3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999,
4 * 2000, 2001, 2002, 2003, 2004, 2005, 2006, 2007, by Larry Wall and others
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
12 * "It all comes from here, the stench and the peril." --Frodo
16 * This file is the lexer for Perl. It's closely linked to the
19 * The main routine is yylex(), which returns the next token.
23 #define PERL_IN_TOKE_C
26 #define new_constant(a,b,c,d,e,f,g) \
27 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
29 #define pl_yylval (PL_parser->yylval)
31 /* YYINITDEPTH -- initial size of the parser's stacks. */
32 #define YYINITDEPTH 200
34 /* XXX temporary backwards compatibility */
35 #define PL_lex_brackets (PL_parser->lex_brackets)
36 #define PL_lex_brackstack (PL_parser->lex_brackstack)
37 #define PL_lex_casemods (PL_parser->lex_casemods)
38 #define PL_lex_casestack (PL_parser->lex_casestack)
39 #define PL_lex_defer (PL_parser->lex_defer)
40 #define PL_lex_dojoin (PL_parser->lex_dojoin)
41 #define PL_lex_expect (PL_parser->lex_expect)
42 #define PL_lex_formbrack (PL_parser->lex_formbrack)
43 #define PL_lex_inpat (PL_parser->lex_inpat)
44 #define PL_lex_inwhat (PL_parser->lex_inwhat)
45 #define PL_lex_op (PL_parser->lex_op)
46 #define PL_lex_repl (PL_parser->lex_repl)
47 #define PL_lex_starts (PL_parser->lex_starts)
48 #define PL_lex_stuff (PL_parser->lex_stuff)
49 #define PL_multi_start (PL_parser->multi_start)
50 #define PL_multi_open (PL_parser->multi_open)
51 #define PL_multi_close (PL_parser->multi_close)
52 #define PL_pending_ident (PL_parser->pending_ident)
53 #define PL_preambled (PL_parser->preambled)
54 #define PL_sublex_info (PL_parser->sublex_info)
55 #define PL_linestr (PL_parser->linestr)
56 #define PL_expect (PL_parser->expect)
57 #define PL_copline (PL_parser->copline)
58 #define PL_bufptr (PL_parser->bufptr)
59 #define PL_oldbufptr (PL_parser->oldbufptr)
60 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
61 #define PL_linestart (PL_parser->linestart)
62 #define PL_bufend (PL_parser->bufend)
63 #define PL_last_uni (PL_parser->last_uni)
64 #define PL_last_lop (PL_parser->last_lop)
65 #define PL_last_lop_op (PL_parser->last_lop_op)
66 #define PL_lex_state (PL_parser->lex_state)
67 #define PL_rsfp (PL_parser->rsfp)
68 #define PL_rsfp_filters (PL_parser->rsfp_filters)
69 #define PL_in_my (PL_parser->in_my)
70 #define PL_in_my_stash (PL_parser->in_my_stash)
71 #define PL_tokenbuf (PL_parser->tokenbuf)
72 #define PL_multi_end (PL_parser->multi_end)
73 #define PL_error_count (PL_parser->error_count)
76 # define PL_endwhite (PL_parser->endwhite)
77 # define PL_faketokens (PL_parser->faketokens)
78 # define PL_lasttoke (PL_parser->lasttoke)
79 # define PL_nextwhite (PL_parser->nextwhite)
80 # define PL_realtokenstart (PL_parser->realtokenstart)
81 # define PL_skipwhite (PL_parser->skipwhite)
82 # define PL_thisclose (PL_parser->thisclose)
83 # define PL_thismad (PL_parser->thismad)
84 # define PL_thisopen (PL_parser->thisopen)
85 # define PL_thisstuff (PL_parser->thisstuff)
86 # define PL_thistoken (PL_parser->thistoken)
87 # define PL_thiswhite (PL_parser->thiswhite)
88 # define PL_thiswhite (PL_parser->thiswhite)
89 # define PL_nexttoke (PL_parser->nexttoke)
90 # define PL_curforce (PL_parser->curforce)
92 # define PL_nexttoke (PL_parser->nexttoke)
93 # define PL_nexttype (PL_parser->nexttype)
94 # define PL_nextval (PL_parser->nextval)
98 S_pending_ident(pTHX);
100 static const char ident_too_long[] = "Identifier too long";
101 static const char commaless_variable_list[] = "comma-less variable list";
103 #ifndef PERL_NO_UTF16_FILTER
104 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
105 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
109 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
110 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
112 # define CURMAD(slot,sv)
113 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
116 #define XFAKEBRACK 128
117 #define XENUMMASK 127
119 #ifdef USE_UTF8_SCRIPTS
120 # define UTF (!IN_BYTES)
122 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
125 /* In variables named $^X, these are the legal values for X.
126 * 1999-02-27 mjd-perl-patch@plover.com */
127 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
129 /* On MacOS, respect nonbreaking spaces */
130 #ifdef MACOS_TRADITIONAL
131 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
133 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
136 /* LEX_* are values for PL_lex_state, the state of the lexer.
137 * They are arranged oddly so that the guard on the switch statement
138 * can get by with a single comparison (if the compiler is smart enough).
141 /* #define LEX_NOTPARSING 11 is done in perl.h. */
143 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
144 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
145 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
146 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
147 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
149 /* at end of code, eg "$x" followed by: */
150 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
151 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
153 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
154 string or after \E, $foo, etc */
155 #define LEX_INTERPCONST 2 /* NOT USED */
156 #define LEX_FORMLINE 1 /* expecting a format line */
157 #define LEX_KNOWNEXT 0 /* next token known; just return it */
161 static const char* const lex_state_names[] = {
180 #include "keywords.h"
182 /* CLINE is a macro that ensures PL_copline has a sane value */
187 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
190 # define SKIPSPACE0(s) skipspace0(s)
191 # define SKIPSPACE1(s) skipspace1(s)
192 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
193 # define PEEKSPACE(s) skipspace2(s,0)
195 # define SKIPSPACE0(s) skipspace(s)
196 # define SKIPSPACE1(s) skipspace(s)
197 # define SKIPSPACE2(s,tsv) skipspace(s)
198 # define PEEKSPACE(s) skipspace(s)
202 * Convenience functions to return different tokens and prime the
203 * lexer for the next token. They all take an argument.
205 * TOKEN : generic token (used for '(', DOLSHARP, etc)
206 * OPERATOR : generic operator
207 * AOPERATOR : assignment operator
208 * PREBLOCK : beginning the block after an if, while, foreach, ...
209 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
210 * PREREF : *EXPR where EXPR is not a simple identifier
211 * TERM : expression term
212 * LOOPX : loop exiting command (goto, last, dump, etc)
213 * FTST : file test operator
214 * FUN0 : zero-argument function
215 * FUN1 : not used, except for not, which isn't a UNIOP
216 * BOop : bitwise or or xor
218 * SHop : shift operator
219 * PWop : power operator
220 * PMop : pattern-matching operator
221 * Aop : addition-level operator
222 * Mop : multiplication-level operator
223 * Eop : equality-testing operator
224 * Rop : relational operator <= != gt
226 * Also see LOP and lop() below.
229 #ifdef DEBUGGING /* Serve -DT. */
230 # define REPORT(retval) tokereport((I32)retval, &pl_yylval)
232 # define REPORT(retval) (retval)
235 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
236 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
237 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
238 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
239 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
240 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
241 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
242 #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
243 #define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
244 #define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
245 #define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
246 #define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
247 #define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
248 #define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
249 #define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
250 #define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
251 #define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
252 #define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
253 #define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
254 #define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
256 /* This bit of chicanery makes a unary function followed by
257 * a parenthesis into a function with one argument, highest precedence.
258 * The UNIDOR macro is for unary functions that can be followed by the //
259 * operator (such as C<shift // 0>).
261 #define UNI2(f,x) { \
262 pl_yylval.ival = f; \
265 PL_last_uni = PL_oldbufptr; \
266 PL_last_lop_op = f; \
268 return REPORT( (int)FUNC1 ); \
270 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
272 #define UNI(f) UNI2(f,XTERM)
273 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
275 #define UNIBRACK(f) { \
276 pl_yylval.ival = f; \
278 PL_last_uni = PL_oldbufptr; \
280 return REPORT( (int)FUNC1 ); \
282 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
285 /* grandfather return to old style */
286 #define OLDLOP(f) return(pl_yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
290 /* how to interpret the pl_yylval associated with the token */
294 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
300 static struct debug_tokens {
302 enum token_type type;
304 } const debug_tokens[] =
306 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
307 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
308 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
309 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
310 { ARROW, TOKENTYPE_NONE, "ARROW" },
311 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
312 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
313 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
314 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
315 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
316 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
317 { DO, TOKENTYPE_NONE, "DO" },
318 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
319 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
320 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
321 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
322 { ELSE, TOKENTYPE_NONE, "ELSE" },
323 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
324 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
325 { FOR, TOKENTYPE_IVAL, "FOR" },
326 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
327 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
328 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
329 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
330 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
331 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
332 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
333 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
334 { IF, TOKENTYPE_IVAL, "IF" },
335 { LABEL, TOKENTYPE_PVAL, "LABEL" },
336 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
337 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
338 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
339 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
340 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
341 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
342 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
343 { MY, TOKENTYPE_IVAL, "MY" },
344 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
345 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
346 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
347 { OROP, TOKENTYPE_IVAL, "OROP" },
348 { OROR, TOKENTYPE_NONE, "OROR" },
349 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
350 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
351 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
352 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
353 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
354 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
355 { PREINC, TOKENTYPE_NONE, "PREINC" },
356 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
357 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
358 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
359 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
360 { SUB, TOKENTYPE_NONE, "SUB" },
361 { THING, TOKENTYPE_OPVAL, "THING" },
362 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
363 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
364 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
365 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
366 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
367 { USE, TOKENTYPE_IVAL, "USE" },
368 { WHEN, TOKENTYPE_IVAL, "WHEN" },
369 { WHILE, TOKENTYPE_IVAL, "WHILE" },
370 { WORD, TOKENTYPE_OPVAL, "WORD" },
371 { 0, TOKENTYPE_NONE, NULL }
374 /* dump the returned token in rv, plus any optional arg in pl_yylval */
377 S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
381 const char *name = NULL;
382 enum token_type type = TOKENTYPE_NONE;
383 const struct debug_tokens *p;
384 SV* const report = newSVpvs("<== ");
386 for (p = debug_tokens; p->token; p++) {
387 if (p->token == (int)rv) {
394 Perl_sv_catpv(aTHX_ report, name);
395 else if ((char)rv > ' ' && (char)rv < '~')
396 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
398 sv_catpvs(report, "EOF");
400 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
403 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
406 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
408 case TOKENTYPE_OPNUM:
409 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
410 PL_op_name[lvalp->ival]);
413 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
415 case TOKENTYPE_OPVAL:
417 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
418 PL_op_name[lvalp->opval->op_type]);
419 if (lvalp->opval->op_type == OP_CONST) {
420 Perl_sv_catpvf(aTHX_ report, " %s",
421 SvPEEK(cSVOPx_sv(lvalp->opval)));
426 sv_catpvs(report, "(opval=null)");
429 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
435 /* print the buffer with suitable escapes */
438 S_printbuf(pTHX_ const char *const fmt, const char *const s)
440 SV* const tmp = newSVpvs("");
441 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
450 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
451 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
455 S_ao(pTHX_ int toketype)
458 if (*PL_bufptr == '=') {
460 if (toketype == ANDAND)
461 pl_yylval.ival = OP_ANDASSIGN;
462 else if (toketype == OROR)
463 pl_yylval.ival = OP_ORASSIGN;
464 else if (toketype == DORDOR)
465 pl_yylval.ival = OP_DORASSIGN;
473 * When Perl expects an operator and finds something else, no_op
474 * prints the warning. It always prints "<something> found where
475 * operator expected. It prints "Missing semicolon on previous line?"
476 * if the surprise occurs at the start of the line. "do you need to
477 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
478 * where the compiler doesn't know if foo is a method call or a function.
479 * It prints "Missing operator before end of line" if there's nothing
480 * after the missing operator, or "... before <...>" if there is something
481 * after the missing operator.
485 S_no_op(pTHX_ const char *const what, char *s)
488 char * const oldbp = PL_bufptr;
489 const bool is_first = (PL_oldbufptr == PL_linestart);
495 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
496 if (ckWARN_d(WARN_SYNTAX)) {
498 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
499 "\t(Missing semicolon on previous line?)\n");
500 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
502 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
504 if (t < PL_bufptr && isSPACE(*t))
505 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
506 "\t(Do you need to predeclare %.*s?)\n",
507 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
511 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
512 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
520 * Complain about missing quote/regexp/heredoc terminator.
521 * If it's called with NULL then it cauterizes the line buffer.
522 * If we're in a delimited string and the delimiter is a control
523 * character, it's reformatted into a two-char sequence like ^C.
528 S_missingterm(pTHX_ char *s)
534 char * const nl = strrchr(s,'\n');
540 iscntrl(PL_multi_close)
542 PL_multi_close < 32 || PL_multi_close == 127
546 tmpbuf[1] = (char)toCTRL(PL_multi_close);
551 *tmpbuf = (char)PL_multi_close;
555 q = strchr(s,'"') ? '\'' : '"';
556 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
559 #define FEATURE_IS_ENABLED(name) \
560 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
561 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
562 /* The longest string we pass in. */
563 #define MAX_FEATURE_LEN (sizeof("switch")-1)
566 * S_feature_is_enabled
567 * Check whether the named feature is enabled.
570 S_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
573 HV * const hinthv = GvHV(PL_hintgv);
574 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
575 assert(namelen <= MAX_FEATURE_LEN);
576 memcpy(&he_name[8], name, namelen);
578 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
586 Perl_deprecate(pTHX_ const char *const s)
588 if (ckWARN(WARN_DEPRECATED))
589 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
593 Perl_deprecate_old(pTHX_ const char *const s)
595 /* This function should NOT be called for any new deprecated warnings */
596 /* Use Perl_deprecate instead */
598 /* It is here to maintain backward compatibility with the pre-5.8 */
599 /* warnings category hierarchy. The "deprecated" category used to */
600 /* live under the "syntax" category. It is now a top-level category */
601 /* in its own right. */
603 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
604 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
605 "Use of %s is deprecated", s);
609 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
610 * utf16-to-utf8-reversed.
613 #ifdef PERL_CR_FILTER
617 register const char *s = SvPVX_const(sv);
618 register const char * const e = s + SvCUR(sv);
619 /* outer loop optimized to do nothing if there are no CR-LFs */
621 if (*s++ == '\r' && *s == '\n') {
622 /* hit a CR-LF, need to copy the rest */
623 register char *d = s - 1;
626 if (*s == '\r' && s[1] == '\n')
637 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
639 const I32 count = FILTER_READ(idx+1, sv, maxlen);
640 if (count > 0 && !maxlen)
651 * Create a parser object and initialise its parser and lexer fields
653 * rsfp is the opened file handle to read from (if any),
655 * line holds any initial content already read from the file (or in
656 * the case of no file, such as an eval, the whole contents);
658 * new_filter indicates that this is a new file and it shouldn't inherit
659 * the filters from the current parser (ie require).
663 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
666 const char *s = NULL;
668 yy_parser *parser, *oparser;
670 /* create and initialise a parser */
672 Newxz(parser, 1, yy_parser);
673 parser->old_parser = oparser = PL_parser;
676 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
677 parser->ps = parser->stack;
678 parser->stack_size = YYINITDEPTH;
680 parser->stack->state = 0;
681 parser->yyerrstatus = 0;
682 parser->yychar = YYEMPTY; /* Cause a token to be read. */
684 /* on scope exit, free this parser and restore any outer one */
686 parser->saved_curcop = PL_curcop;
688 /* initialise lexer state */
691 parser->curforce = -1;
693 parser->nexttoke = 0;
695 parser->copline = NOLINE;
696 parser->lex_state = LEX_NORMAL;
697 parser->expect = XSTATE;
699 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
700 : (AV*)SvREFCNT_inc(oparser->rsfp_filters);
702 Newx(parser->lex_brackstack, 120, char);
703 Newx(parser->lex_casestack, 12, char);
704 *parser->lex_casestack = '\0';
707 s = SvPV_const(line, len);
713 parser->linestr = newSVpvs("\n;");
714 } else if (SvREADONLY(line) || s[len-1] != ';') {
715 parser->linestr = newSVsv(line);
717 sv_catpvs(parser->linestr, "\n;");
720 SvREFCNT_inc_simple_void_NN(line);
721 parser->linestr = line;
723 parser->oldoldbufptr =
726 parser->linestart = SvPVX(parser->linestr);
727 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
728 parser->last_lop = parser->last_uni = NULL;
732 /* delete a parser object */
735 Perl_parser_free(pTHX_ const yy_parser *parser)
737 PL_curcop = parser->saved_curcop;
738 SvREFCNT_dec(parser->linestr);
740 if (parser->rsfp == PerlIO_stdin())
741 PerlIO_clearerr(parser->rsfp);
742 else if (parser->rsfp && parser->old_parser
743 && parser->rsfp != parser->old_parser->rsfp)
744 PerlIO_close(parser->rsfp);
745 SvREFCNT_dec(parser->rsfp_filters);
747 Safefree(parser->stack);
748 Safefree(parser->lex_brackstack);
749 Safefree(parser->lex_casestack);
750 PL_parser = parser->old_parser;
757 * Finalizer for lexing operations. Must be called when the parser is
758 * done with the lexer.
765 PL_doextract = FALSE;
770 * This subroutine has nothing to do with tilting, whether at windmills
771 * or pinball tables. Its name is short for "increment line". It
772 * increments the current line number in CopLINE(PL_curcop) and checks
773 * to see whether the line starts with a comment of the form
774 * # line 500 "foo.pm"
775 * If so, it sets the current line number and file to the values in the comment.
779 S_incline(pTHX_ const char *s)
786 CopLINE_inc(PL_curcop);
789 while (SPACE_OR_TAB(*s))
791 if (strnEQ(s, "line", 4))
795 if (SPACE_OR_TAB(*s))
799 while (SPACE_OR_TAB(*s))
807 while (SPACE_OR_TAB(*s))
809 if (*s == '"' && (t = strchr(s+1, '"'))) {
819 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
821 if (*e != '\n' && *e != '\0')
822 return; /* false alarm */
825 const STRLEN len = t - s;
827 SV *const temp_sv = CopFILESV(PL_curcop);
833 tmplen = SvCUR(temp_sv);
839 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
840 /* must copy *{"::_<(eval N)[oldfilename:L]"}
841 * to *{"::_<newfilename"} */
842 /* However, the long form of evals is only turned on by the
843 debugger - usually they're "(eval %lu)" */
847 STRLEN tmplen2 = len;
848 if (tmplen + 2 <= sizeof smallbuf)
851 Newx(tmpbuf, tmplen + 2, char);
854 memcpy(tmpbuf + 2, cf, tmplen);
856 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
861 if (tmplen2 + 2 <= sizeof smallbuf)
864 Newx(tmpbuf2, tmplen2 + 2, char);
866 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
867 /* Either they malloc'd it, or we malloc'd it,
868 so no prefix is present in ours. */
873 memcpy(tmpbuf2 + 2, s, tmplen2);
876 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
878 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
879 /* adjust ${"::_<newfilename"} to store the new file name */
880 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
881 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
882 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
885 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
887 if (tmpbuf != smallbuf) Safefree(tmpbuf);
890 CopFILE_free(PL_curcop);
891 CopFILE_setn(PL_curcop, s, len);
893 CopLINE_set(PL_curcop, atoi(n)-1);
897 /* skip space before PL_thistoken */
900 S_skipspace0(pTHX_ register char *s)
907 PL_thiswhite = newSVpvs("");
908 sv_catsv(PL_thiswhite, PL_skipwhite);
909 sv_free(PL_skipwhite);
912 PL_realtokenstart = s - SvPVX(PL_linestr);
916 /* skip space after PL_thistoken */
919 S_skipspace1(pTHX_ register char *s)
921 const char *start = s;
922 I32 startoff = start - SvPVX(PL_linestr);
927 start = SvPVX(PL_linestr) + startoff;
928 if (!PL_thistoken && PL_realtokenstart >= 0) {
929 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
930 PL_thistoken = newSVpvn(tstart, start - tstart);
932 PL_realtokenstart = -1;
935 PL_nextwhite = newSVpvs("");
936 sv_catsv(PL_nextwhite, PL_skipwhite);
937 sv_free(PL_skipwhite);
944 S_skipspace2(pTHX_ register char *s, SV **svp)
947 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
948 const I32 startoff = s - SvPVX(PL_linestr);
951 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
952 if (!PL_madskills || !svp)
954 start = SvPVX(PL_linestr) + startoff;
955 if (!PL_thistoken && PL_realtokenstart >= 0) {
956 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
957 PL_thistoken = newSVpvn(tstart, start - tstart);
958 PL_realtokenstart = -1;
963 sv_setsv(*svp, PL_skipwhite);
964 sv_free(PL_skipwhite);
973 S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
975 AV *av = CopFILEAVx(PL_curcop);
977 SV * const sv = newSV_type(SVt_PVMG);
979 sv_setsv(sv, orig_sv);
981 sv_setpvn(sv, buf, len);
984 av_store(av, (I32)CopLINE(PL_curcop), sv);
990 * Called to gobble the appropriate amount and type of whitespace.
991 * Skips comments as well.
995 S_skipspace(pTHX_ register char *s)
1000 int startoff = s - SvPVX(PL_linestr);
1003 sv_free(PL_skipwhite);
1008 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
1009 while (s < PL_bufend && SPACE_OR_TAB(*s))
1019 SSize_t oldprevlen, oldoldprevlen;
1020 SSize_t oldloplen = 0, oldunilen = 0;
1021 while (s < PL_bufend && isSPACE(*s)) {
1022 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1027 if (s < PL_bufend && *s == '#') {
1028 while (s < PL_bufend && *s != '\n')
1030 if (s < PL_bufend) {
1032 if (PL_in_eval && !PL_rsfp) {
1039 /* only continue to recharge the buffer if we're at the end
1040 * of the buffer, we're not reading from a source filter, and
1041 * we're in normal lexing mode
1043 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1044 PL_lex_state == LEX_FORMLINE)
1051 /* try to recharge the buffer */
1053 curoff = s - SvPVX(PL_linestr);
1056 if ((s = filter_gets(PL_linestr, PL_rsfp,
1057 (prevlen = SvCUR(PL_linestr)))) == NULL)
1060 if (PL_madskills && curoff != startoff) {
1062 PL_skipwhite = newSVpvs("");
1063 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1067 /* mustn't throw out old stuff yet if madpropping */
1068 SvCUR(PL_linestr) = curoff;
1069 s = SvPVX(PL_linestr) + curoff;
1071 if (curoff && s[-1] == '\n')
1075 /* end of file. Add on the -p or -n magic */
1076 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1079 sv_catpvs(PL_linestr,
1080 ";}continue{print or die qq(-p destination: $!\\n);}");
1082 sv_setpvs(PL_linestr,
1083 ";}continue{print or die qq(-p destination: $!\\n);}");
1085 PL_minus_n = PL_minus_p = 0;
1087 else if (PL_minus_n) {
1089 sv_catpvn(PL_linestr, ";}", 2);
1091 sv_setpvn(PL_linestr, ";}", 2);
1097 sv_catpvn(PL_linestr,";", 1);
1099 sv_setpvn(PL_linestr,";", 1);
1102 /* reset variables for next time we lex */
1103 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1109 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1110 PL_last_lop = PL_last_uni = NULL;
1112 /* Close the filehandle. Could be from
1113 * STDIN, or a regular file. If we were reading code from
1114 * STDIN (because the commandline held no -e or filename)
1115 * then we don't close it, we reset it so the code can
1116 * read from STDIN too.
1119 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1120 PerlIO_clearerr(PL_rsfp);
1122 (void)PerlIO_close(PL_rsfp);
1127 /* not at end of file, so we only read another line */
1128 /* make corresponding updates to old pointers, for yyerror() */
1129 oldprevlen = PL_oldbufptr - PL_bufend;
1130 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1132 oldunilen = PL_last_uni - PL_bufend;
1134 oldloplen = PL_last_lop - PL_bufend;
1135 PL_linestart = PL_bufptr = s + prevlen;
1136 PL_bufend = s + SvCUR(PL_linestr);
1138 PL_oldbufptr = s + oldprevlen;
1139 PL_oldoldbufptr = s + oldoldprevlen;
1141 PL_last_uni = s + oldunilen;
1143 PL_last_lop = s + oldloplen;
1146 /* debugger active and we're not compiling the debugger code,
1147 * so store the line into the debugger's array of lines
1149 if (PERLDB_LINE && PL_curstash != PL_debstash)
1150 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1157 PL_skipwhite = newSVpvs("");
1158 curoff = s - SvPVX(PL_linestr);
1159 if (curoff - startoff)
1160 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1169 * Check the unary operators to ensure there's no ambiguity in how they're
1170 * used. An ambiguous piece of code would be:
1172 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1173 * the +5 is its argument.
1183 if (PL_oldoldbufptr != PL_last_uni)
1185 while (isSPACE(*PL_last_uni))
1188 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1190 if ((t = strchr(s, '(')) && t < PL_bufptr)
1193 if (ckWARN_d(WARN_AMBIGUOUS)){
1194 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1195 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1196 (int)(s - PL_last_uni), PL_last_uni);
1201 * LOP : macro to build a list operator. Its behaviour has been replaced
1202 * with a subroutine, S_lop() for which LOP is just another name.
1205 #define LOP(f,x) return lop(f,x,s)
1209 * Build a list operator (or something that might be one). The rules:
1210 * - if we have a next token, then it's a list operator [why?]
1211 * - if the next thing is an opening paren, then it's a function
1212 * - else it's a list operator
1216 S_lop(pTHX_ I32 f, int x, char *s)
1223 PL_last_lop = PL_oldbufptr;
1224 PL_last_lop_op = (OPCODE)f;
1227 return REPORT(LSTOP);
1230 return REPORT(LSTOP);
1233 return REPORT(FUNC);
1236 return REPORT(FUNC);
1238 return REPORT(LSTOP);
1244 * Sets up for an eventual force_next(). start_force(0) basically does
1245 * an unshift, while start_force(-1) does a push. yylex removes items
1250 S_start_force(pTHX_ int where)
1254 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1255 where = PL_lasttoke;
1256 assert(PL_curforce < 0 || PL_curforce == where);
1257 if (PL_curforce != where) {
1258 for (i = PL_lasttoke; i > where; --i) {
1259 PL_nexttoke[i] = PL_nexttoke[i-1];
1263 if (PL_curforce < 0) /* in case of duplicate start_force() */
1264 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1265 PL_curforce = where;
1268 curmad('^', newSVpvs(""));
1269 CURMAD('_', PL_nextwhite);
1274 S_curmad(pTHX_ char slot, SV *sv)
1280 if (PL_curforce < 0)
1281 where = &PL_thismad;
1283 where = &PL_nexttoke[PL_curforce].next_mad;
1286 sv_setpvn(sv, "", 0);
1289 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1291 else if (PL_encoding) {
1292 sv_recode_to_utf8(sv, PL_encoding);
1297 /* keep a slot open for the head of the list? */
1298 if (slot != '_' && *where && (*where)->mad_key == '^') {
1299 (*where)->mad_key = slot;
1300 sv_free((SV*)((*where)->mad_val));
1301 (*where)->mad_val = (void*)sv;
1304 addmad(newMADsv(slot, sv), where, 0);
1307 # define start_force(where) NOOP
1308 # define curmad(slot, sv) NOOP
1313 * When the lexer realizes it knows the next token (for instance,
1314 * it is reordering tokens for the parser) then it can call S_force_next
1315 * to know what token to return the next time the lexer is called. Caller
1316 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1317 * and possibly PL_expect to ensure the lexer handles the token correctly.
1321 S_force_next(pTHX_ I32 type)
1326 PerlIO_printf(Perl_debug_log, "### forced token:\n");
1327 tokereport(THING, &NEXTVAL_NEXTTOKE);
1331 if (PL_curforce < 0)
1332 start_force(PL_lasttoke);
1333 PL_nexttoke[PL_curforce].next_type = type;
1334 if (PL_lex_state != LEX_KNOWNEXT)
1335 PL_lex_defer = PL_lex_state;
1336 PL_lex_state = LEX_KNOWNEXT;
1337 PL_lex_expect = PL_expect;
1340 PL_nexttype[PL_nexttoke] = type;
1342 if (PL_lex_state != LEX_KNOWNEXT) {
1343 PL_lex_defer = PL_lex_state;
1344 PL_lex_expect = PL_expect;
1345 PL_lex_state = LEX_KNOWNEXT;
1351 S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
1354 SV * const sv = newSVpvn_utf8(start, len,
1356 && is_utf8_string((const U8*)start, len));
1362 * When the lexer knows the next thing is a word (for instance, it has
1363 * just seen -> and it knows that the next char is a word char, then
1364 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1368 * char *start : buffer position (must be within PL_linestr)
1369 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1370 * int check_keyword : if true, Perl checks to make sure the word isn't
1371 * a keyword (do this if the word is a label, e.g. goto FOO)
1372 * int allow_pack : if true, : characters will also be allowed (require,
1373 * use, etc. do this)
1374 * int allow_initial_tick : used by the "sub" lexer only.
1378 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1384 start = SKIPSPACE1(start);
1386 if (isIDFIRST_lazy_if(s,UTF) ||
1387 (allow_pack && *s == ':') ||
1388 (allow_initial_tick && *s == '\'') )
1390 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1391 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1393 start_force(PL_curforce);
1395 curmad('X', newSVpvn(start,s-start));
1396 if (token == METHOD) {
1401 PL_expect = XOPERATOR;
1405 curmad('g', newSVpvs( "forced" ));
1406 NEXTVAL_NEXTTOKE.opval
1407 = (OP*)newSVOP(OP_CONST,0,
1408 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1409 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1417 * Called when the lexer wants $foo *foo &foo etc, but the program
1418 * text only contains the "foo" portion. The first argument is a pointer
1419 * to the "foo", and the second argument is the type symbol to prefix.
1420 * Forces the next token to be a "WORD".
1421 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1425 S_force_ident(pTHX_ register const char *s, int kind)
1429 const STRLEN len = strlen(s);
1430 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1431 start_force(PL_curforce);
1432 NEXTVAL_NEXTTOKE.opval = o;
1435 o->op_private = OPpCONST_ENTERED;
1436 /* XXX see note in pp_entereval() for why we forgo typo
1437 warnings if the symbol must be introduced in an eval.
1439 gv_fetchpvn_flags(s, len,
1440 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1442 kind == '$' ? SVt_PV :
1443 kind == '@' ? SVt_PVAV :
1444 kind == '%' ? SVt_PVHV :
1452 Perl_str_to_version(pTHX_ SV *sv)
1457 const char *start = SvPV_const(sv,len);
1458 const char * const end = start + len;
1459 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1460 while (start < end) {
1464 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1469 retval += ((NV)n)/nshift;
1478 * Forces the next token to be a version number.
1479 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1480 * and if "guessing" is TRUE, then no new token is created (and the caller
1481 * must use an alternative parsing method).
1485 S_force_version(pTHX_ char *s, int guessing)
1491 I32 startoff = s - SvPVX(PL_linestr);
1500 while (isDIGIT(*d) || *d == '_' || *d == '.')
1504 start_force(PL_curforce);
1505 curmad('X', newSVpvn(s,d-s));
1508 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1510 s = scan_num(s, &pl_yylval);
1511 version = pl_yylval.opval;
1512 ver = cSVOPx(version)->op_sv;
1513 if (SvPOK(ver) && !SvNIOK(ver)) {
1514 SvUPGRADE(ver, SVt_PVNV);
1515 SvNV_set(ver, str_to_version(ver));
1516 SvNOK_on(ver); /* hint that it is a version */
1519 else if (guessing) {
1522 sv_free(PL_nextwhite); /* let next token collect whitespace */
1524 s = SvPVX(PL_linestr) + startoff;
1532 if (PL_madskills && !version) {
1533 sv_free(PL_nextwhite); /* let next token collect whitespace */
1535 s = SvPVX(PL_linestr) + startoff;
1538 /* NOTE: The parser sees the package name and the VERSION swapped */
1539 start_force(PL_curforce);
1540 NEXTVAL_NEXTTOKE.opval = version;
1548 * Tokenize a quoted string passed in as an SV. It finds the next
1549 * chunk, up to end of string or a backslash. It may make a new
1550 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1555 S_tokeq(pTHX_ SV *sv)
1559 register char *send;
1567 s = SvPV_force(sv, len);
1568 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1571 while (s < send && *s != '\\')
1576 if ( PL_hints & HINT_NEW_STRING ) {
1577 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
1581 if (s + 1 < send && (s[1] == '\\'))
1582 s++; /* all that, just for this */
1587 SvCUR_set(sv, d - SvPVX_const(sv));
1589 if ( PL_hints & HINT_NEW_STRING )
1590 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
1595 * Now come three functions related to double-quote context,
1596 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1597 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1598 * interact with PL_lex_state, and create fake ( ... ) argument lists
1599 * to handle functions and concatenation.
1600 * They assume that whoever calls them will be setting up a fake
1601 * join call, because each subthing puts a ',' after it. This lets
1604 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1606 * (I'm not sure whether the spurious commas at the end of lcfirst's
1607 * arguments and join's arguments are created or not).
1612 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1614 * Pattern matching will set PL_lex_op to the pattern-matching op to
1615 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
1617 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1619 * Everything else becomes a FUNC.
1621 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1622 * had an OP_CONST or OP_READLINE). This just sets us up for a
1623 * call to S_sublex_push().
1627 S_sublex_start(pTHX)
1630 register const I32 op_type = pl_yylval.ival;
1632 if (op_type == OP_NULL) {
1633 pl_yylval.opval = PL_lex_op;
1637 if (op_type == OP_CONST || op_type == OP_READLINE) {
1638 SV *sv = tokeq(PL_lex_stuff);
1640 if (SvTYPE(sv) == SVt_PVIV) {
1641 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1643 const char * const p = SvPV_const(sv, len);
1644 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
1648 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1649 PL_lex_stuff = NULL;
1650 /* Allow <FH> // "foo" */
1651 if (op_type == OP_READLINE)
1652 PL_expect = XTERMORDORDOR;
1655 else if (op_type == OP_BACKTICK && PL_lex_op) {
1656 /* readpipe() vas overriden */
1657 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
1658 pl_yylval.opval = PL_lex_op;
1660 PL_lex_stuff = NULL;
1664 PL_sublex_info.super_state = PL_lex_state;
1665 PL_sublex_info.sub_inwhat = (U16)op_type;
1666 PL_sublex_info.sub_op = PL_lex_op;
1667 PL_lex_state = LEX_INTERPPUSH;
1671 pl_yylval.opval = PL_lex_op;
1681 * Create a new scope to save the lexing state. The scope will be
1682 * ended in S_sublex_done. Returns a '(', starting the function arguments
1683 * to the uc, lc, etc. found before.
1684 * Sets PL_lex_state to LEX_INTERPCONCAT.
1693 PL_lex_state = PL_sublex_info.super_state;
1694 SAVEBOOL(PL_lex_dojoin);
1695 SAVEI32(PL_lex_brackets);
1696 SAVEI32(PL_lex_casemods);
1697 SAVEI32(PL_lex_starts);
1698 SAVEI8(PL_lex_state);
1699 SAVEVPTR(PL_lex_inpat);
1700 SAVEI16(PL_lex_inwhat);
1701 SAVECOPLINE(PL_curcop);
1702 SAVEPPTR(PL_bufptr);
1703 SAVEPPTR(PL_bufend);
1704 SAVEPPTR(PL_oldbufptr);
1705 SAVEPPTR(PL_oldoldbufptr);
1706 SAVEPPTR(PL_last_lop);
1707 SAVEPPTR(PL_last_uni);
1708 SAVEPPTR(PL_linestart);
1709 SAVESPTR(PL_linestr);
1710 SAVEGENERICPV(PL_lex_brackstack);
1711 SAVEGENERICPV(PL_lex_casestack);
1713 PL_linestr = PL_lex_stuff;
1714 PL_lex_stuff = NULL;
1716 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1717 = SvPVX(PL_linestr);
1718 PL_bufend += SvCUR(PL_linestr);
1719 PL_last_lop = PL_last_uni = NULL;
1720 SAVEFREESV(PL_linestr);
1722 PL_lex_dojoin = FALSE;
1723 PL_lex_brackets = 0;
1724 Newx(PL_lex_brackstack, 120, char);
1725 Newx(PL_lex_casestack, 12, char);
1726 PL_lex_casemods = 0;
1727 *PL_lex_casestack = '\0';
1729 PL_lex_state = LEX_INTERPCONCAT;
1730 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1732 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1733 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1734 PL_lex_inpat = PL_sublex_info.sub_op;
1736 PL_lex_inpat = NULL;
1743 * Restores lexer state after a S_sublex_push.
1750 if (!PL_lex_starts++) {
1751 SV * const sv = newSVpvs("");
1752 if (SvUTF8(PL_linestr))
1754 PL_expect = XOPERATOR;
1755 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1759 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1760 PL_lex_state = LEX_INTERPCASEMOD;
1764 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1765 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1766 PL_linestr = PL_lex_repl;
1768 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1769 PL_bufend += SvCUR(PL_linestr);
1770 PL_last_lop = PL_last_uni = NULL;
1771 SAVEFREESV(PL_linestr);
1772 PL_lex_dojoin = FALSE;
1773 PL_lex_brackets = 0;
1774 PL_lex_casemods = 0;
1775 *PL_lex_casestack = '\0';
1777 if (SvEVALED(PL_lex_repl)) {
1778 PL_lex_state = LEX_INTERPNORMAL;
1780 /* we don't clear PL_lex_repl here, so that we can check later
1781 whether this is an evalled subst; that means we rely on the
1782 logic to ensure sublex_done() is called again only via the
1783 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1786 PL_lex_state = LEX_INTERPCONCAT;
1796 PL_endwhite = newSVpvs("");
1797 sv_catsv(PL_endwhite, PL_thiswhite);
1801 sv_setpvn(PL_thistoken,"",0);
1803 PL_realtokenstart = -1;
1807 PL_bufend = SvPVX(PL_linestr);
1808 PL_bufend += SvCUR(PL_linestr);
1809 PL_expect = XOPERATOR;
1810 PL_sublex_info.sub_inwhat = 0;
1818 Extracts a pattern, double-quoted string, or transliteration. This
1821 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1822 processing a pattern (PL_lex_inpat is true), a transliteration
1823 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1825 Returns a pointer to the character scanned up to. If this is
1826 advanced from the start pointer supplied (i.e. if anything was
1827 successfully parsed), will leave an OP for the substring scanned
1828 in pl_yylval. Caller must intuit reason for not parsing further
1829 by looking at the next characters herself.
1833 double-quoted style: \r and \n
1834 regexp special ones: \D \s
1837 case and quoting: \U \Q \E
1838 stops on @ and $, but not for $ as tail anchor
1840 In transliterations:
1841 characters are VERY literal, except for - not at the start or end
1842 of the string, which indicates a range. If the range is in bytes,
1843 scan_const expands the range to the full set of intermediate
1844 characters. If the range is in utf8, the hyphen is replaced with
1845 a certain range mark which will be handled by pmtrans() in op.c.
1847 In double-quoted strings:
1849 double-quoted style: \r and \n
1851 deprecated backrefs: \1 (in substitution replacements)
1852 case and quoting: \U \Q \E
1855 scan_const does *not* construct ops to handle interpolated strings.
1856 It stops processing as soon as it finds an embedded $ or @ variable
1857 and leaves it to the caller to work out what's going on.
1859 embedded arrays (whether in pattern or not) could be:
1860 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1862 $ in double-quoted strings must be the symbol of an embedded scalar.
1864 $ in pattern could be $foo or could be tail anchor. Assumption:
1865 it's a tail anchor if $ is the last thing in the string, or if it's
1866 followed by one of "()| \r\n\t"
1868 \1 (backreferences) are turned into $1
1870 The structure of the code is
1871 while (there's a character to process) {
1872 handle transliteration ranges
1873 skip regexp comments /(?#comment)/ and codes /(?{code})/
1874 skip #-initiated comments in //x patterns
1875 check for embedded arrays
1876 check for embedded scalars
1878 leave intact backslashes from leaveit (below)
1879 deprecate \1 in substitution replacements
1880 handle string-changing backslashes \l \U \Q \E, etc.
1881 switch (what was escaped) {
1882 handle \- in a transliteration (becomes a literal -)
1883 handle \132 (octal characters)
1884 handle \x15 and \x{1234} (hex characters)
1885 handle \N{name} (named characters)
1886 handle \cV (control characters)
1887 handle printf-style backslashes (\f, \r, \n, etc)
1889 } (end if backslash)
1890 } (end while character to read)
1895 S_scan_const(pTHX_ char *start)
1898 register char *send = PL_bufend; /* end of the constant */
1899 SV *sv = newSV(send - start); /* sv for the constant */
1900 register char *s = start; /* start of the constant */
1901 register char *d = SvPVX(sv); /* destination for copies */
1902 bool dorange = FALSE; /* are we in a translit range? */
1903 bool didrange = FALSE; /* did we just finish a range? */
1904 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1905 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1908 UV literal_endpoint = 0;
1909 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1912 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1913 /* If we are doing a trans and we know we want UTF8 set expectation */
1914 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1915 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1919 while (s < send || dorange) {
1920 /* get transliterations out of the way (they're most literal) */
1921 if (PL_lex_inwhat == OP_TRANS) {
1922 /* expand a range A-Z to the full set of characters. AIE! */
1924 I32 i; /* current expanded character */
1925 I32 min; /* first character in range */
1926 I32 max; /* last character in range */
1937 char * const c = (char*)utf8_hop((U8*)d, -1);
1941 *c = (char)UTF_TO_NATIVE(0xff);
1942 /* mark the range as done, and continue */
1948 i = d - SvPVX_const(sv); /* remember current offset */
1951 SvLEN(sv) + (has_utf8 ?
1952 (512 - UTF_CONTINUATION_MARK +
1955 /* How many two-byte within 0..255: 128 in UTF-8,
1956 * 96 in UTF-8-mod. */
1958 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1960 d = SvPVX(sv) + i; /* refresh d after realloc */
1964 for (j = 0; j <= 1; j++) {
1965 char * const c = (char*)utf8_hop((U8*)d, -1);
1966 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
1972 max = (U8)0xff; /* only to \xff */
1973 uvmax = uv; /* \x{100} to uvmax */
1975 d = c; /* eat endpoint chars */
1980 d -= 2; /* eat the first char and the - */
1981 min = (U8)*d; /* first char in range */
1982 max = (U8)d[1]; /* last char in range */
1989 "Invalid range \"%c-%c\" in transliteration operator",
1990 (char)min, (char)max);
1994 if (literal_endpoint == 2 &&
1995 ((isLOWER(min) && isLOWER(max)) ||
1996 (isUPPER(min) && isUPPER(max)))) {
1998 for (i = min; i <= max; i++)
2000 *d++ = NATIVE_TO_NEED(has_utf8,i);
2002 for (i = min; i <= max; i++)
2004 *d++ = NATIVE_TO_NEED(has_utf8,i);
2009 for (i = min; i <= max; i++)
2012 const U8 ch = (U8)NATIVE_TO_UTF(i);
2013 if (UNI_IS_INVARIANT(ch))
2016 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2017 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2026 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2028 *d++ = (char)UTF_TO_NATIVE(0xff);
2030 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2034 /* mark the range as done, and continue */
2038 literal_endpoint = 0;
2043 /* range begins (ignore - as first or last char) */
2044 else if (*s == '-' && s+1 < send && s != start) {
2046 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2053 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2063 literal_endpoint = 0;
2064 native_range = TRUE;
2069 /* if we get here, we're not doing a transliteration */
2071 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2072 except for the last char, which will be done separately. */
2073 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2075 while (s+1 < send && *s != ')')
2076 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2078 else if (s[2] == '{' /* This should match regcomp.c */
2079 || (s[2] == '?' && s[3] == '{'))
2082 char *regparse = s + (s[2] == '{' ? 3 : 4);
2085 while (count && (c = *regparse)) {
2086 if (c == '\\' && regparse[1])
2094 if (*regparse != ')')
2095 regparse--; /* Leave one char for continuation. */
2096 while (s < regparse)
2097 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2101 /* likewise skip #-initiated comments in //x patterns */
2102 else if (*s == '#' && PL_lex_inpat &&
2103 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2104 while (s+1 < send && *s != '\n')
2105 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2108 /* check for embedded arrays
2109 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2111 else if (*s == '@' && s[1]) {
2112 if (isALNUM_lazy_if(s+1,UTF))
2114 if (strchr(":'{$", s[1]))
2116 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2117 break; /* in regexp, neither @+ nor @- are interpolated */
2120 /* check for embedded scalars. only stop if we're sure it's a
2123 else if (*s == '$') {
2124 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2126 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
2127 break; /* in regexp, $ might be tail anchor */
2130 /* End of else if chain - OP_TRANS rejoin rest */
2133 if (*s == '\\' && s+1 < send) {
2136 /* deprecate \1 in strings and substitution replacements */
2137 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2138 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2140 if (ckWARN(WARN_SYNTAX))
2141 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2146 /* string-change backslash escapes */
2147 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2151 /* skip any other backslash escapes in a pattern */
2152 else if (PL_lex_inpat) {
2153 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2154 goto default_action;
2157 /* if we get here, it's either a quoted -, or a digit */
2160 /* quoted - in transliterations */
2162 if (PL_lex_inwhat == OP_TRANS) {
2169 if ((isALPHA(*s) || isDIGIT(*s)) &&
2171 Perl_warner(aTHX_ packWARN(WARN_MISC),
2172 "Unrecognized escape \\%c passed through",
2174 /* default action is to copy the quoted character */
2175 goto default_action;
2178 /* \132 indicates an octal constant */
2179 case '0': case '1': case '2': case '3':
2180 case '4': case '5': case '6': case '7':
2184 uv = grok_oct(s, &len, &flags, NULL);
2187 goto NUM_ESCAPE_INSERT;
2189 /* \x24 indicates a hex constant */
2193 char* const e = strchr(s, '}');
2194 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2195 PERL_SCAN_DISALLOW_PREFIX;
2200 yyerror("Missing right brace on \\x{}");
2204 uv = grok_hex(s, &len, &flags, NULL);
2210 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2211 uv = grok_hex(s, &len, &flags, NULL);
2217 /* Insert oct or hex escaped character.
2218 * There will always enough room in sv since such
2219 * escapes will be longer than any UTF-8 sequence
2220 * they can end up as. */
2222 /* We need to map to chars to ASCII before doing the tests
2225 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2226 if (!has_utf8 && uv > 255) {
2227 /* Might need to recode whatever we have
2228 * accumulated so far if it contains any
2231 * (Can't we keep track of that and avoid
2232 * this rescan? --jhi)
2236 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2237 if (!NATIVE_IS_INVARIANT(*c)) {
2242 const STRLEN offset = d - SvPVX_const(sv);
2244 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2248 while (src >= (const U8 *)SvPVX_const(sv)) {
2249 if (!NATIVE_IS_INVARIANT(*src)) {
2250 const U8 ch = NATIVE_TO_ASCII(*src);
2251 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2252 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2262 if (has_utf8 || uv > 255) {
2263 d = (char*)uvchr_to_utf8((U8*)d, uv);
2265 if (PL_lex_inwhat == OP_TRANS &&
2266 PL_sublex_info.sub_op) {
2267 PL_sublex_info.sub_op->op_private |=
2268 (PL_lex_repl ? OPpTRANS_FROM_UTF
2272 if (uv > 255 && !dorange)
2273 native_range = FALSE;
2285 /* \N{LATIN SMALL LETTER A} is a named character */
2289 char* e = strchr(s, '}');
2295 yyerror("Missing right brace on \\N{}");
2299 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2301 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2302 PERL_SCAN_DISALLOW_PREFIX;
2305 uv = grok_hex(s, &len, &flags, NULL);
2306 if ( e > s && len != (STRLEN)(e - s) ) {
2310 goto NUM_ESCAPE_INSERT;
2312 res = newSVpvn(s + 1, e - s - 1);
2313 res = new_constant( NULL, 0, "charnames",
2314 res, NULL, s - 2, e - s + 3 );
2316 sv_utf8_upgrade(res);
2317 str = SvPV_const(res,len);
2318 #ifdef EBCDIC_NEVER_MIND
2319 /* charnames uses pack U and that has been
2320 * recently changed to do the below uni->native
2321 * mapping, so this would be redundant (and wrong,
2322 * the code point would be doubly converted).
2323 * But leave this in just in case the pack U change
2324 * gets revoked, but the semantics is still
2325 * desireable for charnames. --jhi */
2327 UV uv = utf8_to_uvchr((const U8*)str, 0);
2330 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2332 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2333 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2334 str = SvPV_const(res, len);
2338 if (!has_utf8 && SvUTF8(res)) {
2339 const char * const ostart = SvPVX_const(sv);
2340 SvCUR_set(sv, d - ostart);
2343 sv_utf8_upgrade(sv);
2344 /* this just broke our allocation above... */
2345 SvGROW(sv, (STRLEN)(send - start));
2346 d = SvPVX(sv) + SvCUR(sv);
2349 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2350 const char * const odest = SvPVX_const(sv);
2352 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2353 d = SvPVX(sv) + (d - odest);
2357 native_range = FALSE; /* \N{} is guessed to be Unicode */
2359 Copy(str, d, len, char);
2366 yyerror("Missing braces on \\N{}");
2369 /* \c is a control character */
2378 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2381 yyerror("Missing control char name in \\c");
2385 /* printf-style backslashes, formfeeds, newlines, etc */
2387 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2390 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2393 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2396 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2399 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2402 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2405 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2411 } /* end if (backslash) */
2418 /* If we started with encoded form, or already know we want it
2419 and then encode the next character */
2420 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2422 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2423 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2426 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2427 const STRLEN off = d - SvPVX_const(sv);
2428 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2430 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2433 if (uv > 255 && !dorange)
2434 native_range = FALSE;
2438 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2440 } /* while loop to process each character */
2442 /* terminate the string and set up the sv */
2444 SvCUR_set(sv, d - SvPVX_const(sv));
2445 if (SvCUR(sv) >= SvLEN(sv))
2446 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2449 if (PL_encoding && !has_utf8) {
2450 sv_recode_to_utf8(sv, PL_encoding);
2456 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2457 PL_sublex_info.sub_op->op_private |=
2458 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2462 /* shrink the sv if we allocated more than we used */
2463 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2464 SvPV_shrink_to_cur(sv);
2467 /* return the substring (via pl_yylval) only if we parsed anything */
2468 if (s > PL_bufptr) {
2469 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
2470 const char *const key = PL_lex_inpat ? "qr" : "q";
2471 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
2475 if (PL_lex_inwhat == OP_TRANS) {
2478 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
2486 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
2489 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2496 * Returns TRUE if there's more to the expression (e.g., a subscript),
2499 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2501 * ->[ and ->{ return TRUE
2502 * { and [ outside a pattern are always subscripts, so return TRUE
2503 * if we're outside a pattern and it's not { or [, then return FALSE
2504 * if we're in a pattern and the first char is a {
2505 * {4,5} (any digits around the comma) returns FALSE
2506 * if we're in a pattern and the first char is a [
2508 * [SOMETHING] has a funky algorithm to decide whether it's a
2509 * character class or not. It has to deal with things like
2510 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2511 * anything else returns TRUE
2514 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2517 S_intuit_more(pTHX_ register char *s)
2520 if (PL_lex_brackets)
2522 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2524 if (*s != '{' && *s != '[')
2529 /* In a pattern, so maybe we have {n,m}. */
2546 /* On the other hand, maybe we have a character class */
2549 if (*s == ']' || *s == '^')
2552 /* this is terrifying, and it works */
2553 int weight = 2; /* let's weigh the evidence */
2555 unsigned char un_char = 255, last_un_char;
2556 const char * const send = strchr(s,']');
2557 char tmpbuf[sizeof PL_tokenbuf * 4];
2559 if (!send) /* has to be an expression */
2562 Zero(seen,256,char);
2565 else if (isDIGIT(*s)) {
2567 if (isDIGIT(s[1]) && s[2] == ']')
2573 for (; s < send; s++) {
2574 last_un_char = un_char;
2575 un_char = (unsigned char)*s;
2580 weight -= seen[un_char] * 10;
2581 if (isALNUM_lazy_if(s+1,UTF)) {
2583 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2584 len = (int)strlen(tmpbuf);
2585 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2590 else if (*s == '$' && s[1] &&
2591 strchr("[#!%*<>()-=",s[1])) {
2592 if (/*{*/ strchr("])} =",s[2]))
2601 if (strchr("wds]",s[1]))
2603 else if (seen[(U8)'\''] || seen[(U8)'"'])
2605 else if (strchr("rnftbxcav",s[1]))
2607 else if (isDIGIT(s[1])) {
2609 while (s[1] && isDIGIT(s[1]))
2619 if (strchr("aA01! ",last_un_char))
2621 if (strchr("zZ79~",s[1]))
2623 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2624 weight -= 5; /* cope with negative subscript */
2627 if (!isALNUM(last_un_char)
2628 && !(last_un_char == '$' || last_un_char == '@'
2629 || last_un_char == '&')
2630 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2635 if (keyword(tmpbuf, d - tmpbuf, 0))
2638 if (un_char == last_un_char + 1)
2640 weight -= seen[un_char];
2645 if (weight >= 0) /* probably a character class */
2655 * Does all the checking to disambiguate
2657 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2658 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2660 * First argument is the stuff after the first token, e.g. "bar".
2662 * Not a method if bar is a filehandle.
2663 * Not a method if foo is a subroutine prototyped to take a filehandle.
2664 * Not a method if it's really "Foo $bar"
2665 * Method if it's "foo $bar"
2666 * Not a method if it's really "print foo $bar"
2667 * Method if it's really "foo package::" (interpreted as package->foo)
2668 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2669 * Not a method if bar is a filehandle or package, but is quoted with
2674 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2677 char *s = start + (*start == '$');
2678 char tmpbuf[sizeof PL_tokenbuf];
2686 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2690 const char *proto = SvPVX_const(cv);
2701 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2702 /* start is the beginning of the possible filehandle/object,
2703 * and s is the end of it
2704 * tmpbuf is a copy of it
2707 if (*start == '$') {
2708 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2709 isUPPER(*PL_tokenbuf))
2712 len = start - SvPVX(PL_linestr);
2716 start = SvPVX(PL_linestr) + len;
2720 return *s == '(' ? FUNCMETH : METHOD;
2722 if (!keyword(tmpbuf, len, 0)) {
2723 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2727 soff = s - SvPVX(PL_linestr);
2731 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2732 if (indirgv && GvCVu(indirgv))
2734 /* filehandle or package name makes it a method */
2735 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2737 soff = s - SvPVX(PL_linestr);
2740 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2741 return 0; /* no assumptions -- "=>" quotes bearword */
2743 start_force(PL_curforce);
2744 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2745 newSVpvn(tmpbuf,len));
2746 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2748 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2753 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2755 return *s == '(' ? FUNCMETH : METHOD;
2761 /* Encoded script support. filter_add() effectively inserts a
2762 * 'pre-processing' function into the current source input stream.
2763 * Note that the filter function only applies to the current source file
2764 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2766 * The datasv parameter (which may be NULL) can be used to pass
2767 * private data to this instance of the filter. The filter function
2768 * can recover the SV using the FILTER_DATA macro and use it to
2769 * store private buffers and state information.
2771 * The supplied datasv parameter is upgraded to a PVIO type
2772 * and the IoDIRP/IoANY field is used to store the function pointer,
2773 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2774 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2775 * private use must be set using malloc'd pointers.
2779 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2788 if (!PL_rsfp_filters)
2789 PL_rsfp_filters = newAV();
2792 SvUPGRADE(datasv, SVt_PVIO);
2793 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2794 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2795 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2796 FPTR2DPTR(void *, IoANY(datasv)),
2797 SvPV_nolen(datasv)));
2798 av_unshift(PL_rsfp_filters, 1);
2799 av_store(PL_rsfp_filters, 0, datasv) ;
2804 /* Delete most recently added instance of this filter function. */
2806 Perl_filter_del(pTHX_ filter_t funcp)
2812 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2813 FPTR2DPTR(void*, funcp)));
2815 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2817 /* if filter is on top of stack (usual case) just pop it off */
2818 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2819 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2820 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2821 IoANY(datasv) = (void *)NULL;
2822 sv_free(av_pop(PL_rsfp_filters));
2826 /* we need to search for the correct entry and clear it */
2827 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2831 /* Invoke the idxth filter function for the current rsfp. */
2832 /* maxlen 0 = read one text line */
2834 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2839 /* This API is bad. It should have been using unsigned int for maxlen.
2840 Not sure if we want to change the API, but if not we should sanity
2841 check the value here. */
2842 const unsigned int correct_length
2851 if (!PL_parser || !PL_rsfp_filters)
2853 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2854 /* Provide a default input filter to make life easy. */
2855 /* Note that we append to the line. This is handy. */
2856 DEBUG_P(PerlIO_printf(Perl_debug_log,
2857 "filter_read %d: from rsfp\n", idx));
2858 if (correct_length) {
2861 const int old_len = SvCUR(buf_sv);
2863 /* ensure buf_sv is large enough */
2864 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2865 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2866 correct_length)) <= 0) {
2867 if (PerlIO_error(PL_rsfp))
2868 return -1; /* error */
2870 return 0 ; /* end of file */
2872 SvCUR_set(buf_sv, old_len + len) ;
2875 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2876 if (PerlIO_error(PL_rsfp))
2877 return -1; /* error */
2879 return 0 ; /* end of file */
2882 return SvCUR(buf_sv);
2884 /* Skip this filter slot if filter has been deleted */
2885 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2886 DEBUG_P(PerlIO_printf(Perl_debug_log,
2887 "filter_read %d: skipped (filter deleted)\n",
2889 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2891 /* Get function pointer hidden within datasv */
2892 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2893 DEBUG_P(PerlIO_printf(Perl_debug_log,
2894 "filter_read %d: via function %p (%s)\n",
2895 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2896 /* Call function. The function is expected to */
2897 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2898 /* Return: <0:error, =0:eof, >0:not eof */
2899 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2903 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2906 #ifdef PERL_CR_FILTER
2907 if (!PL_rsfp_filters) {
2908 filter_add(S_cr_textfilter,NULL);
2911 if (PL_rsfp_filters) {
2913 SvCUR_set(sv, 0); /* start with empty line */
2914 if (FILTER_READ(0, sv, 0) > 0)
2915 return ( SvPVX(sv) ) ;
2920 return (sv_gets(sv, fp, append));
2924 S_find_in_my_stash(pTHX_ const char *pkgname, STRLEN len)
2929 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2933 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2934 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
2936 return GvHV(gv); /* Foo:: */
2939 /* use constant CLASS => 'MyClass' */
2940 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
2941 if (gv && GvCV(gv)) {
2942 SV * const sv = cv_const_sv(GvCV(gv));
2944 pkgname = SvPV_const(sv, len);
2947 return gv_stashpvn(pkgname, len, 0);
2951 * S_readpipe_override
2952 * Check whether readpipe() is overriden, and generates the appropriate
2953 * optree, provided sublex_start() is called afterwards.
2956 S_readpipe_override(pTHX)
2959 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
2960 pl_yylval.ival = OP_BACKTICK;
2962 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
2964 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
2965 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
2966 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
2968 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
2969 append_elem(OP_LIST,
2970 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
2971 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
2978 * The intent of this yylex wrapper is to minimize the changes to the
2979 * tokener when we aren't interested in collecting madprops. It remains
2980 * to be seen how successful this strategy will be...
2987 char *s = PL_bufptr;
2989 /* make sure PL_thiswhite is initialized */
2993 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
2994 if (PL_pending_ident)
2995 return S_pending_ident(aTHX);
2997 /* previous token ate up our whitespace? */
2998 if (!PL_lasttoke && PL_nextwhite) {
2999 PL_thiswhite = PL_nextwhite;
3003 /* isolate the token, and figure out where it is without whitespace */
3004 PL_realtokenstart = -1;
3008 assert(PL_curforce < 0);
3010 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3011 if (!PL_thistoken) {
3012 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3013 PL_thistoken = newSVpvs("");
3015 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3016 PL_thistoken = newSVpvn(tstart, s - tstart);
3019 if (PL_thismad) /* install head */
3020 CURMAD('X', PL_thistoken);
3023 /* last whitespace of a sublex? */
3024 if (optype == ')' && PL_endwhite) {
3025 CURMAD('X', PL_endwhite);
3030 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3031 if (!PL_thiswhite && !PL_endwhite && !optype) {
3032 sv_free(PL_thistoken);
3037 /* put off final whitespace till peg */
3038 if (optype == ';' && !PL_rsfp) {
3039 PL_nextwhite = PL_thiswhite;
3042 else if (PL_thisopen) {
3043 CURMAD('q', PL_thisopen);
3045 sv_free(PL_thistoken);
3049 /* Store actual token text as madprop X */
3050 CURMAD('X', PL_thistoken);
3054 /* add preceding whitespace as madprop _ */
3055 CURMAD('_', PL_thiswhite);
3059 /* add quoted material as madprop = */
3060 CURMAD('=', PL_thisstuff);
3064 /* add terminating quote as madprop Q */
3065 CURMAD('Q', PL_thisclose);
3069 /* special processing based on optype */
3073 /* opval doesn't need a TOKEN since it can already store mp */
3083 if (pl_yylval.opval)
3084 append_madprops(PL_thismad, pl_yylval.opval, 0);
3092 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3101 /* remember any fake bracket that lexer is about to discard */
3102 if (PL_lex_brackets == 1 &&
3103 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3106 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3109 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3110 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3113 break; /* don't bother looking for trailing comment */
3122 /* attach a trailing comment to its statement instead of next token */
3126 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3128 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3130 if (*s == '\n' || *s == '#') {
3131 while (s < PL_bufend && *s != '\n')
3135 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3136 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3153 /* Create new token struct. Note: opvals return early above. */
3154 pl_yylval.tkval = newTOKEN(optype, pl_yylval, PL_thismad);
3161 S_tokenize_use(pTHX_ int is_use, char *s) {
3163 if (PL_expect != XSTATE)
3164 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3165 is_use ? "use" : "no"));
3167 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3168 s = force_version(s, TRUE);
3169 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3170 start_force(PL_curforce);
3171 NEXTVAL_NEXTTOKE.opval = NULL;
3174 else if (*s == 'v') {
3175 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3176 s = force_version(s, FALSE);
3180 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3181 s = force_version(s, FALSE);
3183 pl_yylval.ival = is_use;
3187 static const char* const exp_name[] =
3188 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3189 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3196 Works out what to call the token just pulled out of the input
3197 stream. The yacc parser takes care of taking the ops we return and
3198 stitching them into a tree.
3204 if read an identifier
3205 if we're in a my declaration
3206 croak if they tried to say my($foo::bar)
3207 build the ops for a my() declaration
3208 if it's an access to a my() variable
3209 are we in a sort block?
3210 croak if my($a); $a <=> $b
3211 build ops for access to a my() variable
3212 if in a dq string, and they've said @foo and we can't find @foo
3214 build ops for a bareword
3215 if we already built the token before, use it.
3220 #pragma segment Perl_yylex
3226 register char *s = PL_bufptr;
3231 /* orig_keyword, gvp, and gv are initialized here because
3232 * jump to the label just_a_word_zero can bypass their
3233 * initialization later. */
3234 I32 orig_keyword = 0;
3239 SV* tmp = newSVpvs("");
3240 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3241 (IV)CopLINE(PL_curcop),
3242 lex_state_names[PL_lex_state],
3243 exp_name[PL_expect],
3244 pv_display(tmp, s, strlen(s), 0, 60));
3247 /* check if there's an identifier for us to look at */
3248 if (PL_pending_ident)
3249 return REPORT(S_pending_ident(aTHX));
3251 /* no identifier pending identification */
3253 switch (PL_lex_state) {
3255 case LEX_NORMAL: /* Some compilers will produce faster */
3256 case LEX_INTERPNORMAL: /* code if we comment these out. */
3260 /* when we've already built the next token, just pull it out of the queue */
3264 pl_yylval = PL_nexttoke[PL_lasttoke].next_val;
3266 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3267 PL_nexttoke[PL_lasttoke].next_mad = 0;
3268 if (PL_thismad && PL_thismad->mad_key == '_') {
3269 PL_thiswhite = (SV*)PL_thismad->mad_val;
3270 PL_thismad->mad_val = 0;
3271 mad_free(PL_thismad);
3276 PL_lex_state = PL_lex_defer;
3277 PL_expect = PL_lex_expect;
3278 PL_lex_defer = LEX_NORMAL;
3279 if (!PL_nexttoke[PL_lasttoke].next_type)
3284 pl_yylval = PL_nextval[PL_nexttoke];
3286 PL_lex_state = PL_lex_defer;
3287 PL_expect = PL_lex_expect;
3288 PL_lex_defer = LEX_NORMAL;
3292 /* FIXME - can these be merged? */
3293 return(PL_nexttoke[PL_lasttoke].next_type);
3295 return REPORT(PL_nexttype[PL_nexttoke]);
3298 /* interpolated case modifiers like \L \U, including \Q and \E.
3299 when we get here, PL_bufptr is at the \
3301 case LEX_INTERPCASEMOD:
3303 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3304 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3306 /* handle \E or end of string */
3307 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3309 if (PL_lex_casemods) {
3310 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3311 PL_lex_casestack[PL_lex_casemods] = '\0';
3313 if (PL_bufptr != PL_bufend
3314 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3316 PL_lex_state = LEX_INTERPCONCAT;
3319 PL_thistoken = newSVpvs("\\E");
3325 while (PL_bufptr != PL_bufend &&
3326 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3328 PL_thiswhite = newSVpvs("");
3329 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3333 if (PL_bufptr != PL_bufend)
3336 PL_lex_state = LEX_INTERPCONCAT;
3340 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3341 "### Saw case modifier\n"); });
3343 if (s[1] == '\\' && s[2] == 'E') {
3346 PL_thiswhite = newSVpvs("");
3347 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3350 PL_lex_state = LEX_INTERPCONCAT;
3355 if (!PL_madskills) /* when just compiling don't need correct */
3356 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3357 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3358 if ((*s == 'L' || *s == 'U') &&
3359 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3360 PL_lex_casestack[--PL_lex_casemods] = '\0';
3363 if (PL_lex_casemods > 10)
3364 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3365 PL_lex_casestack[PL_lex_casemods++] = *s;
3366 PL_lex_casestack[PL_lex_casemods] = '\0';
3367 PL_lex_state = LEX_INTERPCONCAT;
3368 start_force(PL_curforce);
3369 NEXTVAL_NEXTTOKE.ival = 0;
3371 start_force(PL_curforce);
3373 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3375 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3377 NEXTVAL_NEXTTOKE.ival = OP_LC;
3379 NEXTVAL_NEXTTOKE.ival = OP_UC;
3381 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3383 Perl_croak(aTHX_ "panic: yylex");
3385 SV* const tmpsv = newSVpvs("\\ ");
3386 /* replace the space with the character we want to escape
3388 SvPVX(tmpsv)[1] = *s;
3394 if (PL_lex_starts) {
3400 sv_free(PL_thistoken);
3401 PL_thistoken = newSVpvs("");
3404 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3405 if (PL_lex_casemods == 1 && PL_lex_inpat)
3414 case LEX_INTERPPUSH:
3415 return REPORT(sublex_push());
3417 case LEX_INTERPSTART:
3418 if (PL_bufptr == PL_bufend)
3419 return REPORT(sublex_done());
3420 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3421 "### Interpolated variable\n"); });
3423 PL_lex_dojoin = (*PL_bufptr == '@');
3424 PL_lex_state = LEX_INTERPNORMAL;
3425 if (PL_lex_dojoin) {
3426 start_force(PL_curforce);
3427 NEXTVAL_NEXTTOKE.ival = 0;
3429 start_force(PL_curforce);
3430 force_ident("\"", '$');
3431 start_force(PL_curforce);
3432 NEXTVAL_NEXTTOKE.ival = 0;
3434 start_force(PL_curforce);
3435 NEXTVAL_NEXTTOKE.ival = 0;
3437 start_force(PL_curforce);
3438 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3441 if (PL_lex_starts++) {
3446 sv_free(PL_thistoken);
3447 PL_thistoken = newSVpvs("");
3450 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3451 if (!PL_lex_casemods && PL_lex_inpat)
3458 case LEX_INTERPENDMAYBE:
3459 if (intuit_more(PL_bufptr)) {
3460 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3466 if (PL_lex_dojoin) {
3467 PL_lex_dojoin = FALSE;
3468 PL_lex_state = LEX_INTERPCONCAT;
3472 sv_free(PL_thistoken);
3473 PL_thistoken = newSVpvs("");
3478 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3479 && SvEVALED(PL_lex_repl))
3481 if (PL_bufptr != PL_bufend)
3482 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3486 case LEX_INTERPCONCAT:
3488 if (PL_lex_brackets)
3489 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3491 if (PL_bufptr == PL_bufend)
3492 return REPORT(sublex_done());
3494 if (SvIVX(PL_linestr) == '\'') {
3495 SV *sv = newSVsv(PL_linestr);
3498 else if ( PL_hints & HINT_NEW_RE )
3499 sv = new_constant(NULL, 0, "qr", sv, sv, "q", 1);
3500 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3504 s = scan_const(PL_bufptr);
3506 PL_lex_state = LEX_INTERPCASEMOD;
3508 PL_lex_state = LEX_INTERPSTART;
3511 if (s != PL_bufptr) {
3512 start_force(PL_curforce);
3514 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3516 NEXTVAL_NEXTTOKE = pl_yylval;
3519 if (PL_lex_starts++) {
3523 sv_free(PL_thistoken);
3524 PL_thistoken = newSVpvs("");
3527 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3528 if (!PL_lex_casemods && PL_lex_inpat)
3541 PL_lex_state = LEX_NORMAL;
3542 s = scan_formline(PL_bufptr);
3543 if (!PL_lex_formbrack)
3549 PL_oldoldbufptr = PL_oldbufptr;
3555 sv_free(PL_thistoken);
3558 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3562 if (isIDFIRST_lazy_if(s,UTF))
3564 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3565 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3568 goto fake_eof; /* emulate EOF on ^D or ^Z */
3577 if (PL_lex_brackets) {
3578 yyerror((const char *)
3580 ? "Format not terminated"
3581 : "Missing right curly or square bracket"));
3583 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3584 "### Tokener got EOF\n");
3588 if (s++ < PL_bufend)
3589 goto retry; /* ignore stray nulls */
3592 if (!PL_in_eval && !PL_preambled) {
3593 PL_preambled = TRUE;
3599 /* Generate a string of Perl code to load the debugger.
3600 * If PERL5DB is set, it will return the contents of that,
3601 * otherwise a compile-time require of perl5db.pl. */
3603 const char * const pdb = PerlEnv_getenv("PERL5DB");
3606 sv_setpv(PL_linestr, pdb);
3607 sv_catpvs(PL_linestr,";");
3609 SETERRNO(0,SS_NORMAL);
3610 sv_setpvs(PL_linestr, "BEGIN { require 'perl5db.pl' };");
3613 sv_setpvs(PL_linestr,"");
3614 if (PL_preambleav) {
3615 SV **svp = AvARRAY(PL_preambleav);
3616 SV **const end = svp + AvFILLp(PL_preambleav);
3618 sv_catsv(PL_linestr, *svp);
3620 sv_catpvs(PL_linestr, ";");
3622 sv_free((SV*)PL_preambleav);
3623 PL_preambleav = NULL;
3625 if (PL_minus_n || PL_minus_p) {
3626 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3628 sv_catpvs(PL_linestr,"chomp;");
3631 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3632 || *PL_splitstr == '"')
3633 && strchr(PL_splitstr + 1, *PL_splitstr))
3634 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3636 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3637 bytes can be used as quoting characters. :-) */
3638 const char *splits = PL_splitstr;
3639 sv_catpvs(PL_linestr, "our @F=split(q\0");
3642 if (*splits == '\\')
3643 sv_catpvn(PL_linestr, splits, 1);
3644 sv_catpvn(PL_linestr, splits, 1);
3645 } while (*splits++);
3646 /* This loop will embed the trailing NUL of
3647 PL_linestr as the last thing it does before
3649 sv_catpvs(PL_linestr, ");");
3653 sv_catpvs(PL_linestr,"our @F=split(' ');");
3657 sv_catpvs(PL_linestr,
3658 "use feature ':5." STRINGIFY(PERL_VERSION) "';");
3659 sv_catpvs(PL_linestr, "\n");
3660 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3661 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3662 PL_last_lop = PL_last_uni = NULL;
3663 if (PERLDB_LINE && PL_curstash != PL_debstash)
3664 update_debugger_info(PL_linestr, NULL, 0);
3668 bof = PL_rsfp ? TRUE : FALSE;
3669 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3672 PL_realtokenstart = -1;
3675 if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3676 PerlIO_clearerr(PL_rsfp);
3678 (void)PerlIO_close(PL_rsfp);
3680 PL_doextract = FALSE;
3682 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3688 sv_setpvs(PL_linestr, ";}continue{print;}");
3690 sv_setpvs(PL_linestr, ";}");
3691 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3692 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3693 PL_last_lop = PL_last_uni = NULL;
3694 PL_minus_n = PL_minus_p = 0;
3697 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3698 PL_last_lop = PL_last_uni = NULL;
3699 sv_setpvn(PL_linestr,"",0);
3700 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3702 /* If it looks like the start of a BOM or raw UTF-16,
3703 * check if it in fact is. */
3709 #ifdef PERLIO_IS_STDIO
3710 # ifdef __GNU_LIBRARY__
3711 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3712 # define FTELL_FOR_PIPE_IS_BROKEN
3716 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3717 # define FTELL_FOR_PIPE_IS_BROKEN
3722 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3724 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3725 s = swallow_bom((U8*)s);
3729 /* Incest with pod. */
3732 sv_catsv(PL_thiswhite, PL_linestr);
3734 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3735 sv_setpvn(PL_linestr, "", 0);
3736 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3737 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3738 PL_last_lop = PL_last_uni = NULL;
3739 PL_doextract = FALSE;
3743 } while (PL_doextract);
3744 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3745 if (PERLDB_LINE && PL_curstash != PL_debstash)
3746 update_debugger_info(PL_linestr, NULL, 0);
3747 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3748 PL_last_lop = PL_last_uni = NULL;
3749 if (CopLINE(PL_curcop) == 1) {
3750 while (s < PL_bufend && isSPACE(*s))
3752 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3756 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3760 if (*s == '#' && *(s+1) == '!')
3762 #ifdef ALTERNATE_SHEBANG
3764 static char const as[] = ALTERNATE_SHEBANG;
3765 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3766 d = s + (sizeof(as) - 1);
3768 #endif /* ALTERNATE_SHEBANG */
3777 while (*d && !isSPACE(*d))
3781 #ifdef ARG_ZERO_IS_SCRIPT
3782 if (ipathend > ipath) {
3784 * HP-UX (at least) sets argv[0] to the script name,
3785 * which makes $^X incorrect. And Digital UNIX and Linux,
3786 * at least, set argv[0] to the basename of the Perl
3787 * interpreter. So, having found "#!", we'll set it right.
3789 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3791 assert(SvPOK(x) || SvGMAGICAL(x));
3792 if (sv_eq(x, CopFILESV(PL_curcop))) {
3793 sv_setpvn(x, ipath, ipathend - ipath);
3799 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3800 const char * const lstart = SvPV_const(x,llen);
3802 bstart += blen - llen;
3803 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3804 sv_setpvn(x, ipath, ipathend - ipath);
3809 TAINT_NOT; /* $^X is always tainted, but that's OK */
3811 #endif /* ARG_ZERO_IS_SCRIPT */
3816 d = instr(s,"perl -");
3818 d = instr(s,"perl");
3820 /* avoid getting into infinite loops when shebang
3821 * line contains "Perl" rather than "perl" */
3823 for (d = ipathend-4; d >= ipath; --d) {
3824 if ((*d == 'p' || *d == 'P')
3825 && !ibcmp(d, "perl", 4))
3835 #ifdef ALTERNATE_SHEBANG
3837 * If the ALTERNATE_SHEBANG on this system starts with a
3838 * character that can be part of a Perl expression, then if
3839 * we see it but not "perl", we're probably looking at the
3840 * start of Perl code, not a request to hand off to some
3841 * other interpreter. Similarly, if "perl" is there, but
3842 * not in the first 'word' of the line, we assume the line
3843 * contains the start of the Perl program.
3845 if (d && *s != '#') {
3846 const char *c = ipath;
3847 while (*c && !strchr("; \t\r\n\f\v#", *c))
3850 d = NULL; /* "perl" not in first word; ignore */
3852 *s = '#'; /* Don't try to parse shebang line */
3854 #endif /* ALTERNATE_SHEBANG */
3855 #ifndef MACOS_TRADITIONAL
3860 !instr(s,"indir") &&
3861 instr(PL_origargv[0],"perl"))
3868 while (s < PL_bufend && isSPACE(*s))
3870 if (s < PL_bufend) {
3871 Newxz(newargv,PL_origargc+3,char*);
3873 while (s < PL_bufend && !isSPACE(*s))
3876 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3879 newargv = PL_origargv;
3882 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3884 Perl_croak(aTHX_ "Can't exec %s", ipath);
3888 while (*d && !isSPACE(*d))
3890 while (SPACE_OR_TAB(*d))
3894 const bool switches_done = PL_doswitches;
3895 const U32 oldpdb = PL_perldb;
3896 const bool oldn = PL_minus_n;
3897 const bool oldp = PL_minus_p;
3901 if (*d1 == 'M' || *d1 == 'm' || *d1 == 'C') {
3902 const char * const m = d1;
3903 while (*d1 && !isSPACE(*d1))
3905 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3908 d1 = moreswitches(d1);
3910 if (PL_doswitches && !switches_done) {
3911 int argc = PL_origargc;
3912 char **argv = PL_origargv;
3915 } while (argc && argv[0][0] == '-' && argv[0][1]);
3916 init_argv_symbols(argc,argv);
3918 if ((PERLDB_LINE && !oldpdb) ||
3919 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3920 /* if we have already added "LINE: while (<>) {",
3921 we must not do it again */
3923 sv_setpvn(PL_linestr, "", 0);
3924 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3925 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3926 PL_last_lop = PL_last_uni = NULL;
3927 PL_preambled = FALSE;
3929 (void)gv_fetchfile(PL_origfilename);
3936 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3938 PL_lex_state = LEX_FORMLINE;
3943 #ifdef PERL_STRICT_CR
3944 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3946 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3948 case ' ': case '\t': case '\f': case 013:
3949 #ifdef MACOS_TRADITIONAL
3953 PL_realtokenstart = -1;
3955 PL_thiswhite = newSVpvs("");
3956 sv_catpvn(PL_thiswhite, s, 1);
3963 PL_realtokenstart = -1;
3967 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3968 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3969 /* handle eval qq[#line 1 "foo"\n ...] */
3970 CopLINE_dec(PL_curcop);
3973 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
3975 if (!PL_in_eval || PL_rsfp)
3980 while (d < PL_bufend && *d != '\n')
3984 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
3985 Perl_croak(aTHX_ "panic: input overflow");
3988 PL_thiswhite = newSVpvn(s, d - s);
3993 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3995 PL_lex_state = LEX_FORMLINE;
4001 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
4002 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
4005 TOKEN(PEG); /* make sure any #! line is accessible */
4010 /* if (PL_madskills && PL_lex_formbrack) { */
4012 while (d < PL_bufend && *d != '\n')
4016 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4017 Perl_croak(aTHX_ "panic: input overflow");
4018 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4020 PL_thiswhite = newSVpvs("");
4021 if (CopLINE(PL_curcop) == 1) {
4022 sv_setpvn(PL_thiswhite, "", 0);
4025 sv_catpvn(PL_thiswhite, s, d - s);
4039 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4047 while (s < PL_bufend && SPACE_OR_TAB(*s))
4050 if (strnEQ(s,"=>",2)) {
4051 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4052 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4053 OPERATOR('-'); /* unary minus */
4055 PL_last_uni = PL_oldbufptr;
4057 case 'r': ftst = OP_FTEREAD; break;
4058 case 'w': ftst = OP_FTEWRITE; break;
4059 case 'x': ftst = OP_FTEEXEC; break;
4060 case 'o': ftst = OP_FTEOWNED; break;
4061 case 'R': ftst = OP_FTRREAD; break;
4062 case 'W': ftst = OP_FTRWRITE; break;
4063 case 'X': ftst = OP_FTREXEC; break;
4064 case 'O': ftst = OP_FTROWNED; break;
4065 case 'e': ftst = OP_FTIS; break;
4066 case 'z': ftst = OP_FTZERO; break;
4067 case 's': ftst = OP_FTSIZE; break;
4068 case 'f': ftst = OP_FTFILE; break;
4069 case 'd': ftst = OP_FTDIR; break;
4070 case 'l': ftst = OP_FTLINK; break;
4071 case 'p': ftst = OP_FTPIPE; break;
4072 case 'S': ftst = OP_FTSOCK; break;
4073 case 'u': ftst = OP_FTSUID; break;
4074 case 'g': ftst = OP_FTSGID; break;
4075 case 'k': ftst = OP_FTSVTX; break;
4076 case 'b': ftst = OP_FTBLK; break;
4077 case 'c': ftst = OP_FTCHR; break;
4078 case 't': ftst = OP_FTTTY; break;
4079 case 'T': ftst = OP_FTTEXT; break;
4080 case 'B': ftst = OP_FTBINARY; break;
4081 case 'M': case 'A': case 'C':
4082 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4084 case 'M': ftst = OP_FTMTIME; break;
4085 case 'A': ftst = OP_FTATIME; break;
4086 case 'C': ftst = OP_FTCTIME; break;
4094 PL_last_lop_op = (OPCODE)ftst;
4095 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4096 "### Saw file test %c\n", (int)tmp);
4101 /* Assume it was a minus followed by a one-letter named
4102 * subroutine call (or a -bareword), then. */
4103 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4104 "### '-%c' looked like a file test but was not\n",
4111 const char tmp = *s++;
4114 if (PL_expect == XOPERATOR)
4119 else if (*s == '>') {
4122 if (isIDFIRST_lazy_if(s,UTF)) {
4123 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4131 if (PL_expect == XOPERATOR)
4134 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4136 OPERATOR('-'); /* unary minus */
4142 const char tmp = *s++;
4145 if (PL_expect == XOPERATOR)
4150 if (PL_expect == XOPERATOR)
4153 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4160 if (PL_expect != XOPERATOR) {
4161 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4162 PL_expect = XOPERATOR;
4163 force_ident(PL_tokenbuf, '*');
4176 if (PL_expect == XOPERATOR) {
4180 PL_tokenbuf[0] = '%';
4181 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4182 sizeof PL_tokenbuf - 1, FALSE);
4183 if (!PL_tokenbuf[1]) {
4186 PL_pending_ident = '%';
4197 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4204 const char tmp = *s++;
4210 goto just_a_word_zero_gv;
4213 switch (PL_expect) {
4219 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4221 PL_bufptr = s; /* update in case we back off */
4227 PL_expect = XTERMBLOCK;
4230 stuffstart = s - SvPVX(PL_linestr) - 1;
4234 while (isIDFIRST_lazy_if(s,UTF)) {
4237 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4238 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4239 if (tmp < 0) tmp = -tmp;
4253 sv = newSVpvn(s, len);
4255 d = scan_str(d,TRUE,TRUE);
4257 /* MUST advance bufptr here to avoid bogus
4258 "at end of line" context messages from yyerror().
4260 PL_bufptr = s + len;
4261 yyerror("Unterminated attribute parameter in attribute list");
4265 return REPORT(0); /* EOF indicator */
4269 sv_catsv(sv, PL_lex_stuff);
4270 attrs = append_elem(OP_LIST, attrs,
4271 newSVOP(OP_CONST, 0, sv));
4272 SvREFCNT_dec(PL_lex_stuff);
4273 PL_lex_stuff = NULL;
4276 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4278 if (PL_in_my == KEY_our) {
4280 GvUNIQUE_on(cGVOPx_gv(pl_yylval.opval));
4282 /* skip to avoid loading attributes.pm */
4284 deprecate(":unique");
4287 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4290 /* NOTE: any CV attrs applied here need to be part of
4291 the CVf_BUILTIN_ATTRS define in cv.h! */
4292 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4294 CvLVALUE_on(PL_compcv);
4296 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4298 CvLOCKED_on(PL_compcv);
4300 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4302 CvMETHOD_on(PL_compcv);
4304 /* After we've set the flags, it could be argued that
4305 we don't need to do the attributes.pm-based setting
4306 process, and shouldn't bother appending recognized
4307 flags. To experiment with that, uncomment the
4308 following "else". (Note that's already been
4309 uncommented. That keeps the above-applied built-in
4310 attributes from being intercepted (and possibly
4311 rejected) by a package's attribute routines, but is
4312 justified by the performance win for the common case
4313 of applying only built-in attributes.) */
4315 attrs = append_elem(OP_LIST, attrs,
4316 newSVOP(OP_CONST, 0,
4320 if (*s == ':' && s[1] != ':')
4323 break; /* require real whitespace or :'s */
4324 /* XXX losing whitespace on sequential attributes here */
4328 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4329 if (*s != ';' && *s != '}' && *s != tmp
4330 && (tmp != '=' || *s != ')')) {
4331 const char q = ((*s == '\'') ? '"' : '\'');
4332 /* If here for an expression, and parsed no attrs, back
4334 if (tmp == '=' && !attrs) {
4338 /* MUST advance bufptr here to avoid bogus "at end of line"
4339 context messages from yyerror().
4342 yyerror( (const char *)
4344 ? Perl_form(aTHX_ "Invalid separator character "
4345 "%c%c%c in attribute list", q, *s, q)
4346 : "Unterminated attribute list" ) );
4354 start_force(PL_curforce);
4355 NEXTVAL_NEXTTOKE.opval = attrs;
4356 CURMAD('_', PL_nextwhite);
4361 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4362 (s - SvPVX(PL_linestr)) - stuffstart);
4370 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4371 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4379 const char tmp = *s++;
4384 const char tmp = *s++;
4392 if (PL_lex_brackets <= 0)
4393 yyerror("Unmatched right square bracket");
4396 if (PL_lex_state == LEX_INTERPNORMAL) {
4397 if (PL_lex_brackets == 0) {
4398 if (*s == '-' && s[1] == '>')
4399 PL_lex_state = LEX_INTERPENDMAYBE;
4400 else if (*s != '[' && *s != '{')
4401 PL_lex_state = LEX_INTERPEND;
4408 if (PL_lex_brackets > 100) {
4409 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4411 switch (PL_expect) {
4413 if (PL_lex_formbrack) {
4417 if (PL_oldoldbufptr == PL_last_lop)
4418 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4420 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4421 OPERATOR(HASHBRACK);
4423 while (s < PL_bufend && SPACE_OR_TAB(*s))
4426 PL_tokenbuf[0] = '\0';
4427 if (d < PL_bufend && *d == '-') {
4428 PL_tokenbuf[0] = '-';
4430 while (d < PL_bufend && SPACE_OR_TAB(*d))
4433 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4434 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4436 while (d < PL_bufend && SPACE_OR_TAB(*d))
4439 const char minus = (PL_tokenbuf[0] == '-');
4440 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4448 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4453 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4458 if (PL_oldoldbufptr == PL_last_lop)
4459 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4461 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4464 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4466 /* This hack is to get the ${} in the message. */
4468 yyerror("syntax error");
4471 OPERATOR(HASHBRACK);
4473 /* This hack serves to disambiguate a pair of curlies
4474 * as being a block or an anon hash. Normally, expectation
4475 * determines that, but in cases where we're not in a
4476 * position to expect anything in particular (like inside
4477 * eval"") we have to resolve the ambiguity. This code
4478 * covers the case where the first term in the curlies is a
4479 * quoted string. Most other cases need to be explicitly
4480 * disambiguated by prepending a "+" before the opening
4481 * curly in order to force resolution as an anon hash.
4483 * XXX should probably propagate the outer expectation
4484 * into eval"" to rely less on this hack, but that could
4485 * potentially break current behavior of eval"".
4489 if (*s == '\'' || *s == '"' || *s == '`') {
4490 /* common case: get past first string, handling escapes */
4491 for (t++; t < PL_bufend && *t != *s;)
4492 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4496 else if (*s == 'q') {
4499 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4502 /* skip q//-like construct */
4504 char open, close, term;
4507 while (t < PL_bufend && isSPACE(*t))
4509 /* check for q => */
4510 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4511 OPERATOR(HASHBRACK);
4515 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4519 for (t++; t < PL_bufend; t++) {
4520 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4522 else if (*t == open)
4526 for (t++; t < PL_bufend; t++) {
4527 if (*t == '\\' && t+1 < PL_bufend)
4529 else if (*t == close && --brackets <= 0)
4531 else if (*t == open)
4538 /* skip plain q word */
4539 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4542 else if (isALNUM_lazy_if(t,UTF)) {
4544 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4547 while (t < PL_bufend && isSPACE(*t))
4549 /* if comma follows first term, call it an anon hash */
4550 /* XXX it could be a comma expression with loop modifiers */
4551 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4552 || (*t == '=' && t[1] == '>')))
4553 OPERATOR(HASHBRACK);
4554 if (PL_expect == XREF)
4557 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4563 pl_yylval.ival = CopLINE(PL_curcop);
4564 if (isSPACE(*s) || *s == '#')
4565 PL_copline = NOLINE; /* invalidate current command line number */
4570 if (PL_lex_brackets <= 0)
4571 yyerror("Unmatched right curly bracket");
4573 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4574 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4575 PL_lex_formbrack = 0;
4576 if (PL_lex_state == LEX_INTERPNORMAL) {
4577 if (PL_lex_brackets == 0) {
4578 if (PL_expect & XFAKEBRACK) {
4579 PL_expect &= XENUMMASK;
4580 PL_lex_state = LEX_INTERPEND;
4585 PL_thiswhite = newSVpvs("");
4586 sv_catpvn(PL_thiswhite,"}",1);
4589 return yylex(); /* ignore fake brackets */
4591 if (*s == '-' && s[1] == '>')
4592 PL_lex_state = LEX_INTERPENDMAYBE;
4593 else if (*s != '[' && *s != '{')
4594 PL_lex_state = LEX_INTERPEND;
4597 if (PL_expect & XFAKEBRACK) {
4598 PL_expect &= XENUMMASK;
4600 return yylex(); /* ignore fake brackets */
4602 start_force(PL_curforce);
4604 curmad('X', newSVpvn(s-1,1));
4605 CURMAD('_', PL_thiswhite);
4610 PL_thistoken = newSVpvs("");
4618 if (PL_expect == XOPERATOR) {
4619 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4620 && isIDFIRST_lazy_if(s,UTF))
4622 CopLINE_dec(PL_curcop);
4623 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4624 CopLINE_inc(PL_curcop);
4629 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4631 PL_expect = XOPERATOR;
4632 force_ident(PL_tokenbuf, '&');
4636 pl_yylval.ival = (OPpENTERSUB_AMPER<<8);
4648 const char tmp = *s++;
4655 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4656 && strchr("+-*/%.^&|<",tmp))
4657 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4658 "Reversed %c= operator",(int)tmp);
4660 if (PL_expect == XSTATE && isALPHA(tmp) &&
4661 (s == PL_linestart+1 || s[-2] == '\n') )
4663 if (PL_in_eval && !PL_rsfp) {
4668 if (strnEQ(s,"=cut",4)) {
4684 PL_thiswhite = newSVpvs("");
4685 sv_catpvn(PL_thiswhite, PL_linestart,
4686 PL_bufend - PL_linestart);
4690 PL_doextract = TRUE;
4694 if (PL_lex_brackets < PL_lex_formbrack) {
4696 #ifdef PERL_STRICT_CR
4697 while (SPACE_OR_TAB(*t))
4699 while (SPACE_OR_TAB(*t) || *t == '\r')
4702 if (*t == '\n' || *t == '#') {
4713 const char tmp = *s++;
4715 /* was this !=~ where !~ was meant?
4716 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4718 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4719 const char *t = s+1;
4721 while (t < PL_bufend && isSPACE(*t))
4724 if (*t == '/' || *t == '?' ||
4725 ((*t == 'm' || *t == 's' || *t == 'y')
4726 && !isALNUM(t[1])) ||
4727 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4728 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4729 "!=~ should be !~");
4739 if (PL_expect != XOPERATOR) {
4740 if (s[1] != '<' && !strchr(s,'>'))
4743 s = scan_heredoc(s);
4745 s = scan_inputsymbol(s);
4746 TERM(sublex_start());
4752 SHop(OP_LEFT_SHIFT);
4766 const char tmp = *s++;
4768 SHop(OP_RIGHT_SHIFT);
4769 else if (tmp == '=')
4778 if (PL_expect == XOPERATOR) {
4779 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4781 deprecate_old(commaless_variable_list);
4782 return REPORT(','); /* grandfather non-comma-format format */
4786 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4787 PL_tokenbuf[0] = '@';
4788 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4789 sizeof PL_tokenbuf - 1, FALSE);
4790 if (PL_expect == XOPERATOR)
4791 no_op("Array length", s);
4792 if (!PL_tokenbuf[1])
4794 PL_expect = XOPERATOR;
4795 PL_pending_ident = '#';
4799 PL_tokenbuf[0] = '$';
4800 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4801 sizeof PL_tokenbuf - 1, FALSE);
4802 if (PL_expect == XOPERATOR)
4804 if (!PL_tokenbuf[1]) {
4806 yyerror("Final $ should be \\$ or $name");
4810 /* This kludge not intended to be bulletproof. */
4811 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4812 pl_yylval.opval = newSVOP(OP_CONST, 0,
4813 newSViv(CopARYBASE_get(&PL_compiling)));
4814 pl_yylval.opval->op_private = OPpCONST_ARYBASE;
4820 const char tmp = *s;
4821 if (PL_lex_state == LEX_NORMAL)
4824 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4825 && intuit_more(s)) {
4827 PL_tokenbuf[0] = '@';
4828 if (ckWARN(WARN_SYNTAX)) {
4831 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4834 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4835 while (t < PL_bufend && *t != ']')
4837 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4838 "Multidimensional syntax %.*s not supported",
4839 (int)((t - PL_bufptr) + 1), PL_bufptr);
4843 else if (*s == '{') {
4845 PL_tokenbuf[0] = '%';
4846 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4847 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4849 char tmpbuf[sizeof PL_tokenbuf];
4852 } while (isSPACE(*t));
4853 if (isIDFIRST_lazy_if(t,UTF)) {
4855 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4859 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4860 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4861 "You need to quote \"%s\"",
4868 PL_expect = XOPERATOR;
4869 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4870 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4871 if (!islop || PL_last_lop_op == OP_GREPSTART)
4872 PL_expect = XOPERATOR;
4873 else if (strchr("$@\"'`q", *s))
4874 PL_expect = XTERM; /* e.g. print $fh "foo" */
4875 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4876 PL_expect = XTERM; /* e.g. print $fh &sub */
4877 else if (isIDFIRST_lazy_if(s,UTF)) {
4878 char tmpbuf[sizeof PL_tokenbuf];
4880 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4881 if ((t2 = keyword(tmpbuf, len, 0))) {
4882 /* binary operators exclude handle interpretations */
4894 PL_expect = XTERM; /* e.g. print $fh length() */
4899 PL_expect = XTERM; /* e.g. print $fh subr() */
4902 else if (isDIGIT(*s))
4903 PL_expect = XTERM; /* e.g. print $fh 3 */
4904 else if (*s == '.' && isDIGIT(s[1]))
4905 PL_expect = XTERM; /* e.g. print $fh .3 */
4906 else if ((*s == '?' || *s == '-' || *s == '+')
4907 && !isSPACE(s[1]) && s[1] != '=')
4908 PL_expect = XTERM; /* e.g. print $fh -1 */
4909 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4911 PL_expect = XTERM; /* e.g. print $fh /.../
4912 XXX except DORDOR operator
4914 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4916 PL_expect = XTERM; /* print $fh <<"EOF" */
4919 PL_pending_ident = '$';
4923 if (PL_expect == XOPERATOR)
4925 PL_tokenbuf[0] = '@';
4926 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
4927 if (!PL_tokenbuf[1]) {
4930 if (PL_lex_state == LEX_NORMAL)
4932 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
4934 PL_tokenbuf[0] = '%';
4936 /* Warn about @ where they meant $. */
4937 if (*s == '[' || *s == '{') {
4938 if (ckWARN(WARN_SYNTAX)) {
4939 const char *t = s + 1;
4940 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
4942 if (*t == '}' || *t == ']') {
4944 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4945 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4946 "Scalar value %.*s better written as $%.*s",
4947 (int)(t-PL_bufptr), PL_bufptr,
4948 (int)(t-PL_bufptr-1), PL_bufptr+1);
4953 PL_pending_ident = '@';
4956 case '/': /* may be division, defined-or, or pattern */
4957 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
4961 case '?': /* may either be conditional or pattern */
4962 if(PL_expect == XOPERATOR) {
4970 /* A // operator. */
4980 /* Disable warning on "study /blah/" */
4981 if (PL_oldoldbufptr == PL_last_uni
4982 && (*PL_last_uni != 's' || s - PL_last_uni < 5
4983 || memNE(PL_last_uni, "study", 5)
4984 || isALNUM_lazy_if(PL_last_uni+5,UTF)
4987 s = scan_pat(s,OP_MATCH);
4988 TERM(sublex_start());
4992 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
4993 #ifdef PERL_STRICT_CR
4996 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
4998 && (s == PL_linestart || s[-1] == '\n') )
5000 PL_lex_formbrack = 0;
5004 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5010 pl_yylval.ival = OPf_SPECIAL;
5016 if (PL_expect != XOPERATOR)
5021 case '0': case '1': case '2': case '3': case '4':
5022 case '5': case '6': case '7': case '8': case '9':
5023 s = scan_num(s, &pl_yylval);
5024 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5025 if (PL_expect == XOPERATOR)
5030 s = scan_str(s,!!PL_madskills,FALSE);
5031 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5032 if (PL_expect == XOPERATOR) {
5033 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5035 deprecate_old(commaless_variable_list);
5036 return REPORT(','); /* grandfather non-comma-format format */
5043 pl_yylval.ival = OP_CONST;
5044 TERM(sublex_start());
5047 s = scan_str(s,!!PL_madskills,FALSE);
5048 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5049 if (PL_expect == XOPERATOR) {
5050 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5052 deprecate_old(commaless_variable_list);
5053 return REPORT(','); /* grandfather non-comma-format format */
5060 pl_yylval.ival = OP_CONST;
5061 /* FIXME. I think that this can be const if char *d is replaced by
5062 more localised variables. */
5063 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5064 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5065 pl_yylval.ival = OP_STRINGIFY;
5069 TERM(sublex_start());
5072 s = scan_str(s,!!PL_madskills,FALSE);
5073 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5074 if (PL_expect == XOPERATOR)
5075 no_op("Backticks",s);
5078 readpipe_override();
5079 TERM(sublex_start());
5083 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5084 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5086 if (PL_expect == XOPERATOR)
5087 no_op("Backslash",s);
5091 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5092 char *start = s + 2;
5093 while (isDIGIT(*start) || *start == '_')
5095 if (*start == '.' && isDIGIT(start[1])) {
5096 s = scan_num(s, &pl_yylval);
5099 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5100 else if (!isALPHA(*start) && (PL_expect == XTERM
5101 || PL_expect == XREF || PL_expect == XSTATE
5102 || PL_expect == XTERMORDORDOR)) {
5103 GV *const gv = gv_fetchpvn_flags(s, start - s, 0, SVt_PVCV);
5105 s = scan_num(s, &pl_yylval);
5112 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5154 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5156 /* Some keywords can be followed by any delimiter, including ':' */
5157 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5158 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5159 (PL_tokenbuf[0] == 'q' &&
5160 strchr("qwxr", PL_tokenbuf[1])))));
5162 /* x::* is just a word, unless x is "CORE" */
5163 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5167 while (d < PL_bufend && isSPACE(*d))
5168 d++; /* no comments skipped here, or s### is misparsed */
5170 /* Is this a label? */
5171 if (!tmp && PL_expect == XSTATE
5172 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5174 pl_yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5179 /* Check for keywords */
5180 tmp = keyword(PL_tokenbuf, len, 0);
5182 /* Is this a word before a => operator? */
5183 if (*d == '=' && d[1] == '>') {
5186 = (OP*)newSVOP(OP_CONST, 0,
5187 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5188 pl_yylval.opval->op_private = OPpCONST_BARE;
5192 if (tmp < 0) { /* second-class keyword? */
5193 GV *ogv = NULL; /* override (winner) */
5194 GV *hgv = NULL; /* hidden (loser) */
5195 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5197 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5200 if (GvIMPORTED_CV(gv))
5202 else if (! CvMETHOD(cv))
5206 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5207 (gv = *gvp) && isGV_with_GP(gv) &&
5208 GvCVu(gv) && GvIMPORTED_CV(gv))
5215 tmp = 0; /* overridden by import or by GLOBAL */
5218 && -tmp==KEY_lock /* XXX generalizable kludge */
5221 tmp = 0; /* any sub overrides "weak" keyword */
5223 else { /* no override */
5225 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5226 Perl_warner(aTHX_ packWARN(WARN_MISC),
5227 "dump() better written as CORE::dump()");
5231 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5232 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5233 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5234 "Ambiguous call resolved as CORE::%s(), %s",
5235 GvENAME(hgv), "qualify as such or use &");
5242 default: /* not a keyword */
5243 /* Trade off - by using this evil construction we can pull the
5244 variable gv into the block labelled keylookup. If not, then
5245 we have to give it function scope so that the goto from the
5246 earlier ':' case doesn't bypass the initialisation. */
5248 just_a_word_zero_gv:
5256 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5259 SV *nextPL_nextwhite = 0;
5263 /* Get the rest if it looks like a package qualifier */
5265 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5267 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5270 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5271 *s == '\'' ? "'" : "::");
5276 if (PL_expect == XOPERATOR) {
5277 if (PL_bufptr == PL_linestart) {
5278 CopLINE_dec(PL_curcop);
5279 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
5280 CopLINE_inc(PL_curcop);
5283 no_op("Bareword",s);
5286 /* Look for a subroutine with this name in current package,
5287 unless name is "Foo::", in which case Foo is a bearword
5288 (and a package name). */
5290 if (len > 2 && !PL_madskills &&
5291 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5293 if (ckWARN(WARN_BAREWORD)
5294 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5295 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5296 "Bareword \"%s\" refers to nonexistent package",
5299 PL_tokenbuf[len] = '\0';
5305 /* Mustn't actually add anything to a symbol table.
5306 But also don't want to "initialise" any placeholder
5307 constants that might already be there into full
5308 blown PVGVs with attached PVCV. */
5309 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5310 GV_NOADD_NOINIT, SVt_PVCV);
5315 /* if we saw a global override before, get the right name */
5318 sv = newSVpvs("CORE::GLOBAL::");
5319 sv_catpv(sv,PL_tokenbuf);
5322 /* If len is 0, newSVpv does strlen(), which is correct.
5323 If len is non-zero, then it will be the true length,
5324 and so the scalar will be created correctly. */
5325 sv = newSVpv(PL_tokenbuf,len);
5328 if (PL_madskills && !PL_thistoken) {
5329 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5330 PL_thistoken = newSVpvn(start,s - start);
5331 PL_realtokenstart = s - SvPVX(PL_linestr);
5335 /* Presume this is going to be a bareword of some sort. */
5338 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5339 pl_yylval.opval->op_private = OPpCONST_BARE;
5340 /* UTF-8 package name? */
5341 if (UTF && !IN_BYTES &&
5342 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5345 /* And if "Foo::", then that's what it certainly is. */
5350 /* Do the explicit type check so that we don't need to force
5351 the initialisation of the symbol table to have a real GV.
5352 Beware - gv may not really be a PVGV, cv may not really be
5353 a PVCV, (because of the space optimisations that gv_init
5354 understands) But they're true if for this symbol there is
5355 respectively a typeglob and a subroutine.
5357 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5358 /* Real typeglob, so get the real subroutine: */
5360 /* A proxy for a subroutine in this package? */
5361 : SvOK(gv) ? (CV *) gv : NULL)
5364 /* See if it's the indirect object for a list operator. */
5366 if (PL_oldoldbufptr &&
5367 PL_oldoldbufptr < PL_bufptr &&
5368 (PL_oldoldbufptr == PL_last_lop
5369 || PL_oldoldbufptr == PL_last_uni) &&
5370 /* NO SKIPSPACE BEFORE HERE! */
5371 (PL_expect == XREF ||
5372 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5374 bool immediate_paren = *s == '(';
5376 /* (Now we can afford to cross potential line boundary.) */
5377 s = SKIPSPACE2(s,nextPL_nextwhite);
5379 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5382 /* Two barewords in a row may indicate method call. */
5384 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5385 (tmp = intuit_method(s, gv, cv)))
5388 /* If not a declared subroutine, it's an indirect object. */
5389 /* (But it's an indir obj regardless for sort.) */
5390 /* Also, if "_" follows a filetest operator, it's a bareword */
5393 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5395 (PL_last_lop_op != OP_MAPSTART &&
5396 PL_last_lop_op != OP_GREPSTART))))
5397 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5398 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5401 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5406 PL_expect = XOPERATOR;
5409 s = SKIPSPACE2(s,nextPL_nextwhite);
5410 PL_nextwhite = nextPL_nextwhite;
5415 /* Is this a word before a => operator? */
5416 if (*s == '=' && s[1] == '>' && !pkgname) {
5418 sv_setpv(((SVOP*)pl_yylval.opval)->op_sv, PL_tokenbuf);
5419 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5420 SvUTF8_on(((SVOP*)pl_yylval.opval)->op_sv);
5424 /* If followed by a paren, it's certainly a subroutine. */
5429 while (SPACE_OR_TAB(*d))
5431 if (*d == ')' && (sv = gv_const_sv(gv))) {
5438 PL_nextwhite = PL_thiswhite;
5441 start_force(PL_curforce);
5443 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5444 PL_expect = XOPERATOR;
5447 PL_nextwhite = nextPL_nextwhite;
5448 curmad('X', PL_thistoken);
5449 PL_thistoken = newSVpvs("");
5457 /* If followed by var or block, call it a method (unless sub) */
5459 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5460 PL_last_lop = PL_oldbufptr;
5461 PL_last_lop_op = OP_METHOD;
5465 /* If followed by a bareword, see if it looks like indir obj. */
5468 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5469 && (tmp = intuit_method(s, gv, cv)))
5472 /* Not a method, so call it a subroutine (if defined) */
5475 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5476 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5477 "Ambiguous use of -%s resolved as -&%s()",
5478 PL_tokenbuf, PL_tokenbuf);
5479 /* Check for a constant sub */
5480 if ((sv = gv_const_sv(gv))) {
5482 SvREFCNT_dec(((SVOP*)pl_yylval.opval)->op_sv);
5483 ((SVOP*)pl_yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5484 pl_yylval.opval->op_private = 0;
5488 /* Resolve to GV now. */
5489 if (SvTYPE(gv) != SVt_PVGV) {
5490 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5491 assert (SvTYPE(gv) == SVt_PVGV);
5492 /* cv must have been some sort of placeholder, so
5493 now needs replacing with a real code reference. */
5497 op_free(pl_yylval.opval);
5498 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5499 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5500 PL_last_lop = PL_oldbufptr;
5501 PL_last_lop_op = OP_ENTERSUB;
5502 /* Is there a prototype? */
5510 const char *proto = SvPV_const((SV*)cv, protolen);
5513 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5515 while (*proto == ';')
5517 if (*proto == '&' && *s == '{') {
5519 sv_setpvs(PL_subname, "__ANON__");
5521 sv_setpvs(PL_subname, "__ANON__::__ANON__");
5528 PL_nextwhite = PL_thiswhite;
5531 start_force(PL_curforce);
5532 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5535 PL_nextwhite = nextPL_nextwhite;
5536 curmad('X', PL_thistoken);
5537 PL_thistoken = newSVpvs("");
5544 /* Guess harder when madskills require "best effort". */
5545 if (PL_madskills && (!gv || !GvCVu(gv))) {
5546 int probable_sub = 0;
5547 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5549 else if (isALPHA(*s)) {
5553 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5554 if (!keyword(tmpbuf, tmplen, 0))
5557 while (d < PL_bufend && isSPACE(*d))
5559 if (*d == '=' && d[1] == '>')
5564 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5565 op_free(pl_yylval.opval);
5566 pl_yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5567 pl_yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5568 PL_last_lop = PL_oldbufptr;
5569 PL_last_lop_op = OP_ENTERSUB;
5570 PL_nextwhite = PL_thiswhite;
5572 start_force(PL_curforce);
5573 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5575 PL_nextwhite = nextPL_nextwhite;
5576 curmad('X', PL_thistoken);
5577 PL_thistoken = newSVpvs("");
5582 NEXTVAL_NEXTTOKE.opval = pl_yylval.opval;
5589 /* Call it a bare word */
5591 if (PL_hints & HINT_STRICT_SUBS)
5592 pl_yylval.opval->op_private |= OPpCONST_STRICT;
5595 if (lastchar != '-') {
5596 if (ckWARN(WARN_RESERVED)) {
5600 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5601 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5608 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5609 && ckWARN_d(WARN_AMBIGUOUS)) {
5610 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5611 "Operator or semicolon missing before %c%s",
5612 lastchar, PL_tokenbuf);
5613 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5614 "Ambiguous use of %c resolved as operator %c",
5615 lastchar, lastchar);
5621 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5622 newSVpv(CopFILE(PL_curcop),0));
5626 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5627 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5630 case KEY___PACKAGE__:
5631 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5633 ? newSVhek(HvNAME_HEK(PL_curstash))
5640 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5641 const char *pname = "main";
5642 if (PL_tokenbuf[2] == 'D')
5643 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5644 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5648 GvIOp(gv) = newIO();
5649 IoIFP(GvIOp(gv)) = PL_rsfp;
5650 #if defined(HAS_FCNTL) && defined(F_SETFD)
5652 const int fd = PerlIO_fileno(PL_rsfp);
5653 fcntl(fd,F_SETFD,fd >= 3);
5656 /* Mark this internal pseudo-handle as clean */
5657 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5658 if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5659 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5661 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5662 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5663 /* if the script was opened in binmode, we need to revert
5664 * it to text mode for compatibility; but only iff it has CRs
5665 * XXX this is a questionable hack at best. */
5666 if (PL_bufend-PL_bufptr > 2
5667 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5670 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5671 loc = PerlIO_tell(PL_rsfp);
5672 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5675 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5677 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5678 #endif /* NETWARE */
5679 #ifdef PERLIO_IS_STDIO /* really? */
5680 # if defined(__BORLANDC__)
5681 /* XXX see note in do_binmode() */
5682 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5686 PerlIO_seek(PL_rsfp, loc, 0);
5690 #ifdef PERLIO_LAYERS
5693 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5694 else if (PL_encoding) {
5701 XPUSHs(PL_encoding);
5703 call_method("name", G_SCALAR);
5707 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5708 Perl_form(aTHX_ ":encoding(%"SVf")",
5717 if (PL_realtokenstart >= 0) {
5718 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5720 PL_endwhite = newSVpvs("");
5721 sv_catsv(PL_endwhite, PL_thiswhite);
5723 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5724 PL_realtokenstart = -1;
5726 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5727 SvCUR(PL_endwhite))) != NULL) ;
5742 if (PL_expect == XSTATE) {
5749 if (*s == ':' && s[1] == ':') {
5752 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5753 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5754 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5757 else if (tmp == KEY_require || tmp == KEY_do)
5758 /* that's a way to remember we saw "CORE::" */
5771 LOP(OP_ACCEPT,XTERM);
5777 LOP(OP_ATAN2,XTERM);
5783 LOP(OP_BINMODE,XTERM);
5786 LOP(OP_BLESS,XTERM);
5795 /* When 'use switch' is in effect, continue has a dual
5796 life as a control operator. */
5798 if (!FEATURE_IS_ENABLED("switch"))
5801 /* We have to disambiguate the two senses of
5802 "continue". If the next token is a '{' then
5803 treat it as the start of a continue block;
5804 otherwise treat it as a control operator.
5816 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5833 if (!PL_cryptseen) {
5834 PL_cryptseen = TRUE;
5838 LOP(OP_CRYPT,XTERM);
5841 LOP(OP_CHMOD,XTERM);
5844 LOP(OP_CHOWN,XTERM);
5847 LOP(OP_CONNECT,XTERM);
5866 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5867 if (orig_keyword == KEY_do) {
5876 PL_hints |= HINT_BLOCK_SCOPE;
5886 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5887 LOP(OP_DBMOPEN,XTERM);
5893 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5900 pl_yylval.ival = CopLINE(PL_curcop);
5916 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
5917 UNIBRACK(OP_ENTEREVAL);
5931 case KEY_endhostent:
5937 case KEY_endservent:
5940 case KEY_endprotoent:
5951 pl_yylval.ival = CopLINE(PL_curcop);
5953 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
5956 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
5959 if ((PL_bufend - p) >= 3 &&
5960 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
5962 else if ((PL_bufend - p) >= 4 &&
5963 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
5966 if (isIDFIRST_lazy_if(p,UTF)) {
5967 p = scan_ident(p, PL_bufend,
5968 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
5972 Perl_croak(aTHX_ "Missing $ on loop variable");
5974 s = SvPVX(PL_linestr) + soff;
5980 LOP(OP_FORMLINE,XTERM);
5986 LOP(OP_FCNTL,XTERM);
5992 LOP(OP_FLOCK,XTERM);
6001 LOP(OP_GREPSTART, XREF);
6004 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6019 case KEY_getpriority:
6020 LOP(OP_GETPRIORITY,XTERM);
6022 case KEY_getprotobyname:
6025 case KEY_getprotobynumber:
6026 LOP(OP_GPBYNUMBER,XTERM);
6028 case KEY_getprotoent:
6040 case KEY_getpeername:
6041 UNI(OP_GETPEERNAME);
6043 case KEY_gethostbyname:
6046 case KEY_gethostbyaddr:
6047 LOP(OP_GHBYADDR,XTERM);
6049 case KEY_gethostent:
6052 case KEY_getnetbyname:
6055 case KEY_getnetbyaddr:
6056 LOP(OP_GNBYADDR,XTERM);
6061 case KEY_getservbyname:
6062 LOP(OP_GSBYNAME,XTERM);
6064 case KEY_getservbyport:
6065 LOP(OP_GSBYPORT,XTERM);
6067 case KEY_getservent:
6070 case KEY_getsockname:
6071 UNI(OP_GETSOCKNAME);
6073 case KEY_getsockopt:
6074 LOP(OP_GSOCKOPT,XTERM);
6089 pl_yylval.ival = CopLINE(PL_curcop);
6099 pl_yylval.ival = CopLINE(PL_curcop);
6103 LOP(OP_INDEX,XTERM);
6109 LOP(OP_IOCTL,XTERM);
6121 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6153 LOP(OP_LISTEN,XTERM);
6162 s = scan_pat(s,OP_MATCH);
6163 TERM(sublex_start());
6166 LOP(OP_MAPSTART, XREF);
6169 LOP(OP_MKDIR,XTERM);
6172 LOP(OP_MSGCTL,XTERM);
6175 LOP(OP_MSGGET,XTERM);
6178 LOP(OP_MSGRCV,XTERM);
6181 LOP(OP_MSGSND,XTERM);
6186 PL_in_my = (U16)tmp;
6188 if (isIDFIRST_lazy_if(s,UTF)) {
6192 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6193 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6195 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6196 if (!PL_in_my_stash) {
6199 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6203 if (PL_madskills) { /* just add type to declarator token */
6204 sv_catsv(PL_thistoken, PL_nextwhite);
6206 sv_catpvn(PL_thistoken, start, s - start);
6214 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6221 s = tokenize_use(0, s);
6225 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6232 if (isIDFIRST_lazy_if(s,UTF)) {
6234 for (d = s; isALNUM_lazy_if(d,UTF);)
6236 for (t=d; isSPACE(*t);)
6238 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6240 && !(t[0] == '=' && t[1] == '>')
6242 int parms_len = (int)(d-s);
6243 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6244 "Precedence problem: open %.*s should be open(%.*s)",
6245 parms_len, s, parms_len, s);
6251 pl_yylval.ival = OP_OR;
6261 LOP(OP_OPEN_DIR,XTERM);
6264 checkcomma(s,PL_tokenbuf,"filehandle");
6268 checkcomma(s,PL_tokenbuf,"filehandle");
6287 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6291 LOP(OP_PIPE_OP,XTERM);
6294 s = scan_str(s,!!PL_madskills,FALSE);
6297 pl_yylval.ival = OP_CONST;
6298 TERM(sublex_start());
6304 s = scan_str(s,!!PL_madskills,FALSE);
6307 PL_expect = XOPERATOR;
6309 if (SvCUR(PL_lex_stuff)) {
6312 d = SvPV_force(PL_lex_stuff, len);
6314 for (; isSPACE(*d) && len; --len, ++d)
6319 if (!warned && ckWARN(WARN_QW)) {
6320 for (; !isSPACE(*d) && len; --len, ++d) {
6322 Perl_warner(aTHX_ packWARN(WARN_QW),
6323 "Possible attempt to separate words with commas");
6326 else if (*d == '#') {
6327 Perl_warner(aTHX_ packWARN(WARN_QW),
6328 "Possible attempt to put comments in qw() list");
6334 for (; !isSPACE(*d) && len; --len, ++d)
6337 sv = newSVpvn_utf8(b, d-b, DO_UTF8(PL_lex_stuff));
6338 words = append_elem(OP_LIST, words,
6339 newSVOP(OP_CONST, 0, tokeq(sv)));
6343 start_force(PL_curforce);
6344 NEXTVAL_NEXTTOKE.opval = words;
6349 SvREFCNT_dec(PL_lex_stuff);
6350 PL_lex_stuff = NULL;
6356 s = scan_str(s,!!PL_madskills,FALSE);
6359 pl_yylval.ival = OP_STRINGIFY;
6360 if (SvIVX(PL_lex_stuff) == '\'')
6361 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6362 TERM(sublex_start());
6365 s = scan_pat(s,OP_QR);
6366 TERM(sublex_start());
6369 s = scan_str(s,!!PL_madskills,FALSE);
6372 readpipe_override();
6373 TERM(sublex_start());
6381 s = force_version(s, FALSE);
6383 else if (*s != 'v' || !isDIGIT(s[1])
6384 || (s = force_version(s, TRUE), *s == 'v'))
6386 *PL_tokenbuf = '\0';
6387 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6388 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6389 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6391 yyerror("<> should be quotes");
6393 if (orig_keyword == KEY_require) {
6401 PL_last_uni = PL_oldbufptr;
6402 PL_last_lop_op = OP_REQUIRE;
6404 return REPORT( (int)REQUIRE );
6410 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6414 LOP(OP_RENAME,XTERM);
6423 LOP(OP_RINDEX,XTERM);
6432 UNIDOR(OP_READLINE);
6435 UNIDOR(OP_BACKTICK);
6444 LOP(OP_REVERSE,XTERM);
6447 UNIDOR(OP_READLINK);
6454 if (pl_yylval.opval)
6455 TERM(sublex_start());
6457 TOKEN(1); /* force error */
6460 checkcomma(s,PL_tokenbuf,"filehandle");
6470 LOP(OP_SELECT,XTERM);
6476 LOP(OP_SEMCTL,XTERM);
6479 LOP(OP_SEMGET,XTERM);
6482 LOP(OP_SEMOP,XTERM);
6488 LOP(OP_SETPGRP,XTERM);
6490 case KEY_setpriority:
6491 LOP(OP_SETPRIORITY,XTERM);
6493 case KEY_sethostent:
6499 case KEY_setservent:
6502 case KEY_setprotoent:
6512 LOP(OP_SEEKDIR,XTERM);
6514 case KEY_setsockopt:
6515 LOP(OP_SSOCKOPT,XTERM);
6521 LOP(OP_SHMCTL,XTERM);
6524 LOP(OP_SHMGET,XTERM);
6527 LOP(OP_SHMREAD,XTERM);
6530 LOP(OP_SHMWRITE,XTERM);
6533 LOP(OP_SHUTDOWN,XTERM);
6542 LOP(OP_SOCKET,XTERM);
6544 case KEY_socketpair:
6545 LOP(OP_SOCKPAIR,XTERM);
6548 checkcomma(s,PL_tokenbuf,"subroutine name");
6550 if (*s == ';' || *s == ')') /* probably a close */
6551 Perl_croak(aTHX_ "sort is now a reserved word");
6553 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6557 LOP(OP_SPLIT,XTERM);
6560 LOP(OP_SPRINTF,XTERM);
6563 LOP(OP_SPLICE,XTERM);
6578 LOP(OP_SUBSTR,XTERM);
6584 char tmpbuf[sizeof PL_tokenbuf];
6585 SSize_t tboffset = 0;
6586 expectation attrful;
6587 bool have_name, have_proto;
6588 const int key = tmp;
6593 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6594 SV *subtoken = newSVpvn(tstart, s - tstart);
6598 s = SKIPSPACE2(s,tmpwhite);
6603 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6604 (*s == ':' && s[1] == ':'))
6611 attrful = XATTRBLOCK;
6612 /* remember buffer pos'n for later force_word */
6613 tboffset = s - PL_oldbufptr;
6614 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6617 nametoke = newSVpvn(s, d - s);
6619 if (memchr(tmpbuf, ':', len))
6620 sv_setpvn(PL_subname, tmpbuf, len);
6622 sv_setsv(PL_subname,PL_curstname);
6623 sv_catpvs(PL_subname,"::");
6624 sv_catpvn(PL_subname,tmpbuf,len);
6631 CURMAD('X', nametoke);
6632 CURMAD('_', tmpwhite);
6633 (void) force_word(PL_oldbufptr + tboffset, WORD,
6636 s = SKIPSPACE2(d,tmpwhite);
6643 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6644 PL_expect = XTERMBLOCK;
6645 attrful = XATTRTERM;
6646 sv_setpvn(PL_subname,"?",1);
6650 if (key == KEY_format) {
6652 PL_lex_formbrack = PL_lex_brackets + 1;
6654 PL_thistoken = subtoken;
6658 (void) force_word(PL_oldbufptr + tboffset, WORD,
6664 /* Look for a prototype */
6667 bool bad_proto = FALSE;
6668 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6670 s = scan_str(s,!!PL_madskills,FALSE);
6672 Perl_croak(aTHX_ "Prototype not terminated");
6673 /* strip spaces and check for bad characters */
6674 d = SvPVX(PL_lex_stuff);
6676 for (p = d; *p; ++p) {
6679 if (warnsyntax && !strchr("$@%*;[]&\\_", *p))
6685 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6686 "Illegal character in prototype for %"SVf" : %s",
6687 SVfARG(PL_subname), d);
6688 SvCUR_set(PL_lex_stuff, tmp);
6693 CURMAD('q', PL_thisopen);
6694 CURMAD('_', tmpwhite);
6695 CURMAD('=', PL_thisstuff);
6696 CURMAD('Q', PL_thisclose);
6697 NEXTVAL_NEXTTOKE.opval =
6698 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6699 PL_lex_stuff = NULL;
6702 s = SKIPSPACE2(s,tmpwhite);
6710 if (*s == ':' && s[1] != ':')
6711 PL_expect = attrful;
6712 else if (*s != '{' && key == KEY_sub) {
6714 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6716 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6723 curmad('^', newSVpvs(""));
6724 CURMAD('_', tmpwhite);
6728 PL_thistoken = subtoken;
6731 NEXTVAL_NEXTTOKE.opval =
6732 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6733 PL_lex_stuff = NULL;
6739 sv_setpvs(PL_subname, "__ANON__");
6741 sv_setpvs(PL_subname, "__ANON__::__ANON__");
6745 (void) force_word(PL_oldbufptr + tboffset, WORD,
6754 LOP(OP_SYSTEM,XREF);
6757 LOP(OP_SYMLINK,XTERM);
6760 LOP(OP_SYSCALL,XTERM);
6763 LOP(OP_SYSOPEN,XTERM);
6766 LOP(OP_SYSSEEK,XTERM);
6769 LOP(OP_SYSREAD,XTERM);
6772 LOP(OP_SYSWRITE,XTERM);
6776 TERM(sublex_start());
6797 LOP(OP_TRUNCATE,XTERM);
6809 pl_yylval.ival = CopLINE(PL_curcop);
6813 pl_yylval.ival = CopLINE(PL_curcop);
6817 LOP(OP_UNLINK,XTERM);
6823 LOP(OP_UNPACK,XTERM);
6826 LOP(OP_UTIME,XTERM);
6832 LOP(OP_UNSHIFT,XTERM);
6835 s = tokenize_use(1, s);
6845 pl_yylval.ival = CopLINE(PL_curcop);
6849 pl_yylval.ival = CopLINE(PL_curcop);
6853 PL_hints |= HINT_BLOCK_SCOPE;
6860 LOP(OP_WAITPID,XTERM);
6869 ctl_l[0] = toCTRL('L');
6871 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6874 /* Make sure $^L is defined */
6875 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
6880 if (PL_expect == XOPERATOR)
6886 pl_yylval.ival = OP_XOR;
6891 TERM(sublex_start());
6896 #pragma segment Main
6900 S_pending_ident(pTHX)
6905 /* pit holds the identifier we read and pending_ident is reset */
6906 char pit = PL_pending_ident;
6907 const STRLEN tokenbuf_len = strlen(PL_tokenbuf);
6908 /* All routes through this function want to know if there is a colon. */
6909 const char *const has_colon = (const char*) memchr (PL_tokenbuf, ':', tokenbuf_len);
6910 PL_pending_ident = 0;
6912 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
6913 DEBUG_T({ PerlIO_printf(Perl_debug_log,
6914 "### Pending identifier '%s'\n", PL_tokenbuf); });
6916 /* if we're in a my(), we can't allow dynamics here.
6917 $foo'bar has already been turned into $foo::bar, so
6918 just check for colons.
6920 if it's a legal name, the OP is a PADANY.
6923 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
6925 yyerror(Perl_form(aTHX_ "No package name allowed for "
6926 "variable %s in \"our\"",
6928 tmp = allocmy(PL_tokenbuf);
6932 yyerror(Perl_form(aTHX_ PL_no_myglob,
6933 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
6935 pl_yylval.opval = newOP(OP_PADANY, 0);
6936 pl_yylval.opval->op_targ = allocmy(PL_tokenbuf);
6942 build the ops for accesses to a my() variable.
6944 Deny my($a) or my($b) in a sort block, *if* $a or $b is
6945 then used in a comparison. This catches most, but not
6946 all cases. For instance, it catches
6947 sort { my($a); $a <=> $b }
6949 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
6950 (although why you'd do that is anyone's guess).
6955 tmp = pad_findmy(PL_tokenbuf);
6956 if (tmp != NOT_IN_PAD) {
6957 /* might be an "our" variable" */
6958 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
6959 /* build ops for a bareword */
6960 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
6961 HEK * const stashname = HvNAME_HEK(stash);
6962 SV * const sym = newSVhek(stashname);
6963 sv_catpvs(sym, "::");
6964 sv_catpvn(sym, PL_tokenbuf+1, tokenbuf_len - 1);
6965 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
6966 pl_yylval.opval->op_private = OPpCONST_ENTERED;
6969 ? (GV_ADDMULTI | GV_ADDINEVAL)
6972 ((PL_tokenbuf[0] == '$') ? SVt_PV
6973 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
6978 /* if it's a sort block and they're naming $a or $b */
6979 if (PL_last_lop_op == OP_SORT &&
6980 PL_tokenbuf[0] == '$' &&
6981 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
6984 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
6985 d < PL_bufend && *d != '\n';
6988 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
6989 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
6995 pl_yylval.opval = newOP(OP_PADANY, 0);
6996 pl_yylval.opval->op_targ = tmp;
7002 Whine if they've said @foo in a doublequoted string,
7003 and @foo isn't a variable we can find in the symbol
7006 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7007 GV *const gv = gv_fetchpvn_flags(PL_tokenbuf + 1, tokenbuf_len - 1, 0,
7009 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7010 && ckWARN(WARN_AMBIGUOUS)
7011 /* DO NOT warn for @- and @+ */
7012 && !( PL_tokenbuf[2] == '\0' &&
7013 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7016 /* Downgraded from fatal to warning 20000522 mjd */
7017 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7018 "Possible unintended interpolation of %s in string",
7023 /* build ops for a bareword */
7024 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpvn(PL_tokenbuf + 1,
7026 pl_yylval.opval->op_private = OPpCONST_ENTERED;
7028 PL_tokenbuf + 1, tokenbuf_len - 1,
7029 /* If the identifier refers to a stash, don't autovivify it.
7030 * Change 24660 had the side effect of causing symbol table
7031 * hashes to always be defined, even if they were freshly
7032 * created and the only reference in the entire program was
7033 * the single statement with the defined %foo::bar:: test.
7034 * It appears that all code in the wild doing this actually
7035 * wants to know whether sub-packages have been loaded, so
7036 * by avoiding auto-vivifying symbol tables, we ensure that
7037 * defined %foo::bar:: continues to be false, and the existing
7038 * tests still give the expected answers, even though what
7039 * they're actually testing has now changed subtly.
7041 (*PL_tokenbuf == '%'
7042 && *(d = PL_tokenbuf + tokenbuf_len - 1) == ':'
7045 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7046 ((PL_tokenbuf[0] == '$') ? SVt_PV
7047 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7053 * The following code was generated by perl_keyword.pl.
7057 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7062 case 1: /* 5 tokens of length 1 */
7094 case 2: /* 18 tokens of length 2 */
7240 case 3: /* 29 tokens of length 3 */
7244 if (name[1] == 'N' &&
7307 if (name[1] == 'i' &&
7339 if (name[1] == 'o' &&
7348 if (name[1] == 'e' &&
7357 if (name[1] == 'n' &&
7366 if (name[1] == 'o' &&
7375 if (name[1] == 'a' &&
7384 if (name[1] == 'o' &&
7446 if (name[1] == 'e' &&
7460 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7486 if (name[1] == 'i' &&
7495 if (name[1] == 's' &&
7504 if (name[1] == 'e' &&
7513 if (name[1] == 'o' &&
7525 case 4: /* 41 tokens of length 4 */
7529 if (name[1] == 'O' &&
7539 if (name[1] == 'N' &&
7549 if (name[1] == 'i' &&
7559 if (name[1] == 'h' &&
7569 if (name[1] == 'u' &&
7582 if (name[2] == 'c' &&
7591 if (name[2] == 's' &&
7600 if (name[2] == 'a' &&
7636 if (name[1] == 'o' &&
7649 if (name[2] == 't' &&
7658 if (name[2] == 'o' &&
7667 if (name[2] == 't' &&
7676 if (name[2] == 'e' &&
7689 if (name[1] == 'o' &&
7702 if (name[2] == 'y' &&
7711 if (name[2] == 'l' &&
7727 if (name[2] == 's' &&
7736 if (name[2] == 'n' &&
7745 if (name[2] == 'c' &&
7758 if (name[1] == 'e' &&
7768 if (name[1] == 'p' &&
7781 if (name[2] == 'c' &&
7790 if (name[2] == 'p' &&
7799 if (name[2] == 's' &&
7815 if (name[2] == 'n' &&
7885 if (name[2] == 'r' &&
7894 if (name[2] == 'r' &&
7903 if (name[2] == 'a' &&
7919 if (name[2] == 'l' &&
7981 if (name[2] == 'e' &&
7984 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
7997 case 5: /* 39 tokens of length 5 */
8001 if (name[1] == 'E' &&
8012 if (name[1] == 'H' &&
8026 if (name[2] == 'a' &&
8036 if (name[2] == 'a' &&
8053 if (name[2] == 'e' &&
8063 if (name[2] == 'e' &&
8067 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8083 if (name[3] == 'i' &&
8092 if (name[3] == 'o' &&
8128 if (name[2] == 'o' &&
8138 if (name[2] == 'y' &&
8152 if (name[1] == 'l' &&
8166 if (name[2] == 'n' &&
8176 if (name[2] == 'o' &&
8190 if (name[1] == 'i' &&
8195 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8204 if (name[2] == 'd' &&
8214 if (name[2] == 'c' &&
8231 if (name[2] == 'c' &&
8241 if (name[2] == 't' &&
8255 if (name[1] == 'k' &&
8266 if (name[1] == 'r' &&
8280 if (name[2] == 's' &&
8290 if (name[2] == 'd' &&
8307 if (name[2] == 'm' &&
8317 if (name[2] == 'i' &&
8327 if (name[2] == 'e' &&
8337 if (name[2] == 'l' &&
8347 if (name[2] == 'a' &&
8360 if (name[3] == 't' &&
8363 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8369 if (name[3] == 'd' &&
8386 if (name[1] == 'i' &&
8400 if (name[2] == 'a' &&
8413 if (name[3] == 'e' &&
8448 if (name[2] == 'i' &&
8465 if (name[2] == 'i' &&
8475 if (name[2] == 'i' &&
8492 case 6: /* 33 tokens of length 6 */
8496 if (name[1] == 'c' &&
8511 if (name[2] == 'l' &&
8522 if (name[2] == 'r' &&
8537 if (name[1] == 'e' &&
8552 if (name[2] == 's' &&
8557 if(ckWARN_d(WARN_SYNTAX))
8558 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8564 if (name[2] == 'i' &&
8582 if (name[2] == 'l' &&
8593 if (name[2] == 'r' &&
8608 if (name[1] == 'm' &&
8623 if (name[2] == 'n' &&
8634 if (name[2] == 's' &&
8649 if (name[1] == 's' &&
8655 if (name[4] == 't' &&
8664 if (name[4] == 'e' &&
8673 if (name[4] == 'c' &&
8682 if (name[4] == 'n' &&
8698 if (name[1] == 'r' &&
8716 if (name[3] == 'a' &&
8726 if (name[3] == 'u' &&
8740 if (name[2] == 'n' &&
8758 if (name[2] == 'a' &&
8772 if (name[3] == 'e' &&
8785 if (name[4] == 't' &&
8794 if (name[4] == 'e' &&
8816 if (name[4] == 't' &&
8825 if (name[4] == 'e' &&
8841 if (name[2] == 'c' &&
8852 if (name[2] == 'l' &&
8863 if (name[2] == 'b' &&
8874 if (name[2] == 's' &&
8897 if (name[4] == 's' &&
8906 if (name[4] == 'n' &&
8919 if (name[3] == 'a' &&
8936 if (name[1] == 'a' &&
8951 case 7: /* 29 tokens of length 7 */
8955 if (name[1] == 'E' &&
8968 if (name[1] == '_' &&
8981 if (name[1] == 'i' &&
8988 return -KEY_binmode;
8994 if (name[1] == 'o' &&
9001 return -KEY_connect;
9010 if (name[2] == 'm' &&
9016 return -KEY_dbmopen;
9027 if (name[4] == 'u' &&
9031 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9037 if (name[4] == 'n' &&
9058 if (name[1] == 'o' &&
9071 if (name[1] == 'e' &&
9078 if (name[5] == 'r' &&
9081 return -KEY_getpgrp;
9087 if (name[5] == 'i' &&
9090 return -KEY_getppid;
9103 if (name[1] == 'c' &&
9110 return -KEY_lcfirst;
9116 if (name[1] == 'p' &&
9123 return -KEY_opendir;
9129 if (name[1] == 'a' &&
9147 if (name[3] == 'd' &&
9152 return -KEY_readdir;
9158 if (name[3] == 'u' &&
9169 if (name[3] == 'e' &&
9174 return -KEY_reverse;
9193 if (name[3] == 'k' &&
9198 return -KEY_seekdir;
9204 if (name[3] == 'p' &&
9209 return -KEY_setpgrp;
9219 if (name[2] == 'm' &&
9225 return -KEY_shmread;
9231 if (name[2] == 'r' &&
9237 return -KEY_sprintf;
9246 if (name[3] == 'l' &&
9251 return -KEY_symlink;
9260 if (name[4] == 'a' &&
9264 return -KEY_syscall;
9270 if (name[4] == 'p' &&
9274 return -KEY_sysopen;
9280 if (name[4] == 'e' &&
9284 return -KEY_sysread;
9290 if (name[4] == 'e' &&
9294 return -KEY_sysseek;
9312 if (name[1] == 'e' &&
9319 return -KEY_telldir;
9328 if (name[2] == 'f' &&
9334 return -KEY_ucfirst;
9340 if (name[2] == 's' &&
9346 return -KEY_unshift;
9356 if (name[1] == 'a' &&
9363 return -KEY_waitpid;
9372 case 8: /* 26 tokens of length 8 */
9376 if (name[1] == 'U' &&
9384 return KEY_AUTOLOAD;
9395 if (name[3] == 'A' &&
9401 return KEY___DATA__;
9407 if (name[3] == 'I' &&
9413 return -KEY___FILE__;
9419 if (name[3] == 'I' &&
9425 return -KEY___LINE__;
9441 if (name[2] == 'o' &&
9448 return -KEY_closedir;
9454 if (name[2] == 'n' &&
9461 return -KEY_continue;
9471 if (name[1] == 'b' &&
9479 return -KEY_dbmclose;
9485 if (name[1] == 'n' &&
9491 if (name[4] == 'r' &&
9496 return -KEY_endgrent;
9502 if (name[4] == 'w' &&
9507 return -KEY_endpwent;
9520 if (name[1] == 'o' &&
9528 return -KEY_formline;
9534 if (name[1] == 'e' &&
9545 if (name[6] == 'n' &&
9548 return -KEY_getgrent;
9554 if (name[6] == 'i' &&
9557 return -KEY_getgrgid;
9563 if (name[6] == 'a' &&
9566 return -KEY_getgrnam;
9579 if (name[4] == 'o' &&
9584 return -KEY_getlogin;
9595 if (name[6] == 'n' &&
9598 return -KEY_getpwent;
9604 if (name[6] == 'a' &&
9607 return -KEY_getpwnam;
9613 if (name[6] == 'i' &&
9616 return -KEY_getpwuid;
9636 if (name[1] == 'e' &&
9643 if (name[5] == 'i' &&
9650 return -KEY_readline;
9655 return -KEY_readlink;
9666 if (name[5] == 'i' &&
9670 return -KEY_readpipe;
9691 if (name[4] == 'r' &&
9696 return -KEY_setgrent;
9702 if (name[4] == 'w' &&
9707 return -KEY_setpwent;
9723 if (name[3] == 'w' &&
9729 return -KEY_shmwrite;
9735 if (name[3] == 't' &&
9741 return -KEY_shutdown;
9751 if (name[2] == 's' &&
9758 return -KEY_syswrite;
9768 if (name[1] == 'r' &&
9776 return -KEY_truncate;
9785 case 9: /* 9 tokens of length 9 */
9789 if (name[1] == 'N' &&
9798 return KEY_UNITCHECK;
9804 if (name[1] == 'n' &&
9813 return -KEY_endnetent;
9819 if (name[1] == 'e' &&
9828 return -KEY_getnetent;
9834 if (name[1] == 'o' &&
9843 return -KEY_localtime;
9849 if (name[1] == 'r' &&
9858 return KEY_prototype;
9864 if (name[1] == 'u' &&
9873 return -KEY_quotemeta;
9879 if (name[1] == 'e' &&
9888 return -KEY_rewinddir;
9894 if (name[1] == 'e' &&
9903 return -KEY_setnetent;
9909 if (name[1] == 'a' &&
9918 return -KEY_wantarray;
9927 case 10: /* 9 tokens of length 10 */
9931 if (name[1] == 'n' &&
9937 if (name[4] == 'o' &&
9944 return -KEY_endhostent;
9950 if (name[4] == 'e' &&
9957 return -KEY_endservent;
9970 if (name[1] == 'e' &&
9976 if (name[4] == 'o' &&
9983 return -KEY_gethostent;
9992 if (name[5] == 'r' &&
9998 return -KEY_getservent;
10004 if (name[5] == 'c' &&
10010 return -KEY_getsockopt;
10030 if (name[2] == 't')
10035 if (name[4] == 'o' &&
10042 return -KEY_sethostent;
10051 if (name[5] == 'r' &&
10057 return -KEY_setservent;
10063 if (name[5] == 'c' &&
10069 return -KEY_setsockopt;
10086 if (name[2] == 'c' &&
10095 return -KEY_socketpair;
10108 case 11: /* 8 tokens of length 11 */
10112 if (name[1] == '_' &&
10122 { /* __PACKAGE__ */
10123 return -KEY___PACKAGE__;
10129 if (name[1] == 'n' &&
10139 { /* endprotoent */
10140 return -KEY_endprotoent;
10146 if (name[1] == 'e' &&
10155 if (name[5] == 'e' &&
10161 { /* getpeername */
10162 return -KEY_getpeername;
10171 if (name[6] == 'o' &&
10176 { /* getpriority */
10177 return -KEY_getpriority;
10183 if (name[6] == 't' &&
10188 { /* getprotoent */
10189 return -KEY_getprotoent;
10203 if (name[4] == 'o' &&
10210 { /* getsockname */
10211 return -KEY_getsockname;
10224 if (name[1] == 'e' &&
10232 if (name[6] == 'o' &&
10237 { /* setpriority */
10238 return -KEY_setpriority;
10244 if (name[6] == 't' &&
10249 { /* setprotoent */
10250 return -KEY_setprotoent;
10266 case 12: /* 2 tokens of length 12 */
10267 if (name[0] == 'g' &&
10279 if (name[9] == 'd' &&
10282 { /* getnetbyaddr */
10283 return -KEY_getnetbyaddr;
10289 if (name[9] == 'a' &&
10292 { /* getnetbyname */
10293 return -KEY_getnetbyname;
10305 case 13: /* 4 tokens of length 13 */
10306 if (name[0] == 'g' &&
10313 if (name[4] == 'o' &&
10322 if (name[10] == 'd' &&
10325 { /* gethostbyaddr */
10326 return -KEY_gethostbyaddr;
10332 if (name[10] == 'a' &&
10335 { /* gethostbyname */
10336 return -KEY_gethostbyname;
10349 if (name[4] == 'e' &&
10358 if (name[10] == 'a' &&
10361 { /* getservbyname */
10362 return -KEY_getservbyname;
10368 if (name[10] == 'o' &&
10371 { /* getservbyport */
10372 return -KEY_getservbyport;
10391 case 14: /* 1 tokens of length 14 */
10392 if (name[0] == 'g' &&
10406 { /* getprotobyname */
10407 return -KEY_getprotobyname;
10412 case 16: /* 1 tokens of length 16 */
10413 if (name[0] == 'g' &&
10429 { /* getprotobynumber */
10430 return -KEY_getprotobynumber;
10444 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10448 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10449 if (ckWARN(WARN_SYNTAX)) {
10452 for (w = s+2; *w && level; w++) {
10455 else if (*w == ')')
10458 while (isSPACE(*w))
10460 /* the list of chars below is for end of statements or
10461 * block / parens, boolean operators (&&, ||, //) and branch
10462 * constructs (or, and, if, until, unless, while, err, for).
10463 * Not a very solid hack... */
10464 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10465 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10466 "%s (...) interpreted as function",name);
10469 while (s < PL_bufend && isSPACE(*s))
10473 while (s < PL_bufend && isSPACE(*s))
10475 if (isIDFIRST_lazy_if(s,UTF)) {
10476 const char * const w = s++;
10477 while (isALNUM_lazy_if(s,UTF))
10479 while (s < PL_bufend && isSPACE(*s))
10483 if (keyword(w, s - w, 0))
10486 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10487 if (gv && GvCVu(gv))
10489 Perl_croak(aTHX_ "No comma allowed after %s", what);
10494 /* Either returns sv, or mortalizes sv and returns a new SV*.
10495 Best used as sv=new_constant(..., sv, ...).
10496 If s, pv are NULL, calls subroutine with one argument,
10497 and type is used with error messages only. */
10500 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, STRLEN keylen,
10501 SV *sv, SV *pv, const char *type, STRLEN typelen)
10504 HV * const table = GvHV(PL_hintgv); /* ^H */
10508 const char *why1 = "", *why2 = "", *why3 = "";
10510 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10513 why2 = (const char *)
10514 (strEQ(key,"charnames")
10515 ? "(possibly a missing \"use charnames ...\")"
10517 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10518 (type ? type: "undef"), why2);
10520 /* This is convoluted and evil ("goto considered harmful")
10521 * but I do not understand the intricacies of all the different
10522 * failure modes of %^H in here. The goal here is to make
10523 * the most probable error message user-friendly. --jhi */
10528 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10529 (type ? type: "undef"), why1, why2, why3);
10531 yyerror(SvPVX_const(msg));
10535 cvp = hv_fetch(table, key, keylen, FALSE);
10536 if (!cvp || !SvOK(*cvp)) {
10539 why3 = "} is not defined";
10542 sv_2mortal(sv); /* Parent created it permanently */
10545 pv = newSVpvn_flags(s, len, SVs_TEMP);
10547 typesv = newSVpvn_flags(type, typelen, SVs_TEMP);
10549 typesv = &PL_sv_undef;
10551 PUSHSTACKi(PERLSI_OVERLOAD);
10563 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10567 /* Check the eval first */
10568 if (!PL_in_eval && SvTRUE(ERRSV)) {
10569 sv_catpvs(ERRSV, "Propagated");
10570 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10572 res = SvREFCNT_inc_simple(sv);
10576 SvREFCNT_inc_simple_void(res);
10585 why1 = "Call to &{$^H{";
10587 why3 = "}} did not return a defined value";
10595 /* Returns a NUL terminated string, with the length of the string written to
10599 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10602 register char *d = dest;
10603 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10606 Perl_croak(aTHX_ ident_too_long);
10607 if (isALNUM(*s)) /* UTF handled below */
10609 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10614 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10618 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10619 char *t = s + UTF8SKIP(s);
10621 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10625 Perl_croak(aTHX_ ident_too_long);
10626 Copy(s, d, len, char);
10639 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10642 char *bracket = NULL;
10644 register char *d = dest;
10645 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10650 while (isDIGIT(*s)) {
10652 Perl_croak(aTHX_ ident_too_long);
10659 Perl_croak(aTHX_ ident_too_long);
10660 if (isALNUM(*s)) /* UTF handled below */
10662 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10667 else if (*s == ':' && s[1] == ':') {
10671 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10672 char *t = s + UTF8SKIP(s);
10673 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10675 if (d + (t - s) > e)
10676 Perl_croak(aTHX_ ident_too_long);
10677 Copy(s, d, t - s, char);
10688 if (PL_lex_state != LEX_NORMAL)
10689 PL_lex_state = LEX_INTERPENDMAYBE;
10692 if (*s == '$' && s[1] &&
10693 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10706 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10711 if (isSPACE(s[-1])) {
10713 const char ch = *s++;
10714 if (!SPACE_OR_TAB(ch)) {
10720 if (isIDFIRST_lazy_if(d,UTF)) {
10724 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10725 end += UTF8SKIP(end);
10726 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10727 end += UTF8SKIP(end);
10729 Copy(s, d, end - s, char);
10734 while ((isALNUM(*s) || *s == ':') && d < e)
10737 Perl_croak(aTHX_ ident_too_long);
10740 while (s < send && SPACE_OR_TAB(*s))
10742 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10743 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10744 const char * const brack =
10746 ((*s == '[') ? "[...]" : "{...}");
10747 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10748 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10749 funny, dest, brack, funny, dest, brack);
10752 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10756 /* Handle extended ${^Foo} variables
10757 * 1999-02-27 mjd-perl-patch@plover.com */
10758 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10762 while (isALNUM(*s) && d < e) {
10766 Perl_croak(aTHX_ ident_too_long);
10771 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10772 PL_lex_state = LEX_INTERPEND;
10775 if (PL_lex_state == LEX_NORMAL) {
10776 if (ckWARN(WARN_AMBIGUOUS) &&
10777 (keyword(dest, d - dest, 0)
10778 || get_cvn_flags(dest, d - dest, 0)))
10782 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10783 "Ambiguous use of %c{%s} resolved to %c%s",
10784 funny, dest, funny, dest);
10789 s = bracket; /* let the parser handle it */
10793 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10794 PL_lex_state = LEX_INTERPEND;
10799 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10801 PERL_UNUSED_CONTEXT;
10803 const char c = (char)ch;
10805 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10806 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10807 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10808 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10809 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10815 S_scan_pat(pTHX_ char *start, I32 type)
10819 char *s = scan_str(start,!!PL_madskills,FALSE);
10820 const char * const valid_flags =
10821 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10828 const char * const delimiter = skipspace(start);
10832 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10833 : "Search pattern not terminated" ));
10836 pm = (PMOP*)newPMOP(type, 0);
10837 if (PL_multi_open == '?') {
10838 /* This is the only point in the code that sets PMf_ONCE: */
10839 pm->op_pmflags |= PMf_ONCE;
10841 /* Hence it's safe to do this bit of PMOP book-keeping here, which
10842 allows us to restrict the list needed by reset to just the ??
10844 assert(type != OP_TRANS);
10846 MAGIC *mg = mg_find((SV*)PL_curstash, PERL_MAGIC_symtab);
10849 mg = sv_magicext((SV*)PL_curstash, 0, PERL_MAGIC_symtab, 0, 0,
10852 elements = mg->mg_len / sizeof(PMOP**);
10853 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
10854 ((PMOP**)mg->mg_ptr) [elements++] = pm;
10855 mg->mg_len = elements * sizeof(PMOP**);
10856 PmopSTASH_set(pm,PL_curstash);
10862 while (*s && strchr(valid_flags, *s))
10863 pmflag(&pm->op_pmflags,*s++);
10865 if (PL_madskills && modstart != s) {
10866 SV* tmptoken = newSVpvn(modstart, s - modstart);
10867 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
10870 /* issue a warning if /c is specified,but /g is not */
10871 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
10872 && ckWARN(WARN_REGEXP))
10874 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
10875 "Use of /c modifier is meaningless without /g" );
10878 PL_lex_op = (OP*)pm;
10879 pl_yylval.ival = OP_MATCH;
10884 S_scan_subst(pTHX_ char *start)
10895 pl_yylval.ival = OP_NULL;
10897 s = scan_str(start,!!PL_madskills,FALSE);
10900 Perl_croak(aTHX_ "Substitution pattern not terminated");
10902 if (s[-1] == PL_multi_open)
10905 if (PL_madskills) {
10906 CURMAD('q', PL_thisopen);
10907 CURMAD('_', PL_thiswhite);
10908 CURMAD('E', PL_thisstuff);
10909 CURMAD('Q', PL_thisclose);
10910 PL_realtokenstart = s - SvPVX(PL_linestr);
10914 first_start = PL_multi_start;
10915 s = scan_str(s,!!PL_madskills,FALSE);
10917 if (PL_lex_stuff) {
10918 SvREFCNT_dec(PL_lex_stuff);
10919 PL_lex_stuff = NULL;
10921 Perl_croak(aTHX_ "Substitution replacement not terminated");
10923 PL_multi_start = first_start; /* so whole substitution is taken together */
10925 pm = (PMOP*)newPMOP(OP_SUBST, 0);
10928 if (PL_madskills) {
10929 CURMAD('z', PL_thisopen);
10930 CURMAD('R', PL_thisstuff);
10931 CURMAD('Z', PL_thisclose);
10937 if (*s == EXEC_PAT_MOD) {
10941 else if (strchr(S_PAT_MODS, *s))
10942 pmflag(&pm->op_pmflags,*s++);
10948 if (PL_madskills) {
10950 curmad('m', newSVpvn(modstart, s - modstart));
10951 append_madprops(PL_thismad, (OP*)pm, 0);
10955 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
10956 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
10960 SV * const repl = newSVpvs("");
10962 PL_sublex_info.super_bufptr = s;
10963 PL_sublex_info.super_bufend = PL_bufend;
10965 pm->op_pmflags |= PMf_EVAL;
10968 sv_catpvs(repl, "eval ");
10970 sv_catpvs(repl, "do ");
10972 sv_catpvs(repl, "{");
10973 sv_catsv(repl, PL_lex_repl);
10974 if (strchr(SvPVX(PL_lex_repl), '#'))
10975 sv_catpvs(repl, "\n");
10976 sv_catpvs(repl, "}");
10978 SvREFCNT_dec(PL_lex_repl);
10979 PL_lex_repl = repl;
10982 PL_lex_op = (OP*)pm;
10983 pl_yylval.ival = OP_SUBST;
10988 S_scan_trans(pTHX_ char *start)
11001 pl_yylval.ival = OP_NULL;
11003 s = scan_str(start,!!PL_madskills,FALSE);
11005 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11007 if (s[-1] == PL_multi_open)
11010 if (PL_madskills) {
11011 CURMAD('q', PL_thisopen);
11012 CURMAD('_', PL_thiswhite);
11013 CURMAD('E', PL_thisstuff);
11014 CURMAD('Q', PL_thisclose);
11015 PL_realtokenstart = s - SvPVX(PL_linestr);
11019 s = scan_str(s,!!PL_madskills,FALSE);
11021 if (PL_lex_stuff) {
11022 SvREFCNT_dec(PL_lex_stuff);
11023 PL_lex_stuff = NULL;
11025 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11027 if (PL_madskills) {
11028 CURMAD('z', PL_thisopen);
11029 CURMAD('R', PL_thisstuff);
11030 CURMAD('Z', PL_thisclose);
11033 complement = del = squash = 0;
11040 complement = OPpTRANS_COMPLEMENT;
11043 del = OPpTRANS_DELETE;
11046 squash = OPpTRANS_SQUASH;
11055 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11056 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11057 o->op_private &= ~OPpTRANS_ALL;
11058 o->op_private |= del|squash|complement|
11059 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11060 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11063 pl_yylval.ival = OP_TRANS;
11066 if (PL_madskills) {
11068 curmad('m', newSVpvn(modstart, s - modstart));
11069 append_madprops(PL_thismad, o, 0);
11078 S_scan_heredoc(pTHX_ register char *s)
11082 I32 op_type = OP_SCALAR;
11086 const char *found_newline;
11090 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11092 I32 stuffstart = s - SvPVX(PL_linestr);
11095 PL_realtokenstart = -1;
11100 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11104 while (SPACE_OR_TAB(*peek))
11106 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11109 s = delimcpy(d, e, s, PL_bufend, term, &len);
11119 if (!isALNUM_lazy_if(s,UTF))
11120 deprecate_old("bare << to mean <<\"\"");
11121 for (; isALNUM_lazy_if(s,UTF); s++) {
11126 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11127 Perl_croak(aTHX_ "Delimiter for here document is too long");
11130 len = d - PL_tokenbuf;
11133 if (PL_madskills) {
11134 tstart = PL_tokenbuf + !outer;
11135 PL_thisclose = newSVpvn(tstart, len - !outer);
11136 tstart = SvPVX(PL_linestr) + stuffstart;
11137 PL_thisopen = newSVpvn(tstart, s - tstart);
11138 stuffstart = s - SvPVX(PL_linestr);
11141 #ifndef PERL_STRICT_CR
11142 d = strchr(s, '\r');
11144 char * const olds = s;
11146 while (s < PL_bufend) {
11152 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11161 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11168 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11169 herewas = newSVpvn(s,PL_bufend-s);
11173 herewas = newSVpvn(s-1,found_newline-s+1);
11176 herewas = newSVpvn(s,found_newline-s);
11180 if (PL_madskills) {
11181 tstart = SvPVX(PL_linestr) + stuffstart;
11183 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11185 PL_thisstuff = newSVpvn(tstart, s - tstart);
11188 s += SvCUR(herewas);
11191 stuffstart = s - SvPVX(PL_linestr);
11197 tmpstr = newSV_type(SVt_PVIV);
11198 SvGROW(tmpstr, 80);
11199 if (term == '\'') {
11200 op_type = OP_CONST;
11201 SvIV_set(tmpstr, -1);
11203 else if (term == '`') {
11204 op_type = OP_BACKTICK;
11205 SvIV_set(tmpstr, '\\');
11209 PL_multi_start = CopLINE(PL_curcop);
11210 PL_multi_open = PL_multi_close = '<';
11211 term = *PL_tokenbuf;
11212 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11213 char * const bufptr = PL_sublex_info.super_bufptr;
11214 char * const bufend = PL_sublex_info.super_bufend;
11215 char * const olds = s - SvCUR(herewas);
11216 s = strchr(bufptr, '\n');
11220 while (s < bufend &&
11221 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11223 CopLINE_inc(PL_curcop);
11226 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11227 missingterm(PL_tokenbuf);
11229 sv_setpvn(herewas,bufptr,d-bufptr+1);
11230 sv_setpvn(tmpstr,d+1,s-d);
11232 sv_catpvn(herewas,s,bufend-s);
11233 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11240 while (s < PL_bufend &&
11241 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11243 CopLINE_inc(PL_curcop);
11245 if (s >= PL_bufend) {
11246 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11247 missingterm(PL_tokenbuf);
11249 sv_setpvn(tmpstr,d+1,s-d);
11251 if (PL_madskills) {
11253 sv_catpvn(PL_thisstuff, d + 1, s - d);
11255 PL_thisstuff = newSVpvn(d + 1, s - d);
11256 stuffstart = s - SvPVX(PL_linestr);
11260 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11262 sv_catpvn(herewas,s,PL_bufend-s);
11263 sv_setsv(PL_linestr,herewas);
11264 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11265 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11266 PL_last_lop = PL_last_uni = NULL;
11269 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
11270 while (s >= PL_bufend) { /* multiple line string? */
11272 if (PL_madskills) {
11273 tstart = SvPVX(PL_linestr) + stuffstart;
11275 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11277 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11281 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11282 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11283 missingterm(PL_tokenbuf);
11286 stuffstart = s - SvPVX(PL_linestr);
11288 CopLINE_inc(PL_curcop);
11289 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11290 PL_last_lop = PL_last_uni = NULL;
11291 #ifndef PERL_STRICT_CR
11292 if (PL_bufend - PL_linestart >= 2) {
11293 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11294 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11296 PL_bufend[-2] = '\n';
11298 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11300 else if (PL_bufend[-1] == '\r')
11301 PL_bufend[-1] = '\n';
11303 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11304 PL_bufend[-1] = '\n';
11306 if (PERLDB_LINE && PL_curstash != PL_debstash)
11307 update_debugger_info(PL_linestr, NULL, 0);
11308 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11309 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11310 *(SvPVX(PL_linestr) + off ) = ' ';
11311 sv_catsv(PL_linestr,herewas);
11312 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11313 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11317 sv_catsv(tmpstr,PL_linestr);
11322 PL_multi_end = CopLINE(PL_curcop);
11323 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11324 SvPV_shrink_to_cur(tmpstr);
11326 SvREFCNT_dec(herewas);
11328 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11330 else if (PL_encoding)
11331 sv_recode_to_utf8(tmpstr, PL_encoding);
11333 PL_lex_stuff = tmpstr;
11334 pl_yylval.ival = op_type;
11338 /* scan_inputsymbol
11339 takes: current position in input buffer
11340 returns: new position in input buffer
11341 side-effects: pl_yylval and lex_op are set.
11346 <FH> read from filehandle
11347 <pkg::FH> read from package qualified filehandle
11348 <pkg'FH> read from package qualified filehandle
11349 <$fh> read from filehandle in $fh
11350 <*.h> filename glob
11355 S_scan_inputsymbol(pTHX_ char *start)
11358 register char *s = start; /* current position in buffer */
11362 char *d = PL_tokenbuf; /* start of temp holding space */
11363 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11365 end = strchr(s, '\n');
11368 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11370 /* die if we didn't have space for the contents of the <>,
11371 or if it didn't end, or if we see a newline
11374 if (len >= (I32)sizeof PL_tokenbuf)
11375 Perl_croak(aTHX_ "Excessively long <> operator");
11377 Perl_croak(aTHX_ "Unterminated <> operator");
11382 Remember, only scalar variables are interpreted as filehandles by
11383 this code. Anything more complex (e.g., <$fh{$num}>) will be
11384 treated as a glob() call.
11385 This code makes use of the fact that except for the $ at the front,
11386 a scalar variable and a filehandle look the same.
11388 if (*d == '$' && d[1]) d++;
11390 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11391 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11394 /* If we've tried to read what we allow filehandles to look like, and
11395 there's still text left, then it must be a glob() and not a getline.
11396 Use scan_str to pull out the stuff between the <> and treat it
11397 as nothing more than a string.
11400 if (d - PL_tokenbuf != len) {
11401 pl_yylval.ival = OP_GLOB;
11402 s = scan_str(start,!!PL_madskills,FALSE);
11404 Perl_croak(aTHX_ "Glob not terminated");
11408 bool readline_overriden = FALSE;
11411 /* we're in a filehandle read situation */
11414 /* turn <> into <ARGV> */
11416 Copy("ARGV",d,5,char);
11418 /* Check whether readline() is overriden */
11419 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11421 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11423 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11424 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11425 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11426 readline_overriden = TRUE;
11428 /* if <$fh>, create the ops to turn the variable into a
11432 /* try to find it in the pad for this block, otherwise find
11433 add symbol table ops
11435 const PADOFFSET tmp = pad_findmy(d);
11436 if (tmp != NOT_IN_PAD) {
11437 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11438 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11439 HEK * const stashname = HvNAME_HEK(stash);
11440 SV * const sym = sv_2mortal(newSVhek(stashname));
11441 sv_catpvs(sym, "::");
11442 sv_catpv(sym, d+1);
11447 OP * const o = newOP(OP_PADSV, 0);
11449 PL_lex_op = readline_overriden
11450 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11451 append_elem(OP_LIST, o,
11452 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11453 : (OP*)newUNOP(OP_READLINE, 0, o);
11462 ? (GV_ADDMULTI | GV_ADDINEVAL)
11465 PL_lex_op = readline_overriden
11466 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11467 append_elem(OP_LIST,
11468 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11469 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11470 : (OP*)newUNOP(OP_READLINE, 0,
11471 newUNOP(OP_RV2SV, 0,
11472 newGVOP(OP_GV, 0, gv)));
11474 if (!readline_overriden)
11475 PL_lex_op->op_flags |= OPf_SPECIAL;
11476 /* we created the ops in PL_lex_op, so make pl_yylval.ival a null op */
11477 pl_yylval.ival = OP_NULL;
11480 /* If it's none of the above, it must be a literal filehandle
11481 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11483 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11484 PL_lex_op = readline_overriden
11485 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11486 append_elem(OP_LIST,
11487 newGVOP(OP_GV, 0, gv),
11488 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11489 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11490 pl_yylval.ival = OP_NULL;
11499 takes: start position in buffer
11500 keep_quoted preserve \ on the embedded delimiter(s)
11501 keep_delims preserve the delimiters around the string
11502 returns: position to continue reading from buffer
11503 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11504 updates the read buffer.
11506 This subroutine pulls a string out of the input. It is called for:
11507 q single quotes q(literal text)
11508 ' single quotes 'literal text'
11509 qq double quotes qq(interpolate $here please)
11510 " double quotes "interpolate $here please"
11511 qx backticks qx(/bin/ls -l)
11512 ` backticks `/bin/ls -l`
11513 qw quote words @EXPORT_OK = qw( func() $spam )
11514 m// regexp match m/this/
11515 s/// regexp substitute s/this/that/
11516 tr/// string transliterate tr/this/that/
11517 y/// string transliterate y/this/that/
11518 ($*@) sub prototypes sub foo ($)
11519 (stuff) sub attr parameters sub foo : attr(stuff)
11520 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11522 In most of these cases (all but <>, patterns and transliterate)
11523 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11524 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11525 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11528 It skips whitespace before the string starts, and treats the first
11529 character as the delimiter. If the delimiter is one of ([{< then
11530 the corresponding "close" character )]}> is used as the closing
11531 delimiter. It allows quoting of delimiters, and if the string has
11532 balanced delimiters ([{<>}]) it allows nesting.
11534 On success, the SV with the resulting string is put into lex_stuff or,
11535 if that is already non-NULL, into lex_repl. The second case occurs only
11536 when parsing the RHS of the special constructs s/// and tr/// (y///).
11537 For convenience, the terminating delimiter character is stuffed into
11542 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11545 SV *sv; /* scalar value: string */
11546 const char *tmps; /* temp string, used for delimiter matching */
11547 register char *s = start; /* current position in the buffer */
11548 register char term; /* terminating character */
11549 register char *to; /* current position in the sv's data */
11550 I32 brackets = 1; /* bracket nesting level */
11551 bool has_utf8 = FALSE; /* is there any utf8 content? */
11552 I32 termcode; /* terminating char. code */
11553 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11554 STRLEN termlen; /* length of terminating string */
11555 int last_off = 0; /* last position for nesting bracket */
11561 /* skip space before the delimiter */
11567 if (PL_realtokenstart >= 0) {
11568 stuffstart = PL_realtokenstart;
11569 PL_realtokenstart = -1;
11572 stuffstart = start - SvPVX(PL_linestr);
11574 /* mark where we are, in case we need to report errors */
11577 /* after skipping whitespace, the next character is the terminator */
11580 termcode = termstr[0] = term;
11584 termcode = utf8_to_uvchr((U8*)s, &termlen);
11585 Copy(s, termstr, termlen, U8);
11586 if (!UTF8_IS_INVARIANT(term))
11590 /* mark where we are */
11591 PL_multi_start = CopLINE(PL_curcop);
11592 PL_multi_open = term;
11594 /* find corresponding closing delimiter */
11595 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11596 termcode = termstr[0] = term = tmps[5];
11598 PL_multi_close = term;
11600 /* create a new SV to hold the contents. 79 is the SV's initial length.
11601 What a random number. */
11602 sv = newSV_type(SVt_PVIV);
11604 SvIV_set(sv, termcode);
11605 (void)SvPOK_only(sv); /* validate pointer */
11607 /* move past delimiter and try to read a complete string */
11609 sv_catpvn(sv, s, termlen);
11612 tstart = SvPVX(PL_linestr) + stuffstart;
11613 if (!PL_thisopen && !keep_delims) {
11614 PL_thisopen = newSVpvn(tstart, s - tstart);
11615 stuffstart = s - SvPVX(PL_linestr);
11619 if (PL_encoding && !UTF) {
11623 int offset = s - SvPVX_const(PL_linestr);
11624 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11625 &offset, (char*)termstr, termlen);
11626 const char * const ns = SvPVX_const(PL_linestr) + offset;
11627 char * const svlast = SvEND(sv) - 1;
11629 for (; s < ns; s++) {
11630 if (*s == '\n' && !PL_rsfp)
11631 CopLINE_inc(PL_curcop);
11634 goto read_more_line;
11636 /* handle quoted delimiters */
11637 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11639 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11641 if ((svlast-1 - t) % 2) {
11642 if (!keep_quoted) {
11643 *(svlast-1) = term;
11645 SvCUR_set(sv, SvCUR(sv) - 1);
11650 if (PL_multi_open == PL_multi_close) {
11656 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11657 /* At here, all closes are "was quoted" one,
11658 so we don't check PL_multi_close. */
11660 if (!keep_quoted && *(t+1) == PL_multi_open)
11665 else if (*t == PL_multi_open)
11673 SvCUR_set(sv, w - SvPVX_const(sv));
11675 last_off = w - SvPVX(sv);
11676 if (--brackets <= 0)
11681 if (!keep_delims) {
11682 SvCUR_set(sv, SvCUR(sv) - 1);
11688 /* extend sv if need be */
11689 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11690 /* set 'to' to the next character in the sv's string */
11691 to = SvPVX(sv)+SvCUR(sv);
11693 /* if open delimiter is the close delimiter read unbridle */
11694 if (PL_multi_open == PL_multi_close) {
11695 for (; s < PL_bufend; s++,to++) {
11696 /* embedded newlines increment the current line number */
11697 if (*s == '\n' && !PL_rsfp)
11698 CopLINE_inc(PL_curcop);
11699 /* handle quoted delimiters */
11700 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11701 if (!keep_quoted && s[1] == term)
11703 /* any other quotes are simply copied straight through */
11707 /* terminate when run out of buffer (the for() condition), or
11708 have found the terminator */
11709 else if (*s == term) {
11712 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11715 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11721 /* if the terminator isn't the same as the start character (e.g.,
11722 matched brackets), we have to allow more in the quoting, and
11723 be prepared for nested brackets.
11726 /* read until we run out of string, or we find the terminator */
11727 for (; s < PL_bufend; s++,to++) {
11728 /* embedded newlines increment the line count */
11729 if (*s == '\n' && !PL_rsfp)
11730 CopLINE_inc(PL_curcop);
11731 /* backslashes can escape the open or closing characters */
11732 if (*s == '\\' && s+1 < PL_bufend) {
11733 if (!keep_quoted &&
11734 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11739 /* allow nested opens and closes */
11740 else if (*s == PL_multi_close && --brackets <= 0)
11742 else if (*s == PL_multi_open)
11744 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11749 /* terminate the copied string and update the sv's end-of-string */
11751 SvCUR_set(sv, to - SvPVX_const(sv));
11754 * this next chunk reads more into the buffer if we're not done yet
11758 break; /* handle case where we are done yet :-) */
11760 #ifndef PERL_STRICT_CR
11761 if (to - SvPVX_const(sv) >= 2) {
11762 if ((to[-2] == '\r' && to[-1] == '\n') ||
11763 (to[-2] == '\n' && to[-1] == '\r'))
11767 SvCUR_set(sv, to - SvPVX_const(sv));
11769 else if (to[-1] == '\r')
11772 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11777 /* if we're out of file, or a read fails, bail and reset the current
11778 line marker so we can report where the unterminated string began
11781 if (PL_madskills) {
11782 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11784 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11786 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11790 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11792 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11798 /* we read a line, so increment our line counter */
11799 CopLINE_inc(PL_curcop);
11801 /* update debugger info */
11802 if (PERLDB_LINE && PL_curstash != PL_debstash)
11803 update_debugger_info(PL_linestr, NULL, 0);
11805 /* having changed the buffer, we must update PL_bufend */
11806 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11807 PL_last_lop = PL_last_uni = NULL;
11810 /* at this point, we have successfully read the delimited string */
11812 if (!PL_encoding || UTF) {
11814 if (PL_madskills) {
11815 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11816 const int len = s - tstart;
11818 sv_catpvn(PL_thisstuff, tstart, len);
11820 PL_thisstuff = newSVpvn(tstart, len);
11821 if (!PL_thisclose && !keep_delims)
11822 PL_thisclose = newSVpvn(s,termlen);
11827 sv_catpvn(sv, s, termlen);
11832 if (PL_madskills) {
11833 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11834 const int len = s - tstart - termlen;
11836 sv_catpvn(PL_thisstuff, tstart, len);
11838 PL_thisstuff = newSVpvn(tstart, len);
11839 if (!PL_thisclose && !keep_delims)
11840 PL_thisclose = newSVpvn(s - termlen,termlen);
11844 if (has_utf8 || PL_encoding)
11847 PL_multi_end = CopLINE(PL_curcop);
11849 /* if we allocated too much space, give some back */
11850 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11851 SvLEN_set(sv, SvCUR(sv) + 1);
11852 SvPV_renew(sv, SvLEN(sv));
11855 /* decide whether this is the first or second quoted string we've read
11868 takes: pointer to position in buffer
11869 returns: pointer to new position in buffer
11870 side-effects: builds ops for the constant in pl_yylval.op
11872 Read a number in any of the formats that Perl accepts:
11874 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
11875 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
11878 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
11880 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
11883 If it reads a number without a decimal point or an exponent, it will
11884 try converting the number to an integer and see if it can do so
11885 without loss of precision.
11889 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
11892 register const char *s = start; /* current position in buffer */
11893 register char *d; /* destination in temp buffer */
11894 register char *e; /* end of temp buffer */
11895 NV nv; /* number read, as a double */
11896 SV *sv = NULL; /* place to put the converted number */
11897 bool floatit; /* boolean: int or float? */
11898 const char *lastub = NULL; /* position of last underbar */
11899 static char const number_too_long[] = "Number too long";
11901 /* We use the first character to decide what type of number this is */
11905 Perl_croak(aTHX_ "panic: scan_num");
11907 /* if it starts with a 0, it could be an octal number, a decimal in
11908 0.13 disguise, or a hexadecimal number, or a binary number. */
11912 u holds the "number so far"
11913 shift the power of 2 of the base
11914 (hex == 4, octal == 3, binary == 1)
11915 overflowed was the number more than we can hold?
11917 Shift is used when we add a digit. It also serves as an "are
11918 we in octal/hex/binary?" indicator to disallow hex characters
11919 when in octal mode.
11924 bool overflowed = FALSE;
11925 bool just_zero = TRUE; /* just plain 0 or binary number? */
11926 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
11927 static const char* const bases[5] =
11928 { "", "binary", "", "octal", "hexadecimal" };
11929 static const char* const Bases[5] =
11930 { "", "Binary", "", "Octal", "Hexadecimal" };
11931 static const char* const maxima[5] =
11933 "0b11111111111111111111111111111111",
11937 const char *base, *Base, *max;
11939 /* check for hex */
11944 } else if (s[1] == 'b') {
11949 /* check for a decimal in disguise */
11950 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
11952 /* so it must be octal */
11959 if (ckWARN(WARN_SYNTAX))
11960 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11961 "Misplaced _ in number");
11965 base = bases[shift];
11966 Base = Bases[shift];
11967 max = maxima[shift];
11969 /* read the rest of the number */
11971 /* x is used in the overflow test,
11972 b is the digit we're adding on. */
11977 /* if we don't mention it, we're done */
11981 /* _ are ignored -- but warned about if consecutive */
11983 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
11984 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11985 "Misplaced _ in number");
11989 /* 8 and 9 are not octal */
11990 case '8': case '9':
11992 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
11996 case '2': case '3': case '4':
11997 case '5': case '6': case '7':
11999 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
12002 case '0': case '1':
12003 b = *s++ & 15; /* ASCII digit -> value of digit */
12007 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12008 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12009 /* make sure they said 0x */
12012 b = (*s++ & 7) + 9;
12014 /* Prepare to put the digit we have onto the end
12015 of the number so far. We check for overflows.
12021 x = u << shift; /* make room for the digit */
12023 if ((x >> shift) != u
12024 && !(PL_hints & HINT_NEW_BINARY)) {
12027 if (ckWARN_d(WARN_OVERFLOW))
12028 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12029 "Integer overflow in %s number",
12032 u = x | b; /* add the digit to the end */
12035 n *= nvshift[shift];
12036 /* If an NV has not enough bits in its
12037 * mantissa to represent an UV this summing of
12038 * small low-order numbers is a waste of time
12039 * (because the NV cannot preserve the
12040 * low-order bits anyway): we could just
12041 * remember when did we overflow and in the
12042 * end just multiply n by the right
12050 /* if we get here, we had success: make a scalar value from
12055 /* final misplaced underbar check */
12056 if (s[-1] == '_') {
12057 if (ckWARN(WARN_SYNTAX))
12058 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12063 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12064 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12065 "%s number > %s non-portable",
12071 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12072 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12073 "%s number > %s non-portable",
12078 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12079 sv = new_constant(start, s - start, "integer",
12080 sv, NULL, NULL, 0);
12081 else if (PL_hints & HINT_NEW_BINARY)
12082 sv = new_constant(start, s - start, "binary", sv, NULL, NULL, 0);
12087 handle decimal numbers.
12088 we're also sent here when we read a 0 as the first digit
12090 case '1': case '2': case '3': case '4': case '5':
12091 case '6': case '7': case '8': case '9': case '.':
12094 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12097 /* read next group of digits and _ and copy into d */
12098 while (isDIGIT(*s) || *s == '_') {
12099 /* skip underscores, checking for misplaced ones
12103 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12104 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12105 "Misplaced _ in number");
12109 /* check for end of fixed-length buffer */
12111 Perl_croak(aTHX_ number_too_long);
12112 /* if we're ok, copy the character */
12117 /* final misplaced underbar check */
12118 if (lastub && s == lastub + 1) {
12119 if (ckWARN(WARN_SYNTAX))
12120 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12123 /* read a decimal portion if there is one. avoid
12124 3..5 being interpreted as the number 3. followed
12127 if (*s == '.' && s[1] != '.') {
12132 if (ckWARN(WARN_SYNTAX))
12133 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12134 "Misplaced _ in number");
12138 /* copy, ignoring underbars, until we run out of digits.
12140 for (; isDIGIT(*s) || *s == '_'; s++) {
12141 /* fixed length buffer check */
12143 Perl_croak(aTHX_ number_too_long);
12145 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12146 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12147 "Misplaced _ in number");
12153 /* fractional part ending in underbar? */
12154 if (s[-1] == '_') {
12155 if (ckWARN(WARN_SYNTAX))
12156 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12157 "Misplaced _ in number");
12159 if (*s == '.' && isDIGIT(s[1])) {
12160 /* oops, it's really a v-string, but without the "v" */
12166 /* read exponent part, if present */
12167 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12171 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12172 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12174 /* stray preinitial _ */
12176 if (ckWARN(WARN_SYNTAX))
12177 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12178 "Misplaced _ in number");
12182 /* allow positive or negative exponent */
12183 if (*s == '+' || *s == '-')
12186 /* stray initial _ */
12188 if (ckWARN(WARN_SYNTAX))
12189 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12190 "Misplaced _ in number");
12194 /* read digits of exponent */
12195 while (isDIGIT(*s) || *s == '_') {
12198 Perl_croak(aTHX_ number_too_long);
12202 if (((lastub && s == lastub + 1) ||
12203 (!isDIGIT(s[1]) && s[1] != '_'))
12204 && ckWARN(WARN_SYNTAX))
12205 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12206 "Misplaced _ in number");
12213 /* make an sv from the string */
12217 We try to do an integer conversion first if no characters
12218 indicating "float" have been found.
12223 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12225 if (flags == IS_NUMBER_IN_UV) {
12227 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12230 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12231 if (uv <= (UV) IV_MIN)
12232 sv_setiv(sv, -(IV)uv);
12239 /* terminate the string */
12241 nv = Atof(PL_tokenbuf);
12246 ? (PL_hints & HINT_NEW_FLOAT) : (PL_hints & HINT_NEW_INTEGER) ) {
12247 const char *const key = floatit ? "float" : "integer";
12248 const STRLEN keylen = floatit ? 5 : 7;
12249 sv = S_new_constant(aTHX_ PL_tokenbuf, d - PL_tokenbuf,
12250 key, keylen, sv, NULL, NULL, 0);
12254 /* if it starts with a v, it could be a v-string */
12257 sv = newSV(5); /* preallocate storage space */
12258 s = scan_vstring(s, PL_bufend, sv);
12262 /* make the op for the constant and return */
12265 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12267 lvalp->opval = NULL;
12273 S_scan_formline(pTHX_ register char *s)
12276 register char *eol;
12278 SV * const stuff = newSVpvs("");
12279 bool needargs = FALSE;
12280 bool eofmt = FALSE;
12282 char *tokenstart = s;
12285 if (PL_madskills) {
12286 savewhite = PL_thiswhite;
12291 while (!needargs) {
12294 #ifdef PERL_STRICT_CR
12295 while (SPACE_OR_TAB(*t))
12298 while (SPACE_OR_TAB(*t) || *t == '\r')
12301 if (*t == '\n' || t == PL_bufend) {
12306 if (PL_in_eval && !PL_rsfp) {
12307 eol = (char *) memchr(s,'\n',PL_bufend-s);
12312 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12314 for (t = s; t < eol; t++) {
12315 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12317 goto enough; /* ~~ must be first line in formline */
12319 if (*t == '@' || *t == '^')
12323 sv_catpvn(stuff, s, eol-s);
12324 #ifndef PERL_STRICT_CR
12325 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12326 char *end = SvPVX(stuff) + SvCUR(stuff);
12329 SvCUR_set(stuff, SvCUR(stuff) - 1);
12339 if (PL_madskills) {
12341 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12343 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12346 s = filter_gets(PL_linestr, PL_rsfp, 0);
12348 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12350 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12352 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12353 PL_last_lop = PL_last_uni = NULL;
12362 if (SvCUR(stuff)) {
12365 PL_lex_state = LEX_NORMAL;
12366 start_force(PL_curforce);
12367 NEXTVAL_NEXTTOKE.ival = 0;
12371 PL_lex_state = LEX_FORMLINE;
12373 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12375 else if (PL_encoding)
12376 sv_recode_to_utf8(stuff, PL_encoding);
12378 start_force(PL_curforce);
12379 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12381 start_force(PL_curforce);
12382 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12386 SvREFCNT_dec(stuff);
12388 PL_lex_formbrack = 0;
12392 if (PL_madskills) {
12394 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12396 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12397 PL_thiswhite = savewhite;
12404 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12407 const I32 oldsavestack_ix = PL_savestack_ix;
12408 CV* const outsidecv = PL_compcv;
12411 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12413 SAVEI32(PL_subline);
12414 save_item(PL_subname);
12415 SAVESPTR(PL_compcv);
12417 PL_compcv = (CV*)newSV_type(is_format ? SVt_PVFM : SVt_PVCV);
12418 CvFLAGS(PL_compcv) |= flags;
12420 PL_subline = CopLINE(PL_curcop);
12421 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12422 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc_simple(outsidecv);
12423 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12425 return oldsavestack_ix;
12429 #pragma segment Perl_yylex
12432 Perl_yywarn(pTHX_ const char *const s)
12435 PL_in_eval |= EVAL_WARNONLY;
12437 PL_in_eval &= ~EVAL_WARNONLY;
12442 Perl_yyerror(pTHX_ const char *const s)
12445 const char *where = NULL;
12446 const char *context = NULL;
12449 int yychar = PL_parser->yychar;
12451 if (!yychar || (yychar == ';' && !PL_rsfp))
12453 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12454 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12455 PL_oldbufptr != PL_bufptr) {
12458 The code below is removed for NetWare because it abends/crashes on NetWare
12459 when the script has error such as not having the closing quotes like:
12460 if ($var eq "value)
12461 Checking of white spaces is anyway done in NetWare code.
12464 while (isSPACE(*PL_oldoldbufptr))
12467 context = PL_oldoldbufptr;
12468 contlen = PL_bufptr - PL_oldoldbufptr;
12470 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12471 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12474 The code below is removed for NetWare because it abends/crashes on NetWare
12475 when the script has error such as not having the closing quotes like:
12476 if ($var eq "value)
12477 Checking of white spaces is anyway done in NetWare code.
12480 while (isSPACE(*PL_oldbufptr))
12483 context = PL_oldbufptr;
12484 contlen = PL_bufptr - PL_oldbufptr;
12486 else if (yychar > 255)
12487 where = "next token ???";
12488 else if (yychar == -2) { /* YYEMPTY */
12489 if (PL_lex_state == LEX_NORMAL ||
12490 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12491 where = "at end of line";
12492 else if (PL_lex_inpat)
12493 where = "within pattern";
12495 where = "within string";
12498 SV * const where_sv = newSVpvs_flags("next char ", SVs_TEMP);
12500 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12501 else if (isPRINT_LC(yychar)) {
12502 const char string = yychar;
12503 sv_catpvn(where_sv, &string, 1);
12506 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12507 where = SvPVX_const(where_sv);
12509 msg = sv_2mortal(newSVpv(s, 0));
12510 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12511 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12513 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12515 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12516 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12517 Perl_sv_catpvf(aTHX_ msg,
12518 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12519 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12522 if (PL_in_eval & EVAL_WARNONLY) {
12523 if (ckWARN_d(WARN_SYNTAX))
12524 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12528 if (PL_error_count >= 10) {
12529 if (PL_in_eval && SvCUR(ERRSV))
12530 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12531 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12533 Perl_croak(aTHX_ "%s has too many errors.\n",
12534 OutCopFILE(PL_curcop));
12537 PL_in_my_stash = NULL;
12541 #pragma segment Main
12545 S_swallow_bom(pTHX_ U8 *s)
12548 const STRLEN slen = SvCUR(PL_linestr);
12551 if (s[1] == 0xFE) {
12552 /* UTF-16 little-endian? (or UTF32-LE?) */
12553 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12554 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12555 #ifndef PERL_NO_UTF16_FILTER
12556 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12559 if (PL_bufend > (char*)s) {
12563 filter_add(utf16rev_textfilter, NULL);
12564 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12565 utf16_to_utf8_reversed(s, news,
12566 PL_bufend - (char*)s - 1,
12568 sv_setpvn(PL_linestr, (const char*)news, newlen);
12570 s = (U8*)SvPVX(PL_linestr);
12571 Copy(news, s, newlen, U8);
12575 SvUTF8_on(PL_linestr);
12576 s = (U8*)SvPVX(PL_linestr);
12578 /* FIXME - is this a general bug fix? */
12581 PL_bufend = SvPVX(PL_linestr) + newlen;
12584 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12589 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12590 #ifndef PERL_NO_UTF16_FILTER
12591 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12594 if (PL_bufend > (char *)s) {
12598 filter_add(utf16_textfilter, NULL);
12599 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12600 utf16_to_utf8(s, news,
12601 PL_bufend - (char*)s,
12603 sv_setpvn(PL_linestr, (const char*)news, newlen);
12605 SvUTF8_on(PL_linestr);
12606 s = (U8*)SvPVX(PL_linestr);
12607 PL_bufend = SvPVX(PL_linestr) + newlen;
12610 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12615 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12616 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12617 s += 3; /* UTF-8 */
12623 if (s[2] == 0xFE && s[3] == 0xFF) {
12624 /* UTF-32 big-endian */
12625 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12628 else if (s[2] == 0 && s[3] != 0) {
12631 * are a good indicator of UTF-16BE. */
12632 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12638 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12639 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12640 s += 4; /* UTF-8 */
12646 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12649 * are a good indicator of UTF-16LE. */
12650 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12658 #ifndef PERL_NO_UTF16_FILTER
12660 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12663 const STRLEN old = SvCUR(sv);
12664 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12665 DEBUG_P(PerlIO_printf(Perl_debug_log,
12666 "utf16_textfilter(%p): %d %d (%d)\n",
12667 FPTR2DPTR(void *, utf16_textfilter),
12668 idx, maxlen, (int) count));
12672 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12673 Copy(SvPVX_const(sv), tmps, old, char);
12674 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12675 SvCUR(sv) - old, &newlen);
12676 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12678 DEBUG_P({sv_dump(sv);});
12683 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12686 const STRLEN old = SvCUR(sv);
12687 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12688 DEBUG_P(PerlIO_printf(Perl_debug_log,
12689 "utf16rev_textfilter(%p): %d %d (%d)\n",
12690 FPTR2DPTR(void *, utf16rev_textfilter),
12691 idx, maxlen, (int) count));
12695 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12696 Copy(SvPVX_const(sv), tmps, old, char);
12697 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12698 SvCUR(sv) - old, &newlen);
12699 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12701 DEBUG_P({ sv_dump(sv); });
12707 Returns a pointer to the next character after the parsed
12708 vstring, as well as updating the passed in sv.
12710 Function must be called like
12713 s = scan_vstring(s,e,sv);
12715 where s and e are the start and end of the string.
12716 The sv should already be large enough to store the vstring
12717 passed in, for performance reasons.
12722 Perl_scan_vstring(pTHX_ const char *s, const char *const e, SV *sv)
12725 const char *pos = s;
12726 const char *start = s;
12727 if (*pos == 'v') pos++; /* get past 'v' */
12728 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12730 if ( *pos != '.') {
12731 /* this may not be a v-string if followed by => */
12732 const char *next = pos;
12733 while (next < e && isSPACE(*next))
12735 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12736 /* return string not v-string */
12737 sv_setpvn(sv,(char *)s,pos-s);
12738 return (char *)pos;
12742 if (!isALPHA(*pos)) {
12743 U8 tmpbuf[UTF8_MAXBYTES+1];
12746 s++; /* get past 'v' */
12748 sv_setpvn(sv, "", 0);
12751 /* this is atoi() that tolerates underscores */
12754 const char *end = pos;
12756 while (--end >= s) {
12758 const UV orev = rev;
12759 rev += (*end - '0') * mult;
12761 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12762 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12763 "Integer overflow in decimal number");
12767 if (rev > 0x7FFFFFFF)
12768 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12770 /* Append native character for the rev point */
12771 tmpend = uvchr_to_utf8(tmpbuf, rev);
12772 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12773 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12775 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12781 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12785 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12793 * c-indentation-style: bsd
12794 * c-basic-offset: 4
12795 * indent-tabs-mode: t
12798 * ex: set ts=8 sts=4 sw=4 noet: