This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
toke.c: Add comment
[perl5.git] / toke.c
CommitLineData
a0d0e21e 1/* toke.c
a687059c 2 *
1129b882
NC
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
a687059c 5 *
d48672a2
LW
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.
378cc40b 8 *
a0d0e21e
LW
9 */
10
11/*
4ac71550
TC
12 * 'It all comes from here, the stench and the peril.' --Frodo
13 *
14 * [p.719 of _The Lord of the Rings_, IV/ix: "Shelob's Lair"]
378cc40b
LW
15 */
16
9cbb5ea2
GS
17/*
18 * This file is the lexer for Perl. It's closely linked to the
4e553d73 19 * parser, perly.y.
ffb4593c
NT
20 *
21 * The main routine is yylex(), which returns the next token.
22 */
23
f0e67a1d
Z
24/*
25=head1 Lexer interface
26
27This is the lower layer of the Perl parser, managing characters and tokens.
28
29=for apidoc AmU|yy_parser *|PL_parser
30
31Pointer to a structure encapsulating the state of the parsing operation
32currently in progress. The pointer can be locally changed to perform
33a nested parse without interfering with the state of an outer parse.
34Individual members of C<PL_parser> have their own documentation.
35
36=cut
37*/
38
378cc40b 39#include "EXTERN.h"
864dbfa3 40#define PERL_IN_TOKE_C
378cc40b 41#include "perl.h"
04e98a4d 42#include "dquote_static.c"
378cc40b 43
eb0d8d16
NC
44#define new_constant(a,b,c,d,e,f,g) \
45 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
46
6154021b 47#define pl_yylval (PL_parser->yylval)
d3b6f988 48
199e78b7
DM
49/* XXX temporary backwards compatibility */
50#define PL_lex_brackets (PL_parser->lex_brackets)
78cdf107
Z
51#define PL_lex_allbrackets (PL_parser->lex_allbrackets)
52#define PL_lex_fakeeof (PL_parser->lex_fakeeof)
199e78b7
DM
53#define PL_lex_brackstack (PL_parser->lex_brackstack)
54#define PL_lex_casemods (PL_parser->lex_casemods)
55#define PL_lex_casestack (PL_parser->lex_casestack)
56#define PL_lex_defer (PL_parser->lex_defer)
57#define PL_lex_dojoin (PL_parser->lex_dojoin)
58#define PL_lex_expect (PL_parser->lex_expect)
59#define PL_lex_formbrack (PL_parser->lex_formbrack)
60#define PL_lex_inpat (PL_parser->lex_inpat)
61#define PL_lex_inwhat (PL_parser->lex_inwhat)
62#define PL_lex_op (PL_parser->lex_op)
63#define PL_lex_repl (PL_parser->lex_repl)
64#define PL_lex_starts (PL_parser->lex_starts)
65#define PL_lex_stuff (PL_parser->lex_stuff)
66#define PL_multi_start (PL_parser->multi_start)
67#define PL_multi_open (PL_parser->multi_open)
68#define PL_multi_close (PL_parser->multi_close)
69#define PL_pending_ident (PL_parser->pending_ident)
70#define PL_preambled (PL_parser->preambled)
71#define PL_sublex_info (PL_parser->sublex_info)
bdc0bf6f 72#define PL_linestr (PL_parser->linestr)
c2598295
DM
73#define PL_expect (PL_parser->expect)
74#define PL_copline (PL_parser->copline)
f06b5848
DM
75#define PL_bufptr (PL_parser->bufptr)
76#define PL_oldbufptr (PL_parser->oldbufptr)
77#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
78#define PL_linestart (PL_parser->linestart)
79#define PL_bufend (PL_parser->bufend)
80#define PL_last_uni (PL_parser->last_uni)
81#define PL_last_lop (PL_parser->last_lop)
82#define PL_last_lop_op (PL_parser->last_lop_op)
bc177e6b 83#define PL_lex_state (PL_parser->lex_state)
2f9285f8 84#define PL_rsfp (PL_parser->rsfp)
5486870f 85#define PL_rsfp_filters (PL_parser->rsfp_filters)
12bd6ede
DM
86#define PL_in_my (PL_parser->in_my)
87#define PL_in_my_stash (PL_parser->in_my_stash)
14047fc9 88#define PL_tokenbuf (PL_parser->tokenbuf)
670a9cb2 89#define PL_multi_end (PL_parser->multi_end)
13765c85 90#define PL_error_count (PL_parser->error_count)
199e78b7
DM
91
92#ifdef PERL_MAD
93# define PL_endwhite (PL_parser->endwhite)
94# define PL_faketokens (PL_parser->faketokens)
95# define PL_lasttoke (PL_parser->lasttoke)
96# define PL_nextwhite (PL_parser->nextwhite)
97# define PL_realtokenstart (PL_parser->realtokenstart)
98# define PL_skipwhite (PL_parser->skipwhite)
99# define PL_thisclose (PL_parser->thisclose)
100# define PL_thismad (PL_parser->thismad)
101# define PL_thisopen (PL_parser->thisopen)
102# define PL_thisstuff (PL_parser->thisstuff)
103# define PL_thistoken (PL_parser->thistoken)
104# define PL_thiswhite (PL_parser->thiswhite)
fb205e7a
DM
105# define PL_thiswhite (PL_parser->thiswhite)
106# define PL_nexttoke (PL_parser->nexttoke)
107# define PL_curforce (PL_parser->curforce)
108#else
109# define PL_nexttoke (PL_parser->nexttoke)
110# define PL_nexttype (PL_parser->nexttype)
111# define PL_nextval (PL_parser->nextval)
199e78b7
DM
112#endif
113
16173588
NC
114/* This can't be done with embed.fnc, because struct yy_parser contains a
115 member named pending_ident, which clashes with the generated #define */
3cbf51f5
DM
116static int
117S_pending_ident(pTHX);
199e78b7 118
0bd48802 119static const char ident_too_long[] = "Identifier too long";
8903cb82 120
29595ff2 121#ifdef PERL_MAD
29595ff2 122# define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
cd81e915 123# define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
9ded7720 124#else
5db06880 125# define CURMAD(slot,sv)
9ded7720 126# define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
29595ff2
NC
127#endif
128
a7aaec61
Z
129#define XENUMMASK 0x3f
130#define XFAKEEOF 0x40
131#define XFAKEBRACK 0x80
9059aa12 132
39e02b42
JH
133#ifdef USE_UTF8_SCRIPTS
134# define UTF (!IN_BYTES)
2b9d42f0 135#else
746b446a 136# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8))
2b9d42f0 137#endif
a0ed51b3 138
b1fc3636
CJ
139/* The maximum number of characters preceding the unrecognized one to display */
140#define UNRECOGNIZED_PRECEDE_COUNT 10
141
61f0cdd9 142/* In variables named $^X, these are the legal values for X.
2b92dfce
GS
143 * 1999-02-27 mjd-perl-patch@plover.com */
144#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
145
bf4acbe4 146#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
bf4acbe4 147
ffb4593c
NT
148/* LEX_* are values for PL_lex_state, the state of the lexer.
149 * They are arranged oddly so that the guard on the switch statement
79072805
LW
150 * can get by with a single comparison (if the compiler is smart enough).
151 */
152
fb73857a
PP
153/* #define LEX_NOTPARSING 11 is done in perl.h. */
154
b6007c36
DM
155#define LEX_NORMAL 10 /* normal code (ie not within "...") */
156#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
157#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
158#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
159#define LEX_INTERPSTART 6 /* expecting the start of a $var */
160
161 /* at end of code, eg "$x" followed by: */
162#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
163#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
164
165#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
166 string or after \E, $foo, etc */
167#define LEX_INTERPCONST 2 /* NOT USED */
168#define LEX_FORMLINE 1 /* expecting a format line */
169#define LEX_KNOWNEXT 0 /* next token known; just return it */
170
79072805 171
bbf60fe6 172#ifdef DEBUGGING
27da23d5 173static const char* const lex_state_names[] = {
bbf60fe6
DM
174 "KNOWNEXT",
175 "FORMLINE",
176 "INTERPCONST",
177 "INTERPCONCAT",
178 "INTERPENDMAYBE",
179 "INTERPEND",
180 "INTERPSTART",
181 "INTERPPUSH",
182 "INTERPCASEMOD",
183 "INTERPNORMAL",
184 "NORMAL"
185};
186#endif
187
79072805
LW
188#ifdef ff_next
189#undef ff_next
d48672a2
LW
190#endif
191
79072805 192#include "keywords.h"
fe14fcc3 193
ffb4593c
NT
194/* CLINE is a macro that ensures PL_copline has a sane value */
195
ae986130
LW
196#ifdef CLINE
197#undef CLINE
198#endif
57843af0 199#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
3280af22 200
5db06880 201#ifdef PERL_MAD
29595ff2
NC
202# define SKIPSPACE0(s) skipspace0(s)
203# define SKIPSPACE1(s) skipspace1(s)
204# define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
205# define PEEKSPACE(s) skipspace2(s,0)
206#else
207# define SKIPSPACE0(s) skipspace(s)
208# define SKIPSPACE1(s) skipspace(s)
209# define SKIPSPACE2(s,tsv) skipspace(s)
210# define PEEKSPACE(s) skipspace(s)
211#endif
212
ffb4593c
NT
213/*
214 * Convenience functions to return different tokens and prime the
9cbb5ea2 215 * lexer for the next token. They all take an argument.
ffb4593c
NT
216 *
217 * TOKEN : generic token (used for '(', DOLSHARP, etc)
218 * OPERATOR : generic operator
219 * AOPERATOR : assignment operator
220 * PREBLOCK : beginning the block after an if, while, foreach, ...
221 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
222 * PREREF : *EXPR where EXPR is not a simple identifier
223 * TERM : expression term
224 * LOOPX : loop exiting command (goto, last, dump, etc)
225 * FTST : file test operator
226 * FUN0 : zero-argument function
2d2e263d 227 * FUN1 : not used, except for not, which isn't a UNIOP
ffb4593c
NT
228 * BOop : bitwise or or xor
229 * BAop : bitwise and
230 * SHop : shift operator
231 * PWop : power operator
9cbb5ea2 232 * PMop : pattern-matching operator
ffb4593c
NT
233 * Aop : addition-level operator
234 * Mop : multiplication-level operator
235 * Eop : equality-testing operator
e5edeb50 236 * Rop : relational operator <= != gt
ffb4593c
NT
237 *
238 * Also see LOP and lop() below.
239 */
240
998054bd 241#ifdef DEBUGGING /* Serve -DT. */
704d4215 242# define REPORT(retval) tokereport((I32)retval, &pl_yylval)
998054bd 243#else
bbf60fe6 244# define REPORT(retval) (retval)
998054bd
SC
245#endif
246
bbf60fe6
DM
247#define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
248#define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
249#define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
250#define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
251#define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
252#define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
253#define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
6154021b
RGS
254#define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
255#define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
256#define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
257#define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
258#define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
259#define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
260#define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
261#define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
262#define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
263#define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
264#define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
265#define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
266#define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
2f3197b3 267
a687059c
LW
268/* This bit of chicanery makes a unary function followed by
269 * a parenthesis into a function with one argument, highest precedence.
6f33ba73
RGS
270 * The UNIDOR macro is for unary functions that can be followed by the //
271 * operator (such as C<shift // 0>).
a687059c 272 */
376fcdbf 273#define UNI2(f,x) { \
6154021b 274 pl_yylval.ival = f; \
376fcdbf
AL
275 PL_expect = x; \
276 PL_bufptr = s; \
277 PL_last_uni = PL_oldbufptr; \
278 PL_last_lop_op = f; \
279 if (*s == '(') \
280 return REPORT( (int)FUNC1 ); \
29595ff2 281 s = PEEKSPACE(s); \
376fcdbf
AL
282 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
283 }
6f33ba73
RGS
284#define UNI(f) UNI2(f,XTERM)
285#define UNIDOR(f) UNI2(f,XTERMORDORDOR)
a687059c 286
376fcdbf 287#define UNIBRACK(f) { \
6154021b 288 pl_yylval.ival = f; \
376fcdbf
AL
289 PL_bufptr = s; \
290 PL_last_uni = PL_oldbufptr; \
291 if (*s == '(') \
292 return REPORT( (int)FUNC1 ); \
29595ff2 293 s = PEEKSPACE(s); \
376fcdbf
AL
294 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
295 }
79072805 296
9f68db38 297/* grandfather return to old style */
78cdf107
Z
298#define OLDLOP(f) \
299 do { \
300 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) \
301 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; \
302 pl_yylval.ival = (f); \
303 PL_expect = XTERM; \
304 PL_bufptr = s; \
305 return (int)LSTOP; \
306 } while(0)
79072805 307
8fa7f367
JH
308#ifdef DEBUGGING
309
6154021b 310/* how to interpret the pl_yylval associated with the token */
bbf60fe6
DM
311enum token_type {
312 TOKENTYPE_NONE,
313 TOKENTYPE_IVAL,
6154021b 314 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
bbf60fe6
DM
315 TOKENTYPE_PVAL,
316 TOKENTYPE_OPVAL,
317 TOKENTYPE_GVVAL
318};
319
6d4a66ac
NC
320static struct debug_tokens {
321 const int token;
322 enum token_type type;
323 const char *name;
324} const debug_tokens[] =
9041c2e3 325{
bbf60fe6
DM
326 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
327 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
328 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
329 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
330 { ARROW, TOKENTYPE_NONE, "ARROW" },
331 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
332 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
333 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
334 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
335 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
0d863452 336 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
bbf60fe6
DM
337 { DO, TOKENTYPE_NONE, "DO" },
338 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
339 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
340 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
341 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
342 { ELSE, TOKENTYPE_NONE, "ELSE" },
343 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
344 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
345 { FOR, TOKENTYPE_IVAL, "FOR" },
346 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
347 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
348 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
349 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
350 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
351 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
0d863452 352 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
bbf60fe6
DM
353 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
354 { IF, TOKENTYPE_IVAL, "IF" },
355 { LABEL, TOKENTYPE_PVAL, "LABEL" },
356 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
357 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
358 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
359 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
360 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
361 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
362 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
363 { MY, TOKENTYPE_IVAL, "MY" },
364 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
365 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
366 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
367 { OROP, TOKENTYPE_IVAL, "OROP" },
368 { OROR, TOKENTYPE_NONE, "OROR" },
369 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
88e1f1a2
JV
370 { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" },
371 { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" },
bbf60fe6
DM
372 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
373 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
374 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
375 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
376 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
377 { PREINC, TOKENTYPE_NONE, "PREINC" },
378 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
379 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
380 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
381 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
382 { SUB, TOKENTYPE_NONE, "SUB" },
383 { THING, TOKENTYPE_OPVAL, "THING" },
384 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
385 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
386 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
387 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
388 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
389 { USE, TOKENTYPE_IVAL, "USE" },
0d863452 390 { WHEN, TOKENTYPE_IVAL, "WHEN" },
bbf60fe6
DM
391 { WHILE, TOKENTYPE_IVAL, "WHILE" },
392 { WORD, TOKENTYPE_OPVAL, "WORD" },
be25f609 393 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
c35e046a 394 { 0, TOKENTYPE_NONE, NULL }
bbf60fe6
DM
395};
396
6154021b 397/* dump the returned token in rv, plus any optional arg in pl_yylval */
998054bd 398
bbf60fe6 399STATIC int
704d4215 400S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
bbf60fe6 401{
97aff369 402 dVAR;
7918f24d
NC
403
404 PERL_ARGS_ASSERT_TOKEREPORT;
405
bbf60fe6 406 if (DEBUG_T_TEST) {
bd61b366 407 const char *name = NULL;
bbf60fe6 408 enum token_type type = TOKENTYPE_NONE;
f54cb97a 409 const struct debug_tokens *p;
396482e1 410 SV* const report = newSVpvs("<== ");
bbf60fe6 411
f54cb97a 412 for (p = debug_tokens; p->token; p++) {
bbf60fe6
DM
413 if (p->token == (int)rv) {
414 name = p->name;
415 type = p->type;
416 break;
417 }
418 }
419 if (name)
54667de8 420 Perl_sv_catpv(aTHX_ report, name);
bbf60fe6
DM
421 else if ((char)rv > ' ' && (char)rv < '~')
422 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
423 else if (!rv)
396482e1 424 sv_catpvs(report, "EOF");
bbf60fe6
DM
425 else
426 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
427 switch (type) {
428 case TOKENTYPE_NONE:
429 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
430 break;
431 case TOKENTYPE_IVAL:
704d4215 432 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
bbf60fe6
DM
433 break;
434 case TOKENTYPE_OPNUM:
435 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
704d4215 436 PL_op_name[lvalp->ival]);
bbf60fe6
DM
437 break;
438 case TOKENTYPE_PVAL:
704d4215 439 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
bbf60fe6
DM
440 break;
441 case TOKENTYPE_OPVAL:
704d4215 442 if (lvalp->opval) {
401441c0 443 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
704d4215
GG
444 PL_op_name[lvalp->opval->op_type]);
445 if (lvalp->opval->op_type == OP_CONST) {
b6007c36 446 Perl_sv_catpvf(aTHX_ report, " %s",
704d4215 447 SvPEEK(cSVOPx_sv(lvalp->opval)));
b6007c36
DM
448 }
449
450 }
401441c0 451 else
396482e1 452 sv_catpvs(report, "(opval=null)");
bbf60fe6
DM
453 break;
454 }
b6007c36 455 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
bbf60fe6
DM
456 };
457 return (int)rv;
998054bd
SC
458}
459
b6007c36
DM
460
461/* print the buffer with suitable escapes */
462
463STATIC void
15f169a1 464S_printbuf(pTHX_ const char *const fmt, const char *const s)
b6007c36 465{
396482e1 466 SV* const tmp = newSVpvs("");
7918f24d
NC
467
468 PERL_ARGS_ASSERT_PRINTBUF;
469
b6007c36
DM
470 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
471 SvREFCNT_dec(tmp);
472}
473
8fa7f367
JH
474#endif
475
8290c323
NC
476static int
477S_deprecate_commaless_var_list(pTHX) {
478 PL_expect = XTERM;
479 deprecate("comma-less variable list");
480 return REPORT(','); /* grandfather non-comma-format format */
481}
482
ffb4593c
NT
483/*
484 * S_ao
485 *
c963b151
BD
486 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
487 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
ffb4593c
NT
488 */
489
76e3520e 490STATIC int
cea2e8a9 491S_ao(pTHX_ int toketype)
a0d0e21e 492{
97aff369 493 dVAR;
3280af22
NIS
494 if (*PL_bufptr == '=') {
495 PL_bufptr++;
a0d0e21e 496 if (toketype == ANDAND)
6154021b 497 pl_yylval.ival = OP_ANDASSIGN;
a0d0e21e 498 else if (toketype == OROR)
6154021b 499 pl_yylval.ival = OP_ORASSIGN;
c963b151 500 else if (toketype == DORDOR)
6154021b 501 pl_yylval.ival = OP_DORASSIGN;
a0d0e21e
LW
502 toketype = ASSIGNOP;
503 }
504 return toketype;
505}
506
ffb4593c
NT
507/*
508 * S_no_op
509 * When Perl expects an operator and finds something else, no_op
510 * prints the warning. It always prints "<something> found where
511 * operator expected. It prints "Missing semicolon on previous line?"
512 * if the surprise occurs at the start of the line. "do you need to
513 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
514 * where the compiler doesn't know if foo is a method call or a function.
515 * It prints "Missing operator before end of line" if there's nothing
516 * after the missing operator, or "... before <...>" if there is something
517 * after the missing operator.
518 */
519
76e3520e 520STATIC void
15f169a1 521S_no_op(pTHX_ const char *const what, char *s)
463ee0b2 522{
97aff369 523 dVAR;
9d4ba2ae
AL
524 char * const oldbp = PL_bufptr;
525 const bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 526
7918f24d
NC
527 PERL_ARGS_ASSERT_NO_OP;
528
1189a94a
GS
529 if (!s)
530 s = oldbp;
07c798fb 531 else
1189a94a 532 PL_bufptr = s;
cea2e8a9 533 yywarn(Perl_form(aTHX_ "%s found where operator expected", what));
56da5a46
RGS
534 if (ckWARN_d(WARN_SYNTAX)) {
535 if (is_first)
536 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
537 "\t(Missing semicolon on previous line?)\n");
538 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
f54cb97a 539 const char *t;
c35e046a
AL
540 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':'); t++)
541 NOOP;
56da5a46
RGS
542 if (t < PL_bufptr && isSPACE(*t))
543 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
544 "\t(Do you need to predeclare %.*s?)\n",
551405c4 545 (int)(t - PL_oldoldbufptr), PL_oldoldbufptr);
56da5a46
RGS
546 }
547 else {
548 assert(s >= oldbp);
549 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
551405c4 550 "\t(Missing operator before %.*s?)\n", (int)(s - oldbp), oldbp);
56da5a46 551 }
07c798fb 552 }
3280af22 553 PL_bufptr = oldbp;
8990e307
LW
554}
555
ffb4593c
NT
556/*
557 * S_missingterm
558 * Complain about missing quote/regexp/heredoc terminator.
d4c19fe8 559 * If it's called with NULL then it cauterizes the line buffer.
ffb4593c
NT
560 * If we're in a delimited string and the delimiter is a control
561 * character, it's reformatted into a two-char sequence like ^C.
562 * This is fatal.
563 */
564
76e3520e 565STATIC void
cea2e8a9 566S_missingterm(pTHX_ char *s)
8990e307 567{
97aff369 568 dVAR;
8990e307
LW
569 char tmpbuf[3];
570 char q;
571 if (s) {
9d4ba2ae 572 char * const nl = strrchr(s,'\n');
d2719217 573 if (nl)
8990e307
LW
574 *nl = '\0';
575 }
463559e7 576 else if (isCNTRL(PL_multi_close)) {
8990e307 577 *tmpbuf = '^';
585ec06d 578 tmpbuf[1] = (char)toCTRL(PL_multi_close);
8990e307
LW
579 tmpbuf[2] = '\0';
580 s = tmpbuf;
581 }
582 else {
eb160463 583 *tmpbuf = (char)PL_multi_close;
8990e307
LW
584 tmpbuf[1] = '\0';
585 s = tmpbuf;
586 }
587 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 588 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 589}
79072805 590
0d863452 591/*
0d863452
RH
592 * Check whether the named feature is enabled.
593 */
26ea9e12
NC
594bool
595Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
0d863452 596{
97aff369 597 dVAR;
0d863452 598 HV * const hinthv = GvHV(PL_hintgv);
4a731d7b 599 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
7918f24d
NC
600
601 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
602
26ea9e12
NC
603 if (namelen > MAX_FEATURE_LEN)
604 return FALSE;
4a731d7b 605 memcpy(&he_name[8], name, namelen);
d4c19fe8 606
7b9ef140 607 return (hinthv && hv_exists(hinthv, he_name, 8 + namelen));
0d863452
RH
608}
609
ffb4593c 610/*
9cbb5ea2
GS
611 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
612 * utf16-to-utf8-reversed.
ffb4593c
NT
613 */
614
c39cd008
GS
615#ifdef PERL_CR_FILTER
616static void
617strip_return(SV *sv)
618{
95a20fc0 619 register const char *s = SvPVX_const(sv);
9d4ba2ae 620 register const char * const e = s + SvCUR(sv);
7918f24d
NC
621
622 PERL_ARGS_ASSERT_STRIP_RETURN;
623
c39cd008
GS
624 /* outer loop optimized to do nothing if there are no CR-LFs */
625 while (s < e) {
626 if (*s++ == '\r' && *s == '\n') {
627 /* hit a CR-LF, need to copy the rest */
628 register char *d = s - 1;
629 *d++ = *s++;
630 while (s < e) {
631 if (*s == '\r' && s[1] == '\n')
632 s++;
633 *d++ = *s++;
634 }
635 SvCUR(sv) -= s - d;
636 return;
637 }
638 }
639}
a868473f 640
76e3520e 641STATIC I32
c39cd008 642S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 643{
f54cb97a 644 const I32 count = FILTER_READ(idx+1, sv, maxlen);
c39cd008
GS
645 if (count > 0 && !maxlen)
646 strip_return(sv);
647 return count;
a868473f
NIS
648}
649#endif
650
ffb4593c 651/*
8eaa0acf
Z
652=for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags
653
654Creates and initialises a new lexer/parser state object, supplying
655a context in which to lex and parse from a new source of Perl code.
656A pointer to the new state object is placed in L</PL_parser>. An entry
657is made on the save stack so that upon unwinding the new state object
658will be destroyed and the former value of L</PL_parser> will be restored.
659Nothing else need be done to clean up the parsing context.
660
661The code to be parsed comes from I<line> and I<rsfp>. I<line>, if
662non-null, provides a string (in SV form) containing code to be parsed.
663A copy of the string is made, so subsequent modification of I<line>
664does not affect parsing. I<rsfp>, if non-null, provides an input stream
665from which code will be read to be parsed. If both are non-null, the
666code in I<line> comes first and must consist of complete lines of input,
667and I<rsfp> supplies the remainder of the source.
668
669The I<flags> parameter is reserved for future use, and must always
670be zero.
671
672=cut
673*/
ffb4593c 674
a0d0e21e 675void
8eaa0acf 676Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags)
79072805 677{
97aff369 678 dVAR;
6ef55633 679 const char *s = NULL;
8990e307 680 STRLEN len;
5486870f 681 yy_parser *parser, *oparser;
8eaa0acf
Z
682 if (flags)
683 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start");
acdf0a21
DM
684
685 /* create and initialise a parser */
686
199e78b7 687 Newxz(parser, 1, yy_parser);
5486870f 688 parser->old_parser = oparser = PL_parser;
acdf0a21
DM
689 PL_parser = parser;
690
28ac2b49
Z
691 parser->stack = NULL;
692 parser->ps = NULL;
693 parser->stack_size = 0;
acdf0a21 694
e3abe207
DM
695 /* on scope exit, free this parser and restore any outer one */
696 SAVEPARSER(parser);
7c4baf47 697 parser->saved_curcop = PL_curcop;
e3abe207 698
acdf0a21 699 /* initialise lexer state */
8990e307 700
fb205e7a
DM
701#ifdef PERL_MAD
702 parser->curforce = -1;
703#else
704 parser->nexttoke = 0;
705#endif
ca4cfd28 706 parser->error_count = oparser ? oparser->error_count : 0;
c2598295 707 parser->copline = NOLINE;
5afb0a62 708 parser->lex_state = LEX_NORMAL;
c2598295 709 parser->expect = XSTATE;
2f9285f8 710 parser->rsfp = rsfp;
f07ec6dd 711 parser->rsfp_filters = newAV();
2f9285f8 712
199e78b7
DM
713 Newx(parser->lex_brackstack, 120, char);
714 Newx(parser->lex_casestack, 12, char);
715 *parser->lex_casestack = '\0';
02b34bbe 716
10efb74f
NC
717 if (line) {
718 s = SvPV_const(line, len);
719 } else {
720 len = 0;
721 }
bdc0bf6f 722
10efb74f 723 if (!len) {
bdc0bf6f 724 parser->linestr = newSVpvs("\n;");
805700c1 725 } else {
719a9bb0 726 parser->linestr = newSVpvn_flags(s, len, SvUTF8(line));
10efb74f 727 if (s[len-1] != ';')
bdc0bf6f 728 sv_catpvs(parser->linestr, "\n;");
8990e307 729 }
f06b5848
DM
730 parser->oldoldbufptr =
731 parser->oldbufptr =
732 parser->bufptr =
733 parser->linestart = SvPVX(parser->linestr);
734 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
735 parser->last_lop = parser->last_uni = NULL;
737c24fc
Z
736
737 parser->in_pod = 0;
79072805 738}
a687059c 739
e3abe207
DM
740
741/* delete a parser object */
742
743void
744Perl_parser_free(pTHX_ const yy_parser *parser)
745{
7918f24d
NC
746 PERL_ARGS_ASSERT_PARSER_FREE;
747
7c4baf47 748 PL_curcop = parser->saved_curcop;
bdc0bf6f
DM
749 SvREFCNT_dec(parser->linestr);
750
2f9285f8
DM
751 if (parser->rsfp == PerlIO_stdin())
752 PerlIO_clearerr(parser->rsfp);
799361c3
SH
753 else if (parser->rsfp && (!parser->old_parser ||
754 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
2f9285f8 755 PerlIO_close(parser->rsfp);
5486870f 756 SvREFCNT_dec(parser->rsfp_filters);
2f9285f8 757
e3abe207
DM
758 Safefree(parser->lex_brackstack);
759 Safefree(parser->lex_casestack);
760 PL_parser = parser->old_parser;
761 Safefree(parser);
762}
763
764
ffb4593c 765/*
f0e67a1d
Z
766=for apidoc AmxU|SV *|PL_parser-E<gt>linestr
767
768Buffer scalar containing the chunk currently under consideration of the
769text currently being lexed. This is always a plain string scalar (for
770which C<SvPOK> is true). It is not intended to be used as a scalar by
771normal scalar means; instead refer to the buffer directly by the pointer
772variables described below.
773
774The lexer maintains various C<char*> pointers to things in the
775C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever
776reallocated, all of these pointers must be updated. Don't attempt to
777do this manually, but rather use L</lex_grow_linestr> if you need to
778reallocate the buffer.
779
780The content of the text chunk in the buffer is commonly exactly one
781complete line of input, up to and including a newline terminator,
782but there are situations where it is otherwise. The octets of the
783buffer may be intended to be interpreted as either UTF-8 or Latin-1.
784The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8>
785flag on this scalar, which may disagree with it.
786
787For direct examination of the buffer, the variable
788L</PL_parser-E<gt>bufend> points to the end of the buffer. The current
789lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use
790of these pointers is usually preferable to examination of the scalar
791through normal scalar means.
792
793=for apidoc AmxU|char *|PL_parser-E<gt>bufend
794
795Direct pointer to the end of the chunk of text currently being lexed, the
796end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr)
797+ SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is
798always located at the end of the buffer, and does not count as part of
799the buffer's contents.
800
801=for apidoc AmxU|char *|PL_parser-E<gt>bufptr
802
803Points to the current position of lexing inside the lexer buffer.
804Characters around this point may be freely examined, within
805the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
806L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be
807interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
808
809Lexing code (whether in the Perl core or not) moves this pointer past
810the characters that it consumes. It is also expected to perform some
811bookkeeping whenever a newline character is consumed. This movement
812can be more conveniently performed by the function L</lex_read_to>,
813which handles newlines appropriately.
814
815Interpretation of the buffer's octets can be abstracted out by
816using the slightly higher-level functions L</lex_peek_unichar> and
817L</lex_read_unichar>.
818
819=for apidoc AmxU|char *|PL_parser-E<gt>linestart
820
821Points to the start of the current line inside the lexer buffer.
822This is useful for indicating at which column an error occurred, and
823not much else. This must be updated by any lexing code that consumes
824a newline; the function L</lex_read_to> handles this detail.
825
826=cut
827*/
828
829/*
830=for apidoc Amx|bool|lex_bufutf8
831
832Indicates whether the octets in the lexer buffer
833(L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
834of Unicode characters. If not, they should be interpreted as Latin-1
835characters. This is analogous to the C<SvUTF8> flag for scalars.
836
837In UTF-8 mode, it is not guaranteed that the lexer buffer actually
838contains valid UTF-8. Lexing code must be robust in the face of invalid
839encoding.
840
841The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
842is significant, but not the whole story regarding the input character
843encoding. Normally, when a file is being read, the scalar contains octets
844and its C<SvUTF8> flag is off, but the octets should be interpreted as
845UTF-8 if the C<use utf8> pragma is in effect. During a string eval,
846however, the scalar may have the C<SvUTF8> flag on, and in this case its
847octets should be interpreted as UTF-8 unless the C<use bytes> pragma
848is in effect. This logic may change in the future; use this function
849instead of implementing the logic yourself.
850
851=cut
852*/
853
854bool
855Perl_lex_bufutf8(pTHX)
856{
857 return UTF;
858}
859
860/*
861=for apidoc Amx|char *|lex_grow_linestr|STRLEN len
862
863Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
864at least I<len> octets (including terminating NUL). Returns a
865pointer to the reallocated buffer. This is necessary before making
866any direct modification of the buffer that would increase its length.
867L</lex_stuff_pvn> provides a more convenient way to insert text into
868the buffer.
869
870Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
871this function updates all of the lexer's variables that point directly
872into the buffer.
873
874=cut
875*/
876
877char *
878Perl_lex_grow_linestr(pTHX_ STRLEN len)
879{
880 SV *linestr;
881 char *buf;
882 STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
883 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
884 linestr = PL_parser->linestr;
885 buf = SvPVX(linestr);
886 if (len <= SvLEN(linestr))
887 return buf;
888 bufend_pos = PL_parser->bufend - buf;
889 bufptr_pos = PL_parser->bufptr - buf;
890 oldbufptr_pos = PL_parser->oldbufptr - buf;
891 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
892 linestart_pos = PL_parser->linestart - buf;
893 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
894 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
895 buf = sv_grow(linestr, len);
896 PL_parser->bufend = buf + bufend_pos;
897 PL_parser->bufptr = buf + bufptr_pos;
898 PL_parser->oldbufptr = buf + oldbufptr_pos;
899 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
900 PL_parser->linestart = buf + linestart_pos;
901 if (PL_parser->last_uni)
902 PL_parser->last_uni = buf + last_uni_pos;
903 if (PL_parser->last_lop)
904 PL_parser->last_lop = buf + last_lop_pos;
905 return buf;
906}
907
908/*
83aa740e 909=for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags
f0e67a1d
Z
910
911Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
912immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
913reallocating the buffer if necessary. This means that lexing code that
914runs later will see the characters as if they had appeared in the input.
915It is not recommended to do this as part of normal parsing, and most
916uses of this facility run the risk of the inserted characters being
917interpreted in an unintended manner.
918
919The string to be inserted is represented by I<len> octets starting
920at I<pv>. These octets are interpreted as either UTF-8 or Latin-1,
921according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
922The characters are recoded for the lexer buffer, according to how the
923buffer is currently being interpreted (L</lex_bufutf8>). If a string
9dcc53ea 924to be inserted is available as a Perl scalar, the L</lex_stuff_sv>
f0e67a1d
Z
925function is more convenient.
926
927=cut
928*/
929
930void
83aa740e 931Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags)
f0e67a1d 932{
749123ff 933 dVAR;
f0e67a1d
Z
934 char *bufptr;
935 PERL_ARGS_ASSERT_LEX_STUFF_PVN;
936 if (flags & ~(LEX_STUFF_UTF8))
937 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
938 if (UTF) {
939 if (flags & LEX_STUFF_UTF8) {
940 goto plain_copy;
941 } else {
942 STRLEN highhalf = 0;
83aa740e 943 const char *p, *e = pv+len;
f0e67a1d
Z
944 for (p = pv; p != e; p++)
945 highhalf += !!(((U8)*p) & 0x80);
946 if (!highhalf)
947 goto plain_copy;
948 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
949 bufptr = PL_parser->bufptr;
950 Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
951 SvCUR_set(PL_parser->linestr,
952 SvCUR(PL_parser->linestr) + len+highhalf);
f0e67a1d
Z
953 PL_parser->bufend += len+highhalf;
954 for (p = pv; p != e; p++) {
955 U8 c = (U8)*p;
956 if (c & 0x80) {
957 *bufptr++ = (char)(0xc0 | (c >> 6));
958 *bufptr++ = (char)(0x80 | (c & 0x3f));
959 } else {
960 *bufptr++ = (char)c;
961 }
962 }
963 }
964 } else {
965 if (flags & LEX_STUFF_UTF8) {
966 STRLEN highhalf = 0;
83aa740e 967 const char *p, *e = pv+len;
f0e67a1d
Z
968 for (p = pv; p != e; p++) {
969 U8 c = (U8)*p;
970 if (c >= 0xc4) {
971 Perl_croak(aTHX_ "Lexing code attempted to stuff "
972 "non-Latin-1 character into Latin-1 input");
973 } else if (c >= 0xc2 && p+1 != e &&
974 (((U8)p[1]) & 0xc0) == 0x80) {
975 p++;
976 highhalf++;
977 } else if (c >= 0x80) {
978 /* malformed UTF-8 */
979 ENTER;
980 SAVESPTR(PL_warnhook);
981 PL_warnhook = PERL_WARNHOOK_FATAL;
982 utf8n_to_uvuni((U8*)p, e-p, NULL, 0);
983 LEAVE;
984 }
985 }
986 if (!highhalf)
987 goto plain_copy;
988 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
989 bufptr = PL_parser->bufptr;
990 Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
991 SvCUR_set(PL_parser->linestr,
992 SvCUR(PL_parser->linestr) + len-highhalf);
f0e67a1d
Z
993 PL_parser->bufend += len-highhalf;
994 for (p = pv; p != e; p++) {
995 U8 c = (U8)*p;
996 if (c & 0x80) {
997 *bufptr++ = (char)(((c & 0x3) << 6) | (p[1] & 0x3f));
998 p++;
999 } else {
1000 *bufptr++ = (char)c;
1001 }
1002 }
1003 } else {
1004 plain_copy:
1005 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1006 bufptr = PL_parser->bufptr;
1007 Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
255fdf19 1008 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len);
f0e67a1d
Z
1009 PL_parser->bufend += len;
1010 Copy(pv, bufptr, len, char);
1011 }
1012 }
1013}
1014
1015/*
9dcc53ea
Z
1016=for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags
1017
1018Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1019immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1020reallocating the buffer if necessary. This means that lexing code that
1021runs later will see the characters as if they had appeared in the input.
1022It is not recommended to do this as part of normal parsing, and most
1023uses of this facility run the risk of the inserted characters being
1024interpreted in an unintended manner.
1025
1026The string to be inserted is represented by octets starting at I<pv>
1027and continuing to the first nul. These octets are interpreted as either
1028UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set
1029in I<flags>. The characters are recoded for the lexer buffer, according
1030to how the buffer is currently being interpreted (L</lex_bufutf8>).
1031If it is not convenient to nul-terminate a string to be inserted, the
1032L</lex_stuff_pvn> function is more appropriate.
1033
1034=cut
1035*/
1036
1037void
1038Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags)
1039{
1040 PERL_ARGS_ASSERT_LEX_STUFF_PV;
1041 lex_stuff_pvn(pv, strlen(pv), flags);
1042}
1043
1044/*
f0e67a1d
Z
1045=for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1046
1047Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1048immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1049reallocating the buffer if necessary. This means that lexing code that
1050runs later will see the characters as if they had appeared in the input.
1051It is not recommended to do this as part of normal parsing, and most
1052uses of this facility run the risk of the inserted characters being
1053interpreted in an unintended manner.
1054
1055The string to be inserted is the string value of I<sv>. The characters
1056are recoded for the lexer buffer, according to how the buffer is currently
9dcc53ea 1057being interpreted (L</lex_bufutf8>). If a string to be inserted is
f0e67a1d
Z
1058not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1059need to construct a scalar.
1060
1061=cut
1062*/
1063
1064void
1065Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1066{
1067 char *pv;
1068 STRLEN len;
1069 PERL_ARGS_ASSERT_LEX_STUFF_SV;
1070 if (flags)
1071 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1072 pv = SvPV(sv, len);
1073 lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1074}
1075
1076/*
1077=for apidoc Amx|void|lex_unstuff|char *ptr
1078
1079Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1080I<ptr>. Text following I<ptr> will be moved, and the buffer shortened.
1081This hides the discarded text from any lexing code that runs later,
1082as if the text had never appeared.
1083
1084This is not the normal way to consume lexed text. For that, use
1085L</lex_read_to>.
1086
1087=cut
1088*/
1089
1090void
1091Perl_lex_unstuff(pTHX_ char *ptr)
1092{
1093 char *buf, *bufend;
1094 STRLEN unstuff_len;
1095 PERL_ARGS_ASSERT_LEX_UNSTUFF;
1096 buf = PL_parser->bufptr;
1097 if (ptr < buf)
1098 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1099 if (ptr == buf)
1100 return;
1101 bufend = PL_parser->bufend;
1102 if (ptr > bufend)
1103 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1104 unstuff_len = ptr - buf;
1105 Move(ptr, buf, bufend+1-ptr, char);
1106 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1107 PL_parser->bufend = bufend - unstuff_len;
1108}
1109
1110/*
1111=for apidoc Amx|void|lex_read_to|char *ptr
1112
1113Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1114to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1115performing the correct bookkeeping whenever a newline character is passed.
1116This is the normal way to consume lexed text.
1117
1118Interpretation of the buffer's octets can be abstracted out by
1119using the slightly higher-level functions L</lex_peek_unichar> and
1120L</lex_read_unichar>.
1121
1122=cut
1123*/
1124
1125void
1126Perl_lex_read_to(pTHX_ char *ptr)
1127{
1128 char *s;
1129 PERL_ARGS_ASSERT_LEX_READ_TO;
1130 s = PL_parser->bufptr;
1131 if (ptr < s || ptr > PL_parser->bufend)
1132 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1133 for (; s != ptr; s++)
1134 if (*s == '\n') {
1135 CopLINE_inc(PL_curcop);
1136 PL_parser->linestart = s+1;
1137 }
1138 PL_parser->bufptr = ptr;
1139}
1140
1141/*
1142=for apidoc Amx|void|lex_discard_to|char *ptr
1143
1144Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1145up to I<ptr>. The remaining content of the buffer will be moved, and
1146all pointers into the buffer updated appropriately. I<ptr> must not
1147be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1148it is not permitted to discard text that has yet to be lexed.
1149
1150Normally it is not necessarily to do this directly, because it suffices to
1151use the implicit discarding behaviour of L</lex_next_chunk> and things
1152based on it. However, if a token stretches across multiple lines,
1f317c95 1153and the lexing code has kept multiple lines of text in the buffer for
f0e67a1d
Z
1154that purpose, then after completion of the token it would be wise to
1155explicitly discard the now-unneeded earlier lines, to avoid future
1156multi-line tokens growing the buffer without bound.
1157
1158=cut
1159*/
1160
1161void
1162Perl_lex_discard_to(pTHX_ char *ptr)
1163{
1164 char *buf;
1165 STRLEN discard_len;
1166 PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1167 buf = SvPVX(PL_parser->linestr);
1168 if (ptr < buf)
1169 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1170 if (ptr == buf)
1171 return;
1172 if (ptr > PL_parser->bufptr)
1173 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1174 discard_len = ptr - buf;
1175 if (PL_parser->oldbufptr < ptr)
1176 PL_parser->oldbufptr = ptr;
1177 if (PL_parser->oldoldbufptr < ptr)
1178 PL_parser->oldoldbufptr = ptr;
1179 if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1180 PL_parser->last_uni = NULL;
1181 if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1182 PL_parser->last_lop = NULL;
1183 Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1184 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1185 PL_parser->bufend -= discard_len;
1186 PL_parser->bufptr -= discard_len;
1187 PL_parser->oldbufptr -= discard_len;
1188 PL_parser->oldoldbufptr -= discard_len;
1189 if (PL_parser->last_uni)
1190 PL_parser->last_uni -= discard_len;
1191 if (PL_parser->last_lop)
1192 PL_parser->last_lop -= discard_len;
1193}
1194
1195/*
1196=for apidoc Amx|bool|lex_next_chunk|U32 flags
1197
1198Reads in the next chunk of text to be lexed, appending it to
1199L</PL_parser-E<gt>linestr>. This should be called when lexing code has
1200looked to the end of the current chunk and wants to know more. It is
1201usual, but not necessary, for lexing to have consumed the entirety of
1202the current chunk at this time.
1203
1204If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1205chunk (i.e., the current chunk has been entirely consumed), normally the
1206current chunk will be discarded at the same time that the new chunk is
1207read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1208will not be discarded. If the current chunk has not been entirely
1209consumed, then it will not be discarded regardless of the flag.
1210
1211Returns true if some new text was added to the buffer, or false if the
1212buffer has reached the end of the input text.
1213
1214=cut
1215*/
1216
1217#define LEX_FAKE_EOF 0x80000000
1218
1219bool
1220Perl_lex_next_chunk(pTHX_ U32 flags)
1221{
1222 SV *linestr;
1223 char *buf;
1224 STRLEN old_bufend_pos, new_bufend_pos;
1225 STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1226 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
17cc9359 1227 bool got_some_for_debugger = 0;
f0e67a1d
Z
1228 bool got_some;
1229 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF))
1230 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
f0e67a1d
Z
1231 linestr = PL_parser->linestr;
1232 buf = SvPVX(linestr);
1233 if (!(flags & LEX_KEEP_PREVIOUS) &&
1234 PL_parser->bufptr == PL_parser->bufend) {
1235 old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1236 linestart_pos = 0;
1237 if (PL_parser->last_uni != PL_parser->bufend)
1238 PL_parser->last_uni = NULL;
1239 if (PL_parser->last_lop != PL_parser->bufend)
1240 PL_parser->last_lop = NULL;
1241 last_uni_pos = last_lop_pos = 0;
1242 *buf = 0;
1243 SvCUR(linestr) = 0;
1244 } else {
1245 old_bufend_pos = PL_parser->bufend - buf;
1246 bufptr_pos = PL_parser->bufptr - buf;
1247 oldbufptr_pos = PL_parser->oldbufptr - buf;
1248 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1249 linestart_pos = PL_parser->linestart - buf;
1250 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1251 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1252 }
1253 if (flags & LEX_FAKE_EOF) {
1254 goto eof;
1255 } else if (!PL_parser->rsfp) {
1256 got_some = 0;
1257 } else if (filter_gets(linestr, old_bufend_pos)) {
1258 got_some = 1;
17cc9359 1259 got_some_for_debugger = 1;
f0e67a1d 1260 } else {
580561a3
Z
1261 if (!SvPOK(linestr)) /* can get undefined by filter_gets */
1262 sv_setpvs(linestr, "");
f0e67a1d
Z
1263 eof:
1264 /* End of real input. Close filehandle (unless it was STDIN),
1265 * then add implicit termination.
1266 */
1267 if ((PerlIO*)PL_parser->rsfp == PerlIO_stdin())
1268 PerlIO_clearerr(PL_parser->rsfp);
1269 else if (PL_parser->rsfp)
1270 (void)PerlIO_close(PL_parser->rsfp);
1271 PL_parser->rsfp = NULL;
737c24fc 1272 PL_parser->in_pod = 0;
f0e67a1d
Z
1273#ifdef PERL_MAD
1274 if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1275 PL_faketokens = 1;
1276#endif
1277 if (!PL_in_eval && PL_minus_p) {
1278 sv_catpvs(linestr,
1279 /*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1280 PL_minus_n = PL_minus_p = 0;
1281 } else if (!PL_in_eval && PL_minus_n) {
1282 sv_catpvs(linestr, /*{*/";}");
1283 PL_minus_n = 0;
1284 } else
1285 sv_catpvs(linestr, ";");
1286 got_some = 1;
1287 }
1288 buf = SvPVX(linestr);
1289 new_bufend_pos = SvCUR(linestr);
1290 PL_parser->bufend = buf + new_bufend_pos;
1291 PL_parser->bufptr = buf + bufptr_pos;
1292 PL_parser->oldbufptr = buf + oldbufptr_pos;
1293 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1294 PL_parser->linestart = buf + linestart_pos;
1295 if (PL_parser->last_uni)
1296 PL_parser->last_uni = buf + last_uni_pos;
1297 if (PL_parser->last_lop)
1298 PL_parser->last_lop = buf + last_lop_pos;
17cc9359 1299 if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
f0e67a1d
Z
1300 PL_curstash != PL_debstash) {
1301 /* debugger active and we're not compiling the debugger code,
1302 * so store the line into the debugger's array of lines
1303 */
1304 update_debugger_info(NULL, buf+old_bufend_pos,
1305 new_bufend_pos-old_bufend_pos);
1306 }
1307 return got_some;
1308}
1309
1310/*
1311=for apidoc Amx|I32|lex_peek_unichar|U32 flags
1312
1313Looks ahead one (Unicode) character in the text currently being lexed.
1314Returns the codepoint (unsigned integer value) of the next character,
1315or -1 if lexing has reached the end of the input text. To consume the
1316peeked character, use L</lex_read_unichar>.
1317
1318If the next character is in (or extends into) the next chunk of input
1319text, the next chunk will be read in. Normally the current chunk will be
1320discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1321then the current chunk will not be discarded.
1322
1323If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1324is encountered, an exception is generated.
1325
1326=cut
1327*/
1328
1329I32
1330Perl_lex_peek_unichar(pTHX_ U32 flags)
1331{
749123ff 1332 dVAR;
f0e67a1d
Z
1333 char *s, *bufend;
1334 if (flags & ~(LEX_KEEP_PREVIOUS))
1335 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1336 s = PL_parser->bufptr;
1337 bufend = PL_parser->bufend;
1338 if (UTF) {
1339 U8 head;
1340 I32 unichar;
1341 STRLEN len, retlen;
1342 if (s == bufend) {
1343 if (!lex_next_chunk(flags))
1344 return -1;
1345 s = PL_parser->bufptr;
1346 bufend = PL_parser->bufend;
1347 }
1348 head = (U8)*s;
1349 if (!(head & 0x80))
1350 return head;
1351 if (head & 0x40) {
1352 len = PL_utf8skip[head];
1353 while ((STRLEN)(bufend-s) < len) {
1354 if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1355 break;
1356 s = PL_parser->bufptr;
1357 bufend = PL_parser->bufend;
1358 }
1359 }
1360 unichar = utf8n_to_uvuni((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
1361 if (retlen == (STRLEN)-1) {
1362 /* malformed UTF-8 */
1363 ENTER;
1364 SAVESPTR(PL_warnhook);
1365 PL_warnhook = PERL_WARNHOOK_FATAL;
1366 utf8n_to_uvuni((U8*)s, bufend-s, NULL, 0);
1367 LEAVE;
1368 }
1369 return unichar;
1370 } else {
1371 if (s == bufend) {
1372 if (!lex_next_chunk(flags))
1373 return -1;
1374 s = PL_parser->bufptr;
1375 }
1376 return (U8)*s;
1377 }
1378}
1379
1380/*
1381=for apidoc Amx|I32|lex_read_unichar|U32 flags
1382
1383Reads the next (Unicode) character in the text currently being lexed.
1384Returns the codepoint (unsigned integer value) of the character read,
1385and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1386if lexing has reached the end of the input text. To non-destructively
1387examine the next character, use L</lex_peek_unichar> instead.
1388
1389If the next character is in (or extends into) the next chunk of input
1390text, the next chunk will be read in. Normally the current chunk will be
1391discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1392then the current chunk will not be discarded.
1393
1394If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1395is encountered, an exception is generated.
1396
1397=cut
1398*/
1399
1400I32
1401Perl_lex_read_unichar(pTHX_ U32 flags)
1402{
1403 I32 c;
1404 if (flags & ~(LEX_KEEP_PREVIOUS))
1405 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1406 c = lex_peek_unichar(flags);
1407 if (c != -1) {
1408 if (c == '\n')
1409 CopLINE_inc(PL_curcop);
1410 PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1411 }
1412 return c;
1413}
1414
1415/*
1416=for apidoc Amx|void|lex_read_space|U32 flags
1417
1418Reads optional spaces, in Perl style, in the text currently being
1419lexed. The spaces may include ordinary whitespace characters and
1420Perl-style comments. C<#line> directives are processed if encountered.
1421L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1422at a non-space character (or the end of the input text).
1423
1424If spaces extend into the next chunk of input text, the next chunk will
1425be read in. Normally the current chunk will be discarded at the same
1426time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1427chunk will not be discarded.
1428
1429=cut
1430*/
1431
f0998909
Z
1432#define LEX_NO_NEXT_CHUNK 0x80000000
1433
f0e67a1d
Z
1434void
1435Perl_lex_read_space(pTHX_ U32 flags)
1436{
1437 char *s, *bufend;
1438 bool need_incline = 0;
f0998909 1439 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK))
f0e67a1d
Z
1440 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1441#ifdef PERL_MAD
1442 if (PL_skipwhite) {
1443 sv_free(PL_skipwhite);
1444 PL_skipwhite = NULL;
1445 }
1446 if (PL_madskills)
1447 PL_skipwhite = newSVpvs("");
1448#endif /* PERL_MAD */
1449 s = PL_parser->bufptr;
1450 bufend = PL_parser->bufend;
1451 while (1) {
1452 char c = *s;
1453 if (c == '#') {
1454 do {
1455 c = *++s;
1456 } while (!(c == '\n' || (c == 0 && s == bufend)));
1457 } else if (c == '\n') {
1458 s++;
1459 PL_parser->linestart = s;
1460 if (s == bufend)
1461 need_incline = 1;
1462 else
1463 incline(s);
1464 } else if (isSPACE(c)) {
1465 s++;
1466 } else if (c == 0 && s == bufend) {
1467 bool got_more;
1468#ifdef PERL_MAD
1469 if (PL_madskills)
1470 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1471#endif /* PERL_MAD */
f0998909
Z
1472 if (flags & LEX_NO_NEXT_CHUNK)
1473 break;
f0e67a1d
Z
1474 PL_parser->bufptr = s;
1475 CopLINE_inc(PL_curcop);
1476 got_more = lex_next_chunk(flags);
1477 CopLINE_dec(PL_curcop);
1478 s = PL_parser->bufptr;
1479 bufend = PL_parser->bufend;
1480 if (!got_more)
1481 break;
1482 if (need_incline && PL_parser->rsfp) {
1483 incline(s);
1484 need_incline = 0;
1485 }
1486 } else {
1487 break;
1488 }
1489 }
1490#ifdef PERL_MAD
1491 if (PL_madskills)
1492 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1493#endif /* PERL_MAD */
1494 PL_parser->bufptr = s;
1495}
1496
1497/*
ffb4593c
NT
1498 * S_incline
1499 * This subroutine has nothing to do with tilting, whether at windmills
1500 * or pinball tables. Its name is short for "increment line". It
57843af0 1501 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 1502 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
1503 * # line 500 "foo.pm"
1504 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
1505 */
1506
76e3520e 1507STATIC void
d9095cec 1508S_incline(pTHX_ const char *s)
463ee0b2 1509{
97aff369 1510 dVAR;
d9095cec
NC
1511 const char *t;
1512 const char *n;
1513 const char *e;
463ee0b2 1514
7918f24d
NC
1515 PERL_ARGS_ASSERT_INCLINE;
1516
57843af0 1517 CopLINE_inc(PL_curcop);
463ee0b2
LW
1518 if (*s++ != '#')
1519 return;
d4c19fe8
AL
1520 while (SPACE_OR_TAB(*s))
1521 s++;
73659bf1
GS
1522 if (strnEQ(s, "line", 4))
1523 s += 4;
1524 else
1525 return;
084592ab 1526 if (SPACE_OR_TAB(*s))
73659bf1 1527 s++;
4e553d73 1528 else
73659bf1 1529 return;
d4c19fe8
AL
1530 while (SPACE_OR_TAB(*s))
1531 s++;
463ee0b2
LW
1532 if (!isDIGIT(*s))
1533 return;
d4c19fe8 1534
463ee0b2
LW
1535 n = s;
1536 while (isDIGIT(*s))
1537 s++;
07714eb4 1538 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
26b6dc3f 1539 return;
bf4acbe4 1540 while (SPACE_OR_TAB(*s))
463ee0b2 1541 s++;
73659bf1 1542 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 1543 s++;
73659bf1
GS
1544 e = t + 1;
1545 }
463ee0b2 1546 else {
c35e046a
AL
1547 t = s;
1548 while (!isSPACE(*t))
1549 t++;
73659bf1 1550 e = t;
463ee0b2 1551 }
bf4acbe4 1552 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
1553 e++;
1554 if (*e != '\n' && *e != '\0')
1555 return; /* false alarm */
1556
f4dd75d9 1557 if (t - s > 0) {
d9095cec 1558 const STRLEN len = t - s;
8a5ee598 1559#ifndef USE_ITHREADS
19bad673
NC
1560 SV *const temp_sv = CopFILESV(PL_curcop);
1561 const char *cf;
1562 STRLEN tmplen;
1563
1564 if (temp_sv) {
1565 cf = SvPVX(temp_sv);
1566 tmplen = SvCUR(temp_sv);
1567 } else {
1568 cf = NULL;
1569 tmplen = 0;
1570 }
1571
42d9b98d 1572 if (tmplen > 7 && strnEQ(cf, "(eval ", 6)) {
e66cf94c
RGS
1573 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1574 * to *{"::_<newfilename"} */
44867030
NC
1575 /* However, the long form of evals is only turned on by the
1576 debugger - usually they're "(eval %lu)" */
1577 char smallbuf[128];
1578 char *tmpbuf;
1579 GV **gvp;
d9095cec 1580 STRLEN tmplen2 = len;
798b63bc 1581 if (tmplen + 2 <= sizeof smallbuf)
e66cf94c
RGS
1582 tmpbuf = smallbuf;
1583 else
2ae0db35 1584 Newx(tmpbuf, tmplen + 2, char);
44867030
NC
1585 tmpbuf[0] = '_';
1586 tmpbuf[1] = '<';
2ae0db35 1587 memcpy(tmpbuf + 2, cf, tmplen);
44867030 1588 tmplen += 2;
8a5ee598
RGS
1589 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
1590 if (gvp) {
44867030
NC
1591 char *tmpbuf2;
1592 GV *gv2;
1593
1594 if (tmplen2 + 2 <= sizeof smallbuf)
1595 tmpbuf2 = smallbuf;
1596 else
1597 Newx(tmpbuf2, tmplen2 + 2, char);
1598
1599 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
1600 /* Either they malloc'd it, or we malloc'd it,
1601 so no prefix is present in ours. */
1602 tmpbuf2[0] = '_';
1603 tmpbuf2[1] = '<';
1604 }
1605
1606 memcpy(tmpbuf2 + 2, s, tmplen2);
1607 tmplen2 += 2;
1608
8a5ee598 1609 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
e5527e4b 1610 if (!isGV(gv2)) {
8a5ee598 1611 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
e5527e4b
RGS
1612 /* adjust ${"::_<newfilename"} to store the new file name */
1613 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
3cb1dbc6
NC
1614 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
1615 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
e5527e4b 1616 }
44867030
NC
1617
1618 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
8a5ee598 1619 }
e66cf94c 1620 if (tmpbuf != smallbuf) Safefree(tmpbuf);
e66cf94c 1621 }
8a5ee598 1622#endif
05ec9bb3 1623 CopFILE_free(PL_curcop);
d9095cec 1624 CopFILE_setn(PL_curcop, s, len);
f4dd75d9 1625 }
57843af0 1626 CopLINE_set(PL_curcop, atoi(n)-1);
463ee0b2
LW
1627}
1628
29595ff2 1629#ifdef PERL_MAD
cd81e915 1630/* skip space before PL_thistoken */
29595ff2
NC
1631
1632STATIC char *
1633S_skipspace0(pTHX_ register char *s)
1634{
7918f24d
NC
1635 PERL_ARGS_ASSERT_SKIPSPACE0;
1636
29595ff2
NC
1637 s = skipspace(s);
1638 if (!PL_madskills)
1639 return s;
cd81e915
NC
1640 if (PL_skipwhite) {
1641 if (!PL_thiswhite)
6b29d1f5 1642 PL_thiswhite = newSVpvs("");
cd81e915
NC
1643 sv_catsv(PL_thiswhite, PL_skipwhite);
1644 sv_free(PL_skipwhite);
1645 PL_skipwhite = 0;
1646 }
1647 PL_realtokenstart = s - SvPVX(PL_linestr);
29595ff2
NC
1648 return s;
1649}
1650
cd81e915 1651/* skip space after PL_thistoken */
29595ff2
NC
1652
1653STATIC char *
1654S_skipspace1(pTHX_ register char *s)
1655{
d4c19fe8 1656 const char *start = s;
29595ff2
NC
1657 I32 startoff = start - SvPVX(PL_linestr);
1658
7918f24d
NC
1659 PERL_ARGS_ASSERT_SKIPSPACE1;
1660
29595ff2
NC
1661 s = skipspace(s);
1662 if (!PL_madskills)
1663 return s;
1664 start = SvPVX(PL_linestr) + startoff;
cd81e915 1665 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1666 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1667 PL_thistoken = newSVpvn(tstart, start - tstart);
1668 }
1669 PL_realtokenstart = -1;
1670 if (PL_skipwhite) {
1671 if (!PL_nextwhite)
6b29d1f5 1672 PL_nextwhite = newSVpvs("");
cd81e915
NC
1673 sv_catsv(PL_nextwhite, PL_skipwhite);
1674 sv_free(PL_skipwhite);
1675 PL_skipwhite = 0;
29595ff2
NC
1676 }
1677 return s;
1678}
1679
1680STATIC char *
1681S_skipspace2(pTHX_ register char *s, SV **svp)
1682{
c35e046a
AL
1683 char *start;
1684 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
1685 const I32 startoff = s - SvPVX(PL_linestr);
1686
7918f24d
NC
1687 PERL_ARGS_ASSERT_SKIPSPACE2;
1688
29595ff2
NC
1689 s = skipspace(s);
1690 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
1691 if (!PL_madskills || !svp)
1692 return s;
1693 start = SvPVX(PL_linestr) + startoff;
cd81e915 1694 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1695 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1696 PL_thistoken = newSVpvn(tstart, start - tstart);
1697 PL_realtokenstart = -1;
29595ff2 1698 }
cd81e915 1699 if (PL_skipwhite) {
29595ff2 1700 if (!*svp)
6b29d1f5 1701 *svp = newSVpvs("");
cd81e915
NC
1702 sv_setsv(*svp, PL_skipwhite);
1703 sv_free(PL_skipwhite);
1704 PL_skipwhite = 0;
29595ff2
NC
1705 }
1706
1707 return s;
1708}
1709#endif
1710
80a702cd 1711STATIC void
15f169a1 1712S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
80a702cd
RGS
1713{
1714 AV *av = CopFILEAVx(PL_curcop);
1715 if (av) {
b9f83d2f 1716 SV * const sv = newSV_type(SVt_PVMG);
5fa550fb
NC
1717 if (orig_sv)
1718 sv_setsv(sv, orig_sv);
1719 else
1720 sv_setpvn(sv, buf, len);
80a702cd
RGS
1721 (void)SvIOK_on(sv);
1722 SvIV_set(sv, 0);
1723 av_store(av, (I32)CopLINE(PL_curcop), sv);
1724 }
1725}
1726
ffb4593c
NT
1727/*
1728 * S_skipspace
1729 * Called to gobble the appropriate amount and type of whitespace.
1730 * Skips comments as well.
1731 */
1732
76e3520e 1733STATIC char *
cea2e8a9 1734S_skipspace(pTHX_ register char *s)
a687059c 1735{
5db06880 1736#ifdef PERL_MAD
f0e67a1d
Z
1737 char *start = s;
1738#endif /* PERL_MAD */
7918f24d 1739 PERL_ARGS_ASSERT_SKIPSPACE;
f0e67a1d 1740#ifdef PERL_MAD
cd81e915
NC
1741 if (PL_skipwhite) {
1742 sv_free(PL_skipwhite);
f0e67a1d 1743 PL_skipwhite = NULL;
5db06880 1744 }
f0e67a1d 1745#endif /* PERL_MAD */
3280af22 1746 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 1747 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2 1748 s++;
f0e67a1d
Z
1749 } else {
1750 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1751 PL_bufptr = s;
f0998909
Z
1752 lex_read_space(LEX_KEEP_PREVIOUS |
1753 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1754 LEX_NO_NEXT_CHUNK : 0));
3280af22 1755 s = PL_bufptr;
f0e67a1d
Z
1756 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1757 if (PL_linestart > PL_bufptr)
1758 PL_bufptr = PL_linestart;
1759 return s;
463ee0b2 1760 }
5db06880 1761#ifdef PERL_MAD
f0e67a1d
Z
1762 if (PL_madskills)
1763 PL_skipwhite = newSVpvn(start, s-start);
1764#endif /* PERL_MAD */
5db06880 1765 return s;
a687059c 1766}
378cc40b 1767
ffb4593c
NT
1768/*
1769 * S_check_uni
1770 * Check the unary operators to ensure there's no ambiguity in how they're
1771 * used. An ambiguous piece of code would be:
1772 * rand + 5
1773 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1774 * the +5 is its argument.
1775 */
1776
76e3520e 1777STATIC void
cea2e8a9 1778S_check_uni(pTHX)
ba106d47 1779{
97aff369 1780 dVAR;
d4c19fe8
AL
1781 const char *s;
1782 const char *t;
2f3197b3 1783
3280af22 1784 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 1785 return;
3280af22
NIS
1786 while (isSPACE(*PL_last_uni))
1787 PL_last_uni++;
c35e046a
AL
1788 s = PL_last_uni;
1789 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1790 s++;
3280af22 1791 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 1792 return;
6136c704 1793
9b387841
NC
1794 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1795 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1796 (int)(s - PL_last_uni), PL_last_uni);
2f3197b3
LW
1797}
1798
ffb4593c
NT
1799/*
1800 * LOP : macro to build a list operator. Its behaviour has been replaced
1801 * with a subroutine, S_lop() for which LOP is just another name.
1802 */
1803
a0d0e21e
LW
1804#define LOP(f,x) return lop(f,x,s)
1805
ffb4593c
NT
1806/*
1807 * S_lop
1808 * Build a list operator (or something that might be one). The rules:
1809 * - if we have a next token, then it's a list operator [why?]
1810 * - if the next thing is an opening paren, then it's a function
1811 * - else it's a list operator
1812 */
1813
76e3520e 1814STATIC I32
a0be28da 1815S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 1816{
97aff369 1817 dVAR;
7918f24d
NC
1818
1819 PERL_ARGS_ASSERT_LOP;
1820
6154021b 1821 pl_yylval.ival = f;
35c8bce7 1822 CLINE;
3280af22
NIS
1823 PL_expect = x;
1824 PL_bufptr = s;
1825 PL_last_lop = PL_oldbufptr;
eb160463 1826 PL_last_lop_op = (OPCODE)f;
5db06880
NC
1827#ifdef PERL_MAD
1828 if (PL_lasttoke)
78cdf107 1829 goto lstop;
5db06880 1830#else
3280af22 1831 if (PL_nexttoke)
78cdf107 1832 goto lstop;
5db06880 1833#endif
79072805 1834 if (*s == '(')
bbf60fe6 1835 return REPORT(FUNC);
29595ff2 1836 s = PEEKSPACE(s);
79072805 1837 if (*s == '(')
bbf60fe6 1838 return REPORT(FUNC);
78cdf107
Z
1839 else {
1840 lstop:
1841 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
1842 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
bbf60fe6 1843 return REPORT(LSTOP);
78cdf107 1844 }
79072805
LW
1845}
1846
5db06880
NC
1847#ifdef PERL_MAD
1848 /*
1849 * S_start_force
1850 * Sets up for an eventual force_next(). start_force(0) basically does
1851 * an unshift, while start_force(-1) does a push. yylex removes items
1852 * on the "pop" end.
1853 */
1854
1855STATIC void
1856S_start_force(pTHX_ int where)
1857{
1858 int i;
1859
cd81e915 1860 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
5db06880 1861 where = PL_lasttoke;
cd81e915
NC
1862 assert(PL_curforce < 0 || PL_curforce == where);
1863 if (PL_curforce != where) {
5db06880
NC
1864 for (i = PL_lasttoke; i > where; --i) {
1865 PL_nexttoke[i] = PL_nexttoke[i-1];
1866 }
1867 PL_lasttoke++;
1868 }
cd81e915 1869 if (PL_curforce < 0) /* in case of duplicate start_force() */
5db06880 1870 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
cd81e915
NC
1871 PL_curforce = where;
1872 if (PL_nextwhite) {
5db06880 1873 if (PL_madskills)
6b29d1f5 1874 curmad('^', newSVpvs(""));
cd81e915 1875 CURMAD('_', PL_nextwhite);
5db06880
NC
1876 }
1877}
1878
1879STATIC void
1880S_curmad(pTHX_ char slot, SV *sv)
1881{
1882 MADPROP **where;
1883
1884 if (!sv)
1885 return;
cd81e915
NC
1886 if (PL_curforce < 0)
1887 where = &PL_thismad;
5db06880 1888 else
cd81e915 1889 where = &PL_nexttoke[PL_curforce].next_mad;
5db06880 1890
cd81e915 1891 if (PL_faketokens)
76f68e9b 1892 sv_setpvs(sv, "");
5db06880
NC
1893 else {
1894 if (!IN_BYTES) {
1895 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1896 SvUTF8_on(sv);
1897 else if (PL_encoding) {
1898 sv_recode_to_utf8(sv, PL_encoding);
1899 }
1900 }
1901 }
1902
1903 /* keep a slot open for the head of the list? */
1904 if (slot != '_' && *where && (*where)->mad_key == '^') {
1905 (*where)->mad_key = slot;
daba3364 1906 sv_free(MUTABLE_SV(((*where)->mad_val)));
5db06880
NC
1907 (*where)->mad_val = (void*)sv;
1908 }
1909 else
1910 addmad(newMADsv(slot, sv), where, 0);
1911}
1912#else
b3f24c00
MHM
1913# define start_force(where) NOOP
1914# define curmad(slot, sv) NOOP
5db06880
NC
1915#endif
1916
ffb4593c
NT
1917/*
1918 * S_force_next
9cbb5ea2 1919 * When the lexer realizes it knows the next token (for instance,
ffb4593c 1920 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2 1921 * to know what token to return the next time the lexer is called. Caller
5db06880
NC
1922 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1923 * and possibly PL_expect to ensure the lexer handles the token correctly.
ffb4593c
NT
1924 */
1925
4e553d73 1926STATIC void
cea2e8a9 1927S_force_next(pTHX_ I32 type)
79072805 1928{
97aff369 1929 dVAR;
704d4215
GG
1930#ifdef DEBUGGING
1931 if (DEBUG_T_TEST) {
1932 PerlIO_printf(Perl_debug_log, "### forced token:\n");
f05d7009 1933 tokereport(type, &NEXTVAL_NEXTTOKE);
704d4215
GG
1934 }
1935#endif
5db06880 1936#ifdef PERL_MAD
cd81e915 1937 if (PL_curforce < 0)
5db06880 1938 start_force(PL_lasttoke);
cd81e915 1939 PL_nexttoke[PL_curforce].next_type = type;
5db06880
NC
1940 if (PL_lex_state != LEX_KNOWNEXT)
1941 PL_lex_defer = PL_lex_state;
1942 PL_lex_state = LEX_KNOWNEXT;
1943 PL_lex_expect = PL_expect;
cd81e915 1944 PL_curforce = -1;
5db06880 1945#else
3280af22
NIS
1946 PL_nexttype[PL_nexttoke] = type;
1947 PL_nexttoke++;
1948 if (PL_lex_state != LEX_KNOWNEXT) {
1949 PL_lex_defer = PL_lex_state;
1950 PL_lex_expect = PL_expect;
1951 PL_lex_state = LEX_KNOWNEXT;
79072805 1952 }
5db06880 1953#endif
79072805
LW
1954}
1955
28ac2b49
Z
1956void
1957Perl_yyunlex(pTHX)
1958{
a7aaec61
Z
1959 int yyc = PL_parser->yychar;
1960 if (yyc != YYEMPTY) {
1961 if (yyc) {
1962 start_force(-1);
1963 NEXTVAL_NEXTTOKE = PL_parser->yylval;
1964 if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
78cdf107 1965 PL_lex_allbrackets--;
a7aaec61 1966 PL_lex_brackets--;
78cdf107
Z
1967 yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
1968 } else if (yyc == '('/*)*/) {
1969 PL_lex_allbrackets--;
1970 yyc |= (2<<24);
a7aaec61
Z
1971 }
1972 force_next(yyc);
1973 }
28ac2b49
Z
1974 PL_parser->yychar = YYEMPTY;
1975 }
1976}
1977
d0a148a6 1978STATIC SV *
15f169a1 1979S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
d0a148a6 1980{
97aff369 1981 dVAR;
740cce10 1982 SV * const sv = newSVpvn_utf8(start, len,
eaf7a4d2
CS
1983 !IN_BYTES
1984 && UTF
1985 && !is_ascii_string((const U8*)start, len)
740cce10 1986 && is_utf8_string((const U8*)start, len));
d0a148a6
NC
1987 return sv;
1988}
1989
ffb4593c
NT
1990/*
1991 * S_force_word
1992 * When the lexer knows the next thing is a word (for instance, it has
1993 * just seen -> and it knows that the next char is a word char, then
02b34bbe
DM
1994 * it calls S_force_word to stick the next word into the PL_nexttoke/val
1995 * lookahead.
ffb4593c
NT
1996 *
1997 * Arguments:
b1b65b59 1998 * char *start : buffer position (must be within PL_linestr)
02b34bbe 1999 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
ffb4593c
NT
2000 * int check_keyword : if true, Perl checks to make sure the word isn't
2001 * a keyword (do this if the word is a label, e.g. goto FOO)
2002 * int allow_pack : if true, : characters will also be allowed (require,
2003 * use, etc. do this)
9cbb5ea2 2004 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
2005 */
2006
76e3520e 2007STATIC char *
cea2e8a9 2008S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
79072805 2009{
97aff369 2010 dVAR;
463ee0b2
LW
2011 register char *s;
2012 STRLEN len;
4e553d73 2013
7918f24d
NC
2014 PERL_ARGS_ASSERT_FORCE_WORD;
2015
29595ff2 2016 start = SKIPSPACE1(start);
463ee0b2 2017 s = start;
7e2040f0 2018 if (isIDFIRST_lazy_if(s,UTF) ||
a0d0e21e 2019 (allow_pack && *s == ':') ||
15f0808c 2020 (allow_initial_tick && *s == '\'') )
a0d0e21e 2021 {
3280af22 2022 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
5458a98a 2023 if (check_keyword && keyword(PL_tokenbuf, len, 0))
463ee0b2 2024 return start;
cd81e915 2025 start_force(PL_curforce);
5db06880
NC
2026 if (PL_madskills)
2027 curmad('X', newSVpvn(start,s-start));
463ee0b2 2028 if (token == METHOD) {
29595ff2 2029 s = SKIPSPACE1(s);
463ee0b2 2030 if (*s == '(')
3280af22 2031 PL_expect = XTERM;
463ee0b2 2032 else {
3280af22 2033 PL_expect = XOPERATOR;
463ee0b2 2034 }
79072805 2035 }
e74e6b3d 2036 if (PL_madskills)
63575281 2037 curmad('g', newSVpvs( "forced" ));
9ded7720 2038 NEXTVAL_NEXTTOKE.opval
d0a148a6
NC
2039 = (OP*)newSVOP(OP_CONST,0,
2040 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
9ded7720 2041 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
79072805
LW
2042 force_next(token);
2043 }
2044 return s;
2045}
2046
ffb4593c
NT
2047/*
2048 * S_force_ident
9cbb5ea2 2049 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
2050 * text only contains the "foo" portion. The first argument is a pointer
2051 * to the "foo", and the second argument is the type symbol to prefix.
2052 * Forces the next token to be a "WORD".
9cbb5ea2 2053 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
2054 */
2055
76e3520e 2056STATIC void
bfed75c6 2057S_force_ident(pTHX_ register const char *s, int kind)
79072805 2058{
97aff369 2059 dVAR;
7918f24d
NC
2060
2061 PERL_ARGS_ASSERT_FORCE_IDENT;
2062
c35e046a 2063 if (*s) {
90e5519e
NC
2064 const STRLEN len = strlen(s);
2065 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn(s, len));
cd81e915 2066 start_force(PL_curforce);
9ded7720 2067 NEXTVAL_NEXTTOKE.opval = o;
79072805 2068 force_next(WORD);
748a9306 2069 if (kind) {
11343788 2070 o->op_private = OPpCONST_ENTERED;
55497cff
PP
2071 /* XXX see note in pp_entereval() for why we forgo typo
2072 warnings if the symbol must be introduced in an eval.
2073 GSAR 96-10-12 */
90e5519e
NC
2074 gv_fetchpvn_flags(s, len,
2075 PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2076 : GV_ADD,
2077 kind == '$' ? SVt_PV :
2078 kind == '@' ? SVt_PVAV :
2079 kind == '%' ? SVt_PVHV :
a0d0e21e 2080 SVt_PVGV
90e5519e 2081 );
748a9306 2082 }
79072805
LW
2083 }
2084}
2085
1571675a
GS
2086NV
2087Perl_str_to_version(pTHX_ SV *sv)
2088{
2089 NV retval = 0.0;
2090 NV nshift = 1.0;
2091 STRLEN len;
cfd0369c 2092 const char *start = SvPV_const(sv,len);
9d4ba2ae 2093 const char * const end = start + len;
504618e9 2094 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
7918f24d
NC
2095
2096 PERL_ARGS_ASSERT_STR_TO_VERSION;
2097
1571675a 2098 while (start < end) {
ba210ebe 2099 STRLEN skip;
1571675a
GS
2100 UV n;
2101 if (utf)
9041c2e3 2102 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
2103 else {
2104 n = *(U8*)start;
2105 skip = 1;
2106 }
2107 retval += ((NV)n)/nshift;
2108 start += skip;
2109 nshift *= 1000;
2110 }
2111 return retval;
2112}
2113
4e553d73 2114/*
ffb4593c
NT
2115 * S_force_version
2116 * Forces the next token to be a version number.
e759cc13
RGS
2117 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2118 * and if "guessing" is TRUE, then no new token is created (and the caller
2119 * must use an alternative parsing method).
ffb4593c
NT
2120 */
2121
76e3520e 2122STATIC char *
e759cc13 2123S_force_version(pTHX_ char *s, int guessing)
89bfa8cd 2124{
97aff369 2125 dVAR;
5f66b61c 2126 OP *version = NULL;
44dcb63b 2127 char *d;
5db06880
NC
2128#ifdef PERL_MAD
2129 I32 startoff = s - SvPVX(PL_linestr);
2130#endif
89bfa8cd 2131
7918f24d
NC
2132 PERL_ARGS_ASSERT_FORCE_VERSION;
2133
29595ff2 2134 s = SKIPSPACE1(s);
89bfa8cd 2135
44dcb63b 2136 d = s;
dd629d5b 2137 if (*d == 'v')
44dcb63b 2138 d++;
44dcb63b 2139 if (isDIGIT(*d)) {
e759cc13
RGS
2140 while (isDIGIT(*d) || *d == '_' || *d == '.')
2141 d++;
5db06880
NC
2142#ifdef PERL_MAD
2143 if (PL_madskills) {
cd81e915 2144 start_force(PL_curforce);
5db06880
NC
2145 curmad('X', newSVpvn(s,d-s));
2146 }
2147#endif
4e4da3ac 2148 if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
dd629d5b 2149 SV *ver;
8d08d9ba
DG
2150#ifdef USE_LOCALE_NUMERIC
2151 char *loc = setlocale(LC_NUMERIC, "C");
2152#endif
6154021b 2153 s = scan_num(s, &pl_yylval);
8d08d9ba
DG
2154#ifdef USE_LOCALE_NUMERIC
2155 setlocale(LC_NUMERIC, loc);
2156#endif
6154021b 2157 version = pl_yylval.opval;
dd629d5b
GS
2158 ver = cSVOPx(version)->op_sv;
2159 if (SvPOK(ver) && !SvNIOK(ver)) {
862a34c6 2160 SvUPGRADE(ver, SVt_PVNV);
9d6ce603 2161 SvNV_set(ver, str_to_version(ver));
1571675a 2162 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 2163 }
89bfa8cd 2164 }
5db06880
NC
2165 else if (guessing) {
2166#ifdef PERL_MAD
2167 if (PL_madskills) {
cd81e915
NC
2168 sv_free(PL_nextwhite); /* let next token collect whitespace */
2169 PL_nextwhite = 0;
5db06880
NC
2170 s = SvPVX(PL_linestr) + startoff;
2171 }
2172#endif
e759cc13 2173 return s;
5db06880 2174 }
89bfa8cd
PP
2175 }
2176
5db06880
NC
2177#ifdef PERL_MAD
2178 if (PL_madskills && !version) {
cd81e915
NC
2179 sv_free(PL_nextwhite); /* let next token collect whitespace */
2180 PL_nextwhite = 0;
5db06880
NC
2181 s = SvPVX(PL_linestr) + startoff;
2182 }
2183#endif
89bfa8cd 2184 /* NOTE: The parser sees the package name and the VERSION swapped */
cd81e915 2185 start_force(PL_curforce);
9ded7720 2186 NEXTVAL_NEXTTOKE.opval = version;
4e553d73 2187 force_next(WORD);
89bfa8cd 2188
e759cc13 2189 return s;
89bfa8cd
PP
2190}
2191
ffb4593c 2192/*
91152fc1
DG
2193 * S_force_strict_version
2194 * Forces the next token to be a version number using strict syntax rules.
2195 */
2196
2197STATIC char *
2198S_force_strict_version(pTHX_ char *s)
2199{
2200 dVAR;
2201 OP *version = NULL;
2202#ifdef PERL_MAD
2203 I32 startoff = s - SvPVX(PL_linestr);
2204#endif
2205 const char *errstr = NULL;
2206
2207 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2208
2209 while (isSPACE(*s)) /* leading whitespace */
2210 s++;
2211
2212 if (is_STRICT_VERSION(s,&errstr)) {
2213 SV *ver = newSV(0);
2214 s = (char *)scan_version(s, ver, 0);
2215 version = newSVOP(OP_CONST, 0, ver);
2216 }
4e4da3ac
Z
2217 else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2218 (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2219 {
91152fc1
DG
2220 PL_bufptr = s;
2221 if (errstr)
2222 yyerror(errstr); /* version required */
2223 return s;
2224 }
2225
2226#ifdef PERL_MAD
2227 if (PL_madskills && !version) {
2228 sv_free(PL_nextwhite); /* let next token collect whitespace */
2229 PL_nextwhite = 0;
2230 s = SvPVX(PL_linestr) + startoff;
2231 }
2232#endif
2233 /* NOTE: The parser sees the package name and the VERSION swapped */
2234 start_force(PL_curforce);
2235 NEXTVAL_NEXTTOKE.opval = version;
2236 force_next(WORD);
2237
2238 return s;
2239}
2240
2241/*
ffb4593c
NT
2242 * S_tokeq
2243 * Tokenize a quoted string passed in as an SV. It finds the next
2244 * chunk, up to end of string or a backslash. It may make a new
2245 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2246 * turns \\ into \.
2247 */
2248
76e3520e 2249STATIC SV *
cea2e8a9 2250S_tokeq(pTHX_ SV *sv)
79072805 2251{
97aff369 2252 dVAR;
79072805
LW
2253 register char *s;
2254 register char *send;
2255 register char *d;
b3ac6de7
IZ
2256 STRLEN len = 0;
2257 SV *pv = sv;
79072805 2258
7918f24d
NC
2259 PERL_ARGS_ASSERT_TOKEQ;
2260
79072805 2261 if (!SvLEN(sv))
b3ac6de7 2262 goto finish;
79072805 2263
a0d0e21e 2264 s = SvPV_force(sv, len);
21a311ee 2265 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 2266 goto finish;
463ee0b2 2267 send = s + len;
dcb21ed6
NC
2268 /* This is relying on the SV being "well formed" with a trailing '\0' */
2269 while (s < send && !(*s == '\\' && s[1] == '\\'))
79072805
LW
2270 s++;
2271 if (s == send)
b3ac6de7 2272 goto finish;
79072805 2273 d = s;
be4731d2 2274 if ( PL_hints & HINT_NEW_STRING ) {
59cd0e26 2275 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
be4731d2 2276 }
79072805
LW
2277 while (s < send) {
2278 if (*s == '\\') {
a0d0e21e 2279 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
2280 s++; /* all that, just for this */
2281 }
2282 *d++ = *s++;
2283 }
2284 *d = '\0';
95a20fc0 2285 SvCUR_set(sv, d - SvPVX_const(sv));
b3ac6de7 2286 finish:
3280af22 2287 if ( PL_hints & HINT_NEW_STRING )
eb0d8d16 2288 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
79072805
LW
2289 return sv;
2290}
2291
ffb4593c
NT
2292/*
2293 * Now come three functions related to double-quote context,
2294 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2295 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2296 * interact with PL_lex_state, and create fake ( ... ) argument lists
2297 * to handle functions and concatenation.
2298 * They assume that whoever calls them will be setting up a fake
2299 * join call, because each subthing puts a ',' after it. This lets
2300 * "lower \luPpEr"
2301 * become
2302 * join($, , 'lower ', lcfirst( 'uPpEr', ) ,)
2303 *
2304 * (I'm not sure whether the spurious commas at the end of lcfirst's
2305 * arguments and join's arguments are created or not).
2306 */
2307
2308/*
2309 * S_sublex_start
6154021b 2310 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
ffb4593c
NT
2311 *
2312 * Pattern matching will set PL_lex_op to the pattern-matching op to
6154021b 2313 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
ffb4593c
NT
2314 *
2315 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2316 *
2317 * Everything else becomes a FUNC.
2318 *
2319 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2320 * had an OP_CONST or OP_READLINE). This just sets us up for a
2321 * call to S_sublex_push().
2322 */
2323
76e3520e 2324STATIC I32
cea2e8a9 2325S_sublex_start(pTHX)
79072805 2326{
97aff369 2327 dVAR;
6154021b 2328 register const I32 op_type = pl_yylval.ival;
79072805
LW
2329
2330 if (op_type == OP_NULL) {
6154021b 2331 pl_yylval.opval = PL_lex_op;
5f66b61c 2332 PL_lex_op = NULL;
79072805
LW
2333 return THING;
2334 }
2335 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 2336 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
2337
2338 if (SvTYPE(sv) == SVt_PVIV) {
2339 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2340 STRLEN len;
96a5add6 2341 const char * const p = SvPV_const(sv, len);
740cce10 2342 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
b3ac6de7
IZ
2343 SvREFCNT_dec(sv);
2344 sv = nsv;
4e553d73 2345 }
6154021b 2346 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
a0714e2c 2347 PL_lex_stuff = NULL;
6f33ba73
RGS
2348 /* Allow <FH> // "foo" */
2349 if (op_type == OP_READLINE)
2350 PL_expect = XTERMORDORDOR;
79072805
LW
2351 return THING;
2352 }
e3f73d4e
RGS
2353 else if (op_type == OP_BACKTICK && PL_lex_op) {
2354 /* readpipe() vas overriden */
2355 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
6154021b 2356 pl_yylval.opval = PL_lex_op;
9b201d7d 2357 PL_lex_op = NULL;
e3f73d4e
RGS
2358 PL_lex_stuff = NULL;
2359 return THING;
2360 }
79072805 2361
3280af22 2362 PL_sublex_info.super_state = PL_lex_state;
eac04b2e 2363 PL_sublex_info.sub_inwhat = (U16)op_type;
3280af22
NIS
2364 PL_sublex_info.sub_op = PL_lex_op;
2365 PL_lex_state = LEX_INTERPPUSH;
55497cff 2366
3280af22
NIS
2367 PL_expect = XTERM;
2368 if (PL_lex_op) {
6154021b 2369 pl_yylval.opval = PL_lex_op;
5f66b61c 2370 PL_lex_op = NULL;
55497cff
PP
2371 return PMFUNC;
2372 }
2373 else
2374 return FUNC;
2375}
2376
ffb4593c
NT
2377/*
2378 * S_sublex_push
2379 * Create a new scope to save the lexing state. The scope will be
2380 * ended in S_sublex_done. Returns a '(', starting the function arguments
2381 * to the uc, lc, etc. found before.
2382 * Sets PL_lex_state to LEX_INTERPCONCAT.
2383 */
2384
76e3520e 2385STATIC I32
cea2e8a9 2386S_sublex_push(pTHX)
55497cff 2387{
27da23d5 2388 dVAR;
f46d017c 2389 ENTER;
55497cff 2390
3280af22 2391 PL_lex_state = PL_sublex_info.super_state;
651b5b28 2392 SAVEBOOL(PL_lex_dojoin);
3280af22 2393 SAVEI32(PL_lex_brackets);
78cdf107
Z
2394 SAVEI32(PL_lex_allbrackets);
2395 SAVEI8(PL_lex_fakeeof);
3280af22
NIS
2396 SAVEI32(PL_lex_casemods);
2397 SAVEI32(PL_lex_starts);
651b5b28 2398 SAVEI8(PL_lex_state);
7766f137 2399 SAVEVPTR(PL_lex_inpat);
98246f1e 2400 SAVEI16(PL_lex_inwhat);
57843af0 2401 SAVECOPLINE(PL_curcop);
3280af22 2402 SAVEPPTR(PL_bufptr);
8452ff4b 2403 SAVEPPTR(PL_bufend);
3280af22
NIS
2404 SAVEPPTR(PL_oldbufptr);
2405 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
2406 SAVEPPTR(PL_last_lop);
2407 SAVEPPTR(PL_last_uni);
3280af22
NIS
2408 SAVEPPTR(PL_linestart);
2409 SAVESPTR(PL_linestr);
8edd5f42
RGS
2410 SAVEGENERICPV(PL_lex_brackstack);
2411 SAVEGENERICPV(PL_lex_casestack);
3280af22
NIS
2412
2413 PL_linestr = PL_lex_stuff;
a0714e2c 2414 PL_lex_stuff = NULL;
3280af22 2415
9cbb5ea2
GS
2416 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2417 = SvPVX(PL_linestr);
3280af22 2418 PL_bufend += SvCUR(PL_linestr);
bd61b366 2419 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2420 SAVEFREESV(PL_linestr);
2421
2422 PL_lex_dojoin = FALSE;
2423 PL_lex_brackets = 0;
78cdf107
Z
2424 PL_lex_allbrackets = 0;
2425 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
a02a5408
JC
2426 Newx(PL_lex_brackstack, 120, char);
2427 Newx(PL_lex_casestack, 12, char);
3280af22
NIS
2428 PL_lex_casemods = 0;
2429 *PL_lex_casestack = '\0';
2430 PL_lex_starts = 0;
2431 PL_lex_state = LEX_INTERPCONCAT;
eb160463 2432 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
3280af22
NIS
2433
2434 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
bb16bae8 2435 if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
3280af22
NIS
2436 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2437 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 2438 else
5f66b61c 2439 PL_lex_inpat = NULL;
79072805 2440
55497cff 2441 return '(';
79072805
LW
2442}
2443
ffb4593c
NT
2444/*
2445 * S_sublex_done
2446 * Restores lexer state after a S_sublex_push.
2447 */
2448
76e3520e 2449STATIC I32
cea2e8a9 2450S_sublex_done(pTHX)
79072805 2451{
27da23d5 2452 dVAR;
3280af22 2453 if (!PL_lex_starts++) {
396482e1 2454 SV * const sv = newSVpvs("");
9aa983d2
JH
2455 if (SvUTF8(PL_linestr))
2456 SvUTF8_on(sv);
3280af22 2457 PL_expect = XOPERATOR;
6154021b 2458 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
2459 return THING;
2460 }
2461
3280af22
NIS
2462 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2463 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 2464 return yylex();
79072805
LW
2465 }
2466
ffb4593c 2467 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
bb16bae8 2468 assert(PL_lex_inwhat != OP_TRANSR);
3280af22
NIS
2469 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2470 PL_linestr = PL_lex_repl;
2471 PL_lex_inpat = 0;
2472 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2473 PL_bufend += SvCUR(PL_linestr);
bd61b366 2474 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2475 SAVEFREESV(PL_linestr);
2476 PL_lex_dojoin = FALSE;
2477 PL_lex_brackets = 0;
78cdf107
Z
2478 PL_lex_allbrackets = 0;
2479 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
3280af22
NIS
2480 PL_lex_casemods = 0;
2481 *PL_lex_casestack = '\0';
2482 PL_lex_starts = 0;
25da4f38 2483 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
2484 PL_lex_state = LEX_INTERPNORMAL;
2485 PL_lex_starts++;
e9fa98b2
HS
2486 /* we don't clear PL_lex_repl here, so that we can check later
2487 whether this is an evalled subst; that means we rely on the
2488 logic to ensure sublex_done() is called again only via the
2489 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 2490 }
e9fa98b2 2491 else {
3280af22 2492 PL_lex_state = LEX_INTERPCONCAT;
a0714e2c 2493 PL_lex_repl = NULL;
e9fa98b2 2494 }
79072805 2495 return ',';
ffed7fef
LW
2496 }
2497 else {
5db06880
NC
2498#ifdef PERL_MAD
2499 if (PL_madskills) {
cd81e915
NC
2500 if (PL_thiswhite) {
2501 if (!PL_endwhite)
6b29d1f5 2502 PL_endwhite = newSVpvs("");
cd81e915
NC
2503 sv_catsv(PL_endwhite, PL_thiswhite);
2504 PL_thiswhite = 0;
2505 }
2506 if (PL_thistoken)
76f68e9b 2507 sv_setpvs(PL_thistoken,"");
5db06880 2508 else
cd81e915 2509 PL_realtokenstart = -1;
5db06880
NC
2510 }
2511#endif
f46d017c 2512 LEAVE;
3280af22
NIS
2513 PL_bufend = SvPVX(PL_linestr);
2514 PL_bufend += SvCUR(PL_linestr);
2515 PL_expect = XOPERATOR;
09bef843 2516 PL_sublex_info.sub_inwhat = 0;
79072805 2517 return ')';
ffed7fef
LW
2518 }
2519}
2520
02aa26ce
NT
2521/*
2522 scan_const
2523
2524 Extracts a pattern, double-quoted string, or transliteration. This
2525 is terrifying code.
2526
94def140 2527 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
3280af22 2528 processing a pattern (PL_lex_inpat is true), a transliteration
94def140 2529 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
02aa26ce 2530
94def140
ST
2531 Returns a pointer to the character scanned up to. If this is
2532 advanced from the start pointer supplied (i.e. if anything was
9b599b2a 2533 successfully parsed), will leave an OP for the substring scanned
6154021b 2534 in pl_yylval. Caller must intuit reason for not parsing further
9b599b2a
GS
2535 by looking at the next characters herself.
2536
02aa26ce
NT
2537 In patterns:
2538 backslashes:
ff3f963a 2539 constants: \N{NAME} only
02aa26ce
NT
2540 case and quoting: \U \Q \E
2541 stops on @ and $, but not for $ as tail anchor
2542
2543 In transliterations:
2544 characters are VERY literal, except for - not at the start or end
94def140
ST
2545 of the string, which indicates a range. If the range is in bytes,
2546 scan_const expands the range to the full set of intermediate
2547 characters. If the range is in utf8, the hyphen is replaced with
2548 a certain range mark which will be handled by pmtrans() in op.c.
02aa26ce
NT
2549
2550 In double-quoted strings:
2551 backslashes:
2552 double-quoted style: \r and \n
ff3f963a 2553 constants: \x31, etc.
94def140 2554 deprecated backrefs: \1 (in substitution replacements)
02aa26ce
NT
2555 case and quoting: \U \Q \E
2556 stops on @ and $
2557
2558 scan_const does *not* construct ops to handle interpolated strings.
2559 It stops processing as soon as it finds an embedded $ or @ variable
2560 and leaves it to the caller to work out what's going on.
2561
94def140
ST
2562 embedded arrays (whether in pattern or not) could be:
2563 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2564
2565 $ in double-quoted strings must be the symbol of an embedded scalar.
02aa26ce
NT
2566
2567 $ in pattern could be $foo or could be tail anchor. Assumption:
2568 it's a tail anchor if $ is the last thing in the string, or if it's
94def140 2569 followed by one of "()| \r\n\t"
02aa26ce
NT
2570
2571 \1 (backreferences) are turned into $1
2572
2573 The structure of the code is
2574 while (there's a character to process) {
94def140
ST
2575 handle transliteration ranges
2576 skip regexp comments /(?#comment)/ and codes /(?{code})/
2577 skip #-initiated comments in //x patterns
2578 check for embedded arrays
02aa26ce
NT
2579 check for embedded scalars
2580 if (backslash) {
94def140 2581 deprecate \1 in substitution replacements
02aa26ce
NT
2582 handle string-changing backslashes \l \U \Q \E, etc.
2583 switch (what was escaped) {
94def140 2584 handle \- in a transliteration (becomes a literal -)
ff3f963a 2585 if a pattern and not \N{, go treat as regular character
94def140
ST
2586 handle \132 (octal characters)
2587 handle \x15 and \x{1234} (hex characters)
ff3f963a 2588 handle \N{name} (named characters, also \N{3,5} in a pattern)
94def140
ST
2589 handle \cV (control characters)
2590 handle printf-style backslashes (\f, \r, \n, etc)
02aa26ce 2591 } (end switch)
77a135fe 2592 continue
02aa26ce 2593 } (end if backslash)
77a135fe 2594 handle regular character
02aa26ce 2595 } (end while character to read)
4e553d73 2596
02aa26ce
NT
2597*/
2598
76e3520e 2599STATIC char *
cea2e8a9 2600S_scan_const(pTHX_ char *start)
79072805 2601{
97aff369 2602 dVAR;
3280af22 2603 register char *send = PL_bufend; /* end of the constant */
77a135fe
KW
2604 SV *sv = newSV(send - start); /* sv for the constant. See
2605 note below on sizing. */
02aa26ce
NT
2606 register char *s = start; /* start of the constant */
2607 register char *d = SvPVX(sv); /* destination for copies */
2608 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 2609 bool didrange = FALSE; /* did we just finish a range? */
2b9d42f0 2610 I32 has_utf8 = FALSE; /* Output constant is UTF8 */
77a135fe
KW
2611 I32 this_utf8 = UTF; /* Is the source string assumed
2612 to be UTF8? But, this can
2613 show as true when the source
2614 isn't utf8, as for example
2615 when it is entirely composed
2616 of hex constants */
2617
2618 /* Note on sizing: The scanned constant is placed into sv, which is
2619 * initialized by newSV() assuming one byte of output for every byte of
2620 * input. This routine expects newSV() to allocate an extra byte for a
2621 * trailing NUL, which this routine will append if it gets to the end of
2622 * the input. There may be more bytes of input than output (eg., \N{LATIN
2623 * CAPITAL LETTER A}), or more output than input if the constant ends up
2624 * recoded to utf8, but each time a construct is found that might increase
2625 * the needed size, SvGROW() is called. Its size parameter each time is
2626 * based on the best guess estimate at the time, namely the length used so
2627 * far, plus the length the current construct will occupy, plus room for
2628 * the trailing NUL, plus one byte for every input byte still unscanned */
2629
012bcf8d 2630 UV uv;
4c3a8340
ST
2631#ifdef EBCDIC
2632 UV literal_endpoint = 0;
e294cc5d 2633 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
4c3a8340 2634#endif
012bcf8d 2635
7918f24d
NC
2636 PERL_ARGS_ASSERT_SCAN_CONST;
2637
bb16bae8 2638 assert(PL_lex_inwhat != OP_TRANSR);
2b9d42f0
NIS
2639 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2640 /* If we are doing a trans and we know we want UTF8 set expectation */
2641 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
2642 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2643 }
2644
2645
79072805 2646 while (s < send || dorange) {
ff3f963a 2647
02aa26ce 2648 /* get transliterations out of the way (they're most literal) */
3280af22 2649 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 2650 /* expand a range A-Z to the full set of characters. AIE! */
79072805 2651 if (dorange) {
1ba5c669
JH
2652 I32 i; /* current expanded character */
2653 I32 min; /* first character in range */
2654 I32 max; /* last character in range */
02aa26ce 2655
e294cc5d
JH
2656#ifdef EBCDIC
2657 UV uvmax = 0;
2658#endif
2659
2660 if (has_utf8
2661#ifdef EBCDIC
2662 && !native_range
2663#endif
2664 ) {
9d4ba2ae 2665 char * const c = (char*)utf8_hop((U8*)d, -1);
8973db79
JH
2666 char *e = d++;
2667 while (e-- > c)
2668 *(e + 1) = *e;
25716404 2669 *c = (char)UTF_TO_NATIVE(0xff);
8973db79
JH
2670 /* mark the range as done, and continue */
2671 dorange = FALSE;
2672 didrange = TRUE;
2673 continue;
2674 }
2b9d42f0 2675
95a20fc0 2676 i = d - SvPVX_const(sv); /* remember current offset */
e294cc5d
JH
2677#ifdef EBCDIC
2678 SvGROW(sv,
2679 SvLEN(sv) + (has_utf8 ?
2680 (512 - UTF_CONTINUATION_MARK +
2681 UNISKIP(0x100))
2682 : 256));
2683 /* How many two-byte within 0..255: 128 in UTF-8,
2684 * 96 in UTF-8-mod. */
2685#else
9cbb5ea2 2686 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
e294cc5d 2687#endif
9cbb5ea2 2688 d = SvPVX(sv) + i; /* refresh d after realloc */
e294cc5d
JH
2689#ifdef EBCDIC
2690 if (has_utf8) {
2691 int j;
2692 for (j = 0; j <= 1; j++) {
2693 char * const c = (char*)utf8_hop((U8*)d, -1);
2694 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2695 if (j)
2696 min = (U8)uv;
2697 else if (uv < 256)
2698 max = (U8)uv;
2699 else {
2700 max = (U8)0xff; /* only to \xff */
2701 uvmax = uv; /* \x{100} to uvmax */
2702 }
2703 d = c; /* eat endpoint chars */
2704 }
2705 }
2706 else {
2707#endif
2708 d -= 2; /* eat the first char and the - */
2709 min = (U8)*d; /* first char in range */
2710 max = (U8)d[1]; /* last char in range */
2711#ifdef EBCDIC
2712 }
2713#endif
8ada0baa 2714
c2e66d9e 2715 if (min > max) {
01ec43d0 2716 Perl_croak(aTHX_
d1573ac7 2717 "Invalid range \"%c-%c\" in transliteration operator",
1ba5c669 2718 (char)min, (char)max);
c2e66d9e
GS
2719 }
2720
c7f1f016 2721#ifdef EBCDIC
4c3a8340
ST
2722 if (literal_endpoint == 2 &&
2723 ((isLOWER(min) && isLOWER(max)) ||
2724 (isUPPER(min) && isUPPER(max)))) {
8ada0baa
JH
2725 if (isLOWER(min)) {
2726 for (i = min; i <= max; i++)
2727 if (isLOWER(i))
db42d148 2728 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
2729 } else {
2730 for (i = min; i <= max; i++)
2731 if (isUPPER(i))
db42d148 2732 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
2733 }
2734 }
2735 else
2736#endif
2737 for (i = min; i <= max; i++)
e294cc5d
JH
2738#ifdef EBCDIC
2739 if (has_utf8) {
2740 const U8 ch = (U8)NATIVE_TO_UTF(i);
2741 if (UNI_IS_INVARIANT(ch))
2742 *d++ = (U8)i;
2743 else {
2744 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2745 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2746 }
2747 }
2748 else
2749#endif
2750 *d++ = (char)i;
2751
2752#ifdef EBCDIC
2753 if (uvmax) {
2754 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2755 if (uvmax > 0x101)
2756 *d++ = (char)UTF_TO_NATIVE(0xff);
2757 if (uvmax > 0x100)
2758 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2759 }
2760#endif
02aa26ce
NT
2761
2762 /* mark the range as done, and continue */
79072805 2763 dorange = FALSE;
01ec43d0 2764 didrange = TRUE;
4c3a8340
ST
2765#ifdef EBCDIC
2766 literal_endpoint = 0;
2767#endif
79072805 2768 continue;
4e553d73 2769 }
02aa26ce
NT
2770
2771 /* range begins (ignore - as first or last char) */
79072805 2772 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 2773 if (didrange) {
1fafa243 2774 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 2775 }
e294cc5d
JH
2776 if (has_utf8
2777#ifdef EBCDIC
2778 && !native_range
2779#endif
2780 ) {
25716404 2781 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
2782 s++;
2783 continue;
2784 }
79072805
LW
2785 dorange = TRUE;
2786 s++;
01ec43d0
GS
2787 }
2788 else {
2789 didrange = FALSE;
4c3a8340
ST
2790#ifdef EBCDIC
2791 literal_endpoint = 0;
e294cc5d 2792 native_range = TRUE;
4c3a8340 2793#endif
01ec43d0 2794 }
79072805 2795 }
02aa26ce
NT
2796
2797 /* if we get here, we're not doing a transliteration */
2798
0f5d15d6
IZ
2799 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2800 except for the last char, which will be done separately. */
3280af22 2801 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395 2802 if (s[2] == '#') {
e994fd66 2803 while (s+1 < send && *s != ')')
db42d148 2804 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
155aba94
GS
2805 }
2806 else if (s[2] == '{' /* This should match regcomp.c */
67edc0c9 2807 || (s[2] == '?' && s[3] == '{'))
155aba94 2808 {
cc6b7395 2809 I32 count = 1;
0f5d15d6 2810 char *regparse = s + (s[2] == '{' ? 3 : 4);
cc6b7395
IZ
2811 char c;
2812
d9f97599
GS
2813 while (count && (c = *regparse)) {
2814 if (c == '\\' && regparse[1])
2815 regparse++;
4e553d73 2816 else if (c == '{')
cc6b7395 2817 count++;
4e553d73 2818 else if (c == '}')
cc6b7395 2819 count--;
d9f97599 2820 regparse++;
cc6b7395 2821 }
e994fd66 2822 if (*regparse != ')')
5bdf89e7 2823 regparse--; /* Leave one char for continuation. */
0f5d15d6 2824 while (s < regparse)
db42d148 2825 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
cc6b7395 2826 }
748a9306 2827 }
02aa26ce
NT
2828
2829 /* likewise skip #-initiated comments in //x patterns */
3280af22 2830 else if (*s == '#' && PL_lex_inpat &&
73134a2e 2831 ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
748a9306 2832 while (s+1 < send && *s != '\n')
db42d148 2833 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
748a9306 2834 }
02aa26ce 2835
5d1d4326 2836 /* check for embedded arrays
da6eedaa 2837 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
5d1d4326 2838 */
1749ea0d
ST
2839 else if (*s == '@' && s[1]) {
2840 if (isALNUM_lazy_if(s+1,UTF))
2841 break;
2842 if (strchr(":'{$", s[1]))
2843 break;
2844 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2845 break; /* in regexp, neither @+ nor @- are interpolated */
2846 }
02aa26ce
NT
2847
2848 /* check for embedded scalars. only stop if we're sure it's a
2849 variable.
2850 */
79072805 2851 else if (*s == '$') {
3280af22 2852 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 2853 break;
77772344 2854 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
a2a5de95
NC
2855 if (s[1] == '\\') {
2856 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2857 "Possible unintended interpolation of $\\ in regex");
77772344 2858 }
79072805 2859 break; /* in regexp, $ might be tail anchor */
77772344 2860 }
79072805 2861 }
02aa26ce 2862
2b9d42f0
NIS
2863 /* End of else if chain - OP_TRANS rejoin rest */
2864
02aa26ce 2865 /* backslashes */
79072805 2866 if (*s == '\\' && s+1 < send) {
ff3f963a
KW
2867 char* e; /* Can be used for ending '}', etc. */
2868
79072805 2869 s++;
02aa26ce 2870
7d0fc23c
KW
2871 /* warn on \1 - \9 in substitution replacements, but note that \11
2872 * is an octal; and \19 is \1 followed by '9' */
3280af22 2873 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 2874 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 2875 {
a2a5de95 2876 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
79072805
LW
2877 *--s = '$';
2878 break;
2879 }
02aa26ce
NT
2880
2881 /* string-change backslash escapes */
3280af22 2882 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) {
79072805
LW
2883 --s;
2884 break;
2885 }
ff3f963a
KW
2886 /* In a pattern, process \N, but skip any other backslash escapes.
2887 * This is because we don't want to translate an escape sequence
2888 * into a meta symbol and have the regex compiler use the meta
2889 * symbol meaning, e.g. \x{2E} would be confused with a dot. But
2890 * in spite of this, we do have to process \N here while the proper
2891 * charnames handler is in scope. See bugs #56444 and #62056.
2892 * There is a complication because \N in a pattern may also stand
2893 * for 'match a non-nl', and not mean a charname, in which case its
2894 * processing should be deferred to the regex compiler. To be a
2895 * charname it must be followed immediately by a '{', and not look
2896 * like \N followed by a curly quantifier, i.e., not something like
2897 * \N{3,}. regcurly returns a boolean indicating if it is a legal
2898 * quantifier */
2899 else if (PL_lex_inpat
2900 && (*s != 'N'
2901 || s[1] != '{'
2902 || regcurly(s + 1)))
2903 {
cc74c5bd
ST
2904 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2905 goto default_action;
2906 }
02aa26ce 2907
79072805 2908 switch (*s) {
02aa26ce
NT
2909
2910 /* quoted - in transliterations */
79072805 2911 case '-':
3280af22 2912 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
2913 *d++ = *s++;
2914 continue;
2915 }
2916 /* FALL THROUGH */
2917 default:
11b8faa4 2918 {
a2a5de95
NC
2919 if ((isALPHA(*s) || isDIGIT(*s)))
2920 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
2921 "Unrecognized escape \\%c passed through",
2922 *s);
11b8faa4 2923 /* default action is to copy the quoted character */
f9a63242 2924 goto default_action;
11b8faa4 2925 }
02aa26ce 2926
632403cc 2927 /* eg. \132 indicates the octal constant 0132 */
79072805
LW
2928 case '0': case '1': case '2': case '3':
2929 case '4': case '5': case '6': case '7':
ba210ebe 2930 {
53305cf1
NC
2931 I32 flags = 0;
2932 STRLEN len = 3;
77a135fe 2933 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
ba210ebe
JH
2934 s += len;
2935 }
012bcf8d 2936 goto NUM_ESCAPE_INSERT;
02aa26ce 2937
f0a2b745
KW
2938 /* eg. \o{24} indicates the octal constant \024 */
2939 case 'o':
2940 {
2941 STRLEN len;
454155d9 2942 const char* error;
f0a2b745 2943
454155d9 2944 bool valid = grok_bslash_o(s, &uv, &len, &error, 1);
f0a2b745 2945 s += len;
454155d9 2946 if (! valid) {
f0a2b745
KW
2947 yyerror(error);
2948 continue;
2949 }
2950 goto NUM_ESCAPE_INSERT;
2951 }
2952
77a135fe 2953 /* eg. \x24 indicates the hex constant 0x24 */
79072805 2954 case 'x':
a0ed51b3
LW
2955 ++s;
2956 if (*s == '{') {
9d4ba2ae 2957 char* const e = strchr(s, '}');
a4c04bdc
NC
2958 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
2959 PERL_SCAN_DISALLOW_PREFIX;
53305cf1 2960 STRLEN len;
355860ce 2961
53305cf1 2962 ++s;
adaeee49 2963 if (!e) {
a0ed51b3 2964 yyerror("Missing right brace on \\x{}");
355860ce 2965 continue;
ba210ebe 2966 }
53305cf1 2967 len = e - s;
77a135fe 2968 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
ba210ebe 2969 s = e + 1;
a0ed51b3
LW
2970 }
2971 else {
ba210ebe 2972 {
53305cf1 2973 STRLEN len = 2;
a4c04bdc 2974 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
77a135fe 2975 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
ba210ebe
JH
2976 s += len;
2977 }
012bcf8d
GS
2978 }
2979
2980 NUM_ESCAPE_INSERT:
ff3f963a
KW
2981 /* Insert oct or hex escaped character. There will always be
2982 * enough room in sv since such escapes will be longer than any
2983 * UTF-8 sequence they can end up as, except if they force us
2984 * to recode the rest of the string into utf8 */
ba7cea30 2985
77a135fe 2986 /* Here uv is the ordinal of the next character being added in
ff3f963a 2987 * unicode (converted from native). */
77a135fe 2988 if (!UNI_IS_INVARIANT(uv)) {
9aa983d2 2989 if (!has_utf8 && uv > 255) {
77a135fe
KW
2990 /* Might need to recode whatever we have accumulated so
2991 * far if it contains any chars variant in utf8 or
2992 * utf-ebcdic. */
2993
2994 SvCUR_set(sv, d - SvPVX_const(sv));
2995 SvPOK_on(sv);
2996 *d = '\0';
77a135fe 2997 /* See Note on sizing above. */
7bf79863
KW
2998 sv_utf8_upgrade_flags_grow(sv,
2999 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3000 UNISKIP(uv) + (STRLEN)(send - s) + 1);
77a135fe
KW
3001 d = SvPVX(sv) + SvCUR(sv);
3002 has_utf8 = TRUE;
012bcf8d
GS
3003 }
3004
77a135fe
KW
3005 if (has_utf8) {
3006 d = (char*)uvuni_to_utf8((U8*)d, uv);
f9a63242
JH
3007 if (PL_lex_inwhat == OP_TRANS &&
3008 PL_sublex_info.sub_op) {
3009 PL_sublex_info.sub_op->op_private |=
3010 (PL_lex_repl ? OPpTRANS_FROM_UTF
3011 : OPpTRANS_TO_UTF);
f9a63242 3012 }
e294cc5d
JH
3013#ifdef EBCDIC
3014 if (uv > 255 && !dorange)
3015 native_range = FALSE;
3016#endif
012bcf8d 3017 }
a0ed51b3 3018 else {
012bcf8d 3019 *d++ = (char)uv;
a0ed51b3 3020 }
012bcf8d
GS
3021 }
3022 else {
c4d5f83a 3023 *d++ = (char) uv;
a0ed51b3 3024 }
79072805 3025 continue;
02aa26ce 3026
4a2d328f 3027 case 'N':
ff3f963a
KW
3028 /* In a non-pattern \N must be a named character, like \N{LATIN
3029 * SMALL LETTER A} or \N{U+0041}. For patterns, it also can
3030 * mean to match a non-newline. For non-patterns, named
3031 * characters are converted to their string equivalents. In
3032 * patterns, named characters are not converted to their
3033 * ultimate forms for the same reasons that other escapes
3034 * aren't. Instead, they are converted to the \N{U+...} form
3035 * to get the value from the charnames that is in effect right
3036 * now, while preserving the fact that it was a named character
3037 * so that the regex compiler knows this */
3038
3039 /* This section of code doesn't generally use the
3040 * NATIVE_TO_NEED() macro to transform the input. I (khw) did
3041 * a close examination of this macro and determined it is a
3042 * no-op except on utfebcdic variant characters. Every
3043 * character generated by this that would normally need to be
3044 * enclosed by this macro is invariant, so the macro is not
7538f724
KW
3045 * needed, and would complicate use of copy(). XXX There are
3046 * other parts of this file where the macro is used
3047 * inconsistently, but are saved by it being a no-op */
ff3f963a
KW
3048
3049 /* The structure of this section of code (besides checking for
3050 * errors and upgrading to utf8) is:
3051 * Further disambiguate between the two meanings of \N, and if
3052 * not a charname, go process it elsewhere
0a96133f
KW
3053 * If of form \N{U+...}, pass it through if a pattern;
3054 * otherwise convert to utf8
3055 * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3056 * pattern; otherwise convert to utf8 */
ff3f963a
KW
3057
3058 /* Here, s points to the 'N'; the test below is guaranteed to
3059 * succeed if we are being called on a pattern as we already
3060 * know from a test above that the next character is a '{'.
3061 * On a non-pattern \N must mean 'named sequence, which
3062 * requires braces */
3063 s++;
3064 if (*s != '{') {
3065 yyerror("Missing braces on \\N{}");
3066 continue;
3067 }
3068 s++;
3069
0a96133f 3070 /* If there is no matching '}', it is an error. */
ff3f963a
KW
3071 if (! (e = strchr(s, '}'))) {
3072 if (! PL_lex_inpat) {
5777a3f7 3073 yyerror("Missing right brace on \\N{}");
0a96133f
KW
3074 } else {
3075 yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
dbc0d4f2 3076 }
0a96133f 3077 continue;
ff3f963a 3078 }
cddc7ef4 3079
ff3f963a 3080 /* Here it looks like a named character */
cddc7ef4 3081
ff3f963a
KW
3082 if (PL_lex_inpat) {
3083
3084 /* XXX This block is temporary code. \N{} implies that the
3085 * pattern is to have Unicode semantics, and therefore
3086 * currently has to be encoded in utf8. By putting it in
3087 * utf8 now, we save a whole pass in the regular expression
3088 * compiler. Once that code is changed so Unicode
3089 * semantics doesn't necessarily have to be in utf8, this
da3a4baf
KW
3090 * block should be removed. However, the code that parses
3091 * the output of this would have to be changed to not
3092 * necessarily expect utf8 */
ff3f963a 3093 if (!has_utf8) {
77a135fe 3094 SvCUR_set(sv, d - SvPVX_const(sv));
f08d6ad9 3095 SvPOK_on(sv);
e4f3eed8 3096 *d = '\0';
77a135fe 3097 /* See Note on sizing above. */
7bf79863 3098 sv_utf8_upgrade_flags_grow(sv,
ff3f963a
KW
3099 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3100 /* 5 = '\N{' + cur char + NUL */
3101 (STRLEN)(send - s) + 5);
f08d6ad9 3102 d = SvPVX(sv) + SvCUR(sv);
89491803 3103 has_utf8 = TRUE;
ff3f963a
KW
3104 }
3105 }
423cee85 3106
ff3f963a
KW
3107 if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3108 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3109 | PERL_SCAN_DISALLOW_PREFIX;
3110 STRLEN len;
3111
3112 /* For \N{U+...}, the '...' is a unicode value even on
3113 * EBCDIC machines */
3114 s += 2; /* Skip to next char after the 'U+' */
3115 len = e - s;
3116 uv = grok_hex(s, &len, &flags, NULL);
3117 if (len == 0 || len != (STRLEN)(e - s)) {
3118 yyerror("Invalid hexadecimal number in \\N{U+...}");
3119 s = e + 1;
3120 continue;
3121 }
3122
3123 if (PL_lex_inpat) {
3124
3125 /* Pass through to the regex compiler unchanged. The
3126 * reason we evaluated the number above is to make sure
0a96133f 3127 * there wasn't a syntax error. */
ff3f963a
KW
3128 s -= 5; /* Include the '\N{U+' */
3129 Copy(s, d, e - s + 1, char); /* 1 = include the } */
3130 d += e - s + 1;
3131 }
3132 else { /* Not a pattern: convert the hex to string */
3133
3134 /* If destination is not in utf8, unconditionally
3135 * recode it to be so. This is because \N{} implies
3136 * Unicode semantics, and scalars have to be in utf8
3137 * to guarantee those semantics */
3138 if (! has_utf8) {
3139 SvCUR_set(sv, d - SvPVX_const(sv));
3140 SvPOK_on(sv);
3141 *d = '\0';
3142 /* See Note on sizing above. */
3143 sv_utf8_upgrade_flags_grow(
3144 sv,
3145 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3146 UNISKIP(uv) + (STRLEN)(send - e) + 1);
3147 d = SvPVX(sv) + SvCUR(sv);
3148 has_utf8 = TRUE;
3149 }
3150
3151 /* Add the string to the output */
3152 if (UNI_IS_INVARIANT(uv)) {
3153 *d++ = (char) uv;
3154 }
3155 else d = (char*)uvuni_to_utf8((U8*)d, uv);
3156 }
3157 }
3158 else { /* Here is \N{NAME} but not \N{U+...}. */
3159
3160 SV *res; /* result from charnames */
3161 const char *str; /* the string in 'res' */
3162 STRLEN len; /* its length */
3163
3164 /* Get the value for NAME */
3165 res = newSVpvn(s, e - s);
3166 res = new_constant( NULL, 0, "charnames",
3167 /* includes all of: \N{...} */
3168 res, NULL, s - 3, e - s + 4 );
3169
3170 /* Most likely res will be in utf8 already since the
3171 * standard charnames uses pack U, but a custom translator
3172 * can leave it otherwise, so make sure. XXX This can be
3173 * revisited to not have charnames use utf8 for characters
3174 * that don't need it when regexes don't have to be in utf8
3175 * for Unicode semantics. If doing so, remember EBCDIC */
3176 sv_utf8_upgrade(res);
3177 str = SvPV_const(res, len);
3178
3179 /* Don't accept malformed input */
3180 if (! is_utf8_string((U8 *) str, len)) {
3181 yyerror("Malformed UTF-8 returned by \\N");
3182 }
3183 else if (PL_lex_inpat) {
3184
3185 if (! len) { /* The name resolved to an empty string */
3186 Copy("\\N{}", d, 4, char);
3187 d += 4;
3188 }
3189 else {
3190 /* In order to not lose information for the regex
3191 * compiler, pass the result in the specially made
3192 * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3193 * the code points in hex of each character
3194 * returned by charnames */
3195
3196 const char *str_end = str + len;
3197 STRLEN char_length; /* cur char's byte length */
3198 STRLEN output_length; /* and the number of bytes
3199 after this is translated
3200 into hex digits */
3201 const STRLEN off = d - SvPVX_const(sv);
3202
3203 /* 2 hex per byte; 2 chars for '\N'; 2 chars for
3204 * max('U+', '.'); and 1 for NUL */
3205 char hex_string[2 * UTF8_MAXBYTES + 5];
3206
3207 /* Get the first character of the result. */
3208 U32 uv = utf8n_to_uvuni((U8 *) str,
3209 len,
3210 &char_length,
3211 UTF8_ALLOW_ANYUV);
3212
3213 /* The call to is_utf8_string() above hopefully
3214 * guarantees that there won't be an error. But
3215 * it's easy here to make sure. The function just
3216 * above warns and returns 0 if invalid utf8, but
3217 * it can also return 0 if the input is validly a
3218 * NUL. Disambiguate */
3219 if (uv == 0 && NATIVE_TO_ASCII(*str) != '\0') {
3220 uv = UNICODE_REPLACEMENT;
3221 }
3222
3223 /* Convert first code point to hex, including the
3224 * boiler plate before it */
78c35590 3225 output_length =
3353de27
NC
3226 my_snprintf(hex_string, sizeof(hex_string),
3227 "\\N{U+%X", (unsigned int) uv);
ff3f963a
KW
3228
3229 /* Make sure there is enough space to hold it */
3230 d = off + SvGROW(sv, off
3231 + output_length
3232 + (STRLEN)(send - e)
3233 + 2); /* '}' + NUL */
3234 /* And output it */
3235 Copy(hex_string, d, output_length, char);
3236 d += output_length;
3237
3238 /* For each subsequent character, append dot and
3239 * its ordinal in hex */
3240 while ((str += char_length) < str_end) {
3241 const STRLEN off = d - SvPVX_const(sv);
3242 U32 uv = utf8n_to_uvuni((U8 *) str,
3243 str_end - str,
3244 &char_length,
3245 UTF8_ALLOW_ANYUV);
3246 if (uv == 0 && NATIVE_TO_ASCII(*str) != '\0') {
3247 uv = UNICODE_REPLACEMENT;
3248 }
3249
78c35590 3250 output_length =
3353de27
NC
3251 my_snprintf(hex_string, sizeof(hex_string),
3252 ".%X", (unsigned int) uv);
ff3f963a
KW
3253
3254 d = off + SvGROW(sv, off
3255 + output_length
3256 + (STRLEN)(send - e)
3257 + 2); /* '}' + NUL */
3258 Copy(hex_string, d, output_length, char);
3259 d += output_length;
3260 }
3261
3262 *d++ = '}'; /* Done. Add the trailing brace */
3263 }
3264 }
3265 else { /* Here, not in a pattern. Convert the name to a
3266 * string. */
3267
3268 /* If destination is not in utf8, unconditionally
3269 * recode it to be so. This is because \N{} implies
3270 * Unicode semantics, and scalars have to be in utf8
3271 * to guarantee those semantics */
3272 if (! has_utf8) {
3273 SvCUR_set(sv, d - SvPVX_const(sv));
3274 SvPOK_on(sv);
3275 *d = '\0';
3276 /* See Note on sizing above. */
3277 sv_utf8_upgrade_flags_grow(sv,
3278 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3279 len + (STRLEN)(send - s) + 1);
3280 d = SvPVX(sv) + SvCUR(sv);
3281 has_utf8 = TRUE;
3282 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
3283
3284 /* See Note on sizing above. (NOTE: SvCUR() is not
3285 * set correctly here). */
3286 const STRLEN off = d - SvPVX_const(sv);
3287 d = off + SvGROW(sv, off + len + (STRLEN)(send - s) + 1);
3288 }
3289 Copy(str, d, len, char);
3290 d += len;
423cee85 3291 }
423cee85 3292 SvREFCNT_dec(res);
cb233ae3
KW
3293
3294 /* Deprecate non-approved name syntax */
3295 if (ckWARN_d(WARN_DEPRECATED)) {
3296 bool problematic = FALSE;
3297 char* i = s;
3298
3299 /* For non-ut8 input, look to see that the first
3300 * character is an alpha, then loop through the rest
3301 * checking that each is a continuation */
3302 if (! this_utf8) {
3303 if (! isALPHAU(*i)) problematic = TRUE;
3304 else for (i = s + 1; i < e; i++) {
3305 if (isCHARNAME_CONT(*i)) continue;
3306 problematic = TRUE;
3307 break;
3308 }
3309 }
3310 else {
3311 /* Similarly for utf8. For invariants can check
3312 * directly. We accept anything above the latin1
3313 * range because it is immaterial to Perl if it is
3314 * correct or not, and is expensive to check. But
3315 * it is fairly easy in the latin1 range to convert
3316 * the variants into a single character and check
3317 * those */
3318 if (UTF8_IS_INVARIANT(*i)) {
3319 if (! isALPHAU(*i)) problematic = TRUE;
3320 } else if (UTF8_IS_DOWNGRADEABLE_START(*i)) {
81c14aa2 3321 if (! isALPHAU(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*i,
cb233ae3
KW
3322 *(i+1)))))
3323 {
3324 problematic = TRUE;
3325 }
3326 }
3327 if (! problematic) for (i = s + UTF8SKIP(s);
3328 i < e;
3329 i+= UTF8SKIP(i))
3330 {
3331 if (UTF8_IS_INVARIANT(*i)) {
3332 if (isCHARNAME_CONT(*i)) continue;
3333 } else if (! UTF8_IS_DOWNGRADEABLE_START(*i)) {
3334 continue;
3335 } else if (isCHARNAME_CONT(
3336 UNI_TO_NATIVE(
81c14aa2 3337 TWO_BYTE_UTF8_TO_UNI(*i, *(i+1)))))
cb233ae3
KW
3338 {
3339 continue;
3340 }
3341 problematic = TRUE;
3342 break;
3343 }
3344 }
3345 if (problematic) {
6e1bad6c
KW
3346 /* The e-i passed to the final %.*s makes sure that
3347 * should the trailing NUL be missing that this
3348 * print won't run off the end of the string */
cb233ae3 3349 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
b00fc8d4
NC
3350 "Deprecated character in \\N{...}; marked by <-- HERE in \\N{%.*s<-- HERE %.*s",
3351 (int)(i - s + 1), s, (int)(e - i), i + 1);
cb233ae3
KW
3352 }
3353 }
3354 } /* End \N{NAME} */
ff3f963a
KW
3355#ifdef EBCDIC
3356 if (!dorange)
3357 native_range = FALSE; /* \N{} is defined to be Unicode */
3358#endif
3359 s = e + 1; /* Point to just after the '}' */
423cee85
JH
3360 continue;
3361
02aa26ce 3362 /* \c is a control character */
79072805
LW
3363 case 'c':
3364 s++;
961ce445 3365 if (s < send) {
f9d13529 3366 *d++ = grok_bslash_c(*s++, 1);
ba210ebe 3367 }
961ce445
RGS
3368 else {
3369 yyerror("Missing control char name in \\c");
3370 }
79072805 3371 continue;
02aa26ce
NT
3372
3373 /* printf-style backslashes, formfeeds, newlines, etc */
79072805 3374 case 'b':
db42d148 3375 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
79072805
LW
3376 break;
3377 case 'n':
db42d148 3378 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
79072805
LW
3379 break;
3380 case 'r':
db42d148 3381 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
79072805
LW
3382 break;
3383 case 'f':
db42d148 3384 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
79072805
LW
3385 break;
3386 case 't':
db42d148 3387 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
79072805 3388 break;
34a3fe2a 3389 case 'e':
db42d148 3390 *d++ = ASCII_TO_NEED(has_utf8,'\033');
34a3fe2a
PP
3391 break;
3392 case 'a':
db42d148 3393 *d++ = ASCII_TO_NEED(has_utf8,'\007');
79072805 3394 break;
02aa26ce
NT
3395 } /* end switch */
3396
79072805
LW
3397 s++;
3398 continue;
02aa26ce 3399 } /* end if (backslash) */
4c3a8340
ST
3400#ifdef EBCDIC
3401 else
3402 literal_endpoint++;
3403#endif
02aa26ce 3404
f9a63242 3405 default_action:
77a135fe
KW
3406 /* If we started with encoded form, or already know we want it,
3407 then encode the next character */
3408 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
2b9d42f0 3409 STRLEN len = 1;
77a135fe
KW
3410
3411
3412 /* One might think that it is wasted effort in the case of the
3413 * source being utf8 (this_utf8 == TRUE) to take the next character
3414 * in the source, convert it to an unsigned value, and then convert
3415 * it back again. But the source has not been validated here. The
3416 * routine that does the conversion checks for errors like
3417 * malformed utf8 */
3418
5f66b61c
AL
3419 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
3420 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
77a135fe
KW
3421 if (!has_utf8) {
3422 SvCUR_set(sv, d - SvPVX_const(sv));
3423 SvPOK_on(sv);
3424 *d = '\0';
77a135fe 3425 /* See Note on sizing above. */
7bf79863
KW
3426 sv_utf8_upgrade_flags_grow(sv,
3427 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3428 need + (STRLEN)(send - s) + 1);
77a135fe
KW
3429 d = SvPVX(sv) + SvCUR(sv);
3430 has_utf8 = TRUE;
3431 } else if (need > len) {
3432 /* encoded value larger than old, may need extra space (NOTE:
3433 * SvCUR() is not set correctly here). See Note on sizing
3434 * above. */
9d4ba2ae 3435 const STRLEN off = d - SvPVX_const(sv);
77a135fe 3436 d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
2b9d42f0 3437 }
77a135fe
KW
3438 s += len;
3439
5f66b61c 3440 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
e294cc5d
JH
3441#ifdef EBCDIC
3442 if (uv > 255 && !dorange)
3443 native_range = FALSE;
3444#endif
2b9d42f0
NIS
3445 }
3446 else {
3447 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
3448 }
02aa26ce
NT
3449 } /* while loop to process each character */
3450
3451 /* terminate the string and set up the sv */
79072805 3452 *d = '\0';
95a20fc0 3453 SvCUR_set(sv, d - SvPVX_const(sv));
2b9d42f0 3454 if (SvCUR(sv) >= SvLEN(sv))
d0063567 3455 Perl_croak(aTHX_ "panic: constant overflowed allocated space");
2b9d42f0 3456
79072805 3457 SvPOK_on(sv);
9f4817db 3458 if (PL_encoding && !has_utf8) {
d0063567
DK
3459 sv_recode_to_utf8(sv, PL_encoding);
3460 if (SvUTF8(sv))
3461 has_utf8 = TRUE;
9f4817db 3462 }
2b9d42f0 3463 if (has_utf8) {
7e2040f0 3464 SvUTF8_on(sv);
2b9d42f0 3465 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
d0063567 3466 PL_sublex_info.sub_op->op_private |=
2b9d42f0
NIS
3467 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3468 }
3469 }
79072805 3470
02aa26ce 3471 /* shrink the sv if we allocated more than we used */
79072805 3472 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1da4ca5f 3473 SvPV_shrink_to_cur(sv);
79072805 3474 }
02aa26ce 3475
6154021b 3476 /* return the substring (via pl_yylval) only if we parsed anything */
3280af22 3477 if (s > PL_bufptr) {
eb0d8d16
NC
3478 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
3479 const char *const key = PL_lex_inpat ? "qr" : "q";
3480 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
3481 const char *type;
3482 STRLEN typelen;
3483
3484 if (PL_lex_inwhat == OP_TRANS) {
3485 type = "tr";
3486 typelen = 2;
3487 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
3488 type = "s";
3489 typelen = 1;
3490 } else {
3491 type = "qq";
3492 typelen = 2;
3493 }
3494
3495 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
3496 type, typelen);
3497 }
6154021b 3498 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
b3ac6de7 3499 } else
8990e307 3500 SvREFCNT_dec(sv);
79072805
LW
3501 return s;
3502}
3503
ffb4593c
NT
3504/* S_intuit_more
3505 * Returns TRUE if there's more to the expression (e.g., a subscript),
3506 * FALSE otherwise.
ffb4593c
NT
3507 *
3508 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
3509 *
3510 * ->[ and ->{ return TRUE
3511 * { and [ outside a pattern are always subscripts, so return TRUE
3512 * if we're outside a pattern and it's not { or [, then return FALSE
3513 * if we're in a pattern and the first char is a {
3514 * {4,5} (any digits around the comma) returns FALSE
3515 * if we're in a pattern and the first char is a [
3516 * [] returns FALSE
3517 * [SOMETHING] has a funky algorithm to decide whether it's a
3518 * character class or not. It has to deal with things like
3519 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
3520 * anything else returns TRUE
3521 */
3522
9cbb5ea2
GS
3523/* This is the one truly awful dwimmer necessary to conflate C and sed. */
3524
76e3520e 3525STATIC int
cea2e8a9 3526S_intuit_more(pTHX_ register char *s)
79072805 3527{
97aff369 3528 dVAR;
7918f24d
NC
3529
3530 PERL_ARGS_ASSERT_INTUIT_MORE;
3531
3280af22 3532 if (PL_lex_brackets)
79072805
LW
3533 return TRUE;
3534 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
3535 return TRUE;
3536 if (*s != '{' && *s != '[')
3537 return FALSE;
3280af22 3538 if (!PL_lex_inpat)
79072805
LW
3539 return TRUE;
3540
3541 /* In a pattern, so maybe we have {n,m}. */
3542 if (*s == '{') {
b3155d95 3543 if (regcurly(s)) {
79072805 3544 return FALSE;
b3155d95 3545 }
79072805 3546 return TRUE;
79072805
LW
3547 }
3548
3549 /* On the other hand, maybe we have a character class */
3550
3551 s++;
3552 if (*s == ']' || *s == '^')
3553 return FALSE;
3554 else {
ffb4593c 3555 /* this is terrifying, and it works */
79072805
LW
3556 int weight = 2; /* let's weigh the evidence */
3557 char seen[256];
f27ffc4a 3558 unsigned char un_char = 255, last_un_char;
9d4ba2ae 3559 const char * const send = strchr(s,']');
3280af22 3560 char tmpbuf[sizeof PL_tokenbuf * 4];
79072805
LW
3561
3562 if (!send) /* has to be an expression */
3563 return TRUE;
3564
3565 Zero(seen,256,char);
3566 if (*s == '$')
3567 weight -= 3;
3568 else if (isDIGIT(*s)) {
3569 if (s[1] != ']') {
3570 if (isDIGIT(s[1]) && s[2] == ']')
3571 weight -= 10;
3572 }
3573 else
3574 weight -= 100;
3575 }
3576 for (; s < send; s++) {
3577 last_un_char = un_char;
3578 un_char = (unsigned char)*s;
3579 switch (*s) {
3580 case '@':
3581 case '&':
3582 case '$':
3583 weight -= seen[un_char] * 10;
7e2040f0 3584 if (isALNUM_lazy_if(s+1,UTF)) {
90e5519e 3585 int len;
8903cb82 3586 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
90e5519e
NC
3587 len = (int)strlen(tmpbuf);
3588 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PV))
79072805
LW
3589 weight -= 100;
3590 else
3591 weight -= 10;
3592 }
3593 else if (*s == '$' && s[1] &&
93a17b20
LW
3594 strchr("[#!%*<>()-=",s[1])) {
3595 if (/*{*/ strchr("])} =",s[2]))
79072805
LW
3596 weight -= 10;
3597 else
3598 weight -= 1;
3599 }
3600 break;
3601 case '\\':
3602 un_char = 254;
3603 if (s[1]) {
93a17b20 3604 if (strchr("wds]",s[1]))
79072805 3605 weight += 100;
10edeb5d 3606 else if (seen[(U8)'\''] || seen[(U8)'"'])
79072805 3607 weight += 1;
93a17b20 3608 else if (strchr("rnftbxcav",s[1]))
79072805
LW
3609 weight += 40;
3610 else if (isDIGIT(s[1])) {
3611 weight += 40;
3612 while (s[1] && isDIGIT(s[1]))
3613 s++;
3614 }
3615 }
3616 else
3617 weight += 100;
3618 break;
3619 case '-':
3620 if (s[1] == '\\')
3621 weight += 50;
93a17b20 3622 if (strchr("aA01! ",last_un_char))
79072805 3623 weight += 30;
93a17b20 3624 if (strchr("zZ79~",s[1]))
79072805 3625 weight += 30;
f27ffc4a
GS
3626 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
3627 weight -= 5; /* cope with negative subscript */
79072805
LW
3628 break;
3629 default:
3792a11b
NC
3630 if (!isALNUM(last_un_char)
3631 && !(last_un_char == '$' || last_un_char == '@'
3632 || last_un_char == '&')
3633 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
79072805
LW
3634 char *d = tmpbuf;
3635 while (isALPHA(*s))
3636 *d++ = *s++;
3637 *d = '\0';
5458a98a 3638 if (keyword(tmpbuf, d - tmpbuf, 0))
79072805
LW
3639 weight -= 150;
3640 }
3641 if (un_char == last_un_char + 1)
3642 weight += 5;
3643 weight -= seen[un_char];
3644 break;
3645 }
3646 seen[un_char]++;
3647 }
3648 if (weight >= 0) /* probably a character class */
3649 return FALSE;
3650 }
3651
3652 return TRUE;
3653}
ffed7fef 3654
ffb4593c
NT
3655/*
3656 * S_intuit_method
3657 *
3658 * Does all the checking to disambiguate
3659 * foo bar
3660 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
3661 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
3662 *
3663 * First argument is the stuff after the first token, e.g. "bar".
3664 *
3665 * Not a method if bar is a filehandle.
3666 * Not a method if foo is a subroutine prototyped to take a filehandle.
3667 * Not a method if it's really "Foo $bar"
3668 * Method if it's "foo $bar"
3669 * Not a method if it's really "print foo $bar"
3670 * Method if it's really "foo package::" (interpreted as package->foo)
8f8cf39c 3671 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
3cb0bbe5 3672 * Not a method if bar is a filehandle or package, but is quoted with
ffb4593c
NT
3673 * =>
3674 */
3675
76e3520e 3676STATIC int
62d55b22 3677S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
a0d0e21e 3678{
97aff369 3679 dVAR;
a0d0e21e 3680 char *s = start + (*start == '$');
3280af22 3681 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
3682 STRLEN len;
3683 GV* indirgv;
5db06880
NC
3684#ifdef PERL_MAD
3685 int soff;
3686#endif
a0d0e21e 3687
7918f24d
NC
3688 PERL_ARGS_ASSERT_INTUIT_METHOD;
3689
a0d0e21e 3690 if (gv) {
62d55b22 3691 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
a0d0e21e 3692 return 0;
62d55b22
NC
3693 if (cv) {
3694 if (SvPOK(cv)) {
3695 const char *proto = SvPVX_const(cv);
3696 if (proto) {
3697 if (*proto == ';')
3698 proto++;
3699 if (*proto == '*')
3700 return 0;
3701 }
b6c543e3
IZ
3702 }
3703 } else
c35e046a 3704 gv = NULL;
a0d0e21e 3705 }
8903cb82 3706 s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len);
ffb4593c
NT
3707 /* start is the beginning of the possible filehandle/object,
3708 * and s is the end of it
3709 * tmpbuf is a copy of it
3710 */
3711
a0d0e21e 3712 if (*start == '$') {
3ef1310e
RGS
3713 if (gv || PL_last_lop_op == OP_PRINT || PL_last_lop_op == OP_SAY ||
3714 isUPPER(*PL_tokenbuf))
a0d0e21e 3715 return 0;
5db06880
NC
3716#ifdef PERL_MAD
3717 len = start - SvPVX(PL_linestr);
3718#endif
29595ff2 3719 s = PEEKSPACE(s);
f0092767 3720#ifdef PERL_MAD
5db06880
NC
3721 start = SvPVX(PL_linestr) + len;
3722#endif
3280af22
NIS
3723 PL_bufptr = start;
3724 PL_expect = XREF;
a0d0e21e
LW
3725 return *s == '(' ? FUNCMETH : METHOD;
3726 }
5458a98a 3727 if (!keyword(tmpbuf, len, 0)) {
c3e0f903
GS
3728 if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') {
3729 len -= 2;
3730 tmpbuf[len] = '\0';
5db06880
NC
3731#ifdef PERL_MAD
3732 soff = s - SvPVX(PL_linestr);
3733#endif
c3e0f903
GS
3734 goto bare_package;
3735 }
90e5519e 3736 indirgv = gv_fetchpvn_flags(tmpbuf, len, 0, SVt_PVCV);
8ebc5c01 3737 if (indirgv && GvCVu(indirgv))
a0d0e21e
LW
3738 return 0;
3739 /* filehandle or package name makes it a method */
da51bb9b 3740 if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, 0)) {
5db06880
NC
3741#ifdef PERL_MAD
3742 soff = s - SvPVX(PL_linestr);
3743#endif
29595ff2 3744 s = PEEKSPACE(s);
3280af22 3745 if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>')
486ec47a 3746 return 0; /* no assumptions -- "=>" quotes bareword */
c3e0f903 3747 bare_package:
cd81e915 3748 start_force(PL_curforce);
9ded7720 3749 NEXTVAL_NEXTTOKE.opval = (OP*)newSVOP(OP_CONST, 0,
64142370 3750 S_newSV_maybe_utf8(aTHX_ tmpbuf, len));
9ded7720 3751 NEXTVAL_NEXTTOKE.opval->op_private = OPpCONST_BARE;
5db06880
NC
3752 if (PL_madskills)
3753 curmad('X', newSVpvn(start,SvPVX(PL_linestr) + soff - start));
3280af22 3754 PL_expect = XTERM;
a0d0e21e 3755 force_next(WORD);
3280af22 3756 PL_bufptr = s;
5db06880
NC
3757#ifdef PERL_MAD
3758 PL_bufptr = SvPVX(PL_linestr) + soff; /* restart before space */
3759#endif
a0d0e21e
LW
3760 return *s == '(' ? FUNCMETH : METHOD;
3761 }
3762 }
3763 return 0;
3764}
3765
16d20bd9 3766/* Encoded script support. filter_add() effectively inserts a
4e553d73 3767 * 'pre-processing' function into the current source input stream.
16d20bd9
AD
3768 * Note that the filter function only applies to the current source file
3769 * (e.g., it will not affect files 'require'd or 'use'd by this one).
3770 *
3771 * The datasv parameter (which may be NULL) can be used to pass
3772 * private data to this instance of the filter. The filter function