This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
ignore the build product lib/version.pod
[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;
c3eee33f 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++;
c3eee33f
FC
1542 PL_parser->linestart = s;
1543 if (s == bufend)
1544 need_incline = 1;
1545 else
1546 incline(s);
f0e67a1d
Z
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;
c3eee33f 1558 COPLINE_INC_WITH_HERELINES;
f0e67a1d 1559 got_more = lex_next_chunk(flags);
c3eee33f 1560 CopLINE_dec(PL_curcop);
f0e67a1d
Z
1561 s = PL_parser->bufptr;
1562 bufend = PL_parser->bufend;
1563 if (!got_more)
1564 break;
c3eee33f 1565 if (need_incline && PL_parser->rsfp) {
f0e67a1d
Z
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/*
fe788d6b
PM
1581
1582=for apidoc EXMp|bool|validate_proto|SV *name|SV *proto|bool warn
1583
1584This function performs syntax checking on a prototype, C<proto>.
1585If C<warn> is true, any illegal characters or mismatched brackets
1586will trigger illegalproto warnings, declaring that they were
1587detected in the prototype for C<name>.
1588
1589The return value is C<true> if this is a valid prototype, and
1590C<false> if it is not, regardless of whether C<warn> was C<true> or
1591C<false>.
1592
1593Note that C<NULL> is a valid C<proto> and will always return C<true>.
1594
1595=cut
1596
1597 */
1598
1599bool
1600Perl_validate_proto(pTHX_ SV *name, SV *proto, bool warn)
1601{
1602 STRLEN len, origlen;
1603 char *p = proto ? SvPV(proto, len) : NULL;
1604 bool bad_proto = FALSE;
1605 bool in_brackets = FALSE;
1606 bool after_slash = FALSE;
1607 char greedy_proto = ' ';
1608 bool proto_after_greedy_proto = FALSE;
1609 bool must_be_last = FALSE;
1610 bool underscore = FALSE;
f791a21a 1611 bool bad_proto_after_underscore = FALSE;
fe788d6b
PM
1612
1613 PERL_ARGS_ASSERT_VALIDATE_PROTO;
1614
1615 if (!proto)
1616 return TRUE;
1617
1618 origlen = len;
1619 for (; len--; p++) {
1620 if (!isSPACE(*p)) {
1621 if (must_be_last)
1622 proto_after_greedy_proto = TRUE;
f791a21a
PM
1623 if (underscore) {
1624 if (!strchr(";@%", *p))
1625 bad_proto_after_underscore = TRUE;
1626 underscore = FALSE;
1627 }
fe788d6b
PM
1628 if (!strchr("$@%*;[]&\\_+", *p) || *p == '\0') {
1629 bad_proto = TRUE;
1630 }
1631 else {
fe788d6b
PM
1632 if (*p == '[')
1633 in_brackets = TRUE;
1634 else if (*p == ']')
1635 in_brackets = FALSE;
1636 else if ((*p == '@' || *p == '%') &&
1637 !after_slash &&
1638 !in_brackets ) {
1639 must_be_last = TRUE;
1640 greedy_proto = *p;
1641 }
1642 else if (*p == '_')
f791a21a 1643 underscore = TRUE;
fe788d6b
PM
1644 }
1645 if (*p == '\\')
1646 after_slash = TRUE;
1647 else
1648 after_slash = FALSE;
1649 }
1650 }
1651
1652 if (warn) {
b54d603d 1653 SV *tmpsv = newSVpvs_flags("", SVs_TEMP);
fe788d6b 1654 p -= origlen;
b54d603d
PM
1655 p = SvUTF8(proto)
1656 ? sv_uni_display(tmpsv, newSVpvn_flags(p, origlen, SVs_TEMP | SVf_UTF8),
1657 origlen, UNI_DISPLAY_ISPRINT)
1658 : pv_pretty(tmpsv, p, origlen, 60, NULL, NULL, PERL_PV_ESCAPE_NONASCII);
1659
fe788d6b
PM
1660 if (proto_after_greedy_proto)
1661 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1662 "Prototype after '%c' for %"SVf" : %s",
1663 greedy_proto, SVfARG(name), p);
50278ed0
PM
1664 if (in_brackets)
1665 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1666 "Missing ']' in prototype for %"SVf" : %s",
1667 SVfARG(name), p);
b54d603d 1668 if (bad_proto)
fe788d6b 1669 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
f791a21a
PM
1670 "Illegal character in prototype for %"SVf" : %s",
1671 SVfARG(name), p);
1672 if (bad_proto_after_underscore)
1673 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1674 "Illegal character after '_' in prototype for %"SVf" : %s",
1675 SVfARG(name), p);
fe788d6b
PM
1676 }
1677
1678 return (! (proto_after_greedy_proto || bad_proto) );
1679}
1680
1681/*
ffb4593c
NT
1682 * S_incline
1683 * This subroutine has nothing to do with tilting, whether at windmills
1684 * or pinball tables. Its name is short for "increment line". It
57843af0 1685 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 1686 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
1687 * # line 500 "foo.pm"
1688 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
1689 */
1690
76e3520e 1691STATIC void
d9095cec 1692S_incline(pTHX_ const char *s)
463ee0b2 1693{
97aff369 1694 dVAR;
d9095cec
NC
1695 const char *t;
1696 const char *n;
1697 const char *e;
8818d409 1698 line_t line_num;
463ee0b2 1699
7918f24d
NC
1700 PERL_ARGS_ASSERT_INCLINE;
1701
83944c01 1702 COPLINE_INC_WITH_HERELINES;
451f421f
FC
1703 if (!PL_rsfp && !PL_parser->filtered && PL_lex_state == LEX_NORMAL
1704 && s+1 == PL_bufend && *s == ';') {
1705 /* fake newline in string eval */
1706 CopLINE_dec(PL_curcop);
1707 return;
1708 }
463ee0b2
LW
1709 if (*s++ != '#')
1710 return;
d4c19fe8
AL
1711 while (SPACE_OR_TAB(*s))
1712 s++;
73659bf1
GS
1713 if (strnEQ(s, "line", 4))
1714 s += 4;
1715 else
1716 return;
084592ab 1717 if (SPACE_OR_TAB(*s))
73659bf1 1718 s++;
4e553d73 1719 else
73659bf1 1720 return;
d4c19fe8
AL
1721 while (SPACE_OR_TAB(*s))
1722 s++;
463ee0b2
LW
1723 if (!isDIGIT(*s))
1724 return;
d4c19fe8 1725
463ee0b2
LW
1726 n = s;
1727 while (isDIGIT(*s))
1728 s++;
07714eb4 1729 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
26b6dc3f 1730 return;
bf4acbe4 1731 while (SPACE_OR_TAB(*s))
463ee0b2 1732 s++;
73659bf1 1733 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 1734 s++;
73659bf1
GS
1735 e = t + 1;
1736 }
463ee0b2 1737 else {
c35e046a
AL
1738 t = s;
1739 while (!isSPACE(*t))
1740 t++;
73659bf1 1741 e = t;
463ee0b2 1742 }
bf4acbe4 1743 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
1744 e++;
1745 if (*e != '\n' && *e != '\0')
1746 return; /* false alarm */
1747
8818d409
FC
1748 line_num = atoi(n)-1;
1749
f4dd75d9 1750 if (t - s > 0) {
d9095cec 1751 const STRLEN len = t - s;
19bad673
NC
1752 SV *const temp_sv = CopFILESV(PL_curcop);
1753 const char *cf;
1754 STRLEN tmplen;
1755
1756 if (temp_sv) {
1757 cf = SvPVX(temp_sv);
1758 tmplen = SvCUR(temp_sv);
1759 } else {
1760 cf = NULL;
1761 tmplen = 0;
1762 }
1763
d1299d44 1764 if (!PL_rsfp && !PL_parser->filtered) {
e66cf94c
RGS
1765 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1766 * to *{"::_<newfilename"} */
44867030
NC
1767 /* However, the long form of evals is only turned on by the
1768 debugger - usually they're "(eval %lu)" */
1769 char smallbuf[128];
1770 char *tmpbuf;
1771 GV **gvp;
d9095cec 1772 STRLEN tmplen2 = len;
798b63bc 1773 if (tmplen + 2 <= sizeof smallbuf)
e66cf94c
RGS
1774 tmpbuf = smallbuf;
1775 else
2ae0db35 1776 Newx(tmpbuf, tmplen + 2, char);
44867030
NC
1777 tmpbuf[0] = '_';
1778 tmpbuf[1] = '<';
2ae0db35 1779 memcpy(tmpbuf + 2, cf, tmplen);
44867030 1780 tmplen += 2;
8a5ee598
RGS
1781 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
1782 if (gvp) {
44867030
NC
1783 char *tmpbuf2;
1784 GV *gv2;
1785
1786 if (tmplen2 + 2 <= sizeof smallbuf)
1787 tmpbuf2 = smallbuf;
1788 else
1789 Newx(tmpbuf2, tmplen2 + 2, char);
1790
1791 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
1792 /* Either they malloc'd it, or we malloc'd it,
1793 so no prefix is present in ours. */
1794 tmpbuf2[0] = '_';
1795 tmpbuf2[1] = '<';
1796 }
1797
1798 memcpy(tmpbuf2 + 2, s, tmplen2);
1799 tmplen2 += 2;
1800
8a5ee598 1801 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
e5527e4b 1802 if (!isGV(gv2)) {
8a5ee598 1803 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
e5527e4b
RGS
1804 /* adjust ${"::_<newfilename"} to store the new file name */
1805 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
8818d409
FC
1806 /* The line number may differ. If that is the case,
1807 alias the saved lines that are in the array.
1808 Otherwise alias the whole array. */
1809 if (CopLINE(PL_curcop) == line_num) {
1810 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
1811 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
1812 }
1813 else if (GvAV(*gvp)) {
1814 AV * const av = GvAV(*gvp);
1815 const I32 start = CopLINE(PL_curcop)+1;
1816 I32 items = AvFILLp(av) - start;
1817 if (items > 0) {
1818 AV * const av2 = GvAVn(gv2);
1819 SV **svp = AvARRAY(av) + start;
1820 I32 l = (I32)line_num+1;
1821 while (items--)
1822 av_store(av2, l++, SvREFCNT_inc(*svp++));
1823 }
1824 }
e5527e4b 1825 }
44867030
NC
1826
1827 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
8a5ee598 1828 }
e66cf94c 1829 if (tmpbuf != smallbuf) Safefree(tmpbuf);
e66cf94c 1830 }
05ec9bb3 1831 CopFILE_free(PL_curcop);
d9095cec 1832 CopFILE_setn(PL_curcop, s, len);
f4dd75d9 1833 }
8818d409 1834 CopLINE_set(PL_curcop, line_num);
463ee0b2
LW
1835}
1836
29595ff2 1837#ifdef PERL_MAD
cd81e915 1838/* skip space before PL_thistoken */
29595ff2
NC
1839
1840STATIC char *
5aaab254 1841S_skipspace0(pTHX_ char *s)
29595ff2 1842{
7918f24d
NC
1843 PERL_ARGS_ASSERT_SKIPSPACE0;
1844
29595ff2
NC
1845 s = skipspace(s);
1846 if (!PL_madskills)
1847 return s;
cd81e915
NC
1848 if (PL_skipwhite) {
1849 if (!PL_thiswhite)
6b29d1f5 1850 PL_thiswhite = newSVpvs("");
cd81e915
NC
1851 sv_catsv(PL_thiswhite, PL_skipwhite);
1852 sv_free(PL_skipwhite);
1853 PL_skipwhite = 0;
1854 }
1855 PL_realtokenstart = s - SvPVX(PL_linestr);
29595ff2
NC
1856 return s;
1857}
1858
cd81e915 1859/* skip space after PL_thistoken */
29595ff2
NC
1860
1861STATIC char *
5aaab254 1862S_skipspace1(pTHX_ char *s)
29595ff2 1863{
d4c19fe8 1864 const char *start = s;
29595ff2
NC
1865 I32 startoff = start - SvPVX(PL_linestr);
1866
7918f24d
NC
1867 PERL_ARGS_ASSERT_SKIPSPACE1;
1868
29595ff2
NC
1869 s = skipspace(s);
1870 if (!PL_madskills)
1871 return s;
1872 start = SvPVX(PL_linestr) + startoff;
cd81e915 1873 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1874 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1875 PL_thistoken = newSVpvn(tstart, start - tstart);
1876 }
1877 PL_realtokenstart = -1;
1878 if (PL_skipwhite) {
1879 if (!PL_nextwhite)
6b29d1f5 1880 PL_nextwhite = newSVpvs("");
cd81e915
NC
1881 sv_catsv(PL_nextwhite, PL_skipwhite);
1882 sv_free(PL_skipwhite);
1883 PL_skipwhite = 0;
29595ff2
NC
1884 }
1885 return s;
1886}
1887
1888STATIC char *
5aaab254 1889S_skipspace2(pTHX_ char *s, SV **svp)
29595ff2 1890{
c35e046a
AL
1891 char *start;
1892 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
1893 const I32 startoff = s - SvPVX(PL_linestr);
1894
7918f24d
NC
1895 PERL_ARGS_ASSERT_SKIPSPACE2;
1896
29595ff2
NC
1897 s = skipspace(s);
1898 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
1899 if (!PL_madskills || !svp)
1900 return s;
1901 start = SvPVX(PL_linestr) + startoff;
cd81e915 1902 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1903 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1904 PL_thistoken = newSVpvn(tstart, start - tstart);
1905 PL_realtokenstart = -1;
29595ff2 1906 }
cd81e915 1907 if (PL_skipwhite) {
29595ff2 1908 if (!*svp)
6b29d1f5 1909 *svp = newSVpvs("");
cd81e915
NC
1910 sv_setsv(*svp, PL_skipwhite);
1911 sv_free(PL_skipwhite);
1912 PL_skipwhite = 0;
29595ff2
NC
1913 }
1914
1915 return s;
1916}
1917#endif
1918
80a702cd 1919STATIC void
15f169a1 1920S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
80a702cd
RGS
1921{
1922 AV *av = CopFILEAVx(PL_curcop);
1923 if (av) {
b9f83d2f 1924 SV * const sv = newSV_type(SVt_PVMG);
5fa550fb 1925 if (orig_sv)
4e917a04 1926 sv_setsv_flags(sv, orig_sv, 0); /* no cow */
5fa550fb
NC
1927 else
1928 sv_setpvn(sv, buf, len);
80a702cd
RGS
1929 (void)SvIOK_on(sv);
1930 SvIV_set(sv, 0);
1931 av_store(av, (I32)CopLINE(PL_curcop), sv);
1932 }
1933}
1934
ffb4593c
NT
1935/*
1936 * S_skipspace
1937 * Called to gobble the appropriate amount and type of whitespace.
1938 * Skips comments as well.
1939 */
1940
76e3520e 1941STATIC char *
c3eee33f 1942S_skipspace(pTHX_ char *s)
a687059c 1943{
5db06880 1944#ifdef PERL_MAD
f0e67a1d
Z
1945 char *start = s;
1946#endif /* PERL_MAD */
c3eee33f 1947 PERL_ARGS_ASSERT_SKIPSPACE;
f0e67a1d 1948#ifdef PERL_MAD
cd81e915
NC
1949 if (PL_skipwhite) {
1950 sv_free(PL_skipwhite);
f0e67a1d 1951 PL_skipwhite = NULL;
5db06880 1952 }
f0e67a1d 1953#endif /* PERL_MAD */
3280af22 1954 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 1955 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2 1956 s++;
f0e67a1d
Z
1957 } else {
1958 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1959 PL_bufptr = s;
c3eee33f 1960 lex_read_space(LEX_KEEP_PREVIOUS |
f0998909
Z
1961 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1962 LEX_NO_NEXT_CHUNK : 0));
3280af22 1963 s = PL_bufptr;
f0e67a1d
Z
1964 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1965 if (PL_linestart > PL_bufptr)
1966 PL_bufptr = PL_linestart;
1967 return s;
463ee0b2 1968 }
5db06880 1969#ifdef PERL_MAD
f0e67a1d
Z
1970 if (PL_madskills)
1971 PL_skipwhite = newSVpvn(start, s-start);
1972#endif /* PERL_MAD */
5db06880 1973 return s;
a687059c 1974}
378cc40b 1975
ffb4593c
NT
1976/*
1977 * S_check_uni
1978 * Check the unary operators to ensure there's no ambiguity in how they're
1979 * used. An ambiguous piece of code would be:
1980 * rand + 5
1981 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1982 * the +5 is its argument.
1983 */
1984
76e3520e 1985STATIC void
cea2e8a9 1986S_check_uni(pTHX)
ba106d47 1987{
97aff369 1988 dVAR;
d4c19fe8
AL
1989 const char *s;
1990 const char *t;
2f3197b3 1991
3280af22 1992 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 1993 return;
3280af22
NIS
1994 while (isSPACE(*PL_last_uni))
1995 PL_last_uni++;
c35e046a 1996 s = PL_last_uni;
8a2bca12 1997 while (isWORDCHAR_lazy_if(s,UTF) || *s == '-')
c35e046a 1998 s++;
3280af22 1999 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 2000 return;
6136c704 2001
9b387841
NC
2002 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
2003 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
2004 (int)(s - PL_last_uni), PL_last_uni);
2f3197b3
LW
2005}
2006
ffb4593c
NT
2007/*
2008 * LOP : macro to build a list operator. Its behaviour has been replaced
2009 * with a subroutine, S_lop() for which LOP is just another name.
2010 */
2011
a0d0e21e
LW
2012#define LOP(f,x) return lop(f,x,s)
2013
ffb4593c
NT
2014/*
2015 * S_lop
2016 * Build a list operator (or something that might be one). The rules:
2017 * - if we have a next token, then it's a list operator [why?]
2018 * - if the next thing is an opening paren, then it's a function
2019 * - else it's a list operator
2020 */
2021
76e3520e 2022STATIC I32
a0be28da 2023S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 2024{
97aff369 2025 dVAR;
7918f24d
NC
2026
2027 PERL_ARGS_ASSERT_LOP;
2028
6154021b 2029 pl_yylval.ival = f;
35c8bce7 2030 CLINE;
3280af22
NIS
2031 PL_expect = x;
2032 PL_bufptr = s;
2033 PL_last_lop = PL_oldbufptr;
eb160463 2034 PL_last_lop_op = (OPCODE)f;
5db06880
NC
2035#ifdef PERL_MAD
2036 if (PL_lasttoke)
78cdf107 2037 goto lstop;
5db06880 2038#else
3280af22 2039 if (PL_nexttoke)
78cdf107 2040 goto lstop;
5db06880 2041#endif
79072805 2042 if (*s == '(')
bbf60fe6 2043 return REPORT(FUNC);
29595ff2 2044 s = PEEKSPACE(s);
79072805 2045 if (*s == '(')
bbf60fe6 2046 return REPORT(FUNC);
78cdf107
Z
2047 else {
2048 lstop:
2049 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
2050 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
bbf60fe6 2051 return REPORT(LSTOP);
78cdf107 2052 }
79072805
LW
2053}
2054
5db06880
NC
2055#ifdef PERL_MAD
2056 /*
2057 * S_start_force
2058 * Sets up for an eventual force_next(). start_force(0) basically does
2059 * an unshift, while start_force(-1) does a push. yylex removes items
2060 * on the "pop" end.
2061 */
2062
2063STATIC void
2064S_start_force(pTHX_ int where)
2065{
2066 int i;
2067
cd81e915 2068 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
5db06880 2069 where = PL_lasttoke;
cd81e915
NC
2070 assert(PL_curforce < 0 || PL_curforce == where);
2071 if (PL_curforce != where) {
5db06880
NC
2072 for (i = PL_lasttoke; i > where; --i) {
2073 PL_nexttoke[i] = PL_nexttoke[i-1];
2074 }
2075 PL_lasttoke++;
2076 }
cd81e915 2077 if (PL_curforce < 0) /* in case of duplicate start_force() */
5db06880 2078 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
cd81e915
NC
2079 PL_curforce = where;
2080 if (PL_nextwhite) {
5db06880 2081 if (PL_madskills)
6b29d1f5 2082 curmad('^', newSVpvs(""));
cd81e915 2083 CURMAD('_', PL_nextwhite);
5db06880
NC
2084 }
2085}
2086
2087STATIC void
2088S_curmad(pTHX_ char slot, SV *sv)
2089{
2090 MADPROP **where;
2091
2092 if (!sv)
2093 return;
cd81e915
NC
2094 if (PL_curforce < 0)
2095 where = &PL_thismad;
5db06880 2096 else
cd81e915 2097 where = &PL_nexttoke[PL_curforce].next_mad;
5db06880 2098
cd81e915 2099 if (PL_faketokens)
76f68e9b 2100 sv_setpvs(sv, "");
5db06880
NC
2101 else {
2102 if (!IN_BYTES) {
2103 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2104 SvUTF8_on(sv);
2105 else if (PL_encoding) {
2106 sv_recode_to_utf8(sv, PL_encoding);
2107 }
2108 }
2109 }
2110
2111 /* keep a slot open for the head of the list? */
2112 if (slot != '_' && *where && (*where)->mad_key == '^') {
2113 (*where)->mad_key = slot;
daba3364 2114 sv_free(MUTABLE_SV(((*where)->mad_val)));
5db06880
NC
2115 (*where)->mad_val = (void*)sv;
2116 }
2117 else
2118 addmad(newMADsv(slot, sv), where, 0);
2119}
2120#else
b3f24c00
MHM
2121# define start_force(where) NOOP
2122# define curmad(slot, sv) NOOP
5db06880
NC
2123#endif
2124
ffb4593c
NT
2125/*
2126 * S_force_next
9cbb5ea2 2127 * When the lexer realizes it knows the next token (for instance,
ffb4593c 2128 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2 2129 * to know what token to return the next time the lexer is called. Caller
5db06880
NC
2130 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
2131 * and possibly PL_expect to ensure the lexer handles the token correctly.
ffb4593c
NT
2132 */
2133
4e553d73 2134STATIC void
cea2e8a9 2135S_force_next(pTHX_ I32 type)
79072805 2136{
97aff369 2137 dVAR;
704d4215
GG
2138#ifdef DEBUGGING
2139 if (DEBUG_T_TEST) {
2140 PerlIO_printf(Perl_debug_log, "### forced token:\n");
f05d7009 2141 tokereport(type, &NEXTVAL_NEXTTOKE);
704d4215
GG
2142 }
2143#endif
5db06880 2144#ifdef PERL_MAD
cd81e915 2145 if (PL_curforce < 0)
5db06880 2146 start_force(PL_lasttoke);
cd81e915 2147 PL_nexttoke[PL_curforce].next_type = type;
5db06880
NC
2148 if (PL_lex_state != LEX_KNOWNEXT)
2149 PL_lex_defer = PL_lex_state;
2150 PL_lex_state = LEX_KNOWNEXT;
2151 PL_lex_expect = PL_expect;
cd81e915 2152 PL_curforce = -1;
5db06880 2153#else
3280af22
NIS
2154 PL_nexttype[PL_nexttoke] = type;
2155 PL_nexttoke++;
2156 if (PL_lex_state != LEX_KNOWNEXT) {
2157 PL_lex_defer = PL_lex_state;
2158 PL_lex_expect = PL_expect;
2159 PL_lex_state = LEX_KNOWNEXT;
79072805 2160 }
5db06880 2161#endif
79072805
LW
2162}
2163
28ac2b49
Z
2164void
2165Perl_yyunlex(pTHX)
2166{
a7aaec61
Z
2167 int yyc = PL_parser->yychar;
2168 if (yyc != YYEMPTY) {
2169 if (yyc) {
2170 start_force(-1);
2171 NEXTVAL_NEXTTOKE = PL_parser->yylval;
2172 if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
78cdf107 2173 PL_lex_allbrackets--;
a7aaec61 2174 PL_lex_brackets--;
78cdf107
Z
2175 yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
2176 } else if (yyc == '('/*)*/) {
2177 PL_lex_allbrackets--;
2178 yyc |= (2<<24);
a7aaec61
Z
2179 }
2180 force_next(yyc);
2181 }
28ac2b49
Z
2182 PL_parser->yychar = YYEMPTY;
2183 }
2184}
2185
d0a148a6 2186STATIC SV *
15f169a1 2187S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
d0a148a6 2188{
97aff369 2189 dVAR;
740cce10 2190 SV * const sv = newSVpvn_utf8(start, len,
eaf7a4d2
CS
2191 !IN_BYTES
2192 && UTF
2193 && !is_ascii_string((const U8*)start, len)
740cce10 2194 && is_utf8_string((const U8*)start, len));
d0a148a6
NC
2195 return sv;
2196}
2197
ffb4593c
NT
2198/*
2199 * S_force_word
2200 * When the lexer knows the next thing is a word (for instance, it has
2201 * just seen -> and it knows that the next char is a word char, then
02b34bbe
DM
2202 * it calls S_force_word to stick the next word into the PL_nexttoke/val
2203 * lookahead.
ffb4593c
NT
2204 *
2205 * Arguments:
b1b65b59 2206 * char *start : buffer position (must be within PL_linestr)
02b34bbe 2207 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
ffb4593c
NT
2208 * int check_keyword : if true, Perl checks to make sure the word isn't
2209 * a keyword (do this if the word is a label, e.g. goto FOO)
2210 * int allow_pack : if true, : characters will also be allowed (require,
2211 * use, etc. do this)
9cbb5ea2 2212 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
2213 */
2214
76e3520e 2215STATIC char *
345b3785 2216S_force_word(pTHX_ char *start, int token, int check_keyword, int allow_pack)
79072805 2217{
97aff369 2218 dVAR;
eb578fdb 2219 char *s;
463ee0b2 2220 STRLEN len;
4e553d73 2221
7918f24d
NC
2222 PERL_ARGS_ASSERT_FORCE_WORD;
2223
29595ff2 2224 start = SKIPSPACE1(start);
463ee0b2 2225 s = start;
7e2040f0 2226 if (isIDFIRST_lazy_if(s,UTF) ||
345b3785 2227 (allow_pack && *s == ':') )
a0d0e21e 2228 {
3280af22 2229 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
01b5ef50
FC
2230 if (check_keyword) {
2231 char *s2 = PL_tokenbuf;
2232 if (allow_pack && len > 6 && strnEQ(s2, "CORE::", 6))
2233 s2 += 6, len -= 6;
2234 if (keyword(s2, len, 0))
463ee0b2 2235 return start;
01b5ef50 2236 }
cd81e915 2237 start_force(PL_curforce);
5db06880
NC
2238 if (PL_madskills)
2239 curmad('X', newSVpvn(start,s-start));
463ee0b2 2240 if (token == METHOD) {
29595ff2 2241 s = SKIPSPACE1(s);
463ee0b2 2242 if (*s == '(')
3280af22 2243 PL_expect = XTERM;
463ee0b2 2244 else {
3280af22 2245 PL_expect = XOPERATOR;
463ee0b2 2246 }
79072805 2247 }
e74e6b3d 2248 if (PL_madskills)
63575281 2249 curmad('g', newSVpvs( "forced" ));
9ded7720 2250 NEXTVAL_NEXTTOKE.opval
d0a148a6
NC
2251 = (OP*)newSVOP(OP_CONST,0,
2252 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
9ded7720 2253 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
79072805
LW
2254 force_next(token);
2255 }
2256 return s;
2257}
2258
ffb4593c
NT
2259/*
2260 * S_force_ident
9cbb5ea2 2261 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
2262 * text only contains the "foo" portion. The first argument is a pointer
2263 * to the "foo", and the second argument is the type symbol to prefix.
2264 * Forces the next token to be a "WORD".
9cbb5ea2 2265 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
2266 */
2267
76e3520e 2268STATIC void
5aaab254 2269S_force_ident(pTHX_ const char *s, int kind)
79072805 2270{
97aff369 2271 dVAR;
7918f24d
NC
2272
2273 PERL_ARGS_ASSERT_FORCE_IDENT;
2274
c9b48522
DD
2275 if (s[0]) {
2276 const STRLEN len = s[1] ? strlen(s) : 1; /* s = "\"" see yylex */
728847b1
BF
2277 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len,
2278 UTF ? SVf_UTF8 : 0));
cd81e915 2279 start_force(PL_curforce);
9ded7720 2280 NEXTVAL_NEXTTOKE.opval = o;
79072805 2281 force_next(WORD);
748a9306 2282 if (kind) {
11343788 2283 o->op_private = OPpCONST_ENTERED;
55497cff
PP
2284 /* XXX see note in pp_entereval() for why we forgo typo
2285 warnings if the symbol must be introduced in an eval.
2286 GSAR 96-10-12 */
90e5519e 2287 gv_fetchpvn_flags(s, len,
728847b1
BF
2288 (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2289 : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ),
90e5519e
NC
2290 kind == '$' ? SVt_PV :
2291 kind == '@' ? SVt_PVAV :
2292 kind == '%' ? SVt_PVHV :
a0d0e21e 2293 SVt_PVGV
90e5519e 2294 );
748a9306 2295 }
79072805
LW
2296 }
2297}
2298
3f33d153
FC
2299static void
2300S_force_ident_maybe_lex(pTHX_ char pit)
2301{
2302 start_force(PL_curforce);
2303 NEXTVAL_NEXTTOKE.ival = pit;
2304 force_next('p');
2305}
2306
1571675a
GS
2307NV
2308Perl_str_to_version(pTHX_ SV *sv)
2309{
2310 NV retval = 0.0;
2311 NV nshift = 1.0;
2312 STRLEN len;
cfd0369c 2313 const char *start = SvPV_const(sv,len);
9d4ba2ae 2314 const char * const end = start + len;
504618e9 2315 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
7918f24d
NC
2316
2317 PERL_ARGS_ASSERT_STR_TO_VERSION;
2318
1571675a 2319 while (start < end) {
ba210ebe 2320 STRLEN skip;
1571675a
GS
2321 UV n;
2322 if (utf)
9041c2e3 2323 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
2324 else {
2325 n = *(U8*)start;
2326 skip = 1;
2327 }
2328 retval += ((NV)n)/nshift;
2329 start += skip;
2330 nshift *= 1000;
2331 }
2332 return retval;
2333}
2334
4e553d73 2335/*
ffb4593c
NT
2336 * S_force_version
2337 * Forces the next token to be a version number.
e759cc13
RGS
2338 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2339 * and if "guessing" is TRUE, then no new token is created (and the caller
2340 * must use an alternative parsing method).
ffb4593c
NT
2341 */
2342
76e3520e 2343STATIC char *
e759cc13 2344S_force_version(pTHX_ char *s, int guessing)
89bfa8cd 2345{
97aff369 2346 dVAR;
5f66b61c 2347 OP *version = NULL;
44dcb63b 2348 char *d;
5db06880
NC
2349#ifdef PERL_MAD
2350 I32 startoff = s - SvPVX(PL_linestr);
2351#endif
89bfa8cd 2352
7918f24d
NC
2353 PERL_ARGS_ASSERT_FORCE_VERSION;
2354
29595ff2 2355 s = SKIPSPACE1(s);
89bfa8cd 2356
44dcb63b 2357 d = s;
dd629d5b 2358 if (*d == 'v')
44dcb63b 2359 d++;
44dcb63b 2360 if (isDIGIT(*d)) {
e759cc13
RGS
2361 while (isDIGIT(*d) || *d == '_' || *d == '.')
2362 d++;
5db06880
NC
2363#ifdef PERL_MAD
2364 if (PL_madskills) {
cd81e915 2365 start_force(PL_curforce);
5db06880
NC
2366 curmad('X', newSVpvn(s,d-s));
2367 }
2368#endif
4e4da3ac 2369 if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
dd629d5b 2370 SV *ver;
8d08d9ba 2371#ifdef USE_LOCALE_NUMERIC
909d3787
KW
2372 char *loc = savepv(setlocale(LC_NUMERIC, NULL));
2373 setlocale(LC_NUMERIC, "C");
8d08d9ba 2374#endif
6154021b 2375 s = scan_num(s, &pl_yylval);
8d08d9ba
DG
2376#ifdef USE_LOCALE_NUMERIC
2377 setlocale(LC_NUMERIC, loc);
909d3787 2378 Safefree(loc);
8d08d9ba 2379#endif
6154021b 2380 version = pl_yylval.opval;
dd629d5b
GS
2381 ver = cSVOPx(version)->op_sv;
2382 if (SvPOK(ver) && !SvNIOK(ver)) {
862a34c6 2383 SvUPGRADE(ver, SVt_PVNV);
9d6ce603 2384 SvNV_set(ver, str_to_version(ver));
1571675a 2385 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 2386 }
89bfa8cd 2387 }
5db06880
NC
2388 else if (guessing) {
2389#ifdef PERL_MAD
2390 if (PL_madskills) {
cd81e915
NC
2391 sv_free(PL_nextwhite); /* let next token collect whitespace */
2392 PL_nextwhite = 0;
5db06880
NC
2393 s = SvPVX(PL_linestr) + startoff;
2394 }
2395#endif
e759cc13 2396 return s;
5db06880 2397 }
89bfa8cd
PP
2398 }
2399
5db06880
NC
2400#ifdef PERL_MAD
2401 if (PL_madskills && !version) {
cd81e915
NC
2402 sv_free(PL_nextwhite); /* let next token collect whitespace */
2403 PL_nextwhite = 0;
5db06880
NC
2404 s = SvPVX(PL_linestr) + startoff;
2405 }
2406#endif
89bfa8cd 2407 /* NOTE: The parser sees the package name and the VERSION swapped */
cd81e915 2408 start_force(PL_curforce);
9ded7720 2409 NEXTVAL_NEXTTOKE.opval = version;
4e553d73 2410 force_next(WORD);
89bfa8cd 2411
e759cc13 2412 return s;
89bfa8cd
PP
2413}
2414
ffb4593c 2415/*
91152fc1
DG
2416 * S_force_strict_version
2417 * Forces the next token to be a version number using strict syntax rules.
2418 */
2419
2420STATIC char *
2421S_force_strict_version(pTHX_ char *s)
2422{
2423 dVAR;
2424 OP *version = NULL;
2425#ifdef PERL_MAD
2426 I32 startoff = s - SvPVX(PL_linestr);
2427#endif
2428 const char *errstr = NULL;
2429
2430 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2431
2432 while (isSPACE(*s)) /* leading whitespace */
2433 s++;
2434
2435 if (is_STRICT_VERSION(s,&errstr)) {
2436 SV *ver = newSV(0);
2437 s = (char *)scan_version(s, ver, 0);
2438 version = newSVOP(OP_CONST, 0, ver);
2439 }
4e4da3ac
Z
2440 else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2441 (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2442 {
91152fc1
DG
2443 PL_bufptr = s;
2444 if (errstr)
2445 yyerror(errstr); /* version required */
2446 return s;
2447 }
2448
2449#ifdef PERL_MAD
2450 if (PL_madskills && !version) {
2451 sv_free(PL_nextwhite); /* let next token collect whitespace */
2452 PL_nextwhite = 0;
2453 s = SvPVX(PL_linestr) + startoff;
2454 }
2455#endif
2456 /* NOTE: The parser sees the package name and the VERSION swapped */
2457 start_force(PL_curforce);
2458 NEXTVAL_NEXTTOKE.opval = version;
2459 force_next(WORD);
2460
2461 return s;
2462}
2463
2464/*
ffb4593c
NT
2465 * S_tokeq
2466 * Tokenize a quoted string passed in as an SV. It finds the next
2467 * chunk, up to end of string or a backslash. It may make a new
2468 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2469 * turns \\ into \.
2470 */
2471
76e3520e 2472STATIC SV *
cea2e8a9 2473S_tokeq(pTHX_ SV *sv)
79072805 2474{
97aff369 2475 dVAR;
eb578fdb
KW
2476 char *s;
2477 char *send;
2478 char *d;
b3ac6de7
IZ
2479 STRLEN len = 0;
2480 SV *pv = sv;
79072805 2481
7918f24d
NC
2482 PERL_ARGS_ASSERT_TOKEQ;
2483
79072805 2484 if (!SvLEN(sv))
b3ac6de7 2485 goto finish;
79072805 2486
a0d0e21e 2487 s = SvPV_force(sv, len);
21a311ee 2488 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 2489 goto finish;
463ee0b2 2490 send = s + len;
dcb21ed6
NC
2491 /* This is relying on the SV being "well formed" with a trailing '\0' */
2492 while (s < send && !(*s == '\\' && s[1] == '\\'))
79072805
LW
2493 s++;
2494 if (s == send)
b3ac6de7 2495 goto finish;
79072805 2496 d = s;
be4731d2 2497 if ( PL_hints & HINT_NEW_STRING ) {
59cd0e26 2498 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
be4731d2 2499 }
79072805
LW
2500 while (s < send) {
2501 if (*s == '\\') {
a0d0e21e 2502 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
2503 s++; /* all that, just for this */
2504 }
2505 *d++ = *s++;
2506 }
2507 *d = '\0';
95a20fc0 2508 SvCUR_set(sv, d - SvPVX_const(sv));
b3ac6de7 2509 finish:
3280af22 2510 if ( PL_hints & HINT_NEW_STRING )
eb0d8d16 2511 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
79072805
LW
2512 return sv;
2513}
2514
ffb4593c
NT
2515/*
2516 * Now come three functions related to double-quote context,
2517 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2518 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2519 * interact with PL_lex_state, and create fake ( ... ) argument lists
2520 * to handle functions and concatenation.
ecd24171
DM
2521 * For example,
2522 * "foo\lbar"
2523 * is tokenised as
2524 * stringify ( const[foo] concat lcfirst ( const[bar] ) )
ffb4593c
NT
2525 */
2526
2527/*
2528 * S_sublex_start
6154021b 2529 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
ffb4593c
NT
2530 *
2531 * Pattern matching will set PL_lex_op to the pattern-matching op to
6154021b 2532 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
ffb4593c
NT
2533 *
2534 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2535 *
2536 * Everything else becomes a FUNC.
2537 *
2538 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2539 * had an OP_CONST or OP_READLINE). This just sets us up for a
2540 * call to S_sublex_push().
2541 */
2542
76e3520e 2543STATIC I32
cea2e8a9 2544S_sublex_start(pTHX)
79072805 2545{
97aff369 2546 dVAR;
eb578fdb 2547 const I32 op_type = pl_yylval.ival;
79072805
LW
2548
2549 if (op_type == OP_NULL) {
6154021b 2550 pl_yylval.opval = PL_lex_op;
5f66b61c 2551 PL_lex_op = NULL;
79072805
LW
2552 return THING;
2553 }
2554 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 2555 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
2556
2557 if (SvTYPE(sv) == SVt_PVIV) {
2558 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2559 STRLEN len;
96a5add6 2560 const char * const p = SvPV_const(sv, len);
740cce10 2561 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
b3ac6de7
IZ
2562 SvREFCNT_dec(sv);
2563 sv = nsv;
4e553d73 2564 }
6154021b 2565 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
a0714e2c 2566 PL_lex_stuff = NULL;
6f33ba73
RGS
2567 /* Allow <FH> // "foo" */
2568 if (op_type == OP_READLINE)
2569 PL_expect = XTERMORDORDOR;
79072805
LW
2570 return THING;
2571 }
e3f73d4e
RGS
2572 else if (op_type == OP_BACKTICK && PL_lex_op) {
2573 /* readpipe() vas overriden */
2574 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
6154021b 2575 pl_yylval.opval = PL_lex_op;
9b201d7d 2576 PL_lex_op = NULL;
e3f73d4e
RGS
2577 PL_lex_stuff = NULL;
2578 return THING;
2579 }
79072805 2580
3280af22 2581 PL_sublex_info.super_state = PL_lex_state;
eac04b2e 2582 PL_sublex_info.sub_inwhat = (U16)op_type;
3280af22
NIS
2583 PL_sublex_info.sub_op = PL_lex_op;
2584 PL_lex_state = LEX_INTERPPUSH;
55497cff 2585
3280af22
NIS
2586 PL_expect = XTERM;
2587 if (PL_lex_op) {
6154021b 2588 pl_yylval.opval = PL_lex_op;
5f66b61c 2589 PL_lex_op = NULL;
55497cff
PP
2590 return PMFUNC;
2591 }
2592 else
2593 return FUNC;
2594}
2595
ffb4593c
NT
2596/*
2597 * S_sublex_push
2598 * Create a new scope to save the lexing state. The scope will be
2599 * ended in S_sublex_done. Returns a '(', starting the function arguments
2600 * to the uc, lc, etc. found before.
2601 * Sets PL_lex_state to LEX_INTERPCONCAT.
2602 */
2603
76e3520e 2604STATIC I32
cea2e8a9 2605S_sublex_push(pTHX)
55497cff 2606{
27da23d5 2607 dVAR;
78a635de 2608 LEXSHARED *shared;
f46d017c 2609 ENTER;
55497cff 2610
3280af22 2611 PL_lex_state = PL_sublex_info.super_state;
651b5b28 2612 SAVEBOOL(PL_lex_dojoin);
3280af22 2613 SAVEI32(PL_lex_brackets);
78cdf107 2614 SAVEI32(PL_lex_allbrackets);
b27dce25 2615 SAVEI32(PL_lex_formbrack);
78cdf107 2616 SAVEI8(PL_lex_fakeeof);
3280af22
NIS
2617 SAVEI32(PL_lex_casemods);
2618 SAVEI32(PL_lex_starts);
651b5b28 2619 SAVEI8(PL_lex_state);
7cc34111 2620 SAVESPTR(PL_lex_repl);
7766f137 2621 SAVEVPTR(PL_lex_inpat);
98246f1e 2622 SAVEI16(PL_lex_inwhat);
57843af0 2623 SAVECOPLINE(PL_curcop);
3280af22 2624 SAVEPPTR(PL_bufptr);
8452ff4b 2625 SAVEPPTR(PL_bufend);
3280af22
NIS
2626 SAVEPPTR(PL_oldbufptr);
2627 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
2628 SAVEPPTR(PL_last_lop);
2629 SAVEPPTR(PL_last_uni);
3280af22
NIS
2630 SAVEPPTR(PL_linestart);
2631 SAVESPTR(PL_linestr);
8edd5f42
RGS
2632 SAVEGENERICPV(PL_lex_brackstack);
2633 SAVEGENERICPV(PL_lex_casestack);
78a635de 2634 SAVEGENERICPV(PL_parser->lex_shared);
3a54fd60 2635 SAVEBOOL(PL_parser->lex_re_reparsing);
3280af22 2636
99bd9d90 2637 /* The here-doc parser needs to be able to peek into outer lexing
60f40a38
FC
2638 scopes to find the body of the here-doc. So we put PL_linestr and
2639 PL_bufptr into lex_shared, to ‘share’ those values.
99bd9d90 2640 */
60f40a38
FC
2641 PL_parser->lex_shared->ls_linestr = PL_linestr;
2642 PL_parser->lex_shared->ls_bufptr = PL_bufptr;
99bd9d90 2643
3280af22 2644 PL_linestr = PL_lex_stuff;
7cc34111 2645 PL_lex_repl = PL_sublex_info.repl;
a0714e2c 2646 PL_lex_stuff = NULL;
7cc34111 2647 PL_sublex_info.repl = NULL;
3280af22 2648
9cbb5ea2
GS
2649 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2650 = SvPVX(PL_linestr);
3280af22 2651 PL_bufend += SvCUR(PL_linestr);
bd61b366 2652 PL_last_lop = PL_last_uni = NULL;
3280af22 2653 SAVEFREESV(PL_linestr);
4dc843bc 2654 if (PL_lex_repl) SAVEFREESV(PL_lex_repl);
3280af22
NIS
2655
2656 PL_lex_dojoin = FALSE;
b27dce25 2657 PL_lex_brackets = PL_lex_formbrack = 0;
78cdf107
Z
2658 PL_lex_allbrackets = 0;
2659 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
a02a5408
JC
2660 Newx(PL_lex_brackstack, 120, char);
2661 Newx(PL_lex_casestack, 12, char);
3280af22
NIS
2662 PL_lex_casemods = 0;
2663 *PL_lex_casestack = '\0';
2664 PL_lex_starts = 0;
2665 PL_lex_state = LEX_INTERPCONCAT;
eb160463 2666 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
78a635de
FC
2667
2668 Newxz(shared, 1, LEXSHARED);
2669 shared->ls_prev = PL_parser->lex_shared;
2670 PL_parser->lex_shared = shared;
3280af22
NIS
2671
2672 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
bb16bae8 2673 if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
3280af22
NIS
2674 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2675 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 2676 else
5f66b61c 2677 PL_lex_inpat = NULL;
79072805 2678
3a54fd60
DM
2679 PL_parser->lex_re_reparsing = cBOOL(PL_in_eval & EVAL_RE_REPARSING);
2680 PL_in_eval &= ~EVAL_RE_REPARSING;
2681
55497cff 2682 return '(';
79072805
LW
2683}
2684
ffb4593c
NT
2685/*
2686 * S_sublex_done
2687 * Restores lexer state after a S_sublex_push.
2688 */
2689
76e3520e 2690STATIC I32
cea2e8a9 2691S_sublex_done(pTHX)
79072805 2692{
27da23d5 2693 dVAR;
3280af22 2694 if (!PL_lex_starts++) {
396482e1 2695 SV * const sv = newSVpvs("");
9aa983d2
JH
2696 if (SvUTF8(PL_linestr))
2697 SvUTF8_on(sv);
3280af22 2698 PL_expect = XOPERATOR;
6154021b 2699 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
2700 return THING;
2701 }
2702
3280af22
NIS
2703 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2704 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 2705 return yylex();
79072805
LW
2706 }
2707
ffb4593c 2708 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
bb16bae8 2709 assert(PL_lex_inwhat != OP_TRANSR);
3280af22
NIS
2710 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2711 PL_linestr = PL_lex_repl;
2712 PL_lex_inpat = 0;
2713 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2714 PL_bufend += SvCUR(PL_linestr);
bd61b366 2715 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2716 PL_lex_dojoin = FALSE;
2717 PL_lex_brackets = 0;
78cdf107
Z
2718 PL_lex_allbrackets = 0;
2719 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
3280af22
NIS
2720 PL_lex_casemods = 0;
2721 *PL_lex_casestack = '\0';
2722 PL_lex_starts = 0;
25da4f38 2723 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
2724 PL_lex_state = LEX_INTERPNORMAL;
2725 PL_lex_starts++;
e9fa98b2
HS
2726 /* we don't clear PL_lex_repl here, so that we can check later
2727 whether this is an evalled subst; that means we rely on the
2728 logic to ensure sublex_done() is called again only via the
2729 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 2730 }
e9fa98b2 2731 else {
3280af22 2732 PL_lex_state = LEX_INTERPCONCAT;
a0714e2c 2733 PL_lex_repl = NULL;
e9fa98b2 2734 }
79072805 2735 return ',';
ffed7fef
LW
2736 }
2737 else {
5db06880
NC
2738#ifdef PERL_MAD
2739 if (PL_madskills) {
cd81e915
NC
2740 if (PL_thiswhite) {
2741 if (!PL_endwhite)
6b29d1f5 2742 PL_endwhite = newSVpvs("");
cd81e915
NC
2743 sv_catsv(PL_endwhite, PL_thiswhite);
2744 PL_thiswhite = 0;
2745 }
2746 if (PL_thistoken)
76f68e9b 2747 sv_setpvs(PL_thistoken,"");
5db06880 2748 else
cd81e915 2749 PL_realtokenstart = -1;
5db06880
NC
2750 }
2751#endif
f46d017c 2752 LEAVE;
3280af22
NIS
2753 PL_bufend = SvPVX(PL_linestr);
2754 PL_bufend += SvCUR(PL_linestr);
2755 PL_expect = XOPERATOR;
09bef843 2756 PL_sublex_info.sub_inwhat = 0;
79072805 2757 return ')';
ffed7fef
LW
2758 }
2759}
2760
6f613c73
KW
2761PERL_STATIC_INLINE SV*
2762S_get_and_check_backslash_N_name(pTHX_ const char* s, const char* const e)
2763{
140b12ad
KW
2764 /* <s> points to first character of interior of \N{}, <e> to one beyond the
2765 * interior, hence to the "}". Finds what the name resolves to, returning
2766 * an SV* containing it; NULL if no valid one found */
2767
dd2b1b72 2768 SV* res = newSVpvn_flags(s, e - s, UTF ? SVf_UTF8 : 0);
6f613c73 2769
0c415a79
KW
2770 HV * table;
2771 SV **cvp;
2772 SV *cv;
2773 SV *rv;
2774 HV *stash;
2775 const U8* first_bad_char_loc;
2776 const char* backslash_ptr = s - 3; /* Points to the <\> of \N{... */
2777
6f613c73
KW
2778 PERL_ARGS_ASSERT_GET_AND_CHECK_BACKSLASH_N_NAME;
2779
107160e2
KW
2780 if (UTF && ! is_utf8_string_loc((U8 *) backslash_ptr,
2781 e - backslash_ptr,
2782 &first_bad_char_loc))
2783 {
2784 /* If warnings are on, this will print a more detailed analysis of what
2785 * is wrong than the error message below */
2786 utf8n_to_uvuni(first_bad_char_loc,
2787 e - ((char *) first_bad_char_loc),
2788 NULL, 0);
2789
2790 /* We deliberately don't try to print the malformed character, which
2791 * might not print very well; it also may be just the first of many
2792 * malformations, so don't print what comes after it */
2793 yyerror(Perl_form(aTHX_
2794 "Malformed UTF-8 character immediately after '%.*s'",
2795 (int) (first_bad_char_loc - (U8 *) backslash_ptr), backslash_ptr));
2796 return NULL;
2797 }
2798
2799 res = new_constant( NULL, 0, "charnames", res, NULL, backslash_ptr,
2800 /* include the <}> */
2801 e - backslash_ptr + 1);
6f613c73 2802 if (! SvPOK(res)) {
b6407c49 2803 SvREFCNT_dec_NN(res);
6f613c73
KW
2804 return NULL;
2805 }
2806
0c415a79
KW
2807 /* See if the charnames handler is the Perl core's, and if so, we can skip
2808 * the validation needed for a user-supplied one, as Perl's does its own
2809 * validation. */
2810 table = GvHV(PL_hintgv); /* ^H */
2811 cvp = hv_fetchs(table, "charnames", FALSE);
67a057d6
FC
2812 if (cvp && (cv = *cvp) && SvROK(cv) && ((rv = SvRV(cv)) != NULL)
2813 && SvTYPE(rv) == SVt_PVCV && ((stash = CvSTASH(rv)) != NULL))
0c415a79
KW
2814 {
2815 const char * const name = HvNAME(stash);
2816 if strEQ(name, "_charnames") {
2817 return res;
2818 }
2819 }
2820
bde9e88d
KW
2821 /* Here, it isn't Perl's charname handler. We can't rely on a
2822 * user-supplied handler to validate the input name. For non-ut8 input,
2823 * look to see that the first character is legal. Then loop through the
2824 * rest checking that each is a continuation */
6f613c73 2825
b6ba1137
KW
2826 /* This code needs to be sync'ed with a regex in _charnames.pm which does
2827 * the same thing */
2828
b6ba1137 2829 if (! UTF) {
bde9e88d 2830 if (! isALPHAU(*s)) {
b6ba1137
KW
2831 goto bad_charname;
2832 }
bde9e88d
KW
2833 s++;
2834 while (s < e) {
2835 if (! isCHARNAME_CONT(*s)) {
b6ba1137
KW
2836 goto bad_charname;
2837 }
3036c853 2838 if (*s == ' ' && *(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2839 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2840 "A sequence of multiple spaces in a charnames "
2841 "alias definition is deprecated");
bd299e29 2842 }
bde9e88d 2843 s++;
b6ba1137 2844 }
3036c853 2845 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2846 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2847 "Trailing white-space in a charnames alias "
2848 "definition is deprecated");
bd299e29 2849 }
b6ba1137
KW
2850 }
2851 else {
bde9e88d
KW
2852 /* Similarly for utf8. For invariants can check directly; for other
2853 * Latin1, can calculate their code point and check; otherwise use a
2854 * swash */
2855 if (UTF8_IS_INVARIANT(*s)) {
2856 if (! isALPHAU(*s)) {
140b12ad
KW
2857 goto bad_charname;
2858 }
bde9e88d
KW
2859 s++;
2860 } else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
2861 if (! isALPHAU(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s, *(s+1))))) {
b6ba1137 2862 goto bad_charname;
6f613c73 2863 }
bde9e88d 2864 s += 2;
6f613c73 2865 }
bde9e88d
KW
2866 else {
2867 if (! PL_utf8_charname_begin) {
2868 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2869 PL_utf8_charname_begin = _core_swash_init("utf8",
2870 "_Perl_Charname_Begin",
2871 &PL_sv_undef,
2872 1, 0, NULL, &flags);
2873 }
2874 if (! swash_fetch(PL_utf8_charname_begin, (U8 *) s, TRUE)) {
2875 goto bad_charname;
2876 }
2877 s += UTF8SKIP(s);
2878 }
2879
2880 while (s < e) {
2881 if (UTF8_IS_INVARIANT(*s)) {
2882 if (! isCHARNAME_CONT(*s)) {
2883 goto bad_charname;
2884 }
3036c853
FC
2885 if (*s == ' ' && *(s-1) == ' '
2886 && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2887 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2888 "A sequence of multiple spaces in a charnam"
2889 "es alias definition is deprecated");
bd299e29 2890 }
bde9e88d
KW
2891 s++;
2892 }
2893 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
2894 if (! isCHARNAME_CONT(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s,
2895 *(s+1)))))
2896 {
2897 goto bad_charname;
2898 }
2899 s += 2;
2900 }
2901 else {
2902 if (! PL_utf8_charname_continue) {
2903 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2904 PL_utf8_charname_continue = _core_swash_init("utf8",
2905 "_Perl_Charname_Continue",
2906 &PL_sv_undef,
2907 1, 0, NULL, &flags);
2908 }
2909 if (! swash_fetch(PL_utf8_charname_continue, (U8 *) s, TRUE)) {
2910 goto bad_charname;
2911 }
2912 s += UTF8SKIP(s);
6f613c73
KW
2913 }
2914 }
3036c853 2915 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2916 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2917 "Trailing white-space in a charnames alias "
2918 "definition is deprecated");
bd299e29 2919 }
6f613c73
KW
2920 }
2921
94ca1619 2922 if (SvUTF8(res)) { /* Don't accept malformed input */
bde9e88d
KW
2923 const U8* first_bad_char_loc;
2924 STRLEN len;
2925 const char* const str = SvPV_const(res, len);
2926 if (! is_utf8_string_loc((U8 *) str, len, &first_bad_char_loc)) {
2927 /* If warnings are on, this will print a more detailed analysis of
2928 * what is wrong than the error message below */
2929 utf8n_to_uvuni(first_bad_char_loc,
2930 (char *) first_bad_char_loc - str,
2931 NULL, 0);
2932
2933 /* We deliberately don't try to print the malformed character,
2934 * which might not print very well; it also may be just the first
2935 * of many malformations, so don't print what comes after it */
2936 yyerror_pv(
2937 Perl_form(aTHX_
2938 "Malformed UTF-8 returned by %.*s immediately after '%.*s'",
2939 (int) (e - backslash_ptr + 1), backslash_ptr,
2940 (int) ((char *) first_bad_char_loc - str), str
2941 ),
2942 SVf_UTF8);
2943 return NULL;
2944 }
2945 }
140b12ad 2946
bde9e88d 2947 return res;
140b12ad 2948
bde9e88d
KW
2949 bad_charname: {
2950 int bad_char_size = ((UTF) ? UTF8SKIP(s) : 1);
2951
2952 /* The final %.*s makes sure that should the trailing NUL be missing
2953 * that this print won't run off the end of the string */
2954 yyerror_pv(
2955 Perl_form(aTHX_
2956 "Invalid character in \\N{...}; marked by <-- HERE in %.*s<-- HERE %.*s",
2957 (int)(s - backslash_ptr + bad_char_size), backslash_ptr,
2958 (int)(e - s + bad_char_size), s + bad_char_size
2959 ),
2960 UTF ? SVf_UTF8 : 0);
2961 return NULL;
2962 }
6f613c73
KW
2963}
2964
02aa26ce
NT
2965/*
2966 scan_const
2967
9da1dd8f
DM
2968 Extracts the next constant part of a pattern, double-quoted string,
2969 or transliteration. This is terrifying code.
2970
2971 For example, in parsing the double-quoted string "ab\x63$d", it would
2972 stop at the '$' and return an OP_CONST containing 'abc'.
02aa26ce 2973
94def140 2974 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
3280af22 2975 processing a pattern (PL_lex_inpat is true), a transliteration
94def140 2976 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
02aa26ce 2977
94def140
ST
2978 Returns a pointer to the character scanned up to. If this is
2979 advanced from the start pointer supplied (i.e. if anything was
9da1dd8f 2980 successfully parsed), will leave an OP_CONST for the substring scanned
6154021b 2981 in pl_yylval. Caller must intuit reason for not parsing further
9b599b2a
GS
2982 by looking at the next characters herself.
2983
02aa26ce 2984 In patterns:
9da1dd8f 2985 expand:
537124e4
KW
2986 \N{FOO} => \N{U+hex_for_character_FOO}
2987 (if FOO expands to multiple characters, expands to \N{U+xx.XX.yy ...})
9da1dd8f
DM
2988
2989 pass through:
2990 all other \-char, including \N and \N{ apart from \N{ABC}
2991
2992 stops on:
2993 @ and $ where it appears to be a var, but not for $ as tail anchor
2994 \l \L \u \U \Q \E
2995 (?{ or (??{
2996
02aa26ce
NT
2997
2998 In transliterations:
2999 characters are VERY literal, except for - not at the start or end
94def140
ST
3000 of the string, which indicates a range. If the range is in bytes,
3001 scan_const expands the range to the full set of intermediate
3002 characters. If the range is in utf8, the hyphen is replaced with
3003 a certain range mark which will be handled by pmtrans() in op.c.
02aa26ce
NT
3004
3005 In double-quoted strings:
3006 backslashes:
3007 double-quoted style: \r and \n
ff3f963a 3008 constants: \x31, etc.
94def140 3009 deprecated backrefs: \1 (in substitution replacements)
02aa26ce
NT
3010 case and quoting: \U \Q \E
3011 stops on @ and $
3012
3013 scan_const does *not* construct ops to handle interpolated strings.
3014 It stops processing as soon as it finds an embedded $ or @ variable
3015 and leaves it to the caller to work out what's going on.
3016
94def140
ST
3017 embedded arrays (whether in pattern or not) could be:
3018 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
3019
3020 $ in double-quoted strings must be the symbol of an embedded scalar.
02aa26ce
NT
3021
3022 $ in pattern could be $foo or could be tail anchor. Assumption:
3023 it's a tail anchor if $ is the last thing in the string, or if it's
94def140 3024 followed by one of "()| \r\n\t"
02aa26ce 3025
9da1dd8f 3026 \1 (backreferences) are turned into $1 in substitutions
02aa26ce
NT
3027
3028 The structure of the code is
3029 while (there's a character to process) {
94def140
ST
3030 handle transliteration ranges
3031 skip regexp comments /(?#comment)/ and codes /(?{code})/
3032 skip #-initiated comments in //x patterns
3033 check for embedded arrays
02aa26ce
NT
3034 check for embedded scalars
3035 if (backslash) {
94def140 3036 deprecate \1 in substitution replacements
02aa26ce
NT
3037 handle string-changing backslashes \l \U \Q \E, etc.
3038 switch (what was escaped) {
94def140 3039 handle \- in a transliteration (becomes a literal -)
ff3f963a 3040 if a pattern and not \N{, go treat as regular character
94def140
ST
3041 handle \132 (octal characters)
3042 handle \x15 and \x{1234} (hex characters)
ff3f963a 3043 handle \N{name} (named characters, also \N{3,5} in a pattern)
94def140
ST
3044 handle \cV (control characters)
3045 handle printf-style backslashes (\f, \r, \n, etc)
02aa26ce 3046 } (end switch)
77a135fe 3047 continue
02aa26ce 3048 } (end if backslash)
77a135fe 3049 handle regular character
02aa26ce 3050 } (end while character to read)
4e553d73 3051
02aa26ce
NT
3052*/
3053
76e3520e 3054STATIC char *
cea2e8a9 3055S_scan_const(pTHX_ char *start)
79072805 3056{
97aff369 3057 dVAR;
eb578fdb 3058 char *send = PL_bufend; /* end of the constant */
77a135fe
KW
3059 SV *sv = newSV(send - start); /* sv for the constant. See
3060 note below on sizing. */
eb578fdb
KW
3061 char *s = start; /* start of the constant */
3062 char *d = SvPVX(sv); /* destination for copies */
02aa26ce 3063 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 3064 bool didrange = FALSE; /* did we just finish a range? */
2866decb 3065 bool in_charclass = FALSE; /* within /[...]/ */
b953e60c
KW
3066 bool has_utf8 = FALSE; /* Output constant is UTF8 */
3067 bool this_utf8 = cBOOL(UTF); /* Is the source string assumed
77a135fe
KW
3068 to be UTF8? But, this can
3069 show as true when the source
3070 isn't utf8, as for example
3071 when it is entirely composed
3072 of hex constants */
6f613c73 3073 SV *res; /* result from charnames */
77a135fe
KW
3074
3075 /* Note on sizing: The scanned constant is placed into sv, which is
3076 * initialized by newSV() assuming one byte of output for every byte of
3077 * input. This routine expects newSV() to allocate an extra byte for a
3078 * trailing NUL, which this routine will append if it gets to the end of
3079 * the input. There may be more bytes of input than output (eg., \N{LATIN
3080 * CAPITAL LETTER A}), or more output than input if the constant ends up
3081 * recoded to utf8, but each time a construct is found that might increase
3082 * the needed size, SvGROW() is called. Its size parameter each time is
3083 * based on the best guess estimate at the time, namely the length used so
3084 * far, plus the length the current construct will occupy, plus room for
3085 * the trailing NUL, plus one byte for every input byte still unscanned */
3086
c3320c2a
KW
3087 UV uv = UV_MAX; /* Initialize to weird value to try to catch any uses
3088 before set */
4c3a8340
ST
3089#ifdef EBCDIC
3090 UV literal_endpoint = 0;
e294cc5d 3091 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
4c3a8340 3092#endif
012bcf8d 3093
7918f24d
NC
3094 PERL_ARGS_ASSERT_SCAN_CONST;
3095
bb16bae8 3096 assert(PL_lex_inwhat != OP_TRANSR);
2b9d42f0
NIS
3097 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
3098 /* If we are doing a trans and we know we want UTF8 set expectation */
3099 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
3100 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3101 }
3102
b899e89d
FC
3103 /* Protect sv from errors and fatal warnings. */
3104 ENTER_with_name("scan_const");
3105 SAVEFREESV(sv);
2b9d42f0 3106
79072805 3107 while (s < send || dorange) {
ff3f963a 3108
02aa26ce 3109 /* get transliterations out of the way (they're most literal) */
3280af22 3110 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 3111 /* expand a range A-Z to the full set of characters. AIE! */
79072805 3112 if (dorange) {
1ba5c669
JH
3113 I32 i; /* current expanded character */
3114 I32 min; /* first character in range */
3115 I32 max; /* last character in range */
02aa26ce 3116
e294cc5d
JH
3117#ifdef EBCDIC
3118 UV uvmax = 0;
3119#endif
3120
3121 if (has_utf8
3122#ifdef EBCDIC
3123 && !native_range
3124#endif
1953db30 3125 ) {
9d4ba2ae 3126 char * const c = (char*)utf8_hop((U8*)d, -1);
8973db79
JH
3127 char *e = d++;
3128 while (e-- > c)
3129 *(e + 1) = *e;
25716404 3130 *c = (char)UTF_TO_NATIVE(0xff);
8973db79
JH
3131 /* mark the range as done, and continue */
3132 dorange = FALSE;
3133 didrange = TRUE;
3134 continue;
3135 }
2b9d42f0 3136
95a20fc0 3137 i = d - SvPVX_const(sv); /* remember current offset */
e294cc5d
JH
3138#ifdef EBCDIC
3139 SvGROW(sv,
3140 SvLEN(sv) + (has_utf8 ?
3141 (512 - UTF_CONTINUATION_MARK +
3142 UNISKIP(0x100))
3143 : 256));
3144 /* How many two-byte within 0..255: 128 in UTF-8,
3145 * 96 in UTF-8-mod. */
3146#else
9cbb5ea2 3147 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
e294cc5d 3148#endif
9cbb5ea2 3149 d = SvPVX(sv) + i; /* refresh d after realloc */
e294cc5d
JH
3150#ifdef EBCDIC
3151 if (has_utf8) {
3152 int j;
3153 for (j = 0; j <= 1; j++) {
3154 char * const c = (char*)utf8_hop((U8*)d, -1);
3155 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
3156 if (j)
3157 min = (U8)uv;
3158 else if (uv < 256)
3159 max = (U8)uv;
3160 else {
3161 max = (U8)0xff; /* only to \xff */
3162 uvmax = uv; /* \x{100} to uvmax */
3163 }
3164 d = c; /* eat endpoint chars */
3165 }
3166 }
3167 else {
3168#endif
3169 d -= 2; /* eat the first char and the - */
3170 min = (U8)*d; /* first char in range */
3171 max = (U8)d[1]; /* last char in range */
3172#ifdef EBCDIC
3173 }
3174#endif
8ada0baa 3175
c2e66d9e 3176 if (min > max) {
01ec43d0 3177 Perl_croak(aTHX_
d1573ac7 3178 "Invalid range \"%c-%c\" in transliteration operator",
1ba5c669 3179 (char)min, (char)max);
c2e66d9e
GS
3180 }
3181
c7f1f016 3182#ifdef EBCDIC
4c3a8340
ST
3183 if (literal_endpoint == 2 &&
3184 ((isLOWER(min) && isLOWER(max)) ||
3185 (isUPPER(min) && isUPPER(max)))) {
8ada0baa
JH
3186 if (isLOWER(min)) {
3187 for (i = min; i <= max; i++)
3188 if (isLOWER(i))
db42d148 3189 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
3190 } else {
3191 for (i = min; i <= max; i++)
3192 if (isUPPER(i))
db42d148 3193 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
3194 }
3195 }
3196 else
3197#endif
3198 for (i = min; i <= max; i++)
e294cc5d
JH
3199#ifdef EBCDIC
3200 if (has_utf8) {
3201 const U8 ch = (U8)NATIVE_TO_UTF(i);
3202 if (UNI_IS_INVARIANT(ch))
3203 *d++ = (U8)i;
3204 else {
3205 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
3206 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
3207 }
3208 }
3209 else
3210#endif
3211 *d++ = (char)i;
3212
3213#ifdef EBCDIC
3214 if (uvmax) {
3215 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
3216 if (uvmax > 0x101)
3217 *d++ = (char)UTF_TO_NATIVE(0xff);
3218 if (uvmax > 0x100)
3219 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
3220 }
3221#endif
02aa26ce
NT
3222
3223 /* mark the range as done, and continue */
79072805 3224 dorange = FALSE;
01ec43d0 3225 didrange = TRUE;
4c3a8340
ST
3226#ifdef EBCDIC
3227 literal_endpoint = 0;
3228#endif
79072805 3229 continue;
4e553d73 3230 }
02aa26ce
NT
3231
3232 /* range begins (ignore - as first or last char) */
79072805 3233 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 3234 if (didrange) {
1fafa243 3235 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 3236 }
e294cc5d
JH
3237 if (has_utf8
3238#ifdef EBCDIC
3239 && !native_range
3240#endif
3241 ) {
25716404 3242 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
3243 s++;
3244 continue;
3245 }
79072805
LW
3246 dorange = TRUE;
3247 s++;
01ec43d0
GS
3248 }
3249 else {
3250 didrange = FALSE;
4c3a8340
ST
3251#ifdef EBCDIC
3252 literal_endpoint = 0;
e294cc5d 3253 native_range = TRUE;
4c3a8340 3254#endif
01ec43d0 3255 }
79072805 3256 }
02aa26ce
NT
3257
3258 /* if we get here, we're not doing a transliteration */
3259
e4a2df84
DM
3260 else if (*s == '[' && PL_lex_inpat && !in_charclass) {
3261 char *s1 = s-1;
3262 int esc = 0;
3263 while (s1 >= start && *s1-- == '\\')
3264 esc = !esc;
3265 if (!esc)
3266 in_charclass = TRUE;
3267 }
2866decb 3268
e4a2df84
DM
3269 else if (*s == ']' && PL_lex_inpat && in_charclass) {
3270 char *s1 = s-1;
3271 int esc = 0;
3272 while (s1 >= start && *s1-- == '\\')
3273 esc = !esc;
3274 if (!esc)
3275 in_charclass = FALSE;
3276 }
2866decb 3277
9da1dd8f
DM
3278 /* skip for regexp comments /(?#comment)/, except for the last
3279 * char, which will be done separately.
3280 * Stop on (?{..}) and friends */
3281
3280af22 3282 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395 3283 if (s[2] == '#') {
e994fd66 3284 while (s+1 < send && *s != ')')
db42d148 3285 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
155aba94 3286 }
2866decb 3287 else if (!PL_lex_casemods && !in_charclass &&
d3cec5e5
DM
3288 ( s[2] == '{' /* This should match regcomp.c */
3289 || (s[2] == '?' && s[3] == '{')))
155aba94 3290 {
9da1dd8f 3291 break;
cc6b7395 3292 }
748a9306 3293 }
02aa26ce
NT
3294
3295 /* likewise skip #-initiated comments in //x patterns */
3280af22 3296 else if (*s == '#' && PL_lex_inpat &&
73134a2e 3297 ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
748a9306 3298 while (s+1 < send && *s != '\n')
db42d148 3299 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
748a9306 3300 }
02aa26ce 3301
9da1dd8f
DM
3302 /* no further processing of single-quoted regex */
3303 else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'')
3304 goto default_action;
3305
5d1d4326 3306 /* check for embedded arrays
da6eedaa 3307 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
5d1d4326 3308 */
1749ea0d 3309 else if (*s == '@' && s[1]) {
8a2bca12 3310 if (isWORDCHAR_lazy_if(s+1,UTF))
1749ea0d
ST
3311 break;
3312 if (strchr(":'{$", s[1]))
3313 break;
3314 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
3315 break; /* in regexp, neither @+ nor @- are interpolated */
3316 }
02aa26ce
NT
3317
3318 /* check for embedded scalars. only stop if we're sure it's a
3319 variable.
3320 */
79072805 3321 else if (*s == '$') {
3280af22 3322 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 3323 break;
77772344 3324 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
a2a5de95
NC
3325 if (s[1] == '\\') {
3326 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3327 "Possible unintended interpolation of $\\ in regex");
77772344 3328 }
79072805 3329 break; /* in regexp, $ might be tail anchor */
77772344 3330 }
79072805 3331 }
02aa26ce 3332
2b9d42f0
NIS
3333 /* End of else if chain - OP_TRANS rejoin rest */
3334
02aa26ce 3335 /* backslashes */
79072805 3336 if (*s == '\\' && s+1 < send) {
ff3f963a
KW
3337 char* e; /* Can be used for ending '}', etc. */
3338
79072805 3339 s++;
02aa26ce 3340
7d0fc23c
KW
3341 /* warn on \1 - \9 in substitution replacements, but note that \11
3342 * is an octal; and \19 is \1 followed by '9' */
3280af22 3343 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 3344 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 3345 {
a2a5de95 3346 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
79072805
LW
3347 *--s = '$';
3348 break;
3349 }
02aa26ce
NT
3350
3351 /* string-change backslash escapes */
838f2281 3352 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) {
79072805
LW
3353 --s;
3354 break;
3355 }
ff3f963a
KW
3356 /* In a pattern, process \N, but skip any other backslash escapes.
3357 * This is because we don't want to translate an escape sequence
3358 * into a meta symbol and have the regex compiler use the meta
3359 * symbol meaning, e.g. \x{2E} would be confused with a dot. But
3360 * in spite of this, we do have to process \N here while the proper
3361 * charnames handler is in scope. See bugs #56444 and #62056.
3362 * There is a complication because \N in a pattern may also stand
3363 * for 'match a non-nl', and not mean a charname, in which case its
3364 * processing should be deferred to the regex compiler. To be a
3365 * charname it must be followed immediately by a '{', and not look
3366 * like \N followed by a curly quantifier, i.e., not something like
3367 * \N{3,}. regcurly returns a boolean indicating if it is a legal
3368 * quantifier */
3369 else if (PL_lex_inpat
3370 && (*s != 'N'
3371 || s[1] != '{'
4d68ffa0 3372 || regcurly(s + 1, FALSE)))
ff3f963a 3373 {
cc74c5bd
ST
3374 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
3375 goto default_action;
3376 }
02aa26ce 3377
79072805 3378 switch (*s) {
02aa26ce
NT
3379
3380 /* quoted - in transliterations */
79072805 3381 case '-':
3280af22 3382 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
3383 *d++ = *s++;
3384 continue;
3385 }
3386 /* FALL THROUGH */
3387 default:
11b8faa4 3388 {
15861f94 3389 if ((isALPHANUMERIC(*s)))
a2a5de95
NC
3390 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
3391 "Unrecognized escape \\%c passed through",
3392 *s);
11b8faa4 3393 /* default action is to copy the quoted character */
f9a63242 3394 goto default_action;
11b8faa4 3395 }
02aa26ce 3396
632403cc 3397 /* eg. \132 indicates the octal constant 0132 */
79072805
LW
3398 case '0': case '1': case '2': case '3':
3399 case '4': case '5': case '6': case '7':
ba210ebe 3400 {
5e0a247b 3401 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
53305cf1 3402 STRLEN len = 3;
77a135fe 3403 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
ba210ebe 3404 s += len;
5e0a247b
KW
3405 if (len < 3 && s < send && isDIGIT(*s)
3406 && ckWARN(WARN_MISC))
3407 {
3408 Perl_warner(aTHX_ packWARN(WARN_MISC),
3409 "%s", form_short_octal_warning(s, len));
3410 }
ba210ebe 3411 }
012bcf8d 3412 goto NUM_ESCAPE_INSERT;
02aa26ce 3413
f0a2b745
KW
3414 /* eg. \o{24} indicates the octal constant \024 */
3415 case 'o':
3416 {
454155d9 3417 const char* error;
f0a2b745 3418
00ce5563 3419 bool valid = grok_bslash_o(&s, &uv, &error,
80f4111b
KW
3420 TRUE, /* Output warning */
3421 FALSE, /* Not strict */
17896a85
KW
3422 TRUE, /* Output warnings for
3423 non-portables */
80f4111b 3424 UTF);
454155d9 3425 if (! valid) {
f0a2b745
KW
3426 yyerror(error);
3427 continue;
3428 }
3429 goto NUM_ESCAPE_INSERT;
3430 }
3431
77a135fe 3432 /* eg. \x24 indicates the hex constant 0x24 */
79072805 3433 case 'x':
a0481293 3434 {
a0481293 3435 const char* error;
355860ce 3436
00ce5563 3437 bool valid = grok_bslash_x(&s, &uv, &error,
80f4111b
KW
3438 TRUE, /* Output warning */
3439 FALSE, /* Not strict */
17896a85
KW
3440 TRUE, /* Output warnings for
3441 non-portables */
80f4111b 3442 UTF);
a0481293
KW
3443 if (! valid) {
3444 yyerror(error);
355860ce 3445 continue;
ba210ebe 3446 }
012bcf8d
GS
3447 }
3448
3449 NUM_ESCAPE_INSERT:
ff3f963a
KW
3450 /* Insert oct or hex escaped character. There will always be
3451 * enough room in sv since such escapes will be longer than any
3452 * UTF-8 sequence they can end up as, except if they force us
3453 * to recode the rest of the string into utf8 */
ba7cea30 3454
77a135fe 3455 /* Here uv is the ordinal of the next character being added in
ff3f963a 3456 * unicode (converted from native). */
77a135fe 3457 if (!UNI_IS_INVARIANT(uv)) {
9aa983d2 3458 if (!has_utf8 && uv > 255) {
77a135fe
KW
3459 /* Might need to recode whatever we have accumulated so
3460 * far if it contains any chars variant in utf8 or
3461 * utf-ebcdic. */
3462
3463 SvCUR_set(sv, d - SvPVX_const(sv));
3464 SvPOK_on(sv);
3465 *d = '\0';
77a135fe 3466 /* See Note on sizing above. */
7bf79863
KW
3467 sv_utf8_upgrade_flags_grow(sv,
3468 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3469 UNISKIP(uv) + (STRLEN)(send - s) + 1);
77a135fe
KW
3470 d = SvPVX(sv) + SvCUR(sv);
3471 has_utf8 = TRUE;
012bcf8d
GS
3472 }
3473
77a135fe
KW
3474 if (has_utf8) {
3475 d = (char*)uvuni_to_utf8((U8*)d, uv);
f9a63242
JH
3476 if (PL_lex_inwhat == OP_TRANS &&
3477 PL_sublex_info.sub_op) {
3478 PL_sublex_info.sub_op->op_private |=
3479 (PL_lex_repl ? OPpTRANS_FROM_UTF
3480 : OPpTRANS_TO_UTF);
f9a63242 3481 }
e294cc5d
JH
3482#ifdef EBCDIC
3483 if (uv > 255 && !dorange)
3484 native_range = FALSE;
3485#endif
012bcf8d 3486 }
a0ed51b3 3487 else {
012bcf8d 3488 *d++ = (char)uv;
a0ed51b3 3489 }
012bcf8d
GS
3490 }
3491 else {
c4d5f83a 3492 *d++ = (char) uv;
a0ed51b3 3493 }
79072805 3494 continue;
02aa26ce 3495
4a2d328f 3496 case 'N':
ff3f963a
KW
3497 /* In a non-pattern \N must be a named character, like \N{LATIN
3498 * SMALL LETTER A} or \N{U+0041}. For patterns, it also can
3499 * mean to match a non-newline. For non-patterns, named
3500 * characters are converted to their string equivalents. In
3501 * patterns, named characters are not converted to their
3502 * ultimate forms for the same reasons that other escapes
3503 * aren't. Instead, they are converted to the \N{U+...} form
3504 * to get the value from the charnames that is in effect right
3505 * now, while preserving the fact that it was a named character
3506 * so that the regex compiler knows this */
3507
3508 /* This section of code doesn't generally use the
3509 * NATIVE_TO_NEED() macro to transform the input. I (khw) did
3510 * a close examination of this macro and determined it is a
3511 * no-op except on utfebcdic variant characters. Every
3512 * character generated by this that would normally need to be
3513 * enclosed by this macro is invariant, so the macro is not
7538f724
KW
3514 * needed, and would complicate use of copy(). XXX There are
3515 * other parts of this file where the macro is used
3516 * inconsistently, but are saved by it being a no-op */
ff3f963a
KW
3517
3518 /* The structure of this section of code (besides checking for
3519 * errors and upgrading to utf8) is:
3520 * Further disambiguate between the two meanings of \N, and if
3521 * not a charname, go process it elsewhere
0a96133f
KW
3522 * If of form \N{U+...}, pass it through if a pattern;
3523 * otherwise convert to utf8
3524 * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3525 * pattern; otherwise convert to utf8 */
ff3f963a
KW
3526
3527 /* Here, s points to the 'N'; the test below is guaranteed to
3528 * succeed if we are being called on a pattern as we already
3529 * know from a test above that the next character is a '{'.
3530 * On a non-pattern \N must mean 'named sequence, which
3531 * requires braces */
3532 s++;
3533 if (*s != '{') {
3534 yyerror("Missing braces on \\N{}");
3535 continue;
3536 }
3537 s++;
3538
0a96133f 3539 /* If there is no matching '}', it is an error. */
ff3f963a
KW
3540 if (! (e = strchr(s, '}'))) {
3541 if (! PL_lex_inpat) {
5777a3f7 3542 yyerror("Missing right brace on \\N{}");
0a96133f
KW
3543 } else {
3544 yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
dbc0d4f2 3545 }
0a96133f 3546 continue;
ff3f963a 3547 }
cddc7ef4 3548
ff3f963a 3549 /* Here it looks like a named character */
cddc7ef4 3550
ff3f963a
KW
3551 if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3552 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3553 | PERL_SCAN_DISALLOW_PREFIX;
3554 STRLEN len;
3555
3556 /* For \N{U+...}, the '...' is a unicode value even on
3557 * EBCDIC machines */
3558 s += 2; /* Skip to next char after the 'U+' */
3559 len = e - s;
3560 uv = grok_hex(s, &len, &flags, NULL);
3561 if (len == 0 || len != (STRLEN)(e - s)) {
3562 yyerror("Invalid hexadecimal number in \\N{U+...}");
3563 s = e + 1;
3564 continue;
3565 }
3566
3567 if (PL_lex_inpat) {
3568
e2a7e165
KW
3569 /* On non-EBCDIC platforms, pass through to the regex
3570 * compiler unchanged. The reason we evaluated the
3571 * number above is to make sure there wasn't a syntax
3572 * error. But on EBCDIC we convert to native so
3573 * downstream code can continue to assume it's native
3574 */
ff3f963a 3575 s -= 5; /* Include the '\N{U+' */
e2a7e165
KW
3576#ifdef EBCDIC
3577 d += my_snprintf(d, e - s + 1 + 1, /* includes the }
3578 and the \0 */
3579 "\\N{U+%X}",
3580 (unsigned int) UNI_TO_NATIVE(uv));
3581#else
ff3f963a
KW
3582 Copy(s, d, e - s + 1, char); /* 1 = include the } */
3583 d += e - s + 1;
e2a7e165 3584#endif
ff3f963a
KW
3585 }
3586 else { /* Not a pattern: convert the hex to string */
3587
3588 /* If destination is not in utf8, unconditionally
3589 * recode it to be so. This is because \N{} implies
3590 * Unicode semantics, and scalars have to be in utf8
3591 * to guarantee those semantics */
3592 if (! has_utf8) {
3593 SvCUR_set(sv, d - SvPVX_const(sv));
3594 SvPOK_on(sv);
3595 *d = '\0';
3596 /* See Note on sizing above. */
3597 sv_utf8_upgrade_flags_grow(
3598 sv,
3599 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3600 UNISKIP(uv) + (STRLEN)(send - e) + 1);
3601 d = SvPVX(sv) + SvCUR(sv);
3602 has_utf8 = TRUE;
3603 }
3604
3605 /* Add the string to the output */
3606 if (UNI_IS_INVARIANT(uv)) {
3607 *d++ = (char) uv;
3608 }
3609 else d = (char*)uvuni_to_utf8((U8*)d, uv);
3610 }
3611 }
6f613c73
KW
3612 else /* Here is \N{NAME} but not \N{U+...}. */
3613 if ((res = get_and_check_backslash_N_name(s, e)))
3614 {
3615 STRLEN len;
3616 const char *str = SvPV_const(res, len);
3617 if (PL_lex_inpat) {
ff3f963a
KW
3618
3619 if (! len) { /* The name resolved to an empty string */
3620 Copy("\\N{}", d, 4, char);
3621 d += 4;
3622 }
3623 else {
3624 /* In order to not lose information for the regex
3625 * compiler, pass the result in the specially made
3626 * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3627 * the code points in hex of each character
3628 * returned by charnames */
3629
3630 const char *str_end = str + len;
3b721c4f 3631 const STRLEN off = d - SvPVX_const(sv);
94ca1619
KW
3632
3633 if (! SvUTF8(res)) {
3634 /* For the non-UTF-8 case, we can determine the
3635 * exact length needed without having to parse
3636 * through the string. Each character takes up
3637 * 2 hex digits plus either a trailing dot or
3638 * the "}" */
3639 d = off + SvGROW(sv, off
3640 + 3 * len
3641 + 6 /* For the "\N{U+", and
3642 trailing NUL */
3643 + (STRLEN)(send - e));
3644 Copy("\\N{U+", d, 5, char);
3645 d += 5;
3646 while (str < str_end) {
3647 char hex_string[4];
3648 my_snprintf(hex_string, sizeof(hex_string),
3649 "%02X.", (U8) *str);
3650 Copy(hex_string, d, 3, char);
3651 d += 3;
3652 str++;
3653 }
3654 d--; /* We will overwrite below the final
3655 dot with a right brace */
3656 }
3657 else {
1953db30
KW
3658 STRLEN char_length; /* cur char's byte length */
3659
3660 /* and the number of bytes after this is
3661 * translated into hex digits */
3662 STRLEN output_length;
3663
3664 /* 2 hex per byte; 2 chars for '\N'; 2 chars
3665 * for max('U+', '.'); and 1 for NUL */
3666 char hex_string[2 * UTF8_MAXBYTES + 5];
3667
3668 /* Get the first character of the result. */
3669 U32 uv = utf8n_to_uvuni((U8 *) str,
3670 len,
3671 &char_length,
3672 UTF8_ALLOW_ANYUV);
3673 /* Convert first code point to hex, including
3674 * the boiler plate before it. For all these,
3675 * we convert to native format so that
3676 * downstream code can continue to assume the
3677 * input is native */
3678 output_length =
3679 my_snprintf(hex_string, sizeof(hex_string),
3680 "\\N{U+%X",
3681 (unsigned int) UNI_TO_NATIVE(uv));
3682
3683 /* Make sure there is enough space to hold it */
3684 d = off + SvGROW(sv, off
3685 + output_length
3686 + (STRLEN)(send - e)
3687 + 2); /* '}' + NUL */
3688 /* And output it */
3689 Copy(hex_string, d, output_length, char);
3690 d += output_length;
3691
3692 /* For each subsequent character, append dot and
3693 * its ordinal in hex */
3694 while ((str += char_length) < str_end) {
3695 const STRLEN off = d - SvPVX_const(sv);
3696 U32 uv = utf8n_to_uvuni((U8 *) str,
3697 str_end - str,
3698 &char_length,
3699 UTF8_ALLOW_ANYUV);
3700 output_length =
3701 my_snprintf(hex_string,
3702 sizeof(hex_string),
3703 ".%X",
3704 (unsigned int) UNI_TO_NATIVE(uv));
3705
3706 d = off + SvGROW(sv, off
3707 + output_length
3708 + (STRLEN)(send - e)
3709 + 2); /* '}' + NUL */
3710 Copy(hex_string, d, output_length, char);
3711 d += output_length;
3712 }
94ca1619 3713 }
ff3f963a
KW
3714
3715 *d++ = '}'; /* Done. Add the trailing brace */
3716 }
3717 }
3718 else { /* Here, not in a pattern. Convert the name to a
3719 * string. */
3720
3721 /* If destination is not in utf8, unconditionally