This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
minor clean up of the refactoring of d6a4f4b531
[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)
199e78b7
DM
69#define PL_preambled (PL_parser->preambled)
70#define PL_sublex_info (PL_parser->sublex_info)
bdc0bf6f 71#define PL_linestr (PL_parser->linestr)
c2598295
DM
72#define PL_expect (PL_parser->expect)
73#define PL_copline (PL_parser->copline)
f06b5848
DM
74#define PL_bufptr (PL_parser->bufptr)
75#define PL_oldbufptr (PL_parser->oldbufptr)
76#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
77#define PL_linestart (PL_parser->linestart)
78#define PL_bufend (PL_parser->bufend)
79#define PL_last_uni (PL_parser->last_uni)
80#define PL_last_lop (PL_parser->last_lop)
81#define PL_last_lop_op (PL_parser->last_lop_op)
bc177e6b 82#define PL_lex_state (PL_parser->lex_state)
2f9285f8 83#define PL_rsfp (PL_parser->rsfp)
5486870f 84#define PL_rsfp_filters (PL_parser->rsfp_filters)
12bd6ede
DM
85#define PL_in_my (PL_parser->in_my)
86#define PL_in_my_stash (PL_parser->in_my_stash)
14047fc9 87#define PL_tokenbuf (PL_parser->tokenbuf)
670a9cb2 88#define PL_multi_end (PL_parser->multi_end)
13765c85 89#define PL_error_count (PL_parser->error_count)
199e78b7
DM
90
91#ifdef PERL_MAD
92# define PL_endwhite (PL_parser->endwhite)
93# define PL_faketokens (PL_parser->faketokens)
94# define PL_lasttoke (PL_parser->lasttoke)
95# define PL_nextwhite (PL_parser->nextwhite)
96# define PL_realtokenstart (PL_parser->realtokenstart)
97# define PL_skipwhite (PL_parser->skipwhite)
98# define PL_thisclose (PL_parser->thisclose)
99# define PL_thismad (PL_parser->thismad)
100# define PL_thisopen (PL_parser->thisopen)
101# define PL_thisstuff (PL_parser->thisstuff)
102# define PL_thistoken (PL_parser->thistoken)
103# define PL_thiswhite (PL_parser->thiswhite)
fb205e7a
DM
104# define PL_thiswhite (PL_parser->thiswhite)
105# define PL_nexttoke (PL_parser->nexttoke)
106# define PL_curforce (PL_parser->curforce)
107#else
108# define PL_nexttoke (PL_parser->nexttoke)
109# define PL_nexttype (PL_parser->nexttype)
110# define PL_nextval (PL_parser->nextval)
199e78b7
DM
111#endif
112
a1894d81 113static const char* const ident_too_long = "Identifier too long";
8903cb82 114
29595ff2 115#ifdef PERL_MAD
29595ff2 116# define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
cd81e915 117# define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
9ded7720 118#else
5db06880 119# define CURMAD(slot,sv)
9ded7720 120# define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
29595ff2
NC
121#endif
122
a7aaec61
Z
123#define XENUMMASK 0x3f
124#define XFAKEEOF 0x40
125#define XFAKEBRACK 0x80
9059aa12 126
39e02b42
JH
127#ifdef USE_UTF8_SCRIPTS
128# define UTF (!IN_BYTES)
2b9d42f0 129#else
802a15e9 130# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || ( !(PL_parser->lex_flags & LEX_IGNORE_UTF8_HINTS) && (PL_hints & HINT_UTF8)))
2b9d42f0 131#endif
a0ed51b3 132
b1fc3636
CJ
133/* The maximum number of characters preceding the unrecognized one to display */
134#define UNRECOGNIZED_PRECEDE_COUNT 10
135
61f0cdd9 136/* In variables named $^X, these are the legal values for X.
2b92dfce
GS
137 * 1999-02-27 mjd-perl-patch@plover.com */
138#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
139
14bd96d0 140#define SPACE_OR_TAB(c) isBLANK_A(c)
bf4acbe4 141
ffb4593c
NT
142/* LEX_* are values for PL_lex_state, the state of the lexer.
143 * They are arranged oddly so that the guard on the switch statement
79072805 144 * can get by with a single comparison (if the compiler is smart enough).
9da1dd8f
DM
145 *
146 * These values refer to the various states within a sublex parse,
147 * i.e. within a double quotish string
79072805
LW
148 */
149
fb73857a
PP
150/* #define LEX_NOTPARSING 11 is done in perl.h. */
151
b6007c36
DM
152#define LEX_NORMAL 10 /* normal code (ie not within "...") */
153#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
154#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
155#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
156#define LEX_INTERPSTART 6 /* expecting the start of a $var */
157
158 /* at end of code, eg "$x" followed by: */
159#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
160#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
161
162#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
163 string or after \E, $foo, etc */
164#define LEX_INTERPCONST 2 /* NOT USED */
165#define LEX_FORMLINE 1 /* expecting a format line */
166#define LEX_KNOWNEXT 0 /* next token known; just return it */
167
79072805 168
bbf60fe6 169#ifdef DEBUGGING
27da23d5 170static const char* const lex_state_names[] = {
bbf60fe6
DM
171 "KNOWNEXT",
172 "FORMLINE",
173 "INTERPCONST",
174 "INTERPCONCAT",
175 "INTERPENDMAYBE",
176 "INTERPEND",
177 "INTERPSTART",
178 "INTERPPUSH",
179 "INTERPCASEMOD",
180 "INTERPNORMAL",
181 "NORMAL"
182};
183#endif
184
79072805
LW
185#ifdef ff_next
186#undef ff_next
d48672a2
LW
187#endif
188
79072805 189#include "keywords.h"
fe14fcc3 190
ffb4593c
NT
191/* CLINE is a macro that ensures PL_copline has a sane value */
192
ae986130
LW
193#ifdef CLINE
194#undef CLINE
195#endif
57843af0 196#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
3280af22 197
5db06880 198#ifdef PERL_MAD
29595ff2
NC
199# define SKIPSPACE0(s) skipspace0(s)
200# define SKIPSPACE1(s) skipspace1(s)
201# define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
202# define PEEKSPACE(s) skipspace2(s,0)
203#else
204# define SKIPSPACE0(s) skipspace(s)
205# define SKIPSPACE1(s) skipspace(s)
206# define SKIPSPACE2(s,tsv) skipspace(s)
207# define PEEKSPACE(s) skipspace(s)
208#endif
209
ffb4593c
NT
210/*
211 * Convenience functions to return different tokens and prime the
9cbb5ea2 212 * lexer for the next token. They all take an argument.
ffb4593c
NT
213 *
214 * TOKEN : generic token (used for '(', DOLSHARP, etc)
215 * OPERATOR : generic operator
216 * AOPERATOR : assignment operator
217 * PREBLOCK : beginning the block after an if, while, foreach, ...
218 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
219 * PREREF : *EXPR where EXPR is not a simple identifier
220 * TERM : expression term
221 * LOOPX : loop exiting command (goto, last, dump, etc)
222 * FTST : file test operator
223 * FUN0 : zero-argument function
7eb971ee 224 * FUN0OP : zero-argument function, with its op created in this file
2d2e263d 225 * FUN1 : not used, except for not, which isn't a UNIOP
ffb4593c
NT
226 * BOop : bitwise or or xor
227 * BAop : bitwise and
228 * SHop : shift operator
229 * PWop : power operator
9cbb5ea2 230 * PMop : pattern-matching operator
ffb4593c
NT
231 * Aop : addition-level operator
232 * Mop : multiplication-level operator
233 * Eop : equality-testing operator
e5edeb50 234 * Rop : relational operator <= != gt
ffb4593c
NT
235 *
236 * Also see LOP and lop() below.
237 */
238
998054bd 239#ifdef DEBUGGING /* Serve -DT. */
704d4215 240# define REPORT(retval) tokereport((I32)retval, &pl_yylval)
998054bd 241#else
bbf60fe6 242# define REPORT(retval) (retval)
998054bd
SC
243#endif
244
bbf60fe6
DM
245#define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
246#define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
247#define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
248#define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
249#define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
250#define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
251#define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
6154021b
RGS
252#define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
253#define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
254#define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
7eb971ee 255#define FUN0OP(f) return (pl_yylval.opval=f, CLINE, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0OP))
6154021b
RGS
256#define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
257#define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
258#define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
259#define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
260#define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
261#define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
262#define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
263#define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
264#define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
265#define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
2f3197b3 266
a687059c
LW
267/* This bit of chicanery makes a unary function followed by
268 * a parenthesis into a function with one argument, highest precedence.
6f33ba73
RGS
269 * The UNIDOR macro is for unary functions that can be followed by the //
270 * operator (such as C<shift // 0>).
a687059c 271 */
d68ce4ac 272#define UNI3(f,x,have_x) { \
6154021b 273 pl_yylval.ival = f; \
d68ce4ac 274 if (have_x) PL_expect = x; \
376fcdbf
AL
275 PL_bufptr = s; \
276 PL_last_uni = PL_oldbufptr; \
277 PL_last_lop_op = f; \
278 if (*s == '(') \
279 return REPORT( (int)FUNC1 ); \
29595ff2 280 s = PEEKSPACE(s); \
376fcdbf
AL
281 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
282 }
d68ce4ac
FC
283#define UNI(f) UNI3(f,XTERM,1)
284#define UNIDOR(f) UNI3(f,XTERMORDORDOR,1)
b5fb7ce3
FC
285#define UNIPROTO(f,optional) { \
286 if (optional) PL_last_uni = PL_oldbufptr; \
22393538
MH
287 OPERATOR(f); \
288 }
a687059c 289
d68ce4ac 290#define UNIBRACK(f) UNI3(f,0,0)
79072805 291
9f68db38 292/* grandfather return to old style */
78cdf107
Z
293#define OLDLOP(f) \
294 do { \
295 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) \
296 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; \
297 pl_yylval.ival = (f); \
298 PL_expect = XTERM; \
299 PL_bufptr = s; \
300 return (int)LSTOP; \
301 } while(0)
79072805 302
83944c01
FC
303#define COPLINE_INC_WITH_HERELINES \
304 STMT_START { \
305 CopLINE_inc(PL_curcop); \
d794b522
FC
306 if (PL_parser->lex_shared->herelines) \
307 CopLINE(PL_curcop) += PL_parser->lex_shared->herelines, \
308 PL_parser->lex_shared->herelines = 0; \
83944c01
FC
309 } STMT_END
310
311
8fa7f367
JH
312#ifdef DEBUGGING
313
6154021b 314/* how to interpret the pl_yylval associated with the token */
bbf60fe6
DM
315enum token_type {
316 TOKENTYPE_NONE,
317 TOKENTYPE_IVAL,
6154021b 318 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
bbf60fe6 319 TOKENTYPE_PVAL,
aeaef349 320 TOKENTYPE_OPVAL
bbf60fe6
DM
321};
322
6d4a66ac
NC
323static struct debug_tokens {
324 const int token;
325 enum token_type type;
326 const char *name;
327} const debug_tokens[] =
9041c2e3 328{
bbf60fe6
DM
329 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
330 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
331 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
332 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
333 { ARROW, TOKENTYPE_NONE, "ARROW" },
334 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
335 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
336 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
337 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
338 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
0d863452 339 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
bbf60fe6
DM
340 { DO, TOKENTYPE_NONE, "DO" },
341 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
342 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
343 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
344 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
345 { ELSE, TOKENTYPE_NONE, "ELSE" },
346 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
347 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
348 { FOR, TOKENTYPE_IVAL, "FOR" },
349 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
705fe0e5
FC
350 { FORMLBRACK, TOKENTYPE_NONE, "FORMLBRACK" },
351 { FORMRBRACK, TOKENTYPE_NONE, "FORMRBRACK" },
bbf60fe6
DM
352 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
353 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
7eb971ee 354 { FUNC0OP, TOKENTYPE_OPVAL, "FUNC0OP" },
bbf60fe6
DM
355 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
356 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
357 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
0d863452 358 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
bbf60fe6
DM
359 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
360 { IF, TOKENTYPE_IVAL, "IF" },
5504e6cf 361 { LABEL, TOKENTYPE_PVAL, "LABEL" },
bbf60fe6
DM
362 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
363 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
364 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
365 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
366 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
367 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
368 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
369 { MY, TOKENTYPE_IVAL, "MY" },
bbf60fe6
DM
370 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
371 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
372 { OROP, TOKENTYPE_IVAL, "OROP" },
373 { OROR, TOKENTYPE_NONE, "OROR" },
374 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
f3f204dc 375 { PEG, TOKENTYPE_NONE, "PEG" },
88e1f1a2
JV
376 { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" },
377 { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" },
bbf60fe6
DM
378 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
379 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
380 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
381 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
382 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
383 { PREINC, TOKENTYPE_NONE, "PREINC" },
384 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
f3f204dc 385 { QWLIST, TOKENTYPE_OPVAL, "QWLIST" },
bbf60fe6
DM
386 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
387 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
f3f204dc 388 { REQUIRE, TOKENTYPE_NONE, "REQUIRE" },
bbf60fe6
DM
389 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
390 { SUB, TOKENTYPE_NONE, "SUB" },
391 { THING, TOKENTYPE_OPVAL, "THING" },
392 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
393 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
394 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
395 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
396 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
397 { USE, TOKENTYPE_IVAL, "USE" },
0d863452 398 { WHEN, TOKENTYPE_IVAL, "WHEN" },
bbf60fe6
DM
399 { WHILE, TOKENTYPE_IVAL, "WHILE" },
400 { WORD, TOKENTYPE_OPVAL, "WORD" },
be25f609 401 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
c35e046a 402 { 0, TOKENTYPE_NONE, NULL }
bbf60fe6
DM
403};
404
6154021b 405/* dump the returned token in rv, plus any optional arg in pl_yylval */
998054bd 406
bbf60fe6 407STATIC int
704d4215 408S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
bbf60fe6 409{
97aff369 410 dVAR;
7918f24d
NC
411
412 PERL_ARGS_ASSERT_TOKEREPORT;
413
bbf60fe6 414 if (DEBUG_T_TEST) {
bd61b366 415 const char *name = NULL;
bbf60fe6 416 enum token_type type = TOKENTYPE_NONE;
f54cb97a 417 const struct debug_tokens *p;
396482e1 418 SV* const report = newSVpvs("<== ");
bbf60fe6 419
f54cb97a 420 for (p = debug_tokens; p->token; p++) {
bbf60fe6
DM
421 if (p->token == (int)rv) {
422 name = p->name;
423 type = p->type;
424 break;
425 }
426 }
427 if (name)
54667de8 428 Perl_sv_catpv(aTHX_ report, name);
74736ae6 429 else if ((char)rv > ' ' && (char)rv <= '~')
4ebc7986 430 {
bbf60fe6 431 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
4ebc7986
FC
432 if ((char)rv == 'p')
433 sv_catpvs(report, " (pending identifier)");
434 }
bbf60fe6 435 else if (!rv)
396482e1 436 sv_catpvs(report, "EOF");
bbf60fe6
DM
437 else
438 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
439 switch (type) {
440 case TOKENTYPE_NONE:
bbf60fe6
DM
441 break;
442 case TOKENTYPE_IVAL:
704d4215 443 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
bbf60fe6
DM
444 break;
445 case TOKENTYPE_OPNUM:
446 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
704d4215 447 PL_op_name[lvalp->ival]);
bbf60fe6
DM
448 break;
449 case TOKENTYPE_PVAL:
704d4215 450 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
bbf60fe6
DM
451 break;
452 case TOKENTYPE_OPVAL:
704d4215 453 if (lvalp->opval) {
401441c0 454 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
704d4215
GG
455 PL_op_name[lvalp->opval->op_type]);
456 if (lvalp->opval->op_type == OP_CONST) {
b6007c36 457 Perl_sv_catpvf(aTHX_ report, " %s",
704d4215 458 SvPEEK(cSVOPx_sv(lvalp->opval)));
b6007c36
DM
459 }
460
461 }
401441c0 462 else
396482e1 463 sv_catpvs(report, "(opval=null)");
bbf60fe6
DM
464 break;
465 }
b6007c36 466 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
bbf60fe6
DM
467 };
468 return (int)rv;
998054bd
SC
469}
470
b6007c36
DM
471
472/* print the buffer with suitable escapes */
473
474STATIC void
15f169a1 475S_printbuf(pTHX_ const char *const fmt, const char *const s)
b6007c36 476{
396482e1 477 SV* const tmp = newSVpvs("");
7918f24d
NC
478
479 PERL_ARGS_ASSERT_PRINTBUF;
480
b6007c36
DM
481 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
482 SvREFCNT_dec(tmp);
483}
484
8fa7f367
JH
485#endif
486
8290c323
NC
487static int
488S_deprecate_commaless_var_list(pTHX) {
489 PL_expect = XTERM;
490 deprecate("comma-less variable list");
491 return REPORT(','); /* grandfather non-comma-format format */
492}
493
ffb4593c
NT
494/*
495 * S_ao
496 *
c963b151
BD
497 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
498 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
ffb4593c
NT
499 */
500
76e3520e 501STATIC int
cea2e8a9 502S_ao(pTHX_ int toketype)
a0d0e21e 503{
97aff369 504 dVAR;
3280af22
NIS
505 if (*PL_bufptr == '=') {
506 PL_bufptr++;
a0d0e21e 507 if (toketype == ANDAND)
6154021b 508 pl_yylval.ival = OP_ANDASSIGN;
a0d0e21e 509 else if (toketype == OROR)
6154021b 510 pl_yylval.ival = OP_ORASSIGN;
c963b151 511 else if (toketype == DORDOR)
6154021b 512 pl_yylval.ival = OP_DORASSIGN;
a0d0e21e
LW
513 toketype = ASSIGNOP;
514 }
515 return toketype;
516}
517
ffb4593c
NT
518/*
519 * S_no_op
520 * When Perl expects an operator and finds something else, no_op
521 * prints the warning. It always prints "<something> found where
522 * operator expected. It prints "Missing semicolon on previous line?"
523 * if the surprise occurs at the start of the line. "do you need to
524 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
525 * where the compiler doesn't know if foo is a method call or a function.
526 * It prints "Missing operator before end of line" if there's nothing
527 * after the missing operator, or "... before <...>" if there is something
528 * after the missing operator.
529 */
530
76e3520e 531STATIC void
15f169a1 532S_no_op(pTHX_ const char *const what, char *s)
463ee0b2 533{
97aff369 534 dVAR;
9d4ba2ae
AL
535 char * const oldbp = PL_bufptr;
536 const bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 537
7918f24d
NC
538 PERL_ARGS_ASSERT_NO_OP;
539
1189a94a
GS
540 if (!s)
541 s = oldbp;
07c798fb 542 else
1189a94a 543 PL_bufptr = s;
734ab321 544 yywarn(Perl_form(aTHX_ "%s found where operator expected", what), UTF ? SVf_UTF8 : 0);
56da5a46
RGS
545 if (ckWARN_d(WARN_SYNTAX)) {
546 if (is_first)
547 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
548 "\t(Missing semicolon on previous line?)\n");
549 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
f54cb97a 550 const char *t;
8a2bca12 551 for (t = PL_oldoldbufptr; (isWORDCHAR_lazy_if(t,UTF) || *t == ':');
734ab321 552 t += UTF ? UTF8SKIP(t) : 1)
c35e046a 553 NOOP;
56da5a46
RGS
554 if (t < PL_bufptr && isSPACE(*t))
555 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
b17a0679
FC
556 "\t(Do you need to predeclare %"UTF8f"?)\n",
557 UTF8fARG(UTF, t - PL_oldoldbufptr, PL_oldoldbufptr));
56da5a46
RGS
558 }
559 else {
560 assert(s >= oldbp);
561 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
b17a0679
FC
562 "\t(Missing operator before %"UTF8f"?)\n",
563 UTF8fARG(UTF, s - oldbp, oldbp));
56da5a46 564 }
07c798fb 565 }
3280af22 566 PL_bufptr = oldbp;
8990e307
LW
567}
568
ffb4593c
NT
569/*
570 * S_missingterm
571 * Complain about missing quote/regexp/heredoc terminator.
d4c19fe8 572 * If it's called with NULL then it cauterizes the line buffer.
ffb4593c
NT
573 * If we're in a delimited string and the delimiter is a control
574 * character, it's reformatted into a two-char sequence like ^C.
575 * This is fatal.
576 */
577
76e3520e 578STATIC void
cea2e8a9 579S_missingterm(pTHX_ char *s)
8990e307 580{
97aff369 581 dVAR;
8990e307
LW
582 char tmpbuf[3];
583 char q;
584 if (s) {
9d4ba2ae 585 char * const nl = strrchr(s,'\n');
d2719217 586 if (nl)
8990e307
LW
587 *nl = '\0';
588 }
463559e7 589 else if (isCNTRL(PL_multi_close)) {
8990e307 590 *tmpbuf = '^';
585ec06d 591 tmpbuf[1] = (char)toCTRL(PL_multi_close);
8990e307
LW
592 tmpbuf[2] = '\0';
593 s = tmpbuf;
594 }
595 else {
eb160463 596 *tmpbuf = (char)PL_multi_close;
8990e307
LW
597 tmpbuf[1] = '\0';
598 s = tmpbuf;
599 }
600 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 601 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 602}
79072805 603
dd0ac2b9
FC
604#include "feature.h"
605
0d863452 606/*
0d863452
RH
607 * Check whether the named feature is enabled.
608 */
26ea9e12 609bool
3fff3427 610Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
0d863452 611{
97aff369 612 dVAR;
4a731d7b 613 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
7918f24d
NC
614
615 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
ca4d40c4
FC
616
617 assert(CURRENT_FEATURE_BUNDLE == FEATURE_BUNDLE_CUSTOM);
7918f24d 618
26ea9e12
NC
619 if (namelen > MAX_FEATURE_LEN)
620 return FALSE;
3fff3427 621 memcpy(&he_name[8], name, namelen);
7d69d4a6 622
c8ca97b0
NC
623 return cBOOL(cop_hints_fetch_pvn(PL_curcop, he_name, 8 + namelen, 0,
624 REFCOUNTED_HE_EXISTS));
0d863452
RH
625}
626
ffb4593c 627/*
9cbb5ea2
GS
628 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
629 * utf16-to-utf8-reversed.
ffb4593c
NT
630 */
631
c39cd008
GS
632#ifdef PERL_CR_FILTER
633static void
634strip_return(SV *sv)
635{
eb578fdb
KW
636 const char *s = SvPVX_const(sv);
637 const char * const e = s + SvCUR(sv);
7918f24d
NC
638
639 PERL_ARGS_ASSERT_STRIP_RETURN;
640
c39cd008
GS
641 /* outer loop optimized to do nothing if there are no CR-LFs */
642 while (s < e) {
643 if (*s++ == '\r' && *s == '\n') {
644 /* hit a CR-LF, need to copy the rest */
eb578fdb 645 char *d = s - 1;
c39cd008
GS
646 *d++ = *s++;
647 while (s < e) {
648 if (*s == '\r' && s[1] == '\n')
649 s++;
650 *d++ = *s++;
651 }
652 SvCUR(sv) -= s - d;
653 return;
654 }
655 }
656}
a868473f 657
76e3520e 658STATIC I32
c39cd008 659S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 660{
f54cb97a 661 const I32 count = FILTER_READ(idx+1, sv, maxlen);
c39cd008
GS
662 if (count > 0 && !maxlen)
663 strip_return(sv);
664 return count;
a868473f
NIS
665}
666#endif
667
ffb4593c 668/*
8eaa0acf
Z
669=for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags
670
671Creates and initialises a new lexer/parser state object, supplying
672a context in which to lex and parse from a new source of Perl code.
673A pointer to the new state object is placed in L</PL_parser>. An entry
674is made on the save stack so that upon unwinding the new state object
675will be destroyed and the former value of L</PL_parser> will be restored.
676Nothing else need be done to clean up the parsing context.
677
678The code to be parsed comes from I<line> and I<rsfp>. I<line>, if
679non-null, provides a string (in SV form) containing code to be parsed.
680A copy of the string is made, so subsequent modification of I<line>
681does not affect parsing. I<rsfp>, if non-null, provides an input stream
682from which code will be read to be parsed. If both are non-null, the
683code in I<line> comes first and must consist of complete lines of input,
684and I<rsfp> supplies the remainder of the source.
685
e368b3bd
FC
686The I<flags> parameter is reserved for future use. Currently it is only
687used by perl internally, so extensions should always pass zero.
8eaa0acf
Z
688
689=cut
690*/
ffb4593c 691
27fcb6ee 692/* LEX_START_SAME_FILTER indicates that this is not a new file, so it
87606032
NC
693 can share filters with the current parser.
694 LEX_START_DONT_CLOSE indicates that the file handle wasn't opened by the
695 caller, hence isn't owned by the parser, so shouldn't be closed on parser
696 destruction. This is used to handle the case of defaulting to reading the
697 script from the standard input because no filename was given on the command
698 line (without getting confused by situation where STDIN has been closed, so
699 the script handle is opened on fd 0) */
27fcb6ee 700
a0d0e21e 701void
8eaa0acf 702Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags)
79072805 703{
97aff369 704 dVAR;
6ef55633 705 const char *s = NULL;
5486870f 706 yy_parser *parser, *oparser;
60d63348 707 if (flags && flags & ~LEX_START_FLAGS)
8eaa0acf 708 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start");
acdf0a21
DM
709
710 /* create and initialise a parser */
711
199e78b7 712 Newxz(parser, 1, yy_parser);
5486870f 713 parser->old_parser = oparser = PL_parser;
acdf0a21
DM
714 PL_parser = parser;
715
28ac2b49
Z
716 parser->stack = NULL;
717 parser->ps = NULL;
718 parser->stack_size = 0;
acdf0a21 719
e3abe207
DM
720 /* on scope exit, free this parser and restore any outer one */
721 SAVEPARSER(parser);
7c4baf47 722 parser->saved_curcop = PL_curcop;
e3abe207 723
acdf0a21 724 /* initialise lexer state */
8990e307 725
fb205e7a
DM
726#ifdef PERL_MAD
727 parser->curforce = -1;
728#else
729 parser->nexttoke = 0;
730#endif
ca4cfd28 731 parser->error_count = oparser ? oparser->error_count : 0;
c2598295 732 parser->copline = NOLINE;
5afb0a62 733 parser->lex_state = LEX_NORMAL;
c2598295 734 parser->expect = XSTATE;
2f9285f8 735 parser->rsfp = rsfp;
27fcb6ee
FC
736 parser->rsfp_filters =
737 !(flags & LEX_START_SAME_FILTER) || !oparser
d3cd8e11
FC
738 ? NULL
739 : MUTABLE_AV(SvREFCNT_inc(
740 oparser->rsfp_filters
741 ? oparser->rsfp_filters
742 : (oparser->rsfp_filters = newAV())
743 ));
2f9285f8 744
199e78b7
DM
745 Newx(parser->lex_brackstack, 120, char);
746 Newx(parser->lex_casestack, 12, char);
747 *parser->lex_casestack = '\0';
d794b522 748 Newxz(parser->lex_shared, 1, LEXSHARED);
02b34bbe 749
10efb74f 750 if (line) {
0528fd32 751 STRLEN len;
10efb74f 752 s = SvPV_const(line, len);
0abcdfa4
FC
753 parser->linestr = flags & LEX_START_COPIED
754 ? SvREFCNT_inc_simple_NN(line)
755 : newSVpvn_flags(s, len, SvUTF8(line));
11076590 756 sv_catpvs(parser->linestr, "\n;");
0abcdfa4
FC
757 } else {
758 parser->linestr = newSVpvs("\n;");
8990e307 759 }
f06b5848
DM
760 parser->oldoldbufptr =
761 parser->oldbufptr =
762 parser->bufptr =
763 parser->linestart = SvPVX(parser->linestr);
764 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
765 parser->last_lop = parser->last_uni = NULL;
87606032
NC
766 parser->lex_flags = flags & (LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES
767 |LEX_DONT_CLOSE_RSFP);
737c24fc 768
60d63348 769 parser->in_pod = parser->filtered = 0;
79072805 770}
a687059c 771
e3abe207
DM
772
773/* delete a parser object */
774
775void
776Perl_parser_free(pTHX_ const yy_parser *parser)
777{
7918f24d
NC
778 PERL_ARGS_ASSERT_PARSER_FREE;
779
7c4baf47 780 PL_curcop = parser->saved_curcop;
bdc0bf6f
DM
781 SvREFCNT_dec(parser->linestr);
782
87606032 783 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
2f9285f8 784 PerlIO_clearerr(parser->rsfp);
799361c3
SH
785 else if (parser->rsfp && (!parser->old_parser ||
786 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
2f9285f8 787 PerlIO_close(parser->rsfp);
5486870f 788 SvREFCNT_dec(parser->rsfp_filters);
10002bc1
FC
789 SvREFCNT_dec(parser->lex_stuff);
790 SvREFCNT_dec(parser->sublex_info.repl);
3ac7ff8f
FC
791
792 Safefree(parser->lex_brackstack);
793 Safefree(parser->lex_casestack);
794 Safefree(parser->lex_shared);
795 PL_parser = parser->old_parser;
796 Safefree(parser);
797}
798
799void
800Perl_parser_free_nexttoke_ops(pTHX_ yy_parser *parser, OPSLAB *slab)
801{
802#ifdef PERL_MAD
803 I32 nexttoke = parser->lasttoke;
804#else
805 I32 nexttoke = parser->nexttoke;
806#endif
807 PERL_ARGS_ASSERT_PARSER_FREE_NEXTTOKE_OPS;
3ce3dcd9
FC
808 while (nexttoke--) {
809#ifdef PERL_MAD
810 if (S_is_opval_token(parser->nexttoke[nexttoke].next_type
3ac7ff8f
FC
811 & 0xffff)
812 && parser->nexttoke[nexttoke].next_val.opval
813 && parser->nexttoke[nexttoke].next_val.opval->op_slabbed
814 && OpSLAB(parser->nexttoke[nexttoke].next_val.opval) == slab) {
815 op_free(parser->nexttoke[nexttoke].next_val.opval);
816 parser->nexttoke[nexttoke].next_val.opval = NULL;
817 }
3ce3dcd9 818#else
3ac7ff8f
FC
819 if (S_is_opval_token(parser->nexttype[nexttoke] & 0xffff)
820 && parser->nextval[nexttoke].opval
821 && parser->nextval[nexttoke].opval->op_slabbed
822 && OpSLAB(parser->nextval[nexttoke].opval) == slab) {
3ce3dcd9 823 op_free(parser->nextval[nexttoke].opval);
3ac7ff8f
FC
824 parser->nextval[nexttoke].opval = NULL;
825 }
3ce3dcd9
FC
826#endif
827 }
e3abe207
DM
828}
829
830
ffb4593c 831/*
f0e67a1d
Z
832=for apidoc AmxU|SV *|PL_parser-E<gt>linestr
833
834Buffer scalar containing the chunk currently under consideration of the
835text currently being lexed. This is always a plain string scalar (for
836which C<SvPOK> is true). It is not intended to be used as a scalar by
837normal scalar means; instead refer to the buffer directly by the pointer
838variables described below.
839
840The lexer maintains various C<char*> pointers to things in the
841C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever
842reallocated, all of these pointers must be updated. Don't attempt to
843do this manually, but rather use L</lex_grow_linestr> if you need to
844reallocate the buffer.
845
846The content of the text chunk in the buffer is commonly exactly one
847complete line of input, up to and including a newline terminator,
848but there are situations where it is otherwise. The octets of the
849buffer may be intended to be interpreted as either UTF-8 or Latin-1.
850The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8>
851flag on this scalar, which may disagree with it.
852
853For direct examination of the buffer, the variable
854L</PL_parser-E<gt>bufend> points to the end of the buffer. The current
855lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use
856of these pointers is usually preferable to examination of the scalar
857through normal scalar means.
858
859=for apidoc AmxU|char *|PL_parser-E<gt>bufend
860
861Direct pointer to the end of the chunk of text currently being lexed, the
862end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr)
863+ SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is
864always located at the end of the buffer, and does not count as part of
865the buffer's contents.
866
867=for apidoc AmxU|char *|PL_parser-E<gt>bufptr
868
869Points to the current position of lexing inside the lexer buffer.
870Characters around this point may be freely examined, within
871the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
872L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be
873interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
874
875Lexing code (whether in the Perl core or not) moves this pointer past
876the characters that it consumes. It is also expected to perform some
877bookkeeping whenever a newline character is consumed. This movement
878can be more conveniently performed by the function L</lex_read_to>,
879which handles newlines appropriately.
880
881Interpretation of the buffer's octets can be abstracted out by
882using the slightly higher-level functions L</lex_peek_unichar> and
883L</lex_read_unichar>.
884
885=for apidoc AmxU|char *|PL_parser-E<gt>linestart
886
887Points to the start of the current line inside the lexer buffer.
888This is useful for indicating at which column an error occurred, and
889not much else. This must be updated by any lexing code that consumes
890a newline; the function L</lex_read_to> handles this detail.
891
892=cut
893*/
894
895/*
896=for apidoc Amx|bool|lex_bufutf8
897
898Indicates whether the octets in the lexer buffer
899(L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
900of Unicode characters. If not, they should be interpreted as Latin-1
901characters. This is analogous to the C<SvUTF8> flag for scalars.
902
903In UTF-8 mode, it is not guaranteed that the lexer buffer actually
904contains valid UTF-8. Lexing code must be robust in the face of invalid
905encoding.
906
907The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
908is significant, but not the whole story regarding the input character
909encoding. Normally, when a file is being read, the scalar contains octets
910and its C<SvUTF8> flag is off, but the octets should be interpreted as
911UTF-8 if the C<use utf8> pragma is in effect. During a string eval,
912however, the scalar may have the C<SvUTF8> flag on, and in this case its
913octets should be interpreted as UTF-8 unless the C<use bytes> pragma
914is in effect. This logic may change in the future; use this function
915instead of implementing the logic yourself.
916
917=cut
918*/
919
920bool
921Perl_lex_bufutf8(pTHX)
922{
923 return UTF;
924}
925
926/*
927=for apidoc Amx|char *|lex_grow_linestr|STRLEN len
928
929Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
930at least I<len> octets (including terminating NUL). Returns a
931pointer to the reallocated buffer. This is necessary before making
932any direct modification of the buffer that would increase its length.
933L</lex_stuff_pvn> provides a more convenient way to insert text into
934the buffer.
935
936Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
937this function updates all of the lexer's variables that point directly
938into the buffer.
939
940=cut
941*/
942
943char *
944Perl_lex_grow_linestr(pTHX_ STRLEN len)
945{
946 SV *linestr;
947 char *buf;
948 STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
c7641931 949 STRLEN linestart_pos, last_uni_pos, last_lop_pos, re_eval_start_pos;
f0e67a1d
Z
950 linestr = PL_parser->linestr;
951 buf = SvPVX(linestr);
952 if (len <= SvLEN(linestr))
953 return buf;
954 bufend_pos = PL_parser->bufend - buf;
955 bufptr_pos = PL_parser->bufptr - buf;
956 oldbufptr_pos = PL_parser->oldbufptr - buf;
957 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
958 linestart_pos = PL_parser->linestart - buf;
959 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
960 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
3328ab5a
FC
961 re_eval_start_pos = PL_parser->lex_shared->re_eval_start ?
962 PL_parser->lex_shared->re_eval_start - buf : 0;
c7641931 963
f0e67a1d 964 buf = sv_grow(linestr, len);
c7641931 965
f0e67a1d
Z
966 PL_parser->bufend = buf + bufend_pos;
967 PL_parser->bufptr = buf + bufptr_pos;
968 PL_parser->oldbufptr = buf + oldbufptr_pos;
969 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
970 PL_parser->linestart = buf + linestart_pos;
971 if (PL_parser->last_uni)
972 PL_parser->last_uni = buf + last_uni_pos;
973 if (PL_parser->last_lop)
974 PL_parser->last_lop = buf + last_lop_pos;
3328ab5a
FC
975 if (PL_parser->lex_shared->re_eval_start)
976 PL_parser->lex_shared->re_eval_start = buf + re_eval_start_pos;
f0e67a1d
Z
977 return buf;
978}
979
980/*
83aa740e 981=for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags
f0e67a1d
Z
982
983Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
984immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
985reallocating the buffer if necessary. This means that lexing code that
986runs later will see the characters as if they had appeared in the input.
987It is not recommended to do this as part of normal parsing, and most
988uses of this facility run the risk of the inserted characters being
989interpreted in an unintended manner.
990
991The string to be inserted is represented by I<len> octets starting
992at I<pv>. These octets are interpreted as either UTF-8 or Latin-1,
993according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
994The characters are recoded for the lexer buffer, according to how the
995buffer is currently being interpreted (L</lex_bufutf8>). If a string
9dcc53ea 996to be inserted is available as a Perl scalar, the L</lex_stuff_sv>
f0e67a1d
Z
997function is more convenient.
998
999=cut
1000*/
1001
1002void
83aa740e 1003Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags)
f0e67a1d 1004{
749123ff 1005 dVAR;
f0e67a1d
Z
1006 char *bufptr;
1007 PERL_ARGS_ASSERT_LEX_STUFF_PVN;
1008 if (flags & ~(LEX_STUFF_UTF8))
1009 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
1010 if (UTF) {
1011 if (flags & LEX_STUFF_UTF8) {
1012 goto plain_copy;
1013 } else {
54d004e8 1014 STRLEN highhalf = 0; /* Count of variants */
83aa740e 1015 const char *p, *e = pv+len;
54d004e8
KW
1016 for (p = pv; p != e; p++) {
1017 if (! UTF8_IS_INVARIANT(*p)) {
1018 highhalf++;
1019 }
1020 }
f0e67a1d
Z
1021 if (!highhalf)
1022 goto plain_copy;
1023 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
1024 bufptr = PL_parser->bufptr;
1025 Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1026 SvCUR_set(PL_parser->linestr,
1027 SvCUR(PL_parser->linestr) + len+highhalf);
f0e67a1d
Z
1028 PL_parser->bufend += len+highhalf;
1029 for (p = pv; p != e; p++) {
1030 U8 c = (U8)*p;
54d004e8
KW
1031 if (! UTF8_IS_INVARIANT(c)) {
1032 *bufptr++ = UTF8_TWO_BYTE_HI(c);
1033 *bufptr++ = UTF8_TWO_BYTE_LO(c);
f0e67a1d
Z
1034 } else {
1035 *bufptr++ = (char)c;
1036 }
1037 }
1038 }
1039 } else {
1040 if (flags & LEX_STUFF_UTF8) {
1041 STRLEN highhalf = 0;
83aa740e 1042 const char *p, *e = pv+len;
f0e67a1d
Z
1043 for (p = pv; p != e; p++) {
1044 U8 c = (U8)*p;
54d004e8 1045 if (UTF8_IS_ABOVE_LATIN1(c)) {
f0e67a1d
Z
1046 Perl_croak(aTHX_ "Lexing code attempted to stuff "
1047 "non-Latin-1 character into Latin-1 input");
54d004e8 1048 } else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(p, e)) {
f0e67a1d
Z
1049 p++;
1050 highhalf++;
54d004e8 1051 } else if (! UTF8_IS_INVARIANT(c)) {
f0e67a1d
Z
1052 /* malformed UTF-8 */
1053 ENTER;
1054 SAVESPTR(PL_warnhook);
1055 PL_warnhook = PERL_WARNHOOK_FATAL;
1056 utf8n_to_uvuni((U8*)p, e-p, NULL, 0);
1057 LEAVE;
1058 }
1059 }
1060 if (!highhalf)
1061 goto plain_copy;
1062 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
1063 bufptr = PL_parser->bufptr;
1064 Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1065 SvCUR_set(PL_parser->linestr,
1066 SvCUR(PL_parser->linestr) + len-highhalf);
f0e67a1d 1067 PL_parser->bufend += len-highhalf;
54d004e8
KW
1068 p = pv;
1069 while (p < e) {
1070 if (UTF8_IS_INVARIANT(*p)) {
1071 *bufptr++ = *p;
1072 p++;
f0e67a1d 1073 }
54d004e8
KW
1074 else {
1075 assert(p < e -1 );
1076 *bufptr++ = TWO_BYTE_UTF8_TO_UNI(*p, *(p+1));
1077 p += 2;
1078 }
f0e67a1d
Z
1079 }
1080 } else {
54d004e8 1081 plain_copy:
f0e67a1d
Z
1082 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1083 bufptr = PL_parser->bufptr;
1084 Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
255fdf19 1085 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len);
f0e67a1d
Z
1086 PL_parser->bufend += len;
1087 Copy(pv, bufptr, len, char);
1088 }
1089 }
1090}
1091
1092/*
9dcc53ea
Z
1093=for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags
1094
1095Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1096immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1097reallocating the buffer if necessary. This means that lexing code that
1098runs later will see the characters as if they had appeared in the input.
1099It is not recommended to do this as part of normal parsing, and most
1100uses of this facility run the risk of the inserted characters being
1101interpreted in an unintended manner.
1102
1103The string to be inserted is represented by octets starting at I<pv>
1104and continuing to the first nul. These octets are interpreted as either
1105UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set
1106in I<flags>. The characters are recoded for the lexer buffer, according
1107to how the buffer is currently being interpreted (L</lex_bufutf8>).
1108If it is not convenient to nul-terminate a string to be inserted, the
1109L</lex_stuff_pvn> function is more appropriate.
1110
1111=cut
1112*/
1113
1114void
1115Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags)
1116{
1117 PERL_ARGS_ASSERT_LEX_STUFF_PV;
1118 lex_stuff_pvn(pv, strlen(pv), flags);
1119}
1120
1121/*
f0e67a1d
Z
1122=for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1123
1124Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1125immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1126reallocating the buffer if necessary. This means that lexing code that
1127runs later will see the characters as if they had appeared in the input.
1128It is not recommended to do this as part of normal parsing, and most
1129uses of this facility run the risk of the inserted characters being
1130interpreted in an unintended manner.
1131
1132The string to be inserted is the string value of I<sv>. The characters
1133are recoded for the lexer buffer, according to how the buffer is currently
9dcc53ea 1134being interpreted (L</lex_bufutf8>). If a string to be inserted is
f0e67a1d
Z
1135not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1136need to construct a scalar.
1137
1138=cut
1139*/
1140
1141void
1142Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1143{
1144 char *pv;
1145 STRLEN len;
1146 PERL_ARGS_ASSERT_LEX_STUFF_SV;
1147 if (flags)
1148 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1149 pv = SvPV(sv, len);
1150 lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1151}
1152
1153/*
1154=for apidoc Amx|void|lex_unstuff|char *ptr
1155
1156Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1157I<ptr>. Text following I<ptr> will be moved, and the buffer shortened.
1158This hides the discarded text from any lexing code that runs later,
1159as if the text had never appeared.
1160
1161This is not the normal way to consume lexed text. For that, use
1162L</lex_read_to>.
1163
1164=cut
1165*/
1166
1167void
1168Perl_lex_unstuff(pTHX_ char *ptr)
1169{
1170 char *buf, *bufend;
1171 STRLEN unstuff_len;
1172 PERL_ARGS_ASSERT_LEX_UNSTUFF;
1173 buf = PL_parser->bufptr;
1174 if (ptr < buf)
1175 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1176 if (ptr == buf)
1177 return;
1178 bufend = PL_parser->bufend;
1179 if (ptr > bufend)
1180 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1181 unstuff_len = ptr - buf;
1182 Move(ptr, buf, bufend+1-ptr, char);
1183 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1184 PL_parser->bufend = bufend - unstuff_len;
1185}
1186
1187/*
1188=for apidoc Amx|void|lex_read_to|char *ptr
1189
1190Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1191to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1192performing the correct bookkeeping whenever a newline character is passed.
1193This is the normal way to consume lexed text.
1194
1195Interpretation of the buffer's octets can be abstracted out by
1196using the slightly higher-level functions L</lex_peek_unichar> and
1197L</lex_read_unichar>.
1198
1199=cut
1200*/
1201
1202void
1203Perl_lex_read_to(pTHX_ char *ptr)
1204{
1205 char *s;
1206 PERL_ARGS_ASSERT_LEX_READ_TO;
1207 s = PL_parser->bufptr;
1208 if (ptr < s || ptr > PL_parser->bufend)
1209 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1210 for (; s != ptr; s++)
1211 if (*s == '\n') {
83944c01 1212 COPLINE_INC_WITH_HERELINES;
f0e67a1d
Z
1213 PL_parser->linestart = s+1;
1214 }
1215 PL_parser->bufptr = ptr;
1216}
1217
1218/*
1219=for apidoc Amx|void|lex_discard_to|char *ptr
1220
1221Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1222up to I<ptr>. The remaining content of the buffer will be moved, and
1223all pointers into the buffer updated appropriately. I<ptr> must not
1224be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1225it is not permitted to discard text that has yet to be lexed.
1226
1227Normally it is not necessarily to do this directly, because it suffices to
1228use the implicit discarding behaviour of L</lex_next_chunk> and things
1229based on it. However, if a token stretches across multiple lines,
1f317c95 1230and the lexing code has kept multiple lines of text in the buffer for
f0e67a1d
Z
1231that purpose, then after completion of the token it would be wise to
1232explicitly discard the now-unneeded earlier lines, to avoid future
1233multi-line tokens growing the buffer without bound.
1234
1235=cut
1236*/
1237
1238void
1239Perl_lex_discard_to(pTHX_ char *ptr)
1240{
1241 char *buf;
1242 STRLEN discard_len;
1243 PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1244 buf = SvPVX(PL_parser->linestr);
1245 if (ptr < buf)
1246 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1247 if (ptr == buf)
1248 return;
1249 if (ptr > PL_parser->bufptr)
1250 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1251 discard_len = ptr - buf;
1252 if (PL_parser->oldbufptr < ptr)
1253 PL_parser->oldbufptr = ptr;
1254 if (PL_parser->oldoldbufptr < ptr)
1255 PL_parser->oldoldbufptr = ptr;
1256 if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1257 PL_parser->last_uni = NULL;
1258 if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1259 PL_parser->last_lop = NULL;
1260 Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1261 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1262 PL_parser->bufend -= discard_len;
1263 PL_parser->bufptr -= discard_len;
1264 PL_parser->oldbufptr -= discard_len;
1265 PL_parser->oldoldbufptr -= discard_len;
1266 if (PL_parser->last_uni)
1267 PL_parser->last_uni -= discard_len;
1268 if (PL_parser->last_lop)
1269 PL_parser->last_lop -= discard_len;
1270}
1271
1272/*
1273=for apidoc Amx|bool|lex_next_chunk|U32 flags
1274
1275Reads in the next chunk of text to be lexed, appending it to
1276L</PL_parser-E<gt>linestr>. This should be called when lexing code has
1277looked to the end of the current chunk and wants to know more. It is
1278usual, but not necessary, for lexing to have consumed the entirety of
1279the current chunk at this time.
1280
1281If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1282chunk (i.e., the current chunk has been entirely consumed), normally the
1283current chunk will be discarded at the same time that the new chunk is
1284read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1285will not be discarded. If the current chunk has not been entirely
1286consumed, then it will not be discarded regardless of the flag.
1287
1288Returns true if some new text was added to the buffer, or false if the
1289buffer has reached the end of the input text.
1290
1291=cut
1292*/
1293
1294#define LEX_FAKE_EOF 0x80000000
112d1284 1295#define LEX_NO_TERM 0x40000000
f0e67a1d
Z
1296
1297bool
1298Perl_lex_next_chunk(pTHX_ U32 flags)
1299{
1300 SV *linestr;
1301 char *buf;
1302 STRLEN old_bufend_pos, new_bufend_pos;
1303 STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1304 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
17cc9359 1305 bool got_some_for_debugger = 0;
f0e67a1d 1306 bool got_some;
112d1284 1307 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF|LEX_NO_TERM))
f0e67a1d 1308 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
f0e67a1d
Z
1309 linestr = PL_parser->linestr;
1310 buf = SvPVX(linestr);
1311 if (!(flags & LEX_KEEP_PREVIOUS) &&
1312 PL_parser->bufptr == PL_parser->bufend) {
1313 old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1314 linestart_pos = 0;
1315 if (PL_parser->last_uni != PL_parser->bufend)
1316 PL_parser->last_uni = NULL;
1317 if (PL_parser->last_lop != PL_parser->bufend)
1318 PL_parser->last_lop = NULL;
1319 last_uni_pos = last_lop_pos = 0;
1320 *buf = 0;
1321 SvCUR(linestr) = 0;
1322 } else {
1323 old_bufend_pos = PL_parser->bufend - buf;
1324 bufptr_pos = PL_parser->bufptr - buf;
1325 oldbufptr_pos = PL_parser->oldbufptr - buf;
1326 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1327 linestart_pos = PL_parser->linestart - buf;
1328 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1329 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1330 }
1331 if (flags & LEX_FAKE_EOF) {
1332 goto eof;
60d63348 1333 } else if (!PL_parser->rsfp && !PL_parser->filtered) {
f0e67a1d
Z
1334 got_some = 0;
1335 } else if (filter_gets(linestr, old_bufend_pos)) {
1336 got_some = 1;
17cc9359 1337 got_some_for_debugger = 1;
112d1284
FC
1338 } else if (flags & LEX_NO_TERM) {
1339 got_some = 0;
f0e67a1d 1340 } else {
580561a3
Z
1341 if (!SvPOK(linestr)) /* can get undefined by filter_gets */
1342 sv_setpvs(linestr, "");
f0e67a1d
Z
1343 eof:
1344 /* End of real input. Close filehandle (unless it was STDIN),
1345 * then add implicit termination.
1346 */
87606032 1347 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
f0e67a1d
Z
1348 PerlIO_clearerr(PL_parser->rsfp);
1349 else if (PL_parser->rsfp)
1350 (void)PerlIO_close(PL_parser->rsfp);
1351 PL_parser->rsfp = NULL;
60d63348 1352 PL_parser->in_pod = PL_parser->filtered = 0;
f0e67a1d
Z
1353#ifdef PERL_MAD
1354 if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1355 PL_faketokens = 1;
1356#endif
1357 if (!PL_in_eval && PL_minus_p) {
1358 sv_catpvs(linestr,
1359 /*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1360 PL_minus_n = PL_minus_p = 0;
1361 } else if (!PL_in_eval && PL_minus_n) {
1362 sv_catpvs(linestr, /*{*/";}");
1363 PL_minus_n = 0;
1364 } else
1365 sv_catpvs(linestr, ";");
1366 got_some = 1;
1367 }
1368 buf = SvPVX(linestr);
1369 new_bufend_pos = SvCUR(linestr);
1370 PL_parser->bufend = buf + new_bufend_pos;
1371 PL_parser->bufptr = buf + bufptr_pos;
1372 PL_parser->oldbufptr = buf + oldbufptr_pos;
1373 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1374 PL_parser->linestart = buf + linestart_pos;
1375 if (PL_parser->last_uni)
1376 PL_parser->last_uni = buf + last_uni_pos;
1377 if (PL_parser->last_lop)
1378 PL_parser->last_lop = buf + last_lop_pos;
17cc9359 1379 if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
f0e67a1d
Z
1380 PL_curstash != PL_debstash) {
1381 /* debugger active and we're not compiling the debugger code,
1382 * so store the line into the debugger's array of lines
1383 */
1384 update_debugger_info(NULL, buf+old_bufend_pos,
1385 new_bufend_pos-old_bufend_pos);
1386 }
1387 return got_some;
1388}
1389
1390/*
1391=for apidoc Amx|I32|lex_peek_unichar|U32 flags
1392
1393Looks ahead one (Unicode) character in the text currently being lexed.
1394Returns the codepoint (unsigned integer value) of the next character,
1395or -1 if lexing has reached the end of the input text. To consume the
1396peeked character, use L</lex_read_unichar>.
1397
1398If the next character is in (or extends into) the next chunk of input
1399text, the next chunk will be read in. Normally the current chunk will be
1400discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1401then the current chunk will not be discarded.
1402
1403If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1404is encountered, an exception is generated.
1405
1406=cut
1407*/
1408
1409I32
1410Perl_lex_peek_unichar(pTHX_ U32 flags)
1411{
749123ff 1412 dVAR;
f0e67a1d
Z
1413 char *s, *bufend;
1414 if (flags & ~(LEX_KEEP_PREVIOUS))
1415 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1416 s = PL_parser->bufptr;
1417 bufend = PL_parser->bufend;
1418 if (UTF) {
1419 U8 head;
1420 I32 unichar;
1421 STRLEN len, retlen;
1422 if (s == bufend) {
1423 if (!lex_next_chunk(flags))
1424 return -1;
1425 s = PL_parser->bufptr;
1426 bufend = PL_parser->bufend;
1427 }
1428 head = (U8)*s;
54d004e8 1429 if (UTF8_IS_INVARIANT(head))
f0e67a1d 1430 return head;
54d004e8
KW
1431 if (UTF8_IS_START(head)) {
1432 len = UTF8SKIP(&head);
f0e67a1d
Z
1433 while ((STRLEN)(bufend-s) < len) {
1434 if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1435 break;
1436 s = PL_parser->bufptr;
1437 bufend = PL_parser->bufend;
1438 }
1439 }
1440 unichar = utf8n_to_uvuni((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
1441 if (retlen == (STRLEN)-1) {
1442 /* malformed UTF-8 */
1443 ENTER;
1444 SAVESPTR(PL_warnhook);
1445 PL_warnhook = PERL_WARNHOOK_FATAL;
1446 utf8n_to_uvuni((U8*)s, bufend-s, NULL, 0);
1447 LEAVE;
1448 }
1449 return unichar;
1450 } else {
1451 if (s == bufend) {
1452 if (!lex_next_chunk(flags))
1453 return -1;
1454 s = PL_parser->bufptr;
1455 }
1456 return (U8)*s;
1457 }
1458}
1459
1460/*
1461=for apidoc Amx|I32|lex_read_unichar|U32 flags
1462
1463Reads the next (Unicode) character in the text currently being lexed.
1464Returns the codepoint (unsigned integer value) of the character read,
1465and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1466if lexing has reached the end of the input text. To non-destructively
1467examine the next character, use L</lex_peek_unichar> instead.
1468
1469If the next character is in (or extends into) the next chunk of input
1470text, the next chunk will be read in. Normally the current chunk will be
1471discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1472then the current chunk will not be discarded.
1473
1474If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1475is encountered, an exception is generated.
1476
1477=cut
1478*/
1479
1480I32
1481Perl_lex_read_unichar(pTHX_ U32 flags)
1482{
1483 I32 c;
1484 if (flags & ~(LEX_KEEP_PREVIOUS))
1485 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1486 c = lex_peek_unichar(flags);
1487 if (c != -1) {
1488 if (c == '\n')
83944c01 1489 COPLINE_INC_WITH_HERELINES;
d9018cbe
EB
1490 if (UTF)
1491 PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1492 else
1493 ++(PL_parser->bufptr);
f0e67a1d
Z
1494 }
1495 return c;
1496}
1497
1498/*
1499=for apidoc Amx|void|lex_read_space|U32 flags
1500
1501Reads optional spaces, in Perl style, in the text currently being
1502lexed. The spaces may include ordinary whitespace characters and
1503Perl-style comments. C<#line> directives are processed if encountered.
1504L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1505at a non-space character (or the end of the input text).
1506
1507If spaces extend into the next chunk of input text, the next chunk will
1508be read in. Normally the current chunk will be discarded at the same
1509time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1510chunk will not be discarded.
1511
1512=cut
1513*/
1514
f0998909
Z
1515#define LEX_NO_NEXT_CHUNK 0x80000000
1516
f0e67a1d
Z
1517void
1518Perl_lex_read_space(pTHX_ U32 flags)
1519{
1520 char *s, *bufend;
1521 bool need_incline = 0;
f0998909 1522 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK))
f0e67a1d
Z
1523 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1524#ifdef PERL_MAD
1525 if (PL_skipwhite) {
1526 sv_free(PL_skipwhite);
1527 PL_skipwhite = NULL;
1528 }
1529 if (PL_madskills)
1530 PL_skipwhite = newSVpvs("");
1531#endif /* PERL_MAD */
1532 s = PL_parser->bufptr;
1533 bufend = PL_parser->bufend;
1534 while (1) {
1535 char c = *s;
1536 if (c == '#') {
1537 do {
1538 c = *++s;
1539 } while (!(c == '\n' || (c == 0 && s == bufend)));
1540 } else if (c == '\n') {
1541 s++;
1542 PL_parser->linestart = s;
1543 if (s == bufend)
1544 need_incline = 1;
1545 else
1546 incline(s);
1547 } else if (isSPACE(c)) {
1548 s++;
1549 } else if (c == 0 && s == bufend) {
1550 bool got_more;
1551#ifdef PERL_MAD
1552 if (PL_madskills)
1553 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1554#endif /* PERL_MAD */
f0998909
Z
1555 if (flags & LEX_NO_NEXT_CHUNK)
1556 break;
f0e67a1d 1557 PL_parser->bufptr = s;
83944c01 1558 COPLINE_INC_WITH_HERELINES;
f0e67a1d
Z
1559 got_more = lex_next_chunk(flags);
1560 CopLINE_dec(PL_curcop);
1561 s = PL_parser->bufptr;
1562 bufend = PL_parser->bufend;
1563 if (!got_more)
1564 break;
1565 if (need_incline && PL_parser->rsfp) {
1566 incline(s);
1567 need_incline = 0;
1568 }
1569 } else {
1570 break;
1571 }
1572 }
1573#ifdef PERL_MAD
1574 if (PL_madskills)
1575 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1576#endif /* PERL_MAD */
1577 PL_parser->bufptr = s;
1578}
1579
1580/*
ffb4593c
NT
1581 * S_incline
1582 * This subroutine has nothing to do with tilting, whether at windmills
1583 * or pinball tables. Its name is short for "increment line". It
57843af0 1584 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 1585 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
1586 * # line 500 "foo.pm"
1587 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
1588 */
1589
76e3520e 1590STATIC void
d9095cec 1591S_incline(pTHX_ const char *s)
463ee0b2 1592{
97aff369 1593 dVAR;
d9095cec
NC
1594 const char *t;
1595 const char *n;
1596 const char *e;
8818d409 1597 line_t line_num;
463ee0b2 1598
7918f24d
NC
1599 PERL_ARGS_ASSERT_INCLINE;
1600
83944c01 1601 COPLINE_INC_WITH_HERELINES;
451f421f
FC
1602 if (!PL_rsfp && !PL_parser->filtered && PL_lex_state == LEX_NORMAL
1603 && s+1 == PL_bufend && *s == ';') {
1604 /* fake newline in string eval */
1605 CopLINE_dec(PL_curcop);
1606 return;
1607 }
463ee0b2
LW
1608 if (*s++ != '#')
1609 return;
d4c19fe8
AL
1610 while (SPACE_OR_TAB(*s))
1611 s++;
73659bf1
GS
1612 if (strnEQ(s, "line", 4))
1613 s += 4;
1614 else
1615 return;
084592ab 1616 if (SPACE_OR_TAB(*s))
73659bf1 1617 s++;
4e553d73 1618 else
73659bf1 1619 return;
d4c19fe8
AL
1620 while (SPACE_OR_TAB(*s))
1621 s++;
463ee0b2
LW
1622 if (!isDIGIT(*s))
1623 return;
d4c19fe8 1624
463ee0b2
LW
1625 n = s;
1626 while (isDIGIT(*s))
1627 s++;
07714eb4 1628 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
26b6dc3f 1629 return;
bf4acbe4 1630 while (SPACE_OR_TAB(*s))
463ee0b2 1631 s++;
73659bf1 1632 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 1633 s++;
73659bf1
GS
1634 e = t + 1;
1635 }
463ee0b2 1636 else {
c35e046a
AL
1637 t = s;
1638 while (!isSPACE(*t))
1639 t++;
73659bf1 1640 e = t;
463ee0b2 1641 }
bf4acbe4 1642 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
1643 e++;
1644 if (*e != '\n' && *e != '\0')
1645 return; /* false alarm */
1646
8818d409
FC
1647 line_num = atoi(n)-1;
1648
f4dd75d9 1649 if (t - s > 0) {
d9095cec 1650 const STRLEN len = t - s;
19bad673
NC
1651 SV *const temp_sv = CopFILESV(PL_curcop);
1652 const char *cf;
1653 STRLEN tmplen;
1654
1655 if (temp_sv) {
1656 cf = SvPVX(temp_sv);
1657 tmplen = SvCUR(temp_sv);
1658 } else {
1659 cf = NULL;
1660 tmplen = 0;
1661 }
1662
d1299d44 1663 if (!PL_rsfp && !PL_parser->filtered) {
e66cf94c
RGS
1664 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1665 * to *{"::_<newfilename"} */
44867030
NC
1666 /* However, the long form of evals is only turned on by the
1667 debugger - usually they're "(eval %lu)" */
1668 char smallbuf[128];
1669 char *tmpbuf;
1670 GV **gvp;
d9095cec 1671 STRLEN tmplen2 = len;
798b63bc 1672 if (tmplen + 2 <= sizeof smallbuf)
e66cf94c
RGS
1673 tmpbuf = smallbuf;
1674 else
2ae0db35 1675 Newx(tmpbuf, tmplen + 2, char);
44867030
NC
1676 tmpbuf[0] = '_';
1677 tmpbuf[1] = '<';
2ae0db35 1678 memcpy(tmpbuf + 2, cf, tmplen);
44867030 1679 tmplen += 2;
8a5ee598
RGS
1680 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
1681 if (gvp) {
44867030
NC
1682 char *tmpbuf2;
1683 GV *gv2;
1684
1685 if (tmplen2 + 2 <= sizeof smallbuf)
1686 tmpbuf2 = smallbuf;
1687 else
1688 Newx(tmpbuf2, tmplen2 + 2, char);
1689
1690 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
1691 /* Either they malloc'd it, or we malloc'd it,
1692 so no prefix is present in ours. */
1693 tmpbuf2[0] = '_';
1694 tmpbuf2[1] = '<';
1695 }
1696
1697 memcpy(tmpbuf2 + 2, s, tmplen2);
1698 tmplen2 += 2;
1699
8a5ee598 1700 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
e5527e4b 1701 if (!isGV(gv2)) {
8a5ee598 1702 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
e5527e4b
RGS
1703 /* adjust ${"::_<newfilename"} to store the new file name */
1704 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
8818d409
FC
1705 /* The line number may differ. If that is the case,
1706 alias the saved lines that are in the array.
1707 Otherwise alias the whole array. */
1708 if (CopLINE(PL_curcop) == line_num) {
1709 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
1710 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
1711 }
1712 else if (GvAV(*gvp)) {
1713 AV * const av = GvAV(*gvp);
1714 const I32 start = CopLINE(PL_curcop)+1;
1715 I32 items = AvFILLp(av) - start;
1716 if (items > 0) {
1717 AV * const av2 = GvAVn(gv2);
1718 SV **svp = AvARRAY(av) + start;
1719 I32 l = (I32)line_num+1;
1720 while (items--)
1721 av_store(av2, l++, SvREFCNT_inc(*svp++));
1722 }
1723 }
e5527e4b 1724 }
44867030
NC
1725
1726 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
8a5ee598 1727 }
e66cf94c 1728 if (tmpbuf != smallbuf) Safefree(tmpbuf);
e66cf94c 1729 }
05ec9bb3 1730 CopFILE_free(PL_curcop);
d9095cec 1731 CopFILE_setn(PL_curcop, s, len);
f4dd75d9 1732 }
8818d409 1733 CopLINE_set(PL_curcop, line_num);
463ee0b2
LW
1734}
1735
29595ff2 1736#ifdef PERL_MAD
cd81e915 1737/* skip space before PL_thistoken */
29595ff2
NC
1738
1739STATIC char *
5aaab254 1740S_skipspace0(pTHX_ char *s)
29595ff2 1741{
7918f24d
NC
1742 PERL_ARGS_ASSERT_SKIPSPACE0;
1743
29595ff2
NC
1744 s = skipspace(s);
1745 if (!PL_madskills)
1746 return s;
cd81e915
NC
1747 if (PL_skipwhite) {
1748 if (!PL_thiswhite)
6b29d1f5 1749 PL_thiswhite = newSVpvs("");
cd81e915
NC
1750 sv_catsv(PL_thiswhite, PL_skipwhite);
1751 sv_free(PL_skipwhite);
1752 PL_skipwhite = 0;
1753 }
1754 PL_realtokenstart = s - SvPVX(PL_linestr);
29595ff2
NC
1755 return s;
1756}
1757
cd81e915 1758/* skip space after PL_thistoken */
29595ff2
NC
1759
1760STATIC char *
5aaab254 1761S_skipspace1(pTHX_ char *s)
29595ff2 1762{
d4c19fe8 1763 const char *start = s;
29595ff2
NC
1764 I32 startoff = start - SvPVX(PL_linestr);
1765
7918f24d
NC
1766 PERL_ARGS_ASSERT_SKIPSPACE1;
1767
29595ff2
NC
1768 s = skipspace(s);
1769 if (!PL_madskills)
1770 return s;
1771 start = SvPVX(PL_linestr) + startoff;
cd81e915 1772 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1773 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1774 PL_thistoken = newSVpvn(tstart, start - tstart);
1775 }
1776 PL_realtokenstart = -1;
1777 if (PL_skipwhite) {
1778 if (!PL_nextwhite)
6b29d1f5 1779 PL_nextwhite = newSVpvs("");
cd81e915
NC
1780 sv_catsv(PL_nextwhite, PL_skipwhite);
1781 sv_free(PL_skipwhite);
1782 PL_skipwhite = 0;
29595ff2
NC
1783 }
1784 return s;
1785}
1786
1787STATIC char *
5aaab254 1788S_skipspace2(pTHX_ char *s, SV **svp)
29595ff2 1789{
c35e046a
AL
1790 char *start;
1791 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
1792 const I32 startoff = s - SvPVX(PL_linestr);
1793
7918f24d
NC
1794 PERL_ARGS_ASSERT_SKIPSPACE2;
1795
29595ff2
NC
1796 s = skipspace(s);
1797 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
1798 if (!PL_madskills || !svp)
1799 return s;
1800 start = SvPVX(PL_linestr) + startoff;
cd81e915 1801 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1802 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1803 PL_thistoken = newSVpvn(tstart, start - tstart);
1804 PL_realtokenstart = -1;
29595ff2 1805 }
cd81e915 1806 if (PL_skipwhite) {
29595ff2 1807 if (!*svp)
6b29d1f5 1808 *svp = newSVpvs("");
cd81e915
NC
1809 sv_setsv(*svp, PL_skipwhite);
1810 sv_free(PL_skipwhite);
1811 PL_skipwhite = 0;
29595ff2
NC
1812 }
1813
1814 return s;
1815}
1816#endif
1817
80a702cd 1818STATIC void
15f169a1 1819S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
80a702cd
RGS
1820{
1821 AV *av = CopFILEAVx(PL_curcop);
1822 if (av) {
b9f83d2f 1823 SV * const sv = newSV_type(SVt_PVMG);
5fa550fb
NC
1824 if (orig_sv)
1825 sv_setsv(sv, orig_sv);
1826 else
1827 sv_setpvn(sv, buf, len);
80a702cd
RGS
1828 (void)SvIOK_on(sv);
1829 SvIV_set(sv, 0);
1830 av_store(av, (I32)CopLINE(PL_curcop), sv);
1831 }
1832}
1833
ffb4593c
NT
1834/*
1835 * S_skipspace
1836 * Called to gobble the appropriate amount and type of whitespace.
1837 * Skips comments as well.
1838 */
1839
76e3520e 1840STATIC char *
5aaab254 1841S_skipspace(pTHX_ char *s)
a687059c 1842{
5db06880 1843#ifdef PERL_MAD
f0e67a1d
Z
1844 char *start = s;
1845#endif /* PERL_MAD */
7918f24d 1846 PERL_ARGS_ASSERT_SKIPSPACE;
f0e67a1d 1847#ifdef PERL_MAD
cd81e915
NC
1848 if (PL_skipwhite) {
1849 sv_free(PL_skipwhite);
f0e67a1d 1850 PL_skipwhite = NULL;
5db06880 1851 }
f0e67a1d 1852#endif /* PERL_MAD */
3280af22 1853 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 1854 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2 1855 s++;
f0e67a1d
Z
1856 } else {
1857 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1858 PL_bufptr = s;
f0998909
Z
1859 lex_read_space(LEX_KEEP_PREVIOUS |
1860 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1861 LEX_NO_NEXT_CHUNK : 0));
3280af22 1862 s = PL_bufptr;
f0e67a1d
Z
1863 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1864 if (PL_linestart > PL_bufptr)
1865 PL_bufptr = PL_linestart;
1866 return s;
463ee0b2 1867 }
5db06880 1868#ifdef PERL_MAD
f0e67a1d
Z
1869 if (PL_madskills)
1870 PL_skipwhite = newSVpvn(start, s-start);
1871#endif /* PERL_MAD */
5db06880 1872 return s;
a687059c 1873}
378cc40b 1874
ffb4593c
NT
1875/*
1876 * S_check_uni
1877 * Check the unary operators to ensure there's no ambiguity in how they're
1878 * used. An ambiguous piece of code would be:
1879 * rand + 5
1880 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1881 * the +5 is its argument.
1882 */
1883
76e3520e 1884STATIC void
cea2e8a9 1885S_check_uni(pTHX)
ba106d47 1886{
97aff369 1887 dVAR;
d4c19fe8
AL
1888 const char *s;
1889 const char *t;
2f3197b3 1890
3280af22 1891 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 1892 return;
3280af22
NIS
1893 while (isSPACE(*PL_last_uni))
1894 PL_last_uni++;
c35e046a 1895 s = PL_last_uni;
8a2bca12 1896 while (isWORDCHAR_lazy_if(s,UTF) || *s == '-')
c35e046a 1897 s++;
3280af22 1898 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 1899 return;
6136c704 1900
9b387841
NC
1901 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1902 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1903 (int)(s - PL_last_uni), PL_last_uni);
2f3197b3
LW
1904}
1905
ffb4593c
NT
1906/*
1907 * LOP : macro to build a list operator. Its behaviour has been replaced
1908 * with a subroutine, S_lop() for which LOP is just another name.
1909 */
1910
a0d0e21e
LW
1911#define LOP(f,x) return lop(f,x,s)
1912
ffb4593c
NT
1913/*
1914 * S_lop
1915 * Build a list operator (or something that might be one). The rules:
1916 * - if we have a next token, then it's a list operator [why?]
1917 * - if the next thing is an opening paren, then it's a function
1918 * - else it's a list operator
1919 */
1920
76e3520e 1921STATIC I32
a0be28da 1922S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 1923{
97aff369 1924 dVAR;
7918f24d
NC
1925
1926 PERL_ARGS_ASSERT_LOP;
1927
6154021b 1928 pl_yylval.ival = f;
35c8bce7 1929 CLINE;
3280af22
NIS
1930 PL_expect = x;
1931 PL_bufptr = s;
1932 PL_last_lop = PL_oldbufptr;
eb160463 1933 PL_last_lop_op = (OPCODE)f;
5db06880
NC
1934#ifdef PERL_MAD
1935 if (PL_lasttoke)
78cdf107 1936 goto lstop;
5db06880 1937#else
3280af22 1938 if (PL_nexttoke)
78cdf107 1939 goto lstop;
5db06880 1940#endif
79072805 1941 if (*s == '(')
bbf60fe6 1942 return REPORT(FUNC);
29595ff2 1943 s = PEEKSPACE(s);
79072805 1944 if (*s == '(')
bbf60fe6 1945 return REPORT(FUNC);
78cdf107
Z
1946 else {
1947 lstop:
1948 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
1949 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
bbf60fe6 1950 return REPORT(LSTOP);
78cdf107 1951 }
79072805
LW
1952}
1953
5db06880
NC
1954#ifdef PERL_MAD
1955 /*
1956 * S_start_force
1957 * Sets up for an eventual force_next(). start_force(0) basically does
1958 * an unshift, while start_force(-1) does a push. yylex removes items
1959 * on the "pop" end.
1960 */
1961
1962STATIC void
1963S_start_force(pTHX_ int where)
1964{
1965 int i;
1966
cd81e915 1967 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
5db06880 1968 where = PL_lasttoke;
cd81e915
NC
1969 assert(PL_curforce < 0 || PL_curforce == where);
1970 if (PL_curforce != where) {
5db06880
NC
1971 for (i = PL_lasttoke; i > where; --i) {
1972 PL_nexttoke[i] = PL_nexttoke[i-1];
1973 }
1974 PL_lasttoke++;
1975 }
cd81e915 1976 if (PL_curforce < 0) /* in case of duplicate start_force() */
5db06880 1977 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
cd81e915
NC
1978 PL_curforce = where;
1979 if (PL_nextwhite) {
5db06880 1980 if (PL_madskills)
6b29d1f5 1981 curmad('^', newSVpvs(""));
cd81e915 1982 CURMAD('_', PL_nextwhite);
5db06880
NC
1983 }
1984}
1985
1986STATIC void
1987S_curmad(pTHX_ char slot, SV *sv)
1988{
1989 MADPROP **where;
1990
1991 if (!sv)
1992 return;
cd81e915
NC
1993 if (PL_curforce < 0)
1994 where = &PL_thismad;
5db06880 1995 else
cd81e915 1996 where = &PL_nexttoke[PL_curforce].next_mad;
5db06880 1997
cd81e915 1998 if (PL_faketokens)
76f68e9b 1999 sv_setpvs(sv, "");
5db06880
NC
2000 else {
2001 if (!IN_BYTES) {
2002 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2003 SvUTF8_on(sv);
2004 else if (PL_encoding) {
2005 sv_recode_to_utf8(sv, PL_encoding);
2006 }
2007 }
2008 }
2009
2010 /* keep a slot open for the head of the list? */
2011 if (slot != '_' && *where && (*where)->mad_key == '^') {
2012 (*where)->mad_key = slot;
daba3364 2013 sv_free(MUTABLE_SV(((*where)->mad_val)));
5db06880
NC
2014 (*where)->mad_val = (void*)sv;
2015 }
2016 else
2017 addmad(newMADsv(slot, sv), where, 0);
2018}
2019#else
b3f24c00
MHM
2020# define start_force(where) NOOP
2021# define curmad(slot, sv) NOOP
5db06880
NC
2022#endif
2023
ffb4593c
NT
2024/*
2025 * S_force_next
9cbb5ea2 2026 * When the lexer realizes it knows the next token (for instance,
ffb4593c 2027 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2 2028 * to know what token to return the next time the lexer is called. Caller
5db06880
NC
2029 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
2030 * and possibly PL_expect to ensure the lexer handles the token correctly.
ffb4593c
NT
2031 */
2032
4e553d73 2033STATIC void
cea2e8a9 2034S_force_next(pTHX_ I32 type)
79072805 2035{
97aff369 2036 dVAR;
704d4215
GG
2037#ifdef DEBUGGING
2038 if (DEBUG_T_TEST) {
2039 PerlIO_printf(Perl_debug_log, "### forced token:\n");
f05d7009 2040 tokereport(type, &NEXTVAL_NEXTTOKE);
704d4215
GG
2041 }
2042#endif
5db06880 2043#ifdef PERL_MAD
cd81e915 2044 if (PL_curforce < 0)
5db06880 2045 start_force(PL_lasttoke);
cd81e915 2046 PL_nexttoke[PL_curforce].next_type = type;
5db06880
NC
2047 if (PL_lex_state != LEX_KNOWNEXT)
2048 PL_lex_defer = PL_lex_state;
2049 PL_lex_state = LEX_KNOWNEXT;
2050 PL_lex_expect = PL_expect;
cd81e915 2051 PL_curforce = -1;
5db06880 2052#else
3280af22
NIS
2053 PL_nexttype[PL_nexttoke] = type;
2054 PL_nexttoke++;
2055 if (PL_lex_state != LEX_KNOWNEXT) {
2056 PL_lex_defer = PL_lex_state;
2057 PL_lex_expect = PL_expect;
2058 PL_lex_state = LEX_KNOWNEXT;
79072805 2059 }
5db06880 2060#endif
79072805
LW
2061}
2062
28ac2b49
Z
2063void
2064Perl_yyunlex(pTHX)
2065{
a7aaec61
Z
2066 int yyc = PL_parser->yychar;
2067 if (yyc != YYEMPTY) {
2068 if (yyc) {
2069 start_force(-1);
2070 NEXTVAL_NEXTTOKE = PL_parser->yylval;
2071 if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
78cdf107 2072 PL_lex_allbrackets--;
a7aaec61 2073 PL_lex_brackets--;
78cdf107
Z
2074 yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
2075 } else if (yyc == '('/*)*/) {
2076 PL_lex_allbrackets--;
2077 yyc |= (2<<24);
a7aaec61
Z
2078 }
2079 force_next(yyc);
2080 }
28ac2b49
Z
2081 PL_parser->yychar = YYEMPTY;
2082 }
2083}
2084
d0a148a6 2085STATIC SV *
15f169a1 2086S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
d0a148a6 2087{
97aff369 2088 dVAR;
740cce10 2089 SV * const sv = newSVpvn_utf8(start, len,
eaf7a4d2
CS
2090 !IN_BYTES
2091 && UTF
2092 && !is_ascii_string((const U8*)start, len)
740cce10 2093 && is_utf8_string((const U8*)start, len));
d0a148a6
NC
2094 return sv;
2095}
2096
ffb4593c
NT
2097/*
2098 * S_force_word
2099 * When the lexer knows the next thing is a word (for instance, it has
2100 * just seen -> and it knows that the next char is a word char, then
02b34bbe
DM
2101 * it calls S_force_word to stick the next word into the PL_nexttoke/val
2102 * lookahead.
ffb4593c
NT
2103 *
2104 * Arguments:
b1b65b59 2105 * char *start : buffer position (must be within PL_linestr)
02b34bbe 2106 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
ffb4593c
NT
2107 * int check_keyword : if true, Perl checks to make sure the word isn't
2108 * a keyword (do this if the word is a label, e.g. goto FOO)
2109 * int allow_pack : if true, : characters will also be allowed (require,
2110 * use, etc. do this)
9cbb5ea2 2111 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
2112 */
2113
76e3520e 2114STATIC char *
345b3785 2115S_force_word(pTHX_ char *start, int token, int check_keyword, int allow_pack)
79072805 2116{
97aff369 2117 dVAR;
eb578fdb 2118 char *s;
463ee0b2 2119 STRLEN len;
4e553d73 2120
7918f24d
NC
2121 PERL_ARGS_ASSERT_FORCE_WORD;
2122
29595ff2 2123 start = SKIPSPACE1(start);
463ee0b2 2124 s = start;
7e2040f0 2125 if (isIDFIRST_lazy_if(s,UTF) ||
345b3785 2126 (allow_pack && *s == ':') )
a0d0e21e 2127 {
3280af22 2128 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
01b5ef50
FC
2129 if (check_keyword) {
2130 char *s2 = PL_tokenbuf;
2131 if (allow_pack && len > 6 && strnEQ(s2, "CORE::", 6))
2132 s2 += 6, len -= 6;
2133 if (keyword(s2, len, 0))
463ee0b2 2134 return start;
01b5ef50 2135 }
cd81e915 2136 start_force(PL_curforce);
5db06880
NC
2137 if (PL_madskills)
2138 curmad('X', newSVpvn(start,s-start));
463ee0b2 2139 if (token == METHOD) {
29595ff2 2140 s = SKIPSPACE1(s);
463ee0b2 2141 if (*s == '(')
3280af22 2142 PL_expect = XTERM;
463ee0b2 2143 else {
3280af22 2144 PL_expect = XOPERATOR;
463ee0b2 2145 }
79072805 2146 }
e74e6b3d 2147 if (PL_madskills)
63575281 2148 curmad('g', newSVpvs( "forced" ));
9ded7720 2149 NEXTVAL_NEXTTOKE.opval
d0a148a6
NC
2150 = (OP*)newSVOP(OP_CONST,0,
2151 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
9ded7720 2152 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
79072805
LW
2153 force_next(token);
2154 }
2155 return s;
2156}
2157
ffb4593c
NT
2158/*
2159 * S_force_ident
9cbb5ea2 2160 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
2161 * text only contains the "foo" portion. The first argument is a pointer
2162 * to the "foo", and the second argument is the type symbol to prefix.
2163 * Forces the next token to be a "WORD".
9cbb5ea2 2164 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
2165 */
2166
76e3520e 2167STATIC void
5aaab254 2168S_force_ident(pTHX_ const char *s, int kind)
79072805 2169{
97aff369 2170 dVAR;
7918f24d
NC
2171
2172 PERL_ARGS_ASSERT_FORCE_IDENT;
2173
c9b48522
DD
2174 if (s[0]) {
2175 const STRLEN len = s[1] ? strlen(s) : 1; /* s = "\"" see yylex */
728847b1
BF
2176 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len,
2177 UTF ? SVf_UTF8 : 0));
cd81e915 2178 start_force(PL_curforce);
9ded7720 2179 NEXTVAL_NEXTTOKE.opval = o;
79072805 2180 force_next(WORD);
748a9306 2181 if (kind) {
11343788 2182 o->op_private = OPpCONST_ENTERED;
55497cff
PP
2183 /* XXX see note in pp_entereval() for why we forgo typo
2184 warnings if the symbol must be introduced in an eval.
2185 GSAR 96-10-12 */
90e5519e 2186 gv_fetchpvn_flags(s, len,
728847b1
BF
2187 (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2188 : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ),
90e5519e
NC
2189 kind == '$' ? SVt_PV :
2190 kind == '@' ? SVt_PVAV :
2191 kind == '%' ? SVt_PVHV :
a0d0e21e 2192 SVt_PVGV
90e5519e 2193 );
748a9306 2194 }
79072805
LW
2195 }
2196}
2197
3f33d153
FC
2198static void
2199S_force_ident_maybe_lex(pTHX_ char pit)
2200{
2201 start_force(PL_curforce);
2202 NEXTVAL_NEXTTOKE.ival = pit;
2203 force_next('p');
2204}
2205
1571675a
GS
2206NV
2207Perl_str_to_version(pTHX_ SV *sv)
2208{
2209 NV retval = 0.0;
2210 NV nshift = 1.0;
2211 STRLEN len;
cfd0369c 2212 const char *start = SvPV_const(sv,len);
9d4ba2ae 2213 const char * const end = start + len;
504618e9 2214 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
7918f24d
NC
2215
2216 PERL_ARGS_ASSERT_STR_TO_VERSION;
2217
1571675a 2218 while (start < end) {
ba210ebe 2219 STRLEN skip;
1571675a
GS
2220 UV n;
2221 if (utf)
9041c2e3 2222 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
2223 else {
2224 n = *(U8*)start;
2225 skip = 1;
2226 }
2227 retval += ((NV)n)/nshift;
2228 start += skip;
2229 nshift *= 1000;
2230 }
2231 return retval;
2232}
2233
4e553d73 2234/*
ffb4593c
NT
2235 * S_force_version
2236 * Forces the next token to be a version number.
e759cc13
RGS
2237 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2238 * and if "guessing" is TRUE, then no new token is created (and the caller
2239 * must use an alternative parsing method).
ffb4593c
NT
2240 */
2241
76e3520e 2242STATIC char *
e759cc13 2243S_force_version(pTHX_ char *s, int guessing)
89bfa8cd 2244{
97aff369 2245 dVAR;
5f66b61c 2246 OP *version = NULL;
44dcb63b 2247 char *d;
5db06880
NC
2248#ifdef PERL_MAD
2249 I32 startoff = s - SvPVX(PL_linestr);
2250#endif
89bfa8cd 2251
7918f24d
NC
2252 PERL_ARGS_ASSERT_FORCE_VERSION;
2253
29595ff2 2254 s = SKIPSPACE1(s);
89bfa8cd 2255
44dcb63b 2256 d = s;
dd629d5b 2257 if (*d == 'v')
44dcb63b 2258 d++;
44dcb63b 2259 if (isDIGIT(*d)) {
e759cc13
RGS
2260 while (isDIGIT(*d) || *d == '_' || *d == '.')
2261 d++;
5db06880
NC
2262#ifdef PERL_MAD
2263 if (PL_madskills) {
cd81e915 2264 start_force(PL_curforce);
5db06880
NC
2265 curmad('X', newSVpvn(s,d-s));
2266 }
2267#endif
4e4da3ac 2268 if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
dd629d5b 2269 SV *ver;
8d08d9ba 2270#ifdef USE_LOCALE_NUMERIC
909d3787
KW
2271 char *loc = savepv(setlocale(LC_NUMERIC, NULL));
2272 setlocale(LC_NUMERIC, "C");
8d08d9ba 2273#endif
6154021b 2274 s = scan_num(s, &pl_yylval);
8d08d9ba
DG
2275#ifdef USE_LOCALE_NUMERIC
2276 setlocale(LC_NUMERIC, loc);
909d3787 2277 Safefree(loc);
8d08d9ba 2278#endif
6154021b 2279 version = pl_yylval.opval;
dd629d5b
GS
2280 ver = cSVOPx(version)->op_sv;
2281 if (SvPOK(ver) && !SvNIOK(ver)) {
862a34c6 2282 SvUPGRADE(ver, SVt_PVNV);
9d6ce603 2283 SvNV_set(ver, str_to_version(ver));
1571675a 2284 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 2285 }
89bfa8cd 2286 }
5db06880
NC
2287 else if (guessing) {
2288#ifdef PERL_MAD
2289 if (PL_madskills) {
cd81e915
NC
2290 sv_free(PL_nextwhite); /* let next token collect whitespace */
2291 PL_nextwhite = 0;
5db06880
NC
2292 s = SvPVX(PL_linestr) + startoff;
2293 }
2294#endif
e759cc13 2295 return s;
5db06880 2296 }
89bfa8cd
PP
2297 }
2298
5db06880
NC
2299#ifdef PERL_MAD
2300 if (PL_madskills && !version) {
cd81e915
NC
2301 sv_free(PL_nextwhite); /* let next token collect whitespace */
2302 PL_nextwhite = 0;
5db06880
NC
2303 s = SvPVX(PL_linestr) + startoff;
2304 }
2305#endif
89bfa8cd 2306 /* NOTE: The parser sees the package name and the VERSION swapped */
cd81e915 2307 start_force(PL_curforce);
9ded7720 2308 NEXTVAL_NEXTTOKE.opval = version;
4e553d73 2309 force_next(WORD);
89bfa8cd 2310
e759cc13 2311 return s;
89bfa8cd
PP
2312}
2313
ffb4593c 2314/*
91152fc1
DG
2315 * S_force_strict_version
2316 * Forces the next token to be a version number using strict syntax rules.
2317 */
2318
2319STATIC char *
2320S_force_strict_version(pTHX_ char *s)
2321{
2322 dVAR;
2323 OP *version = NULL;
2324#ifdef PERL_MAD
2325 I32 startoff = s - SvPVX(PL_linestr);
2326#endif
2327 const char *errstr = NULL;
2328
2329 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2330
2331 while (isSPACE(*s)) /* leading whitespace */
2332 s++;
2333
2334 if (is_STRICT_VERSION(s,&errstr)) {
2335 SV *ver = newSV(0);
2336 s = (char *)scan_version(s, ver, 0);
2337 version = newSVOP(OP_CONST, 0, ver);
2338 }
4e4da3ac
Z
2339 else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2340 (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2341 {
91152fc1
DG
2342 PL_bufptr = s;
2343 if (errstr)
2344 yyerror(errstr); /* version required */
2345 return s;
2346 }
2347
2348#ifdef PERL_MAD
2349 if (PL_madskills && !version) {
2350 sv_free(PL_nextwhite); /* let next token collect whitespace */
2351 PL_nextwhite = 0;
2352 s = SvPVX(PL_linestr) + startoff;
2353 }
2354#endif
2355 /* NOTE: The parser sees the package name and the VERSION swapped */
2356 start_force(PL_curforce);
2357 NEXTVAL_NEXTTOKE.opval = version;
2358 force_next(WORD);
2359
2360 return s;
2361}
2362
2363/*
ffb4593c
NT
2364 * S_tokeq
2365 * Tokenize a quoted string passed in as an SV. It finds the next
2366 * chunk, up to end of string or a backslash. It may make a new
2367 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2368 * turns \\ into \.
2369 */
2370
76e3520e 2371STATIC SV *
cea2e8a9 2372S_tokeq(pTHX_ SV *sv)
79072805 2373{
97aff369 2374 dVAR;
eb578fdb
KW
2375 char *s;
2376 char *send;
2377 char *d;
b3ac6de7
IZ
2378 STRLEN len = 0;
2379 SV *pv = sv;
79072805 2380
7918f24d
NC
2381 PERL_ARGS_ASSERT_TOKEQ;
2382
79072805 2383 if (!SvLEN(sv))
b3ac6de7 2384 goto finish;
79072805 2385
a0d0e21e 2386 s = SvPV_force(sv, len);
21a311ee 2387 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 2388 goto finish;
463ee0b2 2389 send = s + len;
dcb21ed6
NC
2390 /* This is relying on the SV being "well formed" with a trailing '\0' */
2391 while (s < send && !(*s == '\\' && s[1] == '\\'))
79072805
LW
2392 s++;
2393 if (s == send)
b3ac6de7 2394 goto finish;
79072805 2395 d = s;
be4731d2 2396 if ( PL_hints & HINT_NEW_STRING ) {
59cd0e26 2397 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
be4731d2 2398 }
79072805
LW
2399 while (s < send) {
2400 if (*s == '\\') {
a0d0e21e 2401 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
2402 s++; /* all that, just for this */
2403 }
2404 *d++ = *s++;
2405 }
2406 *d = '\0';
95a20fc0 2407 SvCUR_set(sv, d - SvPVX_const(sv));
b3ac6de7 2408 finish:
3280af22 2409 if ( PL_hints & HINT_NEW_STRING )
eb0d8d16 2410 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
79072805
LW
2411 return sv;
2412}
2413
ffb4593c
NT
2414/*
2415 * Now come three functions related to double-quote context,
2416 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2417 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2418 * interact with PL_lex_state, and create fake ( ... ) argument lists
2419 * to handle functions and concatenation.
ecd24171
DM
2420 * For example,
2421 * "foo\lbar"
2422 * is tokenised as
2423 * stringify ( const[foo] concat lcfirst ( const[bar] ) )
ffb4593c
NT
2424 */
2425
2426/*
2427 * S_sublex_start
6154021b 2428 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
ffb4593c
NT
2429 *
2430 * Pattern matching will set PL_lex_op to the pattern-matching op to
6154021b 2431 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
ffb4593c
NT
2432 *
2433 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2434 *
2435 * Everything else becomes a FUNC.
2436 *
2437 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2438 * had an OP_CONST or OP_READLINE). This just sets us up for a
2439 * call to S_sublex_push().
2440 */
2441
76e3520e 2442STATIC I32
cea2e8a9 2443S_sublex_start(pTHX)
79072805 2444{
97aff369 2445 dVAR;
eb578fdb 2446 const I32 op_type = pl_yylval.ival;
79072805
LW
2447
2448 if (op_type == OP_NULL) {
6154021b 2449 pl_yylval.opval = PL_lex_op;
5f66b61c 2450 PL_lex_op = NULL;
79072805
LW
2451 return THING;
2452 }
2453 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 2454 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
2455
2456 if (SvTYPE(sv) == SVt_PVIV) {
2457 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2458 STRLEN len;
96a5add6 2459 const char * const p = SvPV_const(sv, len);
740cce10 2460 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
b3ac6de7
IZ
2461 SvREFCNT_dec(sv);
2462 sv = nsv;
4e553d73 2463 }
6154021b 2464 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
a0714e2c 2465 PL_lex_stuff = NULL;
6f33ba73
RGS
2466 /* Allow <FH> // "foo" */
2467 if (op_type == OP_READLINE)
2468 PL_expect = XTERMORDORDOR;
79072805
LW
2469 return THING;
2470 }
e3f73d4e
RGS
2471 else if (op_type == OP_BACKTICK && PL_lex_op) {
2472 /* readpipe() vas overriden */
2473 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
6154021b 2474 pl_yylval.opval = PL_lex_op;
9b201d7d 2475 PL_lex_op = NULL;
e3f73d4e
RGS
2476 PL_lex_stuff = NULL;
2477 return THING;
2478 }
79072805 2479
3280af22 2480 PL_sublex_info.super_state = PL_lex_state;
eac04b2e 2481 PL_sublex_info.sub_inwhat = (U16)op_type;
3280af22
NIS
2482 PL_sublex_info.sub_op = PL_lex_op;
2483 PL_lex_state = LEX_INTERPPUSH;
55497cff 2484
3280af22
NIS
2485 PL_expect = XTERM;
2486 if (PL_lex_op) {
6154021b 2487 pl_yylval.opval = PL_lex_op;
5f66b61c 2488 PL_lex_op = NULL;
55497cff
PP
2489 return PMFUNC;
2490 }
2491 else
2492 return FUNC;
2493}
2494
ffb4593c
NT
2495/*
2496 * S_sublex_push
2497 * Create a new scope to save the lexing state. The scope will be
2498 * ended in S_sublex_done. Returns a '(', starting the function arguments
2499 * to the uc, lc, etc. found before.
2500 * Sets PL_lex_state to LEX_INTERPCONCAT.
2501 */
2502
76e3520e 2503STATIC I32
cea2e8a9 2504S_sublex_push(pTHX)
55497cff 2505{
27da23d5 2506 dVAR;
78a635de 2507 LEXSHARED *shared;
f46d017c 2508 ENTER;
55497cff 2509
3280af22 2510 PL_lex_state = PL_sublex_info.super_state;
651b5b28 2511 SAVEBOOL(PL_lex_dojoin);
3280af22 2512 SAVEI32(PL_lex_brackets);
78cdf107 2513 SAVEI32(PL_lex_allbrackets);
b27dce25 2514 SAVEI32(PL_lex_formbrack);
78cdf107 2515 SAVEI8(PL_lex_fakeeof);
3280af22
NIS
2516 SAVEI32(PL_lex_casemods);
2517 SAVEI32(PL_lex_starts);
651b5b28 2518 SAVEI8(PL_lex_state);
7cc34111 2519 SAVESPTR(PL_lex_repl);
7766f137 2520 SAVEVPTR(PL_lex_inpat);
98246f1e 2521 SAVEI16(PL_lex_inwhat);
57843af0 2522 SAVECOPLINE(PL_curcop);
3280af22 2523 SAVEPPTR(PL_bufptr);
8452ff4b 2524 SAVEPPTR(PL_bufend);
3280af22
NIS
2525 SAVEPPTR(PL_oldbufptr);
2526 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
2527 SAVEPPTR(PL_last_lop);
2528 SAVEPPTR(PL_last_uni);
3280af22
NIS
2529 SAVEPPTR(PL_linestart);
2530 SAVESPTR(PL_linestr);
8edd5f42
RGS
2531 SAVEGENERICPV(PL_lex_brackstack);
2532 SAVEGENERICPV(PL_lex_casestack);
78a635de 2533 SAVEGENERICPV(PL_parser->lex_shared);
3a54fd60 2534 SAVEBOOL(PL_parser->lex_re_reparsing);
3280af22 2535
99bd9d90 2536 /* The here-doc parser needs to be able to peek into outer lexing
60f40a38
FC
2537 scopes to find the body of the here-doc. So we put PL_linestr and
2538 PL_bufptr into lex_shared, to ‘share’ those values.
99bd9d90 2539 */
60f40a38
FC
2540 PL_parser->lex_shared->ls_linestr = PL_linestr;
2541 PL_parser->lex_shared->ls_bufptr = PL_bufptr;
99bd9d90 2542
3280af22 2543 PL_linestr = PL_lex_stuff;
7cc34111 2544 PL_lex_repl = PL_sublex_info.repl;
a0714e2c 2545 PL_lex_stuff = NULL;
7cc34111 2546 PL_sublex_info.repl = NULL;
3280af22 2547
9cbb5ea2
GS
2548 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2549 = SvPVX(PL_linestr);
3280af22 2550 PL_bufend += SvCUR(PL_linestr);
bd61b366 2551 PL_last_lop = PL_last_uni = NULL;
3280af22 2552 SAVEFREESV(PL_linestr);
4dc843bc 2553 if (PL_lex_repl) SAVEFREESV(PL_lex_repl);
3280af22
NIS
2554
2555 PL_lex_dojoin = FALSE;
b27dce25 2556 PL_lex_brackets = PL_lex_formbrack = 0;
78cdf107
Z
2557 PL_lex_allbrackets = 0;
2558 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
a02a5408
JC
2559 Newx(PL_lex_brackstack, 120, char);
2560 Newx(PL_lex_casestack, 12, char);
3280af22
NIS
2561 PL_lex_casemods = 0;
2562 *PL_lex_casestack = '\0';
2563 PL_lex_starts = 0;
2564 PL_lex_state = LEX_INTERPCONCAT;
eb160463 2565 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
78a635de
FC
2566
2567 Newxz(shared, 1, LEXSHARED);
2568 shared->ls_prev = PL_parser->lex_shared;
2569 PL_parser->lex_shared = shared;
3280af22
NIS
2570
2571 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
bb16bae8 2572 if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
3280af22
NIS
2573 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2574 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 2575 else
5f66b61c 2576 PL_lex_inpat = NULL;
79072805 2577
3a54fd60
DM
2578 PL_parser->lex_re_reparsing = cBOOL(PL_in_eval & EVAL_RE_REPARSING);
2579 PL_in_eval &= ~EVAL_RE_REPARSING;
2580
55497cff 2581 return '(';
79072805
LW
2582}
2583
ffb4593c
NT
2584/*
2585 * S_sublex_done
2586 * Restores lexer state after a S_sublex_push.
2587 */
2588
76e3520e 2589STATIC I32
cea2e8a9 2590S_sublex_done(pTHX)
79072805 2591{
27da23d5 2592 dVAR;
3280af22 2593 if (!PL_lex_starts++) {
396482e1 2594 SV * const sv = newSVpvs("");
9aa983d2
JH
2595 if (SvUTF8(PL_linestr))
2596 SvUTF8_on(sv);
3280af22 2597 PL_expect = XOPERATOR;
6154021b 2598 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
2599 return THING;
2600 }
2601
3280af22
NIS
2602 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2603 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 2604 return yylex();
79072805
LW
2605 }
2606
ffb4593c 2607 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
bb16bae8 2608 assert(PL_lex_inwhat != OP_TRANSR);
3280af22
NIS
2609 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2610 PL_linestr = PL_lex_repl;
2611 PL_lex_inpat = 0;
2612 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2613 PL_bufend += SvCUR(PL_linestr);
bd61b366 2614 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2615 PL_lex_dojoin = FALSE;
2616 PL_lex_brackets = 0;
78cdf107
Z
2617 PL_lex_allbrackets = 0;
2618 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
3280af22
NIS
2619 PL_lex_casemods = 0;
2620 *PL_lex_casestack = '\0';
2621 PL_lex_starts = 0;
25da4f38 2622 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
2623 PL_lex_state = LEX_INTERPNORMAL;
2624 PL_lex_starts++;
e9fa98b2
HS
2625 /* we don't clear PL_lex_repl here, so that we can check later
2626 whether this is an evalled subst; that means we rely on the
2627 logic to ensure sublex_done() is called again only via the
2628 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 2629 }
e9fa98b2 2630 else {
3280af22 2631 PL_lex_state = LEX_INTERPCONCAT;
a0714e2c 2632 PL_lex_repl = NULL;
e9fa98b2 2633 }
79072805 2634 return ',';
ffed7fef
LW
2635 }
2636 else {
5db06880
NC
2637#ifdef PERL_MAD
2638 if (PL_madskills) {
cd81e915
NC
2639 if (PL_thiswhite) {
2640 if (!PL_endwhite)
6b29d1f5 2641 PL_endwhite = newSVpvs("");
cd81e915
NC
2642 sv_catsv(PL_endwhite, PL_thiswhite);
2643 PL_thiswhite = 0;
2644 }
2645 if (PL_thistoken)
76f68e9b 2646 sv_setpvs(PL_thistoken,"");
5db06880 2647 else
cd81e915 2648 PL_realtokenstart = -1;
5db06880
NC
2649 }
2650#endif
f46d017c 2651 LEAVE;
3280af22
NIS
2652 PL_bufend = SvPVX(PL_linestr);
2653 PL_bufend += SvCUR(PL_linestr);
2654 PL_expect = XOPERATOR;
09bef843 2655 PL_sublex_info.sub_inwhat = 0;
79072805 2656 return ')';
ffed7fef
LW
2657 }
2658}
2659
6f613c73
KW
2660PERL_STATIC_INLINE SV*
2661S_get_and_check_backslash_N_name(pTHX_ const char* s, const char* const e)
2662{
140b12ad
KW
2663 /* <s> points to first character of interior of \N{}, <e> to one beyond the
2664 * interior, hence to the "}". Finds what the name resolves to, returning
2665 * an SV* containing it; NULL if no valid one found */
2666
dd2b1b72 2667 SV* res = newSVpvn_flags(s, e - s, UTF ? SVf_UTF8 : 0);
6f613c73 2668
0c415a79
KW
2669 HV * table;
2670 SV **cvp;
2671 SV *cv;
2672 SV *rv;
2673 HV *stash;
2674 const U8* first_bad_char_loc;
2675 const char* backslash_ptr = s - 3; /* Points to the <\> of \N{... */
2676
6f613c73
KW
2677 PERL_ARGS_ASSERT_GET_AND_CHECK_BACKSLASH_N_NAME;
2678
107160e2
KW
2679 if (UTF && ! is_utf8_string_loc((U8 *) backslash_ptr,
2680 e - backslash_ptr,
2681 &first_bad_char_loc))
2682 {
2683 /* If warnings are on, this will print a more detailed analysis of what
2684 * is wrong than the error message below */
2685 utf8n_to_uvuni(first_bad_char_loc,
2686 e - ((char *) first_bad_char_loc),
2687 NULL, 0);
2688
2689 /* We deliberately don't try to print the malformed character, which
2690 * might not print very well; it also may be just the first of many
2691 * malformations, so don't print what comes after it */
2692 yyerror(Perl_form(aTHX_
2693 "Malformed UTF-8 character immediately after '%.*s'",
2694 (int) (first_bad_char_loc - (U8 *) backslash_ptr), backslash_ptr));
2695 return NULL;
2696 }
2697
2698 res = new_constant( NULL, 0, "charnames", res, NULL, backslash_ptr,
2699 /* include the <}> */
2700 e - backslash_ptr + 1);
6f613c73 2701 if (! SvPOK(res)) {
b6407c49 2702 SvREFCNT_dec_NN(res);
6f613c73
KW
2703 return NULL;
2704 }
2705
0c415a79
KW
2706 /* See if the charnames handler is the Perl core's, and if so, we can skip
2707 * the validation needed for a user-supplied one, as Perl's does its own
2708 * validation. */
2709 table = GvHV(PL_hintgv); /* ^H */
2710 cvp = hv_fetchs(table, "charnames", FALSE);
67a057d6
FC
2711 if (cvp && (cv = *cvp) && SvROK(cv) && ((rv = SvRV(cv)) != NULL)
2712 && SvTYPE(rv) == SVt_PVCV && ((stash = CvSTASH(rv)) != NULL))
0c415a79
KW
2713 {
2714 const char * const name = HvNAME(stash);
2715 if strEQ(name, "_charnames") {
2716 return res;
2717 }
2718 }
2719
bde9e88d
KW
2720 /* Here, it isn't Perl's charname handler. We can't rely on a
2721 * user-supplied handler to validate the input name. For non-ut8 input,
2722 * look to see that the first character is legal. Then loop through the
2723 * rest checking that each is a continuation */
6f613c73 2724
b6ba1137
KW
2725 /* This code needs to be sync'ed with a regex in _charnames.pm which does
2726 * the same thing */
2727
b6ba1137 2728 if (! UTF) {
bde9e88d 2729 if (! isALPHAU(*s)) {
b6ba1137
KW
2730 goto bad_charname;
2731 }
bde9e88d
KW
2732 s++;
2733 while (s < e) {
2734 if (! isCHARNAME_CONT(*s)) {
b6ba1137
KW
2735 goto bad_charname;
2736 }
3036c853 2737 if (*s == ' ' && *(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2738 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2739 "A sequence of multiple spaces in a charnames "
2740 "alias definition is deprecated");
bd299e29 2741 }
bde9e88d 2742 s++;
b6ba1137 2743 }
3036c853 2744 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2745 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2746 "Trailing white-space in a charnames alias "
2747 "definition is deprecated");
bd299e29 2748 }
b6ba1137
KW
2749 }
2750 else {
bde9e88d
KW
2751 /* Similarly for utf8. For invariants can check directly; for other
2752 * Latin1, can calculate their code point and check; otherwise use a
2753 * swash */
2754 if (UTF8_IS_INVARIANT(*s)) {
2755 if (! isALPHAU(*s)) {
140b12ad
KW
2756 goto bad_charname;
2757 }
bde9e88d
KW
2758 s++;
2759 } else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
2760 if (! isALPHAU(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s, *(s+1))))) {
b6ba1137 2761 goto bad_charname;
6f613c73 2762 }
bde9e88d 2763 s += 2;
6f613c73 2764 }
bde9e88d
KW
2765 else {
2766 if (! PL_utf8_charname_begin) {
2767 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2768 PL_utf8_charname_begin = _core_swash_init("utf8",
2769 "_Perl_Charname_Begin",
2770 &PL_sv_undef,
2771 1, 0, NULL, &flags);
2772 }
2773 if (! swash_fetch(PL_utf8_charname_begin, (U8 *) s, TRUE)) {
2774 goto bad_charname;
2775 }
2776 s += UTF8SKIP(s);
2777 }
2778
2779 while (s < e) {
2780 if (UTF8_IS_INVARIANT(*s)) {
2781 if (! isCHARNAME_CONT(*s)) {
2782 goto bad_charname;
2783 }
3036c853
FC
2784 if (*s == ' ' && *(s-1) == ' '
2785 && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2786 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2787 "A sequence of multiple spaces in a charnam"
2788 "es alias definition is deprecated");
bd299e29 2789 }
bde9e88d
KW
2790 s++;
2791 }
2792 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
2793 if (! isCHARNAME_CONT(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s,
2794 *(s+1)))))
2795 {
2796 goto bad_charname;
2797 }
2798 s += 2;
2799 }
2800 else {
2801 if (! PL_utf8_charname_continue) {
2802 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2803 PL_utf8_charname_continue = _core_swash_init("utf8",
2804 "_Perl_Charname_Continue",
2805 &PL_sv_undef,
2806 1, 0, NULL, &flags);
2807 }
2808 if (! swash_fetch(PL_utf8_charname_continue, (U8 *) s, TRUE)) {
2809 goto bad_charname;
2810 }
2811 s += UTF8SKIP(s);
6f613c73
KW
2812 }
2813 }
3036c853 2814 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2815 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2816 "Trailing white-space in a charnames alias "
2817 "definition is deprecated");
bd299e29 2818 }
6f613c73
KW
2819 }
2820
94ca1619 2821 if (SvUTF8(res)) { /* Don't accept malformed input */
bde9e88d
KW
2822 const U8* first_bad_char_loc;
2823 STRLEN len;
2824 const char* const str = SvPV_const(res, len);
2825 if (! is_utf8_string_loc((U8 *) str, len, &first_bad_char_loc)) {
2826 /* If warnings are on, this will print a more detailed analysis of
2827 * what is wrong than the error message below */
2828 utf8n_to_uvuni(first_bad_char_loc,
2829 (char *) first_bad_char_loc - str,
2830 NULL, 0);
2831
2832 /* We deliberately don't try to print the malformed character,
2833 * which might not print very well; it also may be just the first
2834 * of many malformations, so don't print what comes after it */
2835 yyerror_pv(
2836 Perl_form(aTHX_
2837 "Malformed UTF-8 returned by %.*s immediately after '%.*s'",
2838 (int) (e - backslash_ptr + 1), backslash_ptr,
2839 (int) ((char *) first_bad_char_loc - str), str
2840 ),
2841 SVf_UTF8);
2842 return NULL;
2843 }
2844 }
140b12ad 2845
bde9e88d 2846 return res;
140b12ad 2847
bde9e88d
KW
2848 bad_charname: {
2849 int bad_char_size = ((UTF) ? UTF8SKIP(s) : 1);
2850
2851 /* The final %.*s makes sure that should the trailing NUL be missing
2852 * that this print won't run off the end of the string */
2853 yyerror_pv(
2854 Perl_form(aTHX_
2855 "Invalid character in \\N{...}; marked by <-- HERE in %.*s<-- HERE %.*s",
2856 (int)(s - backslash_ptr + bad_char_size), backslash_ptr,
2857 (int)(e - s + bad_char_size), s + bad_char_size
2858 ),
2859 UTF ? SVf_UTF8 : 0);
2860 return NULL;
2861 }
6f613c73
KW
2862}
2863
02aa26ce
NT
2864/*
2865 scan_const
2866
9da1dd8f
DM
2867 Extracts the next constant part of a pattern, double-quoted string,
2868 or transliteration. This is terrifying code.
2869
2870 For example, in parsing the double-quoted string "ab\x63$d", it would
2871 stop at the '$' and return an OP_CONST containing 'abc'.
02aa26ce 2872
94def140 2873 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
3280af22 2874 processing a pattern (PL_lex_inpat is true), a transliteration
94def140 2875 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
02aa26ce 2876
94def140
ST
2877 Returns a pointer to the character scanned up to. If this is
2878 advanced from the start pointer supplied (i.e. if anything was
9da1dd8f 2879 successfully parsed), will leave an OP_CONST for the substring scanned
6154021b 2880 in pl_yylval. Caller must intuit reason for not parsing further
9b599b2a
GS
2881 by looking at the next characters herself.
2882
02aa26ce 2883 In patterns:
9da1dd8f 2884 expand:
537124e4
KW
2885 \N{FOO} => \N{U+hex_for_character_FOO}
2886 (if FOO expands to multiple characters, expands to \N{U+xx.XX.yy ...})
9da1dd8f
DM
2887
2888 pass through:
2889 all other \-char, including \N and \N{ apart from \N{ABC}
2890
2891 stops on:
2892 @ and $ where it appears to be a var, but not for $ as tail anchor
2893 \l \L \u \U \Q \E
2894 (?{ or (??{
2895
02aa26ce
NT
2896
2897 In transliterations:
2898 characters are VERY literal, except for - not at the start or end
94def140
ST
2899 of the string, which indicates a range. If the range is in bytes,
2900 scan_const expands the range to the full set of intermediate
2901 characters. If the range is in utf8, the hyphen is replaced with
2902 a certain range mark which will be handled by pmtrans() in op.c.
02aa26ce
NT
2903
2904 In double-quoted strings:
2905 backslashes:
2906 double-quoted style: \r and \n
ff3f963a 2907 constants: \x31, etc.
94def140 2908 deprecated backrefs: \1 (in substitution replacements)
02aa26ce
NT
2909 case and quoting: \U \Q \E
2910 stops on @ and $
2911
2912 scan_const does *not* construct ops to handle interpolated strings.
2913 It stops processing as soon as it finds an embedded $ or @ variable
2914 and leaves it to the caller to work out what's going on.
2915
94def140
ST
2916 embedded arrays (whether in pattern or not) could be:
2917 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2918
2919 $ in double-quoted strings must be the symbol of an embedded scalar.
02aa26ce
NT
2920
2921 $ in pattern could be $foo or could be tail anchor. Assumption:
2922 it's a tail anchor if $ is the last thing in the string, or if it's
94def140 2923 followed by one of "()| \r\n\t"
02aa26ce 2924
9da1dd8f 2925 \1 (backreferences) are turned into $1 in substitutions
02aa26ce
NT
2926
2927 The structure of the code is
2928 while (there's a character to process) {
94def140
ST
2929 handle transliteration ranges
2930 skip regexp comments /(?#comment)/ and codes /(?{code})/
2931 skip #-initiated comments in //x patterns
2932 check for embedded arrays
02aa26ce
NT
2933 check for embedded scalars
2934 if (backslash) {
94def140 2935 deprecate \1 in substitution replacements
02aa26ce
NT
2936 handle string-changing backslashes \l \U \Q \E, etc.
2937 switch (what was escaped) {
94def140 2938 handle \- in a transliteration (becomes a literal -)
ff3f963a 2939 if a pattern and not \N{, go treat as regular character
94def140
ST
2940 handle \132 (octal characters)
2941 handle \x15 and \x{1234} (hex characters)
ff3f963a 2942 handle \N{name} (named characters, also \N{3,5} in a pattern)
94def140
ST
2943 handle \cV (control characters)
2944 handle printf-style backslashes (\f, \r, \n, etc)
02aa26ce 2945 } (end switch)
77a135fe 2946 continue
02aa26ce 2947 } (end if backslash)
77a135fe 2948 handle regular character
02aa26ce 2949 } (end while character to read)
4e553d73 2950
02aa26ce
NT
2951*/
2952
76e3520e 2953STATIC char *
cea2e8a9 2954S_scan_const(pTHX_ char *start)
79072805 2955{
97aff369 2956 dVAR;
eb578fdb 2957 char *send = PL_bufend; /* end of the constant */
77a135fe
KW
2958 SV *sv = newSV(send - start); /* sv for the constant. See
2959 note below on sizing. */
eb578fdb
KW
2960 char *s = start; /* start of the constant */
2961 char *d = SvPVX(sv); /* destination for copies */
02aa26ce 2962 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 2963 bool didrange = FALSE; /* did we just finish a range? */
2866decb 2964 bool in_charclass = FALSE; /* within /[...]/ */
b953e60c
KW
2965 bool has_utf8 = FALSE; /* Output constant is UTF8 */
2966 bool this_utf8 = cBOOL(UTF); /* Is the source string assumed
77a135fe
KW
2967 to be UTF8? But, this can
2968 show as true when the source
2969 isn't utf8, as for example
2970 when it is entirely composed
2971 of hex constants */
6f613c73 2972 SV *res; /* result from charnames */
77a135fe
KW
2973
2974 /* Note on sizing: The scanned constant is placed into sv, which is
2975 * initialized by newSV() assuming one byte of output for every byte of
2976 * input. This routine expects newSV() to allocate an extra byte for a
2977 * trailing NUL, which this routine will append if it gets to the end of
2978 * the input. There may be more bytes of input than output (eg., \N{LATIN
2979 * CAPITAL LETTER A}), or more output than input if the constant ends up
2980 * recoded to utf8, but each time a construct is found that might increase
2981 * the needed size, SvGROW() is called. Its size parameter each time is
2982 * based on the best guess estimate at the time, namely the length used so
2983 * far, plus the length the current construct will occupy, plus room for
2984 * the trailing NUL, plus one byte for every input byte still unscanned */
2985
c3320c2a
KW
2986 UV uv = UV_MAX; /* Initialize to weird value to try to catch any uses
2987 before set */
4c3a8340
ST
2988#ifdef EBCDIC
2989 UV literal_endpoint = 0;
e294cc5d 2990 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
4c3a8340 2991#endif
012bcf8d 2992
7918f24d
NC
2993 PERL_ARGS_ASSERT_SCAN_CONST;
2994
bb16bae8 2995 assert(PL_lex_inwhat != OP_TRANSR);
2b9d42f0
NIS
2996 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2997 /* If we are doing a trans and we know we want UTF8 set expectation */
2998 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
2999 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3000 }
3001
b899e89d
FC
3002 /* Protect sv from errors and fatal warnings. */
3003 ENTER_with_name("scan_const");
3004 SAVEFREESV(sv);
2b9d42f0 3005
79072805 3006 while (s < send || dorange) {
ff3f963a 3007
02aa26ce 3008 /* get transliterations out of the way (they're most literal) */
3280af22 3009 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 3010 /* expand a range A-Z to the full set of characters. AIE! */
79072805 3011 if (dorange) {
1ba5c669
JH
3012 I32 i; /* current expanded character */
3013 I32 min; /* first character in range */
3014 I32 max; /* last character in range */
02aa26ce 3015
e294cc5d
JH
3016#ifdef EBCDIC
3017 UV uvmax = 0;
3018#endif
3019
3020 if (has_utf8
3021#ifdef EBCDIC
3022 && !native_range
3023#endif
1953db30 3024 ) {
9d4ba2ae 3025 char * const c = (char*)utf8_hop((U8*)d, -1);
8973db79
JH
3026 char *e = d++;
3027 while (e-- > c)
3028 *(e + 1) = *e;
25716404 3029 *c = (char)UTF_TO_NATIVE(0xff);
8973db79
JH
3030 /* mark the range as done, and continue */
3031 dorange = FALSE;
3032 didrange = TRUE;
3033 continue;
3034 }
2b9d42f0 3035
95a20fc0 3036 i = d - SvPVX_const(sv); /* remember current offset */
e294cc5d
JH
3037#ifdef EBCDIC
3038 SvGROW(sv,
3039 SvLEN(sv) + (has_utf8 ?
3040 (512 - UTF_CONTINUATION_MARK +
3041 UNISKIP(0x100))
3042 : 256));
3043 /* How many two-byte within 0..255: 128 in UTF-8,
3044 * 96 in UTF-8-mod. */
3045#else
9cbb5ea2 3046 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
e294cc5d 3047#endif
9cbb5ea2 3048 d = SvPVX(sv) + i; /* refresh d after realloc */
e294cc5d
JH
3049#ifdef EBCDIC
3050 if (has_utf8) {
3051 int j;
3052 for (j = 0; j <= 1; j++) {
3053 char * const c = (char*)utf8_hop((U8*)d, -1);
3054 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
3055 if (j)
3056 min = (U8)uv;
3057 else if (uv < 256)
3058 max = (U8)uv;
3059 else {
3060 max = (U8)0xff; /* only to \xff */
3061 uvmax = uv; /* \x{100} to uvmax */
3062 }
3063 d = c; /* eat endpoint chars */
3064 }
3065 }
3066 else {
3067#endif
3068 d -= 2; /* eat the first char and the - */
3069 min = (U8)*d; /* first char in range */
3070 max = (U8)d[1]; /* last char in range */
3071#ifdef EBCDIC
3072 }
3073#endif
8ada0baa 3074
c2e66d9e 3075 if (min > max) {
01ec43d0 3076 Perl_croak(aTHX_
d1573ac7 3077 "Invalid range \"%c-%c\" in transliteration operator",
1ba5c669 3078 (char)min, (char)max);
c2e66d9e
GS
3079 }
3080
c7f1f016 3081#ifdef EBCDIC
4c3a8340
ST
3082 if (literal_endpoint == 2 &&
3083 ((isLOWER(min) && isLOWER(max)) ||
3084 (isUPPER(min) && isUPPER(max)))) {
8ada0baa
JH
3085 if (isLOWER(min)) {
3086 for (i = min; i <= max; i++)
3087 if (isLOWER(i))
db42d148 3088 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
3089 } else {
3090 for (i = min; i <= max; i++)
3091 if (isUPPER(i))
db42d148 3092 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
3093 }
3094 }
3095 else
3096#endif
3097 for (i = min; i <= max; i++)
e294cc5d
JH
3098#ifdef EBCDIC
3099 if (has_utf8) {
3100 const U8 ch = (U8)NATIVE_TO_UTF(i);
3101 if (UNI_IS_INVARIANT(ch))
3102 *d++ = (U8)i;
3103 else {
3104 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
3105 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
3106 }
3107 }
3108 else
3109#endif
3110 *d++ = (char)i;
3111
3112#ifdef EBCDIC
3113 if (uvmax) {
3114 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
3115 if (uvmax > 0x101)
3116 *d++ = (char)UTF_TO_NATIVE(0xff);
3117 if (uvmax > 0x100)
3118 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
3119 }
3120#endif
02aa26ce
NT
3121
3122 /* mark the range as done, and continue */
79072805 3123 dorange = FALSE;
01ec43d0 3124 didrange = TRUE;
4c3a8340
ST
3125#ifdef EBCDIC
3126 literal_endpoint = 0;
3127#endif
79072805 3128 continue;
4e553d73 3129 }
02aa26ce
NT
3130
3131 /* range begins (ignore - as first or last char) */
79072805 3132 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 3133 if (didrange) {
1fafa243 3134 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 3135 }
e294cc5d
JH
3136 if (has_utf8
3137#ifdef EBCDIC
3138 && !native_range
3139#endif
3140 ) {
25716404 3141 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
3142 s++;
3143 continue;
3144 }
79072805
LW
3145 dorange = TRUE;
3146 s++;
01ec43d0
GS
3147 }
3148 else {
3149 didrange = FALSE;
4c3a8340
ST
3150#ifdef EBCDIC
3151 literal_endpoint = 0;
e294cc5d 3152 native_range = TRUE;
4c3a8340 3153#endif
01ec43d0 3154 }
79072805 3155 }
02aa26ce
NT
3156
3157 /* if we get here, we're not doing a transliteration */
3158
e4a2df84
DM
3159 else if (*s == '[' && PL_lex_inpat && !in_charclass) {
3160 char *s1 = s-1;
3161 int esc = 0;
3162 while (s1 >= start && *s1-- == '\\')
3163 esc = !esc;
3164 if (!esc)
3165 in_charclass = TRUE;
3166 }
2866decb 3167
e4a2df84
DM
3168 else if (*s == ']' && PL_lex_inpat && in_charclass) {
3169 char *s1 = s-1;
3170 int esc = 0;
3171 while (s1 >= start && *s1-- == '\\')
3172 esc = !esc;
3173 if (!esc)
3174 in_charclass = FALSE;
3175 }
2866decb 3176
9da1dd8f
DM
3177 /* skip for regexp comments /(?#comment)/, except for the last
3178 * char, which will be done separately.
3179 * Stop on (?{..}) and friends */
3180
3280af22 3181 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395 3182 if (s[2] == '#') {
e994fd66 3183 while (s+1 < send && *s != ')')
db42d148 3184 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
155aba94 3185 }
2866decb 3186 else if (!PL_lex_casemods && !in_charclass &&
d3cec5e5
DM
3187 ( s[2] == '{' /* This should match regcomp.c */
3188 || (s[2] == '?' && s[3] == '{')))
155aba94 3189 {
9da1dd8f 3190 break;
cc6b7395 3191 }
748a9306 3192 }
02aa26ce
NT
3193
3194 /* likewise skip #-initiated comments in //x patterns */
3280af22 3195 else if (*s == '#' && PL_lex_inpat &&
73134a2e 3196 ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
748a9306 3197 while (s+1 < send && *s != '\n')
db42d148 3198 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
748a9306 3199 }
02aa26ce 3200
9da1dd8f
DM
3201 /* no further processing of single-quoted regex */
3202 else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'')
3203 goto default_action;
3204
5d1d4326 3205 /* check for embedded arrays
da6eedaa 3206 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
5d1d4326 3207 */
1749ea0d 3208 else if (*s == '@' && s[1]) {
8a2bca12 3209 if (isWORDCHAR_lazy_if(s+1,UTF))
1749ea0d
ST
3210 break;
3211 if (strchr(":'{$", s[1]))
3212 break;
3213 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
3214 break; /* in regexp, neither @+ nor @- are interpolated */
3215 }
02aa26ce
NT
3216
3217 /* check for embedded scalars. only stop if we're sure it's a
3218 variable.
3219 */
79072805 3220 else if (*s == '$') {
3280af22 3221 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 3222 break;
77772344 3223 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
a2a5de95
NC
3224 if (s[1] == '\\') {
3225 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3226 "Possible unintended interpolation of $\\ in regex");
77772344 3227 }
79072805 3228 break; /* in regexp, $ might be tail anchor */
77772344 3229 }
79072805 3230 }
02aa26ce 3231
2b9d42f0
NIS
3232 /* End of else if chain - OP_TRANS rejoin rest */
3233
02aa26ce 3234 /* backslashes */
79072805 3235 if (*s == '\\' && s+1 < send) {
ff3f963a
KW
3236 char* e; /* Can be used for ending '}', etc. */
3237
79072805 3238 s++;
02aa26ce 3239
7d0fc23c
KW
3240 /* warn on \1 - \9 in substitution replacements, but note that \11
3241 * is an octal; and \19 is \1 followed by '9' */
3280af22 3242 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 3243 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 3244 {
a2a5de95 3245 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
79072805
LW
3246 *--s = '$';
3247 break;
3248 }
02aa26ce
NT
3249
3250 /* string-change backslash escapes */
838f2281 3251 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) {
79072805
LW
3252 --s;
3253 break;
3254 }
ff3f963a
KW
3255 /* In a pattern, process \N, but skip any other backslash escapes.
3256 * This is because we don't want to translate an escape sequence
3257 * into a meta symbol and have the regex compiler use the meta
3258 * symbol meaning, e.g. \x{2E} would be confused with a dot. But
3259 * in spite of this, we do have to process \N here while the proper
3260 * charnames handler is in scope. See bugs #56444 and #62056.
3261 * There is a complication because \N in a pattern may also stand
3262 * for 'match a non-nl', and not mean a charname, in which case its
3263 * processing should be deferred to the regex compiler. To be a
3264 * charname it must be followed immediately by a '{', and not look
3265 * like \N followed by a curly quantifier, i.e., not something like
3266 * \N{3,}. regcurly returns a boolean indicating if it is a legal
3267 * quantifier */
3268 else if (PL_lex_inpat
3269 && (*s != 'N'
3270 || s[1] != '{'
4d68ffa0 3271 || regcurly(s + 1, FALSE)))
ff3f963a 3272 {
cc74c5bd
ST
3273 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
3274 goto default_action;
3275 }
02aa26ce 3276
79072805 3277 switch (*s) {
02aa26ce
NT
3278
3279 /* quoted - in transliterations */
79072805 3280 case '-':
3280af22 3281 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
3282 *d++ = *s++;
3283 continue;
3284 }
3285 /* FALL THROUGH */
3286 default:
11b8faa4 3287 {
15861f94 3288 if ((isALPHANUMERIC(*s)))
a2a5de95
NC
3289 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
3290 "Unrecognized escape \\%c passed through",
3291 *s);
11b8faa4 3292 /* default action is to copy the quoted character */
f9a63242 3293 goto default_action;
11b8faa4 3294 }
02aa26ce 3295
632403cc 3296 /* eg. \132 indicates the octal constant 0132 */
79072805
LW
3297 case '0': case '1': case '2': case '3':
3298 case '4': case '5': case '6': case '7':
ba210ebe 3299 {
5e0a247b 3300 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
53305cf1 3301 STRLEN len = 3;
77a135fe 3302 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
ba210ebe 3303 s += len;
5e0a247b
KW
3304 if (len < 3 && s < send && isDIGIT(*s)
3305 && ckWARN(WARN_MISC))
3306 {
3307 Perl_warner(aTHX_ packWARN(WARN_MISC),
3308 "%s", form_short_octal_warning(s, len));
3309 }
ba210ebe 3310 }
012bcf8d 3311 goto NUM_ESCAPE_INSERT;
02aa26ce 3312
f0a2b745
KW
3313 /* eg. \o{24} indicates the octal constant \024 */
3314 case 'o':
3315 {
454155d9 3316 const char* error;
f0a2b745 3317
00ce5563 3318 bool valid = grok_bslash_o(&s, &uv, &error,
80f4111b
KW
3319 TRUE, /* Output warning */
3320 FALSE, /* Not strict */
17896a85
KW
3321 TRUE, /* Output warnings for
3322 non-portables */
80f4111b 3323 UTF);
454155d9 3324 if (! valid) {
f0a2b745
KW
3325 yyerror(error);
3326 continue;
3327 }
3328 goto NUM_ESCAPE_INSERT;
3329 }
3330
77a135fe 3331 /* eg. \x24 indicates the hex constant 0x24 */
79072805 3332 case 'x':
a0481293 3333 {
a0481293 3334 const char* error;
355860ce 3335
00ce5563 3336 bool valid = grok_bslash_x(&s, &uv, &error,
80f4111b
KW
3337 TRUE, /* Output warning */
3338 FALSE, /* Not strict */
17896a85
KW
3339 TRUE, /* Output warnings for
3340 non-portables */
80f4111b 3341 UTF);
a0481293
KW
3342 if (! valid) {
3343 yyerror(error);
355860ce 3344 continue;
ba210ebe 3345 }
012bcf8d
GS
3346 }
3347
3348 NUM_ESCAPE_INSERT:
ff3f963a
KW
3349 /* Insert oct or hex escaped character. There will always be
3350 * enough room in sv since such escapes will be longer than any
3351 * UTF-8 sequence they can end up as, except if they force us
3352 * to recode the rest of the string into utf8 */
ba7cea30 3353
77a135fe 3354 /* Here uv is the ordinal of the next character being added in
ff3f963a 3355 * unicode (converted from native). */
77a135fe 3356 if (!UNI_IS_INVARIANT(uv)) {
9aa983d2 3357 if (!has_utf8 && uv > 255) {
77a135fe
KW
3358 /* Might need to recode whatever we have accumulated so
3359 * far if it contains any chars variant in utf8 or
3360 * utf-ebcdic. */
3361
3362 SvCUR_set(sv, d - SvPVX_const(sv));
3363 SvPOK_on(sv);
3364 *d = '\0';
77a135fe 3365 /* See Note on sizing above. */
7bf79863
KW
3366 sv_utf8_upgrade_flags_grow(sv,
3367 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3368 UNISKIP(uv) + (STRLEN)(send - s) + 1);
77a135fe
KW
3369 d = SvPVX(sv) + SvCUR(sv);
3370 has_utf8 = TRUE;
012bcf8d
GS
3371 }
3372
77a135fe
KW
3373 if (has_utf8) {
3374 d = (char*)uvuni_to_utf8((U8*)d, uv);
f9a63242
JH
3375 if (PL_lex_inwhat == OP_TRANS &&
3376 PL_sublex_info.sub_op) {
3377 PL_sublex_info.sub_op->op_private |=
3378 (PL_lex_repl ? OPpTRANS_FROM_UTF
3379 : OPpTRANS_TO_UTF);
f9a63242 3380 }
e294cc5d
JH
3381#ifdef EBCDIC
3382 if (uv > 255 && !dorange)
3383 native_range = FALSE;
3384#endif
012bcf8d 3385 }
a0ed51b3 3386 else {
012bcf8d 3387 *d++ = (char)uv;
a0ed51b3 3388 }
012bcf8d
GS
3389 }
3390 else {
c4d5f83a 3391 *d++ = (char) uv;
a0ed51b3 3392 }
79072805 3393 continue;
02aa26ce 3394
4a2d328f 3395 case 'N':
ff3f963a
KW
3396 /* In a non-pattern \N must be a named character, like \N{LATIN
3397 * SMALL LETTER A} or \N{U+0041}. For patterns, it also can
3398 * mean to match a non-newline. For non-patterns, named
3399 * characters are converted to their string equivalents. In
3400 * patterns, named characters are not converted to their
3401 * ultimate forms for the same reasons that other escapes
3402 * aren't. Instead, they are converted to the \N{U+...} form
3403 * to get the value from the charnames that is in effect right
3404 * now, while preserving the fact that it was a named character
3405 * so that the regex compiler knows this */
3406
3407 /* This section of code doesn't generally use the
3408 * NATIVE_TO_NEED() macro to transform the input. I (khw) did
3409 * a close examination of this macro and determined it is a
3410 * no-op except on utfebcdic variant characters. Every
3411 * character generated by this that would normally need to be
3412 * enclosed by this macro is invariant, so the macro is not
7538f724
KW
3413 * needed, and would complicate use of copy(). XXX There are
3414 * other parts of this file where the macro is used
3415 * inconsistently, but are saved by it being a no-op */
ff3f963a
KW
3416
3417 /* The structure of this section of code (besides checking for
3418 * errors and upgrading to utf8) is:
3419 * Further disambiguate between the two meanings of \N, and if
3420 * not a charname, go process it elsewhere
0a96133f
KW
3421 * If of form \N{U+...}, pass it through if a pattern;
3422 * otherwise convert to utf8
3423 * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3424 * pattern; otherwise convert to utf8 */
ff3f963a
KW
3425
3426 /* Here, s points to the 'N'; the test below is guaranteed to
3427 * succeed if we are being called on a pattern as we already
3428 * know from a test above that the next character is a '{'.
3429 * On a non-pattern \N must mean 'named sequence, which
3430 * requires braces */
3431 s++;
3432 if (*s != '{') {
3433 yyerror("Missing braces on \\N{}");
3434 continue;
3435 }
3436 s++;
3437
0a96133f 3438 /* If there is no matching '}', it is an error. */
ff3f963a
KW
3439 if (! (e = strchr(s, '}'))) {
3440 if (! PL_lex_inpat) {
5777a3f7 3441 yyerror("Missing right brace on \\N{}");
0a96133f
KW
3442 } else {
3443 yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
dbc0d4f2 3444 }
0a96133f 3445 continue;
ff3f963a 3446 }
cddc7ef4 3447
ff3f963a 3448 /* Here it looks like a named character */
cddc7ef4 3449
ff3f963a
KW
3450 if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3451 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3452 | PERL_SCAN_DISALLOW_PREFIX;
3453 STRLEN len;
3454
3455 /* For \N{U+...}, the '...' is a unicode value even on
3456 * EBCDIC machines */
3457 s += 2; /* Skip to next char after the 'U+' */
3458 len = e - s;
3459 uv = grok_hex(s, &len, &flags, NULL);
3460 if (len == 0 || len != (STRLEN)(e - s)) {
3461 yyerror("Invalid hexadecimal number in \\N{U+...}");
3462 s = e + 1;
3463 continue;
3464 }
3465
3466 if (PL_lex_inpat) {
3467
e2a7e165
KW
3468 /* On non-EBCDIC platforms, pass through to the regex
3469 * compiler unchanged. The reason we evaluated the
3470 * number above is to make sure there wasn't a syntax
3471 * error. But on EBCDIC we convert to native so
3472 * downstream code can continue to assume it's native
3473 */
ff3f963a 3474 s -= 5; /* Include the '\N{U+' */
e2a7e165
KW
3475#ifdef EBCDIC
3476 d += my_snprintf(d, e - s + 1 + 1, /* includes the }
3477 and the \0 */
3478 "\\N{U+%X}",
3479 (unsigned int) UNI_TO_NATIVE(uv));
3480#else
ff3f963a
KW
3481 Copy(s, d, e - s + 1, char); /* 1 = include the } */
3482 d += e - s + 1;
e2a7e165 3483#endif
ff3f963a
KW
3484 }
3485 else { /* Not a pattern: convert the hex to string */
3486
3487 /* If destination is not in utf8, unconditionally
3488 * recode it to be so. This is because \N{} implies
3489 * Unicode semantics, and scalars have to be in utf8
3490 * to guarantee those semantics */
3491 if (! has_utf8) {
3492 SvCUR_set(sv, d - SvPVX_const(sv));
3493 SvPOK_on(sv);
3494 *d = '\0';
3495 /* See Note on sizing above. */
3496 sv_utf8_upgrade_flags_grow(
3497 sv,
3498 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3499 UNISKIP(uv) + (STRLEN)(send - e) + 1);
3500 d = SvPVX(sv) + SvCUR(sv);
3501 has_utf8 = TRUE;
3502 }
3503
3504 /* Add the string to the output */
3505 if (UNI_IS_INVARIANT(uv)) {
3506 *d++ = (char) uv;
3507 }
3508 else d = (char*)uvuni_to_utf8((U8*)d, uv);
3509 }
3510 }
6f613c73
KW
3511 else /* Here is \N{NAME} but not \N{U+...}. */
3512 if ((res = get_and_check_backslash_N_name(s, e)))
3513 {
3514 STRLEN len;
3515 const char *str = SvPV_const(res, len);
3516 if (PL_lex_inpat) {
ff3f963a
KW
3517
3518 if (! len) { /* The name resolved to an empty string */
3519 Copy("\\N{}", d, 4, char);
3520 d += 4;
3521 }
3522 else {
3523 /* In order to not lose information for the regex
3524 * compiler, pass the result in the specially made
3525 * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3526 * the code points in hex of each character
3527 * returned by charnames */
3528
3529 const char *str_end = str + len;
3b721c4f 3530 const STRLEN off = d - SvPVX_const(sv);
94ca1619
KW
3531
3532 if (! SvUTF8(res)) {
3533 /* For the non-UTF-8 case, we can determine the
3534 * exact length needed without having to parse
3535 * through the string. Each character takes up
3536 * 2 hex digits plus either a trailing dot or
3537 * the "}" */
3538 d = off + SvGROW(sv, off
3539 + 3 * len
3540 + 6 /* For the "\N{U+", and
3541 trailing NUL */
3542 + (STRLEN)(send - e));
3543 Copy("\\N{U+", d, 5, char);
3544 d += 5;
3545 while (str < str_end) {
3546 char hex_string[4];
3547 my_snprintf(hex_string, sizeof(hex_string),
3548 "%02X.", (U8) *str);
3549 Copy(hex_string, d, 3, char);
3550 d += 3;
3551 str++;
3552 }
3553 d--; /* We will overwrite below the final
3554 dot with a right brace */
3555 }
3556 else {
1953db30
KW
3557 STRLEN char_length; /* cur char's byte length */
3558
3559 /* and the number of bytes after this is
3560 * translated into hex digits */
3561 STRLEN output_length;
3562
3563 /* 2 hex per byte; 2 chars for '\N'; 2 chars
3564 * for max('U+', '.'); and 1 for NUL */
3565 char hex_string[2 * UTF8_MAXBYTES + 5];
3566
3567 /* Get the first character of the result. */
3568 U32 uv = utf8n_to_uvuni((U8 *) str,
3569 len,
3570 &char_length,
3571 UTF8_ALLOW_ANYUV);
3572 /* Convert first code point to hex, including
3573 * the boiler plate before it. For all these,
3574 * we convert to native format so that
3575 * downstream code can continue to assume the
3576 * input is native */
3577 output_length =
3578 my_snprintf(hex_string, sizeof(hex_string),
3579 "\\N{U+%X",
3580 (unsigned int) UNI_TO_NATIVE(uv));
3581
3582 /* Make sure there is enough space to hold it */
3583 d = off + SvGROW(sv, off
3584 + output_length
3585 + (STRLEN)(send - e)
3586 + 2); /* '}' + NUL */
3587 /* And output it */
3588 Copy(hex_string, d, output_length, char);
3589 d += output_length;
3590
3591 /* For each subsequent character, append dot and
3592 * its ordinal in hex */
3593 while ((str += char_length) < str_end) {
3594 const STRLEN off = d - SvPVX_const(sv);
3595 U32 uv = utf8n_to_uvuni((U8 *) str,
3596 str_end - str,
3597 &char_length,
3598 UTF8_ALLOW_ANYUV);
3599 output_length =
3600 my_snprintf(hex_string,
3601 sizeof(hex_string),
3602 ".%X",
3603 (unsigned int) UNI_TO_NATIVE(uv));
3604
3605 d = off + SvGROW(sv, off
3606 + output_length
3607 + (STRLEN)(send - e)
3608 + 2); /* '}' + NUL */
3609 Copy(hex_string, d, output_length, char);
3610 d += output_length;
3611 }
94ca1619 3612 }
ff3f963a
KW
3613
3614 *d++ = '}'; /* Done. Add the trailing brace */
3615 }
3616 }
3617 else { /* Here, not in a pattern. Convert the name to a
3618 * string. */
3619
3620 /* If destination is not in utf8, unconditionally
3621 * recode it to be so. This is because \N{} implies
3622 * Unicode semantics, and scalars have to be in utf8
3623 * to guarantee those semantics */
3624 if (! has_utf8) {
3625 SvCUR_set(sv, d - SvPVX_const(sv));
3626 SvPOK_on(sv);
3627 *d = '\0';
3628 /* See Note on sizing above. */
3629 sv_utf8_upgrade_flags_grow(sv,
3630 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3631 len + (STRLEN)(send - s) + 1);
3632 d = SvPVX(sv) + SvCUR(sv);
3633 has_utf8 = TRUE;
3634 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
3635
3636 /* See Note on sizing above. (NOTE: SvCUR() is not
3637 * set correctly here). */
3638 const STRLEN off = d - SvPVX_const(sv);
3639 d = off + SvGROW(sv, off + len + (STRLEN)(send - s) + 1);
3640 }
3641 Copy(str, d, len, char);
3642 d += len;
423cee85 3643 }
6f613c73 3644
423cee85 3645 SvREFCNT_dec(res);
cb233ae3 3646
cb233ae3 3647 } /* End \N{NAME} */
ff3f963a
KW
3648#ifdef EBCDIC
3649 if (!dorange)
3650 native_range = FALSE; /* \N{} is defined to be Unicode */
3651#endif
3652 s = e + 1; /* Point to just after the '}' */
423cee85
JH
3653 continue;
3654
02aa26ce 3655 /* \c is a control character */
79072805
LW
3656 case 'c':
3657 s++;
961ce445 3658 if (s < send) {
17a3df4c 3659 *d++ = grok_bslash_c(*s++, has_utf8, 1);
ba210ebe 3660 }
961ce445
RGS
3661 else {
3662 yyerror("Missing control char name in \\c");
3663 }
79072805 3664 continue;
02aa26ce
NT
3665
3666 /* printf-style backslashes, formfeeds, newlines, etc */
79072805 3667 case 'b':
db42d148 3668 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
79072805
LW
3669 break;
3670 case 'n':
db42d148 3671 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
79072805
LW
3672 break;
3673 case 'r':
db42d148 3674 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
79072805
LW
3675 break;
3676 case 'f':
db42d148 3677 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
79072805
LW
3678 break;
3679 case 't':
db42d148 3680 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
79072805 3681 break;
34a3fe2a 3682 case 'e':
db42d148 3683 *d++ = ASCII_TO_NEED(has_utf8,'\033');
34a3fe2a
PP
3684 break;
3685 case 'a':
db42d148 3686 *d++ = ASCII_TO_NEED(has_utf8,'\007');
79072805 3687 break;
02aa26ce
NT
3688 } /* end switch */
3689
79072805
LW
3690 s++;
3691 continue;
02aa26ce 3692 } /* end if (backslash) */
4c3a8340
ST
3693#ifdef EBCDIC
3694 else
3695 literal_endpoint++;
3696#endif
02aa26ce 3697
f9a63242 3698 default_action:
77a135fe
KW
3699 /* If we started with encoded form, or already know we want it,
3700 then encode the next character */
3701 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
2b9d42f0 3702 STRLEN len = 1;
77a135fe
KW
3703
3704
3705 /* One might think that it is wasted effort in the case of the
3706 * source being utf8 (this_utf8 == TRUE) to take the next character
3707 * in the source, convert it to an unsigned value, and then convert
3708 * it back again. But the source has not been validated here. The
3709 * routine that does the conversion checks for errors like
3710 * malformed utf8 */
3711