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 yylval (PL_parser->yylval)
28 /* YYINITDEPTH -- initial size of the parser's stacks. */
29 #define YYINITDEPTH 200
31 /* XXX temporary backwards compatibility */
32 #define PL_lex_brackets (PL_parser->lex_brackets)
33 #define PL_lex_brackstack (PL_parser->lex_brackstack)
34 #define PL_lex_casemods (PL_parser->lex_casemods)
35 #define PL_lex_casestack (PL_parser->lex_casestack)
36 #define PL_lex_defer (PL_parser->lex_defer)
37 #define PL_lex_dojoin (PL_parser->lex_dojoin)
38 #define PL_lex_expect (PL_parser->lex_expect)
39 #define PL_lex_formbrack (PL_parser->lex_formbrack)
40 #define PL_lex_inpat (PL_parser->lex_inpat)
41 #define PL_lex_inwhat (PL_parser->lex_inwhat)
42 #define PL_lex_op (PL_parser->lex_op)
43 #define PL_lex_repl (PL_parser->lex_repl)
44 #define PL_lex_starts (PL_parser->lex_starts)
45 #define PL_lex_stuff (PL_parser->lex_stuff)
46 #define PL_multi_start (PL_parser->multi_start)
47 #define PL_multi_open (PL_parser->multi_open)
48 #define PL_multi_close (PL_parser->multi_close)
49 #define PL_pending_ident (PL_parser->pending_ident)
50 #define PL_preambled (PL_parser->preambled)
51 #define PL_sublex_info (PL_parser->sublex_info)
52 #define PL_linestr (PL_parser->linestr)
53 #define PL_expect (PL_parser->expect)
54 #define PL_copline (PL_parser->copline)
55 #define PL_bufptr (PL_parser->bufptr)
56 #define PL_oldbufptr (PL_parser->oldbufptr)
57 #define PL_oldoldbufptr (PL_parser->oldoldbufptr)
58 #define PL_linestart (PL_parser->linestart)
59 #define PL_bufend (PL_parser->bufend)
60 #define PL_last_uni (PL_parser->last_uni)
61 #define PL_last_lop (PL_parser->last_lop)
62 #define PL_last_lop_op (PL_parser->last_lop_op)
63 #define PL_lex_state (PL_parser->lex_state)
64 #define PL_rsfp (PL_parser->rsfp)
65 #define PL_rsfp_filters (PL_parser->rsfp_filters)
66 #define PL_in_my (PL_parser->in_my)
67 #define PL_in_my_stash (PL_parser->in_my_stash)
68 #define PL_tokenbuf (PL_parser->tokenbuf)
69 #define PL_multi_end (PL_parser->multi_end)
70 #define PL_error_count (PL_parser->error_count)
73 # define PL_endwhite (PL_parser->endwhite)
74 # define PL_faketokens (PL_parser->faketokens)
75 # define PL_lasttoke (PL_parser->lasttoke)
76 # define PL_nextwhite (PL_parser->nextwhite)
77 # define PL_realtokenstart (PL_parser->realtokenstart)
78 # define PL_skipwhite (PL_parser->skipwhite)
79 # define PL_thisclose (PL_parser->thisclose)
80 # define PL_thismad (PL_parser->thismad)
81 # define PL_thisopen (PL_parser->thisopen)
82 # define PL_thisstuff (PL_parser->thisstuff)
83 # define PL_thistoken (PL_parser->thistoken)
84 # define PL_thiswhite (PL_parser->thiswhite)
85 # define PL_thiswhite (PL_parser->thiswhite)
86 # define PL_nexttoke (PL_parser->nexttoke)
87 # define PL_curforce (PL_parser->curforce)
89 # define PL_nexttoke (PL_parser->nexttoke)
90 # define PL_nexttype (PL_parser->nexttype)
91 # define PL_nextval (PL_parser->nextval)
95 S_pending_ident(pTHX);
97 static const char ident_too_long[] = "Identifier too long";
98 static const char commaless_variable_list[] = "comma-less variable list";
100 #ifndef PERL_NO_UTF16_FILTER
101 static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen);
102 static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen);
106 # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
107 # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
109 # define CURMAD(slot,sv)
110 # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
113 #define XFAKEBRACK 128
114 #define XENUMMASK 127
116 #ifdef USE_UTF8_SCRIPTS
117 # define UTF (!IN_BYTES)
119 # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
122 /* In variables named $^X, these are the legal values for X.
123 * 1999-02-27 mjd-perl-patch@plover.com */
124 #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
126 /* On MacOS, respect nonbreaking spaces */
127 #ifdef MACOS_TRADITIONAL
128 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t')
130 #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
133 /* LEX_* are values for PL_lex_state, the state of the lexer.
134 * They are arranged oddly so that the guard on the switch statement
135 * can get by with a single comparison (if the compiler is smart enough).
138 /* #define LEX_NOTPARSING 11 is done in perl.h. */
140 #define LEX_NORMAL 10 /* normal code (ie not within "...") */
141 #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
142 #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
143 #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
144 #define LEX_INTERPSTART 6 /* expecting the start of a $var */
146 /* at end of code, eg "$x" followed by: */
147 #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
148 #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
150 #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
151 string or after \E, $foo, etc */
152 #define LEX_INTERPCONST 2 /* NOT USED */
153 #define LEX_FORMLINE 1 /* expecting a format line */
154 #define LEX_KNOWNEXT 0 /* next token known; just return it */
158 static const char* const lex_state_names[] = {
177 #include "keywords.h"
179 /* CLINE is a macro that ensures PL_copline has a sane value */
184 #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
187 # define SKIPSPACE0(s) skipspace0(s)
188 # define SKIPSPACE1(s) skipspace1(s)
189 # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
190 # define PEEKSPACE(s) skipspace2(s,0)
192 # define SKIPSPACE0(s) skipspace(s)
193 # define SKIPSPACE1(s) skipspace(s)
194 # define SKIPSPACE2(s,tsv) skipspace(s)
195 # define PEEKSPACE(s) skipspace(s)
199 * Convenience functions to return different tokens and prime the
200 * lexer for the next token. They all take an argument.
202 * TOKEN : generic token (used for '(', DOLSHARP, etc)
203 * OPERATOR : generic operator
204 * AOPERATOR : assignment operator
205 * PREBLOCK : beginning the block after an if, while, foreach, ...
206 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
207 * PREREF : *EXPR where EXPR is not a simple identifier
208 * TERM : expression term
209 * LOOPX : loop exiting command (goto, last, dump, etc)
210 * FTST : file test operator
211 * FUN0 : zero-argument function
212 * FUN1 : not used, except for not, which isn't a UNIOP
213 * BOop : bitwise or or xor
215 * SHop : shift operator
216 * PWop : power operator
217 * PMop : pattern-matching operator
218 * Aop : addition-level operator
219 * Mop : multiplication-level operator
220 * Eop : equality-testing operator
221 * Rop : relational operator <= != gt
223 * Also see LOP and lop() below.
226 #ifdef DEBUGGING /* Serve -DT. */
227 # define REPORT(retval) tokereport((I32)retval)
229 # define REPORT(retval) (retval)
232 #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
233 #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
234 #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
235 #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
236 #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
237 #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
238 #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
239 #define LOOPX(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
240 #define FTST(f) return (yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
241 #define FUN0(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
242 #define FUN1(f) return (yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
243 #define BOop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
244 #define BAop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
245 #define SHop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
246 #define PWop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
247 #define PMop(f) return(yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
248 #define Aop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
249 #define Mop(f) return ao((yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
250 #define Eop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
251 #define Rop(f) return (yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
253 /* This bit of chicanery makes a unary function followed by
254 * a parenthesis into a function with one argument, highest precedence.
255 * The UNIDOR macro is for unary functions that can be followed by the //
256 * operator (such as C<shift // 0>).
258 #define UNI2(f,x) { \
262 PL_last_uni = PL_oldbufptr; \
263 PL_last_lop_op = f; \
265 return REPORT( (int)FUNC1 ); \
267 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
269 #define UNI(f) UNI2(f,XTERM)
270 #define UNIDOR(f) UNI2(f,XTERMORDORDOR)
272 #define UNIBRACK(f) { \
275 PL_last_uni = PL_oldbufptr; \
277 return REPORT( (int)FUNC1 ); \
279 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
282 /* grandfather return to old style */
283 #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP)
287 /* how to interpret the yylval associated with the token */
291 TOKENTYPE_OPNUM, /* yylval.ival contains an opcode number */
297 static struct debug_tokens {
299 enum token_type type;
301 } const debug_tokens[] =
303 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
304 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
305 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
306 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
307 { ARROW, TOKENTYPE_NONE, "ARROW" },
308 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
309 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
310 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
311 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
312 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
313 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
314 { DO, TOKENTYPE_NONE, "DO" },
315 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
316 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
317 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
318 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
319 { ELSE, TOKENTYPE_NONE, "ELSE" },
320 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
321 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
322 { FOR, TOKENTYPE_IVAL, "FOR" },
323 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
324 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
325 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
326 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
327 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
328 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
329 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
330 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
331 { IF, TOKENTYPE_IVAL, "IF" },
332 { LABEL, TOKENTYPE_PVAL, "LABEL" },
333 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
334 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
335 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
336 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
337 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
338 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
339 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
340 { MY, TOKENTYPE_IVAL, "MY" },
341 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
342 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
343 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
344 { OROP, TOKENTYPE_IVAL, "OROP" },
345 { OROR, TOKENTYPE_NONE, "OROR" },
346 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
347 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
348 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
349 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
350 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
351 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
352 { PREINC, TOKENTYPE_NONE, "PREINC" },
353 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
354 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
355 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
356 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
357 { SUB, TOKENTYPE_NONE, "SUB" },
358 { THING, TOKENTYPE_OPVAL, "THING" },
359 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
360 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
361 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
362 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
363 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
364 { USE, TOKENTYPE_IVAL, "USE" },
365 { WHEN, TOKENTYPE_IVAL, "WHEN" },
366 { WHILE, TOKENTYPE_IVAL, "WHILE" },
367 { WORD, TOKENTYPE_OPVAL, "WORD" },
368 { 0, TOKENTYPE_NONE, NULL }
371 /* dump the returned token in rv, plus any optional arg in yylval */
374 S_tokereport(pTHX_ I32 rv)
378 const char *name = NULL;
379 enum token_type type = TOKENTYPE_NONE;
380 const struct debug_tokens *p;
381 SV* const report = newSVpvs("<== ");
383 for (p = debug_tokens; p->token; p++) {
384 if (p->token == (int)rv) {
391 Perl_sv_catpv(aTHX_ report, name);
392 else if ((char)rv > ' ' && (char)rv < '~')
393 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
395 sv_catpvs(report, "EOF");
397 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
400 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
403 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)yylval.ival);
405 case TOKENTYPE_OPNUM:
406 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
407 PL_op_name[yylval.ival]);
410 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", yylval.pval);
412 case TOKENTYPE_OPVAL:
414 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
415 PL_op_name[yylval.opval->op_type]);
416 if (yylval.opval->op_type == OP_CONST) {
417 Perl_sv_catpvf(aTHX_ report, " %s",
418 SvPEEK(cSVOPx_sv(yylval.opval)));
423 sv_catpvs(report, "(opval=null)");
426 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
432 /* print the buffer with suitable escapes */
435 S_printbuf(pTHX_ const char* fmt, const char* s)
437 SV* const tmp = newSVpvs("");
438 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
447 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
448 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
452 S_ao(pTHX_ int toketype)
455 if (*PL_bufptr == '=') {
457 if (toketype == ANDAND)
458 yylval.ival = OP_ANDASSIGN;
459 else if (toketype == OROR)
460 yylval.ival = OP_ORASSIGN;
461 else if (toketype == DORDOR)
462 yylval.ival = OP_DORASSIGN;
470 * When Perl expects an operator and finds something else, no_op
471 * prints the warning. It always prints "<something> found where
472 * operator expected. It prints "Missing semicolon on previous line?"
473 * if the surprise occurs at the start of the line. "do you need to
474 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
475 * where the compiler doesn't know if foo is a method call or a function.
476 * It prints "Missing operator before end of line" if there's nothing
477 * after the missing operator, or "... before <...>" if there is something
478 * after the missing operator.
482 S_no_op(pTHX_ const char *what, char *s)
485 char * const oldbp = PL_bufptr;
486 const bool is_first = (PL_oldbufptr == PL_linestart);
492 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
493 if (ckWARN_d(WARN_SYNTAX)) {
495 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
496 "\t(Missing semicolon on previous line?)\n");
497 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
499 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
501 if (t < PL_bufptr && isSPACE(*t))
502 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
503 "\t(Do you need to predeclare %.*s?)\n",
504 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
508 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
509 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
517 * Complain about missing quote/regexp/heredoc terminator.
518 * If it's called with NULL then it cauterizes the line buffer.
519 * If we're in a delimited string and the delimiter is a control
520 * character, it's reformatted into a two-char sequence like ^C.
525 S_missingterm(pTHX_ char *s)
531 char * const nl = strrchr(s,'\n');
537 iscntrl(PL_multi_close)
539 PL_multi_close < 32 || PL_multi_close == 127
543 tmpbuf[1] = (char)toCTRL(PL_multi_close);
548 *tmpbuf = (char)PL_multi_close;
552 q = strchr(s,'"') ? '\'' : '"';
553 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
556 #define FEATURE_IS_ENABLED(name) \
557 ((0 != (PL_hints & HINT_LOCALIZE_HH)) \
558 && S_feature_is_enabled(aTHX_ STR_WITH_LEN(name)))
560 * S_feature_is_enabled
561 * Check whether the named feature is enabled.
564 S_feature_is_enabled(pTHX_ const char *name, STRLEN namelen)
567 HV * const hinthv = GvHV(PL_hintgv);
568 char he_name[32] = "feature_";
569 (void) my_strlcpy(&he_name[8], name, 24);
571 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
579 Perl_deprecate(pTHX_ const char *s)
581 if (ckWARN(WARN_DEPRECATED))
582 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), "Use of %s is deprecated", s);
586 Perl_deprecate_old(pTHX_ const char *s)
588 /* This function should NOT be called for any new deprecated warnings */
589 /* Use Perl_deprecate instead */
591 /* It is here to maintain backward compatibility with the pre-5.8 */
592 /* warnings category hierarchy. The "deprecated" category used to */
593 /* live under the "syntax" category. It is now a top-level category */
594 /* in its own right. */
596 if (ckWARN2(WARN_DEPRECATED, WARN_SYNTAX))
597 Perl_warner(aTHX_ packWARN2(WARN_DEPRECATED, WARN_SYNTAX),
598 "Use of %s is deprecated", s);
602 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
603 * utf16-to-utf8-reversed.
606 #ifdef PERL_CR_FILTER
610 register const char *s = SvPVX_const(sv);
611 register const char * const e = s + SvCUR(sv);
612 /* outer loop optimized to do nothing if there are no CR-LFs */
614 if (*s++ == '\r' && *s == '\n') {
615 /* hit a CR-LF, need to copy the rest */
616 register char *d = s - 1;
619 if (*s == '\r' && s[1] == '\n')
630 S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
632 const I32 count = FILTER_READ(idx+1, sv, maxlen);
633 if (count > 0 && !maxlen)
644 * Create a parser object and initialise its parser and lexer fields
646 * rsfp is the opened file handle to read from (if any),
648 * line holds any initial content already read from the file (or in
649 * the case of no file, such as an eval, the whole contents);
651 * new_filter indicates that this is a new file and it shouldn't inherit
652 * the filters from the current parser (ie require).
656 Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, bool new_filter)
659 const char *s = NULL;
661 yy_parser *parser, *oparser;
663 /* create and initialise a parser */
665 Newxz(parser, 1, yy_parser);
666 parser->old_parser = oparser = PL_parser;
669 Newx(parser->stack, YYINITDEPTH, yy_stack_frame);
670 parser->ps = parser->stack;
671 parser->stack_size = YYINITDEPTH;
673 parser->stack->state = 0;
674 parser->yyerrstatus = 0;
675 parser->yychar = YYEMPTY; /* Cause a token to be read. */
677 /* on scope exit, free this parser and restore any outer one */
679 parser->saved_curcop = PL_curcop;
681 /* initialise lexer state */
684 parser->curforce = -1;
686 parser->nexttoke = 0;
688 parser->copline = NOLINE;
689 parser->lex_state = LEX_NORMAL;
690 parser->expect = XSTATE;
692 parser->rsfp_filters = (new_filter || !oparser) ? newAV()
693 : (AV*)SvREFCNT_inc(oparser->rsfp_filters);
695 Newx(parser->lex_brackstack, 120, char);
696 Newx(parser->lex_casestack, 12, char);
697 *parser->lex_casestack = '\0';
700 s = SvPV_const(line, len);
706 parser->linestr = newSVpvs("\n;");
707 } else if (SvREADONLY(line) || s[len-1] != ';') {
708 parser->linestr = newSVsv(line);
710 sv_catpvs(parser->linestr, "\n;");
713 SvREFCNT_inc_simple_void_NN(line);
714 parser->linestr = line;
716 parser->oldoldbufptr =
719 parser->linestart = SvPVX(parser->linestr);
720 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
721 parser->last_lop = parser->last_uni = NULL;
725 /* delete a parser object */
728 Perl_parser_free(pTHX_ const yy_parser *parser)
730 PL_curcop = parser->saved_curcop;
731 SvREFCNT_dec(parser->linestr);
733 if (parser->rsfp == PerlIO_stdin())
734 PerlIO_clearerr(parser->rsfp);
735 else if (parser->rsfp && parser->old_parser
736 && parser->rsfp != parser->old_parser->rsfp)
737 PerlIO_close(parser->rsfp);
738 SvREFCNT_dec(parser->rsfp_filters);
740 Safefree(parser->stack);
741 Safefree(parser->lex_brackstack);
742 Safefree(parser->lex_casestack);
743 PL_parser = parser->old_parser;
750 * Finalizer for lexing operations. Must be called when the parser is
751 * done with the lexer.
758 PL_doextract = FALSE;
763 * This subroutine has nothing to do with tilting, whether at windmills
764 * or pinball tables. Its name is short for "increment line". It
765 * increments the current line number in CopLINE(PL_curcop) and checks
766 * to see whether the line starts with a comment of the form
767 * # line 500 "foo.pm"
768 * If so, it sets the current line number and file to the values in the comment.
772 S_incline(pTHX_ const char *s)
779 CopLINE_inc(PL_curcop);
782 while (SPACE_OR_TAB(*s))
784 if (strnEQ(s, "line", 4))
788 if (SPACE_OR_TAB(*s))
792 while (SPACE_OR_TAB(*s))
800 while (SPACE_OR_TAB(*s))
802 if (*s == '"' && (t = strchr(s+1, '"'))) {
812 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
814 if (*e != '\n' && *e != '\0')
815 return; /* false alarm */
818 const STRLEN len = t - s;
820 const char * const cf = CopFILE(PL_curcop);
821 STRLEN tmplen = cf ? strlen(cf) : 0;
822 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
823 /* must copy *{"::_<(eval N)[oldfilename:L]"}
824 * to *{"::_<newfilename"} */
825 /* However, the long form of evals is only turned on by the
826 debugger - usually they're "(eval %lu)" */
830 STRLEN tmplen2 = len;
831 if (tmplen + 2 <= sizeof smallbuf)
834 Newx(tmpbuf, tmplen + 2, char);
837 memcpy(tmpbuf + 2, cf, tmplen);
839 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
844 if (tmplen2 + 2 <= sizeof smallbuf)
847 Newx(tmpbuf2, tmplen2 + 2, char);
849 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
850 /* Either they malloc'd it, or we malloc'd it,
851 so no prefix is present in ours. */
856 memcpy(tmpbuf2 + 2, s, tmplen2);
859 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
861 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
862 /* adjust ${"::_<newfilename"} to store the new file name */
863 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
864 GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp));
865 GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp));
868 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
870 if (tmpbuf != smallbuf) Safefree(tmpbuf);
873 CopFILE_free(PL_curcop);
874 CopFILE_setn(PL_curcop, s, len);
876 CopLINE_set(PL_curcop, atoi(n)-1);
880 /* skip space before PL_thistoken */
883 S_skipspace0(pTHX_ register char *s)
890 PL_thiswhite = newSVpvs("");
891 sv_catsv(PL_thiswhite, PL_skipwhite);
892 sv_free(PL_skipwhite);
895 PL_realtokenstart = s - SvPVX(PL_linestr);
899 /* skip space after PL_thistoken */
902 S_skipspace1(pTHX_ register char *s)
904 const char *start = s;
905 I32 startoff = start - SvPVX(PL_linestr);
910 start = SvPVX(PL_linestr) + startoff;
911 if (!PL_thistoken && PL_realtokenstart >= 0) {
912 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
913 PL_thistoken = newSVpvn(tstart, start - tstart);
915 PL_realtokenstart = -1;
918 PL_nextwhite = newSVpvs("");
919 sv_catsv(PL_nextwhite, PL_skipwhite);
920 sv_free(PL_skipwhite);
927 S_skipspace2(pTHX_ register char *s, SV **svp)
930 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
931 const I32 startoff = s - SvPVX(PL_linestr);
934 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
935 if (!PL_madskills || !svp)
937 start = SvPVX(PL_linestr) + startoff;
938 if (!PL_thistoken && PL_realtokenstart >= 0) {
939 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
940 PL_thistoken = newSVpvn(tstart, start - tstart);
941 PL_realtokenstart = -1;
946 sv_setsv(*svp, PL_skipwhite);
947 sv_free(PL_skipwhite);
956 S_update_debugger_info(pTHX_ SV *orig_sv, const char *buf, STRLEN len)
958 AV *av = CopFILEAVx(PL_curcop);
960 SV * const sv = newSV_type(SVt_PVMG);
962 sv_setsv(sv, orig_sv);
964 sv_setpvn(sv, buf, len);
967 av_store(av, (I32)CopLINE(PL_curcop), sv);
973 * Called to gobble the appropriate amount and type of whitespace.
974 * Skips comments as well.
978 S_skipspace(pTHX_ register char *s)
983 int startoff = s - SvPVX(PL_linestr);
986 sv_free(PL_skipwhite);
991 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
992 while (s < PL_bufend && SPACE_OR_TAB(*s))
1002 SSize_t oldprevlen, oldoldprevlen;
1003 SSize_t oldloplen = 0, oldunilen = 0;
1004 while (s < PL_bufend && isSPACE(*s)) {
1005 if (*s++ == '\n' && PL_in_eval && !PL_rsfp)
1010 if (s < PL_bufend && *s == '#') {
1011 while (s < PL_bufend && *s != '\n')
1013 if (s < PL_bufend) {
1015 if (PL_in_eval && !PL_rsfp) {
1022 /* only continue to recharge the buffer if we're at the end
1023 * of the buffer, we're not reading from a source filter, and
1024 * we're in normal lexing mode
1026 if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat ||
1027 PL_lex_state == LEX_FORMLINE)
1034 /* try to recharge the buffer */
1036 curoff = s - SvPVX(PL_linestr);
1039 if ((s = filter_gets(PL_linestr, PL_rsfp,
1040 (prevlen = SvCUR(PL_linestr)))) == NULL)
1043 if (PL_madskills && curoff != startoff) {
1045 PL_skipwhite = newSVpvs("");
1046 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1050 /* mustn't throw out old stuff yet if madpropping */
1051 SvCUR(PL_linestr) = curoff;
1052 s = SvPVX(PL_linestr) + curoff;
1054 if (curoff && s[-1] == '\n')
1058 /* end of file. Add on the -p or -n magic */
1059 /* XXX these shouldn't really be added here, can't set PL_faketokens */
1062 sv_catpvs(PL_linestr,
1063 ";}continue{print or die qq(-p destination: $!\\n);}");
1065 sv_setpvs(PL_linestr,
1066 ";}continue{print or die qq(-p destination: $!\\n);}");
1068 PL_minus_n = PL_minus_p = 0;
1070 else if (PL_minus_n) {
1072 sv_catpvn(PL_linestr, ";}", 2);
1074 sv_setpvn(PL_linestr, ";}", 2);
1080 sv_catpvn(PL_linestr,";", 1);
1082 sv_setpvn(PL_linestr,";", 1);
1085 /* reset variables for next time we lex */
1086 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart
1092 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
1093 PL_last_lop = PL_last_uni = NULL;
1095 /* Close the filehandle. Could be from -P preprocessor,
1096 * STDIN, or a regular file. If we were reading code from
1097 * STDIN (because the commandline held no -e or filename)
1098 * then we don't close it, we reset it so the code can
1099 * read from STDIN too.
1102 if (PL_preprocess && !PL_in_eval)
1103 (void)PerlProc_pclose(PL_rsfp);
1104 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
1105 PerlIO_clearerr(PL_rsfp);
1107 (void)PerlIO_close(PL_rsfp);
1112 /* not at end of file, so we only read another line */
1113 /* make corresponding updates to old pointers, for yyerror() */
1114 oldprevlen = PL_oldbufptr - PL_bufend;
1115 oldoldprevlen = PL_oldoldbufptr - PL_bufend;
1117 oldunilen = PL_last_uni - PL_bufend;
1119 oldloplen = PL_last_lop - PL_bufend;
1120 PL_linestart = PL_bufptr = s + prevlen;
1121 PL_bufend = s + SvCUR(PL_linestr);
1123 PL_oldbufptr = s + oldprevlen;
1124 PL_oldoldbufptr = s + oldoldprevlen;
1126 PL_last_uni = s + oldunilen;
1128 PL_last_lop = s + oldloplen;
1131 /* debugger active and we're not compiling the debugger code,
1132 * so store the line into the debugger's array of lines
1134 if (PERLDB_LINE && PL_curstash != PL_debstash)
1135 update_debugger_info(NULL, PL_bufptr, PL_bufend - PL_bufptr);
1142 PL_skipwhite = newSVpvs("");
1143 curoff = s - SvPVX(PL_linestr);
1144 if (curoff - startoff)
1145 sv_catpvn(PL_skipwhite, SvPVX(PL_linestr) + startoff,
1154 * Check the unary operators to ensure there's no ambiguity in how they're
1155 * used. An ambiguous piece of code would be:
1157 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1158 * the +5 is its argument.
1168 if (PL_oldoldbufptr != PL_last_uni)
1170 while (isSPACE(*PL_last_uni))
1173 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1175 if ((t = strchr(s, '(')) && t < PL_bufptr)
1178 if (ckWARN_d(WARN_AMBIGUOUS)){
1179 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
1180 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1181 (int)(s - PL_last_uni), PL_last_uni);
1186 * LOP : macro to build a list operator. Its behaviour has been replaced
1187 * with a subroutine, S_lop() for which LOP is just another name.
1190 #define LOP(f,x) return lop(f,x,s)
1194 * Build a list operator (or something that might be one). The rules:
1195 * - if we have a next token, then it's a list operator [why?]
1196 * - if the next thing is an opening paren, then it's a function
1197 * - else it's a list operator
1201 S_lop(pTHX_ I32 f, int x, char *s)
1208 PL_last_lop = PL_oldbufptr;
1209 PL_last_lop_op = (OPCODE)f;
1212 return REPORT(LSTOP);
1215 return REPORT(LSTOP);
1218 return REPORT(FUNC);
1221 return REPORT(FUNC);
1223 return REPORT(LSTOP);
1229 * Sets up for an eventual force_next(). start_force(0) basically does
1230 * an unshift, while start_force(-1) does a push. yylex removes items
1235 S_start_force(pTHX_ int where)
1239 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
1240 where = PL_lasttoke;
1241 assert(PL_curforce < 0 || PL_curforce == where);
1242 if (PL_curforce != where) {
1243 for (i = PL_lasttoke; i > where; --i) {
1244 PL_nexttoke[i] = PL_nexttoke[i-1];
1248 if (PL_curforce < 0) /* in case of duplicate start_force() */
1249 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
1250 PL_curforce = where;
1253 curmad('^', newSVpvs(""));
1254 CURMAD('_', PL_nextwhite);
1259 S_curmad(pTHX_ char slot, SV *sv)
1265 if (PL_curforce < 0)
1266 where = &PL_thismad;
1268 where = &PL_nexttoke[PL_curforce].next_mad;
1271 sv_setpvn(sv, "", 0);
1274 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1276 else if (PL_encoding) {
1277 sv_recode_to_utf8(sv, PL_encoding);
1282 /* keep a slot open for the head of the list? */
1283 if (slot != '_' && *where && (*where)->mad_key == '^') {
1284 (*where)->mad_key = slot;
1285 sv_free((SV*)((*where)->mad_val));
1286 (*where)->mad_val = (void*)sv;
1289 addmad(newMADsv(slot, sv), where, 0);
1292 # define start_force(where) NOOP
1293 # define curmad(slot, sv) NOOP
1298 * When the lexer realizes it knows the next token (for instance,
1299 * it is reordering tokens for the parser) then it can call S_force_next
1300 * to know what token to return the next time the lexer is called. Caller
1301 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1302 * and possibly PL_expect to ensure the lexer handles the token correctly.
1306 S_force_next(pTHX_ I32 type)
1310 if (PL_curforce < 0)
1311 start_force(PL_lasttoke);
1312 PL_nexttoke[PL_curforce].next_type = type;
1313 if (PL_lex_state != LEX_KNOWNEXT)
1314 PL_lex_defer = PL_lex_state;
1315 PL_lex_state = LEX_KNOWNEXT;
1316 PL_lex_expect = PL_expect;
1319 PL_nexttype[PL_nexttoke] = type;
1321 if (PL_lex_state != LEX_KNOWNEXT) {
1322 PL_lex_defer = PL_lex_state;
1323 PL_lex_expect = PL_expect;
1324 PL_lex_state = LEX_KNOWNEXT;
1330 S_newSV_maybe_utf8(pTHX_ const char *start, STRLEN len)
1333 SV * const sv = newSVpvn(start,len);
1334 if (UTF && !IN_BYTES && is_utf8_string((const U8*)start, len))
1341 * When the lexer knows the next thing is a word (for instance, it has
1342 * just seen -> and it knows that the next char is a word char, then
1343 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1347 * char *start : buffer position (must be within PL_linestr)
1348 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
1349 * int check_keyword : if true, Perl checks to make sure the word isn't
1350 * a keyword (do this if the word is a label, e.g. goto FOO)
1351 * int allow_pack : if true, : characters will also be allowed (require,
1352 * use, etc. do this)
1353 * int allow_initial_tick : used by the "sub" lexer only.
1357 S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
1363 start = SKIPSPACE1(start);
1365 if (isIDFIRST_lazy_if(s,UTF) ||
1366 (allow_pack && *s == ':') ||
1367 (allow_initial_tick && *s == '\'') )
1369 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
1370 if (check_keyword && keyword(PL_tokenbuf, len, 0))
1372 start_force(PL_curforce);
1374 curmad('X', newSVpvn(start,s-start));
1375 if (token == METHOD) {
1380 PL_expect = XOPERATOR;
1384 curmad('g', newSVpvs( "forced" ));
1385 NEXTVAL_NEXTTOKE.opval
1386 = (OP*)newSVOP(OP_CONST,0,
1387 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
1388 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
1396 * Called when the lexer wants $foo *foo &foo etc, but the program
1397 * text only contains the "foo" portion. The first argument is a pointer
1398 * to the "foo", and the second argument is the type symbol to prefix.
1399 * Forces the next token to be a "WORD".
1400 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
1404 S_force_ident(pTHX_ register const char *s, int kind)
1408 const STRLEN len = strlen(s);
1409 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
1410 start_force(PL_curforce);
1411 NEXTVAL_NEXTTOKE.opval = o;
1414 o->op_private = OPpCONST_ENTERED;
1415 /* XXX see note in pp_entereval() for why we forgo typo
1416 warnings if the symbol must be introduced in an eval.
1418 gv_fetchpvn_flags(s, len,
1419 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
1421 kind == '$' ? SVt_PV :
1422 kind == '@' ? SVt_PVAV :
1423 kind == '%' ? SVt_PVHV :
1431 Perl_str_to_version(pTHX_ SV *sv)
1436 const char *start = SvPV_const(sv,len);
1437 const char * const end = start + len;
1438 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
1439 while (start < end) {
1443 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1448 retval += ((NV)n)/nshift;
1457 * Forces the next token to be a version number.
1458 * If the next token appears to be an invalid version number, (e.g. "v2b"),
1459 * and if "guessing" is TRUE, then no new token is created (and the caller
1460 * must use an alternative parsing method).
1464 S_force_version(pTHX_ char *s, int guessing)
1470 I32 startoff = s - SvPVX(PL_linestr);
1479 while (isDIGIT(*d) || *d == '_' || *d == '.')
1483 start_force(PL_curforce);
1484 curmad('X', newSVpvn(s,d-s));
1487 if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) {
1489 s = scan_num(s, &yylval);
1490 version = yylval.opval;
1491 ver = cSVOPx(version)->op_sv;
1492 if (SvPOK(ver) && !SvNIOK(ver)) {
1493 SvUPGRADE(ver, SVt_PVNV);
1494 SvNV_set(ver, str_to_version(ver));
1495 SvNOK_on(ver); /* hint that it is a version */
1498 else if (guessing) {
1501 sv_free(PL_nextwhite); /* let next token collect whitespace */
1503 s = SvPVX(PL_linestr) + startoff;
1511 if (PL_madskills && !version) {
1512 sv_free(PL_nextwhite); /* let next token collect whitespace */
1514 s = SvPVX(PL_linestr) + startoff;
1517 /* NOTE: The parser sees the package name and the VERSION swapped */
1518 start_force(PL_curforce);
1519 NEXTVAL_NEXTTOKE.opval = version;
1527 * Tokenize a quoted string passed in as an SV. It finds the next
1528 * chunk, up to end of string or a backslash. It may make a new
1529 * SV containing that chunk (if HINT_NEW_STRING is on). It also
1534 S_tokeq(pTHX_ SV *sv)
1538 register char *send;
1546 s = SvPV_force(sv, len);
1547 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
1550 while (s < send && *s != '\\')
1555 if ( PL_hints & HINT_NEW_STRING ) {
1556 pv = sv_2mortal(newSVpvn(SvPVX_const(pv), len));
1562 if (s + 1 < send && (s[1] == '\\'))
1563 s++; /* all that, just for this */
1568 SvCUR_set(sv, d - SvPVX_const(sv));
1570 if ( PL_hints & HINT_NEW_STRING )
1571 return new_constant(NULL, 0, "q", sv, pv, "q");
1576 * Now come three functions related to double-quote context,
1577 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
1578 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
1579 * interact with PL_lex_state, and create fake ( ... ) argument lists
1580 * to handle functions and concatenation.
1581 * They assume that whoever calls them will be setting up a fake
1582 * join call, because each subthing puts a ',' after it. This lets
1585 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
1587 * (I'm not sure whether the spurious commas at the end of lcfirst's
1588 * arguments and join's arguments are created or not).
1593 * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST).
1595 * Pattern matching will set PL_lex_op to the pattern-matching op to
1596 * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise).
1598 * OP_CONST and OP_READLINE are easy--just make the new op and return.
1600 * Everything else becomes a FUNC.
1602 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
1603 * had an OP_CONST or OP_READLINE). This just sets us up for a
1604 * call to S_sublex_push().
1608 S_sublex_start(pTHX)
1611 register const I32 op_type = yylval.ival;
1613 if (op_type == OP_NULL) {
1614 yylval.opval = PL_lex_op;
1618 if (op_type == OP_CONST || op_type == OP_READLINE) {
1619 SV *sv = tokeq(PL_lex_stuff);
1621 if (SvTYPE(sv) == SVt_PVIV) {
1622 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
1624 const char * const p = SvPV_const(sv, len);
1625 SV * const nsv = newSVpvn(p, len);
1631 yylval.opval = (OP*)newSVOP(op_type, 0, sv);
1632 PL_lex_stuff = NULL;
1633 /* Allow <FH> // "foo" */
1634 if (op_type == OP_READLINE)
1635 PL_expect = XTERMORDORDOR;
1638 else if (op_type == OP_BACKTICK && PL_lex_op) {
1639 /* readpipe() vas overriden */
1640 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
1641 yylval.opval = PL_lex_op;
1643 PL_lex_stuff = NULL;
1647 PL_sublex_info.super_state = PL_lex_state;
1648 PL_sublex_info.sub_inwhat = (U16)op_type;
1649 PL_sublex_info.sub_op = PL_lex_op;
1650 PL_lex_state = LEX_INTERPPUSH;
1654 yylval.opval = PL_lex_op;
1664 * Create a new scope to save the lexing state. The scope will be
1665 * ended in S_sublex_done. Returns a '(', starting the function arguments
1666 * to the uc, lc, etc. found before.
1667 * Sets PL_lex_state to LEX_INTERPCONCAT.
1676 PL_lex_state = PL_sublex_info.super_state;
1677 SAVEBOOL(PL_lex_dojoin);
1678 SAVEI32(PL_lex_brackets);
1679 SAVEI32(PL_lex_casemods);
1680 SAVEI32(PL_lex_starts);
1681 SAVEI8(PL_lex_state);
1682 SAVEVPTR(PL_lex_inpat);
1683 SAVEI16(PL_lex_inwhat);
1684 SAVECOPLINE(PL_curcop);
1685 SAVEPPTR(PL_bufptr);
1686 SAVEPPTR(PL_bufend);
1687 SAVEPPTR(PL_oldbufptr);
1688 SAVEPPTR(PL_oldoldbufptr);
1689 SAVEPPTR(PL_last_lop);
1690 SAVEPPTR(PL_last_uni);
1691 SAVEPPTR(PL_linestart);
1692 SAVESPTR(PL_linestr);
1693 SAVEGENERICPV(PL_lex_brackstack);
1694 SAVEGENERICPV(PL_lex_casestack);
1696 PL_linestr = PL_lex_stuff;
1697 PL_lex_stuff = NULL;
1699 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
1700 = SvPVX(PL_linestr);
1701 PL_bufend += SvCUR(PL_linestr);
1702 PL_last_lop = PL_last_uni = NULL;
1703 SAVEFREESV(PL_linestr);
1705 PL_lex_dojoin = FALSE;
1706 PL_lex_brackets = 0;
1707 Newx(PL_lex_brackstack, 120, char);
1708 Newx(PL_lex_casestack, 12, char);
1709 PL_lex_casemods = 0;
1710 *PL_lex_casestack = '\0';
1712 PL_lex_state = LEX_INTERPCONCAT;
1713 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
1715 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
1716 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
1717 PL_lex_inpat = PL_sublex_info.sub_op;
1719 PL_lex_inpat = NULL;
1726 * Restores lexer state after a S_sublex_push.
1733 if (!PL_lex_starts++) {
1734 SV * const sv = newSVpvs("");
1735 if (SvUTF8(PL_linestr))
1737 PL_expect = XOPERATOR;
1738 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
1742 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
1743 PL_lex_state = LEX_INTERPCASEMOD;
1747 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
1748 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
1749 PL_linestr = PL_lex_repl;
1751 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
1752 PL_bufend += SvCUR(PL_linestr);
1753 PL_last_lop = PL_last_uni = NULL;
1754 SAVEFREESV(PL_linestr);
1755 PL_lex_dojoin = FALSE;
1756 PL_lex_brackets = 0;
1757 PL_lex_casemods = 0;
1758 *PL_lex_casestack = '\0';
1760 if (SvEVALED(PL_lex_repl)) {
1761 PL_lex_state = LEX_INTERPNORMAL;
1763 /* we don't clear PL_lex_repl here, so that we can check later
1764 whether this is an evalled subst; that means we rely on the
1765 logic to ensure sublex_done() is called again only via the
1766 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
1769 PL_lex_state = LEX_INTERPCONCAT;
1779 PL_endwhite = newSVpvs("");
1780 sv_catsv(PL_endwhite, PL_thiswhite);
1784 sv_setpvn(PL_thistoken,"",0);
1786 PL_realtokenstart = -1;
1790 PL_bufend = SvPVX(PL_linestr);
1791 PL_bufend += SvCUR(PL_linestr);
1792 PL_expect = XOPERATOR;
1793 PL_sublex_info.sub_inwhat = 0;
1801 Extracts a pattern, double-quoted string, or transliteration. This
1804 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
1805 processing a pattern (PL_lex_inpat is true), a transliteration
1806 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
1808 Returns a pointer to the character scanned up to. If this is
1809 advanced from the start pointer supplied (i.e. if anything was
1810 successfully parsed), will leave an OP for the substring scanned
1811 in yylval. Caller must intuit reason for not parsing further
1812 by looking at the next characters herself.
1816 double-quoted style: \r and \n
1817 regexp special ones: \D \s
1820 case and quoting: \U \Q \E
1821 stops on @ and $, but not for $ as tail anchor
1823 In transliterations:
1824 characters are VERY literal, except for - not at the start or end
1825 of the string, which indicates a range. If the range is in bytes,
1826 scan_const expands the range to the full set of intermediate
1827 characters. If the range is in utf8, the hyphen is replaced with
1828 a certain range mark which will be handled by pmtrans() in op.c.
1830 In double-quoted strings:
1832 double-quoted style: \r and \n
1834 deprecated backrefs: \1 (in substitution replacements)
1835 case and quoting: \U \Q \E
1838 scan_const does *not* construct ops to handle interpolated strings.
1839 It stops processing as soon as it finds an embedded $ or @ variable
1840 and leaves it to the caller to work out what's going on.
1842 embedded arrays (whether in pattern or not) could be:
1843 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
1845 $ in double-quoted strings must be the symbol of an embedded scalar.
1847 $ in pattern could be $foo or could be tail anchor. Assumption:
1848 it's a tail anchor if $ is the last thing in the string, or if it's
1849 followed by one of "()| \r\n\t"
1851 \1 (backreferences) are turned into $1
1853 The structure of the code is
1854 while (there's a character to process) {
1855 handle transliteration ranges
1856 skip regexp comments /(?#comment)/ and codes /(?{code})/
1857 skip #-initiated comments in //x patterns
1858 check for embedded arrays
1859 check for embedded scalars
1861 leave intact backslashes from leaveit (below)
1862 deprecate \1 in substitution replacements
1863 handle string-changing backslashes \l \U \Q \E, etc.
1864 switch (what was escaped) {
1865 handle \- in a transliteration (becomes a literal -)
1866 handle \132 (octal characters)
1867 handle \x15 and \x{1234} (hex characters)
1868 handle \N{name} (named characters)
1869 handle \cV (control characters)
1870 handle printf-style backslashes (\f, \r, \n, etc)
1872 } (end if backslash)
1873 } (end while character to read)
1878 S_scan_const(pTHX_ char *start)
1881 register char *send = PL_bufend; /* end of the constant */
1882 SV *sv = newSV(send - start); /* sv for the constant */
1883 register char *s = start; /* start of the constant */
1884 register char *d = SvPVX(sv); /* destination for copies */
1885 bool dorange = FALSE; /* are we in a translit range? */
1886 bool didrange = FALSE; /* did we just finish a range? */
1887 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
1888 I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */
1891 UV literal_endpoint = 0;
1892 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
1895 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
1896 /* If we are doing a trans and we know we want UTF8 set expectation */
1897 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
1898 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
1902 while (s < send || dorange) {
1903 /* get transliterations out of the way (they're most literal) */
1904 if (PL_lex_inwhat == OP_TRANS) {
1905 /* expand a range A-Z to the full set of characters. AIE! */
1907 I32 i; /* current expanded character */
1908 I32 min; /* first character in range */
1909 I32 max; /* last character in range */
1920 char * const c = (char*)utf8_hop((U8*)d, -1);
1924 *c = (char)UTF_TO_NATIVE(0xff);
1925 /* mark the range as done, and continue */
1931 i = d - SvPVX_const(sv); /* remember current offset */
1934 SvLEN(sv) + (has_utf8 ?
1935 (512 - UTF_CONTINUATION_MARK +
1938 /* How many two-byte within 0..255: 128 in UTF-8,
1939 * 96 in UTF-8-mod. */
1941 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
1943 d = SvPVX(sv) + i; /* refresh d after realloc */
1947 for (j = 0; j <= 1; j++) {
1948 char * const c = (char*)utf8_hop((U8*)d, -1);
1949 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
1955 max = (U8)0xff; /* only to \xff */
1956 uvmax = uv; /* \x{100} to uvmax */
1958 d = c; /* eat endpoint chars */
1963 d -= 2; /* eat the first char and the - */
1964 min = (U8)*d; /* first char in range */
1965 max = (U8)d[1]; /* last char in range */
1972 "Invalid range \"%c-%c\" in transliteration operator",
1973 (char)min, (char)max);
1977 if (literal_endpoint == 2 &&
1978 ((isLOWER(min) && isLOWER(max)) ||
1979 (isUPPER(min) && isUPPER(max)))) {
1981 for (i = min; i <= max; i++)
1983 *d++ = NATIVE_TO_NEED(has_utf8,i);
1985 for (i = min; i <= max; i++)
1987 *d++ = NATIVE_TO_NEED(has_utf8,i);
1992 for (i = min; i <= max; i++)
1995 const U8 ch = (U8)NATIVE_TO_UTF(i);
1996 if (UNI_IS_INVARIANT(ch))
1999 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2000 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2009 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2011 *d++ = (char)UTF_TO_NATIVE(0xff);
2013 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2017 /* mark the range as done, and continue */
2021 literal_endpoint = 0;
2026 /* range begins (ignore - as first or last char) */
2027 else if (*s == '-' && s+1 < send && s != start) {
2029 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
2036 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
2046 literal_endpoint = 0;
2047 native_range = TRUE;
2052 /* if we get here, we're not doing a transliteration */
2054 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2055 except for the last char, which will be done separately. */
2056 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
2058 while (s+1 < send && *s != ')')
2059 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2061 else if (s[2] == '{' /* This should match regcomp.c */
2062 || (s[2] == '?' && s[3] == '{'))
2065 char *regparse = s + (s[2] == '{' ? 3 : 4);
2068 while (count && (c = *regparse)) {
2069 if (c == '\\' && regparse[1])
2077 if (*regparse != ')')
2078 regparse--; /* Leave one char for continuation. */
2079 while (s < regparse)
2080 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2084 /* likewise skip #-initiated comments in //x patterns */
2085 else if (*s == '#' && PL_lex_inpat &&
2086 ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) {
2087 while (s+1 < send && *s != '\n')
2088 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2091 /* check for embedded arrays
2092 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
2094 else if (*s == '@' && s[1]) {
2095 if (isALNUM_lazy_if(s+1,UTF))
2097 if (strchr(":'{$", s[1]))
2099 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2100 break; /* in regexp, neither @+ nor @- are interpolated */
2103 /* check for embedded scalars. only stop if we're sure it's a
2106 else if (*s == '$') {
2107 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
2109 if (s + 1 < send && !strchr("()| \r\n\t", s[1]))
2110 break; /* in regexp, $ might be tail anchor */
2113 /* End of else if chain - OP_TRANS rejoin rest */
2116 if (*s == '\\' && s+1 < send) {
2119 /* deprecate \1 in strings and substitution replacements */
2120 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
2121 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
2123 if (ckWARN(WARN_SYNTAX))
2124 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
2129 /* string-change backslash escapes */
2130 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
2134 /* skip any other backslash escapes in a pattern */
2135 else if (PL_lex_inpat) {
2136 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2137 goto default_action;
2140 /* if we get here, it's either a quoted -, or a digit */
2143 /* quoted - in transliterations */
2145 if (PL_lex_inwhat == OP_TRANS) {
2152 if ((isALPHA(*s) || isDIGIT(*s)) &&
2154 Perl_warner(aTHX_ packWARN(WARN_MISC),
2155 "Unrecognized escape \\%c passed through",
2157 /* default action is to copy the quoted character */
2158 goto default_action;
2161 /* \132 indicates an octal constant */
2162 case '0': case '1': case '2': case '3':
2163 case '4': case '5': case '6': case '7':
2167 uv = grok_oct(s, &len, &flags, NULL);
2170 goto NUM_ESCAPE_INSERT;
2172 /* \x24 indicates a hex constant */
2176 char* const e = strchr(s, '}');
2177 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2178 PERL_SCAN_DISALLOW_PREFIX;
2183 yyerror("Missing right brace on \\x{}");
2187 uv = grok_hex(s, &len, &flags, NULL);
2193 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
2194 uv = grok_hex(s, &len, &flags, NULL);
2200 /* Insert oct or hex escaped character.
2201 * There will always enough room in sv since such
2202 * escapes will be longer than any UTF-8 sequence
2203 * they can end up as. */
2205 /* We need to map to chars to ASCII before doing the tests
2208 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) {
2209 if (!has_utf8 && uv > 255) {
2210 /* Might need to recode whatever we have
2211 * accumulated so far if it contains any
2214 * (Can't we keep track of that and avoid
2215 * this rescan? --jhi)
2219 for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) {
2220 if (!NATIVE_IS_INVARIANT(*c)) {
2225 const STRLEN offset = d - SvPVX_const(sv);
2227 d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset;
2231 while (src >= (const U8 *)SvPVX_const(sv)) {
2232 if (!NATIVE_IS_INVARIANT(*src)) {
2233 const U8 ch = NATIVE_TO_ASCII(*src);
2234 *dst-- = (U8)UTF8_EIGHT_BIT_LO(ch);
2235 *dst-- = (U8)UTF8_EIGHT_BIT_HI(ch);
2245 if (has_utf8 || uv > 255) {
2246 d = (char*)uvchr_to_utf8((U8*)d, uv);
2248 if (PL_lex_inwhat == OP_TRANS &&
2249 PL_sublex_info.sub_op) {
2250 PL_sublex_info.sub_op->op_private |=
2251 (PL_lex_repl ? OPpTRANS_FROM_UTF
2255 if (uv > 255 && !dorange)
2256 native_range = FALSE;
2268 /* \N{LATIN SMALL LETTER A} is a named character */
2272 char* e = strchr(s, '}');
2279 yyerror("Missing right brace on \\N{}");
2283 if (e > s + 2 && s[1] == 'U' && s[2] == '+') {
2285 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2286 PERL_SCAN_DISALLOW_PREFIX;
2289 uv = grok_hex(s, &len, &flags, NULL);
2290 if ( e > s && len != (STRLEN)(e - s) ) {
2294 goto NUM_ESCAPE_INSERT;
2296 res = newSVpvn(s + 1, e - s - 1);
2297 type = newSVpvn(s - 2,e - s + 3);
2298 res = new_constant( NULL, 0, "charnames",
2299 res, NULL, SvPVX(type) );
2302 sv_utf8_upgrade(res);
2303 str = SvPV_const(res,len);
2304 #ifdef EBCDIC_NEVER_MIND
2305 /* charnames uses pack U and that has been
2306 * recently changed to do the below uni->native
2307 * mapping, so this would be redundant (and wrong,
2308 * the code point would be doubly converted).
2309 * But leave this in just in case the pack U change
2310 * gets revoked, but the semantics is still
2311 * desireable for charnames. --jhi */
2313 UV uv = utf8_to_uvchr((const U8*)str, 0);
2316 U8 tmpbuf[UTF8_MAXBYTES+1], *d;
2318 d = uvchr_to_utf8(tmpbuf, UNI_TO_NATIVE(uv));
2319 sv_setpvn(res, (char *)tmpbuf, d - tmpbuf);
2320 str = SvPV_const(res, len);
2324 if (!has_utf8 && SvUTF8(res)) {
2325 const char * const ostart = SvPVX_const(sv);
2326 SvCUR_set(sv, d - ostart);
2329 sv_utf8_upgrade(sv);
2330 /* this just broke our allocation above... */
2331 SvGROW(sv, (STRLEN)(send - start));
2332 d = SvPVX(sv) + SvCUR(sv);
2335 if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
2336 const char * const odest = SvPVX_const(sv);
2338 SvGROW(sv, (SvLEN(sv) + len - (e - s + 4)));
2339 d = SvPVX(sv) + (d - odest);
2343 native_range = FALSE; /* \N{} is guessed to be Unicode */
2345 Copy(str, d, len, char);
2352 yyerror("Missing braces on \\N{}");
2355 /* \c is a control character */
2364 *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c));
2367 yyerror("Missing control char name in \\c");
2371 /* printf-style backslashes, formfeeds, newlines, etc */
2373 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
2376 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
2379 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
2382 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
2385 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
2388 *d++ = ASCII_TO_NEED(has_utf8,'\033');
2391 *d++ = ASCII_TO_NEED(has_utf8,'\007');
2397 } /* end if (backslash) */
2404 /* If we started with encoded form, or already know we want it
2405 and then encode the next character */
2406 if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) {
2408 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
2409 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
2412 /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */
2413 const STRLEN off = d - SvPVX_const(sv);
2414 d = SvGROW(sv, SvLEN(sv) + (need-len)) + off;
2416 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
2419 if (uv > 255 && !dorange)
2420 native_range = FALSE;
2424 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
2426 } /* while loop to process each character */
2428 /* terminate the string and set up the sv */
2430 SvCUR_set(sv, d - SvPVX_const(sv));
2431 if (SvCUR(sv) >= SvLEN(sv))
2432 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2435 if (PL_encoding && !has_utf8) {
2436 sv_recode_to_utf8(sv, PL_encoding);
2442 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2443 PL_sublex_info.sub_op->op_private |=
2444 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2448 /* shrink the sv if we allocated more than we used */
2449 if (SvCUR(sv) + 5 < SvLEN(sv)) {
2450 SvPV_shrink_to_cur(sv);
2453 /* return the substring (via yylval) only if we parsed anything */
2454 if (s > PL_bufptr) {
2455 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) )
2456 sv = new_constant(start, s - start,
2457 (const char *)(PL_lex_inpat ? "qr" : "q"),
2460 (( PL_lex_inwhat == OP_TRANS
2462 : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat)
2465 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
2472 * Returns TRUE if there's more to the expression (e.g., a subscript),
2475 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
2477 * ->[ and ->{ return TRUE
2478 * { and [ outside a pattern are always subscripts, so return TRUE
2479 * if we're outside a pattern and it's not { or [, then return FALSE
2480 * if we're in a pattern and the first char is a {
2481 * {4,5} (any digits around the comma) returns FALSE
2482 * if we're in a pattern and the first char is a [
2484 * [SOMETHING] has a funky algorithm to decide whether it's a
2485 * character class or not. It has to deal with things like
2486 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
2487 * anything else returns TRUE
2490 /* This is the one truly awful dwimmer necessary to conflate C and sed. */
2493 S_intuit_more(pTHX_ register char *s)
2496 if (PL_lex_brackets)
2498 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
2500 if (*s != '{' && *s != '[')
2505 /* In a pattern, so maybe we have {n,m}. */
2522 /* On the other hand, maybe we have a character class */
2525 if (*s == ']' || *s == '^')
2528 /* this is terrifying, and it works */
2529 int weight = 2; /* let's weigh the evidence */
2531 unsigned char un_char = 255, last_un_char;
2532 const char * const send = strchr(s,']');
2533 char tmpbuf[sizeof PL_tokenbuf * 4];
2535 if (!send) /* has to be an expression */
2538 Zero(seen,256,char);
2541 else if (isDIGIT(*s)) {
2543 if (isDIGIT(s[1]) && s[2] == ']')
2549 for (; s < send; s++) {
2550 last_un_char = un_char;
2551 un_char = (unsigned char)*s;
2556 weight -= seen[un_char] * 10;
2557 if (isALNUM_lazy_if(s+1,UTF)) {
2559 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
2560 len = (int)strlen(tmpbuf);
2561 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
2566 else if (*s == '$' && s[1] &&
2567 strchr("[#!%*<>()-=",s[1])) {
2568 if (/*{*/ strchr("])} =",s[2]))
2577 if (strchr("wds]",s[1]))
2579 else if (seen[(U8)'\''] || seen[(U8)'"'])
2581 else if (strchr("rnftbxcav",s[1]))
2583 else if (isDIGIT(s[1])) {
2585 while (s[1] && isDIGIT(s[1]))
2595 if (strchr("aA01! ",last_un_char))
2597 if (strchr("zZ79~",s[1]))
2599 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
2600 weight -= 5; /* cope with negative subscript */
2603 if (!isALNUM(last_un_char)
2604 && !(last_un_char == '$' || last_un_char == '@'
2605 || last_un_char == '&')
2606 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
2611 if (keyword(tmpbuf, d - tmpbuf, 0))
2614 if (un_char == last_un_char + 1)
2616 weight -= seen[un_char];
2621 if (weight >= 0) /* probably a character class */
2631 * Does all the checking to disambiguate
2633 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
2634 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
2636 * First argument is the stuff after the first token, e.g. "bar".
2638 * Not a method if bar is a filehandle.
2639 * Not a method if foo is a subroutine prototyped to take a filehandle.
2640 * Not a method if it's really "Foo $bar"
2641 * Method if it's "foo $bar"
2642 * Not a method if it's really "print foo $bar"
2643 * Method if it's really "foo package::" (interpreted as package->foo)
2644 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
2645 * Not a method if bar is a filehandle or package, but is quoted with
2650 S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
2653 char *s = start + (*start == '$');
2654 char tmpbuf[sizeof PL_tokenbuf];
2662 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
2666 const char *proto = SvPVX_const(cv);
2677 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
2678 /* start is the beginning of the possible filehandle/object,
2679 * and s is the end of it
2680 * tmpbuf is a copy of it
2683 if (*start == '$') {
2684 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
2685 isUPPER(*PL_tokenbuf))
2688 len = start - SvPVX(PL_linestr);
2692 start = SvPVX(PL_linestr) + len;
2696 return *s == '(' ? FUNCMETH : METHOD;
2698 if (!keyword(tmpbuf, len, 0)) {
2699 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
2703 soff = s - SvPVX(PL_linestr);
2707 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
2708 if (indirgv && GvCVu(indirgv))
2710 /* filehandle or package name makes it a method */
2711 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
2713 soff = s - SvPVX(PL_linestr);
2716 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
2717 return 0; /* no assumptions -- "=>" quotes bearword */
2719 start_force(PL_curforce);
2720 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
2721 newSVpvn(tmpbuf,len));
2722 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
2724 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
2729 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
2731 return *s == '(' ? FUNCMETH : METHOD;
2739 * Return a string of Perl code to load the debugger. If PERL5DB
2740 * is set, it will return the contents of that, otherwise a
2741 * compile-time require of perl5db.pl.
2749 const char * const pdb = PerlEnv_getenv("PERL5DB");
2753 SETERRNO(0,SS_NORMAL);
2754 return "BEGIN { require 'perl5db.pl' }";
2760 /* Encoded script support. filter_add() effectively inserts a
2761 * 'pre-processing' function into the current source input stream.
2762 * Note that the filter function only applies to the current source file
2763 * (e.g., it will not affect files 'require'd or 'use'd by this one).
2765 * The datasv parameter (which may be NULL) can be used to pass
2766 * private data to this instance of the filter. The filter function
2767 * can recover the SV using the FILTER_DATA macro and use it to
2768 * store private buffers and state information.
2770 * The supplied datasv parameter is upgraded to a PVIO type
2771 * and the IoDIRP/IoANY field is used to store the function pointer,
2772 * and IOf_FAKE_DIRP is enabled on datasv to mark this as such.
2773 * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for
2774 * private use must be set using malloc'd pointers.
2778 Perl_filter_add(pTHX_ filter_t funcp, SV *datasv)
2787 if (!PL_rsfp_filters)
2788 PL_rsfp_filters = newAV();
2791 SvUPGRADE(datasv, SVt_PVIO);
2792 IoANY(datasv) = FPTR2DPTR(void *, funcp); /* stash funcp into spare field */
2793 IoFLAGS(datasv) |= IOf_FAKE_DIRP;
2794 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n",
2795 FPTR2DPTR(void *, IoANY(datasv)),
2796 SvPV_nolen(datasv)));
2797 av_unshift(PL_rsfp_filters, 1);
2798 av_store(PL_rsfp_filters, 0, datasv) ;
2803 /* Delete most recently added instance of this filter function. */
2805 Perl_filter_del(pTHX_ filter_t funcp)
2811 DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p",
2812 FPTR2DPTR(void*, funcp)));
2814 if (!PL_parser || !PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0)
2816 /* if filter is on top of stack (usual case) just pop it off */
2817 datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters));
2818 if (IoANY(datasv) == FPTR2DPTR(void *, funcp)) {
2819 IoFLAGS(datasv) &= ~IOf_FAKE_DIRP;
2820 IoANY(datasv) = (void *)NULL;
2821 sv_free(av_pop(PL_rsfp_filters));
2825 /* we need to search for the correct entry and clear it */
2826 Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)");
2830 /* Invoke the idxth filter function for the current rsfp. */
2831 /* maxlen 0 = read one text line */
2833 Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen)
2838 /* This API is bad. It should have been using unsigned int for maxlen.
2839 Not sure if we want to change the API, but if not we should sanity
2840 check the value here. */
2841 const unsigned int correct_length
2850 if (!PL_parser || !PL_rsfp_filters)
2852 if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */
2853 /* Provide a default input filter to make life easy. */
2854 /* Note that we append to the line. This is handy. */
2855 DEBUG_P(PerlIO_printf(Perl_debug_log,
2856 "filter_read %d: from rsfp\n", idx));
2857 if (correct_length) {
2860 const int old_len = SvCUR(buf_sv);
2862 /* ensure buf_sv is large enough */
2863 SvGROW(buf_sv, (STRLEN)(old_len + correct_length)) ;
2864 if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len,
2865 correct_length)) <= 0) {
2866 if (PerlIO_error(PL_rsfp))
2867 return -1; /* error */
2869 return 0 ; /* end of file */
2871 SvCUR_set(buf_sv, old_len + len) ;
2874 if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) {
2875 if (PerlIO_error(PL_rsfp))
2876 return -1; /* error */
2878 return 0 ; /* end of file */
2881 return SvCUR(buf_sv);
2883 /* Skip this filter slot if filter has been deleted */
2884 if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) {
2885 DEBUG_P(PerlIO_printf(Perl_debug_log,
2886 "filter_read %d: skipped (filter deleted)\n",
2888 return FILTER_READ(idx+1, buf_sv, correct_length); /* recurse */
2890 /* Get function pointer hidden within datasv */
2891 funcp = DPTR2FPTR(filter_t, IoANY(datasv));
2892 DEBUG_P(PerlIO_printf(Perl_debug_log,
2893 "filter_read %d: via function %p (%s)\n",
2894 idx, (void*)datasv, SvPV_nolen_const(datasv)));
2895 /* Call function. The function is expected to */
2896 /* call "FILTER_READ(idx+1, buf_sv)" first. */
2897 /* Return: <0:error, =0:eof, >0:not eof */
2898 return (*funcp)(aTHX_ idx, buf_sv, correct_length);
2902 S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append)
2905 #ifdef PERL_CR_FILTER
2906 if (!PL_rsfp_filters) {
2907 filter_add(S_cr_textfilter,NULL);
2910 if (PL_rsfp_filters) {
2912 SvCUR_set(sv, 0); /* start with empty line */
2913 if (FILTER_READ(0, sv, 0) > 0)
2914 return ( SvPVX(sv) ) ;
2919 return (sv_gets(sv, fp, append));
2923 S_find_in_my_stash(pTHX_ const char *pkgname, I32 len)
2928 if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__"))
2932 (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') &&
2933 (gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVHV)))
2935 return GvHV(gv); /* Foo:: */
2938 /* use constant CLASS => 'MyClass' */
2939 gv = gv_fetchpvn_flags(pkgname, len, 0, SVt_PVCV);
2940 if (gv && GvCV(gv)) {
2941 SV * const sv = cv_const_sv(GvCV(gv));
2943 pkgname = SvPV_nolen_const(sv);
2946 return gv_stashpv(pkgname, 0);
2950 * S_readpipe_override
2951 * Check whether readpipe() is overriden, and generates the appropriate
2952 * optree, provided sublex_start() is called afterwards.
2955 S_readpipe_override(pTHX)
2958 GV *gv_readpipe = gv_fetchpvs("readpipe", GV_NOTQUAL, SVt_PVCV);
2959 yylval.ival = OP_BACKTICK;
2961 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe))
2963 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readpipe", FALSE))
2964 && (gv_readpipe = *gvp) && isGV_with_GP(gv_readpipe)
2965 && GvCVu(gv_readpipe) && GvIMPORTED_CV(gv_readpipe)))
2967 PL_lex_op = (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
2968 append_elem(OP_LIST,
2969 newSVOP(OP_CONST, 0, &PL_sv_undef), /* value will be read later */
2970 newCVREF(0, newGVOP(OP_GV, 0, gv_readpipe))));
2977 * The intent of this yylex wrapper is to minimize the changes to the
2978 * tokener when we aren't interested in collecting madprops. It remains
2979 * to be seen how successful this strategy will be...
2986 char *s = PL_bufptr;
2988 /* make sure PL_thiswhite is initialized */
2992 /* just do what yylex would do on pending identifier; leave PL_thiswhite alone */
2993 if (PL_pending_ident)
2994 return S_pending_ident(aTHX);
2996 /* previous token ate up our whitespace? */
2997 if (!PL_lasttoke && PL_nextwhite) {
2998 PL_thiswhite = PL_nextwhite;
3002 /* isolate the token, and figure out where it is without whitespace */
3003 PL_realtokenstart = -1;
3007 assert(PL_curforce < 0);
3009 if (!PL_thismad || PL_thismad->mad_key == '^') { /* not forced already? */
3010 if (!PL_thistoken) {
3011 if (PL_realtokenstart < 0 || !CopLINE(PL_curcop))
3012 PL_thistoken = newSVpvs("");
3014 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
3015 PL_thistoken = newSVpvn(tstart, s - tstart);
3018 if (PL_thismad) /* install head */
3019 CURMAD('X', PL_thistoken);
3022 /* last whitespace of a sublex? */
3023 if (optype == ')' && PL_endwhite) {
3024 CURMAD('X', PL_endwhite);
3029 /* if no whitespace and we're at EOF, bail. Otherwise fake EOF below. */
3030 if (!PL_thiswhite && !PL_endwhite && !optype) {
3031 sv_free(PL_thistoken);
3036 /* put off final whitespace till peg */
3037 if (optype == ';' && !PL_rsfp) {
3038 PL_nextwhite = PL_thiswhite;
3041 else if (PL_thisopen) {
3042 CURMAD('q', PL_thisopen);
3044 sv_free(PL_thistoken);
3048 /* Store actual token text as madprop X */
3049 CURMAD('X', PL_thistoken);
3053 /* add preceding whitespace as madprop _ */
3054 CURMAD('_', PL_thiswhite);
3058 /* add quoted material as madprop = */
3059 CURMAD('=', PL_thisstuff);
3063 /* add terminating quote as madprop Q */
3064 CURMAD('Q', PL_thisclose);
3068 /* special processing based on optype */
3072 /* opval doesn't need a TOKEN since it can already store mp */
3083 append_madprops(PL_thismad, yylval.opval, 0);
3091 addmad(newMADsv('p', PL_endwhite), &PL_thismad, 0);
3100 /* remember any fake bracket that lexer is about to discard */
3101 if (PL_lex_brackets == 1 &&
3102 ((expectation)PL_lex_brackstack[0] & XFAKEBRACK))
3105 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3108 PL_thiswhite = newSVpvn(PL_bufptr, ++s - PL_bufptr);
3109 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3112 break; /* don't bother looking for trailing comment */
3121 /* attach a trailing comment to its statement instead of next token */
3125 if (PL_bufptr > PL_oldbufptr && PL_bufptr[-1] == optype) {
3127 while (s < PL_bufend && (*s == ' ' || *s == '\t'))
3129 if (*s == '\n' || *s == '#') {
3130 while (s < PL_bufend && *s != '\n')
3134 PL_thiswhite = newSVpvn(PL_bufptr, s - PL_bufptr);
3135 addmad(newMADsv('#', PL_thiswhite), &PL_thismad, 0);
3152 /* Create new token struct. Note: opvals return early above. */
3153 yylval.tkval = newTOKEN(optype, yylval, PL_thismad);
3160 S_tokenize_use(pTHX_ int is_use, char *s) {
3162 if (PL_expect != XSTATE)
3163 yyerror(Perl_form(aTHX_ "\"%s\" not allowed in expression",
3164 is_use ? "use" : "no"));
3166 if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) {
3167 s = force_version(s, TRUE);
3168 if (*s == ';' || (s = SKIPSPACE1(s), *s == ';')) {
3169 start_force(PL_curforce);
3170 NEXTVAL_NEXTTOKE.opval = NULL;
3173 else if (*s == 'v') {
3174 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3175 s = force_version(s, FALSE);
3179 s = force_word(s,WORD,FALSE,TRUE,FALSE);
3180 s = force_version(s, FALSE);
3182 yylval.ival = is_use;
3186 static const char* const exp_name[] =
3187 { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK",
3188 "ATTRTERM", "TERMBLOCK", "TERMORDORDOR"
3195 Works out what to call the token just pulled out of the input
3196 stream. The yacc parser takes care of taking the ops we return and
3197 stitching them into a tree.
3203 if read an identifier
3204 if we're in a my declaration
3205 croak if they tried to say my($foo::bar)
3206 build the ops for a my() declaration
3207 if it's an access to a my() variable
3208 are we in a sort block?
3209 croak if my($a); $a <=> $b
3210 build ops for access to a my() variable
3211 if in a dq string, and they've said @foo and we can't find @foo
3213 build ops for a bareword
3214 if we already built the token before, use it.
3219 #pragma segment Perl_yylex
3225 register char *s = PL_bufptr;
3230 /* orig_keyword, gvp, and gv are initialized here because
3231 * jump to the label just_a_word_zero can bypass their
3232 * initialization later. */
3233 I32 orig_keyword = 0;
3238 SV* tmp = newSVpvs("");
3239 PerlIO_printf(Perl_debug_log, "### %"IVdf":LEX_%s/X%s %s\n",
3240 (IV)CopLINE(PL_curcop),
3241 lex_state_names[PL_lex_state],
3242 exp_name[PL_expect],
3243 pv_display(tmp, s, strlen(s), 0, 60));
3246 /* check if there's an identifier for us to look at */
3247 if (PL_pending_ident)
3248 return REPORT(S_pending_ident(aTHX));
3250 /* no identifier pending identification */
3252 switch (PL_lex_state) {
3254 case LEX_NORMAL: /* Some compilers will produce faster */
3255 case LEX_INTERPNORMAL: /* code if we comment these out. */
3259 /* when we've already built the next token, just pull it out of the queue */
3263 yylval = PL_nexttoke[PL_lasttoke].next_val;
3265 PL_thismad = PL_nexttoke[PL_lasttoke].next_mad;
3266 PL_nexttoke[PL_lasttoke].next_mad = 0;
3267 if (PL_thismad && PL_thismad->mad_key == '_') {
3268 PL_thiswhite = (SV*)PL_thismad->mad_val;
3269 PL_thismad->mad_val = 0;
3270 mad_free(PL_thismad);
3275 PL_lex_state = PL_lex_defer;
3276 PL_expect = PL_lex_expect;
3277 PL_lex_defer = LEX_NORMAL;
3278 if (!PL_nexttoke[PL_lasttoke].next_type)
3283 yylval = PL_nextval[PL_nexttoke];
3285 PL_lex_state = PL_lex_defer;
3286 PL_expect = PL_lex_expect;
3287 PL_lex_defer = LEX_NORMAL;
3291 /* FIXME - can these be merged? */
3292 return(PL_nexttoke[PL_lasttoke].next_type);
3294 return REPORT(PL_nexttype[PL_nexttoke]);
3297 /* interpolated case modifiers like \L \U, including \Q and \E.
3298 when we get here, PL_bufptr is at the \
3300 case LEX_INTERPCASEMOD:
3302 if (PL_bufptr != PL_bufend && *PL_bufptr != '\\')
3303 Perl_croak(aTHX_ "panic: INTERPCASEMOD");
3305 /* handle \E or end of string */
3306 if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') {
3308 if (PL_lex_casemods) {
3309 const char oldmod = PL_lex_casestack[--PL_lex_casemods];
3310 PL_lex_casestack[PL_lex_casemods] = '\0';
3312 if (PL_bufptr != PL_bufend
3313 && (oldmod == 'L' || oldmod == 'U' || oldmod == 'Q')) {
3315 PL_lex_state = LEX_INTERPCONCAT;
3318 PL_thistoken = newSVpvs("\\E");
3324 while (PL_bufptr != PL_bufend &&
3325 PL_bufptr[0] == '\\' && PL_bufptr[1] == 'E') {
3327 PL_thiswhite = newSVpvs("");
3328 sv_catpvn(PL_thiswhite, PL_bufptr, 2);
3332 if (PL_bufptr != PL_bufend)
3335 PL_lex_state = LEX_INTERPCONCAT;
3339 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3340 "### Saw case modifier\n"); });
3342 if (s[1] == '\\' && s[2] == 'E') {
3345 PL_thiswhite = newSVpvs("");
3346 sv_catpvn(PL_thiswhite, PL_bufptr, 4);
3349 PL_lex_state = LEX_INTERPCONCAT;
3354 if (!PL_madskills) /* when just compiling don't need correct */
3355 if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3))
3356 tmp = *s, *s = s[2], s[2] = (char)tmp; /* misordered... */
3357 if ((*s == 'L' || *s == 'U') &&
3358 (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) {
3359 PL_lex_casestack[--PL_lex_casemods] = '\0';
3362 if (PL_lex_casemods > 10)
3363 Renew(PL_lex_casestack, PL_lex_casemods + 2, char);
3364 PL_lex_casestack[PL_lex_casemods++] = *s;
3365 PL_lex_casestack[PL_lex_casemods] = '\0';
3366 PL_lex_state = LEX_INTERPCONCAT;
3367 start_force(PL_curforce);
3368 NEXTVAL_NEXTTOKE.ival = 0;
3370 start_force(PL_curforce);
3372 NEXTVAL_NEXTTOKE.ival = OP_LCFIRST;
3374 NEXTVAL_NEXTTOKE.ival = OP_UCFIRST;
3376 NEXTVAL_NEXTTOKE.ival = OP_LC;
3378 NEXTVAL_NEXTTOKE.ival = OP_UC;
3380 NEXTVAL_NEXTTOKE.ival = OP_QUOTEMETA;
3382 Perl_croak(aTHX_ "panic: yylex");
3384 SV* const tmpsv = newSVpvs("\\ ");
3385 /* replace the space with the character we want to escape
3387 SvPVX(tmpsv)[1] = *s;
3393 if (PL_lex_starts) {
3399 sv_free(PL_thistoken);
3400 PL_thistoken = newSVpvs("");
3403 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3404 if (PL_lex_casemods == 1 && PL_lex_inpat)
3413 case LEX_INTERPPUSH:
3414 return REPORT(sublex_push());
3416 case LEX_INTERPSTART:
3417 if (PL_bufptr == PL_bufend)
3418 return REPORT(sublex_done());
3419 DEBUG_T({ PerlIO_printf(Perl_debug_log,
3420 "### Interpolated variable\n"); });
3422 PL_lex_dojoin = (*PL_bufptr == '@');
3423 PL_lex_state = LEX_INTERPNORMAL;
3424 if (PL_lex_dojoin) {
3425 start_force(PL_curforce);
3426 NEXTVAL_NEXTTOKE.ival = 0;
3428 start_force(PL_curforce);
3429 force_ident("\"", '$');
3430 start_force(PL_curforce);
3431 NEXTVAL_NEXTTOKE.ival = 0;
3433 start_force(PL_curforce);
3434 NEXTVAL_NEXTTOKE.ival = 0;
3436 start_force(PL_curforce);
3437 NEXTVAL_NEXTTOKE.ival = OP_JOIN; /* emulate join($", ...) */
3440 if (PL_lex_starts++) {
3445 sv_free(PL_thistoken);
3446 PL_thistoken = newSVpvs("");
3449 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3450 if (!PL_lex_casemods && PL_lex_inpat)
3457 case LEX_INTERPENDMAYBE:
3458 if (intuit_more(PL_bufptr)) {
3459 PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */
3465 if (PL_lex_dojoin) {
3466 PL_lex_dojoin = FALSE;
3467 PL_lex_state = LEX_INTERPCONCAT;
3471 sv_free(PL_thistoken);
3472 PL_thistoken = newSVpvs("");
3477 if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl
3478 && SvEVALED(PL_lex_repl))
3480 if (PL_bufptr != PL_bufend)
3481 Perl_croak(aTHX_ "Bad evalled substitution pattern");
3485 case LEX_INTERPCONCAT:
3487 if (PL_lex_brackets)
3488 Perl_croak(aTHX_ "panic: INTERPCONCAT");
3490 if (PL_bufptr == PL_bufend)
3491 return REPORT(sublex_done());
3493 if (SvIVX(PL_linestr) == '\'') {
3494 SV *sv = newSVsv(PL_linestr);
3497 else if ( PL_hints & HINT_NEW_RE )
3498 sv = new_constant(NULL, 0, "qr", sv, sv, "q");
3499 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
3503 s = scan_const(PL_bufptr);
3505 PL_lex_state = LEX_INTERPCASEMOD;
3507 PL_lex_state = LEX_INTERPSTART;
3510 if (s != PL_bufptr) {
3511 start_force(PL_curforce);
3513 curmad('X', newSVpvn(PL_bufptr,s-PL_bufptr));
3515 NEXTVAL_NEXTTOKE = yylval;
3518 if (PL_lex_starts++) {
3522 sv_free(PL_thistoken);
3523 PL_thistoken = newSVpvs("");
3526 /* commas only at base level: /$a\Ub$c/ => ($a,uc(b.$c)) */
3527 if (!PL_lex_casemods && PL_lex_inpat)
3540 PL_lex_state = LEX_NORMAL;
3541 s = scan_formline(PL_bufptr);
3542 if (!PL_lex_formbrack)
3548 PL_oldoldbufptr = PL_oldbufptr;
3554 sv_free(PL_thistoken);
3557 PL_realtokenstart = s - SvPVX(PL_linestr); /* assume but undo on ws */
3561 if (isIDFIRST_lazy_if(s,UTF))
3563 len = UTF ? Perl_utf8_length(aTHX_ (U8 *) PL_linestart, (U8 *) s) : (STRLEN) (s - PL_linestart);
3564 Perl_croak(aTHX_ "Unrecognized character \\x%02X in column %d", *s & 255, (int) len + 1);
3567 goto fake_eof; /* emulate EOF on ^D or ^Z */
3576 if (PL_lex_brackets) {
3577 yyerror((const char *)
3579 ? "Format not terminated"
3580 : "Missing right curly or square bracket"));
3582 DEBUG_T( { PerlIO_printf(Perl_debug_log,
3583 "### Tokener got EOF\n");
3587 if (s++ < PL_bufend)
3588 goto retry; /* ignore stray nulls */
3591 if (!PL_in_eval && !PL_preambled) {
3592 PL_preambled = TRUE;
3597 sv_setpv(PL_linestr,incl_perldb());
3598 if (SvCUR(PL_linestr))
3599 sv_catpvs(PL_linestr,";");
3601 while(AvFILLp(PL_preambleav) >= 0) {
3602 SV *tmpsv = av_shift(PL_preambleav);
3603 sv_catsv(PL_linestr, tmpsv);
3604 sv_catpvs(PL_linestr, ";");
3607 sv_free((SV*)PL_preambleav);
3608 PL_preambleav = NULL;
3610 if (PL_minus_n || PL_minus_p) {
3611 sv_catpvs(PL_linestr, "LINE: while (<>) {");
3613 sv_catpvs(PL_linestr,"chomp;");
3616 if ((*PL_splitstr == '/' || *PL_splitstr == '\''
3617 || *PL_splitstr == '"')
3618 && strchr(PL_splitstr + 1, *PL_splitstr))
3619 Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s);", PL_splitstr);
3621 /* "q\0${splitstr}\0" is legal perl. Yes, even NUL
3622 bytes can be used as quoting characters. :-) */
3623 const char *splits = PL_splitstr;
3624 sv_catpvs(PL_linestr, "our @F=split(q\0");
3627 if (*splits == '\\')
3628 sv_catpvn(PL_linestr, splits, 1);
3629 sv_catpvn(PL_linestr, splits, 1);
3630 } while (*splits++);
3631 /* This loop will embed the trailing NUL of
3632 PL_linestr as the last thing it does before
3634 sv_catpvs(PL_linestr, ");");
3638 sv_catpvs(PL_linestr,"our @F=split(' ');");
3642 sv_catpvs(PL_linestr,"use feature ':5.10';");
3643 sv_catpvs(PL_linestr, "\n");
3644 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3645 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3646 PL_last_lop = PL_last_uni = NULL;
3647 if (PERLDB_LINE && PL_curstash != PL_debstash)
3648 update_debugger_info(PL_linestr, NULL, 0);
3652 bof = PL_rsfp ? TRUE : FALSE;
3653 if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == NULL) {
3656 PL_realtokenstart = -1;
3659 if (PL_preprocess && !PL_in_eval)
3660 (void)PerlProc_pclose(PL_rsfp);
3661 else if ((PerlIO *)PL_rsfp == PerlIO_stdin())
3662 PerlIO_clearerr(PL_rsfp);
3664 (void)PerlIO_close(PL_rsfp);
3666 PL_doextract = FALSE;
3668 if (!PL_in_eval && (PL_minus_n || PL_minus_p)) {
3673 sv_setpv(PL_linestr,
3676 ? ";}continue{print;}" : ";}"));
3677 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3678 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3679 PL_last_lop = PL_last_uni = NULL;
3680 PL_minus_n = PL_minus_p = 0;
3683 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3684 PL_last_lop = PL_last_uni = NULL;
3685 sv_setpvn(PL_linestr,"",0);
3686 TOKEN(';'); /* not infinite loop because rsfp is NULL now */
3688 /* If it looks like the start of a BOM or raw UTF-16,
3689 * check if it in fact is. */
3695 #ifdef PERLIO_IS_STDIO
3696 # ifdef __GNU_LIBRARY__
3697 # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */
3698 # define FTELL_FOR_PIPE_IS_BROKEN
3702 # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */
3703 # define FTELL_FOR_PIPE_IS_BROKEN
3708 #ifdef FTELL_FOR_PIPE_IS_BROKEN
3709 /* This loses the possibility to detect the bof
3710 * situation on perl -P when the libc5 is being used.
3711 * Workaround? Maybe attach some extra state to PL_rsfp?
3714 bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr);
3716 bof = PerlIO_tell(PL_rsfp) == (Off_t)SvCUR(PL_linestr);
3719 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3720 s = swallow_bom((U8*)s);
3724 /* Incest with pod. */
3727 sv_catsv(PL_thiswhite, PL_linestr);
3729 if (*s == '=' && strnEQ(s, "=cut", 4) && !isALPHA(s[4])) {
3730 sv_setpvn(PL_linestr, "", 0);
3731 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3732 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3733 PL_last_lop = PL_last_uni = NULL;
3734 PL_doextract = FALSE;
3738 } while (PL_doextract);
3739 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s;
3740 if (PERLDB_LINE && PL_curstash != PL_debstash)
3741 update_debugger_info(PL_linestr, NULL, 0);
3742 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3743 PL_last_lop = PL_last_uni = NULL;
3744 if (CopLINE(PL_curcop) == 1) {
3745 while (s < PL_bufend && isSPACE(*s))
3747 if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */
3751 PL_thiswhite = newSVpvn(PL_linestart, s - PL_linestart);
3755 if (*s == '#' && *(s+1) == '!')
3757 #ifdef ALTERNATE_SHEBANG
3759 static char const as[] = ALTERNATE_SHEBANG;
3760 if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1))
3761 d = s + (sizeof(as) - 1);
3763 #endif /* ALTERNATE_SHEBANG */
3772 while (*d && !isSPACE(*d))
3776 #ifdef ARG_ZERO_IS_SCRIPT
3777 if (ipathend > ipath) {
3779 * HP-UX (at least) sets argv[0] to the script name,
3780 * which makes $^X incorrect. And Digital UNIX and Linux,
3781 * at least, set argv[0] to the basename of the Perl
3782 * interpreter. So, having found "#!", we'll set it right.
3784 SV * const x = GvSV(gv_fetchpvs("\030", GV_ADD|GV_NOTQUAL,
3786 assert(SvPOK(x) || SvGMAGICAL(x));
3787 if (sv_eq(x, CopFILESV(PL_curcop))) {
3788 sv_setpvn(x, ipath, ipathend - ipath);
3794 const char *bstart = SvPV_const(CopFILESV(PL_curcop),blen);
3795 const char * const lstart = SvPV_const(x,llen);
3797 bstart += blen - llen;
3798 if (strnEQ(bstart, lstart, llen) && bstart[-1] == '/') {
3799 sv_setpvn(x, ipath, ipathend - ipath);
3804 TAINT_NOT; /* $^X is always tainted, but that's OK */
3806 #endif /* ARG_ZERO_IS_SCRIPT */
3811 d = instr(s,"perl -");
3813 d = instr(s,"perl");
3815 /* avoid getting into infinite loops when shebang
3816 * line contains "Perl" rather than "perl" */
3818 for (d = ipathend-4; d >= ipath; --d) {
3819 if ((*d == 'p' || *d == 'P')
3820 && !ibcmp(d, "perl", 4))
3830 #ifdef ALTERNATE_SHEBANG
3832 * If the ALTERNATE_SHEBANG on this system starts with a
3833 * character that can be part of a Perl expression, then if
3834 * we see it but not "perl", we're probably looking at the
3835 * start of Perl code, not a request to hand off to some
3836 * other interpreter. Similarly, if "perl" is there, but
3837 * not in the first 'word' of the line, we assume the line
3838 * contains the start of the Perl program.
3840 if (d && *s != '#') {
3841 const char *c = ipath;
3842 while (*c && !strchr("; \t\r\n\f\v#", *c))
3845 d = NULL; /* "perl" not in first word; ignore */
3847 *s = '#'; /* Don't try to parse shebang line */
3849 #endif /* ALTERNATE_SHEBANG */
3850 #ifndef MACOS_TRADITIONAL
3855 !instr(s,"indir") &&
3856 instr(PL_origargv[0],"perl"))
3863 while (s < PL_bufend && isSPACE(*s))
3865 if (s < PL_bufend) {
3866 Newxz(newargv,PL_origargc+3,char*);
3868 while (s < PL_bufend && !isSPACE(*s))
3871 Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*);
3874 newargv = PL_origargv;
3877 PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv));
3879 Perl_croak(aTHX_ "Can't exec %s", ipath);
3883 while (*d && !isSPACE(*d))
3885 while (SPACE_OR_TAB(*d))
3889 const bool switches_done = PL_doswitches;
3890 const U32 oldpdb = PL_perldb;
3891 const bool oldn = PL_minus_n;
3892 const bool oldp = PL_minus_p;
3895 if (*d == 'M' || *d == 'm' || *d == 'C') {
3896 const char * const m = d;
3897 while (*d && !isSPACE(*d))
3899 Perl_croak(aTHX_ "Too late for \"-%.*s\" option",
3902 d = moreswitches(d);
3904 if (PL_doswitches && !switches_done) {
3905 int argc = PL_origargc;
3906 char **argv = PL_origargv;
3909 } while (argc && argv[0][0] == '-' && argv[0][1]);
3910 init_argv_symbols(argc,argv);
3912 if ((PERLDB_LINE && !oldpdb) ||
3913 ((PL_minus_n || PL_minus_p) && !(oldn || oldp)))
3914 /* if we have already added "LINE: while (<>) {",
3915 we must not do it again */
3917 sv_setpvn(PL_linestr, "", 0);
3918 PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr);
3919 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
3920 PL_last_lop = PL_last_uni = NULL;
3921 PL_preambled = FALSE;
3923 (void)gv_fetchfile(PL_origfilename);
3930 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3932 PL_lex_state = LEX_FORMLINE;
3937 #ifdef PERL_STRICT_CR
3938 Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r');
3940 "\t(Maybe you didn't strip carriage returns after a network transfer?)\n");
3942 case ' ': case '\t': case '\f': case 013:
3943 #ifdef MACOS_TRADITIONAL
3947 PL_realtokenstart = -1;
3949 PL_thiswhite = newSVpvs("");
3950 sv_catpvn(PL_thiswhite, s, 1);
3957 PL_realtokenstart = -1;
3961 if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) {
3962 if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) {
3963 /* handle eval qq[#line 1 "foo"\n ...] */
3964 CopLINE_dec(PL_curcop);
3967 if (PL_madskills && !PL_lex_formbrack && !PL_in_eval) {
3969 if (!PL_in_eval || PL_rsfp)
3974 while (d < PL_bufend && *d != '\n')
3978 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
3979 Perl_croak(aTHX_ "panic: input overflow");
3982 PL_thiswhite = newSVpvn(s, d - s);
3987 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
3989 PL_lex_state = LEX_FORMLINE;
3995 if (PL_madskills && CopLINE(PL_curcop) >= 1 && !PL_lex_formbrack) {
3996 if (CopLINE(PL_curcop) == 1 && s[0] == '#' && s[1] == '!') {
3999 TOKEN(PEG); /* make sure any #! line is accessible */
4004 /* if (PL_madskills && PL_lex_formbrack) { */
4006 while (d < PL_bufend && *d != '\n')
4010 else if (d > PL_bufend) /* Found by Ilya: feed random input to Perl. */
4011 Perl_croak(aTHX_ "panic: input overflow");
4012 if (PL_madskills && CopLINE(PL_curcop) >= 1) {
4014 PL_thiswhite = newSVpvs("");
4015 if (CopLINE(PL_curcop) == 1) {
4016 sv_setpvn(PL_thiswhite, "", 0);
4019 sv_catpvn(PL_thiswhite, s, d - s);
4033 if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) {
4041 while (s < PL_bufend && SPACE_OR_TAB(*s))
4044 if (strnEQ(s,"=>",2)) {
4045 s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE);
4046 DEBUG_T( { printbuf("### Saw unary minus before =>, forcing word %s\n", s); } );
4047 OPERATOR('-'); /* unary minus */
4049 PL_last_uni = PL_oldbufptr;
4051 case 'r': ftst = OP_FTEREAD; break;
4052 case 'w': ftst = OP_FTEWRITE; break;
4053 case 'x': ftst = OP_FTEEXEC; break;
4054 case 'o': ftst = OP_FTEOWNED; break;
4055 case 'R': ftst = OP_FTRREAD; break;
4056 case 'W': ftst = OP_FTRWRITE; break;
4057 case 'X': ftst = OP_FTREXEC; break;
4058 case 'O': ftst = OP_FTROWNED; break;
4059 case 'e': ftst = OP_FTIS; break;
4060 case 'z': ftst = OP_FTZERO; break;
4061 case 's': ftst = OP_FTSIZE; break;
4062 case 'f': ftst = OP_FTFILE; break;
4063 case 'd': ftst = OP_FTDIR; break;
4064 case 'l': ftst = OP_FTLINK; break;
4065 case 'p': ftst = OP_FTPIPE; break;
4066 case 'S': ftst = OP_FTSOCK; break;
4067 case 'u': ftst = OP_FTSUID; break;
4068 case 'g': ftst = OP_FTSGID; break;
4069 case 'k': ftst = OP_FTSVTX; break;
4070 case 'b': ftst = OP_FTBLK; break;
4071 case 'c': ftst = OP_FTCHR; break;
4072 case 't': ftst = OP_FTTTY; break;
4073 case 'T': ftst = OP_FTTEXT; break;
4074 case 'B': ftst = OP_FTBINARY; break;
4075 case 'M': case 'A': case 'C':
4076 gv_fetchpvs("\024", GV_ADD|GV_NOTQUAL, SVt_PV);
4078 case 'M': ftst = OP_FTMTIME; break;
4079 case 'A': ftst = OP_FTATIME; break;
4080 case 'C': ftst = OP_FTCTIME; break;
4088 PL_last_lop_op = (OPCODE)ftst;
4089 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4090 "### Saw file test %c\n", (int)tmp);
4095 /* Assume it was a minus followed by a one-letter named
4096 * subroutine call (or a -bareword), then. */
4097 DEBUG_T( { PerlIO_printf(Perl_debug_log,
4098 "### '-%c' looked like a file test but was not\n",
4105 const char tmp = *s++;
4108 if (PL_expect == XOPERATOR)
4113 else if (*s == '>') {
4116 if (isIDFIRST_lazy_if(s,UTF)) {
4117 s = force_word(s,METHOD,FALSE,TRUE,FALSE);
4125 if (PL_expect == XOPERATOR)
4128 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4130 OPERATOR('-'); /* unary minus */
4136 const char tmp = *s++;
4139 if (PL_expect == XOPERATOR)
4144 if (PL_expect == XOPERATOR)
4147 if (isSPACE(*s) || !isSPACE(*PL_bufptr))
4154 if (PL_expect != XOPERATOR) {
4155 s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4156 PL_expect = XOPERATOR;
4157 force_ident(PL_tokenbuf, '*');
4170 if (PL_expect == XOPERATOR) {
4174 PL_tokenbuf[0] = '%';
4175 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4176 sizeof PL_tokenbuf - 1, FALSE);
4177 if (!PL_tokenbuf[1]) {
4180 PL_pending_ident = '%';
4191 && (PL_expect == XOPERATOR || PL_expect == XTERMORDORDOR))
4198 const char tmp = *s++;
4204 goto just_a_word_zero_gv;
4207 switch (PL_expect) {
4213 if (!PL_in_my || PL_lex_state != LEX_NORMAL)
4215 PL_bufptr = s; /* update in case we back off */
4221 PL_expect = XTERMBLOCK;
4224 stuffstart = s - SvPVX(PL_linestr) - 1;
4228 while (isIDFIRST_lazy_if(s,UTF)) {
4231 d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
4232 if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len, 0))) {
4233 if (tmp < 0) tmp = -tmp;
4247 sv = newSVpvn(s, len);
4249 d = scan_str(d,TRUE,TRUE);
4251 /* MUST advance bufptr here to avoid bogus
4252 "at end of line" context messages from yyerror().
4254 PL_bufptr = s + len;
4255 yyerror("Unterminated attribute parameter in attribute list");
4259 return REPORT(0); /* EOF indicator */
4263 sv_catsv(sv, PL_lex_stuff);
4264 attrs = append_elem(OP_LIST, attrs,
4265 newSVOP(OP_CONST, 0, sv));
4266 SvREFCNT_dec(PL_lex_stuff);
4267 PL_lex_stuff = NULL;
4270 if (len == 6 && strnEQ(SvPVX(sv), "unique", len)) {
4272 if (PL_in_my == KEY_our) {
4274 GvUNIQUE_on(cGVOPx_gv(yylval.opval));
4276 /* skip to avoid loading attributes.pm */
4278 deprecate(":unique");
4281 Perl_croak(aTHX_ "The 'unique' attribute may only be applied to 'our' variables");
4284 /* NOTE: any CV attrs applied here need to be part of
4285 the CVf_BUILTIN_ATTRS define in cv.h! */
4286 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "lvalue", len)) {
4288 CvLVALUE_on(PL_compcv);
4290 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "locked", len)) {
4292 CvLOCKED_on(PL_compcv);
4294 else if (!PL_in_my && len == 6 && strnEQ(SvPVX(sv), "method", len)) {
4296 CvMETHOD_on(PL_compcv);
4298 /* After we've set the flags, it could be argued that
4299 we don't need to do the attributes.pm-based setting
4300 process, and shouldn't bother appending recognized
4301 flags. To experiment with that, uncomment the
4302 following "else". (Note that's already been
4303 uncommented. That keeps the above-applied built-in
4304 attributes from being intercepted (and possibly
4305 rejected) by a package's attribute routines, but is
4306 justified by the performance win for the common case
4307 of applying only built-in attributes.) */
4309 attrs = append_elem(OP_LIST, attrs,
4310 newSVOP(OP_CONST, 0,
4314 if (*s == ':' && s[1] != ':')
4317 break; /* require real whitespace or :'s */
4318 /* XXX losing whitespace on sequential attributes here */
4322 = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */
4323 if (*s != ';' && *s != '}' && *s != tmp
4324 && (tmp != '=' || *s != ')')) {
4325 const char q = ((*s == '\'') ? '"' : '\'');
4326 /* If here for an expression, and parsed no attrs, back
4328 if (tmp == '=' && !attrs) {
4332 /* MUST advance bufptr here to avoid bogus "at end of line"
4333 context messages from yyerror().
4336 yyerror( (const char *)
4338 ? Perl_form(aTHX_ "Invalid separator character "
4339 "%c%c%c in attribute list", q, *s, q)
4340 : "Unterminated attribute list" ) );
4348 start_force(PL_curforce);
4349 NEXTVAL_NEXTTOKE.opval = attrs;
4350 CURMAD('_', PL_nextwhite);
4355 PL_thistoken = newSVpvn(SvPVX(PL_linestr) + stuffstart,
4356 (s - SvPVX(PL_linestr)) - stuffstart);
4364 if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr)
4365 PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */
4373 const char tmp = *s++;
4378 const char tmp = *s++;
4386 if (PL_lex_brackets <= 0)
4387 yyerror("Unmatched right square bracket");
4390 if (PL_lex_state == LEX_INTERPNORMAL) {
4391 if (PL_lex_brackets == 0) {
4392 if (*s == '-' && s[1] == '>')
4393 PL_lex_state = LEX_INTERPENDMAYBE;
4394 else if (*s != '[' && *s != '{')
4395 PL_lex_state = LEX_INTERPEND;
4402 if (PL_lex_brackets > 100) {
4403 Renew(PL_lex_brackstack, PL_lex_brackets + 10, char);
4405 switch (PL_expect) {
4407 if (PL_lex_formbrack) {
4411 if (PL_oldoldbufptr == PL_last_lop)
4412 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4414 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4415 OPERATOR(HASHBRACK);
4417 while (s < PL_bufend && SPACE_OR_TAB(*s))
4420 PL_tokenbuf[0] = '\0';
4421 if (d < PL_bufend && *d == '-') {
4422 PL_tokenbuf[0] = '-';
4424 while (d < PL_bufend && SPACE_OR_TAB(*d))
4427 if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) {
4428 d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1,
4430 while (d < PL_bufend && SPACE_OR_TAB(*d))
4433 const char minus = (PL_tokenbuf[0] == '-');
4434 s = force_word(s + minus, WORD, FALSE, TRUE, FALSE);
4442 PL_lex_brackstack[PL_lex_brackets++] = XSTATE;
4447 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4452 if (PL_oldoldbufptr == PL_last_lop)
4453 PL_lex_brackstack[PL_lex_brackets++] = XTERM;
4455 PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR;
4458 if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) {
4460 /* This hack is to get the ${} in the message. */
4462 yyerror("syntax error");
4465 OPERATOR(HASHBRACK);
4467 /* This hack serves to disambiguate a pair of curlies
4468 * as being a block or an anon hash. Normally, expectation
4469 * determines that, but in cases where we're not in a
4470 * position to expect anything in particular (like inside
4471 * eval"") we have to resolve the ambiguity. This code
4472 * covers the case where the first term in the curlies is a
4473 * quoted string. Most other cases need to be explicitly
4474 * disambiguated by prepending a "+" before the opening
4475 * curly in order to force resolution as an anon hash.
4477 * XXX should probably propagate the outer expectation
4478 * into eval"" to rely less on this hack, but that could
4479 * potentially break current behavior of eval"".
4483 if (*s == '\'' || *s == '"' || *s == '`') {
4484 /* common case: get past first string, handling escapes */
4485 for (t++; t < PL_bufend && *t != *s;)
4486 if (*t++ == '\\' && (*t == '\\' || *t == *s))
4490 else if (*s == 'q') {
4493 || ((*t == 'q' || *t == 'x') && ++t < PL_bufend
4496 /* skip q//-like construct */
4498 char open, close, term;
4501 while (t < PL_bufend && isSPACE(*t))
4503 /* check for q => */
4504 if (t+1 < PL_bufend && t[0] == '=' && t[1] == '>') {
4505 OPERATOR(HASHBRACK);
4509 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
4513 for (t++; t < PL_bufend; t++) {
4514 if (*t == '\\' && t+1 < PL_bufend && open != '\\')
4516 else if (*t == open)
4520 for (t++; t < PL_bufend; t++) {
4521 if (*t == '\\' && t+1 < PL_bufend)
4523 else if (*t == close && --brackets <= 0)
4525 else if (*t == open)
4532 /* skip plain q word */
4533 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4536 else if (isALNUM_lazy_if(t,UTF)) {
4538 while (t < PL_bufend && isALNUM_lazy_if(t,UTF))
4541 while (t < PL_bufend && isSPACE(*t))
4543 /* if comma follows first term, call it an anon hash */
4544 /* XXX it could be a comma expression with loop modifiers */
4545 if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s)))
4546 || (*t == '=' && t[1] == '>')))
4547 OPERATOR(HASHBRACK);
4548 if (PL_expect == XREF)
4551 PL_lex_brackstack[PL_lex_brackets-1] = XSTATE;
4557 yylval.ival = CopLINE(PL_curcop);
4558 if (isSPACE(*s) || *s == '#')
4559 PL_copline = NOLINE; /* invalidate current command line number */
4564 if (PL_lex_brackets <= 0)
4565 yyerror("Unmatched right curly bracket");
4567 PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets];
4568 if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL)
4569 PL_lex_formbrack = 0;
4570 if (PL_lex_state == LEX_INTERPNORMAL) {
4571 if (PL_lex_brackets == 0) {
4572 if (PL_expect & XFAKEBRACK) {
4573 PL_expect &= XENUMMASK;
4574 PL_lex_state = LEX_INTERPEND;
4579 PL_thiswhite = newSVpvs("");
4580 sv_catpvn(PL_thiswhite,"}",1);
4583 return yylex(); /* ignore fake brackets */
4585 if (*s == '-' && s[1] == '>')
4586 PL_lex_state = LEX_INTERPENDMAYBE;
4587 else if (*s != '[' && *s != '{')
4588 PL_lex_state = LEX_INTERPEND;
4591 if (PL_expect & XFAKEBRACK) {
4592 PL_expect &= XENUMMASK;
4594 return yylex(); /* ignore fake brackets */
4596 start_force(PL_curforce);
4598 curmad('X', newSVpvn(s-1,1));
4599 CURMAD('_', PL_thiswhite);
4604 PL_thistoken = newSVpvs("");
4612 if (PL_expect == XOPERATOR) {
4613 if (PL_bufptr == PL_linestart && ckWARN(WARN_SEMICOLON)
4614 && isIDFIRST_lazy_if(s,UTF))
4616 CopLINE_dec(PL_curcop);
4617 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
4618 CopLINE_inc(PL_curcop);
4623 s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
4625 PL_expect = XOPERATOR;
4626 force_ident(PL_tokenbuf, '&');
4630 yylval.ival = (OPpENTERSUB_AMPER<<8);
4642 const char tmp = *s++;
4649 if (tmp && isSPACE(*s) && ckWARN(WARN_SYNTAX)
4650 && strchr("+-*/%.^&|<",tmp))
4651 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4652 "Reversed %c= operator",(int)tmp);
4654 if (PL_expect == XSTATE && isALPHA(tmp) &&
4655 (s == PL_linestart+1 || s[-2] == '\n') )
4657 if (PL_in_eval && !PL_rsfp) {
4662 if (strnEQ(s,"=cut",4)) {
4678 PL_thiswhite = newSVpvs("");
4679 sv_catpvn(PL_thiswhite, PL_linestart,
4680 PL_bufend - PL_linestart);
4684 PL_doextract = TRUE;
4688 if (PL_lex_brackets < PL_lex_formbrack) {
4690 #ifdef PERL_STRICT_CR
4691 while (SPACE_OR_TAB(*t))
4693 while (SPACE_OR_TAB(*t) || *t == '\r')
4696 if (*t == '\n' || *t == '#') {
4707 const char tmp = *s++;
4709 /* was this !=~ where !~ was meant?
4710 * warn on m:!=~\s+([/?]|[msy]\W|tr\W): */
4712 if (*s == '~' && ckWARN(WARN_SYNTAX)) {
4713 const char *t = s+1;
4715 while (t < PL_bufend && isSPACE(*t))
4718 if (*t == '/' || *t == '?' ||
4719 ((*t == 'm' || *t == 's' || *t == 'y')
4720 && !isALNUM(t[1])) ||
4721 (*t == 't' && t[1] == 'r' && !isALNUM(t[2])))
4722 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4723 "!=~ should be !~");
4733 if (PL_expect != XOPERATOR) {
4734 if (s[1] != '<' && !strchr(s,'>'))
4737 s = scan_heredoc(s);
4739 s = scan_inputsymbol(s);
4740 TERM(sublex_start());
4746 SHop(OP_LEFT_SHIFT);
4760 const char tmp = *s++;
4762 SHop(OP_RIGHT_SHIFT);
4763 else if (tmp == '=')
4772 if (PL_expect == XOPERATOR) {
4773 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
4775 deprecate_old(commaless_variable_list);
4776 return REPORT(','); /* grandfather non-comma-format format */
4780 if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) {
4781 PL_tokenbuf[0] = '@';
4782 s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1,
4783 sizeof PL_tokenbuf - 1, FALSE);
4784 if (PL_expect == XOPERATOR)
4785 no_op("Array length", s);
4786 if (!PL_tokenbuf[1])
4788 PL_expect = XOPERATOR;
4789 PL_pending_ident = '#';
4793 PL_tokenbuf[0] = '$';
4794 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1,
4795 sizeof PL_tokenbuf - 1, FALSE);
4796 if (PL_expect == XOPERATOR)
4798 if (!PL_tokenbuf[1]) {
4800 yyerror("Final $ should be \\$ or $name");
4804 /* This kludge not intended to be bulletproof. */
4805 if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) {
4806 yylval.opval = newSVOP(OP_CONST, 0,
4807 newSViv(CopARYBASE_get(&PL_compiling)));
4808 yylval.opval->op_private = OPpCONST_ARYBASE;
4814 const char tmp = *s;
4815 if (PL_lex_state == LEX_NORMAL)
4818 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop)
4819 && intuit_more(s)) {
4821 PL_tokenbuf[0] = '@';
4822 if (ckWARN(WARN_SYNTAX)) {
4825 while (isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$')
4828 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4829 while (t < PL_bufend && *t != ']')
4831 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4832 "Multidimensional syntax %.*s not supported",
4833 (int)((t - PL_bufptr) + 1), PL_bufptr);
4837 else if (*s == '{') {
4839 PL_tokenbuf[0] = '%';
4840 if (strEQ(PL_tokenbuf+1, "SIG") && ckWARN(WARN_SYNTAX)
4841 && (t = strchr(s, '}')) && (t = strchr(t, '=')))
4843 char tmpbuf[sizeof PL_tokenbuf];
4846 } while (isSPACE(*t));
4847 if (isIDFIRST_lazy_if(t,UTF)) {
4849 t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE,
4853 if (*t == ';' && get_cvn_flags(tmpbuf, len, 0))
4854 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4855 "You need to quote \"%s\"",
4862 PL_expect = XOPERATOR;
4863 if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) {
4864 const bool islop = (PL_last_lop == PL_oldoldbufptr);
4865 if (!islop || PL_last_lop_op == OP_GREPSTART)
4866 PL_expect = XOPERATOR;
4867 else if (strchr("$@\"'`q", *s))
4868 PL_expect = XTERM; /* e.g. print $fh "foo" */
4869 else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF))
4870 PL_expect = XTERM; /* e.g. print $fh &sub */
4871 else if (isIDFIRST_lazy_if(s,UTF)) {
4872 char tmpbuf[sizeof PL_tokenbuf];
4874 scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
4875 if ((t2 = keyword(tmpbuf, len, 0))) {
4876 /* binary operators exclude handle interpretations */
4888 PL_expect = XTERM; /* e.g. print $fh length() */
4893 PL_expect = XTERM; /* e.g. print $fh subr() */
4896 else if (isDIGIT(*s))
4897 PL_expect = XTERM; /* e.g. print $fh 3 */
4898 else if (*s == '.' && isDIGIT(s[1]))
4899 PL_expect = XTERM; /* e.g. print $fh .3 */
4900 else if ((*s == '?' || *s == '-' || *s == '+')
4901 && !isSPACE(s[1]) && s[1] != '=')
4902 PL_expect = XTERM; /* e.g. print $fh -1 */
4903 else if (*s == '/' && !isSPACE(s[1]) && s[1] != '='
4905 PL_expect = XTERM; /* e.g. print $fh /.../
4906 XXX except DORDOR operator
4908 else if (*s == '<' && s[1] == '<' && !isSPACE(s[2])
4910 PL_expect = XTERM; /* print $fh <<"EOF" */
4913 PL_pending_ident = '$';
4917 if (PL_expect == XOPERATOR)
4919 PL_tokenbuf[0] = '@';
4920 s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE);
4921 if (!PL_tokenbuf[1]) {
4924 if (PL_lex_state == LEX_NORMAL)
4926 if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) {
4928 PL_tokenbuf[0] = '%';
4930 /* Warn about @ where they meant $. */
4931 if (*s == '[' || *s == '{') {
4932 if (ckWARN(WARN_SYNTAX)) {
4933 const char *t = s + 1;
4934 while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t)))
4936 if (*t == '}' || *t == ']') {
4938 PL_bufptr = PEEKSPACE(PL_bufptr); /* XXX can realloc */
4939 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
4940 "Scalar value %.*s better written as $%.*s",
4941 (int)(t-PL_bufptr), PL_bufptr,
4942 (int)(t-PL_bufptr-1), PL_bufptr+1);
4947 PL_pending_ident = '@';
4950 case '/': /* may be division, defined-or, or pattern */
4951 if (PL_expect == XTERMORDORDOR && s[1] == '/') {
4955 case '?': /* may either be conditional or pattern */
4956 if(PL_expect == XOPERATOR) {
4964 /* A // operator. */
4974 /* Disable warning on "study /blah/" */
4975 if (PL_oldoldbufptr == PL_last_uni
4976 && (*PL_last_uni != 's' || s - PL_last_uni < 5
4977 || memNE(PL_last_uni, "study", 5)
4978 || isALNUM_lazy_if(PL_last_uni+5,UTF)
4981 s = scan_pat(s,OP_MATCH);
4982 TERM(sublex_start());
4986 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack
4987 #ifdef PERL_STRICT_CR
4990 && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n'))
4992 && (s == PL_linestart || s[-1] == '\n') )
4994 PL_lex_formbrack = 0;
4998 if (PL_expect == XOPERATOR || !isDIGIT(s[1])) {
5004 yylval.ival = OPf_SPECIAL;
5010 if (PL_expect != XOPERATOR)
5015 case '0': case '1': case '2': case '3': case '4':
5016 case '5': case '6': case '7': case '8': case '9':
5017 s = scan_num(s, &yylval);
5018 DEBUG_T( { printbuf("### Saw number in %s\n", s); } );
5019 if (PL_expect == XOPERATOR)
5024 s = scan_str(s,!!PL_madskills,FALSE);
5025 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5026 if (PL_expect == XOPERATOR) {
5027 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5029 deprecate_old(commaless_variable_list);
5030 return REPORT(','); /* grandfather non-comma-format format */
5037 yylval.ival = OP_CONST;
5038 TERM(sublex_start());
5041 s = scan_str(s,!!PL_madskills,FALSE);
5042 DEBUG_T( { printbuf("### Saw string before %s\n", s); } );
5043 if (PL_expect == XOPERATOR) {
5044 if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) {
5046 deprecate_old(commaless_variable_list);
5047 return REPORT(','); /* grandfather non-comma-format format */
5054 yylval.ival = OP_CONST;
5055 /* FIXME. I think that this can be const if char *d is replaced by
5056 more localised variables. */
5057 for (d = SvPV(PL_lex_stuff, len); len; len--, d++) {
5058 if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) {
5059 yylval.ival = OP_STRINGIFY;
5063 TERM(sublex_start());
5066 s = scan_str(s,!!PL_madskills,FALSE);
5067 DEBUG_T( { printbuf("### Saw backtick string before %s\n", s); } );
5068 if (PL_expect == XOPERATOR)
5069 no_op("Backticks",s);
5072 readpipe_override();
5073 TERM(sublex_start());
5077 if (PL_lex_inwhat && isDIGIT(*s) && ckWARN(WARN_SYNTAX))
5078 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),"Can't use \\%c to mean $%c in expression",
5080 if (PL_expect == XOPERATOR)
5081 no_op("Backslash",s);
5085 if (isDIGIT(s[1]) && PL_expect != XOPERATOR) {
5086 char *start = s + 2;
5087 while (isDIGIT(*start) || *start == '_')
5089 if (*start == '.' && isDIGIT(start[1])) {
5090 s = scan_num(s, &yylval);
5093 /* avoid v123abc() or $h{v1}, allow C<print v10;> */
5094 else if (!isALPHA(*start) && (PL_expect == XTERM
5095 || PL_expect == XREF || PL_expect == XSTATE
5096 || PL_expect == XTERMORDORDOR)) {
5097 /* XXX Use gv_fetchpvn rather than stomping on a const string */
5098 const char c = *start;
5101 gv = gv_fetchpv(s, 0, SVt_PVCV);
5104 s = scan_num(s, &yylval);
5111 if (isDIGIT(s[1]) && PL_expect == XOPERATOR) {
5153 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5155 /* Some keywords can be followed by any delimiter, including ':' */
5156 tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) ||
5157 (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') ||
5158 (PL_tokenbuf[0] == 'q' &&
5159 strchr("qwxr", PL_tokenbuf[1])))));
5161 /* x::* is just a word, unless x is "CORE" */
5162 if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE"))
5166 while (d < PL_bufend && isSPACE(*d))
5167 d++; /* no comments skipped here, or s### is misparsed */
5169 /* Is this a label? */
5170 if (!tmp && PL_expect == XSTATE
5171 && d < PL_bufend && *d == ':' && *(d + 1) != ':') {
5173 yylval.pval = CopLABEL_alloc(PL_tokenbuf);
5178 /* Check for keywords */
5179 tmp = keyword(PL_tokenbuf, len, 0);
5181 /* Is this a word before a => operator? */
5182 if (*d == '=' && d[1] == '>') {
5185 = (OP*)newSVOP(OP_CONST, 0,
5186 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
5187 yylval.opval->op_private = OPpCONST_BARE;
5191 if (tmp < 0) { /* second-class keyword? */
5192 GV *ogv = NULL; /* override (winner) */
5193 GV *hgv = NULL; /* hidden (loser) */
5194 if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) {
5196 if ((gv = gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVCV)) &&
5199 if (GvIMPORTED_CV(gv))
5201 else if (! CvMETHOD(cv))
5205 (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) &&
5206 (gv = *gvp) && isGV_with_GP(gv) &&
5207 GvCVu(gv) && GvIMPORTED_CV(gv))
5214 tmp = 0; /* overridden by import or by GLOBAL */
5217 && -tmp==KEY_lock /* XXX generalizable kludge */
5220 tmp = 0; /* any sub overrides "weak" keyword */
5222 else { /* no override */
5224 if (tmp == KEY_dump && ckWARN(WARN_MISC)) {
5225 Perl_warner(aTHX_ packWARN(WARN_MISC),
5226 "dump() better written as CORE::dump()");
5230 if (hgv && tmp != KEY_x && tmp != KEY_CORE
5231 && ckWARN(WARN_AMBIGUOUS)) /* never ambiguous */
5232 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5233 "Ambiguous call resolved as CORE::%s(), %s",
5234 GvENAME(hgv), "qualify as such or use &");
5241 default: /* not a keyword */
5242 /* Trade off - by using this evil construction we can pull the
5243 variable gv into the block labelled keylookup. If not, then
5244 we have to give it function scope so that the goto from the
5245 earlier ':' case doesn't bypass the initialisation. */
5247 just_a_word_zero_gv:
5255 const char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]);
5258 SV *nextPL_nextwhite = 0;
5262 /* Get the rest if it looks like a package qualifier */
5264 if (*s == '\'' || (*s == ':' && s[1] == ':')) {
5266 s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len,
5269 Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf,
5270 *s == '\'' ? "'" : "::");
5275 if (PL_expect == XOPERATOR) {
5276 if (PL_bufptr == PL_linestart) {
5277 CopLINE_dec(PL_curcop);
5278 Perl_warner(aTHX_ packWARN(WARN_SEMICOLON), PL_warn_nosemi);
5279 CopLINE_inc(PL_curcop);
5282 no_op("Bareword",s);
5285 /* Look for a subroutine with this name in current package,
5286 unless name is "Foo::", in which case Foo is a bearword
5287 (and a package name). */
5289 if (len > 2 && !PL_madskills &&
5290 PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':')
5292 if (ckWARN(WARN_BAREWORD)
5293 && ! gv_fetchpvn_flags(PL_tokenbuf, len, 0, SVt_PVHV))
5294 Perl_warner(aTHX_ packWARN(WARN_BAREWORD),
5295 "Bareword \"%s\" refers to nonexistent package",
5298 PL_tokenbuf[len] = '\0';
5304 /* Mustn't actually add anything to a symbol table.
5305 But also don't want to "initialise" any placeholder
5306 constants that might already be there into full
5307 blown PVGVs with attached PVCV. */
5308 gv = gv_fetchpvn_flags(PL_tokenbuf, len,
5309 GV_NOADD_NOINIT, SVt_PVCV);
5314 /* if we saw a global override before, get the right name */
5317 sv = newSVpvs("CORE::GLOBAL::");
5318 sv_catpv(sv,PL_tokenbuf);
5321 /* If len is 0, newSVpv does strlen(), which is correct.
5322 If len is non-zero, then it will be the true length,
5323 and so the scalar will be created correctly. */
5324 sv = newSVpv(PL_tokenbuf,len);
5327 if (PL_madskills && !PL_thistoken) {
5328 char *start = SvPVX(PL_linestr) + PL_realtokenstart;
5329 PL_thistoken = newSVpv(start,s - start);
5330 PL_realtokenstart = s - SvPVX(PL_linestr);
5334 /* Presume this is going to be a bareword of some sort. */
5337 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
5338 yylval.opval->op_private = OPpCONST_BARE;
5339 /* UTF-8 package name? */
5340 if (UTF && !IN_BYTES &&
5341 is_utf8_string((U8*)SvPVX_const(sv), SvCUR(sv)))
5344 /* And if "Foo::", then that's what it certainly is. */
5349 /* Do the explicit type check so that we don't need to force
5350 the initialisation of the symbol table to have a real GV.
5351 Beware - gv may not really be a PVGV, cv may not really be
5352 a PVCV, (because of the space optimisations that gv_init
5353 understands) But they're true if for this symbol there is
5354 respectively a typeglob and a subroutine.
5356 cv = gv ? ((SvTYPE(gv) == SVt_PVGV)
5357 /* Real typeglob, so get the real subroutine: */
5359 /* A proxy for a subroutine in this package? */
5360 : SvOK(gv) ? (CV *) gv : NULL)
5363 /* See if it's the indirect object for a list operator. */
5365 if (PL_oldoldbufptr &&
5366 PL_oldoldbufptr < PL_bufptr &&
5367 (PL_oldoldbufptr == PL_last_lop
5368 || PL_oldoldbufptr == PL_last_uni) &&
5369 /* NO SKIPSPACE BEFORE HERE! */
5370 (PL_expect == XREF ||
5371 ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF))
5373 bool immediate_paren = *s == '(';
5375 /* (Now we can afford to cross potential line boundary.) */
5376 s = SKIPSPACE2(s,nextPL_nextwhite);
5378 PL_nextwhite = nextPL_nextwhite; /* assume no & deception */
5381 /* Two barewords in a row may indicate method call. */
5383 if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') &&
5384 (tmp = intuit_method(s, gv, cv)))
5387 /* If not a declared subroutine, it's an indirect object. */
5388 /* (But it's an indir obj regardless for sort.) */
5389 /* Also, if "_" follows a filetest operator, it's a bareword */
5392 ( !immediate_paren && (PL_last_lop_op == OP_SORT ||
5394 (PL_last_lop_op != OP_MAPSTART &&
5395 PL_last_lop_op != OP_GREPSTART))))
5396 || (PL_tokenbuf[0] == '_' && PL_tokenbuf[1] == '\0'
5397 && ((PL_opargs[PL_last_lop_op] & OA_CLASS_MASK) == OA_FILESTATOP))
5400 PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR;
5405 PL_expect = XOPERATOR;
5408 s = SKIPSPACE2(s,nextPL_nextwhite);
5409 PL_nextwhite = nextPL_nextwhite;
5414 /* Is this a word before a => operator? */
5415 if (*s == '=' && s[1] == '>' && !pkgname) {
5417 sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf);
5418 if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len))
5419 SvUTF8_on(((SVOP*)yylval.opval)->op_sv);
5423 /* If followed by a paren, it's certainly a subroutine. */
5428 while (SPACE_OR_TAB(*d))
5430 if (*d == ')' && (sv = gv_const_sv(gv))) {
5437 PL_nextwhite = PL_thiswhite;
5440 start_force(PL_curforce);
5442 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5443 PL_expect = XOPERATOR;
5446 PL_nextwhite = nextPL_nextwhite;
5447 curmad('X', PL_thistoken);
5448 PL_thistoken = newSVpvs("");
5456 /* If followed by var or block, call it a method (unless sub) */
5458 if ((*s == '$' || *s == '{') && (!gv || !cv)) {
5459 PL_last_lop = PL_oldbufptr;
5460 PL_last_lop_op = OP_METHOD;
5464 /* If followed by a bareword, see if it looks like indir obj. */
5467 && (isIDFIRST_lazy_if(s,UTF) || *s == '$')
5468 && (tmp = intuit_method(s, gv, cv)))
5471 /* Not a method, so call it a subroutine (if defined) */
5474 if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS))
5475 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5476 "Ambiguous use of -%s resolved as -&%s()",
5477 PL_tokenbuf, PL_tokenbuf);
5478 /* Check for a constant sub */
5479 if ((sv = gv_const_sv(gv))) {
5481 SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv);
5482 ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc_simple(sv);
5483 yylval.opval->op_private = 0;
5487 /* Resolve to GV now. */
5488 if (SvTYPE(gv) != SVt_PVGV) {
5489 gv = gv_fetchpv(PL_tokenbuf, 0, SVt_PVCV);
5490 assert (SvTYPE(gv) == SVt_PVGV);
5491 /* cv must have been some sort of placeholder, so
5492 now needs replacing with a real code reference. */
5496 op_free(yylval.opval);
5497 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5498 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5499 PL_last_lop = PL_oldbufptr;
5500 PL_last_lop_op = OP_ENTERSUB;
5501 /* Is there a prototype? */
5509 const char *proto = SvPV_const((SV*)cv, protolen);
5512 if ((*proto == '$' || *proto == '_') && proto[1] == '\0')
5514 while (*proto == ';')
5516 if (*proto == '&' && *s == '{') {
5517 sv_setpv(PL_subname,
5520 "__ANON__" : "__ANON__::__ANON__"));
5527 PL_nextwhite = PL_thiswhite;
5530 start_force(PL_curforce);
5531 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5534 PL_nextwhite = nextPL_nextwhite;
5535 curmad('X', PL_thistoken);
5536 PL_thistoken = newSVpvs("");
5543 /* Guess harder when madskills require "best effort". */
5544 if (PL_madskills && (!gv || !GvCVu(gv))) {
5545 int probable_sub = 0;
5546 if (strchr("\"'`$@%0123456789!*+{[<", *s))
5548 else if (isALPHA(*s)) {
5552 d = scan_word(d, tmpbuf, sizeof tmpbuf, TRUE, &tmplen);
5553 if (!keyword(tmpbuf, tmplen, 0))
5556 while (d < PL_bufend && isSPACE(*d))
5558 if (*d == '=' && d[1] == '>')
5563 gv = gv_fetchpv(PL_tokenbuf, GV_ADD, SVt_PVCV);
5564 op_free(yylval.opval);
5565 yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv));
5566 yylval.opval->op_private |= OPpENTERSUB_NOPAREN;
5567 PL_last_lop = PL_oldbufptr;
5568 PL_last_lop_op = OP_ENTERSUB;
5569 PL_nextwhite = PL_thiswhite;
5571 start_force(PL_curforce);
5572 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5574 PL_nextwhite = nextPL_nextwhite;
5575 curmad('X', PL_thistoken);
5576 PL_thistoken = newSVpvs("");
5581 NEXTVAL_NEXTTOKE.opval = yylval.opval;
5588 /* Call it a bare word */
5590 if (PL_hints & HINT_STRICT_SUBS)
5591 yylval.opval->op_private |= OPpCONST_STRICT;
5594 if (lastchar != '-') {
5595 if (ckWARN(WARN_RESERVED)) {
5599 if (!*d && !gv_stashpv(PL_tokenbuf, 0))
5600 Perl_warner(aTHX_ packWARN(WARN_RESERVED), PL_warn_reserved,
5607 if ((lastchar == '*' || lastchar == '%' || lastchar == '&')
5608 && ckWARN_d(WARN_AMBIGUOUS)) {
5609 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5610 "Operator or semicolon missing before %c%s",
5611 lastchar, PL_tokenbuf);
5612 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
5613 "Ambiguous use of %c resolved as operator %c",
5614 lastchar, lastchar);
5620 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5621 newSVpv(CopFILE(PL_curcop),0));
5625 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5626 Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop)));
5629 case KEY___PACKAGE__:
5630 yylval.opval = (OP*)newSVOP(OP_CONST, 0,
5632 ? newSVhek(HvNAME_HEK(PL_curstash))
5639 if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) {
5640 const char *pname = "main";
5641 if (PL_tokenbuf[2] == 'D')
5642 pname = HvNAME_get(PL_curstash ? PL_curstash : PL_defstash);
5643 gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), GV_ADD,
5647 GvIOp(gv) = newIO();
5648 IoIFP(GvIOp(gv)) = PL_rsfp;
5649 #if defined(HAS_FCNTL) && defined(F_SETFD)
5651 const int fd = PerlIO_fileno(PL_rsfp);
5652 fcntl(fd,F_SETFD,fd >= 3);
5655 /* Mark this internal pseudo-handle as clean */
5656 IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT;
5658 IoTYPE(GvIOp(gv)) = IoTYPE_PIPE;
5659 else if ((PerlIO*)PL_rsfp == PerlIO_stdin())
5660 IoTYPE(GvIOp(gv)) = IoTYPE_STD;
5662 IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY;
5663 #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS)
5664 /* if the script was opened in binmode, we need to revert
5665 * it to text mode for compatibility; but only iff it has CRs
5666 * XXX this is a questionable hack at best. */
5667 if (PL_bufend-PL_bufptr > 2
5668 && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r')
5671 if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) {
5672 loc = PerlIO_tell(PL_rsfp);
5673 (void)PerlIO_seek(PL_rsfp, 0L, 0);
5676 if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) {
5678 if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) {
5679 #endif /* NETWARE */
5680 #ifdef PERLIO_IS_STDIO /* really? */
5681 # if defined(__BORLANDC__)
5682 /* XXX see note in do_binmode() */
5683 ((FILE*)PL_rsfp)->flags &= ~_F_BIN;
5687 PerlIO_seek(PL_rsfp, loc, 0);
5691 #ifdef PERLIO_LAYERS
5694 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8");
5695 else if (PL_encoding) {
5702 XPUSHs(PL_encoding);
5704 call_method("name", G_SCALAR);
5708 PerlIO_apply_layers(aTHX_ PL_rsfp, NULL,
5709 Perl_form(aTHX_ ":encoding(%"SVf")",
5718 if (PL_realtokenstart >= 0) {
5719 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
5721 PL_endwhite = newSVpvs("");
5722 sv_catsv(PL_endwhite, PL_thiswhite);
5724 sv_catpvn(PL_endwhite, tstart, PL_bufend - tstart);
5725 PL_realtokenstart = -1;
5727 while ((s = filter_gets(PL_endwhite, PL_rsfp,
5728 SvCUR(PL_endwhite))) != Nullch) ;
5743 if (PL_expect == XSTATE) {
5750 if (*s == ':' && s[1] == ':') {
5753 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len);
5754 if (!(tmp = keyword(PL_tokenbuf, len, 0)))
5755 Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf);
5758 else if (tmp == KEY_require || tmp == KEY_do)
5759 /* that's a way to remember we saw "CORE::" */
5772 LOP(OP_ACCEPT,XTERM);
5778 LOP(OP_ATAN2,XTERM);
5784 LOP(OP_BINMODE,XTERM);
5787 LOP(OP_BLESS,XTERM);
5796 /* When 'use switch' is in effect, continue has a dual
5797 life as a control operator. */
5799 if (!FEATURE_IS_ENABLED("switch"))
5802 /* We have to disambiguate the two senses of
5803 "continue". If the next token is a '{' then
5804 treat it as the start of a continue block;
5805 otherwise treat it as a control operator.
5817 (void)gv_fetchpvs("ENV", GV_ADD|GV_NOTQUAL, SVt_PVHV);
5834 if (!PL_cryptseen) {
5835 PL_cryptseen = TRUE;
5839 LOP(OP_CRYPT,XTERM);
5842 LOP(OP_CHMOD,XTERM);
5845 LOP(OP_CHOWN,XTERM);
5848 LOP(OP_CONNECT,XTERM);
5867 s = force_word(s,WORD,TRUE,TRUE,FALSE);
5868 if (orig_keyword == KEY_do) {
5877 PL_hints |= HINT_BLOCK_SCOPE;
5887 gv_fetchpvs("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV);
5888 LOP(OP_DBMOPEN,XTERM);
5894 s = force_word(s,WORD,TRUE,FALSE,FALSE);
5901 yylval.ival = CopLINE(PL_curcop);
5917 PL_expect = (*s == '{') ? XTERMBLOCK : XTERM;
5918 UNIBRACK(OP_ENTEREVAL);
5932 case KEY_endhostent:
5938 case KEY_endservent:
5941 case KEY_endprotoent:
5952 yylval.ival = CopLINE(PL_curcop);
5954 if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) {
5957 int soff = s - SvPVX(PL_linestr); /* for skipspace realloc */
5960 if ((PL_bufend - p) >= 3 &&
5961 strnEQ(p, "my", 2) && isSPACE(*(p + 2)))
5963 else if ((PL_bufend - p) >= 4 &&
5964 strnEQ(p, "our", 3) && isSPACE(*(p + 3)))
5967 if (isIDFIRST_lazy_if(p,UTF)) {
5968 p = scan_ident(p, PL_bufend,
5969 PL_tokenbuf, sizeof PL_tokenbuf, TRUE);
5973 Perl_croak(aTHX_ "Missing $ on loop variable");
5975 s = SvPVX(PL_linestr) + soff;
5981 LOP(OP_FORMLINE,XTERM);
5987 LOP(OP_FCNTL,XTERM);
5993 LOP(OP_FLOCK,XTERM);
6002 LOP(OP_GREPSTART, XREF);
6005 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6020 case KEY_getpriority:
6021 LOP(OP_GETPRIORITY,XTERM);
6023 case KEY_getprotobyname:
6026 case KEY_getprotobynumber:
6027 LOP(OP_GPBYNUMBER,XTERM);
6029 case KEY_getprotoent:
6041 case KEY_getpeername:
6042 UNI(OP_GETPEERNAME);
6044 case KEY_gethostbyname:
6047 case KEY_gethostbyaddr:
6048 LOP(OP_GHBYADDR,XTERM);
6050 case KEY_gethostent:
6053 case KEY_getnetbyname:
6056 case KEY_getnetbyaddr:
6057 LOP(OP_GNBYADDR,XTERM);
6062 case KEY_getservbyname:
6063 LOP(OP_GSBYNAME,XTERM);
6065 case KEY_getservbyport:
6066 LOP(OP_GSBYPORT,XTERM);
6068 case KEY_getservent:
6071 case KEY_getsockname:
6072 UNI(OP_GETSOCKNAME);
6074 case KEY_getsockopt:
6075 LOP(OP_GSOCKOPT,XTERM);
6090 yylval.ival = CopLINE(PL_curcop);
6100 yylval.ival = CopLINE(PL_curcop);
6104 LOP(OP_INDEX,XTERM);
6110 LOP(OP_IOCTL,XTERM);
6122 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6154 LOP(OP_LISTEN,XTERM);
6163 s = scan_pat(s,OP_MATCH);
6164 TERM(sublex_start());
6167 LOP(OP_MAPSTART, XREF);
6170 LOP(OP_MKDIR,XTERM);
6173 LOP(OP_MSGCTL,XTERM);
6176 LOP(OP_MSGGET,XTERM);
6179 LOP(OP_MSGRCV,XTERM);
6182 LOP(OP_MSGSND,XTERM);
6187 PL_in_my = (U16)tmp;
6189 if (isIDFIRST_lazy_if(s,UTF)) {
6193 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len);
6194 if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3))
6196 PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len);
6197 if (!PL_in_my_stash) {
6200 my_snprintf(tmpbuf, sizeof(tmpbuf), "No such class %.1000s", PL_tokenbuf);
6204 if (PL_madskills) { /* just add type to declarator token */
6205 sv_catsv(PL_thistoken, PL_nextwhite);
6207 sv_catpvn(PL_thistoken, start, s - start);
6215 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6222 s = tokenize_use(0, s);
6226 if (*s == '(' || (s = SKIPSPACE1(s), *s == '('))
6233 if (isIDFIRST_lazy_if(s,UTF)) {
6235 for (d = s; isALNUM_lazy_if(d,UTF);)
6237 for (t=d; isSPACE(*t);)
6239 if ( *t && strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)
6241 && !(t[0] == '=' && t[1] == '>')
6243 int parms_len = (int)(d-s);
6244 Perl_warner(aTHX_ packWARN(WARN_PRECEDENCE),
6245 "Precedence problem: open %.*s should be open(%.*s)",
6246 parms_len, s, parms_len, s);
6252 yylval.ival = OP_OR;
6262 LOP(OP_OPEN_DIR,XTERM);
6265 checkcomma(s,PL_tokenbuf,"filehandle");
6269 checkcomma(s,PL_tokenbuf,"filehandle");
6288 s = force_word(s,WORD,FALSE,TRUE,FALSE);
6292 LOP(OP_PIPE_OP,XTERM);
6295 s = scan_str(s,!!PL_madskills,FALSE);
6298 yylval.ival = OP_CONST;
6299 TERM(sublex_start());
6305 s = scan_str(s,!!PL_madskills,FALSE);
6308 PL_expect = XOPERATOR;
6310 if (SvCUR(PL_lex_stuff)) {
6313 d = SvPV_force(PL_lex_stuff, len);
6315 for (; isSPACE(*d) && len; --len, ++d)
6320 if (!warned && ckWARN(WARN_QW)) {
6321 for (; !isSPACE(*d) && len; --len, ++d) {
6323 Perl_warner(aTHX_ packWARN(WARN_QW),
6324 "Possible attempt to separate words with commas");
6327 else if (*d == '#') {
6328 Perl_warner(aTHX_ packWARN(WARN_QW),
6329 "Possible attempt to put comments in qw() list");
6335 for (; !isSPACE(*d) && len; --len, ++d)
6338 sv = newSVpvn(b, d-b);
6339 if (DO_UTF8(PL_lex_stuff))
6341 words = append_elem(OP_LIST, words,
6342 newSVOP(OP_CONST, 0, tokeq(sv)));
6346 start_force(PL_curforce);
6347 NEXTVAL_NEXTTOKE.opval = words;
6352 SvREFCNT_dec(PL_lex_stuff);
6353 PL_lex_stuff = NULL;
6359 s = scan_str(s,!!PL_madskills,FALSE);
6362 yylval.ival = OP_STRINGIFY;
6363 if (SvIVX(PL_lex_stuff) == '\'')
6364 SvIV_set(PL_lex_stuff, 0); /* qq'$foo' should intepolate */
6365 TERM(sublex_start());
6368 s = scan_pat(s,OP_QR);
6369 TERM(sublex_start());
6372 s = scan_str(s,!!PL_madskills,FALSE);
6375 readpipe_override();
6376 TERM(sublex_start());
6384 s = force_version(s, FALSE);
6386 else if (*s != 'v' || !isDIGIT(s[1])
6387 || (s = force_version(s, TRUE), *s == 'v'))
6389 *PL_tokenbuf = '\0';
6390 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6391 if (isIDFIRST_lazy_if(PL_tokenbuf,UTF))
6392 gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), GV_ADD);
6394 yyerror("<> should be quotes");
6396 if (orig_keyword == KEY_require) {
6404 PL_last_uni = PL_oldbufptr;
6405 PL_last_lop_op = OP_REQUIRE;
6407 return REPORT( (int)REQUIRE );
6413 s = force_word(s,WORD,TRUE,FALSE,FALSE);
6417 LOP(OP_RENAME,XTERM);
6426 LOP(OP_RINDEX,XTERM);
6435 UNIDOR(OP_READLINE);
6438 UNIDOR(OP_BACKTICK);
6447 LOP(OP_REVERSE,XTERM);
6450 UNIDOR(OP_READLINK);
6458 TERM(sublex_start());
6460 TOKEN(1); /* force error */
6463 checkcomma(s,PL_tokenbuf,"filehandle");
6473 LOP(OP_SELECT,XTERM);
6479 LOP(OP_SEMCTL,XTERM);
6482 LOP(OP_SEMGET,XTERM);
6485 LOP(OP_SEMOP,XTERM);
6491 LOP(OP_SETPGRP,XTERM);
6493 case KEY_setpriority:
6494 LOP(OP_SETPRIORITY,XTERM);
6496 case KEY_sethostent:
6502 case KEY_setservent:
6505 case KEY_setprotoent:
6515 LOP(OP_SEEKDIR,XTERM);
6517 case KEY_setsockopt:
6518 LOP(OP_SSOCKOPT,XTERM);
6524 LOP(OP_SHMCTL,XTERM);
6527 LOP(OP_SHMGET,XTERM);
6530 LOP(OP_SHMREAD,XTERM);
6533 LOP(OP_SHMWRITE,XTERM);
6536 LOP(OP_SHUTDOWN,XTERM);
6545 LOP(OP_SOCKET,XTERM);
6547 case KEY_socketpair:
6548 LOP(OP_SOCKPAIR,XTERM);
6551 checkcomma(s,PL_tokenbuf,"subroutine name");
6553 if (*s == ';' || *s == ')') /* probably a close */
6554 Perl_croak(aTHX_ "sort is now a reserved word");
6556 s = force_word(s,WORD,TRUE,TRUE,FALSE);
6560 LOP(OP_SPLIT,XTERM);
6563 LOP(OP_SPRINTF,XTERM);
6566 LOP(OP_SPLICE,XTERM);
6581 LOP(OP_SUBSTR,XTERM);
6587 char tmpbuf[sizeof PL_tokenbuf];
6588 SSize_t tboffset = 0;
6589 expectation attrful;
6590 bool have_name, have_proto;
6591 const int key = tmp;
6596 char *tstart = SvPVX(PL_linestr) + PL_realtokenstart;
6597 SV *subtoken = newSVpvn(tstart, s - tstart);
6601 s = SKIPSPACE2(s,tmpwhite);
6606 if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' ||
6607 (*s == ':' && s[1] == ':'))
6614 attrful = XATTRBLOCK;
6615 /* remember buffer pos'n for later force_word */
6616 tboffset = s - PL_oldbufptr;
6617 d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
6620 nametoke = newSVpvn(s, d - s);
6622 if (memchr(tmpbuf, ':', len))
6623 sv_setpvn(PL_subname, tmpbuf, len);
6625 sv_setsv(PL_subname,PL_curstname);
6626 sv_catpvs(PL_subname,"::");
6627 sv_catpvn(PL_subname,tmpbuf,len);
6634 CURMAD('X', nametoke);
6635 CURMAD('_', tmpwhite);
6636 (void) force_word(PL_oldbufptr + tboffset, WORD,
6639 s = SKIPSPACE2(d,tmpwhite);
6646 Perl_croak(aTHX_ "Missing name in \"my sub\"");
6647 PL_expect = XTERMBLOCK;
6648 attrful = XATTRTERM;
6649 sv_setpvn(PL_subname,"?",1);
6653 if (key == KEY_format) {
6655 PL_lex_formbrack = PL_lex_brackets + 1;
6657 PL_thistoken = subtoken;
6661 (void) force_word(PL_oldbufptr + tboffset, WORD,
6667 /* Look for a prototype */
6670 bool bad_proto = FALSE;
6671 const bool warnsyntax = ckWARN(WARN_SYNTAX);
6673 s = scan_str(s,!!PL_madskills,FALSE);
6675 Perl_croak(aTHX_ "Prototype not terminated");
6676 /* strip spaces and check for bad characters */
6677 d = SvPVX(PL_lex_stuff);
6679 for (p = d; *p; ++p) {
6682 if (warnsyntax && !strchr("$@%*;[]&\\_", *p))
6688 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
6689 "Illegal character in prototype for %"SVf" : %s",
6690 SVfARG(PL_subname), d);
6691 SvCUR_set(PL_lex_stuff, tmp);
6696 CURMAD('q', PL_thisopen);
6697 CURMAD('_', tmpwhite);
6698 CURMAD('=', PL_thisstuff);
6699 CURMAD('Q', PL_thisclose);
6700 NEXTVAL_NEXTTOKE.opval =
6701 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6702 PL_lex_stuff = Nullsv;
6705 s = SKIPSPACE2(s,tmpwhite);
6713 if (*s == ':' && s[1] != ':')
6714 PL_expect = attrful;
6715 else if (*s != '{' && key == KEY_sub) {
6717 Perl_croak(aTHX_ "Illegal declaration of anonymous subroutine");
6719 Perl_croak(aTHX_ "Illegal declaration of subroutine %"SVf, SVfARG(PL_subname));
6726 curmad('^', newSVpvs(""));
6727 CURMAD('_', tmpwhite);
6731 PL_thistoken = subtoken;
6734 NEXTVAL_NEXTTOKE.opval =
6735 (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff);
6736 PL_lex_stuff = NULL;
6741 sv_setpv(PL_subname,
6743 (PL_curstash ? "__ANON__" : "__ANON__::__ANON__"));
6747 (void) force_word(PL_oldbufptr + tboffset, WORD,
6756 LOP(OP_SYSTEM,XREF);
6759 LOP(OP_SYMLINK,XTERM);
6762 LOP(OP_SYSCALL,XTERM);
6765 LOP(OP_SYSOPEN,XTERM);
6768 LOP(OP_SYSSEEK,XTERM);
6771 LOP(OP_SYSREAD,XTERM);
6774 LOP(OP_SYSWRITE,XTERM);
6778 TERM(sublex_start());
6799 LOP(OP_TRUNCATE,XTERM);
6811 yylval.ival = CopLINE(PL_curcop);
6815 yylval.ival = CopLINE(PL_curcop);
6819 LOP(OP_UNLINK,XTERM);
6825 LOP(OP_UNPACK,XTERM);
6828 LOP(OP_UTIME,XTERM);
6834 LOP(OP_UNSHIFT,XTERM);
6837 s = tokenize_use(1, s);
6847 yylval.ival = CopLINE(PL_curcop);
6851 yylval.ival = CopLINE(PL_curcop);
6855 PL_hints |= HINT_BLOCK_SCOPE;
6862 LOP(OP_WAITPID,XTERM);
6871 ctl_l[0] = toCTRL('L');
6873 gv_fetchpvn_flags(ctl_l, 1, GV_ADD|GV_NOTQUAL, SVt_PV);
6876 /* Make sure $^L is defined */
6877 gv_fetchpvs("\f", GV_ADD|GV_NOTQUAL, SVt_PV);
6882 if (PL_expect == XOPERATOR)
6888 yylval.ival = OP_XOR;
6893 TERM(sublex_start());
6898 #pragma segment Main
6902 S_pending_ident(pTHX)
6907 /* pit holds the identifier we read and pending_ident is reset */
6908 char pit = PL_pending_ident;
6909 PL_pending_ident = 0;
6911 /* PL_realtokenstart = realtokenend = PL_bufptr - SvPVX(PL_linestr); */
6912 DEBUG_T({ PerlIO_printf(Perl_debug_log,
6913 "### Pending identifier '%s'\n", PL_tokenbuf); });
6915 /* if we're in a my(), we can't allow dynamics here.
6916 $foo'bar has already been turned into $foo::bar, so
6917 just check for colons.
6919 if it's a legal name, the OP is a PADANY.
6922 if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */
6923 if (strchr(PL_tokenbuf,':'))
6924 yyerror(Perl_form(aTHX_ "No package name allowed for "
6925 "variable %s in \"our\"",
6927 tmp = allocmy(PL_tokenbuf);
6930 if (strchr(PL_tokenbuf,':'))
6931 yyerror(Perl_form(aTHX_ PL_no_myglob,
6932 PL_in_my == KEY_my ? "my" : "state", PL_tokenbuf));
6934 yylval.opval = newOP(OP_PADANY, 0);
6935 yylval.opval->op_targ = allocmy(PL_tokenbuf);
6941 build the ops for accesses to a my() variable.
6943 Deny my($a) or my($b) in a sort block, *if* $a or $b is
6944 then used in a comparison. This catches most, but not
6945 all cases. For instance, it catches
6946 sort { my($a); $a <=> $b }
6948 sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; }
6949 (although why you'd do that is anyone's guess).
6952 if (!strchr(PL_tokenbuf,':')) {
6954 tmp = pad_findmy(PL_tokenbuf);
6955 if (tmp != NOT_IN_PAD) {
6956 /* might be an "our" variable" */
6957 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
6958 /* build ops for a bareword */
6959 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
6960 HEK * const stashname = HvNAME_HEK(stash);
6961 SV * const sym = newSVhek(stashname);
6962 sv_catpvs(sym, "::");
6963 sv_catpv(sym, PL_tokenbuf+1);
6964 yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym);
6965 yylval.opval->op_private = OPpCONST_ENTERED;
6968 ? (GV_ADDMULTI | GV_ADDINEVAL)
6971 ((PL_tokenbuf[0] == '$') ? SVt_PV
6972 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
6977 /* if it's a sort block and they're naming $a or $b */
6978 if (PL_last_lop_op == OP_SORT &&
6979 PL_tokenbuf[0] == '$' &&
6980 (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b')
6983 for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart;
6984 d < PL_bufend && *d != '\n';
6987 if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) {
6988 Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison",
6994 yylval.opval = newOP(OP_PADANY, 0);
6995 yylval.opval->op_targ = tmp;
7001 Whine if they've said @foo in a doublequoted string,
7002 and @foo isn't a variable we can find in the symbol
7005 if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) {
7006 GV *gv = gv_fetchpv(PL_tokenbuf+1, 0, SVt_PVAV);
7007 if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv)))
7008 && ckWARN(WARN_AMBIGUOUS)
7009 /* DO NOT warn for @- and @+ */
7010 && !( PL_tokenbuf[2] == '\0' &&
7011 ( PL_tokenbuf[1] == '-' || PL_tokenbuf[1] == '+' ))
7014 /* Downgraded from fatal to warning 20000522 mjd */
7015 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
7016 "Possible unintended interpolation of %s in string",
7021 /* build ops for a bareword */
7022 yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0));
7023 yylval.opval->op_private = OPpCONST_ENTERED;
7026 /* If the identifier refers to a stash, don't autovivify it.
7027 * Change 24660 had the side effect of causing symbol table
7028 * hashes to always be defined, even if they were freshly
7029 * created and the only reference in the entire program was
7030 * the single statement with the defined %foo::bar:: test.
7031 * It appears that all code in the wild doing this actually
7032 * wants to know whether sub-packages have been loaded, so
7033 * by avoiding auto-vivifying symbol tables, we ensure that
7034 * defined %foo::bar:: continues to be false, and the existing
7035 * tests still give the expected answers, even though what
7036 * they're actually testing has now changed subtly.
7038 (*PL_tokenbuf == '%' && *(d = PL_tokenbuf + strlen(PL_tokenbuf) - 1) == ':' && d[-1] == ':'
7040 : PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : GV_ADD),
7041 ((PL_tokenbuf[0] == '$') ? SVt_PV
7042 : (PL_tokenbuf[0] == '@') ? SVt_PVAV
7048 * The following code was generated by perl_keyword.pl.
7052 Perl_keyword (pTHX_ const char *name, I32 len, bool all_keywords)
7057 case 1: /* 5 tokens of length 1 */
7089 case 2: /* 18 tokens of length 2 */
7235 case 3: /* 29 tokens of length 3 */
7239 if (name[1] == 'N' &&
7302 if (name[1] == 'i' &&
7334 if (name[1] == 'o' &&
7343 if (name[1] == 'e' &&
7352 if (name[1] == 'n' &&
7361 if (name[1] == 'o' &&
7370 if (name[1] == 'a' &&
7379 if (name[1] == 'o' &&
7441 if (name[1] == 'e' &&
7455 return (all_keywords || FEATURE_IS_ENABLED("say") ? KEY_say : 0);
7481 if (name[1] == 'i' &&
7490 if (name[1] == 's' &&
7499 if (name[1] == 'e' &&
7508 if (name[1] == 'o' &&
7520 case 4: /* 41 tokens of length 4 */
7524 if (name[1] == 'O' &&
7534 if (name[1] == 'N' &&
7544 if (name[1] == 'i' &&
7554 if (name[1] == 'h' &&
7564 if (name[1] == 'u' &&
7577 if (name[2] == 'c' &&
7586 if (name[2] == 's' &&
7595 if (name[2] == 'a' &&
7631 if (name[1] == 'o' &&
7644 if (name[2] == 't' &&
7653 if (name[2] == 'o' &&
7662 if (name[2] == 't' &&
7671 if (name[2] == 'e' &&
7684 if (name[1] == 'o' &&
7697 if (name[2] == 'y' &&
7706 if (name[2] == 'l' &&
7722 if (name[2] == 's' &&
7731 if (name[2] == 'n' &&
7740 if (name[2] == 'c' &&
7753 if (name[1] == 'e' &&
7763 if (name[1] == 'p' &&
7776 if (name[2] == 'c' &&
7785 if (name[2] == 'p' &&
7794 if (name[2] == 's' &&
7810 if (name[2] == 'n' &&
7880 if (name[2] == 'r' &&
7889 if (name[2] == 'r' &&
7898 if (name[2] == 'a' &&
7914 if (name[2] == 'l' &&
7976 if (name[2] == 'e' &&
7979 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_when : 0);
7992 case 5: /* 39 tokens of length 5 */
7996 if (name[1] == 'E' &&
8007 if (name[1] == 'H' &&
8021 if (name[2] == 'a' &&
8031 if (name[2] == 'a' &&
8048 if (name[2] == 'e' &&
8058 if (name[2] == 'e' &&
8062 return (all_keywords || FEATURE_IS_ENABLED("switch") ? -KEY_break : 0);
8078 if (name[3] == 'i' &&
8087 if (name[3] == 'o' &&
8123 if (name[2] == 'o' &&
8133 if (name[2] == 'y' &&
8147 if (name[1] == 'l' &&
8161 if (name[2] == 'n' &&
8171 if (name[2] == 'o' &&
8185 if (name[1] == 'i' &&
8190 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_given : 0);
8199 if (name[2] == 'd' &&
8209 if (name[2] == 'c' &&
8226 if (name[2] == 'c' &&
8236 if (name[2] == 't' &&
8250 if (name[1] == 'k' &&
8261 if (name[1] == 'r' &&
8275 if (name[2] == 's' &&
8285 if (name[2] == 'd' &&
8302 if (name[2] == 'm' &&
8312 if (name[2] == 'i' &&
8322 if (name[2] == 'e' &&
8332 if (name[2] == 'l' &&
8342 if (name[2] == 'a' &&
8355 if (name[3] == 't' &&
8358 return (all_keywords || FEATURE_IS_ENABLED("state") ? KEY_state : 0);
8364 if (name[3] == 'd' &&
8381 if (name[1] == 'i' &&
8395 if (name[2] == 'a' &&
8408 if (name[3] == 'e' &&
8443 if (name[2] == 'i' &&
8460 if (name[2] == 'i' &&
8470 if (name[2] == 'i' &&
8487 case 6: /* 33 tokens of length 6 */
8491 if (name[1] == 'c' &&
8506 if (name[2] == 'l' &&
8517 if (name[2] == 'r' &&
8532 if (name[1] == 'e' &&
8547 if (name[2] == 's' &&
8552 if(ckWARN_d(WARN_SYNTAX))
8553 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "elseif should be elsif");
8559 if (name[2] == 'i' &&
8577 if (name[2] == 'l' &&
8588 if (name[2] == 'r' &&
8603 if (name[1] == 'm' &&
8618 if (name[2] == 'n' &&
8629 if (name[2] == 's' &&
8644 if (name[1] == 's' &&
8650 if (name[4] == 't' &&
8659 if (name[4] == 'e' &&
8668 if (name[4] == 'c' &&
8677 if (name[4] == 'n' &&
8693 if (name[1] == 'r' &&
8711 if (name[3] == 'a' &&
8721 if (name[3] == 'u' &&
8735 if (name[2] == 'n' &&
8753 if (name[2] == 'a' &&
8767 if (name[3] == 'e' &&
8780 if (name[4] == 't' &&
8789 if (name[4] == 'e' &&
8811 if (name[4] == 't' &&
8820 if (name[4] == 'e' &&
8836 if (name[2] == 'c' &&
8847 if (name[2] == 'l' &&
8858 if (name[2] == 'b' &&
8869 if (name[2] == 's' &&
8892 if (name[4] == 's' &&
8901 if (name[4] == 'n' &&
8914 if (name[3] == 'a' &&
8931 if (name[1] == 'a' &&
8946 case 7: /* 29 tokens of length 7 */
8950 if (name[1] == 'E' &&
8963 if (name[1] == '_' &&
8976 if (name[1] == 'i' &&
8983 return -KEY_binmode;
8989 if (name[1] == 'o' &&
8996 return -KEY_connect;
9005 if (name[2] == 'm' &&
9011 return -KEY_dbmopen;
9022 if (name[4] == 'u' &&
9026 return (all_keywords || FEATURE_IS_ENABLED("switch") ? KEY_default : 0);
9032 if (name[4] == 'n' &&
9053 if (name[1] == 'o' &&
9066 if (name[1] == 'e' &&
9073 if (name[5] == 'r' &&
9076 return -KEY_getpgrp;
9082 if (name[5] == 'i' &&
9085 return -KEY_getppid;
9098 if (name[1] == 'c' &&
9105 return -KEY_lcfirst;
9111 if (name[1] == 'p' &&
9118 return -KEY_opendir;
9124 if (name[1] == 'a' &&
9142 if (name[3] == 'd' &&
9147 return -KEY_readdir;
9153 if (name[3] == 'u' &&
9164 if (name[3] == 'e' &&
9169 return -KEY_reverse;
9188 if (name[3] == 'k' &&
9193 return -KEY_seekdir;
9199 if (name[3] == 'p' &&
9204 return -KEY_setpgrp;
9214 if (name[2] == 'm' &&
9220 return -KEY_shmread;
9226 if (name[2] == 'r' &&
9232 return -KEY_sprintf;
9241 if (name[3] == 'l' &&
9246 return -KEY_symlink;
9255 if (name[4] == 'a' &&
9259 return -KEY_syscall;
9265 if (name[4] == 'p' &&
9269 return -KEY_sysopen;
9275 if (name[4] == 'e' &&
9279 return -KEY_sysread;
9285 if (name[4] == 'e' &&
9289 return -KEY_sysseek;
9307 if (name[1] == 'e' &&
9314 return -KEY_telldir;
9323 if (name[2] == 'f' &&
9329 return -KEY_ucfirst;
9335 if (name[2] == 's' &&
9341 return -KEY_unshift;
9351 if (name[1] == 'a' &&
9358 return -KEY_waitpid;
9367 case 8: /* 26 tokens of length 8 */
9371 if (name[1] == 'U' &&
9379 return KEY_AUTOLOAD;
9390 if (name[3] == 'A' &&
9396 return KEY___DATA__;
9402 if (name[3] == 'I' &&
9408 return -KEY___FILE__;
9414 if (name[3] == 'I' &&
9420 return -KEY___LINE__;
9436 if (name[2] == 'o' &&
9443 return -KEY_closedir;
9449 if (name[2] == 'n' &&
9456 return -KEY_continue;
9466 if (name[1] == 'b' &&
9474 return -KEY_dbmclose;
9480 if (name[1] == 'n' &&
9486 if (name[4] == 'r' &&
9491 return -KEY_endgrent;
9497 if (name[4] == 'w' &&
9502 return -KEY_endpwent;
9515 if (name[1] == 'o' &&
9523 return -KEY_formline;
9529 if (name[1] == 'e' &&
9540 if (name[6] == 'n' &&
9543 return -KEY_getgrent;
9549 if (name[6] == 'i' &&
9552 return -KEY_getgrgid;
9558 if (name[6] == 'a' &&
9561 return -KEY_getgrnam;
9574 if (name[4] == 'o' &&
9579 return -KEY_getlogin;
9590 if (name[6] == 'n' &&
9593 return -KEY_getpwent;
9599 if (name[6] == 'a' &&
9602 return -KEY_getpwnam;
9608 if (name[6] == 'i' &&
9611 return -KEY_getpwuid;
9631 if (name[1] == 'e' &&
9638 if (name[5] == 'i' &&
9645 return -KEY_readline;
9650 return -KEY_readlink;
9661 if (name[5] == 'i' &&
9665 return -KEY_readpipe;
9686 if (name[4] == 'r' &&
9691 return -KEY_setgrent;
9697 if (name[4] == 'w' &&
9702 return -KEY_setpwent;
9718 if (name[3] == 'w' &&
9724 return -KEY_shmwrite;
9730 if (name[3] == 't' &&
9736 return -KEY_shutdown;
9746 if (name[2] == 's' &&
9753 return -KEY_syswrite;
9763 if (name[1] == 'r' &&
9771 return -KEY_truncate;
9780 case 9: /* 9 tokens of length 9 */
9784 if (name[1] == 'N' &&
9793 return KEY_UNITCHECK;
9799 if (name[1] == 'n' &&
9808 return -KEY_endnetent;
9814 if (name[1] == 'e' &&
9823 return -KEY_getnetent;
9829 if (name[1] == 'o' &&
9838 return -KEY_localtime;
9844 if (name[1] == 'r' &&
9853 return KEY_prototype;
9859 if (name[1] == 'u' &&
9868 return -KEY_quotemeta;
9874 if (name[1] == 'e' &&
9883 return -KEY_rewinddir;
9889 if (name[1] == 'e' &&
9898 return -KEY_setnetent;
9904 if (name[1] == 'a' &&
9913 return -KEY_wantarray;
9922 case 10: /* 9 tokens of length 10 */
9926 if (name[1] == 'n' &&
9932 if (name[4] == 'o' &&
9939 return -KEY_endhostent;
9945 if (name[4] == 'e' &&
9952 return -KEY_endservent;
9965 if (name[1] == 'e' &&
9971 if (name[4] == 'o' &&
9978 return -KEY_gethostent;
9987 if (name[5] == 'r' &&
9993 return -KEY_getservent;
9999 if (name[5] == 'c' &&
10005 return -KEY_getsockopt;
10025 if (name[2] == 't')
10030 if (name[4] == 'o' &&
10037 return -KEY_sethostent;
10046 if (name[5] == 'r' &&
10052 return -KEY_setservent;
10058 if (name[5] == 'c' &&
10064 return -KEY_setsockopt;
10081 if (name[2] == 'c' &&
10090 return -KEY_socketpair;
10103 case 11: /* 8 tokens of length 11 */
10107 if (name[1] == '_' &&
10117 { /* __PACKAGE__ */
10118 return -KEY___PACKAGE__;
10124 if (name[1] == 'n' &&
10134 { /* endprotoent */
10135 return -KEY_endprotoent;
10141 if (name[1] == 'e' &&
10150 if (name[5] == 'e' &&
10156 { /* getpeername */
10157 return -KEY_getpeername;
10166 if (name[6] == 'o' &&
10171 { /* getpriority */
10172 return -KEY_getpriority;
10178 if (name[6] == 't' &&
10183 { /* getprotoent */
10184 return -KEY_getprotoent;
10198 if (name[4] == 'o' &&
10205 { /* getsockname */
10206 return -KEY_getsockname;
10219 if (name[1] == 'e' &&
10227 if (name[6] == 'o' &&
10232 { /* setpriority */
10233 return -KEY_setpriority;
10239 if (name[6] == 't' &&
10244 { /* setprotoent */
10245 return -KEY_setprotoent;
10261 case 12: /* 2 tokens of length 12 */
10262 if (name[0] == 'g' &&
10274 if (name[9] == 'd' &&
10277 { /* getnetbyaddr */
10278 return -KEY_getnetbyaddr;
10284 if (name[9] == 'a' &&
10287 { /* getnetbyname */
10288 return -KEY_getnetbyname;
10300 case 13: /* 4 tokens of length 13 */
10301 if (name[0] == 'g' &&
10308 if (name[4] == 'o' &&
10317 if (name[10] == 'd' &&
10320 { /* gethostbyaddr */
10321 return -KEY_gethostbyaddr;
10327 if (name[10] == 'a' &&
10330 { /* gethostbyname */
10331 return -KEY_gethostbyname;
10344 if (name[4] == 'e' &&
10353 if (name[10] == 'a' &&
10356 { /* getservbyname */
10357 return -KEY_getservbyname;
10363 if (name[10] == 'o' &&
10366 { /* getservbyport */
10367 return -KEY_getservbyport;
10386 case 14: /* 1 tokens of length 14 */
10387 if (name[0] == 'g' &&
10401 { /* getprotobyname */
10402 return -KEY_getprotobyname;
10407 case 16: /* 1 tokens of length 16 */
10408 if (name[0] == 'g' &&
10424 { /* getprotobynumber */
10425 return -KEY_getprotobynumber;
10439 S_checkcomma(pTHX_ const char *s, const char *name, const char *what)
10443 if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */
10444 if (ckWARN(WARN_SYNTAX)) {
10447 for (w = s+2; *w && level; w++) {
10450 else if (*w == ')')
10453 while (isSPACE(*w))
10455 /* the list of chars below is for end of statements or
10456 * block / parens, boolean operators (&&, ||, //) and branch
10457 * constructs (or, and, if, until, unless, while, err, for).
10458 * Not a very solid hack... */
10459 if (!*w || !strchr(";&/|})]oaiuwef!=", *w))
10460 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
10461 "%s (...) interpreted as function",name);
10464 while (s < PL_bufend && isSPACE(*s))
10468 while (s < PL_bufend && isSPACE(*s))
10470 if (isIDFIRST_lazy_if(s,UTF)) {
10471 const char * const w = s++;
10472 while (isALNUM_lazy_if(s,UTF))
10474 while (s < PL_bufend && isSPACE(*s))
10478 if (keyword(w, s - w, 0))
10481 gv = gv_fetchpvn_flags(w, s - w, 0, SVt_PVCV);
10482 if (gv && GvCVu(gv))
10484 Perl_croak(aTHX_ "No comma allowed after %s", what);
10489 /* Either returns sv, or mortalizes sv and returns a new SV*.
10490 Best used as sv=new_constant(..., sv, ...).
10491 If s, pv are NULL, calls subroutine with one argument,
10492 and type is used with error messages only. */
10495 S_new_constant(pTHX_ const char *s, STRLEN len, const char *key, SV *sv, SV *pv,
10499 HV * const table = GvHV(PL_hintgv); /* ^H */
10503 const char *why1 = "", *why2 = "", *why3 = "";
10505 if (!table || !(PL_hints & HINT_LOCALIZE_HH)) {
10508 why2 = (const char *)
10509 (strEQ(key,"charnames")
10510 ? "(possibly a missing \"use charnames ...\")"
10512 msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s",
10513 (type ? type: "undef"), why2);
10515 /* This is convoluted and evil ("goto considered harmful")
10516 * but I do not understand the intricacies of all the different
10517 * failure modes of %^H in here. The goal here is to make
10518 * the most probable error message user-friendly. --jhi */
10523 msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s",
10524 (type ? type: "undef"), why1, why2, why3);
10526 yyerror(SvPVX_const(msg));
10530 cvp = hv_fetch(table, key, strlen(key), FALSE);
10531 if (!cvp || !SvOK(*cvp)) {
10534 why3 = "} is not defined";
10537 sv_2mortal(sv); /* Parent created it permanently */
10540 pv = sv_2mortal(newSVpvn(s, len));
10542 typesv = sv_2mortal(newSVpv(type, 0));
10544 typesv = &PL_sv_undef;
10546 PUSHSTACKi(PERLSI_OVERLOAD);
10558 call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL));
10562 /* Check the eval first */
10563 if (!PL_in_eval && SvTRUE(ERRSV)) {
10564 sv_catpvs(ERRSV, "Propagated");
10565 yyerror(SvPV_nolen_const(ERRSV)); /* Duplicates the message inside eval */
10567 res = SvREFCNT_inc_simple(sv);
10571 SvREFCNT_inc_simple_void(res);
10580 why1 = "Call to &{$^H{";
10582 why3 = "}} did not return a defined value";
10590 /* Returns a NUL terminated string, with the length of the string written to
10594 S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp)
10597 register char *d = dest;
10598 register char * const e = d + destlen - 3; /* two-character token, ending NUL */
10601 Perl_croak(aTHX_ ident_too_long);
10602 if (isALNUM(*s)) /* UTF handled below */
10604 else if (allow_package && (*s == '\'') && isIDFIRST_lazy_if(s+1,UTF)) {
10609 else if (allow_package && (s[0] == ':') && (s[1] == ':') && (s[2] != '$')) {
10613 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10614 char *t = s + UTF8SKIP(s);
10616 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10620 Perl_croak(aTHX_ ident_too_long);
10621 Copy(s, d, len, char);
10634 S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni)
10637 char *bracket = NULL;
10639 register char *d = dest;
10640 register char * const e = d + destlen + 3; /* two-character token, ending NUL */
10645 while (isDIGIT(*s)) {
10647 Perl_croak(aTHX_ ident_too_long);
10654 Perl_croak(aTHX_ ident_too_long);
10655 if (isALNUM(*s)) /* UTF handled below */
10657 else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) {
10662 else if (*s == ':' && s[1] == ':') {
10666 else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) {
10667 char *t = s + UTF8SKIP(s);
10668 while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t))
10670 if (d + (t - s) > e)
10671 Perl_croak(aTHX_ ident_too_long);
10672 Copy(s, d, t - s, char);
10683 if (PL_lex_state != LEX_NORMAL)
10684 PL_lex_state = LEX_INTERPENDMAYBE;
10687 if (*s == '$' && s[1] &&
10688 (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) )
10701 if (*d == '^' && *s && isCONTROLVAR(*s)) {
10706 if (isSPACE(s[-1])) {
10708 const char ch = *s++;
10709 if (!SPACE_OR_TAB(ch)) {
10715 if (isIDFIRST_lazy_if(d,UTF)) {
10719 while ((end < send && isALNUM_lazy_if(end,UTF)) || *end == ':') {
10720 end += UTF8SKIP(end);
10721 while (end < send && UTF8_IS_CONTINUED(*end) && is_utf8_mark((U8*)end))
10722 end += UTF8SKIP(end);
10724 Copy(s, d, end - s, char);
10729 while ((isALNUM(*s) || *s == ':') && d < e)
10732 Perl_croak(aTHX_ ident_too_long);
10735 while (s < send && SPACE_OR_TAB(*s))
10737 if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) {
10738 if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest, 0)) {
10739 const char * const brack =
10741 ((*s == '[') ? "[...]" : "{...}");
10742 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10743 "Ambiguous use of %c{%s%s} resolved to %c%s%s",
10744 funny, dest, brack, funny, dest, brack);
10747 PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK);
10751 /* Handle extended ${^Foo} variables
10752 * 1999-02-27 mjd-perl-patch@plover.com */
10753 else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */
10757 while (isALNUM(*s) && d < e) {
10761 Perl_croak(aTHX_ ident_too_long);
10766 if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) {
10767 PL_lex_state = LEX_INTERPEND;
10770 if (PL_lex_state == LEX_NORMAL) {
10771 if (ckWARN(WARN_AMBIGUOUS) &&
10772 (keyword(dest, d - dest, 0)
10773 || get_cvn_flags(dest, d - dest, 0)))
10777 Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
10778 "Ambiguous use of %c{%s} resolved to %c%s",
10779 funny, dest, funny, dest);
10784 s = bracket; /* let the parser handle it */
10788 else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s))
10789 PL_lex_state = LEX_INTERPEND;
10794 Perl_pmflag(pTHX_ U32* pmfl, int ch)
10796 PERL_UNUSED_CONTEXT;
10800 CASE_STD_PMMOD_FLAGS_PARSE_SET(pmfl);
10801 case GLOBAL_PAT_MOD: *pmfl |= PMf_GLOBAL; break;
10802 case CONTINUE_PAT_MOD: *pmfl |= PMf_CONTINUE; break;
10803 case ONCE_PAT_MOD: *pmfl |= PMf_KEEP; break;
10804 case KEEPCOPY_PAT_MOD: *pmfl |= PMf_KEEPCOPY; break;
10810 S_scan_pat(pTHX_ char *start, I32 type)
10814 char *s = scan_str(start,!!PL_madskills,FALSE);
10815 const char * const valid_flags =
10816 (const char *)((type == OP_QR) ? QR_PAT_MODS : M_PAT_MODS);
10823 const char * const delimiter = skipspace(start);
10827 ? "Search pattern not terminated or ternary operator parsed as search pattern"
10828 : "Search pattern not terminated" ));
10831 pm = (PMOP*)newPMOP(type, 0);
10832 if (PL_multi_open == '?') {
10833 /* This is the only point in the code that sets PMf_ONCE: */
10834 pm->op_pmflags |= PMf_ONCE;
10836 /* Hence it's safe to do this bit of PMOP book-keeping here, which
10837 allows us to restrict the list needed by reset to just the ??
10839 assert(type != OP_TRANS);
10841 MAGIC *mg = mg_find((SV*)PL_curstash, PERL_MAGIC_symtab);
10844 mg = sv_magicext((SV*)PL_curstash, 0, PERL_MAGIC_symtab, 0, 0,
10847 elements = mg->mg_len / sizeof(PMOP**);
10848 Renewc(mg->mg_ptr, elements + 1, PMOP*, char);
10849 ((PMOP**)mg->mg_ptr) [elements++] = pm;
10850 mg->mg_len = elements * sizeof(PMOP**);
10851 PmopSTASH_set(pm,PL_curstash);
10857 while (*s && strchr(valid_flags, *s))
10858 pmflag(&pm->op_pmflags,*s++);
10860 if (PL_madskills && modstart != s) {
10861 SV* tmptoken = newSVpvn(modstart, s - modstart);
10862 append_madprops(newMADPROP('m', MAD_SV, tmptoken, 0), (OP*)pm, 0);
10865 /* issue a warning if /c is specified,but /g is not */
10866 if ((pm->op_pmflags & PMf_CONTINUE) && !(pm->op_pmflags & PMf_GLOBAL)
10867 && ckWARN(WARN_REGEXP))
10869 Perl_warner(aTHX_ packWARN(WARN_REGEXP),
10870 "Use of /c modifier is meaningless without /g" );
10873 PL_lex_op = (OP*)pm;
10874 yylval.ival = OP_MATCH;
10879 S_scan_subst(pTHX_ char *start)
10890 yylval.ival = OP_NULL;
10892 s = scan_str(start,!!PL_madskills,FALSE);
10895 Perl_croak(aTHX_ "Substitution pattern not terminated");
10897 if (s[-1] == PL_multi_open)
10900 if (PL_madskills) {
10901 CURMAD('q', PL_thisopen);
10902 CURMAD('_', PL_thiswhite);
10903 CURMAD('E', PL_thisstuff);
10904 CURMAD('Q', PL_thisclose);
10905 PL_realtokenstart = s - SvPVX(PL_linestr);
10909 first_start = PL_multi_start;
10910 s = scan_str(s,!!PL_madskills,FALSE);
10912 if (PL_lex_stuff) {
10913 SvREFCNT_dec(PL_lex_stuff);
10914 PL_lex_stuff = NULL;
10916 Perl_croak(aTHX_ "Substitution replacement not terminated");
10918 PL_multi_start = first_start; /* so whole substitution is taken together */
10920 pm = (PMOP*)newPMOP(OP_SUBST, 0);
10923 if (PL_madskills) {
10924 CURMAD('z', PL_thisopen);
10925 CURMAD('R', PL_thisstuff);
10926 CURMAD('Z', PL_thisclose);
10932 if (*s == EXEC_PAT_MOD) {
10936 else if (strchr(S_PAT_MODS, *s))
10937 pmflag(&pm->op_pmflags,*s++);
10943 if (PL_madskills) {
10945 curmad('m', newSVpvn(modstart, s - modstart));
10946 append_madprops(PL_thismad, (OP*)pm, 0);
10950 if ((pm->op_pmflags & PMf_CONTINUE) && ckWARN(WARN_REGEXP)) {
10951 Perl_warner(aTHX_ packWARN(WARN_REGEXP), "Use of /c modifier is meaningless in s///" );
10955 SV * const repl = newSVpvs("");
10957 PL_sublex_info.super_bufptr = s;
10958 PL_sublex_info.super_bufend = PL_bufend;
10960 pm->op_pmflags |= PMf_EVAL;
10963 sv_catpvs(repl, "eval ");
10965 sv_catpvs(repl, "do ");
10967 sv_catpvs(repl, "{");
10968 sv_catsv(repl, PL_lex_repl);
10969 if (strchr(SvPVX(PL_lex_repl), '#'))
10970 sv_catpvs(repl, "\n");
10971 sv_catpvs(repl, "}");
10973 SvREFCNT_dec(PL_lex_repl);
10974 PL_lex_repl = repl;
10977 PL_lex_op = (OP*)pm;
10978 yylval.ival = OP_SUBST;
10983 S_scan_trans(pTHX_ char *start)
10996 yylval.ival = OP_NULL;
10998 s = scan_str(start,!!PL_madskills,FALSE);
11000 Perl_croak(aTHX_ "Transliteration pattern not terminated");
11002 if (s[-1] == PL_multi_open)
11005 if (PL_madskills) {
11006 CURMAD('q', PL_thisopen);
11007 CURMAD('_', PL_thiswhite);
11008 CURMAD('E', PL_thisstuff);
11009 CURMAD('Q', PL_thisclose);
11010 PL_realtokenstart = s - SvPVX(PL_linestr);
11014 s = scan_str(s,!!PL_madskills,FALSE);
11016 if (PL_lex_stuff) {
11017 SvREFCNT_dec(PL_lex_stuff);
11018 PL_lex_stuff = NULL;
11020 Perl_croak(aTHX_ "Transliteration replacement not terminated");
11022 if (PL_madskills) {
11023 CURMAD('z', PL_thisopen);
11024 CURMAD('R', PL_thisstuff);
11025 CURMAD('Z', PL_thisclose);
11028 complement = del = squash = 0;
11035 complement = OPpTRANS_COMPLEMENT;
11038 del = OPpTRANS_DELETE;
11041 squash = OPpTRANS_SQUASH;
11050 tbl = (short *)PerlMemShared_calloc(complement&&!del?258:256, sizeof(short));
11051 o = newPVOP(OP_TRANS, 0, (char*)tbl);
11052 o->op_private &= ~OPpTRANS_ALL;
11053 o->op_private |= del|squash|complement|
11054 (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)|
11055 (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0);
11058 yylval.ival = OP_TRANS;
11061 if (PL_madskills) {
11063 curmad('m', newSVpvn(modstart, s - modstart));
11064 append_madprops(PL_thismad, o, 0);
11073 S_scan_heredoc(pTHX_ register char *s)
11077 I32 op_type = OP_SCALAR;
11081 const char *found_newline;
11085 const int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR));
11087 I32 stuffstart = s - SvPVX(PL_linestr);
11090 PL_realtokenstart = -1;
11095 e = PL_tokenbuf + sizeof PL_tokenbuf - 1;
11099 while (SPACE_OR_TAB(*peek))
11101 if (*peek == '`' || *peek == '\'' || *peek =='"') {
11104 s = delimcpy(d, e, s, PL_bufend, term, &len);
11114 if (!isALNUM_lazy_if(s,UTF))
11115 deprecate_old("bare << to mean <<\"\"");
11116 for (; isALNUM_lazy_if(s,UTF); s++) {
11121 if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1)
11122 Perl_croak(aTHX_ "Delimiter for here document is too long");
11125 len = d - PL_tokenbuf;
11128 if (PL_madskills) {
11129 tstart = PL_tokenbuf + !outer;
11130 PL_thisclose = newSVpvn(tstart, len - !outer);
11131 tstart = SvPVX(PL_linestr) + stuffstart;
11132 PL_thisopen = newSVpvn(tstart, s - tstart);
11133 stuffstart = s - SvPVX(PL_linestr);
11136 #ifndef PERL_STRICT_CR
11137 d = strchr(s, '\r');
11139 char * const olds = s;
11141 while (s < PL_bufend) {
11147 else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */
11156 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11163 if ( outer || !(found_newline = (char*)memchr((void*)s, '\n', PL_bufend - s)) ) {
11164 herewas = newSVpvn(s,PL_bufend-s);
11168 herewas = newSVpvn(s-1,found_newline-s+1);
11171 herewas = newSVpvn(s,found_newline-s);
11175 if (PL_madskills) {
11176 tstart = SvPVX(PL_linestr) + stuffstart;
11178 sv_catpvn(PL_thisstuff, tstart, s - tstart);
11180 PL_thisstuff = newSVpvn(tstart, s - tstart);
11183 s += SvCUR(herewas);
11186 stuffstart = s - SvPVX(PL_linestr);
11192 tmpstr = newSV_type(SVt_PVIV);
11193 SvGROW(tmpstr, 80);
11194 if (term == '\'') {
11195 op_type = OP_CONST;
11196 SvIV_set(tmpstr, -1);
11198 else if (term == '`') {
11199 op_type = OP_BACKTICK;
11200 SvIV_set(tmpstr, '\\');
11204 PL_multi_start = CopLINE(PL_curcop);
11205 PL_multi_open = PL_multi_close = '<';
11206 term = *PL_tokenbuf;
11207 if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) {
11208 char * const bufptr = PL_sublex_info.super_bufptr;
11209 char * const bufend = PL_sublex_info.super_bufend;
11210 char * const olds = s - SvCUR(herewas);
11211 s = strchr(bufptr, '\n');
11215 while (s < bufend &&
11216 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11218 CopLINE_inc(PL_curcop);
11221 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11222 missingterm(PL_tokenbuf);
11224 sv_setpvn(herewas,bufptr,d-bufptr+1);
11225 sv_setpvn(tmpstr,d+1,s-d);
11227 sv_catpvn(herewas,s,bufend-s);
11228 Copy(SvPVX_const(herewas),bufptr,SvCUR(herewas) + 1,char);
11235 while (s < PL_bufend &&
11236 (*s != term || memNE(s,PL_tokenbuf,len)) ) {
11238 CopLINE_inc(PL_curcop);
11240 if (s >= PL_bufend) {
11241 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11242 missingterm(PL_tokenbuf);
11244 sv_setpvn(tmpstr,d+1,s-d);
11246 if (PL_madskills) {
11248 sv_catpvn(PL_thisstuff, d + 1, s - d);
11250 PL_thisstuff = newSVpvn(d + 1, s - d);
11251 stuffstart = s - SvPVX(PL_linestr);
11255 CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */
11257 sv_catpvn(herewas,s,PL_bufend-s);
11258 sv_setsv(PL_linestr,herewas);
11259 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr);
11260 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11261 PL_last_lop = PL_last_uni = NULL;
11264 sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */
11265 while (s >= PL_bufend) { /* multiple line string? */
11267 if (PL_madskills) {
11268 tstart = SvPVX(PL_linestr) + stuffstart;
11270 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11272 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11276 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11277 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11278 missingterm(PL_tokenbuf);
11281 stuffstart = s - SvPVX(PL_linestr);
11283 CopLINE_inc(PL_curcop);
11284 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11285 PL_last_lop = PL_last_uni = NULL;
11286 #ifndef PERL_STRICT_CR
11287 if (PL_bufend - PL_linestart >= 2) {
11288 if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') ||
11289 (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r'))
11291 PL_bufend[-2] = '\n';
11293 SvCUR_set(PL_linestr, PL_bufend - SvPVX_const(PL_linestr));
11295 else if (PL_bufend[-1] == '\r')
11296 PL_bufend[-1] = '\n';
11298 else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r')
11299 PL_bufend[-1] = '\n';
11301 if (PERLDB_LINE && PL_curstash != PL_debstash)
11302 update_debugger_info(PL_linestr, NULL, 0);
11303 if (*s == term && memEQ(s,PL_tokenbuf,len)) {
11304 STRLEN off = PL_bufend - 1 - SvPVX_const(PL_linestr);
11305 *(SvPVX(PL_linestr) + off ) = ' ';
11306 sv_catsv(PL_linestr,herewas);
11307 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11308 s = SvPVX(PL_linestr) + off; /* In case PV of PL_linestr moved. */
11312 sv_catsv(tmpstr,PL_linestr);
11317 PL_multi_end = CopLINE(PL_curcop);
11318 if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) {
11319 SvPV_shrink_to_cur(tmpstr);
11321 SvREFCNT_dec(herewas);
11323 if (UTF && is_utf8_string((U8*)SvPVX_const(tmpstr), SvCUR(tmpstr)))
11325 else if (PL_encoding)
11326 sv_recode_to_utf8(tmpstr, PL_encoding);
11328 PL_lex_stuff = tmpstr;
11329 yylval.ival = op_type;
11333 /* scan_inputsymbol
11334 takes: current position in input buffer
11335 returns: new position in input buffer
11336 side-effects: yylval and lex_op are set.
11341 <FH> read from filehandle
11342 <pkg::FH> read from package qualified filehandle
11343 <pkg'FH> read from package qualified filehandle
11344 <$fh> read from filehandle in $fh
11345 <*.h> filename glob
11350 S_scan_inputsymbol(pTHX_ char *start)
11353 register char *s = start; /* current position in buffer */
11357 char *d = PL_tokenbuf; /* start of temp holding space */
11358 const char * const e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */
11360 end = strchr(s, '\n');
11363 s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */
11365 /* die if we didn't have space for the contents of the <>,
11366 or if it didn't end, or if we see a newline
11369 if (len >= (I32)sizeof PL_tokenbuf)
11370 Perl_croak(aTHX_ "Excessively long <> operator");
11372 Perl_croak(aTHX_ "Unterminated <> operator");
11377 Remember, only scalar variables are interpreted as filehandles by
11378 this code. Anything more complex (e.g., <$fh{$num}>) will be
11379 treated as a glob() call.
11380 This code makes use of the fact that except for the $ at the front,
11381 a scalar variable and a filehandle look the same.
11383 if (*d == '$' && d[1]) d++;
11385 /* allow <Pkg'VALUE> or <Pkg::VALUE> */
11386 while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':'))
11389 /* If we've tried to read what we allow filehandles to look like, and
11390 there's still text left, then it must be a glob() and not a getline.
11391 Use scan_str to pull out the stuff between the <> and treat it
11392 as nothing more than a string.
11395 if (d - PL_tokenbuf != len) {
11396 yylval.ival = OP_GLOB;
11397 s = scan_str(start,!!PL_madskills,FALSE);
11399 Perl_croak(aTHX_ "Glob not terminated");
11403 bool readline_overriden = FALSE;
11406 /* we're in a filehandle read situation */
11409 /* turn <> into <ARGV> */
11411 Copy("ARGV",d,5,char);
11413 /* Check whether readline() is overriden */
11414 gv_readline = gv_fetchpvs("readline", GV_NOTQUAL, SVt_PVCV);
11416 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline))
11418 ((gvp = (GV**)hv_fetchs(PL_globalstash, "readline", FALSE))
11419 && (gv_readline = *gvp) && isGV_with_GP(gv_readline)
11420 && GvCVu(gv_readline) && GvIMPORTED_CV(gv_readline)))
11421 readline_overriden = TRUE;
11423 /* if <$fh>, create the ops to turn the variable into a
11427 /* try to find it in the pad for this block, otherwise find
11428 add symbol table ops
11430 const PADOFFSET tmp = pad_findmy(d);
11431 if (tmp != NOT_IN_PAD) {
11432 if (PAD_COMPNAME_FLAGS_isOUR(tmp)) {
11433 HV * const stash = PAD_COMPNAME_OURSTASH(tmp);
11434 HEK * const stashname = HvNAME_HEK(stash);
11435 SV * const sym = sv_2mortal(newSVhek(stashname));
11436 sv_catpvs(sym, "::");
11437 sv_catpv(sym, d+1);
11442 OP * const o = newOP(OP_PADSV, 0);
11444 PL_lex_op = readline_overriden
11445 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11446 append_elem(OP_LIST, o,
11447 newCVREF(0, newGVOP(OP_GV,0,gv_readline))))
11448 : (OP*)newUNOP(OP_READLINE, 0, o);
11457 ? (GV_ADDMULTI | GV_ADDINEVAL)
11460 PL_lex_op = readline_overriden
11461 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11462 append_elem(OP_LIST,
11463 newUNOP(OP_RV2SV, 0, newGVOP(OP_GV, 0, gv)),
11464 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11465 : (OP*)newUNOP(OP_READLINE, 0,
11466 newUNOP(OP_RV2SV, 0,
11467 newGVOP(OP_GV, 0, gv)));
11469 if (!readline_overriden)
11470 PL_lex_op->op_flags |= OPf_SPECIAL;
11471 /* we created the ops in PL_lex_op, so make yylval.ival a null op */
11472 yylval.ival = OP_NULL;
11475 /* If it's none of the above, it must be a literal filehandle
11476 (<Foo::BAR> or <FOO>) so build a simple readline OP */
11478 GV * const gv = gv_fetchpv(d, GV_ADD, SVt_PVIO);
11479 PL_lex_op = readline_overriden
11480 ? (OP*)newUNOP(OP_ENTERSUB, OPf_STACKED,
11481 append_elem(OP_LIST,
11482 newGVOP(OP_GV, 0, gv),
11483 newCVREF(0, newGVOP(OP_GV, 0, gv_readline))))
11484 : (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv));
11485 yylval.ival = OP_NULL;
11494 takes: start position in buffer
11495 keep_quoted preserve \ on the embedded delimiter(s)
11496 keep_delims preserve the delimiters around the string
11497 returns: position to continue reading from buffer
11498 side-effects: multi_start, multi_close, lex_repl or lex_stuff, and
11499 updates the read buffer.
11501 This subroutine pulls a string out of the input. It is called for:
11502 q single quotes q(literal text)
11503 ' single quotes 'literal text'
11504 qq double quotes qq(interpolate $here please)
11505 " double quotes "interpolate $here please"
11506 qx backticks qx(/bin/ls -l)
11507 ` backticks `/bin/ls -l`
11508 qw quote words @EXPORT_OK = qw( func() $spam )
11509 m// regexp match m/this/
11510 s/// regexp substitute s/this/that/
11511 tr/// string transliterate tr/this/that/
11512 y/// string transliterate y/this/that/
11513 ($*@) sub prototypes sub foo ($)
11514 (stuff) sub attr parameters sub foo : attr(stuff)
11515 <> readline or globs <FOO>, <>, <$fh>, or <*.c>
11517 In most of these cases (all but <>, patterns and transliterate)
11518 yylex() calls scan_str(). m// makes yylex() call scan_pat() which
11519 calls scan_str(). s/// makes yylex() call scan_subst() which calls
11520 scan_str(). tr/// and y/// make yylex() call scan_trans() which
11523 It skips whitespace before the string starts, and treats the first
11524 character as the delimiter. If the delimiter is one of ([{< then
11525 the corresponding "close" character )]}> is used as the closing
11526 delimiter. It allows quoting of delimiters, and if the string has
11527 balanced delimiters ([{<>}]) it allows nesting.
11529 On success, the SV with the resulting string is put into lex_stuff or,
11530 if that is already non-NULL, into lex_repl. The second case occurs only
11531 when parsing the RHS of the special constructs s/// and tr/// (y///).
11532 For convenience, the terminating delimiter character is stuffed into
11537 S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims)
11540 SV *sv; /* scalar value: string */
11541 const char *tmps; /* temp string, used for delimiter matching */
11542 register char *s = start; /* current position in the buffer */
11543 register char term; /* terminating character */
11544 register char *to; /* current position in the sv's data */
11545 I32 brackets = 1; /* bracket nesting level */
11546 bool has_utf8 = FALSE; /* is there any utf8 content? */
11547 I32 termcode; /* terminating char. code */
11548 U8 termstr[UTF8_MAXBYTES]; /* terminating string */
11549 STRLEN termlen; /* length of terminating string */
11550 int last_off = 0; /* last position for nesting bracket */
11556 /* skip space before the delimiter */
11562 if (PL_realtokenstart >= 0) {
11563 stuffstart = PL_realtokenstart;
11564 PL_realtokenstart = -1;
11567 stuffstart = start - SvPVX(PL_linestr);
11569 /* mark where we are, in case we need to report errors */
11572 /* after skipping whitespace, the next character is the terminator */
11575 termcode = termstr[0] = term;
11579 termcode = utf8_to_uvchr((U8*)s, &termlen);
11580 Copy(s, termstr, termlen, U8);
11581 if (!UTF8_IS_INVARIANT(term))
11585 /* mark where we are */
11586 PL_multi_start = CopLINE(PL_curcop);
11587 PL_multi_open = term;
11589 /* find corresponding closing delimiter */
11590 if (term && (tmps = strchr("([{< )]}> )]}>",term)))
11591 termcode = termstr[0] = term = tmps[5];
11593 PL_multi_close = term;
11595 /* create a new SV to hold the contents. 79 is the SV's initial length.
11596 What a random number. */
11597 sv = newSV_type(SVt_PVIV);
11599 SvIV_set(sv, termcode);
11600 (void)SvPOK_only(sv); /* validate pointer */
11602 /* move past delimiter and try to read a complete string */
11604 sv_catpvn(sv, s, termlen);
11607 tstart = SvPVX(PL_linestr) + stuffstart;
11608 if (!PL_thisopen && !keep_delims) {
11609 PL_thisopen = newSVpvn(tstart, s - tstart);
11610 stuffstart = s - SvPVX(PL_linestr);
11614 if (PL_encoding && !UTF) {
11618 int offset = s - SvPVX_const(PL_linestr);
11619 const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr,
11620 &offset, (char*)termstr, termlen);
11621 const char * const ns = SvPVX_const(PL_linestr) + offset;
11622 char * const svlast = SvEND(sv) - 1;
11624 for (; s < ns; s++) {
11625 if (*s == '\n' && !PL_rsfp)
11626 CopLINE_inc(PL_curcop);
11629 goto read_more_line;
11631 /* handle quoted delimiters */
11632 if (SvCUR(sv) > 1 && *(svlast-1) == '\\') {
11634 for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';)
11636 if ((svlast-1 - t) % 2) {
11637 if (!keep_quoted) {
11638 *(svlast-1) = term;
11640 SvCUR_set(sv, SvCUR(sv) - 1);
11645 if (PL_multi_open == PL_multi_close) {
11651 for (t = w = SvPVX(sv)+last_off; t < svlast; w++, t++) {
11652 /* At here, all closes are "was quoted" one,
11653 so we don't check PL_multi_close. */
11655 if (!keep_quoted && *(t+1) == PL_multi_open)
11660 else if (*t == PL_multi_open)
11668 SvCUR_set(sv, w - SvPVX_const(sv));
11670 last_off = w - SvPVX(sv);
11671 if (--brackets <= 0)
11676 if (!keep_delims) {
11677 SvCUR_set(sv, SvCUR(sv) - 1);
11683 /* extend sv if need be */
11684 SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1);
11685 /* set 'to' to the next character in the sv's string */
11686 to = SvPVX(sv)+SvCUR(sv);
11688 /* if open delimiter is the close delimiter read unbridle */
11689 if (PL_multi_open == PL_multi_close) {
11690 for (; s < PL_bufend; s++,to++) {
11691 /* embedded newlines increment the current line number */
11692 if (*s == '\n' && !PL_rsfp)
11693 CopLINE_inc(PL_curcop);
11694 /* handle quoted delimiters */
11695 if (*s == '\\' && s+1 < PL_bufend && term != '\\') {
11696 if (!keep_quoted && s[1] == term)
11698 /* any other quotes are simply copied straight through */
11702 /* terminate when run out of buffer (the for() condition), or
11703 have found the terminator */
11704 else if (*s == term) {
11707 if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen))
11710 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11716 /* if the terminator isn't the same as the start character (e.g.,
11717 matched brackets), we have to allow more in the quoting, and
11718 be prepared for nested brackets.
11721 /* read until we run out of string, or we find the terminator */
11722 for (; s < PL_bufend; s++,to++) {
11723 /* embedded newlines increment the line count */
11724 if (*s == '\n' && !PL_rsfp)
11725 CopLINE_inc(PL_curcop);
11726 /* backslashes can escape the open or closing characters */
11727 if (*s == '\\' && s+1 < PL_bufend) {
11728 if (!keep_quoted &&
11729 ((s[1] == PL_multi_open) || (s[1] == PL_multi_close)))
11734 /* allow nested opens and closes */
11735 else if (*s == PL_multi_close && --brackets <= 0)
11737 else if (*s == PL_multi_open)
11739 else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF)
11744 /* terminate the copied string and update the sv's end-of-string */
11746 SvCUR_set(sv, to - SvPVX_const(sv));
11749 * this next chunk reads more into the buffer if we're not done yet
11753 break; /* handle case where we are done yet :-) */
11755 #ifndef PERL_STRICT_CR
11756 if (to - SvPVX_const(sv) >= 2) {
11757 if ((to[-2] == '\r' && to[-1] == '\n') ||
11758 (to[-2] == '\n' && to[-1] == '\r'))
11762 SvCUR_set(sv, to - SvPVX_const(sv));
11764 else if (to[-1] == '\r')
11767 else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r')
11772 /* if we're out of file, or a read fails, bail and reset the current
11773 line marker so we can report where the unterminated string began
11776 if (PL_madskills) {
11777 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11779 sv_catpvn(PL_thisstuff, tstart, PL_bufend - tstart);
11781 PL_thisstuff = newSVpvn(tstart, PL_bufend - tstart);
11785 !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) {
11787 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
11793 /* we read a line, so increment our line counter */
11794 CopLINE_inc(PL_curcop);
11796 /* update debugger info */
11797 if (PERLDB_LINE && PL_curstash != PL_debstash)
11798 update_debugger_info(PL_linestr, NULL, 0);
11800 /* having changed the buffer, we must update PL_bufend */
11801 PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
11802 PL_last_lop = PL_last_uni = NULL;
11805 /* at this point, we have successfully read the delimited string */
11807 if (!PL_encoding || UTF) {
11809 if (PL_madskills) {
11810 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11811 const int len = s - tstart;
11813 sv_catpvn(PL_thisstuff, tstart, len);
11815 PL_thisstuff = newSVpvn(tstart, len);
11816 if (!PL_thisclose && !keep_delims)
11817 PL_thisclose = newSVpvn(s,termlen);
11822 sv_catpvn(sv, s, termlen);
11827 if (PL_madskills) {
11828 char * const tstart = SvPVX(PL_linestr) + stuffstart;
11829 const int len = s - tstart - termlen;
11831 sv_catpvn(PL_thisstuff, tstart, len);
11833 PL_thisstuff = newSVpvn(tstart, len);
11834 if (!PL_thisclose && !keep_delims)
11835 PL_thisclose = newSVpvn(s - termlen,termlen);
11839 if (has_utf8 || PL_encoding)
11842 PL_multi_end = CopLINE(PL_curcop);
11844 /* if we allocated too much space, give some back */
11845 if (SvCUR(sv) + 5 < SvLEN(sv)) {
11846 SvLEN_set(sv, SvCUR(sv) + 1);
11847 SvPV_renew(sv, SvLEN(sv));
11850 /* decide whether this is the first or second quoted string we've read
11863 takes: pointer to position in buffer
11864 returns: pointer to new position in buffer
11865 side-effects: builds ops for the constant in yylval.op
11867 Read a number in any of the formats that Perl accepts:
11869 \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12.
11870 \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34
11873 0x[0-9A-Fa-f](_?[0-9A-Fa-f])*
11875 Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the
11878 If it reads a number without a decimal point or an exponent, it will
11879 try converting the number to an integer and see if it can do so
11880 without loss of precision.
11884 Perl_scan_num(pTHX_ const char *start, YYSTYPE* lvalp)
11887 register const char *s = start; /* current position in buffer */
11888 register char *d; /* destination in temp buffer */
11889 register char *e; /* end of temp buffer */
11890 NV nv; /* number read, as a double */
11891 SV *sv = NULL; /* place to put the converted number */
11892 bool floatit; /* boolean: int or float? */
11893 const char *lastub = NULL; /* position of last underbar */
11894 static char const number_too_long[] = "Number too long";
11896 /* We use the first character to decide what type of number this is */
11900 Perl_croak(aTHX_ "panic: scan_num");
11902 /* if it starts with a 0, it could be an octal number, a decimal in
11903 0.13 disguise, or a hexadecimal number, or a binary number. */
11907 u holds the "number so far"
11908 shift the power of 2 of the base
11909 (hex == 4, octal == 3, binary == 1)
11910 overflowed was the number more than we can hold?
11912 Shift is used when we add a digit. It also serves as an "are
11913 we in octal/hex/binary?" indicator to disallow hex characters
11914 when in octal mode.
11919 bool overflowed = FALSE;
11920 bool just_zero = TRUE; /* just plain 0 or binary number? */
11921 static const NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 };
11922 static const char* const bases[5] =
11923 { "", "binary", "", "octal", "hexadecimal" };
11924 static const char* const Bases[5] =
11925 { "", "Binary", "", "Octal", "Hexadecimal" };
11926 static const char* const maxima[5] =
11928 "0b11111111111111111111111111111111",
11932 const char *base, *Base, *max;
11934 /* check for hex */
11939 } else if (s[1] == 'b') {
11944 /* check for a decimal in disguise */
11945 else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E')
11947 /* so it must be octal */
11954 if (ckWARN(WARN_SYNTAX))
11955 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11956 "Misplaced _ in number");
11960 base = bases[shift];
11961 Base = Bases[shift];
11962 max = maxima[shift];
11964 /* read the rest of the number */
11966 /* x is used in the overflow test,
11967 b is the digit we're adding on. */
11972 /* if we don't mention it, we're done */
11976 /* _ are ignored -- but warned about if consecutive */
11978 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
11979 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
11980 "Misplaced _ in number");
11984 /* 8 and 9 are not octal */
11985 case '8': case '9':
11987 yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s));
11991 case '2': case '3': case '4':
11992 case '5': case '6': case '7':
11994 yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s));
11997 case '0': case '1':
11998 b = *s++ & 15; /* ASCII digit -> value of digit */
12002 case 'a': case 'b': case 'c': case 'd': case 'e': case 'f':
12003 case 'A': case 'B': case 'C': case 'D': case 'E': case 'F':
12004 /* make sure they said 0x */
12007 b = (*s++ & 7) + 9;
12009 /* Prepare to put the digit we have onto the end
12010 of the number so far. We check for overflows.
12016 x = u << shift; /* make room for the digit */
12018 if ((x >> shift) != u
12019 && !(PL_hints & HINT_NEW_BINARY)) {
12022 if (ckWARN_d(WARN_OVERFLOW))
12023 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12024 "Integer overflow in %s number",
12027 u = x | b; /* add the digit to the end */
12030 n *= nvshift[shift];
12031 /* If an NV has not enough bits in its
12032 * mantissa to represent an UV this summing of
12033 * small low-order numbers is a waste of time
12034 * (because the NV cannot preserve the
12035 * low-order bits anyway): we could just
12036 * remember when did we overflow and in the
12037 * end just multiply n by the right
12045 /* if we get here, we had success: make a scalar value from
12050 /* final misplaced underbar check */
12051 if (s[-1] == '_') {
12052 if (ckWARN(WARN_SYNTAX))
12053 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12058 if (n > 4294967295.0 && ckWARN(WARN_PORTABLE))
12059 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12060 "%s number > %s non-portable",
12066 if (u > 0xffffffff && ckWARN(WARN_PORTABLE))
12067 Perl_warner(aTHX_ packWARN(WARN_PORTABLE),
12068 "%s number > %s non-portable",
12073 if (just_zero && (PL_hints & HINT_NEW_INTEGER))
12074 sv = new_constant(start, s - start, "integer",
12076 else if (PL_hints & HINT_NEW_BINARY)
12077 sv = new_constant(start, s - start, "binary", sv, NULL, NULL);
12082 handle decimal numbers.
12083 we're also sent here when we read a 0 as the first digit
12085 case '1': case '2': case '3': case '4': case '5':
12086 case '6': case '7': case '8': case '9': case '.':
12089 e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */
12092 /* read next group of digits and _ and copy into d */
12093 while (isDIGIT(*s) || *s == '_') {
12094 /* skip underscores, checking for misplaced ones
12098 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12099 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12100 "Misplaced _ in number");
12104 /* check for end of fixed-length buffer */
12106 Perl_croak(aTHX_ number_too_long);
12107 /* if we're ok, copy the character */
12112 /* final misplaced underbar check */
12113 if (lastub && s == lastub + 1) {
12114 if (ckWARN(WARN_SYNTAX))
12115 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "Misplaced _ in number");
12118 /* read a decimal portion if there is one. avoid
12119 3..5 being interpreted as the number 3. followed
12122 if (*s == '.' && s[1] != '.') {
12127 if (ckWARN(WARN_SYNTAX))
12128 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12129 "Misplaced _ in number");
12133 /* copy, ignoring underbars, until we run out of digits.
12135 for (; isDIGIT(*s) || *s == '_'; s++) {
12136 /* fixed length buffer check */
12138 Perl_croak(aTHX_ number_too_long);
12140 if (lastub && s == lastub + 1 && ckWARN(WARN_SYNTAX))
12141 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12142 "Misplaced _ in number");
12148 /* fractional part ending in underbar? */
12149 if (s[-1] == '_') {
12150 if (ckWARN(WARN_SYNTAX))
12151 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12152 "Misplaced _ in number");
12154 if (*s == '.' && isDIGIT(s[1])) {
12155 /* oops, it's really a v-string, but without the "v" */
12161 /* read exponent part, if present */
12162 if ((*s == 'e' || *s == 'E') && strchr("+-0123456789_", s[1])) {
12166 /* regardless of whether user said 3E5 or 3e5, use lower 'e' */
12167 *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */
12169 /* stray preinitial _ */
12171 if (ckWARN(WARN_SYNTAX))
12172 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12173 "Misplaced _ in number");
12177 /* allow positive or negative exponent */
12178 if (*s == '+' || *s == '-')
12181 /* stray initial _ */
12183 if (ckWARN(WARN_SYNTAX))
12184 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12185 "Misplaced _ in number");
12189 /* read digits of exponent */
12190 while (isDIGIT(*s) || *s == '_') {
12193 Perl_croak(aTHX_ number_too_long);
12197 if (((lastub && s == lastub + 1) ||
12198 (!isDIGIT(s[1]) && s[1] != '_'))
12199 && ckWARN(WARN_SYNTAX))
12200 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
12201 "Misplaced _ in number");
12208 /* make an sv from the string */
12212 We try to do an integer conversion first if no characters
12213 indicating "float" have been found.
12218 const int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv);
12220 if (flags == IS_NUMBER_IN_UV) {
12222 sv_setiv(sv, uv); /* Prefer IVs over UVs. */
12225 } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) {
12226 if (uv <= (UV) IV_MIN)
12227 sv_setiv(sv, -(IV)uv);
12234 /* terminate the string */
12236 nv = Atof(PL_tokenbuf);
12240 if ( floatit ? (PL_hints & HINT_NEW_FLOAT) :
12241 (PL_hints & HINT_NEW_INTEGER) )
12242 sv = new_constant(PL_tokenbuf,
12245 (floatit ? "float" : "integer"),
12249 /* if it starts with a v, it could be a v-string */
12252 sv = newSV(5); /* preallocate storage space */
12253 s = scan_vstring(s, PL_bufend, sv);
12257 /* make the op for the constant and return */
12260 lvalp->opval = newSVOP(OP_CONST, 0, sv);
12262 lvalp->opval = NULL;
12268 S_scan_formline(pTHX_ register char *s)
12271 register char *eol;
12273 SV * const stuff = newSVpvs("");
12274 bool needargs = FALSE;
12275 bool eofmt = FALSE;
12277 char *tokenstart = s;
12280 if (PL_madskills) {
12281 savewhite = PL_thiswhite;
12286 while (!needargs) {
12289 #ifdef PERL_STRICT_CR
12290 while (SPACE_OR_TAB(*t))
12293 while (SPACE_OR_TAB(*t) || *t == '\r')
12296 if (*t == '\n' || t == PL_bufend) {
12301 if (PL_in_eval && !PL_rsfp) {
12302 eol = (char *) memchr(s,'\n',PL_bufend-s);
12307 eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr);
12309 for (t = s; t < eol; t++) {
12310 if (*t == '~' && t[1] == '~' && SvCUR(stuff)) {
12312 goto enough; /* ~~ must be first line in formline */
12314 if (*t == '@' || *t == '^')
12318 sv_catpvn(stuff, s, eol-s);
12319 #ifndef PERL_STRICT_CR
12320 if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') {
12321 char *end = SvPVX(stuff) + SvCUR(stuff);
12324 SvCUR_set(stuff, SvCUR(stuff) - 1);
12334 if (PL_madskills) {
12336 sv_catpvn(PL_thistoken, tokenstart, PL_bufend - tokenstart);
12338 PL_thistoken = newSVpvn(tokenstart, PL_bufend - tokenstart);
12341 s = filter_gets(PL_linestr, PL_rsfp, 0);
12343 tokenstart = PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12345 PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr);
12347 PL_bufend = PL_bufptr + SvCUR(PL_linestr);
12348 PL_last_lop = PL_last_uni = NULL;
12357 if (SvCUR(stuff)) {
12360 PL_lex_state = LEX_NORMAL;
12361 start_force(PL_curforce);
12362 NEXTVAL_NEXTTOKE.ival = 0;
12366 PL_lex_state = LEX_FORMLINE;
12368 if (UTF && is_utf8_string((U8*)SvPVX_const(stuff), SvCUR(stuff)))
12370 else if (PL_encoding)
12371 sv_recode_to_utf8(stuff, PL_encoding);
12373 start_force(PL_curforce);
12374 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0, stuff);
12376 start_force(PL_curforce);
12377 NEXTVAL_NEXTTOKE.ival = OP_FORMLINE;
12381 SvREFCNT_dec(stuff);
12383 PL_lex_formbrack = 0;
12387 if (PL_madskills) {
12389 sv_catpvn(PL_thistoken, tokenstart, s - tokenstart);
12391 PL_thistoken = newSVpvn(tokenstart, s - tokenstart);
12392 PL_thiswhite = savewhite;
12399 Perl_start_subparse(pTHX_ I32 is_format, U32 flags)
12402 const I32 oldsavestack_ix = PL_savestack_ix;
12403 CV* const outsidecv = PL_compcv;
12406 assert(SvTYPE(PL_compcv) == SVt_PVCV);
12408 SAVEI32(PL_subline);
12409 save_item(PL_subname);
12410 SAVESPTR(PL_compcv);
12412 PL_compcv = (CV*)newSV_type(is_format ? SVt_PVFM : SVt_PVCV);
12413 CvFLAGS(PL_compcv) |= flags;
12415 PL_subline = CopLINE(PL_curcop);
12416 CvPADLIST(PL_compcv) = pad_new(padnew_SAVE|padnew_SAVESUB);
12417 CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc_simple(outsidecv);
12418 CvOUTSIDE_SEQ(PL_compcv) = PL_cop_seqmax;
12420 return oldsavestack_ix;
12424 #pragma segment Perl_yylex
12427 Perl_yywarn(pTHX_ const char *s)
12430 PL_in_eval |= EVAL_WARNONLY;
12432 PL_in_eval &= ~EVAL_WARNONLY;
12437 Perl_yyerror(pTHX_ const char *s)
12440 const char *where = NULL;
12441 const char *context = NULL;
12444 int yychar = PL_parser->yychar;
12446 if (!yychar || (yychar == ';' && !PL_rsfp))
12448 else if (PL_oldoldbufptr && PL_bufptr > PL_oldoldbufptr &&
12449 PL_bufptr - PL_oldoldbufptr < 200 && PL_oldoldbufptr != PL_oldbufptr &&
12450 PL_oldbufptr != PL_bufptr) {
12453 The code below is removed for NetWare because it abends/crashes on NetWare
12454 when the script has error such as not having the closing quotes like:
12455 if ($var eq "value)
12456 Checking of white spaces is anyway done in NetWare code.
12459 while (isSPACE(*PL_oldoldbufptr))
12462 context = PL_oldoldbufptr;
12463 contlen = PL_bufptr - PL_oldoldbufptr;
12465 else if (PL_oldbufptr && PL_bufptr > PL_oldbufptr &&
12466 PL_bufptr - PL_oldbufptr < 200 && PL_oldbufptr != PL_bufptr) {
12469 The code below is removed for NetWare because it abends/crashes on NetWare
12470 when the script has error such as not having the closing quotes like:
12471 if ($var eq "value)
12472 Checking of white spaces is anyway done in NetWare code.
12475 while (isSPACE(*PL_oldbufptr))
12478 context = PL_oldbufptr;
12479 contlen = PL_bufptr - PL_oldbufptr;
12481 else if (yychar > 255)
12482 where = "next token ???";
12483 else if (yychar == -2) { /* YYEMPTY */
12484 if (PL_lex_state == LEX_NORMAL ||
12485 (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL))
12486 where = "at end of line";
12487 else if (PL_lex_inpat)
12488 where = "within pattern";
12490 where = "within string";
12493 SV * const where_sv = sv_2mortal(newSVpvs("next char "));
12495 Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar));
12496 else if (isPRINT_LC(yychar)) {
12497 const unsigned char string = (unsigned char) yychar;
12498 sv_catpvn(where_sv, &string, 1);
12501 Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255);
12502 where = SvPVX_const(where_sv);
12504 msg = sv_2mortal(newSVpv(s, 0));
12505 Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ",
12506 OutCopFILE(PL_curcop), (IV)CopLINE(PL_curcop));
12508 Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context);
12510 Perl_sv_catpvf(aTHX_ msg, "%s\n", where);
12511 if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) {
12512 Perl_sv_catpvf(aTHX_ msg,
12513 " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n",
12514 (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start);
12517 if (PL_in_eval & EVAL_WARNONLY) {
12518 if (ckWARN_d(WARN_SYNTAX))
12519 Perl_warner(aTHX_ packWARN(WARN_SYNTAX), "%"SVf, SVfARG(msg));
12523 if (PL_error_count >= 10) {
12524 if (PL_in_eval && SvCUR(ERRSV))
12525 Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n",
12526 SVfARG(ERRSV), OutCopFILE(PL_curcop));
12528 Perl_croak(aTHX_ "%s has too many errors.\n",
12529 OutCopFILE(PL_curcop));
12532 PL_in_my_stash = NULL;
12536 #pragma segment Main
12540 S_swallow_bom(pTHX_ U8 *s)
12543 const STRLEN slen = SvCUR(PL_linestr);
12546 if (s[1] == 0xFE) {
12547 /* UTF-16 little-endian? (or UTF32-LE?) */
12548 if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */
12549 Perl_croak(aTHX_ "Unsupported script encoding UTF32-LE");
12550 #ifndef PERL_NO_UTF16_FILTER
12551 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF16-LE script encoding (BOM)\n");
12554 if (PL_bufend > (char*)s) {
12558 filter_add(utf16rev_textfilter, NULL);
12559 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12560 utf16_to_utf8_reversed(s, news,
12561 PL_bufend - (char*)s - 1,
12563 sv_setpvn(PL_linestr, (const char*)news, newlen);
12565 s = (U8*)SvPVX(PL_linestr);
12566 Copy(news, s, newlen, U8);
12570 SvUTF8_on(PL_linestr);
12571 s = (U8*)SvPVX(PL_linestr);
12573 /* FIXME - is this a general bug fix? */
12576 PL_bufend = SvPVX(PL_linestr) + newlen;
12579 Perl_croak(aTHX_ "Unsupported script encoding UTF16-LE");
12584 if (s[1] == 0xFF) { /* UTF-16 big-endian? */
12585 #ifndef PERL_NO_UTF16_FILTER
12586 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (BOM)\n");
12589 if (PL_bufend > (char *)s) {
12593 filter_add(utf16_textfilter, NULL);
12594 Newx(news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8);
12595 utf16_to_utf8(s, news,
12596 PL_bufend - (char*)s,
12598 sv_setpvn(PL_linestr, (const char*)news, newlen);
12600 SvUTF8_on(PL_linestr);
12601 s = (U8*)SvPVX(PL_linestr);
12602 PL_bufend = SvPVX(PL_linestr) + newlen;
12605 Perl_croak(aTHX_ "Unsupported script encoding UTF16-BE");
12610 if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) {
12611 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12612 s += 3; /* UTF-8 */
12618 if (s[2] == 0xFE && s[3] == 0xFF) {
12619 /* UTF-32 big-endian */
12620 Perl_croak(aTHX_ "Unsupported script encoding UTF32-BE");
12623 else if (s[2] == 0 && s[3] != 0) {
12626 * are a good indicator of UTF-16BE. */
12627 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding (no BOM)\n");
12633 if (slen > 3 && s[1] == 0x73 && s[2] == 0x66 && s[3] == 0x73) {
12634 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-8 script encoding (BOM)\n");
12635 s += 4; /* UTF-8 */
12641 if (slen > 3 && s[1] == 0 && s[2] != 0 && s[3] == 0) {
12644 * are a good indicator of UTF-16LE. */
12645 if (DEBUG_p_TEST || DEBUG_T_TEST) PerlIO_printf(Perl_debug_log, "UTF-16LE script encoding (no BOM)\n");
12653 #ifndef PERL_NO_UTF16_FILTER
12655 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12658 const STRLEN old = SvCUR(sv);
12659 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12660 DEBUG_P(PerlIO_printf(Perl_debug_log,
12661 "utf16_textfilter(%p): %d %d (%d)\n",
12662 FPTR2DPTR(void *, utf16_textfilter),
12663 idx, maxlen, (int) count));
12667 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12668 Copy(SvPVX_const(sv), tmps, old, char);
12669 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12670 SvCUR(sv) - old, &newlen);
12671 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12673 DEBUG_P({sv_dump(sv);});
12678 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen)
12681 const STRLEN old = SvCUR(sv);
12682 const I32 count = FILTER_READ(idx+1, sv, maxlen);
12683 DEBUG_P(PerlIO_printf(Perl_debug_log,
12684 "utf16rev_textfilter(%p): %d %d (%d)\n",
12685 FPTR2DPTR(void *, utf16rev_textfilter),
12686 idx, maxlen, (int) count));
12690 Newx(tmps, SvCUR(sv) * 3 / 2 + 1, U8);
12691 Copy(SvPVX_const(sv), tmps, old, char);
12692 utf16_to_utf8((U8*)SvPVX_const(sv) + old, tmps + old,
12693 SvCUR(sv) - old, &newlen);
12694 sv_usepvn(sv, (char*)tmps, (STRLEN)newlen + old);
12696 DEBUG_P({ sv_dump(sv); });
12702 Returns a pointer to the next character after the parsed
12703 vstring, as well as updating the passed in sv.
12705 Function must be called like
12708 s = scan_vstring(s,e,sv);
12710 where s and e are the start and end of the string.
12711 The sv should already be large enough to store the vstring
12712 passed in, for performance reasons.
12717 Perl_scan_vstring(pTHX_ const char *s, const char *e, SV *sv)
12720 const char *pos = s;
12721 const char *start = s;
12722 if (*pos == 'v') pos++; /* get past 'v' */
12723 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12725 if ( *pos != '.') {
12726 /* this may not be a v-string if followed by => */
12727 const char *next = pos;
12728 while (next < e && isSPACE(*next))
12730 if ((e - next) >= 2 && *next == '=' && next[1] == '>' ) {
12731 /* return string not v-string */
12732 sv_setpvn(sv,(char *)s,pos-s);
12733 return (char *)pos;
12737 if (!isALPHA(*pos)) {
12738 U8 tmpbuf[UTF8_MAXBYTES+1];
12741 s++; /* get past 'v' */
12743 sv_setpvn(sv, "", 0);
12746 /* this is atoi() that tolerates underscores */
12749 const char *end = pos;
12751 while (--end >= s) {
12753 const UV orev = rev;
12754 rev += (*end - '0') * mult;
12756 if (orev > rev && ckWARN_d(WARN_OVERFLOW))
12757 Perl_warner(aTHX_ packWARN(WARN_OVERFLOW),
12758 "Integer overflow in decimal number");
12762 if (rev > 0x7FFFFFFF)
12763 Perl_croak(aTHX_ "In EBCDIC the v-string components cannot exceed 2147483647");
12765 /* Append native character for the rev point */
12766 tmpend = uvchr_to_utf8(tmpbuf, rev);
12767 sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf);
12768 if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev)))
12770 if (pos + 1 < e && *pos == '.' && isDIGIT(pos[1]))
12776 while (pos < e && (isDIGIT(*pos) || *pos == '_'))
12780 sv_magic(sv,NULL,PERL_MAGIC_vstring,(const char*)start, pos-start);
12788 * c-indentation-style: bsd
12789 * c-basic-offset: 4
12790 * indent-tabs-mode: t
12793 * ex: set ts=8 sts=4 sw=4 noet: