_DIR,XTERM);
6344 checkcomma(s,PL_tokenbuf,"filehandle");
6348 checkcomma(s,PL_tokenbuf,"filehandle");
6367 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6371 LOP(OP_PIPE_OP,XTERM);
6374 s = scan_str(s,!!PL_madskills,FALSE);
6377 pl_yylval.ival = OP_CONST;
6378 TERM(sublex_start());
6384 s = scan_str(s,!!PL_madskills,FALSE);
6387 PL_expect = XOPERATOR;
6389 if (SvCUR(PL_lex_stuff)) {
6392 d = SvPV_force(PL_lex_stuff, len);
6394 for (; isSPACE(*d) && len; --len, ++d)
6399 if (!warned && ckWARN(WARN_QW)) {
6400 for (; !isSPACE(*d) && len; --len, ++d) {
6402 Perl_warner(aTHX_ packWARN(WARN_QW),
6403 "Possible attempt to separate words with commas");
6406 else if (*d == '#') {
6407 Perl_warner(aTHX_ packWARN(WARN_QW),
6408 "Possible attempt to put comments in qw() list");
6414 for (; !isSPACE(*d) && len; --len, ++d)
6417 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6418 words = append_elem(OP_LIST, words,
6419 newSVOP(OP_CONST, 0, tokeq(sv)));
6423 start_force(PL_curforce);
6424 NEXTVAL_NEXTTOKE.opval = words;
6429 SvREFCNT_dec(PL_lex_stuff);
6430 PL_lex_stuff = NULL;
6436 s = scan_str(s,!!PL_madskills,FALSE);
6439 pl_yylval.ival = OP_STRINGIFY;
6440 if (SvIVX(PL_lex_stuff) == '\'')
6441 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6442 TERM(sublex_start());
6445 s = scan_pat(s,OP_QR);
6446 TERM(sublex_start());
6449 s = scan_str(s,!!PL_madskills,FALSE);
6452 readpipe_override();
6453 TERM(sublex_start());
6461 s = force_version(s, FALSE);
6463 else if (*s != 'v' || !isDIGIT(s[1])
6464 || (s = force_version(s, TRUE), *s == 'v'))
6466 *PL_tokenbuf = '\0';
6467 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6468 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6469 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6471 yyerror("<> should be quotes");
6473 if (orig_keyword == KEY_require) {
6481 PL_last_uni = PL_oldbufptr;
6482 PL_last_lop_op = OP_REQUIRE;
6484 return REPORT( (int)REQUIRE );
6490 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6494 LOP(OP_RENAME,XTERM);
6503 LOP(OP_RINDEX,XTERM);
6512 UNIDOR(OP_READLINE);
6515 UNIDOR(OP_BACKTICK);
6524 LOP(OP_REVERSE,XTERM);
6527 UNIDOR(OP_READLINK);
6534 if (pl_yylval.opval)
6535 TERM(sublex_start());
6537 TOKEN(1); /* force error */
6540 checkcomma(s,PL_tokenbuf,"filehandle");
6550 LOP(OP_SELECT,XTERM);
6556 LOP(OP_SEMCTL,XTERM);
6559 LOP(OP_SEMGET,XTERM);
6562 LOP(OP_SEMOP,XTERM);
6568 LOP(OP_SETPGRP,XTERM);
6570 case KEY_setpriority:
6571 LOP(OP_SETPRIORITY,XTERM);
6573 case KEY_sethostent:
6579 case KEY_setservent:
6582 case KEY_setprotoent:
6592 LOP(OP_SEEKDIR,XTERM);
6594 case KEY_setsockopt:
6595 LOP(OP_SSOCKOPT,XTERM);
6601 LOP(OP_SHMCTL,XTERM);
6604 LOP(OP_SHMGET,XTERM);
6607 LOP(OP_SHMREAD,XTERM);
6610 LOP(OP_SHMWRITE,XTERM);
6613 LOP(OP_SHUTDOWN,XTERM);
6622 LOP(OP_SOCKET,XTERM);
6624 case KEY_socketpair:
6625 LOP(OP_SOCKPAIR,XTERM);
6628 checkcomma(s,PL_tokenbuf,"subroutine name");
6630 if (*s == ';' || *s == ')') /* probably a close */
6631 Perl_croak(aTHX_ "sort is now a reserved word");
6633 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6637 LOP(OP_SPLIT,XTERM);
6640 LOP(OP_SPRINTF,XTERM);
6643 LOP(OP_SPLICE,XTERM);
6658 LOP(OP_SUBSTR,XTERM);
6664 char tmpbuf[sizeof PL_tokenbuf];
6665 SSize_t tboffset = 0;
6666 expectation attrful;
6667 bool have_name, have_proto;
6668 const int key = tmp;
6673 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6674 SV *subtoken = newSVpvn(tstart, s - tstart);
6678 s = SKIPSPACE2(s,tmpwhite);
6683 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6684 (*s == ':' && s[1] == ':'))
6687 SV *nametoke = NULL;
6691 attrful = XATTRBLOCK;
6692 /* remember buffer pos'n for later force_word */
6693 tboffset = s - PL_oldbufptr;
6694 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6697 nametoke = newSVpvn(s, d - s);
6699 if (memchr(tmpbuf, ':', len))
6700 sv_setpvn(PL_subname, tmpbuf, len);
6702 sv_setsv(PL_subname,PL_curstname);
6703 sv_catpvs(PL_subname,"::");
6704 sv_catpvn(PL_subname,tmpbuf,len);
6711 CURMAD('X', nametoke);
6712 CURMAD('_', tmpwhite);
6713 (void) force_word(PL_oldbufptr + tboffset, WORD,
6716 s = SKIPSPACE2(d,tmpwhite);
6723 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6724 PL_expect = XTERMBLOCK;
6725 attrful = XATTRTERM;
6726 sv_setpvs(PL_subname,"?");
6730 if (key == KEY_format) {
6732 PL_lex_formbrack = PL_lex_brackets + 1;
6734 PL_thistoken = subtoken;
6738 (void) force_word(PL_oldbufptr + tboffset, WORD,
6744 /* Look for a prototype */
6747 bool bad_proto = FALSE;
6748 bool in_brackets = FALSE;
6749 char greedy_proto = ' ';
6750 bool proto_after_greedy_proto = FALSE;
6751 bool must_be_last = FALSE;
6752 bool underscore = FALSE;
6753 bool seen_underscore = FALSE;
6754 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6756 s = scan_str(s,!!PL_madskills,FALSE);
6758 Perl_croak(aTHX_ "Prototype not terminated");
6759 /* strip spaces and check for bad characters */
6760 d = SvPVX(PL_lex_stuff);
6762 for (p = d; *p; ++p) {
6768 proto_after_greedy_proto = TRUE;
6769 if (!strchr("$@%*;[]&\\_", *p)) {
6781 else if ( *p == ']' ) {
6782 in_brackets = FALSE;
6784 else if ( (*p == '@' || *p == '%') &&
6785 ( tmp < 2 || d[tmp-2] != '\\' ) &&
6787 must_be_last = TRUE;
6790 else if ( *p == '_' ) {
6791 underscore = seen_underscore = TRUE;
6798 if (proto_after_greedy_proto)
6799 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6800 "Prototype after '%c' for %"SVf" : %s",
6801 greedy_proto, SVfARG(PL_subname), d);
6803 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6804 "Illegal character %sin prototype for %"SVf" : %s",
6805 seen_underscore ? "after '_' " : "",
6806 SVfARG(PL_subname), d);
6807 SvCUR_set(PL_lex_stuff, tmp);
6812 CURMAD('q', PL_thisopen);
6813 CURMAD('_', tmpwhite);
6814 CURMAD('=', PL_thisstuff);
6815 CURMAD('Q', PL_thisclose);
6816 NEXTVAL_NEXTTOKE.opval =
6817 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6818 PL_lex_stuff = NULL;
6821 s = SKIPSPACE2(s,tmpwhite);
6829 if (*s == ':' && s[1] != ':')
6830 PL_expect = attrful;
6831 else if (*s != '{' && key == KEY_sub) {
6833 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6835 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6842 curmad('^', newSVpvs(""));
6843 CURMAD('_', tmpwhite);
6847 PL_thistoken = subtoken;
6850 NEXTVAL_NEXTTOKE.opval =
6851 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6852 PL_lex_stuff = NULL;
6858 sv_setpvs(PL_subname, "__ANON__");
6860 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6864 (void) force_word(PL_oldbufptr + tboffset, WORD,
6873 LOP(OP_SYSTEM,XREF);
6876 LOP(OP_SYMLINK,XTERM);
6879 LOP(OP_SYSCALL,XTERM);
6882 LOP(OP_SYSOPEN,XTERM);
6885 LOP(OP_SYSSEEK,XTERM);
6888 LOP(OP_SYSREAD,XTERM);
6891 LOP(OP_SYSWRITE,XTERM);
6895 TERM(sublex_start());
6916 LOP(OP_TRUNCATE,XTERM);
6928 pl_yylval.ival = CopLINE(PL_curcop);
6932 pl_yylval.ival = CopLINE(PL_curcop);
6936 LOP(OP_UNLINK,XTERM);
6942 LOP(OP_UNPACK,XTERM);
6945 LOP(OP_UTIME,XTERM);
6951 LOP(OP_UNSHIFT,XTERM);
6954 s = tokenize_use(1, s);
6964 pl_yylval.ival = CopLINE(PL_curcop);
6968 pl_yylval.ival = CopLINE(PL_curcop);
6972 PL_hints |= HINT_BLOCK_SCOPE;
6979 LOP(OP_WAITPID,XTERM);
6988 ctl_l[0] = toCTRL('L');
6990 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6993 /* Make sure $^L is defined */
6994 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
6999 if (PL_expect == XOPERATOR)
7005 pl_yylval.ival = OP_XOR;
7010 TERM(sublex_start());
7015 #pragma segment Main
7019 S_pending_ident(pTHX)
7024 /* pit holds the identifier we read and pending_ident is reset */
7025 char pit = PL_pending_ident;
7026 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
7027 /* All routes through this function want to know if there is a colon. */
7028 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
7029 PL_pending_ident = 0;
7031 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
7032 DEBUG_T({ PerlIO_printf(Perl_debug_log,
7033 "### Pending identifier '%s'\n", PL_tokenbuf); });
7035 /* if we're in a my(), we can't allow dynamics here.
7036 $foo'bar has already been turned into $foo::bar, so
7037 just check for colons.
7039 if it's a legal name, the OP is a PADANY.
7042 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
7044 yyerror(Perl_form(aTHX_ "No package name allowed for "
7045 "variable %s in \"our\"",
7047 tmp = allocmy(PL_tokenbuf);
7051 yyerror(Perl_form(aTHX_ PL_no_myglob,
7052 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
7054 pl_yylval.opval = newOP(OP_PADANY, 0);
7055 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
7061 build the ops for accesses to a my() variable.
7063 Deny my($a) or my($b) in a sort block, *if* $a or $b is
7064 then used in a comparison. This catches most, but not
7065 all cases. For instance, it catches
7066 sort { my($a); $a <=> $b }
7068 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
7069 (although why you'd do that is anyone's guess).
7074 tmp = pad_findmy(PL_tokenbuf);
7075 if (tmp != NOT_IN_PAD) {
7076 /* might be an "our" variable" */
7077 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
7078 /* build ops for a bareword */
7079 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
7080 HEK * const stashname = HvNAME_HEK(stash);
7081 SV * const sym = newSVhek(stashname);
7082 sv_catpvs(sym, "::");
7083 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
7084 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
7085 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7088 ? (GV_ADDMULTI | GV_ADDINEVAL)
7091 ((PL_tokenbuf[0] == '$') ? SVt_PV
7092 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7097 /* if it's a sort block and they're naming $a or $b */
7098 if (PL_last_lop_op == OP_SORT &&
7099 PL_tokenbuf[0] == '$' &&
7100 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
7103 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
7104 d < PL_bufend && *d != '\n';
7107 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
7108 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
7114 pl_yylval.opval = newOP(OP_PADANY, 0);
7115 pl_yylval.opval->op_targ = tmp;
7121 Whine if they've said @foo in a doublequoted string,
7122 and @foo isn't a variable we can find in the symbol
7125 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7126 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7128 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7129 && ckWARN(WARN_AMBIGUOUS)
7130 /* DO NOT warn for @- and @+ */
7131 && !( PL_tokenbuf[2] == '\0' &&
7132 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7135 /* Downgraded from fatal to warning 20000522 mjd */
7136 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7137 "Possible unintended interpolation of %s in string",
7142 /* build ops for a bareword */
7143 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7145 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7147 PL_tokenbuf + 1, tokenbuf_len - 1,
7148 /* If the identifier refers to a stash, don't autovivify it.
7149 * Change 24660 had the side effect of causing symbol table
7150 * hashes to always be defined, even if they were freshly
7151 * created and the only reference in the entire program was
7152 * the single statement with the defined %foo::bar:: test.
7153 * It appears that all code in the wild doing this actually
7154 * wants to know whether sub-packages have been loaded, so
7155 * by avoiding auto-vivifying symbol tables, we ensure that
7156 * defined %foo::bar:: continues to be false, and the existing
7157 * tests still give the expected answers, even though what
7158 * they're actually testing has now changed subtly.
7160 (*PL_tokenbuf == '%'
7161 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7164 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7165 ((PL_tokenbuf[0] == '$') ? SVt_PV
7166 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7172 * The following code was generated by perl_keyword.pl.
7176 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7180 PERL_ARGS_ASSERT_KEYWORD;
7184 case 1: /* 5 tokens of length 1 */
7216 case 2: /* 18 tokens of length 2 */
7362 case 3: /* 29 tokens of length 3 */
7366 if (name[1] == 'N' &&
7429 if (name[1] == 'i' &&
7461 if (name[1] == 'o' &&
7470 if (name[1] == 'e' &&
7479 if (name[1] == 'n' &&
7488 if (name[1] == 'o' &&
7497 if (name[1] == 'a' &&
7506 if (name[1] == 'o' &&
7568 if (name[1] == 'e' &&
7582 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7608 if (name[1] == 'i' &&
7617 if (name[1] == 's' &&
7626 if (name[1] == 'e' &&
7635 if (name[1] == 'o' &&
7647 case 4: /* 41 tokens of length 4 */
7651 if (name[1] == 'O' &&
7661 if (name[1] == 'N' &&
7671 if (name[1] == 'i' &&
7681 if (name[1] == 'h' &&
7691 if (name[1] == 'u' &&
7704 if (name[2] == 'c' &&
7713 if (name[2] == 's' &&
7722 if (name[2] == 'a' &&
7758 if (name[1] == 'o' &&
7771 if (name[2] == 't' &&
7780 if (name[2] == 'o' &&
7789 if (name[2] == 't' &&
7798 if (name[2] == 'e' &&
7811 if (name[1] == 'o' &&
7824 if (name[2] == 'y' &&
7833 if (name[2] == 'l' &&
7849 if (name[2] == 's' &&
7858 if (name[2] == 'n' &&
7867 if (name[2] == 'c' &&
7880 if (name[1] == 'e' &&
7890 if (name[1] == 'p' &&
7903 if (name[2] == 'c' &&
7912 if (name[2] == 'p' &&
7921 if (name[2] == 's' &&
7937 if (name[2] == 'n' &&
8007 if (name[2] == 'r' &&
8016 if (name[2] == 'r' &&
8025 if (name[2] == 'a' &&
8041 if (name[2] == 'l' &&
8103 if (name[2] == 'e' &&
8106 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
8119 case 5: /* 39 tokens of length 5 */
8123 if (name[1] == 'E' &&
8134 if (name[1] == 'H' &&
8148 if (name[2] == 'a' &&
8158 if (name[2] == 'a' &&
8175 if (name[2] == 'e' &&
8185 if (name[2] == 'e' &&
8189 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8205 if (name[3] == 'i' &&
8214 if (name[3] == 'o' &&
8250 if (name[2] == 'o' &&
8260 if (name[2] == 'y' &&
8274 if (name[1] == 'l' &&
8288 if (name[2] == 'n' &&
8298 if (name[2] == 'o' &&
8312 if (name[1] == 'i' &&
8317 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8326 if (name[2] == 'd' &&
8336 if (name[2] == 'c' &&
8353 if (name[2] == 'c' &&
8363 if (name[2] == 't' &&
8377 if (name[1] == 'k' &&
8388 if (name[1] == 'r' &&
8402 if (name[2] == 's' &&
8412 if (name[2] == 'd' &&
8429 if (name[2] == 'm' &&
8439 if (name[2] == 'i' &&
8449 if (name[2] == 'e' &&
8459 if (name[2] == 'l' &&
8469 if (name[2] == 'a' &&
8482 if (name[3] == 't' &&
8485 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8491 if (name[3] == 'd' &&
8508 if (name[1] == 'i' &&
8522 if (name[2] == 'a' &&
8535 if (name[3] == 'e' &&
8570 if (name[2] == 'i' &&
8587 if (name[2] == 'i' &&
8597 if (name[2] == 'i' &&
8614 case 6: /* 33 tokens of length 6 */
8618 if (name[1] == 'c' &&
8633 if (name[2] == 'l' &&
8644 if (name[2] == 'r' &&
8659 if (name[1] == 'e' &&
8674 if (name[2] == 's' &&
8679 if(ckWARN_d(WARN_SYNTAX))
8680 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8686 if (name[2] == 'i' &&
8704 if (name[2] == 'l' &&
8715 if (name[2] == 'r' &&
8730 if (name[1] == 'm' &&
8745 if (name[2] == 'n' &&
8756 if (name[2] == 's' &&
8771 if (name[1] == 's' &&
8777 if (name[4] == 't' &&
8786 if (name[4] == 'e' &&
8795 if (name[4] == 'c' &&
8804 if (name[4] == 'n' &&
8820 if (name[1] == 'r' &&
8838 if (name[3] == 'a' &&
8848 if (name[3] == 'u' &&
8862 if (name[2] == 'n' &&
8880 if (name[2] == 'a' &&
8894 if (name[3] == 'e' &&
8907 if (name[4] == 't' &&
8916 if (name[4] == 'e' &&
8938 if (name[4] == 't' &&
8947 if (name[4] == 'e' &&
8963 if (name[2] == 'c' &&
8974 if (name[2] == 'l' &&
8985 if (name[2] == 'b' &&
8996 if (name[2] == 's' &&
9019 if (name[4] == 's' &&
9028 if (name[4] == 'n' &&
9041 if (name[3] == 'a' &&
9058 if (name[1] == 'a' &&
9073 case 7: /* 29 tokens of length 7 */
9077 if (name[1] == 'E' &&
9090 if (name[1] == '_' &&
9103 if (name[1] == 'i' &&
9110 return -KEY_binmode;
9116 if (name[1] == 'o' &&
9123 return -KEY_connect;
9132 if (name[2] == 'm' &&
9138 return -KEY_dbmopen;
9149 if (name[4] == 'u' &&
9153 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9159 if (name[4] == 'n' &&
9180 if (name[1] == 'o' &&
9193 if (name[1] == 'e' &&
9200 if (name[5] == 'r' &&
9203 return -KEY_getpgrp;
9209 if (name[5] == 'i' &&
9212 return -KEY_getppid;
9225 if (name[1] == 'c' &&
9232 return -KEY_lcfirst;
9238 if (name[1] == 'p' &&
9245 return -KEY_opendir;
9251 if (name[1] == 'a' &&
9269 if (name[3] == 'd' &&
9274 return -KEY_readdir;
9280 if (name[3] == 'u' &&
9291 if (name[3] == 'e' &&
9296 return -KEY_reverse;
9315 if (name[3] == 'k' &&
9320 return -KEY_seekdir;
9326 if (name[3] == 'p' &&
9331 return -KEY_setpgrp;
9341 if (name[2] == 'm' &&
9347 return -KEY_shmread;
9353 if (name[2] == 'r' &&
9359 return -KEY_sprintf;
9368 if (name[3] == 'l' &&
9373 return -KEY_symlink;
9382 if (name[4] == 'a' &&
9386 return -KEY_syscall;
9392 if (name[4] == 'p' &&
9396 return -KEY_sysopen;
9402 if (name[4] == 'e' &&
9406 return -KEY_sysread;
9412 if (name[4] == 'e' &&
9416 return -KEY_sysseek;
9434 if (name[1] == 'e' &&
9441 return -KEY_telldir;
9450 if (name[2] == 'f' &&
9456 return -KEY_ucfirst;
9462 if (name[2] == 's' &&
9468 return -KEY_unshift;
9478 if (name[1] == 'a' &&
9485 return -KEY_waitpid;
9494 case 8: /* 26 tokens of length 8 */
9498 if (name[1] == 'U' &&
9506 return KEY_AUTOLOAD;
9517 if (name[3] == 'A' &&
9523 return KEY___DATA__;
9529 if (name[3] == 'I' &&
9535 return -KEY___FILE__;
9541 if (name[3] == 'I' &&
9547 return -KEY___LINE__;
9563 if (name[2] == 'o' &&
9570 return -KEY_closedir;
9576 if (name[2] == 'n' &&
9583 return -KEY_continue;
9593 if (name[1] == 'b' &&
9601 return -KEY_dbmclose;
9607 if (name[1] == 'n' &&
9613 if (name[4] == 'r' &&
9618 return -KEY_endgrent;
9624 if (name[4] == 'w' &&
9629 return -KEY_endpwent;
9642 if (name[1] == 'o' &&
9650 return -KEY_formline;
9656 if (name[1] == 'e' &&
9667 if (name[6] == 'n' &&
9670 return -KEY_getgrent;
9676 if (name[6] == 'i' &&
9679 return -KEY_getgrgid;
9685 if (name[6] == 'a' &&
9688 return -KEY_getgrnam;
9701 if (name[4] == 'o' &&
9706 return -KEY_getlogin;
9717 if (name[6] == 'n' &&
9720 return -KEY_getpwent;
9726 if (name[6] == 'a' &&
9729 return -KEY_getpwnam;
9735 if (name[6] == 'i' &&
9738 return -KEY_getpwuid;
9758 if (name[1] == 'e' &&
9765 if (name[5] == 'i' &&
9772 return -KEY_readline;
9777 return -KEY_readlink;
9788 if (name[5] == 'i' &&
9792 return -KEY_readpipe;
9813 if (name[4] == 'r' &&
9818 return -KEY_setgrent;
9824 if (name[4] == 'w' &&
9829 return -KEY_setpwent;
9845 if (name[3] == 'w' &&
9851 return -KEY_shmwrite;
9857 if (name[3] == 't' &&
9863 return -KEY_shutdown;
9873 if (name[2] == 's' &&
9880 return -KEY_syswrite;
9890 if (name[1] == 'r' &&
9898 return -KEY_truncate;
9907 case 9: /* 9 tokens of length 9 */
9911 if (name[1] == 'N' &&
9920 return KEY_UNITCHECK;
9926 if (name[1] == 'n' &&
9935 return -KEY_endnetent;
9941 if (name[1] == 'e' &&
9950 return -KEY_getnetent;
9956 if (name[1] == 'o' &&
9965 return -KEY_localtime;
9971 if (name[1] == 'r' &&
9980 return KEY_prototype;
9986 if (name[1] == 'u' &&
9995 return -KEY_quotemeta;
10001 if (name[1] == 'e' &&
10010 return -KEY_rewinddir;
10016 if (name[1] == 'e' &&
10025 return -KEY_setnetent;
10031 if (name[1] == 'a' &&
10040 return -KEY_wantarray;
10049 case 10: /* 9 tokens of length 10 */
10053 if (name[1] == 'n' &&
10059 if (name[4] == 'o' &&
10066 return -KEY_endhostent;
10072 if (name[4] == 'e' &&
10079 return -KEY_endservent;
10092 if (name[1] == 'e' &&
10098 if (name[4] == 'o' &&
10105 return -KEY_gethostent;
10114 if (name[5] == 'r' &&
10120 return -KEY_getservent;
10126 if (name[5] == 'c' &&
10132 return -KEY_getsockopt;
10152 if (name[2] == 't')
10157 if (name[4] == 'o' &&
10164 return -KEY_sethostent;
10173 if (name[5] == 'r' &&
10179 return -KEY_setservent;
10185 if (name[5] == 'c' &&
10191 return -KEY_setsockopt;
10208 if (name[2] == 'c' &&
10217 return -KEY_socketpair;
10230 case 11: /* 8 tokens of length 11 */
10234 if (name[1] == '_' &&
10244 { /* __PACKAGE__ */
10245 return -KEY___PACKAGE__;
10251 if (name[1] == 'n' &&
10261 { /* endprotoent */
10262 return -KEY_endprotoent;
10268 if (name[1] == 'e' &&
10277 if (name[5] == 'e' &&
10283 { /* getpeername */
10284 return -KEY_getpeername;
10293 if (name[6] == 'o' &&
10298 { /* getpriority */
10299 return -KEY_getpriority;
10305 if (name[6] == 't' &&
10310 { /* getprotoent */
10311 return -KEY_getprotoent;
10325 if (name[4] == 'o' &&
10332 { /* getsockname */
10333 return -KEY_getsockname;
10346 if (name[1] == 'e' &&
10354 if (name[6] == 'o' &&
10359 { /* setpriority */
10360 return -KEY_setpriority;
10366 if (name[6] == 't' &&
10371 { /* setprotoent */
10372 return -KEY_setprotoent;
10388 case 12: /* 2 tokens of length 12 */
10389 if (name[0] == 'g' &&
10401 if (name[9] == 'd' &&
10404 { /* getnetbyaddr */
10405 return -KEY_getnetbyaddr;
10411 if (name[9] == 'a' &&
10414 { /* getnetbyname */
10415 return -KEY_getnetbyname;
10427 case 13: /* 4 tokens of length 13 */
10428 if (name[0] == 'g' &&
10435 if (name[4] == 'o' &&
10444 if (name[10] == 'd' &&
10447 { /* gethostbyaddr */
10448 return -KEY_gethostbyaddr;
10454 if (name[10] == 'a' &&
10457 { /* gethostbyname */
10458 return -KEY_gethostbyname;
10471 if (name[4] == 'e' &&
10480 if (name[10] == 'a' &&
10483 { /* getservbyname */
10484 return -KEY_getservbyname;
10490 if (name[10] == 'o' &&
10493 { /* getservbyport */
10494 return -KEY_getservbyport;
10513 case 14: /* 1 tokens of length 14 */
10514 if (name[0] == 'g' &&
10528 { /* getprotobyname */
10529 return -KEY_getprotobyname;
10534 case 16: /* 1 tokens of length 16 */
10535 if (name[0] == 'g' &&
10551 { /* getprotobynumber */
10552 return -KEY_getprotobynumber;
10566 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10570 PERL_ARGS_ASSERT_CHECKCOMMA;
10572 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10573 if (ckWARN(WARN_SYNTAX)) {
10576 for (w = s+2; *w && level; w++) {
10579 else if (*w == ')')
10582 while (isSPACE(*w))
10584 /* the list of chars below is for end of statements or
10585 * block / parens, boolean operators (&&, ||, //) and branch
10586 * constructs (or, and, if, until, unless, while, err, for).
10587 * Not a very solid hack... */
10588 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10589 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10590 "%s (...) interpreted as function",name);
10593 while (s < PL_bufend && isSPACE(*s))
10597 while (s < PL_bufend && isSPACE(*s))
10599 if (isIDFIRST_lazy_if(s,UTF)) {
10600 const char * const w = s++;
10601 while (isALNUM_lazy_if(s,UTF))
10603 while (s < PL_bufend && isSPACE(*s))
10607 if (keyword(w, s - w, 0))
10610 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10611 if (gv && GvCVu(gv))
10613 Perl_croak(aTHX_ "No comma allowed after %s", what);
10618 /* Either returns sv, or mortalizes sv and returns a new SV*.
10619 Best used as sv=new_constant(..., sv, ...).
10620 If s, pv are NULL, calls subroutine with one argument,
10621 and type is used with error messages only. */
10624 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10625 SV *sv, SV *pv, const char *type, STRLEN typelen)
10628 HV * const table = GvHV(PL_hintgv); /* ^H */
10632 const char *why1 = "", *why2 = "", *why3 = "";
10634 PERL_ARGS_ASSERT_NEW_CONSTANT;
10636 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10639 why2 = (const char *)
10640 (strEQ(key,"charnames")
10641 ? "(possibly a missing \"use charnames ...\")"
10643 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10644 (type ? type: "undef"), why2);
10646 /* This is convoluted and evil ("goto considered harmful")
10647 * but I do not understand the intricacies of all the different
10648 * failure modes of %^H in here. The goal here is to make
10649 * the most probable error message user-friendly. --jhi */
10654 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10655 (type ? type: "undef"), why1, why2, why3);
10657 yyerror(SvPVX_const(msg));
10661 cvp = hv_fetch(table, key, keylen, FALSE);
10662 if (!cvp || !SvOK(*cvp)) {
10665 why3 = "} is not defined";
10668 sv_2mortal(sv); /* Parent created it permanently */
10671 pv = newSVpvn_flags(s, len, SVs_TEMP);
10673 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10675 typesv = &PL_sv_undef;
10677 PUSHSTACKi(PERLSI_OVERLOAD);
10689 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10693 /* Check the eval first */
10694 if (!PL_in_eval && SvTRUE(ERRSV)) {
10695 sv_catpvs(ERRSV, "Propagated");
10696 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10698 res = SvREFCNT_inc_simple(sv);
10702 SvREFCNT_inc_simple_void(res);
10711 why1 = "Call to &{$^H{";
10713 why3 = "}} did not return a defined value";
10721 /* Returns a NUL terminated string, with the length of the string written to
10725 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10728 register char *d = dest;
10729 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10731 PERL_ARGS_ASSERT_SCAN_WORD;
10735 Perl_croak(aTHX_ ident_too_long);
10736 if (isALNUM(*s)) /* UTF handled below */
10738 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10743 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10747 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10748 char *t = s + UTF8SKIP(s);
10750 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10754 Perl_croak(aTHX_ ident_too_long);
10755 Copy(s, d, len, char);
10768 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10771 char *bracket = NULL;
10773 register char *d = dest;
10774 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10776 PERL_ARGS_ASSERT_SCAN_IDENT;
10781 while (isDIGIT(*s)) {
10783 Perl_croak(aTHX_ ident_too_long);
10790 Perl_croak(aTHX_ ident_too_long);
10791 if (isALNUM(*s)) /* UTF handled below */
10793 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10798 else if (*s == ':' && s[1] == ':') {
10802 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10803 char *t = s + UTF8SKIP(s);
10804 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10806 if (d + (t - s) > e)
10807 Perl_croak(aTHX_ ident_too_long);
10808 Copy(s, d, t - s, char);
10819 if (PL_lex_state != LEX_NORMAL)
10820 PL_lex_state = LEX_INTERPENDMAYBE;
10823 if (*s == '$' && s[1] &&
10824 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10837 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10842 if (isSPACE(s[-1])) {
10844 const char ch = *s++;
10845 if (!SPACE_OR_TAB(ch)) {
10851 if (isIDFIRST_lazy_if(d,UTF)) {
10855 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10856 end += UTF8SKIP(end);
10857 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10858 end += UTF8SKIP(end);
10860 Copy(s, d, end - s, char);
10865 while ((isALNUM(*s) || *s == ':') && d < e)
10868 Perl_croak(aTHX_ ident_too_long);
10871 while (s < send && SPACE_OR_TAB(*s))
10873 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10874 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10875 const char * const brack =
10877 ((*s == '[') ? "[...]" : "{...}");
10878 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10879 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10880 funny, dest, brack, funny, dest, brack);
10883 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10887 /* Handle extended ${^Foo} variables
10888 * 1999-02-27 mjd-perl-patch@plover.com */
10889 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10893 while (isALNUM(*s) && d < e) {
10897 Perl_croak(aTHX_ ident_too_long);
10902 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10903 PL_lex_state = LEX_INTERPEND;
10906 if (PL_lex_state == LEX_NORMAL) {
10907 if (ckWARN(WARN_AMBIGUOUS) &&
10908 (keyword(dest, d - dest, 0)
10909 || get_cvn_flags(dest, d - dest, 0)))
10913 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10914 "Ambiguous use of %c{%s} resolved to %c%s",
10915 funny, dest, funny, dest);
10920 s = bracket; /* let the parser handle it */
10924 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10925 PL_lex_state = LEX_INTERPEND;
10930 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10932 PERL_ARGS_ASSERT_PMFLAG;
10934 PERL_UNUSED_CONTEXT;
10936 const char c = (char)ch;
10938 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10939 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10940 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10941 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10942 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10948 S_scan_pat(pTHX_ char *start, I32 type)
10952 char *s = scan_str(start,!!PL_madskills,FALSE);
10953 const char * const valid_flags =
10954 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10959 PERL_ARGS_ASSERT_SCAN_PAT;
10962 const char * const delimiter = skipspace(start);
10966 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10967 : "Search pattern not terminated" ));
10970 pm = (PMOP*)newPMOP(type, 0);
10971 if (PL_multi_open == '?') {
10972 /* This is the only point in the code that sets PMf_ONCE: */
10973 pm->op_pmflags |= PMf_ONCE;
10975 /* Hence it's safe to do this bit of PMOP book-keeping here, which
10976 allows us to restrict the list needed by reset to just the ??
10978 assert(type != OP_TRANS);
10980 MAGIC *mg = mg_find((const SV *)PL_curstash, PERL_MAGIC_symtab);
10983 mg = sv_magicext(MUTABLE_SV(PL_curstash), 0, PERL_MAGIC_symtab, 0, 0,
10986 elements = mg->mg_len / sizeof(PMOP**);
10987 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
10988 ((PMOP**)mg->mg_ptr) [elements++] = pm;
10989 mg->mg_len = elements * sizeof(PMOP**);
10990 PmopSTASH_set(pm,PL_curstash);
10996 while (*s && strchr(valid_flags, *s))
10997 pmflag(&pm->op_pmflags,*s++);
10999 if (PL_madskills && modstart != s) {
11000 SV* tmptoken = newSVpvn(modstart, s - modstart);
11001 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
11004 /* issue a warning if /c is specified,but /g is not */
11005 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
11006 && ckWARN(WARN_REGEXP))
11008 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
11009 "Use of /c modifier is meaningless without /g" );
11012 PL_lex_op = (OP*)pm;
11013 pl_yylval.ival = OP_MATCH;
11018 S_scan_subst(pTHX_ char *start)
11029 PERL_ARGS_ASSERT_SCAN_SUBST;
11031 pl_yylval.ival = OP_NULL;
11033 s = scan_str(start,!!PL_madskills,FALSE);
11036 Perl_croak(aTHX_ "Substitution pattern not terminated");
11038 if (s[-1] == PL_multi_open)
11041 if (PL_madskills) {
11042 CURMAD('q', PL_thisopen);
11043 CURMAD('_', PL_thiswhite);
11044 CURMAD('E', PL_thisstuff);
11045 CURMAD('Q', PL_thisclose);
11046 PL_realtokenstart = s - SvPVX(PL_linestr);
11050 first_start = PL_multi_start;
11051 s = scan_str(s,!!PL_madskills,FALSE);
11053 if (PL_lex_stuff) {
11054 SvREFCNT_dec(PL_lex_stuff);
11055 PL_lex_stuff = NULL;
11057 Perl_croak(aTHX_ "Substitution replacement not terminated");
11059 PL_multi_start = first_start; /* so whole substitution is taken together */
11061 pm = (PMOP*)newPMOP(OP_SUBST, 0);
11064 if (PL_madskills) {
11065 CURMAD('z', PL_thisopen);
11066 CURMAD('R', PL_thisstuff);
11067 CURMAD('Z', PL_thisclose);
11073 if (*s == EXEC_PAT_MOD) {
11077 else if (strchr(S_PAT_MODS, *s))
11078 pmflag(&pm->op_pmflags,*s++);
11084 if (PL_madskills) {
11086 curmad('m', newSVpvn(modstart, s - modstart));
11087 append_madprops(PL_thismad, (OP*)pm, 0);
11091 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
11092 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
11096 SV * const repl = newSVpvs("");
11098 PL_sublex_info.super_bufptr = s;
11099 PL_sublex_info.super_bufend = PL_bufend;
11101 pm->op_pmflags |= PMf_EVAL;
11104 sv_catpvs(repl, "eval ");
11106 sv_catpvs(repl, "do ");
11108 sv_catpvs(repl, "{");
11109 sv_catsv(repl, PL_lex_repl);
11110 if (strchr(SvPVX(PL_lex_repl), '#'))
11111 sv_catpvs(repl, "\n");
11112 sv_catpvs(repl, "}");
11114 SvREFCNT_dec(PL_lex_repl);
11115 PL_lex_repl = repl;
11118 PL_lex_op = (OP*)pm;
11119 pl_yylval.ival = OP_SUBST;
11124 S_scan_trans(pTHX_ char *start)
11137 PERL_ARGS_ASSERT_SCAN_TRANS;
11139 pl_yylval.ival = OP_NULL;
11141 s = scan_str(start,!!PL_madskills,FALSE);
11143 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11145 if (s[-1] == PL_multi_open)
11148 if (PL_madskills) {
11149 CURMAD('q', PL_thisopen);
11150 CURMAD('_', PL_thiswhite);
11151 CURMAD('E', PL_thisstuff);
11152 CURMAD('Q', PL_thisclose);
11153 PL_realtokenstart = s - SvPVX(PL_linestr);
11157 s = scan_str(s,!!PL_madskills,FALSE);
11159 if (PL_lex_stuff) {
11160 SvREFCNT_dec(PL_lex_stuff);
11161 PL_lex_stuff = NULL;
11163 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11165 if (PL_madskills) {
11166 CURMAD('z', PL_thisopen);
11167 CURMAD('R', PL_thisstuff);
11168 CURMAD('Z', PL_thisclose);
11171 complement = del = squash = 0;
11178 complement = OPpTRANS_COMPLEMENT;
11181 del = OPpTRANS_DELETE;
11184 squash = OPpTRANS_SQUASH;
11193 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11194 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11195 o->op_private &= ~OPpTRANS_ALL;
11196 o->op_private |= del|squash|complement|
11197 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11198 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11201 pl_yylval.ival = OP_TRANS;
11204 if (PL_madskills) {
11206 curmad('m', newSVpvn(modstart, s - modstart));
11207 append_madprops(PL_thismad, o, 0);
11216 S_scan_heredoc(pTHX_ register char *s)
11220 I32 op_type = OP_SCALAR;
11224 const char *found_newline;
11228 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11230 I32 stuffstart = s - SvPVX(PL_linestr);
11233 PL_realtokenstart = -1;
11236 PERL_ARGS_ASSERT_SCAN_HEREDOC;
11240 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11244 while (SPACE_OR_TAB(*peek))
11246 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11249 s = delimcpy(d, e, s, PL_bufend, term, &len);
11259 if (!isALNUM_lazy_if(s,UTF))
11260 deprecate_old("bare << to mean <<\"\"");
11261 for (; isALNUM_lazy_if(s,UTF); s++) {
11266 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11267 Perl_croak(aTHX_ "Delimiter for here document is too long");
11270 len = d - PL_tokenbuf;
11273 if (PL_madskills) {
11274 tstart = PL_tokenbuf + !outer;
11275 PL_thisclose = newSVpvn(tstart, len - !outer);
11276 tstart = SvPVX(PL_linestr) + stuffstart;
11277 PL_thisopen = newSVpvn(tstart, s - tstart);
11278 stuffstart = s - SvPVX(PL_linestr);
11281 #ifndef PERL_STRICT_CR
11282 d = strchr(s, '\r');
11284 char * const olds = s;
11286 while (s < PL_bufend) {
11292 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11301 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11308 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11309 herewas = newSVpvn(s,PL_bufend-s);
11313 herewas = newSVpvn(s-1,found_newline-s+1);
11316 herewas = newSVpvn(s,found_newline-s);
11320 if (PL_madskills) {
11321 tstart = SvPVX(PL_linestr) + stuffstart;
11323 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11325 PL_thisstuff = newSVpvn(tstart, s - tstart);
11328 s += SvCUR(herewas);
11331 stuffstart = s - SvPVX(PL_linestr);
11337 tmpstr = newSV_type(SVt_PVIV);
11338 SvGROW(tmpstr, 80);
11339 if (term == '\'') {
11340 op_type = OP_CONST;
11341 SvIV_set(tmpstr, -1);
11343 else if (term == '`') {
11344 op_type = OP_BACKTICK;
11345 SvIV_set(tmpstr, '\\');
11349 PL_multi_start = CopLINE(PL_curcop);
11350 PL_multi_open = PL_multi_close = '<';
11351 term = *PL_tokenbuf;
11352 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11353 char * const bufptr = PL_sublex_info.super_bufptr;
11354 char * const bufend = PL_sublex_info.super_bufend;
11355 char * const olds = s - SvCUR(herewas);
11356 s = strchr(bufptr, '\n');
11360 while (s < bufend &&
11361 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11363 CopLINE_inc(PL_curcop);
11366 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11367 missingterm(PL_tokenbuf);
11369 sv_setpvn(herewas,bufptr,d-bufptr+1);
11370 sv_setpvn(tmpstr,d+1,s-d);
11372 sv_catpvn(herewas,s,bufend-s);
11373 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11380 while (s < PL_bufend &&
11381 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11383 CopLINE_inc(PL_curcop);
11385 if (s >= PL_bufend) {
11386 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11387 missingterm(PL_tokenbuf);
11389 sv_setpvn(tmpstr,d+1,s-d);
11391 if (PL_madskills) {
11393 sv_catpvn(PL_thisstuff, d + 1, s - d);
11395 PL_thisstuff = newSVpvn(d + 1, s - d);
11396 stuffstart = s - SvPVX(PL_linestr);
11400 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11402 sv_catpvn(herewas,s,PL_bufend-s);
11403 sv_setsv(PL_linestr,herewas);
11404 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11405 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11406 PL_last_lop = PL_last_uni = NULL;
11409 sv_setpvs(tmpstr,""); /* avoid "uninitialized" warning */
11410 while (s >= PL_bufend) { /* multiple line string? */
11412 if (PL_madskills) {
11413 tstart = SvPVX(PL_linestr) + stuffstart;
11415 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11417 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11421 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11422 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11423 missingterm(PL_tokenbuf);
11426 stuffstart = s - SvPVX(PL_linestr);
11428 CopLINE_inc(PL_curcop);
11429 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11430 PL_last_lop = PL_last_uni = NULL;
11431 #ifndef PERL_STRICT_CR
11432 if (PL_bufend - PL_linestart >= 2) {
11433 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11434 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11436 PL_bufend[-2] = '\n';
11438 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11440 else if (PL_bufend[-1] == '\r')
11441 PL_bufend[-1] = '\n';
11443 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11444 PL_bufend[-1] = '\n';
11446 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11447 update_debugger_info(PL_linestr, NULL, 0);
11448 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11449 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11450 *(SvPVX(PL_linestr) + off ) = ' ';
11451 sv_catsv(PL_linestr,herewas);
11452 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11453 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11457 sv_catsv(tmpstr,PL_linestr);
11462 PL_multi_end = CopLINE(PL_curcop);
11463 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11464 SvPV_shrink_to_cur(tmpstr);
11466 SvREFCNT_dec(herewas);
11468 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11470 else if (PL_encoding)
11471 sv_recode_to_utf8(tmpstr, PL_encoding);
11473 PL_lex_stuff = tmpstr;
11474 pl_yylval.ival = op_type;
11478 /* scan_inputsymbol
11479 takes: current position in input buffer
11480 returns: new position in input buffer
11481 side-effects: pl_yylval and lex_op are set.
11486 <FH> read from filehandle
11487 <pkg::FH> read from package qualified filehandle
11488 <pkg'FH> read from package qualified filehandle
11489 <$fh> read from filehandle in $fh
11490 <*.h> filename glob
11495 S_scan_inputsymbol(pTHX_ char *start)
11498 register char *s = start; /* current position in buffer */
11501 char *d = PL_tokenbuf; /* start of temp holding space */
11502 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11504 PERL_ARGS_ASSERT_SCAN_INPUTSYMBOL;
11506 end = strchr(s, '\n');
11509 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11511 /* die if we didn't have space for the contents of the <>,
11512 or if it didn't end, or if we see a newline
11515 if (len >= (I32)sizeof PL_tokenbuf)
11516 Perl_croak(aTHX_ "Excessively long <> operator");
11518 Perl_croak(aTHX_ "Unterminated <> operator");
11523 Remember, only scalar variables are interpreted as filehandles by
11524 this code. Anything more complex (e.g., <$fh{$num}>) will be
11525 treated as a glob() call.
11526 This code makes use of the fact that except for the $ at the front,
11527 a scalar variable and a filehandle look the same.
11529 if (*d == '$' && d[1]) d++;
11531 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11532 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11535 /* If we've tried to read what we allow filehandles to look like, and
11536 there's still text left, then it must be a glob() and not a getline.
11537 Use scan_str to pull out the stuff between the <> and treat it
11538 as nothing more than a string.
11541 if (d - PL_tokenbuf != len) {
11542 pl_yylval.ival = OP_GLOB;
11543 s = scan_str(start,!!PL_madskills,FALSE);
11545 Perl_croak(aTHX_ "Glob not terminated");
11549 bool readline_overriden = FALSE;
11552 /* we're in a filehandle read situation */
11555 /* turn <> into <ARGV> */
11557 Copy("ARGV",d,5,char);
11559 /* Check whether readline() is overriden */
11560 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11562 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11564 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11565 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11566 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11567 readline_overriden = TRUE;
11569 /* if <$fh>, create the ops to turn the variable into a
11573 /* try to find it in the pad for this block, otherwise find
11574 add symbol table ops
11576 const PADOFFSET tmp = pad_findmy(d);
11577 if (tmp != NOT_IN_PAD) {
11578 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11579 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11580 HEK * const stashname = HvNAME_HEK(stash);
11581 SV * const sym = sv_2mortal(newSVhek(stashname));
11582 sv_catpvs(sym, "::");
11583 sv_catpv(sym, d+1);
11588 OP * const o = newOP(OP_PADSV, 0);
11590 PL_lex_op = readline_overriden
11591 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11592 append_elem(OP_LIST, o,
11593 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11594 : (OP*)newUNOP(OP_READLINE, 0, o);
11603 ? (GV_ADDMULTI | GV_ADDINEVAL)
11606 PL_lex_op = readline_overriden
11607 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11608 append_elem(OP_LIST,
11609 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11610 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11611 : (OP*)newUNOP(OP_READLINE, 0,
11612 newUNOP(OP_RV2SV, 0,
11613 newGVOP(OP_GV, 0, gv)));
11615 if (!readline_overriden)
11616 PL_lex_op->op_flags |= OPf_SPECIAL;
11617 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11618 pl_yylval.ival = OP_NULL;
11621 /* If it's none of the above, it must be a literal filehandle
11622 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11624 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11625 PL_lex_op = readline_overriden
11626 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11627 append_elem(OP_LIST,
11628 newGVOP(OP_GV, 0, gv),
11629 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11630 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11631 pl_yylval.ival = OP_NULL;
11640 takes: start position in buffer
11641 keep_quoted preserve \ on the embedded delimiter(s)
11642 keep_delims preserve the delimiters around the string
11643 returns: position to continue reading from buffer
11644 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11645 updates the read buffer.
11647 This subroutine pulls a string out of the input. It is called for:
11648 q single quotes q(literal text)
11649 ' single quotes 'literal text'
11650 qq double quotes qq(interpolate $here please)
11651 " double quotes "interpolate $here please"
11652 qx backticks qx(/bin/ls -l)
11653 ` backticks `/bin/ls -l`
11654 qw quote words @EXPORT_OK = qw( func() $spam )
11655 m// regexp match m/this/
11656 s/// regexp substitute s/this/that/
11657 tr/// string transliterate tr/this/that/
11658 y/// string transliterate y/this/that/
11659 ($*@) sub prototypes sub foo ($)
11660 (stuff) sub attr parameters sub foo : attr(stuff)
11661 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11663 In most of these cases (all but <>, patterns and transliterate)
11664 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11665 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11666 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11669 It skips whitespace before the string starts, and treats the first
11670 character as the delimiter. If the delimiter is one of ([{< then
11671 the corresponding "close" character )]}> is used as the closing
11672 delimiter. It allows quoting of delimiters, and if the string has
11673 balanced delimiters ([{<>}]) it allows nesting.
11675 On success, the SV with the resulting string is put into lex_stuff or,
11676 if that is already non-NULL, into lex_repl. The second case occurs only
11677 when parsing the RHS of the special constructs s/// and tr/// (y///).
11678 For convenience, the terminating delimiter character is stuffed into
11683 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11686 SV *sv; /* scalar value: string */
11687 const char *tmps; /* temp string, used for delimiter matching */
11688 register char *s = start; /* current position in the buffer */
11689 register char term; /* terminating character */
11690 register char *to; /* current position in the sv's data */
11691 I32 brackets = 1; /* bracket nesting level */
11692 bool has_utf8 = FALSE; /* is there any utf8 content? */
11693 I32 termcode; /* terminating char. code */
11694 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11695 STRLEN termlen; /* length of terminating string */
11696 int last_off = 0; /* last position for nesting bracket */
11702 PERL_ARGS_ASSERT_SCAN_STR;
11704 /* skip space before the delimiter */
11710 if (PL_realtokenstart >= 0) {
11711 stuffstart = PL_realtokenstart;
11712 PL_realtokenstart = -1;
11715 stuffstart = start - SvPVX(PL_linestr);
11717 /* mark where we are, in case we need to report errors */
11720 /* after skipping whitespace, the next character is the terminator */
11723 termcode = termstr[0] = term;
11727 termcode = utf8_to_uvchr((U8*)s, &termlen);
11728 Copy(s, termstr, termlen, U8);
11729 if (!UTF8_IS_INVARIANT(term))
11733 /* mark where we are */
11734 PL_multi_start = CopLINE(PL_curcop);
11735 PL_multi_open = term;
11737 /* find corresponding closing delimiter */
11738 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11739 termcode = termstr[0] = term = tmps[5];
11741 PL_multi_close = term;
11743 /* create a new SV to hold the contents. 79 is the SV's initial length.
11744 What a random number. */
11745 sv = newSV_type(SVt_PVIV);
11747 SvIV_set(sv, termcode);
11748 (void)SvPOK_only(sv); /* validate pointer */
11750 /* move past delimiter and try to read a complete string */
11752 sv_catpvn(sv, s, termlen);
11755 tstart = SvPVX(PL_linestr) + stuffstart;
11756 if (!PL_thisopen && !keep_delims) {
11757 PL_thisopen = newSVpvn(tstart, s - tstart);
11758 stuffstart = s - SvPVX(PL_linestr);
11762 if (PL_encoding && !UTF) {
11766 int offset = s - SvPVX_const(PL_linestr);
11767 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11768 &offset, (char*)termstr, termlen);
11769 const char * const ns = SvPVX_const(PL_linestr) + offset;
11770 char * const svlast = SvEND(sv) - 1;
11772 for (; s < ns; s++) {
11773 if (*s == '\n' && !PL_rsfp)
11774 CopLINE_inc(PL_curcop);
11777 goto read_more_line;
11779 /* handle quoted delimiters */
11780 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11782 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11784 if ((svlast-1 - t) % 2) {
11785 if (!keep_quoted) {
11786 *(svlast-1) = term;
11788 SvCUR_set(sv, SvCUR(sv) - 1);
11793 if (PL_multi_open == PL_multi_close) {
11799 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11800 /* At here, all closes are "was quoted" one,
11801 so we don't check PL_multi_close. */
11803 if (!keep_quoted && *(t+1) == PL_multi_open)
11808 else if (*t == PL_multi_open)
11816 SvCUR_set(sv, w - SvPVX_const(sv));
11818 last_off = w - SvPVX(sv);
11819 if (--brackets <= 0)
11824 if (!keep_delims) {
11825 SvCUR_set(sv, SvCUR(sv) - 1);
11831 /* extend sv if need be */
11832 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11833 /* set 'to' to the next character in the sv's string */
11834 to = SvPVX(sv)+SvCUR(sv);
11836 /* if open delimiter is the close delimiter read unbridle */
11837 if (PL_multi_open == PL_multi_close) {
11838 for (; s < PL_bufend; s++,to++) {
11839 /* embedded newlines increment the current line number */
11840 if (*s == '\n' && !PL_rsfp)
11841 CopLINE_inc(PL_curcop);
11842 /* handle quoted delimiters */
11843 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11844 if (!keep_quoted && s[1] == term)
11846 /* any other quotes are simply copied straight through */
11850 /* terminate when run out of buffer (the for() condition), or
11851 have found the terminator */
11852 else if (*s == term) {
11855 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11858 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11864 /* if the terminator isn't the same as the start character (e.g.,
11865 matched brackets), we have to allow more in the quoting, and
11866 be prepared for nested brackets.
11869 /* read until we run out of string, or we find the terminator */
11870 for (; s < PL_bufend; s++,to++) {
11871 /* embedded newlines increment the line count */
11872 if (*s == '\n' && !PL_rsfp)
11873 CopLINE_inc(PL_curcop);
11874 /* backslashes can escape the open or closing characters */
11875 if (*s == '\\' && s+1 < PL_bufend) {
11876 if (!keep_quoted &&
11877 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11882 /* allow nested opens and closes */
11883 else if (*s == PL_multi_close && --brackets <= 0)
11885 else if (*s == PL_multi_open)
11887 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11892 /* terminate the copied string and update the sv's end-of-string */
11894 SvCUR_set(sv, to - SvPVX_const(sv));
11897 * this next chunk reads more into the buffer if we're not done yet
11901 break; /* handle case where we are done yet :-) */
11903 #ifndef PERL_STRICT_CR
11904 if (to - SvPVX_const(sv) >= 2) {
11905 if ((to[-2] == '\r' && to[-1] == '\n') ||
11906 (to[-2] == '\n' && to[-1] == '\r'))
11910 SvCUR_set(sv, to - SvPVX_const(sv));
11912 else if (to[-1] == '\r')
11915 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11920 /* if we're out of file, or a read fails, bail and reset the current
11921 line marker so we can report where the unterminated string began
11924 if (PL_madskills) {
11925 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11927 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11929 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11933 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11935 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11941 /* we read a line, so increment our line counter */
11942 CopLINE_inc(PL_curcop);
11944 /* update debugger info */
11945 if ((PERLDB_LINE || PERLDB_SAVESRC) && PL_curstash != PL_debstash)
11946 update_debugger_info(PL_linestr, NULL, 0);
11948 /* having changed the buffer, we must update PL_bufend */
11949 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11950 PL_last_lop = PL_last_uni = NULL;
11953 /* at this point, we have successfully read the delimited string */
11955 if (!PL_encoding || UTF) {
11957 if (PL_madskills) {
11958 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11959 const int len = s - tstart;
11961 sv_catpvn(PL_thisstuff, tstart, len);
11963 PL_thisstuff = newSVpvn(tstart, len);
11964 if (!PL_thisclose && !keep_delims)
11965 PL_thisclose = newSVpvn(s,termlen);
11970 sv_catpvn(sv, s, termlen);
11975 if (PL_madskills) {
11976 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11977 const int len = s - tstart - termlen;
11979 sv_catpvn(PL_thisstuff, tstart, len);
11981 PL_thisstuff = newSVpvn(tstart, len);
11982 if (!PL_thisclose && !keep_delims)
11983 PL_thisclose = newSVpvn(s - termlen,termlen);
11987 if (has_utf8 || PL_encoding)
11990 PL_multi_end = CopLINE(PL_curcop);
11992 /* if we allocated too much space, give some back */
11993 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11994 SvLEN_set(sv, SvCUR(sv) + 1);
11995 SvPV_renew(sv, SvLEN(sv));
11998 /* decide whether this is the first or second quoted string we've read
12011 takes: pointer to position in buffer
12012 returns: pointer to new position in buffer
12013 side-effects: builds ops for the constant in pl_yylval.op
12015 Read a number in any of the formats that Perl accepts:
12017 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
12018 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
12021 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
12023 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
12026 If it reads a number without a decimal point or an exponent, it will
12027 try converting the number to an integer and see if it can do so
12028 without loss of precision.
12032 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
12035 register const char *s = start; /* current position in buffer */
12036 register char *d; /* destination in temp buffer */
12037 register char *e; /* end of temp buffer */
12038 NV nv; /* number read, as a double */
12039 SV *sv = NULL; /* place to put the converted number */
12040 bool floatit; /* boolean: int or float? */
12041 const char *lastub = NULL; /* position of last underbar */
12042 static char const number_too_long[] = "Number too long";
12044 PERL_ARGS_ASSERT_SCAN_NUM;
12046 /* We use the first character to decide what type of number this is */
12050 Perl_croak(aTHX_ "panic: scan_num");
12052 /* if it starts with a 0, it could be an octal number, a decimal in
12053 0.13 disguise, or a hexadecimal number, or a binary number. */
12057 u holds the "number so far"
12058 shift the power of 2 of the base
12059 (hex == 4, octal == 3, binary == 1)
12060 overflowed was the number more than we can hold?
12062 Shift is used when we add a digit. It also serves as an "are
12063 we in octal/hex/binary?" indicator to disallow hex characters
12064 when in octal mode.
12069 bool overflowed = FALSE;
12070 bool just_zero = TRUE; /* just plain 0 or binary number? */
12071 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
12072 static const char* const bases[5] =
12073 { "", "binary", "", "octal", "hexadecimal" };
12074 static const char* const Bases[5] =
12075 { "", "Binary", "", "Octal", "Hexadecimal" };
12076 static const char* const maxima[5] =
12078 "0b11111111111111111111111111111111",
12082 const char *base, *Base, *max;
12084 /* check for hex */
12089 } else if (s[1] == 'b') {
12094 /* check for a decimal in disguise */
12095 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
12097 /* so it must be octal */
12104 if (ckWARN(WARN_SYNTAX))
12105 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12106 "Misplaced _ in number");
12110 base = bases[shift];
12111 Base = Bases[shift];
12112 max = maxima[shift];
12114 /* read the rest of the number */
12116 /* x is used in the overflow test,
12117 b is the digit we're adding on. */
12122 /* if we don't mention it, we're done */
12126 /* _ are ignored -- but warned about if consecutive */
12128 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12129 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12130 "Misplaced _ in number");
12134 /* 8 and 9 are not octal */
12135 case '8': case '9':
12137 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
12141 case '2': case '3': case '4':
12142 case '5': case '6': case '7':
12144 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12147 case '0': case '1':
12148 b = *s++ & 15; /* ASCII digit -> value of digit */
12152 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12153 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12154 /* make sure they said 0x */
12157 b = (*s++ & 7) + 9;
12159 /* Prepare to put the digit we have onto the end
12160 of the number so far. We check for overflows.
12166 x = u << shift; /* make room for the digit */
12168 if ((x >> shift) != u
12169 && !(PL_hints & HINT_NEW_BINARY)) {
12172 if (ckWARN_d(WARN_OVERFLOW))
12173 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12174 "Integer overflow in %s number",
12177 u = x | b; /* add the digit to the end */
12180 n *= nvshift[shift];
12181 /* If an NV has not enough bits in its
12182 * mantissa to represent an UV this summing of
12183 * small low-order numbers is a waste of time
12184 * (because the NV cannot preserve the
12185 * low-order bits anyway): we could just
12186 * remember when did we overflow and in the
12187 * end just multiply n by the right
12195 /* if we get here, we had success: make a scalar value from
12200 /* final misplaced underbar check */
12201 if (s[-1] == '_') {
12202 if (ckWARN(WARN_SYNTAX))
12203 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12208 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12209 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12210 "%s number > %s non-portable",
12216 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12217 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12218 "%s number > %s non-portable",
12223 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12224 sv = new_constant(start, s - start, "integer",
12225 sv, NULL, NULL, 0);
12226 else if (PL_hints & HINT_NEW_BINARY)
12227 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12232 handle decimal numbers.
12233 we're also sent here when we read a 0 as the first digit
12235 case '1': case '2': case '3': case '4': case '5':
12236 case '6': case '7': case '8': case '9': case '.':
12239 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12242 /* read next group of digits and _ and copy into d */
12243 while (isDIGIT(*s) || *s == '_') {
12244 /* skip underscores, checking for misplaced ones
12248 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12249 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12250 "Misplaced _ in number");
12254 /* check for end of fixed-length buffer */
12256 Perl_croak(aTHX_ number_too_long);
12257 /* if we're ok, copy the character */
12262 /* final misplaced underbar check */
12263 if (lastub && s == lastub + 1) {
12264 if (ckWARN(WARN_SYNTAX))
12265 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12268 /* read a decimal portion if there is one. avoid
12269 3..5 being interpreted as the number 3. followed
12272 if (*s == '.' && s[1] != '.') {
12277 if (ckWARN(WARN_SYNTAX))
12278 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12279 "Misplaced _ in number");
12283 /* copy, ignoring underbars, until we run out of digits.
12285 for (; isDIGIT(*s) || *s == '_'; s++) {
12286 /* fixed length buffer check */
12288 Perl_croak(aTHX_ number_too_long);
12290 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12291 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12292 "Misplaced _ in number");
12298 /* fractional part ending in underbar? */
12299 if (s[-1] == '_') {
12300 if (ckWARN(WARN_SYNTAX))
12301 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12302 "Misplaced _ in number");
12304 if (*s == '.' && isDIGIT(s[1])) {
12305 /* oops, it's really a v-string, but without the "v" */
12311 /* read exponent part, if present */
12312 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12316 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12317 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12319 /* stray preinitial _ */
12321 if (ckWARN(WARN_SYNTAX))
12322 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12323 "Misplaced _ in number");
12327 /* allow positive or negative exponent */
12328 if (*s == '+' || *s == '-')
12331 /* stray initial _ */
12333 if (ckWARN(WARN_SYNTAX))
12334 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12335 "Misplaced _ in number");
12339 /* read digits of exponent */
12340 while (isDIGIT(*s) || *s == '_') {
12343 Perl_croak(aTHX_ number_too_long);
12347 if (((lastub && s == lastub + 1) ||
12348 (!isDIGIT(s[1]) && s[1] != '_'))
12349 && ckWARN(WARN_SYNTAX))
12350 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12351 "Misplaced _ in number");
12358 /* make an sv from the string */
12362 We try to do an integer conversion first if no characters
12363 indicating "float" have been found.
12368 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12370 if (flags == IS_NUMBER_IN_UV) {
12372 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12375 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12376 if (uv <= (UV) IV_MIN)
12377 sv_setiv(sv, -(IV)uv);
12384 /* terminate the string */
12386 nv = Atof(PL_tokenbuf);
12391 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12392 const char *const key = floatit ? "float" : "integer";
12393 const STRLEN keylen = floatit ? 5 : 7;
12394 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12395 key, keylen, sv, NULL, NULL, 0);
12399 /* if it starts with a v, it could be a v-string */
12402 sv = newSV(5); /* preallocate storage space */
12403 s = scan_vstring(s, PL_bufend, sv);
12407 /* make the op for the constant and return */
12410 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12412 lvalp->opval = NULL;
12418 S_scan_formline(pTHX_ register char *s)
12421 register char *eol;
12423 SV * const stuff = newSVpvs("");
12424 bool needargs = FALSE;
12425 bool eofmt = FALSE;
12427 char *tokenstart = s;
12428 SV* savewhite = NULL;
12430 if (PL_madskills) {
12431 savewhite = PL_thiswhite;
12436 PERL_ARGS_ASSERT_SCAN_FORMLINE;
12438 while (!needargs) {
12441 #ifdef PERL_STRICT_CR
12442 while (SPACE_OR_TAB(*t))
12445 while (SPACE_OR_TAB(*t) || *t == '\r')
12448 if (*t == '\n' || t == PL_bufend) {
12453 if (PL_in_eval && !PL_rsfp) {
12454 eol = (char *) memchr(s,'\n',PL_bufend-s);
12459 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12461 for (t = s; t < eol; t++) {
12462 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12464 goto enough; /* ~~ must be first line in formline */
12466 if (*t == '@' || *t == '^')
12470 sv_catpvn(stuff, s, eol-s);
12471 #ifndef PERL_STRICT_CR
12472 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12473 char *end = SvPVX(stuff) + SvCUR(stuff);
12476 SvCUR_set(stuff, SvCUR(stuff) - 1);
12486 if (PL_madskills) {
12488 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12490 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12493 s = filter_gets(PL_linestr, PL_rsfp, 0);
12495 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12497 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12499 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12500 PL_last_lop = PL_last_uni = NULL;
12509 if (SvCUR(stuff)) {
12512 PL_lex_state = LEX_NORMAL;
12513 start_force(PL_curforce);
12514 NEXTVAL_NEXTTOKE.ival = 0;
12518 PL_lex_state = LEX_FORMLINE;
12520 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12522 else if (PL_encoding)
12523 sv_recode_to_utf8(stuff, PL_encoding);
12525 start_force(PL_curforce);
12526 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12528 start_force(PL_curforce);
12529 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12533 SvREFCNT_dec(stuff);
12535 PL_lex_formbrack = 0;
12539 if (PL_madskills) {
12541 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12543 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12544 PL_thiswhite = savewhite;
12551 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12554 const I32 oldsavestack_ix = PL_savestack_ix;
12555 CV* const outsidecv = PL_compcv;
12558 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12560 SAVEI32(PL_subline);
12561 save_item(PL_subname);
12562 SAVESPTR(PL_compcv);
12564 PL_compcv = MUTABLE_CV(newSV_type(is_format ? SVt_PVFM : SVt_PVCV));
12565 CvFLAGS(PL_compcv) |= flags;
12567 PL_subline = CopLINE(PL_curcop);
12568 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12569 CvOUTSIDE(PL_compcv) = MUTABLE_CV(SvREFCNT_inc_simple(outsidecv));
12570 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12572 return oldsavestack_ix;
12576 #pragma segment Perl_yylex
12579 Perl_yywarn(pTHX_ const char *const s)
12583 PERL_ARGS_ASSERT_YYWARN;
12585 PL_in_eval |= EVAL_WARNONLY;
12587 PL_in_eval &= ~EVAL_WARNONLY;
12592 Perl_yyerror(pTHX_ const char *const s)
12595 const char *where = NULL;
12596 const char *context = NULL;
12599 int yychar = PL_parser->yychar;
12601 PERL_ARGS_ASSERT_YYERROR;
12603 if (!yychar || (yychar == ';' && !PL_rsfp))
12605 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12606 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12607 PL_oldbufptr != PL_bufptr) {
12610 The code below is removed for NetWare because it abends/crashes on NetWare
12611 when the script has error such as not having the closing quotes like:
12612 if ($var eq "value)
12613 Checking of white spaces is anyway done in NetWare code.
12616 while (isSPACE(*PL_oldoldbufptr))
12619 context = PL_oldoldbufptr;
12620 contlen = PL_bufptr - PL_oldoldbufptr;
12622 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12623 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12626 The code below is removed for NetWare because it abends/crashes on NetWare
12627 when the script has error such as not having the closing quotes like:
12628 if ($var eq "value)
12629 Checking of white spaces is anyway done in NetWare code.
12632 while (isSPACE(*PL_oldbufptr))
12635 context = PL_oldbufptr;
12636 contlen = PL_bufptr - PL_oldbufptr;
12638 else if (yychar > 255)
12639 where = "next token ???";
12640 else if (yychar == -2) { /* YYEMPTY */
12641 if (PL_lex_state == LEX_NORMAL ||
12642 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12643 where = "at end of line";
12644 else if (PL_lex_inpat)
12645 where = "within pattern";
12647 where = "within string";
12650 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12652 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12653 else if (isPRINT_LC(yychar)) {
12654 const char string = yychar;
12655 sv_catpvn(where_sv, &string, 1);
12658 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12659 where = SvPVX_const(where_sv);
12661 msg = sv_2mortal(newSVpv(s, 0));
12662 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12663 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12665 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12667 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12668 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12669 Perl_sv_catpvf(aTHX_ msg,
12670 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12671 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12674 if (PL_in_eval & EVAL_WARNONLY) {
12675 if (ckWARN_d(WARN_SYNTAX))
12676 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12680 if (PL_error_count >= 10) {
12681 if (PL_in_eval && SvCUR(ERRSV))
12682 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12683 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12685 Perl_croak(aTHX_ "%s has too many errors.\n",
12686 OutCopFILE(PL_curcop));
12689 PL_in_my_stash = NULL;
12693 #pragma segment Main
12697 S_swallow_bom(pTHX_ U8 *s)
12700 const STRLEN slen = SvCUR(PL_linestr);
12702 PERL_ARGS_ASSERT_SWALLOW_BOM;
12706 if (s[1] == 0xFE) {
12707 /* UTF-16 little-endian? (or UTF32-LE?) */
12708 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12709 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12710 #ifndef PERL_NO_UTF16_FILTER
12711 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12714 if (PL_bufend > (char*)s) {
12718 filter_add(utf16rev_textfilter, NULL);
12719 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12720 utf16_to_utf8_reversed(s, news,
12721 PL_bufend - (char*)s - 1,
12723 sv_setpvn(PL_linestr, (const char*)news, newlen);
12725 s = (U8*)SvPVX(PL_linestr);
12726 Copy(news, s, newlen, U8);
12730 SvUTF8_on(PL_linestr);
12731 s = (U8*)SvPVX(PL_linestr);
12733 /* FIXME - is this a general bug fix? */
12736 PL_bufend = SvPVX(PL_linestr) + newlen;
12739 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12744 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12745 #ifndef PERL_NO_UTF16_FILTER
12746 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12749 if (PL_bufend > (char *)s) {
12753 filter_add(utf16_textfilter, NULL);
12754 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12755 utf16_to_utf8(s, news,
12756 PL_bufend - (char*)s,
12758 sv_setpvn(PL_linestr, (const char*)news, newlen);
12760 SvUTF8_on(PL_linestr);
12761 s = (U8*)SvPVX(PL_linestr);
12762 PL_bufend = SvPVX(PL_linestr) + newlen;
12765 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12770 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12771 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12772 s += 3; /* UTF-8 */
12778 if (s[2] == 0xFE && s[3] == 0xFF) {
12779 /* UTF-32 big-endian */
12780 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12783 else if (s[2] == 0 && s[3] != 0) {
12786 * are a good indicator of UTF-16BE. */
12787 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12793 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12794 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12795 s += 4; /* UTF-8 */
12801 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12804 * are a good indicator of UTF-16LE. */
12805 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12813 #ifndef PERL_NO_UTF16_FILTER
12815 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12818 const STRLEN old = SvCUR(sv);
12819 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12820 DEBUG_P(PerlIO_printf(Perl_debug_log,
12821 "utf16_textfilter(%p): %d %d (%d)\n",
12822 FPTR2DPTR(void *, utf16_textfilter),
12823 idx, maxlen, (int) count));
12827 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12828 Copy(SvPVX_const(sv), tmps, old, char);
12829 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12830 SvCUR(sv) - old, &newlen);
12831 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12833 DEBUG_P({sv_dump(sv);});
12838 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12841 const STRLEN old = SvCUR(sv);
12842 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12843 DEBUG_P(PerlIO_printf(Perl_debug_log,
12844 "utf16rev_textfilter(%p): %d %d (%d)\n",
12845 FPTR2DPTR(void *, utf16rev_textfilter),
12846 idx, maxlen, (int) count));
12850 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12851 Copy(SvPVX_const(sv), tmps, old, char);
12852 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12853 SvCUR(sv) - old, &newlen);
12854 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12856 DEBUG_P({ sv_dump(sv); });
12862 Returns a pointer to the next character after the parsed
12863 vstring, as well as updating the passed in sv.
12865 Function must be called like
12868 s = scan_vstring(s,e,sv);
12870 where s and e are the start and end of the string.
12871 The sv should already be large enough to store the vstring
12872 passed in, for performance reasons.
12877 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12880 const char *pos = s;
12881 const char *start = s;
12883 PERL_ARGS_ASSERT_SCAN_VSTRING;
12885 if (*pos == 'v') pos++; /* get past 'v' */
12886 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12888 if ( *pos != '.') {
12889 /* this may not be a v-string if followed by => */
12890 const char *next = pos;
12891 while (next < e && isSPACE(*next))
12893 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12894 /* return string not v-string */
12895 sv_setpvn(sv,(char *)s,pos-s);
12896 return (char *)pos;
12900 if (!isALPHA(*pos)) {
12901 U8 tmpbuf[UTF8_MAXBYTES+1];
12904 s++; /* get past 'v' */
12909 /* this is atoi() that tolerates underscores */
12912 const char *end = pos;
12914 while (--end >= s) {
12916 const UV orev = rev;
12917 rev += (*end - '0') * mult;
12919 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12920 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12921 "Integer overflow in decimal number");
12925 if (rev > 0x7FFFFFFF)
12926 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12928 /* Append native character for the rev point */
12929 tmpend = uvchr_to_utf8(tmpbuf, rev);
12930 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12931 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12933 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12939 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12943 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12951 * c-indentation-style: bsd
12952 * c-basic-offset: 4
12953 * indent-tabs-mode: t
12956 * ex: set ts=8 sts=4 sw=4 noet: