This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
Test line breaks in PERL5DB
[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); \
851b527a
FC
306 if (PL_parser->herelines) \
307 CopLINE(PL_curcop) += PL_parser->herelines, \
308 PL_parser->herelines = 0; \
83944c01 309 } STMT_END
ffdb8b16
FC
310/* Called after scan_str to update CopLINE(PL_curcop), but only when there
311 * is no sublex_push to follow. */
312#define COPLINE_SET_FROM_MULTI_END \
313 STMT_START { \
314 CopLINE_set(PL_curcop, PL_multi_end); \
315 if (PL_multi_end != PL_multi_start) \
851b527a 316 PL_parser->herelines = 0; \
ffdb8b16 317 } STMT_END
83944c01
FC
318
319
8fa7f367
JH
320#ifdef DEBUGGING
321
6154021b 322/* how to interpret the pl_yylval associated with the token */
bbf60fe6
DM
323enum token_type {
324 TOKENTYPE_NONE,
325 TOKENTYPE_IVAL,
6154021b 326 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
bbf60fe6 327 TOKENTYPE_PVAL,
aeaef349 328 TOKENTYPE_OPVAL
bbf60fe6
DM
329};
330
6d4a66ac
NC
331static struct debug_tokens {
332 const int token;
333 enum token_type type;
334 const char *name;
335} const debug_tokens[] =
9041c2e3 336{
bbf60fe6
DM
337 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
338 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
339 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
340 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
341 { ARROW, TOKENTYPE_NONE, "ARROW" },
342 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
343 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
344 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
345 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
346 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
0d863452 347 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
bbf60fe6
DM
348 { DO, TOKENTYPE_NONE, "DO" },
349 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
350 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
351 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
352 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
353 { ELSE, TOKENTYPE_NONE, "ELSE" },
354 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
355 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
356 { FOR, TOKENTYPE_IVAL, "FOR" },
357 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
705fe0e5
FC
358 { FORMLBRACK, TOKENTYPE_NONE, "FORMLBRACK" },
359 { FORMRBRACK, TOKENTYPE_NONE, "FORMRBRACK" },
bbf60fe6
DM
360 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
361 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
7eb971ee 362 { FUNC0OP, TOKENTYPE_OPVAL, "FUNC0OP" },
bbf60fe6
DM
363 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
364 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
365 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
0d863452 366 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
bbf60fe6
DM
367 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
368 { IF, TOKENTYPE_IVAL, "IF" },
5504e6cf 369 { LABEL, TOKENTYPE_PVAL, "LABEL" },
bbf60fe6
DM
370 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
371 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
372 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
373 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
374 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
375 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
376 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
377 { MY, TOKENTYPE_IVAL, "MY" },
bbf60fe6
DM
378 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
379 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
380 { OROP, TOKENTYPE_IVAL, "OROP" },
381 { OROR, TOKENTYPE_NONE, "OROR" },
382 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
f3f204dc 383 { PEG, TOKENTYPE_NONE, "PEG" },
88e1f1a2
JV
384 { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" },
385 { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" },
bbf60fe6
DM
386 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
387 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
388 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
389 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
390 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
391 { PREINC, TOKENTYPE_NONE, "PREINC" },
392 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
f3f204dc 393 { QWLIST, TOKENTYPE_OPVAL, "QWLIST" },
bbf60fe6
DM
394 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
395 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
f3f204dc 396 { REQUIRE, TOKENTYPE_NONE, "REQUIRE" },
bbf60fe6
DM
397 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
398 { SUB, TOKENTYPE_NONE, "SUB" },
399 { THING, TOKENTYPE_OPVAL, "THING" },
400 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
401 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
402 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
403 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
404 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
405 { USE, TOKENTYPE_IVAL, "USE" },
0d863452 406 { WHEN, TOKENTYPE_IVAL, "WHEN" },
bbf60fe6
DM
407 { WHILE, TOKENTYPE_IVAL, "WHILE" },
408 { WORD, TOKENTYPE_OPVAL, "WORD" },
be25f609 409 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
c35e046a 410 { 0, TOKENTYPE_NONE, NULL }
bbf60fe6
DM
411};
412
6154021b 413/* dump the returned token in rv, plus any optional arg in pl_yylval */
998054bd 414
bbf60fe6 415STATIC int
704d4215 416S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
bbf60fe6 417{
97aff369 418 dVAR;
7918f24d
NC
419
420 PERL_ARGS_ASSERT_TOKEREPORT;
421
bbf60fe6 422 if (DEBUG_T_TEST) {
bd61b366 423 const char *name = NULL;
bbf60fe6 424 enum token_type type = TOKENTYPE_NONE;
f54cb97a 425 const struct debug_tokens *p;
396482e1 426 SV* const report = newSVpvs("<== ");
bbf60fe6 427
f54cb97a 428 for (p = debug_tokens; p->token; p++) {
bbf60fe6
DM
429 if (p->token == (int)rv) {
430 name = p->name;
431 type = p->type;
432 break;
433 }
434 }
435 if (name)
54667de8 436 Perl_sv_catpv(aTHX_ report, name);
74736ae6 437 else if ((char)rv > ' ' && (char)rv <= '~')
4ebc7986 438 {
bbf60fe6 439 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
4ebc7986
FC
440 if ((char)rv == 'p')
441 sv_catpvs(report, " (pending identifier)");
442 }
bbf60fe6 443 else if (!rv)
396482e1 444 sv_catpvs(report, "EOF");
bbf60fe6
DM
445 else
446 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
447 switch (type) {
448 case TOKENTYPE_NONE:
bbf60fe6
DM
449 break;
450 case TOKENTYPE_IVAL:
704d4215 451 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
bbf60fe6
DM
452 break;
453 case TOKENTYPE_OPNUM:
454 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
704d4215 455 PL_op_name[lvalp->ival]);
bbf60fe6
DM
456 break;
457 case TOKENTYPE_PVAL:
704d4215 458 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
bbf60fe6
DM
459 break;
460 case TOKENTYPE_OPVAL:
704d4215 461 if (lvalp->opval) {
401441c0 462 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
704d4215
GG
463 PL_op_name[lvalp->opval->op_type]);
464 if (lvalp->opval->op_type == OP_CONST) {
b6007c36 465 Perl_sv_catpvf(aTHX_ report, " %s",
704d4215 466 SvPEEK(cSVOPx_sv(lvalp->opval)));
b6007c36
DM
467 }
468
469 }
401441c0 470 else
396482e1 471 sv_catpvs(report, "(opval=null)");
bbf60fe6
DM
472 break;
473 }
b6007c36 474 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
bbf60fe6
DM
475 };
476 return (int)rv;
998054bd
SC
477}
478
b6007c36
DM
479
480/* print the buffer with suitable escapes */
481
482STATIC void
15f169a1 483S_printbuf(pTHX_ const char *const fmt, const char *const s)
b6007c36 484{
396482e1 485 SV* const tmp = newSVpvs("");
7918f24d
NC
486
487 PERL_ARGS_ASSERT_PRINTBUF;
488
b6007c36
DM
489 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
490 SvREFCNT_dec(tmp);
491}
492
8fa7f367
JH
493#endif
494
8290c323
NC
495static int
496S_deprecate_commaless_var_list(pTHX) {
497 PL_expect = XTERM;
498 deprecate("comma-less variable list");
499 return REPORT(','); /* grandfather non-comma-format format */
500}
501
ffb4593c
NT
502/*
503 * S_ao
504 *
c963b151
BD
505 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
506 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
ffb4593c
NT
507 */
508
76e3520e 509STATIC int
cea2e8a9 510S_ao(pTHX_ int toketype)
a0d0e21e 511{
97aff369 512 dVAR;
3280af22
NIS
513 if (*PL_bufptr == '=') {
514 PL_bufptr++;
a0d0e21e 515 if (toketype == ANDAND)
6154021b 516 pl_yylval.ival = OP_ANDASSIGN;
a0d0e21e 517 else if (toketype == OROR)
6154021b 518 pl_yylval.ival = OP_ORASSIGN;
c963b151 519 else if (toketype == DORDOR)
6154021b 520 pl_yylval.ival = OP_DORASSIGN;
a0d0e21e
LW
521 toketype = ASSIGNOP;
522 }
523 return toketype;
524}
525
ffb4593c
NT
526/*
527 * S_no_op
528 * When Perl expects an operator and finds something else, no_op
529 * prints the warning. It always prints "<something> found where
530 * operator expected. It prints "Missing semicolon on previous line?"
531 * if the surprise occurs at the start of the line. "do you need to
532 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
533 * where the compiler doesn't know if foo is a method call or a function.
534 * It prints "Missing operator before end of line" if there's nothing
535 * after the missing operator, or "... before <...>" if there is something
536 * after the missing operator.
537 */
538
76e3520e 539STATIC void
15f169a1 540S_no_op(pTHX_ const char *const what, char *s)
463ee0b2 541{
97aff369 542 dVAR;
9d4ba2ae
AL
543 char * const oldbp = PL_bufptr;
544 const bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 545
7918f24d
NC
546 PERL_ARGS_ASSERT_NO_OP;
547
1189a94a
GS
548 if (!s)
549 s = oldbp;
07c798fb 550 else
1189a94a 551 PL_bufptr = s;
734ab321 552 yywarn(Perl_form(aTHX_ "%s found where operator expected", what), UTF ? SVf_UTF8 : 0);
56da5a46
RGS
553 if (ckWARN_d(WARN_SYNTAX)) {
554 if (is_first)
555 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
556 "\t(Missing semicolon on previous line?)\n");
557 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
f54cb97a 558 const char *t;
8a2bca12 559 for (t = PL_oldoldbufptr; (isWORDCHAR_lazy_if(t,UTF) || *t == ':');
734ab321 560 t += UTF ? UTF8SKIP(t) : 1)
c35e046a 561 NOOP;
56da5a46
RGS
562 if (t < PL_bufptr && isSPACE(*t))
563 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
b17a0679
FC
564 "\t(Do you need to predeclare %"UTF8f"?)\n",
565 UTF8fARG(UTF, t - PL_oldoldbufptr, PL_oldoldbufptr));
56da5a46
RGS
566 }
567 else {
568 assert(s >= oldbp);
569 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
b17a0679
FC
570 "\t(Missing operator before %"UTF8f"?)\n",
571 UTF8fARG(UTF, s - oldbp, oldbp));
56da5a46 572 }
07c798fb 573 }
3280af22 574 PL_bufptr = oldbp;
8990e307
LW
575}
576
ffb4593c
NT
577/*
578 * S_missingterm
579 * Complain about missing quote/regexp/heredoc terminator.
d4c19fe8 580 * If it's called with NULL then it cauterizes the line buffer.
ffb4593c
NT
581 * If we're in a delimited string and the delimiter is a control
582 * character, it's reformatted into a two-char sequence like ^C.
583 * This is fatal.
584 */
585
76e3520e 586STATIC void
cea2e8a9 587S_missingterm(pTHX_ char *s)
8990e307 588{
97aff369 589 dVAR;
8990e307
LW
590 char tmpbuf[3];
591 char q;
592 if (s) {
9d4ba2ae 593 char * const nl = strrchr(s,'\n');
d2719217 594 if (nl)
8990e307
LW
595 *nl = '\0';
596 }
ca8b19a7 597 else if ((U8) PL_multi_close < 32) {
8990e307 598 *tmpbuf = '^';
585ec06d 599 tmpbuf[1] = (char)toCTRL(PL_multi_close);
8990e307
LW
600 tmpbuf[2] = '\0';
601 s = tmpbuf;
602 }
603 else {
eb160463 604 *tmpbuf = (char)PL_multi_close;
8990e307
LW
605 tmpbuf[1] = '\0';
606 s = tmpbuf;
607 }
608 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 609 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 610}
79072805 611
dd0ac2b9
FC
612#include "feature.h"
613
0d863452 614/*
0d863452
RH
615 * Check whether the named feature is enabled.
616 */
26ea9e12 617bool
3fff3427 618Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
0d863452 619{
97aff369 620 dVAR;
4a731d7b 621 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
7918f24d
NC
622
623 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
ca4d40c4
FC
624
625 assert(CURRENT_FEATURE_BUNDLE == FEATURE_BUNDLE_CUSTOM);
7918f24d 626
26ea9e12
NC
627 if (namelen > MAX_FEATURE_LEN)
628 return FALSE;
3fff3427 629 memcpy(&he_name[8], name, namelen);
7d69d4a6 630
c8ca97b0
NC
631 return cBOOL(cop_hints_fetch_pvn(PL_curcop, he_name, 8 + namelen, 0,
632 REFCOUNTED_HE_EXISTS));
0d863452
RH
633}
634
ffb4593c 635/*
9cbb5ea2
GS
636 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
637 * utf16-to-utf8-reversed.
ffb4593c
NT
638 */
639
c39cd008
GS
640#ifdef PERL_CR_FILTER
641static void
642strip_return(SV *sv)
643{
eb578fdb
KW
644 const char *s = SvPVX_const(sv);
645 const char * const e = s + SvCUR(sv);
7918f24d
NC
646
647 PERL_ARGS_ASSERT_STRIP_RETURN;
648
c39cd008
GS
649 /* outer loop optimized to do nothing if there are no CR-LFs */
650 while (s < e) {
651 if (*s++ == '\r' && *s == '\n') {
652 /* hit a CR-LF, need to copy the rest */
eb578fdb 653 char *d = s - 1;
c39cd008
GS
654 *d++ = *s++;
655 while (s < e) {
656 if (*s == '\r' && s[1] == '\n')
657 s++;
658 *d++ = *s++;
659 }
660 SvCUR(sv) -= s - d;
661 return;
662 }
663 }
664}
a868473f 665
76e3520e 666STATIC I32
c39cd008 667S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 668{
f54cb97a 669 const I32 count = FILTER_READ(idx+1, sv, maxlen);
c39cd008
GS
670 if (count > 0 && !maxlen)
671 strip_return(sv);
672 return count;
a868473f
NIS
673}
674#endif
675
ffb4593c 676/*
8eaa0acf
Z
677=for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags
678
679Creates and initialises a new lexer/parser state object, supplying
680a context in which to lex and parse from a new source of Perl code.
681A pointer to the new state object is placed in L</PL_parser>. An entry
682is made on the save stack so that upon unwinding the new state object
683will be destroyed and the former value of L</PL_parser> will be restored.
684Nothing else need be done to clean up the parsing context.
685
686The code to be parsed comes from I<line> and I<rsfp>. I<line>, if
687non-null, provides a string (in SV form) containing code to be parsed.
688A copy of the string is made, so subsequent modification of I<line>
689does not affect parsing. I<rsfp>, if non-null, provides an input stream
690from which code will be read to be parsed. If both are non-null, the
691code in I<line> comes first and must consist of complete lines of input,
692and I<rsfp> supplies the remainder of the source.
693
e368b3bd
FC
694The I<flags> parameter is reserved for future use. Currently it is only
695used by perl internally, so extensions should always pass zero.
8eaa0acf
Z
696
697=cut
698*/
ffb4593c 699
27fcb6ee 700/* LEX_START_SAME_FILTER indicates that this is not a new file, so it
87606032
NC
701 can share filters with the current parser.
702 LEX_START_DONT_CLOSE indicates that the file handle wasn't opened by the
703 caller, hence isn't owned by the parser, so shouldn't be closed on parser
704 destruction. This is used to handle the case of defaulting to reading the
705 script from the standard input because no filename was given on the command
706 line (without getting confused by situation where STDIN has been closed, so
707 the script handle is opened on fd 0) */
27fcb6ee 708
a0d0e21e 709void
8eaa0acf 710Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags)
79072805 711{
97aff369 712 dVAR;
6ef55633 713 const char *s = NULL;
5486870f 714 yy_parser *parser, *oparser;
60d63348 715 if (flags && flags & ~LEX_START_FLAGS)
8eaa0acf 716 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start");
acdf0a21
DM
717
718 /* create and initialise a parser */
719
199e78b7 720 Newxz(parser, 1, yy_parser);
5486870f 721 parser->old_parser = oparser = PL_parser;
acdf0a21
DM
722 PL_parser = parser;
723
28ac2b49
Z
724 parser->stack = NULL;
725 parser->ps = NULL;
726 parser->stack_size = 0;
acdf0a21 727
e3abe207
DM
728 /* on scope exit, free this parser and restore any outer one */
729 SAVEPARSER(parser);
7c4baf47 730 parser->saved_curcop = PL_curcop;
e3abe207 731
acdf0a21 732 /* initialise lexer state */
8990e307 733
fb205e7a
DM
734#ifdef PERL_MAD
735 parser->curforce = -1;
736#else
737 parser->nexttoke = 0;
738#endif
ca4cfd28 739 parser->error_count = oparser ? oparser->error_count : 0;
c2598295 740 parser->copline = NOLINE;
5afb0a62 741 parser->lex_state = LEX_NORMAL;
c2598295 742 parser->expect = XSTATE;
2f9285f8 743 parser->rsfp = rsfp;
27fcb6ee
FC
744 parser->rsfp_filters =
745 !(flags & LEX_START_SAME_FILTER) || !oparser
d3cd8e11
FC
746 ? NULL
747 : MUTABLE_AV(SvREFCNT_inc(
748 oparser->rsfp_filters
749 ? oparser->rsfp_filters
750 : (oparser->rsfp_filters = newAV())
751 ));
2f9285f8 752
199e78b7
DM
753 Newx(parser->lex_brackstack, 120, char);
754 Newx(parser->lex_casestack, 12, char);
755 *parser->lex_casestack = '\0';
d794b522 756 Newxz(parser->lex_shared, 1, LEXSHARED);
02b34bbe 757
10efb74f 758 if (line) {
0528fd32 759 STRLEN len;
10efb74f 760 s = SvPV_const(line, len);
0abcdfa4
FC
761 parser->linestr = flags & LEX_START_COPIED
762 ? SvREFCNT_inc_simple_NN(line)
763 : newSVpvn_flags(s, len, SvUTF8(line));
bf1b738b 764 sv_catpvn(parser->linestr, "\n;", rsfp ? 1 : 2);
0abcdfa4 765 } else {
bf1b738b 766 parser->linestr = newSVpvn("\n;", rsfp ? 1 : 2);
8990e307 767 }
f06b5848
DM
768 parser->oldoldbufptr =
769 parser->oldbufptr =
770 parser->bufptr =
771 parser->linestart = SvPVX(parser->linestr);
772 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
773 parser->last_lop = parser->last_uni = NULL;
87606032
NC
774 parser->lex_flags = flags & (LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES
775 |LEX_DONT_CLOSE_RSFP);
737c24fc 776
60d63348 777 parser->in_pod = parser->filtered = 0;
79072805 778}
a687059c 779
e3abe207
DM
780
781/* delete a parser object */
782
783void
784Perl_parser_free(pTHX_ const yy_parser *parser)
785{
7918f24d
NC
786 PERL_ARGS_ASSERT_PARSER_FREE;
787
7c4baf47 788 PL_curcop = parser->saved_curcop;
bdc0bf6f
DM
789 SvREFCNT_dec(parser->linestr);
790
87606032 791 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
2f9285f8 792 PerlIO_clearerr(parser->rsfp);
799361c3
SH
793 else if (parser->rsfp && (!parser->old_parser ||
794 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
2f9285f8 795 PerlIO_close(parser->rsfp);
5486870f 796 SvREFCNT_dec(parser->rsfp_filters);
10002bc1
FC
797 SvREFCNT_dec(parser->lex_stuff);
798 SvREFCNT_dec(parser->sublex_info.repl);
3ac7ff8f
FC
799
800 Safefree(parser->lex_brackstack);
801 Safefree(parser->lex_casestack);
802 Safefree(parser->lex_shared);
803 PL_parser = parser->old_parser;
804 Safefree(parser);
805}
806
807void
808Perl_parser_free_nexttoke_ops(pTHX_ yy_parser *parser, OPSLAB *slab)
809{
810#ifdef PERL_MAD
811 I32 nexttoke = parser->lasttoke;
812#else
813 I32 nexttoke = parser->nexttoke;
814#endif
815 PERL_ARGS_ASSERT_PARSER_FREE_NEXTTOKE_OPS;
3ce3dcd9
FC
816 while (nexttoke--) {
817#ifdef PERL_MAD
818 if (S_is_opval_token(parser->nexttoke[nexttoke].next_type
3ac7ff8f
FC
819 & 0xffff)
820 && parser->nexttoke[nexttoke].next_val.opval
821 && parser->nexttoke[nexttoke].next_val.opval->op_slabbed
822 && OpSLAB(parser->nexttoke[nexttoke].next_val.opval) == slab) {
823 op_free(parser->nexttoke[nexttoke].next_val.opval);
824 parser->nexttoke[nexttoke].next_val.opval = NULL;
825 }
3ce3dcd9 826#else
3ac7ff8f
FC
827 if (S_is_opval_token(parser->nexttype[nexttoke] & 0xffff)
828 && parser->nextval[nexttoke].opval
829 && parser->nextval[nexttoke].opval->op_slabbed
830 && OpSLAB(parser->nextval[nexttoke].opval) == slab) {
3ce3dcd9 831 op_free(parser->nextval[nexttoke].opval);
3ac7ff8f
FC
832 parser->nextval[nexttoke].opval = NULL;
833 }
3ce3dcd9
FC
834#endif
835 }
e3abe207
DM
836}
837
838
ffb4593c 839/*
f0e67a1d
Z
840=for apidoc AmxU|SV *|PL_parser-E<gt>linestr
841
842Buffer scalar containing the chunk currently under consideration of the
843text currently being lexed. This is always a plain string scalar (for
844which C<SvPOK> is true). It is not intended to be used as a scalar by
845normal scalar means; instead refer to the buffer directly by the pointer
846variables described below.
847
848The lexer maintains various C<char*> pointers to things in the
849C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever
850reallocated, all of these pointers must be updated. Don't attempt to
851do this manually, but rather use L</lex_grow_linestr> if you need to
852reallocate the buffer.
853
854The content of the text chunk in the buffer is commonly exactly one
855complete line of input, up to and including a newline terminator,
856but there are situations where it is otherwise. The octets of the
857buffer may be intended to be interpreted as either UTF-8 or Latin-1.
858The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8>
859flag on this scalar, which may disagree with it.
860
861For direct examination of the buffer, the variable
862L</PL_parser-E<gt>bufend> points to the end of the buffer. The current
863lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use
864of these pointers is usually preferable to examination of the scalar
865through normal scalar means.
866
867=for apidoc AmxU|char *|PL_parser-E<gt>bufend
868
869Direct pointer to the end of the chunk of text currently being lexed, the
870end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr)
871+ SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is
872always located at the end of the buffer, and does not count as part of
873the buffer's contents.
874
875=for apidoc AmxU|char *|PL_parser-E<gt>bufptr
876
877Points to the current position of lexing inside the lexer buffer.
878Characters around this point may be freely examined, within
879the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
880L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be
881interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
882
883Lexing code (whether in the Perl core or not) moves this pointer past
884the characters that it consumes. It is also expected to perform some
885bookkeeping whenever a newline character is consumed. This movement
886can be more conveniently performed by the function L</lex_read_to>,
887which handles newlines appropriately.
888
889Interpretation of the buffer's octets can be abstracted out by
890using the slightly higher-level functions L</lex_peek_unichar> and
891L</lex_read_unichar>.
892
893=for apidoc AmxU|char *|PL_parser-E<gt>linestart
894
895Points to the start of the current line inside the lexer buffer.
896This is useful for indicating at which column an error occurred, and
897not much else. This must be updated by any lexing code that consumes
898a newline; the function L</lex_read_to> handles this detail.
899
900=cut
901*/
902
903/*
904=for apidoc Amx|bool|lex_bufutf8
905
906Indicates whether the octets in the lexer buffer
907(L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
908of Unicode characters. If not, they should be interpreted as Latin-1
909characters. This is analogous to the C<SvUTF8> flag for scalars.
910
911In UTF-8 mode, it is not guaranteed that the lexer buffer actually
912contains valid UTF-8. Lexing code must be robust in the face of invalid
913encoding.
914
915The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
916is significant, but not the whole story regarding the input character
917encoding. Normally, when a file is being read, the scalar contains octets
918and its C<SvUTF8> flag is off, but the octets should be interpreted as
919UTF-8 if the C<use utf8> pragma is in effect. During a string eval,
920however, the scalar may have the C<SvUTF8> flag on, and in this case its
921octets should be interpreted as UTF-8 unless the C<use bytes> pragma
922is in effect. This logic may change in the future; use this function
923instead of implementing the logic yourself.
924
925=cut
926*/
927
928bool
929Perl_lex_bufutf8(pTHX)
930{
931 return UTF;
932}
933
934/*
935=for apidoc Amx|char *|lex_grow_linestr|STRLEN len
936
937Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
938at least I<len> octets (including terminating NUL). Returns a
939pointer to the reallocated buffer. This is necessary before making
940any direct modification of the buffer that would increase its length.
941L</lex_stuff_pvn> provides a more convenient way to insert text into
942the buffer.
943
944Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
945this function updates all of the lexer's variables that point directly
946into the buffer.
947
948=cut
949*/
950
951char *
952Perl_lex_grow_linestr(pTHX_ STRLEN len)
953{
954 SV *linestr;
955 char *buf;
956 STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
c7641931 957 STRLEN linestart_pos, last_uni_pos, last_lop_pos, re_eval_start_pos;
f0e67a1d
Z
958 linestr = PL_parser->linestr;
959 buf = SvPVX(linestr);
960 if (len <= SvLEN(linestr))
961 return buf;
962 bufend_pos = PL_parser->bufend - buf;
963 bufptr_pos = PL_parser->bufptr - buf;
964 oldbufptr_pos = PL_parser->oldbufptr - buf;
965 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
966 linestart_pos = PL_parser->linestart - buf;
967 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
968 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
3328ab5a
FC
969 re_eval_start_pos = PL_parser->lex_shared->re_eval_start ?
970 PL_parser->lex_shared->re_eval_start - buf : 0;
c7641931 971
f0e67a1d 972 buf = sv_grow(linestr, len);
c7641931 973
f0e67a1d
Z
974 PL_parser->bufend = buf + bufend_pos;
975 PL_parser->bufptr = buf + bufptr_pos;
976 PL_parser->oldbufptr = buf + oldbufptr_pos;
977 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
978 PL_parser->linestart = buf + linestart_pos;
979 if (PL_parser->last_uni)
980 PL_parser->last_uni = buf + last_uni_pos;
981 if (PL_parser->last_lop)
982 PL_parser->last_lop = buf + last_lop_pos;
3328ab5a
FC
983 if (PL_parser->lex_shared->re_eval_start)
984 PL_parser->lex_shared->re_eval_start = buf + re_eval_start_pos;
f0e67a1d
Z
985 return buf;
986}
987
988/*
83aa740e 989=for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags
f0e67a1d
Z
990
991Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
992immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
993reallocating the buffer if necessary. This means that lexing code that
994runs later will see the characters as if they had appeared in the input.
995It is not recommended to do this as part of normal parsing, and most
996uses of this facility run the risk of the inserted characters being
997interpreted in an unintended manner.
998
999The string to be inserted is represented by I<len> octets starting
1000at I<pv>. These octets are interpreted as either UTF-8 or Latin-1,
1001according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
1002The characters are recoded for the lexer buffer, according to how the
1003buffer is currently being interpreted (L</lex_bufutf8>). If a string
9dcc53ea 1004to be inserted is available as a Perl scalar, the L</lex_stuff_sv>
f0e67a1d
Z
1005function is more convenient.
1006
1007=cut
1008*/
1009
1010void
83aa740e 1011Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags)
f0e67a1d 1012{
749123ff 1013 dVAR;
f0e67a1d
Z
1014 char *bufptr;
1015 PERL_ARGS_ASSERT_LEX_STUFF_PVN;
1016 if (flags & ~(LEX_STUFF_UTF8))
1017 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
1018 if (UTF) {
1019 if (flags & LEX_STUFF_UTF8) {
1020 goto plain_copy;
1021 } else {
54d004e8 1022 STRLEN highhalf = 0; /* Count of variants */
83aa740e 1023 const char *p, *e = pv+len;
54d004e8
KW
1024 for (p = pv; p != e; p++) {
1025 if (! UTF8_IS_INVARIANT(*p)) {
1026 highhalf++;
1027 }
1028 }
f0e67a1d
Z
1029 if (!highhalf)
1030 goto plain_copy;
1031 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
1032 bufptr = PL_parser->bufptr;
1033 Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1034 SvCUR_set(PL_parser->linestr,
1035 SvCUR(PL_parser->linestr) + len+highhalf);
f0e67a1d
Z
1036 PL_parser->bufend += len+highhalf;
1037 for (p = pv; p != e; p++) {
1038 U8 c = (U8)*p;
54d004e8
KW
1039 if (! UTF8_IS_INVARIANT(c)) {
1040 *bufptr++ = UTF8_TWO_BYTE_HI(c);
1041 *bufptr++ = UTF8_TWO_BYTE_LO(c);
f0e67a1d
Z
1042 } else {
1043 *bufptr++ = (char)c;
1044 }
1045 }
1046 }
1047 } else {
1048 if (flags & LEX_STUFF_UTF8) {
1049 STRLEN highhalf = 0;
83aa740e 1050 const char *p, *e = pv+len;
f0e67a1d
Z
1051 for (p = pv; p != e; p++) {
1052 U8 c = (U8)*p;
54d004e8 1053 if (UTF8_IS_ABOVE_LATIN1(c)) {
f0e67a1d
Z
1054 Perl_croak(aTHX_ "Lexing code attempted to stuff "
1055 "non-Latin-1 character into Latin-1 input");
54d004e8 1056 } else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(p, e)) {
f0e67a1d
Z
1057 p++;
1058 highhalf++;
54d004e8 1059 } else if (! UTF8_IS_INVARIANT(c)) {
f0e67a1d
Z
1060 /* malformed UTF-8 */
1061 ENTER;
1062 SAVESPTR(PL_warnhook);
1063 PL_warnhook = PERL_WARNHOOK_FATAL;
c80e42f3 1064 utf8n_to_uvchr((U8*)p, e-p, NULL, 0);
f0e67a1d
Z
1065 LEAVE;
1066 }
1067 }
1068 if (!highhalf)
1069 goto plain_copy;
1070 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
1071 bufptr = PL_parser->bufptr;
1072 Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1073 SvCUR_set(PL_parser->linestr,
1074 SvCUR(PL_parser->linestr) + len-highhalf);
f0e67a1d 1075 PL_parser->bufend += len-highhalf;
54d004e8
KW
1076 p = pv;
1077 while (p < e) {
1078 if (UTF8_IS_INVARIANT(*p)) {
1079 *bufptr++ = *p;
1080 p++;
f0e67a1d 1081 }
54d004e8
KW
1082 else {
1083 assert(p < e -1 );
94bb8c36 1084 *bufptr++ = TWO_BYTE_UTF8_TO_NATIVE(*p, *(p+1));
54d004e8
KW
1085 p += 2;
1086 }
f0e67a1d
Z
1087 }
1088 } else {
54d004e8 1089 plain_copy:
f0e67a1d
Z
1090 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1091 bufptr = PL_parser->bufptr;
1092 Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
255fdf19 1093 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len);
f0e67a1d
Z
1094 PL_parser->bufend += len;
1095 Copy(pv, bufptr, len, char);
1096 }
1097 }
1098}
1099
1100/*
9dcc53ea
Z
1101=for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags
1102
1103Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1104immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1105reallocating the buffer if necessary. This means that lexing code that
1106runs later will see the characters as if they had appeared in the input.
1107It is not recommended to do this as part of normal parsing, and most
1108uses of this facility run the risk of the inserted characters being
1109interpreted in an unintended manner.
1110
1111The string to be inserted is represented by octets starting at I<pv>
1112and continuing to the first nul. These octets are interpreted as either
1113UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set
1114in I<flags>. The characters are recoded for the lexer buffer, according
1115to how the buffer is currently being interpreted (L</lex_bufutf8>).
1116If it is not convenient to nul-terminate a string to be inserted, the
1117L</lex_stuff_pvn> function is more appropriate.
1118
1119=cut
1120*/
1121
1122void
1123Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags)
1124{
1125 PERL_ARGS_ASSERT_LEX_STUFF_PV;
1126 lex_stuff_pvn(pv, strlen(pv), flags);
1127}
1128
1129/*
f0e67a1d
Z
1130=for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1131
1132Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1133immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1134reallocating the buffer if necessary. This means that lexing code that
1135runs later will see the characters as if they had appeared in the input.
1136It is not recommended to do this as part of normal parsing, and most
1137uses of this facility run the risk of the inserted characters being
1138interpreted in an unintended manner.
1139
1140The string to be inserted is the string value of I<sv>. The characters
1141are recoded for the lexer buffer, according to how the buffer is currently
9dcc53ea 1142being interpreted (L</lex_bufutf8>). If a string to be inserted is
f0e67a1d
Z
1143not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1144need to construct a scalar.
1145
1146=cut
1147*/
1148
1149void
1150Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1151{
1152 char *pv;
1153 STRLEN len;
1154 PERL_ARGS_ASSERT_LEX_STUFF_SV;
1155 if (flags)
1156 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1157 pv = SvPV(sv, len);
1158 lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1159}
1160
1161/*
1162=for apidoc Amx|void|lex_unstuff|char *ptr
1163
1164Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1165I<ptr>. Text following I<ptr> will be moved, and the buffer shortened.
1166This hides the discarded text from any lexing code that runs later,
1167as if the text had never appeared.
1168
1169This is not the normal way to consume lexed text. For that, use
1170L</lex_read_to>.
1171
1172=cut
1173*/
1174
1175void
1176Perl_lex_unstuff(pTHX_ char *ptr)
1177{
1178 char *buf, *bufend;
1179 STRLEN unstuff_len;
1180 PERL_ARGS_ASSERT_LEX_UNSTUFF;
1181 buf = PL_parser->bufptr;
1182 if (ptr < buf)
1183 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1184 if (ptr == buf)
1185 return;
1186 bufend = PL_parser->bufend;
1187 if (ptr > bufend)
1188 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1189 unstuff_len = ptr - buf;
1190 Move(ptr, buf, bufend+1-ptr, char);
1191 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1192 PL_parser->bufend = bufend - unstuff_len;
1193}
1194
1195/*
1196=for apidoc Amx|void|lex_read_to|char *ptr
1197
1198Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1199to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1200performing the correct bookkeeping whenever a newline character is passed.
1201This is the normal way to consume lexed text.
1202
1203Interpretation of the buffer's octets can be abstracted out by
1204using the slightly higher-level functions L</lex_peek_unichar> and
1205L</lex_read_unichar>.
1206
1207=cut
1208*/
1209
1210void
1211Perl_lex_read_to(pTHX_ char *ptr)
1212{
1213 char *s;
1214 PERL_ARGS_ASSERT_LEX_READ_TO;
1215 s = PL_parser->bufptr;
1216 if (ptr < s || ptr > PL_parser->bufend)
1217 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1218 for (; s != ptr; s++)
1219 if (*s == '\n') {
83944c01 1220 COPLINE_INC_WITH_HERELINES;
f0e67a1d
Z
1221 PL_parser->linestart = s+1;
1222 }
1223 PL_parser->bufptr = ptr;
1224}
1225
1226/*
1227=for apidoc Amx|void|lex_discard_to|char *ptr
1228
1229Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1230up to I<ptr>. The remaining content of the buffer will be moved, and
1231all pointers into the buffer updated appropriately. I<ptr> must not
1232be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1233it is not permitted to discard text that has yet to be lexed.
1234
1235Normally it is not necessarily to do this directly, because it suffices to
1236use the implicit discarding behaviour of L</lex_next_chunk> and things
1237based on it. However, if a token stretches across multiple lines,
1f317c95 1238and the lexing code has kept multiple lines of text in the buffer for
f0e67a1d
Z
1239that purpose, then after completion of the token it would be wise to
1240explicitly discard the now-unneeded earlier lines, to avoid future
1241multi-line tokens growing the buffer without bound.
1242
1243=cut
1244*/
1245
1246void
1247Perl_lex_discard_to(pTHX_ char *ptr)
1248{
1249 char *buf;
1250 STRLEN discard_len;
1251 PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1252 buf = SvPVX(PL_parser->linestr);
1253 if (ptr < buf)
1254 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1255 if (ptr == buf)
1256 return;
1257 if (ptr > PL_parser->bufptr)
1258 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1259 discard_len = ptr - buf;
1260 if (PL_parser->oldbufptr < ptr)
1261 PL_parser->oldbufptr = ptr;
1262 if (PL_parser->oldoldbufptr < ptr)
1263 PL_parser->oldoldbufptr = ptr;
1264 if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1265 PL_parser->last_uni = NULL;
1266 if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1267 PL_parser->last_lop = NULL;
1268 Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1269 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1270 PL_parser->bufend -= discard_len;
1271 PL_parser->bufptr -= discard_len;
1272 PL_parser->oldbufptr -= discard_len;
1273 PL_parser->oldoldbufptr -= discard_len;
1274 if (PL_parser->last_uni)
1275 PL_parser->last_uni -= discard_len;
1276 if (PL_parser->last_lop)
1277 PL_parser->last_lop -= discard_len;
1278}
1279
1280/*
1281=for apidoc Amx|bool|lex_next_chunk|U32 flags
1282
1283Reads in the next chunk of text to be lexed, appending it to
1284L</PL_parser-E<gt>linestr>. This should be called when lexing code has
1285looked to the end of the current chunk and wants to know more. It is
1286usual, but not necessary, for lexing to have consumed the entirety of
1287the current chunk at this time.
1288
1289If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1290chunk (i.e., the current chunk has been entirely consumed), normally the
1291current chunk will be discarded at the same time that the new chunk is
1292read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1293will not be discarded. If the current chunk has not been entirely
1294consumed, then it will not be discarded regardless of the flag.
1295
1296Returns true if some new text was added to the buffer, or false if the
1297buffer has reached the end of the input text.
1298
1299=cut
1300*/
1301
1302#define LEX_FAKE_EOF 0x80000000
112d1284 1303#define LEX_NO_TERM 0x40000000
f0e67a1d
Z
1304
1305bool
1306Perl_lex_next_chunk(pTHX_ U32 flags)
1307{
1308 SV *linestr;
1309 char *buf;
1310 STRLEN old_bufend_pos, new_bufend_pos;
1311 STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1312 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
17cc9359 1313 bool got_some_for_debugger = 0;
f0e67a1d 1314 bool got_some;
112d1284 1315 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF|LEX_NO_TERM))
f0e67a1d 1316 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
f0e67a1d
Z
1317 linestr = PL_parser->linestr;
1318 buf = SvPVX(linestr);
1319 if (!(flags & LEX_KEEP_PREVIOUS) &&
1320 PL_parser->bufptr == PL_parser->bufend) {
1321 old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1322 linestart_pos = 0;
1323 if (PL_parser->last_uni != PL_parser->bufend)
1324 PL_parser->last_uni = NULL;
1325 if (PL_parser->last_lop != PL_parser->bufend)
1326 PL_parser->last_lop = NULL;
1327 last_uni_pos = last_lop_pos = 0;
1328 *buf = 0;
1329 SvCUR(linestr) = 0;
1330 } else {
1331 old_bufend_pos = PL_parser->bufend - buf;
1332 bufptr_pos = PL_parser->bufptr - buf;
1333 oldbufptr_pos = PL_parser->oldbufptr - buf;
1334 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1335 linestart_pos = PL_parser->linestart - buf;
1336 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1337 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1338 }
1339 if (flags & LEX_FAKE_EOF) {
1340 goto eof;
60d63348 1341 } else if (!PL_parser->rsfp && !PL_parser->filtered) {
f0e67a1d
Z
1342 got_some = 0;
1343 } else if (filter_gets(linestr, old_bufend_pos)) {
1344 got_some = 1;
17cc9359 1345 got_some_for_debugger = 1;
112d1284
FC
1346 } else if (flags & LEX_NO_TERM) {
1347 got_some = 0;
f0e67a1d 1348 } else {
580561a3
Z
1349 if (!SvPOK(linestr)) /* can get undefined by filter_gets */
1350 sv_setpvs(linestr, "");
f0e67a1d
Z
1351 eof:
1352 /* End of real input. Close filehandle (unless it was STDIN),
1353 * then add implicit termination.
1354 */
87606032 1355 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
f0e67a1d
Z
1356 PerlIO_clearerr(PL_parser->rsfp);
1357 else if (PL_parser->rsfp)
1358 (void)PerlIO_close(PL_parser->rsfp);
1359 PL_parser->rsfp = NULL;
60d63348 1360 PL_parser->in_pod = PL_parser->filtered = 0;
f0e67a1d
Z
1361#ifdef PERL_MAD
1362 if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1363 PL_faketokens = 1;
1364#endif
1365 if (!PL_in_eval && PL_minus_p) {
1366 sv_catpvs(linestr,
1367 /*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1368 PL_minus_n = PL_minus_p = 0;
1369 } else if (!PL_in_eval && PL_minus_n) {
1370 sv_catpvs(linestr, /*{*/";}");
1371 PL_minus_n = 0;
1372 } else
1373 sv_catpvs(linestr, ";");
1374 got_some = 1;
1375 }
1376 buf = SvPVX(linestr);
1377 new_bufend_pos = SvCUR(linestr);
1378 PL_parser->bufend = buf + new_bufend_pos;
1379 PL_parser->bufptr = buf + bufptr_pos;
1380 PL_parser->oldbufptr = buf + oldbufptr_pos;
1381 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1382 PL_parser->linestart = buf + linestart_pos;
1383 if (PL_parser->last_uni)
1384 PL_parser->last_uni = buf + last_uni_pos;
1385 if (PL_parser->last_lop)
1386 PL_parser->last_lop = buf + last_lop_pos;
17cc9359 1387 if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
f0e67a1d
Z
1388 PL_curstash != PL_debstash) {
1389 /* debugger active and we're not compiling the debugger code,
1390 * so store the line into the debugger's array of lines
1391 */
1392 update_debugger_info(NULL, buf+old_bufend_pos,
1393 new_bufend_pos-old_bufend_pos);
1394 }
1395 return got_some;
1396}
1397
1398/*
1399=for apidoc Amx|I32|lex_peek_unichar|U32 flags
1400
1401Looks ahead one (Unicode) character in the text currently being lexed.
1402Returns the codepoint (unsigned integer value) of the next character,
1403or -1 if lexing has reached the end of the input text. To consume the
1404peeked character, use L</lex_read_unichar>.
1405
1406If the next character is in (or extends into) the next chunk of input
1407text, the next chunk will be read in. Normally the current chunk will be
1408discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1409then the current chunk will not be discarded.
1410
1411If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1412is encountered, an exception is generated.
1413
1414=cut
1415*/
1416
1417I32
1418Perl_lex_peek_unichar(pTHX_ U32 flags)
1419{
749123ff 1420 dVAR;
f0e67a1d
Z
1421 char *s, *bufend;
1422 if (flags & ~(LEX_KEEP_PREVIOUS))
1423 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1424 s = PL_parser->bufptr;
1425 bufend = PL_parser->bufend;
1426 if (UTF) {
1427 U8 head;
1428 I32 unichar;
1429 STRLEN len, retlen;
1430 if (s == bufend) {
1431 if (!lex_next_chunk(flags))
1432 return -1;
1433 s = PL_parser->bufptr;
1434 bufend = PL_parser->bufend;
1435 }
1436 head = (U8)*s;
54d004e8 1437 if (UTF8_IS_INVARIANT(head))
f0e67a1d 1438 return head;
54d004e8
KW
1439 if (UTF8_IS_START(head)) {
1440 len = UTF8SKIP(&head);
f0e67a1d
Z
1441 while ((STRLEN)(bufend-s) < len) {
1442 if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1443 break;
1444 s = PL_parser->bufptr;
1445 bufend = PL_parser->bufend;
1446 }
1447 }
c80e42f3 1448 unichar = utf8n_to_uvchr((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
f0e67a1d
Z
1449 if (retlen == (STRLEN)-1) {
1450 /* malformed UTF-8 */
1451 ENTER;
1452 SAVESPTR(PL_warnhook);
1453 PL_warnhook = PERL_WARNHOOK_FATAL;
c80e42f3 1454 utf8n_to_uvchr((U8*)s, bufend-s, NULL, 0);
f0e67a1d
Z
1455 LEAVE;
1456 }
1457 return unichar;
1458 } else {
1459 if (s == bufend) {
1460 if (!lex_next_chunk(flags))
1461 return -1;
1462 s = PL_parser->bufptr;
1463 }
1464 return (U8)*s;
1465 }
1466}
1467
1468/*
1469=for apidoc Amx|I32|lex_read_unichar|U32 flags
1470
1471Reads the next (Unicode) character in the text currently being lexed.
1472Returns the codepoint (unsigned integer value) of the character read,
1473and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1474if lexing has reached the end of the input text. To non-destructively
1475examine the next character, use L</lex_peek_unichar> instead.
1476
1477If the next character is in (or extends into) the next chunk of input
1478text, the next chunk will be read in. Normally the current chunk will be
1479discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1480then the current chunk will not be discarded.
1481
1482If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1483is encountered, an exception is generated.
1484
1485=cut
1486*/
1487
1488I32
1489Perl_lex_read_unichar(pTHX_ U32 flags)
1490{
1491 I32 c;
1492 if (flags & ~(LEX_KEEP_PREVIOUS))
1493 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1494 c = lex_peek_unichar(flags);
1495 if (c != -1) {
1496 if (c == '\n')
83944c01 1497 COPLINE_INC_WITH_HERELINES;
d9018cbe
EB
1498 if (UTF)
1499 PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1500 else
1501 ++(PL_parser->bufptr);
f0e67a1d
Z
1502 }
1503 return c;
1504}
1505
1506/*
1507=for apidoc Amx|void|lex_read_space|U32 flags
1508
1509Reads optional spaces, in Perl style, in the text currently being
1510lexed. The spaces may include ordinary whitespace characters and
1511Perl-style comments. C<#line> directives are processed if encountered.
1512L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1513at a non-space character (or the end of the input text).
1514
1515If spaces extend into the next chunk of input text, the next chunk will
1516be read in. Normally the current chunk will be discarded at the same
1517time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1518chunk will not be discarded.
1519
1520=cut
1521*/
1522
21791330 1523#define LEX_NO_INCLINE 0x40000000
f0998909
Z
1524#define LEX_NO_NEXT_CHUNK 0x80000000
1525
f0e67a1d
Z
1526void
1527Perl_lex_read_space(pTHX_ U32 flags)
1528{
1529 char *s, *bufend;
21791330 1530 const bool can_incline = !(flags & LEX_NO_INCLINE);
f0e67a1d 1531 bool need_incline = 0;
21791330 1532 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK|LEX_NO_INCLINE))
f0e67a1d
Z
1533 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1534#ifdef PERL_MAD
1535 if (PL_skipwhite) {
1536 sv_free(PL_skipwhite);
1537 PL_skipwhite = NULL;
1538 }
1539 if (PL_madskills)
1540 PL_skipwhite = newSVpvs("");
1541#endif /* PERL_MAD */
1542 s = PL_parser->bufptr;
1543 bufend = PL_parser->bufend;
1544 while (1) {
1545 char c = *s;
1546 if (c == '#') {
1547 do {
1548 c = *++s;
1549 } while (!(c == '\n' || (c == 0 && s == bufend)));
1550 } else if (c == '\n') {
1551 s++;
21791330
FC
1552 if (can_incline) {
1553 PL_parser->linestart = s;
1554 if (s == bufend)
1555 need_incline = 1;
1556 else
1557 incline(s);
1558 }
f0e67a1d
Z
1559 } else if (isSPACE(c)) {
1560 s++;
1561 } else if (c == 0 && s == bufend) {
1562 bool got_more;
65c68e17 1563 line_t l;
f0e67a1d
Z
1564#ifdef PERL_MAD
1565 if (PL_madskills)
1566 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1567#endif /* PERL_MAD */
f0998909
Z
1568 if (flags & LEX_NO_NEXT_CHUNK)
1569 break;
f0e67a1d 1570 PL_parser->bufptr = s;
65c68e17 1571 l = CopLINE(PL_curcop);
851b527a 1572 CopLINE(PL_curcop) += PL_parser->herelines + 1;
f0e67a1d 1573 got_more = lex_next_chunk(flags);
65c68e17 1574 CopLINE_set(PL_curcop, l);
f0e67a1d
Z
1575 s = PL_parser->bufptr;
1576 bufend = PL_parser->bufend;
1577 if (!got_more)
1578 break;
21791330 1579 if (can_incline && need_incline && PL_parser->rsfp) {
f0e67a1d
Z
1580 incline(s);
1581 need_incline = 0;
1582 }
1583 } else {
1584 break;
1585 }
1586 }
1587#ifdef PERL_MAD
1588 if (PL_madskills)
1589 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1590#endif /* PERL_MAD */
1591 PL_parser->bufptr = s;
1592}
1593
1594/*
fe788d6b
PM
1595
1596=for apidoc EXMp|bool|validate_proto|SV *name|SV *proto|bool warn
1597
1598This function performs syntax checking on a prototype, C<proto>.
1599If C<warn> is true, any illegal characters or mismatched brackets
1600will trigger illegalproto warnings, declaring that they were
1601detected in the prototype for C<name>.
1602
1603The return value is C<true> if this is a valid prototype, and
1604C<false> if it is not, regardless of whether C<warn> was C<true> or
1605C<false>.
1606
1607Note that C<NULL> is a valid C<proto> and will always return C<true>.
1608
1609=cut
1610
1611 */
1612
1613bool
1614Perl_validate_proto(pTHX_ SV *name, SV *proto, bool warn)
1615{
1616 STRLEN len, origlen;
1617 char *p = proto ? SvPV(proto, len) : NULL;
1618 bool bad_proto = FALSE;
1619 bool in_brackets = FALSE;
1620 bool after_slash = FALSE;
1621 char greedy_proto = ' ';
1622 bool proto_after_greedy_proto = FALSE;
1623 bool must_be_last = FALSE;
1624 bool underscore = FALSE;
f791a21a 1625 bool bad_proto_after_underscore = FALSE;
fe788d6b
PM
1626
1627 PERL_ARGS_ASSERT_VALIDATE_PROTO;
1628
1629 if (!proto)
1630 return TRUE;
1631
1632 origlen = len;
1633 for (; len--; p++) {
1634 if (!isSPACE(*p)) {
1635 if (must_be_last)
1636 proto_after_greedy_proto = TRUE;
f791a21a
PM
1637 if (underscore) {
1638 if (!strchr(";@%", *p))
1639 bad_proto_after_underscore = TRUE;
1640 underscore = FALSE;
1641 }
fe788d6b
PM
1642 if (!strchr("$@%*;[]&\\_+", *p) || *p == '\0') {
1643 bad_proto = TRUE;
1644 }
1645 else {
fe788d6b
PM
1646 if (*p == '[')
1647 in_brackets = TRUE;
1648 else if (*p == ']')
1649 in_brackets = FALSE;
1650 else if ((*p == '@' || *p == '%') &&
1651 !after_slash &&
1652 !in_brackets ) {
1653 must_be_last = TRUE;
1654 greedy_proto = *p;
1655 }
1656 else if (*p == '_')
f791a21a 1657 underscore = TRUE;
fe788d6b
PM
1658 }
1659 if (*p == '\\')
1660 after_slash = TRUE;
1661 else
1662 after_slash = FALSE;
1663 }
1664 }
1665
1666 if (warn) {
b54d603d 1667 SV *tmpsv = newSVpvs_flags("", SVs_TEMP);
fe788d6b 1668 p -= origlen;
b54d603d
PM
1669 p = SvUTF8(proto)
1670 ? sv_uni_display(tmpsv, newSVpvn_flags(p, origlen, SVs_TEMP | SVf_UTF8),
1671 origlen, UNI_DISPLAY_ISPRINT)
1672 : pv_pretty(tmpsv, p, origlen, 60, NULL, NULL, PERL_PV_ESCAPE_NONASCII);
1673
fe788d6b
PM
1674 if (proto_after_greedy_proto)
1675 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1676 "Prototype after '%c' for %"SVf" : %s",
1677 greedy_proto, SVfARG(name), p);
50278ed0
PM
1678 if (in_brackets)
1679 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1680 "Missing ']' in prototype for %"SVf" : %s",
1681 SVfARG(name), p);
b54d603d 1682 if (bad_proto)
fe788d6b 1683 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
f791a21a
PM
1684 "Illegal character in prototype for %"SVf" : %s",
1685 SVfARG(name), p);
1686 if (bad_proto_after_underscore)
1687 Perl_warner(aTHX_ packWARN(WARN_ILLEGALPROTO),
1688 "Illegal character after '_' in prototype for %"SVf" : %s",
1689 SVfARG(name), p);
fe788d6b
PM
1690 }
1691
1692 return (! (proto_after_greedy_proto || bad_proto) );
1693}
1694
1695/*
ffb4593c
NT
1696 * S_incline
1697 * This subroutine has nothing to do with tilting, whether at windmills
1698 * or pinball tables. Its name is short for "increment line". It
57843af0 1699 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 1700 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
1701 * # line 500 "foo.pm"
1702 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
1703 */
1704
76e3520e 1705STATIC void
d9095cec 1706S_incline(pTHX_ const char *s)
463ee0b2 1707{
97aff369 1708 dVAR;
d9095cec
NC
1709 const char *t;
1710 const char *n;
1711 const char *e;
8818d409 1712 line_t line_num;
463ee0b2 1713
7918f24d
NC
1714 PERL_ARGS_ASSERT_INCLINE;
1715
83944c01 1716 COPLINE_INC_WITH_HERELINES;
451f421f
FC
1717 if (!PL_rsfp && !PL_parser->filtered && PL_lex_state == LEX_NORMAL
1718 && s+1 == PL_bufend && *s == ';') {
1719 /* fake newline in string eval */
1720 CopLINE_dec(PL_curcop);
1721 return;
1722 }
463ee0b2
LW
1723 if (*s++ != '#')
1724 return;
d4c19fe8
AL
1725 while (SPACE_OR_TAB(*s))
1726 s++;
73659bf1
GS
1727 if (strnEQ(s, "line", 4))
1728 s += 4;
1729 else
1730 return;
084592ab 1731 if (SPACE_OR_TAB(*s))
73659bf1 1732 s++;
4e553d73 1733 else
73659bf1 1734 return;
d4c19fe8
AL
1735 while (SPACE_OR_TAB(*s))
1736 s++;
463ee0b2
LW
1737 if (!isDIGIT(*s))
1738 return;
d4c19fe8 1739
463ee0b2
LW
1740 n = s;
1741 while (isDIGIT(*s))
1742 s++;
07714eb4 1743 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
26b6dc3f 1744 return;
bf4acbe4 1745 while (SPACE_OR_TAB(*s))
463ee0b2 1746 s++;
73659bf1 1747 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 1748 s++;
73659bf1
GS
1749 e = t + 1;
1750 }
463ee0b2 1751 else {
c35e046a
AL
1752 t = s;
1753 while (!isSPACE(*t))
1754 t++;
73659bf1 1755 e = t;
463ee0b2 1756 }
bf4acbe4 1757 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
1758 e++;
1759 if (*e != '\n' && *e != '\0')
1760 return; /* false alarm */
1761
8818d409
FC
1762 line_num = atoi(n)-1;
1763
f4dd75d9 1764 if (t - s > 0) {
d9095cec 1765 const STRLEN len = t - s;
3df32bda 1766
d36ee5be 1767 if (!PL_rsfp && !PL_parser->filtered) {
e66cf94c
RGS
1768 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1769 * to *{"::_<newfilename"} */
44867030
NC
1770 /* However, the long form of evals is only turned on by the
1771 debugger - usually they're "(eval %lu)" */
d36ee5be
FC
1772 GV * const cfgv = CopFILEGV(PL_curcop);
1773 if (cfgv) {
38bd7ad8
FC
1774 char smallbuf[128];
1775 STRLEN tmplen2 = len;
44867030 1776 char *tmpbuf2;
449dd039 1777 GV *gv2;
44867030
NC
1778
1779 if (tmplen2 + 2 <= sizeof smallbuf)
1780 tmpbuf2 = smallbuf;
1781 else
1782 Newx(tmpbuf2, tmplen2 + 2, char);
1783
38bd7ad8
FC
1784 tmpbuf2[0] = '_';
1785 tmpbuf2[1] = '<';
44867030
NC
1786
1787 memcpy(tmpbuf2 + 2, s, tmplen2);
1788 tmplen2 += 2;
1789
8a5ee598 1790 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
e5527e4b 1791 if (!isGV(gv2)) {
8a5ee598 1792 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
e5527e4b
RGS
1793 /* adjust ${"::_<newfilename"} to store the new file name */
1794 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
8818d409
FC
1795 /* The line number may differ. If that is the case,
1796 alias the saved lines that are in the array.
1797 Otherwise alias the whole array. */
1798 if (CopLINE(PL_curcop) == line_num) {
38bd7ad8
FC
1799 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(cfgv)));
1800 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(cfgv)));
8818d409 1801 }
38bd7ad8
FC
1802 else if (GvAV(cfgv)) {
1803 AV * const av = GvAV(cfgv);
8818d409
FC
1804 const I32 start = CopLINE(PL_curcop)+1;
1805 I32 items = AvFILLp(av) - start;
1806 if (items > 0) {
1807 AV * const av2 = GvAVn(gv2);
1808 SV **svp = AvARRAY(av) + start;
1809 I32 l = (I32)line_num+1;
1810 while (items--)
1811 av_store(av2, l++, SvREFCNT_inc(*svp++));
1812 }
1813 }
e5527e4b 1814 }
44867030
NC
1815
1816 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
d36ee5be 1817 }
e66cf94c 1818 }
05ec9bb3 1819 CopFILE_free(PL_curcop);
449dd039 1820 CopFILE_setn(PL_curcop, s, len);
f4dd75d9 1821 }
8818d409 1822 CopLINE_set(PL_curcop, line_num);
463ee0b2
LW
1823}
1824
21791330
FC
1825#define skipspace(s) skipspace_flags(s, 0)
1826
29595ff2 1827#ifdef PERL_MAD
cd81e915 1828/* skip space before PL_thistoken */
29595ff2
NC
1829
1830STATIC char *
5aaab254 1831S_skipspace0(pTHX_ char *s)
29595ff2 1832{
7918f24d
NC
1833 PERL_ARGS_ASSERT_SKIPSPACE0;
1834
29595ff2
NC
1835 s = skipspace(s);
1836 if (!PL_madskills)
1837 return s;
cd81e915
NC
1838 if (PL_skipwhite) {
1839 if (!PL_thiswhite)
6b29d1f5 1840 PL_thiswhite = newSVpvs("");
cd81e915
NC
1841 sv_catsv(PL_thiswhite, PL_skipwhite);
1842 sv_free(PL_skipwhite);
1843 PL_skipwhite = 0;
1844 }
1845 PL_realtokenstart = s - SvPVX(PL_linestr);
29595ff2
NC
1846 return s;
1847}
1848
cd81e915 1849/* skip space after PL_thistoken */
29595ff2
NC
1850
1851STATIC char *
5aaab254 1852S_skipspace1(pTHX_ char *s)
29595ff2 1853{
d4c19fe8 1854 const char *start = s;
29595ff2
NC
1855 I32 startoff = start - SvPVX(PL_linestr);
1856
7918f24d
NC
1857 PERL_ARGS_ASSERT_SKIPSPACE1;
1858
29595ff2
NC
1859 s = skipspace(s);
1860 if (!PL_madskills)
1861 return s;
1862 start = SvPVX(PL_linestr) + startoff;
cd81e915 1863 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1864 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1865 PL_thistoken = newSVpvn(tstart, start - tstart);
1866 }
1867 PL_realtokenstart = -1;
1868 if (PL_skipwhite) {
1869 if (!PL_nextwhite)
6b29d1f5 1870 PL_nextwhite = newSVpvs("");
cd81e915
NC
1871 sv_catsv(PL_nextwhite, PL_skipwhite);
1872 sv_free(PL_skipwhite);
1873 PL_skipwhite = 0;
29595ff2
NC
1874 }
1875 return s;
1876}
1877
1878STATIC char *
5aaab254 1879S_skipspace2(pTHX_ char *s, SV **svp)
29595ff2 1880{
c35e046a 1881 char *start;
c35e046a
AL
1882 const I32 startoff = s - SvPVX(PL_linestr);
1883
7918f24d
NC
1884 PERL_ARGS_ASSERT_SKIPSPACE2;
1885
29595ff2 1886 s = skipspace(s);
29595ff2
NC
1887 if (!PL_madskills || !svp)
1888 return s;
1889 start = SvPVX(PL_linestr) + startoff;
cd81e915 1890 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1891 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1892 PL_thistoken = newSVpvn(tstart, start - tstart);
1893 PL_realtokenstart = -1;
29595ff2 1894 }
cd81e915 1895 if (PL_skipwhite) {
29595ff2 1896 if (!*svp)
6b29d1f5 1897 *svp = newSVpvs("");
cd81e915
NC
1898 sv_setsv(*svp, PL_skipwhite);
1899 sv_free(PL_skipwhite);
1900 PL_skipwhite = 0;
29595ff2
NC
1901 }
1902
1903 return s;
1904}
1905#endif
1906
80a702cd 1907STATIC void
15f169a1 1908S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
80a702cd
RGS
1909{
1910 AV *av = CopFILEAVx(PL_curcop);
1911 if (av) {
b9f83d2f 1912 SV * const sv = newSV_type(SVt_PVMG);
5fa550fb 1913 if (orig_sv)
4e917a04 1914 sv_setsv_flags(sv, orig_sv, 0); /* no cow */
5fa550fb
NC
1915 else
1916 sv_setpvn(sv, buf, len);
80a702cd
RGS
1917 (void)SvIOK_on(sv);
1918 SvIV_set(sv, 0);
c70927a6 1919 av_store(av, CopLINE(PL_curcop), sv);
80a702cd
RGS
1920 }
1921}
1922
ffb4593c
NT
1923/*
1924 * S_skipspace
1925 * Called to gobble the appropriate amount and type of whitespace.
1926 * Skips comments as well.
1927 */
1928
76e3520e 1929STATIC char *
21791330 1930S_skipspace_flags(pTHX_ char *s, U32 flags)
a687059c 1931{
5db06880 1932#ifdef PERL_MAD
f0e67a1d
Z
1933 char *start = s;
1934#endif /* PERL_MAD */
21791330 1935 PERL_ARGS_ASSERT_SKIPSPACE_FLAGS;
f0e67a1d 1936#ifdef PERL_MAD
cd81e915
NC
1937 if (PL_skipwhite) {
1938 sv_free(PL_skipwhite);
f0e67a1d 1939 PL_skipwhite = NULL;
5db06880 1940 }
f0e67a1d 1941#endif /* PERL_MAD */
3280af22 1942 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 1943 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2 1944 s++;
f0e67a1d
Z
1945 } else {
1946 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1947 PL_bufptr = s;
21791330 1948 lex_read_space(flags | LEX_KEEP_PREVIOUS |
f0998909
Z
1949 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1950 LEX_NO_NEXT_CHUNK : 0));
3280af22 1951 s = PL_bufptr;
f0e67a1d
Z
1952 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1953 if (PL_linestart > PL_bufptr)
1954 PL_bufptr = PL_linestart;
1955 return s;
463ee0b2 1956 }
5db06880 1957#ifdef PERL_MAD
f0e67a1d
Z
1958 if (PL_madskills)
1959 PL_skipwhite = newSVpvn(start, s-start);
1960#endif /* PERL_MAD */
5db06880 1961 return s;
a687059c 1962}
378cc40b 1963
ffb4593c
NT
1964/*
1965 * S_check_uni
1966 * Check the unary operators to ensure there's no ambiguity in how they're
1967 * used. An ambiguous piece of code would be:
1968 * rand + 5
1969 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1970 * the +5 is its argument.
1971 */
1972
76e3520e 1973STATIC void
cea2e8a9 1974S_check_uni(pTHX)
ba106d47 1975{
97aff369 1976 dVAR;
d4c19fe8
AL
1977 const char *s;
1978 const char *t;
2f3197b3 1979
3280af22 1980 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 1981 return;
3280af22
NIS
1982 while (isSPACE(*PL_last_uni))
1983 PL_last_uni++;
c35e046a 1984 s = PL_last_uni;
8a2bca12 1985 while (isWORDCHAR_lazy_if(s,UTF) || *s == '-')
c35e046a 1986 s++;
3280af22 1987 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 1988 return;
6136c704 1989
9b387841
NC
1990 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1991 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1992 (int)(s - PL_last_uni), PL_last_uni);
2f3197b3
LW
1993}
1994
ffb4593c
NT
1995/*
1996 * LOP : macro to build a list operator. Its behaviour has been replaced
1997 * with a subroutine, S_lop() for which LOP is just another name.
1998 */
1999
a0d0e21e
LW
2000#define LOP(f,x) return lop(f,x,s)
2001
ffb4593c
NT
2002/*
2003 * S_lop
2004 * Build a list operator (or something that might be one). The rules:
2005 * - if we have a next token, then it's a list operator [why?]
2006 * - if the next thing is an opening paren, then it's a function
2007 * - else it's a list operator
2008 */
2009
76e3520e 2010STATIC I32
a0be28da 2011S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 2012{
97aff369 2013 dVAR;
7918f24d
NC
2014
2015 PERL_ARGS_ASSERT_LOP;
2016
6154021b 2017 pl_yylval.ival = f;
35c8bce7 2018 CLINE;
3280af22
NIS
2019 PL_expect = x;
2020 PL_bufptr = s;
2021 PL_last_lop = PL_oldbufptr;
eb160463 2022 PL_last_lop_op = (OPCODE)f;
5db06880
NC
2023#ifdef PERL_MAD
2024 if (PL_lasttoke)
78cdf107 2025 goto lstop;
5db06880 2026#else
3280af22 2027 if (PL_nexttoke)
78cdf107 2028 goto lstop;
5db06880 2029#endif
79072805 2030 if (*s == '(')
bbf60fe6 2031 return REPORT(FUNC);
29595ff2 2032 s = PEEKSPACE(s);
79072805 2033 if (*s == '(')
bbf60fe6 2034 return REPORT(FUNC);
78cdf107
Z
2035 else {
2036 lstop:
2037 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
2038 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
bbf60fe6 2039 return REPORT(LSTOP);
78cdf107 2040 }
79072805
LW
2041}
2042
5db06880
NC
2043#ifdef PERL_MAD
2044 /*
2045 * S_start_force
2046 * Sets up for an eventual force_next(). start_force(0) basically does
2047 * an unshift, while start_force(-1) does a push. yylex removes items
2048 * on the "pop" end.
2049 */
2050
2051STATIC void
2052S_start_force(pTHX_ int where)
2053{
2054 int i;
2055
cd81e915 2056 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
5db06880 2057 where = PL_lasttoke;
cd81e915
NC
2058 assert(PL_curforce < 0 || PL_curforce == where);
2059 if (PL_curforce != where) {
5db06880
NC
2060 for (i = PL_lasttoke; i > where; --i) {
2061 PL_nexttoke[i] = PL_nexttoke[i-1];
2062 }
2063 PL_lasttoke++;
2064 }
cd81e915 2065 if (PL_curforce < 0) /* in case of duplicate start_force() */
5db06880 2066 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
cd81e915
NC
2067 PL_curforce = where;
2068 if (PL_nextwhite) {
5db06880 2069 if (PL_madskills)
6b29d1f5 2070 curmad('^', newSVpvs(""));
cd81e915 2071 CURMAD('_', PL_nextwhite);
5db06880
NC
2072 }
2073}
2074
2075STATIC void
2076S_curmad(pTHX_ char slot, SV *sv)
2077{
2078 MADPROP **where;
2079
2080 if (!sv)
2081 return;
cd81e915
NC
2082 if (PL_curforce < 0)
2083 where = &PL_thismad;
5db06880 2084 else
cd81e915 2085 where = &PL_nexttoke[PL_curforce].next_mad;
5db06880 2086
cd81e915 2087 if (PL_faketokens)
76f68e9b 2088 sv_setpvs(sv, "");
5db06880
NC
2089 else {
2090 if (!IN_BYTES) {
2091 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
2092 SvUTF8_on(sv);
2093 else if (PL_encoding) {
2094 sv_recode_to_utf8(sv, PL_encoding);
2095 }
2096 }
2097 }
2098
2099 /* keep a slot open for the head of the list? */
2100 if (slot != '_' && *where && (*where)->mad_key == '^') {
2101 (*where)->mad_key = slot;
daba3364 2102 sv_free(MUTABLE_SV(((*where)->mad_val)));
5db06880
NC
2103 (*where)->mad_val = (void*)sv;
2104 }
2105 else
2106 addmad(newMADsv(slot, sv), where, 0);
2107}
2108#else
b3f24c00
MHM
2109# define start_force(where) NOOP
2110# define curmad(slot, sv) NOOP
5db06880
NC
2111#endif
2112
ffb4593c
NT
2113/*
2114 * S_force_next
9cbb5ea2 2115 * When the lexer realizes it knows the next token (for instance,
ffb4593c 2116 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2 2117 * to know what token to return the next time the lexer is called. Caller
5db06880
NC
2118 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
2119 * and possibly PL_expect to ensure the lexer handles the token correctly.
ffb4593c
NT
2120 */
2121
4e553d73 2122STATIC void
cea2e8a9 2123S_force_next(pTHX_ I32 type)
79072805 2124{
97aff369 2125 dVAR;
704d4215
GG
2126#ifdef DEBUGGING
2127 if (DEBUG_T_TEST) {
2128 PerlIO_printf(Perl_debug_log, "### forced token:\n");
f05d7009 2129 tokereport(type, &NEXTVAL_NEXTTOKE);
704d4215
GG
2130 }
2131#endif
5db06880 2132#ifdef PERL_MAD
cd81e915 2133 if (PL_curforce < 0)
5db06880 2134 start_force(PL_lasttoke);
cd81e915 2135 PL_nexttoke[PL_curforce].next_type = type;
5db06880
NC
2136 if (PL_lex_state != LEX_KNOWNEXT)
2137 PL_lex_defer = PL_lex_state;
2138 PL_lex_state = LEX_KNOWNEXT;
2139 PL_lex_expect = PL_expect;
cd81e915 2140 PL_curforce = -1;
5db06880 2141#else
3280af22
NIS
2142 PL_nexttype[PL_nexttoke] = type;
2143 PL_nexttoke++;
2144 if (PL_lex_state != LEX_KNOWNEXT) {
2145 PL_lex_defer = PL_lex_state;
2146 PL_lex_expect = PL_expect;
2147 PL_lex_state = LEX_KNOWNEXT;
79072805 2148 }
5db06880 2149#endif
79072805
LW
2150}
2151
28ac2b49
Z
2152void
2153Perl_yyunlex(pTHX)
2154{
a7aaec61
Z
2155 int yyc = PL_parser->yychar;
2156 if (yyc != YYEMPTY) {
2157 if (yyc) {
2158 start_force(-1);
2159 NEXTVAL_NEXTTOKE = PL_parser->yylval;
2160 if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
78cdf107 2161 PL_lex_allbrackets--;
a7aaec61 2162 PL_lex_brackets--;
78cdf107
Z
2163 yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
2164 } else if (yyc == '('/*)*/) {
2165 PL_lex_allbrackets--;
2166 yyc |= (2<<24);
a7aaec61
Z
2167 }
2168 force_next(yyc);
2169 }
28ac2b49
Z
2170 PL_parser->yychar = YYEMPTY;
2171 }
2172}
2173
d0a148a6 2174STATIC SV *
15f169a1 2175S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
d0a148a6 2176{
97aff369 2177 dVAR;
740cce10 2178 SV * const sv = newSVpvn_utf8(start, len,
eaf7a4d2
CS
2179 !IN_BYTES
2180 && UTF
2181 && !is_ascii_string((const U8*)start, len)
740cce10 2182 && is_utf8_string((const U8*)start, len));
d0a148a6
NC
2183 return sv;
2184}
2185
ffb4593c
NT
2186/*
2187 * S_force_word
2188 * When the lexer knows the next thing is a word (for instance, it has
2189 * just seen -> and it knows that the next char is a word char, then
02b34bbe
DM
2190 * it calls S_force_word to stick the next word into the PL_nexttoke/val
2191 * lookahead.
ffb4593c
NT
2192 *
2193 * Arguments:
b1b65b59 2194 * char *start : buffer position (must be within PL_linestr)
02b34bbe 2195 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
ffb4593c
NT
2196 * int check_keyword : if true, Perl checks to make sure the word isn't
2197 * a keyword (do this if the word is a label, e.g. goto FOO)
2198 * int allow_pack : if true, : characters will also be allowed (require,
2199 * use, etc. do this)
9cbb5ea2 2200 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
2201 */
2202
76e3520e 2203STATIC char *
345b3785 2204S_force_word(pTHX_ char *start, int token, int check_keyword, int allow_pack)
79072805 2205{
97aff369 2206 dVAR;
eb578fdb 2207 char *s;
463ee0b2 2208 STRLEN len;
4e553d73 2209
7918f24d
NC
2210 PERL_ARGS_ASSERT_FORCE_WORD;
2211
29595ff2 2212 start = SKIPSPACE1(start);
463ee0b2 2213 s = start;
7e2040f0 2214 if (isIDFIRST_lazy_if(s,UTF) ||
345b3785 2215 (allow_pack && *s == ':') )
a0d0e21e 2216 {
3280af22 2217 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
01b5ef50
FC
2218 if (check_keyword) {
2219 char *s2 = PL_tokenbuf;
2220 if (allow_pack && len > 6 && strnEQ(s2, "CORE::", 6))
2221 s2 += 6, len -= 6;
2222 if (keyword(s2, len, 0))
463ee0b2 2223 return start;
01b5ef50 2224 }
cd81e915 2225 start_force(PL_curforce);
5db06880
NC
2226 if (PL_madskills)
2227 curmad('X', newSVpvn(start,s-start));
463ee0b2 2228 if (token == METHOD) {
29595ff2 2229 s = SKIPSPACE1(s);
463ee0b2 2230 if (*s == '(')
3280af22 2231 PL_expect = XTERM;
463ee0b2 2232 else {
3280af22 2233 PL_expect = XOPERATOR;
463ee0b2 2234 }
79072805 2235 }
e74e6b3d 2236 if (PL_madskills)
63575281 2237 curmad('g', newSVpvs( "forced" ));
9ded7720 2238 NEXTVAL_NEXTTOKE.opval
d0a148a6
NC
2239 = (OP*)newSVOP(OP_CONST,0,
2240 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
9ded7720 2241 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
79072805
LW
2242 force_next(token);
2243 }
2244 return s;
2245}
2246
ffb4593c
NT
2247/*
2248 * S_force_ident
9cbb5ea2 2249 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
2250 * text only contains the "foo" portion. The first argument is a pointer
2251 * to the "foo", and the second argument is the type symbol to prefix.
2252 * Forces the next token to be a "WORD".
9cbb5ea2 2253 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
2254 */
2255
76e3520e 2256STATIC void
5aaab254 2257S_force_ident(pTHX_ const char *s, int kind)
79072805 2258{
97aff369 2259 dVAR;
7918f24d
NC
2260
2261 PERL_ARGS_ASSERT_FORCE_IDENT;
2262
c9b48522
DD
2263 if (s[0]) {
2264 const STRLEN len = s[1] ? strlen(s) : 1; /* s = "\"" see yylex */
728847b1
BF
2265 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len,
2266 UTF ? SVf_UTF8 : 0));
cd81e915 2267 start_force(PL_curforce);
9ded7720 2268 NEXTVAL_NEXTTOKE.opval = o;
79072805 2269 force_next(WORD);
748a9306 2270 if (kind) {
11343788 2271 o->op_private = OPpCONST_ENTERED;
55497cff
PP
2272 /* XXX see note in pp_entereval() for why we forgo typo
2273 warnings if the symbol must be introduced in an eval.
2274 GSAR 96-10-12 */
90e5519e 2275 gv_fetchpvn_flags(s, len,
728847b1
BF
2276 (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2277 : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ),
90e5519e
NC
2278 kind == '$' ? SVt_PV :
2279 kind == '@' ? SVt_PVAV :
2280 kind == '%' ? SVt_PVHV :
a0d0e21e 2281 SVt_PVGV
90e5519e 2282 );
748a9306 2283 }
79072805
LW
2284 }
2285}
2286
3f33d153
FC
2287static void
2288S_force_ident_maybe_lex(pTHX_ char pit)
2289{
2290 start_force(PL_curforce);
2291 NEXTVAL_NEXTTOKE.ival = pit;
2292 force_next('p');
2293}
2294
1571675a
GS
2295NV
2296Perl_str_to_version(pTHX_ SV *sv)
2297{
2298 NV retval = 0.0;
2299 NV nshift = 1.0;
2300 STRLEN len;
cfd0369c 2301 const char *start = SvPV_const(sv,len);
9d4ba2ae 2302 const char * const end = start + len;
504618e9 2303 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
7918f24d
NC
2304
2305 PERL_ARGS_ASSERT_STR_TO_VERSION;
2306
1571675a 2307 while (start < end) {
ba210ebe 2308 STRLEN skip;
1571675a
GS
2309 UV n;
2310 if (utf)
9041c2e3 2311 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
2312 else {
2313 n = *(U8*)start;
2314 skip = 1;
2315 }
2316 retval += ((NV)n)/nshift;
2317 start += skip;
2318 nshift *= 1000;
2319 }
2320 return retval;
2321}
2322
4e553d73 2323/*
ffb4593c
NT
2324 * S_force_version
2325 * Forces the next token to be a version number.
e759cc13
RGS
2326 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2327 * and if "guessing" is TRUE, then no new token is created (and the caller
2328 * must use an alternative parsing method).
ffb4593c
NT
2329 */
2330
76e3520e 2331STATIC char *
e759cc13 2332S_force_version(pTHX_ char *s, int guessing)
89bfa8cd 2333{
97aff369 2334 dVAR;
5f66b61c 2335 OP *version = NULL;
44dcb63b 2336 char *d;
5db06880
NC
2337#ifdef PERL_MAD
2338 I32 startoff = s - SvPVX(PL_linestr);
2339#endif
89bfa8cd 2340
7918f24d
NC
2341 PERL_ARGS_ASSERT_FORCE_VERSION;
2342
29595ff2 2343 s = SKIPSPACE1(s);
89bfa8cd 2344
44dcb63b 2345 d = s;
dd629d5b 2346 if (*d == 'v')
44dcb63b 2347 d++;
44dcb63b 2348 if (isDIGIT(*d)) {
e759cc13
RGS
2349 while (isDIGIT(*d) || *d == '_' || *d == '.')
2350 d++;
5db06880
NC
2351#ifdef PERL_MAD
2352 if (PL_madskills) {
cd81e915 2353 start_force(PL_curforce);
5db06880
NC
2354 curmad('X', newSVpvn(s,d-s));
2355 }
2356#endif
4e4da3ac 2357 if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
dd629d5b 2358 SV *ver;
8d08d9ba 2359#ifdef USE_LOCALE_NUMERIC
909d3787
KW
2360 char *loc = savepv(setlocale(LC_NUMERIC, NULL));
2361 setlocale(LC_NUMERIC, "C");
8d08d9ba 2362#endif
6154021b 2363 s = scan_num(s, &pl_yylval);
8d08d9ba
DG
2364#ifdef USE_LOCALE_NUMERIC
2365 setlocale(LC_NUMERIC, loc);
909d3787 2366 Safefree(loc);
8d08d9ba 2367#endif
6154021b 2368 version = pl_yylval.opval;
dd629d5b
GS
2369 ver = cSVOPx(version)->op_sv;
2370 if (SvPOK(ver) && !SvNIOK(ver)) {
862a34c6 2371 SvUPGRADE(ver, SVt_PVNV);
9d6ce603 2372 SvNV_set(ver, str_to_version(ver));
1571675a 2373 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 2374 }
89bfa8cd 2375 }
5db06880
NC
2376 else if (guessing) {
2377#ifdef PERL_MAD
2378 if (PL_madskills) {
cd81e915
NC
2379 sv_free(PL_nextwhite); /* let next token collect whitespace */
2380 PL_nextwhite = 0;
5db06880
NC
2381 s = SvPVX(PL_linestr) + startoff;
2382 }
2383#endif
e759cc13 2384 return s;
5db06880 2385 }
89bfa8cd
PP
2386 }
2387
5db06880
NC
2388#ifdef PERL_MAD
2389 if (PL_madskills && !version) {
cd81e915
NC
2390 sv_free(PL_nextwhite); /* let next token collect whitespace */
2391 PL_nextwhite = 0;
5db06880
NC
2392 s = SvPVX(PL_linestr) + startoff;
2393 }
2394#endif
89bfa8cd 2395 /* NOTE: The parser sees the package name and the VERSION swapped */
cd81e915 2396 start_force(PL_curforce);
9ded7720 2397 NEXTVAL_NEXTTOKE.opval = version;
4e553d73 2398 force_next(WORD);
89bfa8cd 2399
e759cc13 2400 return s;
89bfa8cd
PP
2401}
2402
ffb4593c 2403/*
91152fc1
DG
2404 * S_force_strict_version
2405 * Forces the next token to be a version number using strict syntax rules.
2406 */
2407
2408STATIC char *
2409S_force_strict_version(pTHX_ char *s)
2410{
2411 dVAR;
2412 OP *version = NULL;
2413#ifdef PERL_MAD
2414 I32 startoff = s - SvPVX(PL_linestr);
2415#endif
2416 const char *errstr = NULL;
2417
2418 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2419
2420 while (isSPACE(*s)) /* leading whitespace */
2421 s++;
2422
2423 if (is_STRICT_VERSION(s,&errstr)) {
2424 SV *ver = newSV(0);
2425 s = (char *)scan_version(s, ver, 0);
2426 version = newSVOP(OP_CONST, 0, ver);
2427 }
4e4da3ac
Z
2428 else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2429 (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2430 {
91152fc1
DG
2431 PL_bufptr = s;
2432 if (errstr)
2433 yyerror(errstr); /* version required */
2434 return s;
2435 }
2436
2437#ifdef PERL_MAD
2438 if (PL_madskills && !version) {
2439 sv_free(PL_nextwhite); /* let next token collect whitespace */
2440 PL_nextwhite = 0;
2441 s = SvPVX(PL_linestr) + startoff;
2442 }
2443#endif
2444 /* NOTE: The parser sees the package name and the VERSION swapped */
2445 start_force(PL_curforce);
2446 NEXTVAL_NEXTTOKE.opval = version;
2447 force_next(WORD);
2448
2449 return s;
2450}
2451
2452/*
ffb4593c
NT
2453 * S_tokeq
2454 * Tokenize a quoted string passed in as an SV. It finds the next
2455 * chunk, up to end of string or a backslash. It may make a new
2456 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2457 * turns \\ into \.
2458 */
2459
76e3520e 2460STATIC SV *
cea2e8a9 2461S_tokeq(pTHX_ SV *sv)
79072805 2462{
97aff369 2463 dVAR;
eb578fdb
KW
2464 char *s;
2465 char *send;
2466 char *d;
b3ac6de7
IZ
2467 STRLEN len = 0;
2468 SV *pv = sv;
79072805 2469
7918f24d
NC
2470 PERL_ARGS_ASSERT_TOKEQ;
2471
79072805 2472 if (!SvLEN(sv))
b3ac6de7 2473 goto finish;
79072805 2474
a0d0e21e 2475 s = SvPV_force(sv, len);
21a311ee 2476 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 2477 goto finish;
463ee0b2 2478 send = s + len;
dcb21ed6
NC
2479 /* This is relying on the SV being "well formed" with a trailing '\0' */
2480 while (s < send && !(*s == '\\' && s[1] == '\\'))
79072805
LW
2481 s++;
2482 if (s == send)
b3ac6de7 2483 goto finish;
79072805 2484 d = s;
be4731d2 2485 if ( PL_hints & HINT_NEW_STRING ) {
59cd0e26 2486 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
be4731d2 2487 }
79072805
LW
2488 while (s < send) {
2489 if (*s == '\\') {
a0d0e21e 2490 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
2491 s++; /* all that, just for this */
2492 }
2493 *d++ = *s++;
2494 }
2495 *d = '\0';
95a20fc0 2496 SvCUR_set(sv, d - SvPVX_const(sv));
b3ac6de7 2497 finish:
3280af22 2498 if ( PL_hints & HINT_NEW_STRING )
eb0d8d16 2499 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
79072805
LW
2500 return sv;
2501}
2502
ffb4593c
NT
2503/*
2504 * Now come three functions related to double-quote context,
2505 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2506 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2507 * interact with PL_lex_state, and create fake ( ... ) argument lists
2508 * to handle functions and concatenation.
ecd24171
DM
2509 * For example,
2510 * "foo\lbar"
2511 * is tokenised as
2512 * stringify ( const[foo] concat lcfirst ( const[bar] ) )
ffb4593c
NT
2513 */
2514
2515/*
2516 * S_sublex_start
6154021b 2517 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
ffb4593c
NT
2518 *
2519 * Pattern matching will set PL_lex_op to the pattern-matching op to
6154021b 2520 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
ffb4593c
NT
2521 *
2522 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2523 *
2524 * Everything else becomes a FUNC.
2525 *
2526 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2527 * had an OP_CONST or OP_READLINE). This just sets us up for a
2528 * call to S_sublex_push().
2529 */
2530
76e3520e 2531STATIC I32
cea2e8a9 2532S_sublex_start(pTHX)
79072805 2533{
97aff369 2534 dVAR;
eb578fdb 2535 const I32 op_type = pl_yylval.ival;
79072805
LW
2536
2537 if (op_type == OP_NULL) {
6154021b 2538 pl_yylval.opval = PL_lex_op;
5f66b61c 2539 PL_lex_op = NULL;
79072805
LW
2540 return THING;
2541 }
2542 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 2543 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
2544
2545 if (SvTYPE(sv) == SVt_PVIV) {
2546 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2547 STRLEN len;
96a5add6 2548 const char * const p = SvPV_const(sv, len);
740cce10 2549 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
b3ac6de7
IZ
2550 SvREFCNT_dec(sv);
2551 sv = nsv;
4e553d73 2552 }
6154021b 2553 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
a0714e2c 2554 PL_lex_stuff = NULL;
6f33ba73
RGS
2555 /* Allow <FH> // "foo" */
2556 if (op_type == OP_READLINE)
2557 PL_expect = XTERMORDORDOR;
79072805
LW
2558 return THING;
2559 }
e3f73d4e 2560 else if (op_type == OP_BACKTICK && PL_lex_op) {
06df4f41 2561 /* readpipe() was overridden */
e3f73d4e 2562 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
6154021b 2563 pl_yylval.opval = PL_lex_op;
9b201d7d 2564 PL_lex_op = NULL;
e3f73d4e
RGS
2565 PL_lex_stuff = NULL;
2566 return THING;
2567 }
79072805 2568
3280af22 2569 PL_sublex_info.super_state = PL_lex_state;
eac04b2e 2570 PL_sublex_info.sub_inwhat = (U16)op_type;
3280af22
NIS
2571 PL_sublex_info.sub_op = PL_lex_op;
2572 PL_lex_state = LEX_INTERPPUSH;
55497cff 2573
3280af22
NIS
2574 PL_expect = XTERM;
2575 if (PL_lex_op) {
6154021b 2576 pl_yylval.opval = PL_lex_op;
5f66b61c 2577 PL_lex_op = NULL;
55497cff
PP
2578 return PMFUNC;
2579 }
2580 else
2581 return FUNC;
2582}
2583
ffb4593c
NT
2584/*
2585 * S_sublex_push
2586 * Create a new scope to save the lexing state. The scope will be
2587 * ended in S_sublex_done. Returns a '(', starting the function arguments
2588 * to the uc, lc, etc. found before.
2589 * Sets PL_lex_state to LEX_INTERPCONCAT.
2590 */
2591
76e3520e 2592STATIC I32
cea2e8a9 2593S_sublex_push(pTHX)
55497cff 2594{
27da23d5 2595 dVAR;
78a635de 2596 LEXSHARED *shared;
801d32ac 2597 const bool is_heredoc = PL_multi_close == '<';
f46d017c 2598 ENTER;
55497cff 2599
3280af22 2600 PL_lex_state = PL_sublex_info.super_state;
651b5b28 2601 SAVEBOOL(PL_lex_dojoin);
3280af22 2602 SAVEI32(PL_lex_brackets);
78cdf107 2603 SAVEI32(PL_lex_allbrackets);
b27dce25 2604 SAVEI32(PL_lex_formbrack);
78cdf107 2605 SAVEI8(PL_lex_fakeeof);
3280af22
NIS
2606 SAVEI32(PL_lex_casemods);
2607 SAVEI32(PL_lex_starts);
651b5b28 2608 SAVEI8(PL_lex_state);
7cc34111 2609 SAVESPTR(PL_lex_repl);
7766f137 2610 SAVEVPTR(PL_lex_inpat);
98246f1e 2611 SAVEI16(PL_lex_inwhat);
ffdb8b16 2612 if (is_heredoc)
b42366d4 2613 {
ffdb8b16 2614 SAVECOPLINE(PL_curcop);
b42366d4 2615 SAVEI32(PL_multi_end);
851b527a
FC
2616 SAVEI32(PL_parser->herelines);
2617 PL_parser->herelines = 0;
b42366d4
FC
2618 }
2619 SAVEI8(PL_multi_close);
3280af22 2620 SAVEPPTR(PL_bufptr);
8452ff4b 2621 SAVEPPTR(PL_bufend);
3280af22
NIS
2622 SAVEPPTR(PL_oldbufptr);
2623 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
2624 SAVEPPTR(PL_last_lop);
2625 SAVEPPTR(PL_last_uni);
3280af22
NIS
2626 SAVEPPTR(PL_linestart);
2627 SAVESPTR(PL_linestr);
8edd5f42
RGS
2628 SAVEGENERICPV(PL_lex_brackstack);
2629 SAVEGENERICPV(PL_lex_casestack);
78a635de 2630 SAVEGENERICPV(PL_parser->lex_shared);
3a54fd60 2631 SAVEBOOL(PL_parser->lex_re_reparsing);
ffdb8b16 2632 SAVEI32(PL_copline);
3280af22 2633
99bd9d90 2634 /* The here-doc parser needs to be able to peek into outer lexing
60f40a38
FC
2635 scopes to find the body of the here-doc. So we put PL_linestr and
2636 PL_bufptr into lex_shared, to ‘share’ those values.
99bd9d90 2637 */
60f40a38
FC
2638 PL_parser->lex_shared->ls_linestr = PL_linestr;
2639 PL_parser->lex_shared->ls_bufptr = PL_bufptr;
99bd9d90 2640
3280af22 2641 PL_linestr = PL_lex_stuff;
7cc34111 2642 PL_lex_repl = PL_sublex_info.repl;
a0714e2c 2643 PL_lex_stuff = NULL;
7cc34111 2644 PL_sublex_info.repl = NULL;
3280af22 2645
9cbb5ea2
GS
2646 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2647 = SvPVX(PL_linestr);
3280af22 2648 PL_bufend += SvCUR(PL_linestr);
bd61b366 2649 PL_last_lop = PL_last_uni = NULL;
3280af22 2650 SAVEFREESV(PL_linestr);
4dc843bc 2651 if (PL_lex_repl) SAVEFREESV(PL_lex_repl);
3280af22
NIS
2652
2653 PL_lex_dojoin = FALSE;
b27dce25 2654 PL_lex_brackets = PL_lex_formbrack = 0;
78cdf107
Z
2655 PL_lex_allbrackets = 0;
2656 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
a02a5408
JC
2657 Newx(PL_lex_brackstack, 120, char);
2658 Newx(PL_lex_casestack, 12, char);
3280af22
NIS
2659 PL_lex_casemods = 0;
2660 *PL_lex_casestack = '\0';
2661 PL_lex_starts = 0;
2662 PL_lex_state = LEX_INTERPCONCAT;
ffdb8b16 2663 if (is_heredoc)
6ddcf93b 2664 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
ffdb8b16 2665 PL_copline = NOLINE;
78a635de
FC
2666
2667 Newxz(shared, 1, LEXSHARED);
2668 shared->ls_prev = PL_parser->lex_shared;
2669 PL_parser->lex_shared = shared;
3280af22
NIS
2670
2671 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
bb16bae8 2672 if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
3280af22
NIS
2673 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2674 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 2675 else
5f66b61c 2676 PL_lex_inpat = NULL;
79072805 2677
3a54fd60
DM
2678 PL_parser->lex_re_reparsing = cBOOL(PL_in_eval & EVAL_RE_REPARSING);
2679 PL_in_eval &= ~EVAL_RE_REPARSING;
2680
55497cff 2681 return '(';
79072805
LW
2682}
2683
ffb4593c
NT
2684/*
2685 * S_sublex_done
2686 * Restores lexer state after a S_sublex_push.
2687 */
2688
76e3520e 2689STATIC I32
cea2e8a9 2690S_sublex_done(pTHX)
79072805 2691{
27da23d5 2692 dVAR;
3280af22 2693 if (!PL_lex_starts++) {
396482e1 2694 SV * const sv = newSVpvs("");
9aa983d2
JH
2695 if (SvUTF8(PL_linestr))
2696 SvUTF8_on(sv);
3280af22 2697 PL_expect = XOPERATOR;
6154021b 2698 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
2699 return THING;
2700 }
2701
3280af22
NIS
2702 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2703 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 2704 return yylex();
79072805
LW
2705 }
2706
ffb4593c 2707 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
bb16bae8 2708 assert(PL_lex_inwhat != OP_TRANSR);
3280af22
NIS
2709 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2710 PL_linestr = PL_lex_repl;
2711 PL_lex_inpat = 0;
2712 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2713 PL_bufend += SvCUR(PL_linestr);
bd61b366 2714 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2715 PL_lex_dojoin = FALSE;
2716 PL_lex_brackets = 0;
78cdf107
Z
2717 PL_lex_allbrackets = 0;
2718 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
3280af22
NIS
2719 PL_lex_casemods = 0;
2720 *PL_lex_casestack = '\0';
2721 PL_lex_starts = 0;
25da4f38 2722 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
2723 PL_lex_state = LEX_INTERPNORMAL;
2724 PL_lex_starts++;
e9fa98b2
HS
2725 /* we don't clear PL_lex_repl here, so that we can check later
2726 whether this is an evalled subst; that means we rely on the
2727 logic to ensure sublex_done() is called again only via the
2728 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 2729 }
e9fa98b2 2730 else {
3280af22 2731 PL_lex_state = LEX_INTERPCONCAT;
a0714e2c 2732 PL_lex_repl = NULL;
e9fa98b2 2733 }
ffdb8b16
FC
2734 if (SvTYPE(PL_linestr) >= SVt_PVNV) {
2735 CopLINE(PL_curcop) +=
2736 ((XPVNV*)SvANY(PL_linestr))->xnv_u.xpad_cop_seq.xlow
851b527a
FC
2737 + PL_parser->herelines;
2738 PL_parser->herelines = 0;
ffdb8b16 2739 }
79072805 2740 return ',';
ffed7fef
LW
2741 }
2742 else {
b42366d4 2743 const line_t l = CopLINE(PL_curcop);
5db06880
NC
2744#ifdef PERL_MAD
2745 if (PL_madskills) {
cd81e915
NC
2746 if (PL_thiswhite) {
2747 if (!PL_endwhite)
6b29d1f5 2748 PL_endwhite = newSVpvs("");
cd81e915
NC
2749 sv_catsv(PL_endwhite, PL_thiswhite);
2750 PL_thiswhite = 0;
2751 }
2752 if (PL_thistoken)
76f68e9b 2753 sv_setpvs(PL_thistoken,"");
5db06880 2754 else
cd81e915 2755 PL_realtokenstart = -1;
5db06880
NC
2756 }
2757#endif
f46d017c 2758 LEAVE;
b42366d4 2759 if (PL_multi_close == '<')
851b527a 2760 PL_parser->herelines += l - PL_multi_end;
3280af22
NIS
2761 PL_bufend = SvPVX(PL_linestr);
2762 PL_bufend += SvCUR(PL_linestr);
2763 PL_expect = XOPERATOR;
09bef843 2764 PL_sublex_info.sub_inwhat = 0;
79072805 2765 return ')';
ffed7fef
LW
2766 }
2767}
2768
6f613c73
KW
2769PERL_STATIC_INLINE SV*
2770S_get_and_check_backslash_N_name(pTHX_ const char* s, const char* const e)
2771{
140b12ad
KW
2772 /* <s> points to first character of interior of \N{}, <e> to one beyond the
2773 * interior, hence to the "}". Finds what the name resolves to, returning
2774 * an SV* containing it; NULL if no valid one found */
2775
dd2b1b72 2776 SV* res = newSVpvn_flags(s, e - s, UTF ? SVf_UTF8 : 0);
6f613c73 2777
0c415a79
KW
2778 HV * table;
2779 SV **cvp;
2780 SV *cv;
2781 SV *rv;
2782 HV *stash;
2783 const U8* first_bad_char_loc;
2784 const char* backslash_ptr = s - 3; /* Points to the <\> of \N{... */
2785
6f613c73
KW
2786 PERL_ARGS_ASSERT_GET_AND_CHECK_BACKSLASH_N_NAME;
2787
107160e2
KW
2788 if (UTF && ! is_utf8_string_loc((U8 *) backslash_ptr,
2789 e - backslash_ptr,
2790 &first_bad_char_loc))
2791 {
2792 /* If warnings are on, this will print a more detailed analysis of what
2793 * is wrong than the error message below */
c80e42f3 2794 utf8n_to_uvchr(first_bad_char_loc,
107160e2
KW
2795 e - ((char *) first_bad_char_loc),
2796 NULL, 0);
2797
2798 /* We deliberately don't try to print the malformed character, which
2799 * might not print very well; it also may be just the first of many
2800 * malformations, so don't print what comes after it */
2801 yyerror(Perl_form(aTHX_
2802 "Malformed UTF-8 character immediately after '%.*s'",
2803 (int) (first_bad_char_loc - (U8 *) backslash_ptr), backslash_ptr));
2804 return NULL;
2805 }
2806
2807 res = new_constant( NULL, 0, "charnames", res, NULL, backslash_ptr,
2808 /* include the <}> */
2809 e - backslash_ptr + 1);
6f613c73 2810 if (! SvPOK(res)) {
b6407c49 2811 SvREFCNT_dec_NN(res);
6f613c73
KW
2812 return NULL;
2813 }
2814
0c415a79
KW
2815 /* See if the charnames handler is the Perl core's, and if so, we can skip
2816 * the validation needed for a user-supplied one, as Perl's does its own
2817 * validation. */
2818 table = GvHV(PL_hintgv); /* ^H */
2819 cvp = hv_fetchs(table, "charnames", FALSE);
67a057d6
FC
2820 if (cvp && (cv = *cvp) && SvROK(cv) && ((rv = SvRV(cv)) != NULL)
2821 && SvTYPE(rv) == SVt_PVCV && ((stash = CvSTASH(rv)) != NULL))
0c415a79
KW
2822 {
2823 const char * const name = HvNAME(stash);
2824 if strEQ(name, "_charnames") {
2825 return res;
2826 }
2827 }
2828
bde9e88d
KW
2829 /* Here, it isn't Perl's charname handler. We can't rely on a
2830 * user-supplied handler to validate the input name. For non-ut8 input,
2831 * look to see that the first character is legal. Then loop through the
2832 * rest checking that each is a continuation */
6f613c73 2833
b6ba1137
KW
2834 /* This code needs to be sync'ed with a regex in _charnames.pm which does
2835 * the same thing */
2836
b6ba1137 2837 if (! UTF) {
bde9e88d 2838 if (! isALPHAU(*s)) {
b6ba1137
KW
2839 goto bad_charname;
2840 }
bde9e88d
KW
2841 s++;
2842 while (s < e) {
2843 if (! isCHARNAME_CONT(*s)) {
b6ba1137
KW
2844 goto bad_charname;
2845 }
3036c853 2846 if (*s == ' ' && *(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2847 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2848 "A sequence of multiple spaces in a charnames "
2849 "alias definition is deprecated");
bd299e29 2850 }
bde9e88d 2851 s++;
b6ba1137 2852 }
3036c853 2853 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2854 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2855 "Trailing white-space in a charnames alias "
2856 "definition is deprecated");
bd299e29 2857 }
b6ba1137
KW
2858 }
2859 else {
bde9e88d
KW
2860 /* Similarly for utf8. For invariants can check directly; for other
2861 * Latin1, can calculate their code point and check; otherwise use a
2862 * swash */
2863 if (UTF8_IS_INVARIANT(*s)) {
2864 if (! isALPHAU(*s)) {
140b12ad
KW
2865 goto bad_charname;
2866 }
bde9e88d
KW
2867 s++;
2868 } else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
94bb8c36 2869 if (! isALPHAU(TWO_BYTE_UTF8_TO_NATIVE(*s, *(s+1)))) {
b6ba1137 2870 goto bad_charname;
6f613c73 2871 }
bde9e88d 2872 s += 2;
6f613c73 2873 }
bde9e88d
KW
2874 else {
2875 if (! PL_utf8_charname_begin) {
2876 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2877 PL_utf8_charname_begin = _core_swash_init("utf8",
2878 "_Perl_Charname_Begin",
2879 &PL_sv_undef,
2880 1, 0, NULL, &flags);
2881 }
2882 if (! swash_fetch(PL_utf8_charname_begin, (U8 *) s, TRUE)) {
2883 goto bad_charname;
2884 }
2885 s += UTF8SKIP(s);
2886 }
2887
2888 while (s < e) {
2889 if (UTF8_IS_INVARIANT(*s)) {
2890 if (! isCHARNAME_CONT(*s)) {
2891 goto bad_charname;
2892 }
3036c853
FC
2893 if (*s == ' ' && *(s-1) == ' '
2894 && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2895 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2896 "A sequence of multiple spaces in a charnam"
2897 "es alias definition is deprecated");
bd299e29 2898 }
bde9e88d
KW
2899 s++;
2900 }
2901 else if (UTF8_IS_DOWNGRADEABLE_START(*s)) {
94bb8c36 2902 if (! isCHARNAME_CONT(TWO_BYTE_UTF8_TO_NATIVE(*s, *(s+1))))
bde9e88d
KW
2903 {
2904 goto bad_charname;
2905 }
2906 s += 2;
2907 }
2908 else {
2909 if (! PL_utf8_charname_continue) {
2910 U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST;
2911 PL_utf8_charname_continue = _core_swash_init("utf8",
2912 "_Perl_Charname_Continue",
2913 &PL_sv_undef,
2914 1, 0, NULL, &flags);
2915 }
2916 if (! swash_fetch(PL_utf8_charname_continue, (U8 *) s, TRUE)) {
2917 goto bad_charname;
2918 }
2919 s += UTF8SKIP(s);
6f613c73
KW
2920 }
2921 }
3036c853 2922 if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) {
94ec3a20
FC
2923 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
2924 "Trailing white-space in a charnames alias "
2925 "definition is deprecated");
bd299e29 2926 }
6f613c73
KW
2927 }
2928
94ca1619 2929 if (SvUTF8(res)) { /* Don't accept malformed input */
bde9e88d
KW
2930 const U8* first_bad_char_loc;
2931 STRLEN len;
2932 const char* const str = SvPV_const(res, len);
2933 if (! is_utf8_string_loc((U8 *) str, len, &first_bad_char_loc)) {
2934 /* If warnings are on, this will print a more detailed analysis of
2935 * what is wrong than the error message below */
c80e42f3 2936 utf8n_to_uvchr(first_bad_char_loc,
bde9e88d
KW
2937 (char *) first_bad_char_loc - str,
2938 NULL, 0);
2939
2940 /* We deliberately don't try to print the malformed character,
2941 * which might not print very well; it also may be just the first
2942 * of many malformations, so don't print what comes after it */
2943 yyerror_pv(
2944 Perl_form(aTHX_
2945 "Malformed UTF-8 returned by %.*s immediately after '%.*s'",
2946 (int) (e - backslash_ptr + 1), backslash_ptr,
2947 (int) ((char *) first_bad_char_loc - str), str
2948 ),
2949 SVf_UTF8);
2950 return NULL;
2951 }
2952 }
140b12ad 2953
bde9e88d 2954 return res;
140b12ad 2955
bde9e88d
KW
2956 bad_charname: {
2957 int bad_char_size = ((UTF) ? UTF8SKIP(s) : 1);
2958
2959 /* The final %.*s makes sure that should the trailing NUL be missing
2960 * that this print won't run off the end of the string */
2961 yyerror_pv(
2962 Perl_form(aTHX_
2963 "Invalid character in \\N{...}; marked by <-- HERE in %.*s<-- HERE %.*s",
2964 (int)(s - backslash_ptr + bad_char_size), backslash_ptr,
2965 (int)(e - s + bad_char_size), s + bad_char_size
2966 ),
2967 UTF ? SVf_UTF8 : 0);
2968 return NULL;
2969 }
6f613c73
KW
2970}
2971
02aa26ce
NT
2972/*
2973 scan_const
2974
9da1dd8f
DM
2975 Extracts the next constant part of a pattern, double-quoted string,
2976 or transliteration. This is terrifying code.
2977
2978 For example, in parsing the double-quoted string "ab\x63$d", it would
2979 stop at the '$' and return an OP_CONST containing 'abc'.
02aa26ce 2980
94def140 2981 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
3280af22 2982 processing a pattern (PL_lex_inpat is true), a transliteration
94def140 2983 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
02aa26ce 2984
94def140
ST
2985 Returns a pointer to the character scanned up to. If this is
2986 advanced from the start pointer supplied (i.e. if anything was
9da1dd8f 2987 successfully parsed), will leave an OP_CONST for the substring scanned
6154021b 2988 in pl_yylval. Caller must intuit reason for not parsing further
9b599b2a
GS
2989 by looking at the next characters herself.
2990
02aa26ce 2991 In patterns:
9da1dd8f 2992 expand:
537124e4
KW
2993 \N{FOO} => \N{U+hex_for_character_FOO}
2994 (if FOO expands to multiple characters, expands to \N{U+xx.XX.yy ...})
9da1dd8f
DM
2995
2996 pass through:
2997 all other \-char, including \N and \N{ apart from \N{ABC}
2998
2999 stops on:
3000 @ and $ where it appears to be a var, but not for $ as tail anchor
3001 \l \L \u \U \Q \E
3002 (?{ or (??{
3003
02aa26ce
NT
3004
3005 In transliterations:
3006 characters are VERY literal, except for - not at the start or end
94def140
ST
3007 of the string, which indicates a range. If the range is in bytes,
3008 scan_const expands the range to the full set of intermediate
3009 characters. If the range is in utf8, the hyphen is replaced with
3010 a certain range mark which will be handled by pmtrans() in op.c.
02aa26ce
NT
3011
3012 In double-quoted strings:
3013 backslashes:
3014 double-quoted style: \r and \n
ff3f963a 3015 constants: \x31, etc.
94def140 3016 deprecated backrefs: \1 (in substitution replacements)
02aa26ce
NT
3017 case and quoting: \U \Q \E
3018 stops on @ and $
3019
3020 scan_const does *not* construct ops to handle interpolated strings.
3021 It stops processing as soon as it finds an embedded $ or @ variable
3022 and leaves it to the caller to work out what's going on.
3023
94def140
ST
3024 embedded arrays (whether in pattern or not) could be:
3025 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
3026
3027 $ in double-quoted strings must be the symbol of an embedded scalar.
02aa26ce
NT
3028
3029 $ in pattern could be $foo or could be tail anchor. Assumption:
3030 it's a tail anchor if $ is the last thing in the string, or if it's
94def140 3031 followed by one of "()| \r\n\t"
02aa26ce 3032
9da1dd8f 3033 \1 (backreferences) are turned into $1 in substitutions
02aa26ce
NT
3034
3035 The structure of the code is
3036 while (there's a character to process) {
94def140
ST
3037 handle transliteration ranges
3038 skip regexp comments /(?#comment)/ and codes /(?{code})/
3039 skip #-initiated comments in //x patterns
3040 check for embedded arrays
02aa26ce
NT
3041 check for embedded scalars
3042 if (backslash) {
94def140 3043 deprecate \1 in substitution replacements
02aa26ce
NT
3044 handle string-changing backslashes \l \U \Q \E, etc.
3045 switch (what was escaped) {
94def140 3046 handle \- in a transliteration (becomes a literal -)
ff3f963a 3047 if a pattern and not \N{, go treat as regular character
94def140
ST
3048 handle \132 (octal characters)
3049 handle \x15 and \x{1234} (hex characters)
ff3f963a 3050 handle \N{name} (named characters, also \N{3,5} in a pattern)
94def140
ST
3051 handle \cV (control characters)
3052 handle printf-style backslashes (\f, \r, \n, etc)
02aa26ce 3053 } (end switch)
77a135fe 3054 continue
02aa26ce 3055 } (end if backslash)
77a135fe 3056 handle regular character
02aa26ce 3057 } (end while character to read)
4e553d73 3058
02aa26ce
NT
3059*/
3060
76e3520e 3061STATIC char *
cea2e8a9 3062S_scan_const(pTHX_ char *start)
79072805 3063{
97aff369 3064 dVAR;
eb578fdb 3065 char *send = PL_bufend; /* end of the constant */
77a135fe
KW
3066 SV *sv = newSV(send - start); /* sv for the constant. See
3067 note below on sizing. */
eb578fdb
KW
3068 char *s = start; /* start of the constant */
3069 char *d = SvPVX(sv); /* destination for copies */
02aa26ce 3070 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 3071 bool didrange = FALSE; /* did we just finish a range? */
2866decb 3072 bool in_charclass = FALSE; /* within /[...]/ */
b953e60c
KW
3073 bool has_utf8 = FALSE; /* Output constant is UTF8 */
3074 bool this_utf8 = cBOOL(UTF); /* Is the source string assumed
77a135fe
KW
3075 to be UTF8? But, this can
3076 show as true when the source
3077 isn't utf8, as for example
3078 when it is entirely composed
3079 of hex constants */
6f613c73 3080 SV *res; /* result from charnames */
77a135fe
KW
3081
3082 /* Note on sizing: The scanned constant is placed into sv, which is
3083 * initialized by newSV() assuming one byte of output for every byte of
3084 * input. This routine expects newSV() to allocate an extra byte for a
3085 * trailing NUL, which this routine will append if it gets to the end of
3086 * the input. There may be more bytes of input than output (eg., \N{LATIN
3087 * CAPITAL LETTER A}), or more output than input if the constant ends up
3088 * recoded to utf8, but each time a construct is found that might increase
3089 * the needed size, SvGROW() is called. Its size parameter each time is
3090 * based on the best guess estimate at the time, namely the length used so
3091 * far, plus the length the current construct will occupy, plus room for
3092 * the trailing NUL, plus one byte for every input byte still unscanned */
3093
c3320c2a
KW
3094 UV uv = UV_MAX; /* Initialize to weird value to try to catch any uses
3095 before set */
4c3a8340
ST
3096#ifdef EBCDIC
3097 UV literal_endpoint = 0;
e294cc5d 3098 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
4c3a8340 3099#endif
012bcf8d 3100
7918f24d
NC
3101 PERL_ARGS_ASSERT_SCAN_CONST;
3102
bb16bae8 3103 assert(PL_lex_inwhat != OP_TRANSR);
2b9d42f0
NIS
3104 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
3105 /* If we are doing a trans and we know we want UTF8 set expectation */
3106 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
3107 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3108 }
3109
b899e89d
FC
3110 /* Protect sv from errors and fatal warnings. */
3111 ENTER_with_name("scan_const");
3112 SAVEFREESV(sv);
2b9d42f0 3113
79072805 3114 while (s < send || dorange) {
ff3f963a 3115
02aa26ce 3116 /* get transliterations out of the way (they're most literal) */
3280af22 3117 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 3118 /* expand a range A-Z to the full set of characters. AIE! */
79072805 3119 if (dorange) {
1ba5c669
JH
3120 I32 i; /* current expanded character */
3121 I32 min; /* first character in range */
3122 I32 max; /* last character in range */
02aa26ce 3123
e294cc5d
JH
3124#ifdef EBCDIC
3125 UV uvmax = 0;
3126#endif
3127
3128 if (has_utf8
3129#ifdef EBCDIC
3130 && !native_range
3131#endif
1953db30 3132 ) {
9d4ba2ae 3133 char * const c = (char*)utf8_hop((U8*)d, -1);
8973db79
JH
3134 char *e = d++;
3135 while (e-- > c)
3136 *(e + 1) = *e;
e7214ce8 3137 *c = (char) ILLEGAL_UTF8_BYTE;
8973db79
JH
3138 /* mark the range as done, and continue */
3139 dorange = FALSE;
3140 didrange = TRUE;
3141 continue;
3142 }
2b9d42f0 3143
95a20fc0 3144 i = d - SvPVX_const(sv); /* remember current offset */
e294cc5d
JH
3145#ifdef EBCDIC
3146 SvGROW(sv,
3147 SvLEN(sv) + (has_utf8 ?
3148 (512 - UTF_CONTINUATION_MARK +
3149 UNISKIP(0x100))
3150 : 256));
3151 /* How many two-byte within 0..255: 128 in UTF-8,
3152 * 96 in UTF-8-mod. */
3153#else
9cbb5ea2 3154 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
e294cc5d 3155#endif
9cbb5ea2 3156 d = SvPVX(sv) + i; /* refresh d after realloc */
e294cc5d
JH
3157#ifdef EBCDIC
3158 if (has_utf8) {
3159 int j;
3160 for (j = 0; j <= 1; j++) {
3161 char * const c = (char*)utf8_hop((U8*)d, -1);
3162 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
3163 if (j)
3164 min = (U8)uv;
3165 else if (uv < 256)
3166 max = (U8)uv;
3167 else {
3168 max = (U8)0xff; /* only to \xff */
3169 uvmax = uv; /* \x{100} to uvmax */
3170 }
3171 d = c; /* eat endpoint chars */
3172 }
3173 }
3174 else {
3175#endif
3176 d -= 2; /* eat the first char and the - */
3177 min = (U8)*d; /* first char in range */
3178 max = (U8)d[1]; /* last char in range */
3179#ifdef EBCDIC
3180 }
3181#endif
8ada0baa 3182
c2e66d9e 3183 if (min > max) {
01ec43d0 3184 Perl_croak(aTHX_
d1573ac7 3185 "Invalid range \"%c-%c\" in transliteration operator",
1ba5c669 3186 (char)min, (char)max);
c2e66d9e
GS
3187 }
3188
c7f1f016 3189#ifdef EBCDIC
4c3a8340 3190 if (literal_endpoint == 2 &&
c286a9a6 3191 ((isLOWER_A(min) && isLOWER_A(max)) ||
75e9e7bf
KW
3192 (isUPPER_A(min) && isUPPER_A(max))))
3193 {
3194 for (i = min; i <= max; i++) {
3195 if (isALPHA_A(i))
3196 *d++ = i;
8ada0baa
JH
3197 }
3198 }
3199 else
3200#endif
3201 for (i = min; i <= max; i++)
e294cc5d
JH
3202#ifdef EBCDIC
3203 if (has_utf8) {
55d09dc8 3204 append_utf8_from_native_byte(i, &d);
e294cc5d
JH
3205 }
3206 else
3207#endif
3208 *d++ = (char)i;
3209
3210#ifdef EBCDIC
3211 if (uvmax) {
3212 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
3213 if (uvmax > 0x101)
e7214ce8 3214 *d++ = (char) ILLEGAL_UTF8_BYTE;
e294cc5d
JH
3215 if (uvmax > 0x100)
3216 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
3217 }
3218#endif
02aa26ce
NT
3219
3220 /* mark the range as done, and continue */
79072805 3221 dorange = FALSE;
01ec43d0 3222 didrange = TRUE;
4c3a8340
ST
3223#ifdef EBCDIC
3224 literal_endpoint = 0;
3225#endif
79072805 3226 continue;
4e553d73 3227 }
02aa26ce
NT
3228
3229 /* range begins (ignore - as first or last char) */
79072805 3230 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 3231 if (didrange) {
1fafa243 3232 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 3233 }
e294cc5d
JH
3234 if (has_utf8
3235#ifdef EBCDIC
3236 && !native_range
3237#endif
3238 ) {
e7214ce8 3239 *d++ = (char) ILLEGAL_UTF8_BYTE; /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
3240 s++;
3241 continue;
3242 }
79072805
LW
3243 dorange = TRUE;
3244 s++;
01ec43d0
GS
3245 }
3246 else {
3247 didrange = FALSE;
4c3a8340
ST
3248#ifdef EBCDIC
3249 literal_endpoint = 0;
e294cc5d 3250 native_range = TRUE;
4c3a8340 3251#endif
01ec43d0 3252 }
79072805 3253 }
02aa26ce
NT
3254
3255 /* if we get here, we're not doing a transliteration */
3256
e4a2df84
DM
3257 else if (*s == '[' && PL_lex_inpat && !in_charclass) {
3258 char *s1 = s-1;
3259 int esc = 0;
3260 while (s1 >= start && *s1-- == '\\')
3261 esc = !esc;
3262 if (!esc)
3263 in_charclass = TRUE;
3264 }
2866decb 3265
e4a2df84
DM
3266 else if (*s == ']' && PL_lex_inpat && in_charclass) {
3267 char *s1 = s-1;
3268 int esc = 0;
3269 while (s1 >= start && *s1-- == '\\')
3270 esc = !esc;
3271 if (!esc)
3272 in_charclass = FALSE;
3273 }
2866decb 3274
9da1dd8f
DM
3275 /* skip for regexp comments /(?#comment)/, except for the last
3276 * char, which will be done separately.
3277 * Stop on (?{..}) and friends */
3278
c30fc27b 3279 else if (*s == '(' && PL_lex_inpat && s[1] == '?' && !in_charclass) {
cc6b7395 3280 if (s[2] == '#') {
e994fd66 3281 while (s+1 < send && *s != ')')
5ff03569 3282 *d++ = *s++;
155aba94 3283 }
c30fc27b 3284 else if (!PL_lex_casemods &&
d3cec5e5
DM
3285 ( s[2] == '{' /* This should match regcomp.c */
3286 || (s[2] == '?' && s[3] == '{')))
155aba94 3287 {
9da1dd8f 3288 break;
cc6b7395 3289 }
748a9306 3290 }
02aa26ce
NT
3291
3292 /* likewise skip #-initiated comments in //x patterns */
c30fc27b 3293 else if (*s == '#' && PL_lex_inpat && !in_charclass &&
73134a2e 3294 ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
748a9306 3295 while (s+1 < send && *s != '\n')
5ff03569 3296 *d++ = *s++;
748a9306 3297 }
02aa26ce 3298
9da1dd8f
DM
3299 /* no further processing of single-quoted regex */
3300 else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'')
3301 goto default_action;
3302
5d1d4326 3303 /* check for embedded arrays
da6eedaa 3304 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
5d1d4326 3305 */
1749ea0d 3306 else if (*s == '@' && s[1]) {
8a2bca12 3307 if (isWORDCHAR_lazy_if(s+1,UTF))
1749ea0d
ST
3308 break;
3309 if (strchr(":'{$", s[1]))
3310 break;
3311 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
3312 break; /* in regexp, neither @+ nor @- are interpolated */
3313 }
02aa26ce
NT
3314
3315 /* check for embedded scalars. only stop if we're sure it's a
3316 variable.
3317 */
79072805 3318 else if (*s == '$') {
3280af22 3319 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 3320 break;
77772344 3321 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
a2a5de95
NC
3322 if (s[1] == '\\') {
3323 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
3324 "Possible unintended interpolation of $\\ in regex");
77772344 3325 }
79072805 3326 break; /* in regexp, $ might be tail anchor */
77772344 3327 }
79072805 3328 }
02aa26ce 3329
2b9d42f0
NIS
3330 /* End of else if chain - OP_TRANS rejoin rest */
3331
02aa26ce 3332 /* backslashes */
79072805 3333 if (*s == '\\' && s+1 < send) {
ff3f963a
KW
3334 char* e; /* Can be used for ending '}', etc. */
3335
79072805 3336 s++;
02aa26ce 3337
7d0fc23c
KW
3338 /* warn on \1 - \9 in substitution replacements, but note that \11
3339 * is an octal; and \19 is \1 followed by '9' */
3280af22 3340 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 3341 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 3342 {
a2a5de95 3343 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
79072805
LW
3344 *--s = '$';
3345 break;
3346 }
02aa26ce
NT
3347
3348 /* string-change backslash escapes */
838f2281 3349 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) {
79072805
LW
3350 --s;
3351 break;
3352 }
ff3f963a
KW
3353 /* In a pattern, process \N, but skip any other backslash escapes.
3354 * This is because we don't want to translate an escape sequence
3355 * into a meta symbol and have the regex compiler use the meta
3356 * symbol meaning, e.g. \x{2E} would be confused with a dot. But
3357 * in spite of this, we do have to process \N here while the proper
3358 * charnames handler is in scope. See bugs #56444 and #62056.
3359 * There is a complication because \N in a pattern may also stand
3360 * for 'match a non-nl', and not mean a charname, in which case its
3361 * processing should be deferred to the regex compiler. To be a
3362 * charname it must be followed immediately by a '{', and not look
3363 * like \N followed by a curly quantifier, i.e., not something like
3364 * \N{3,}. regcurly returns a boolean indicating if it is a legal
3365 * quantifier */
3366 else if (PL_lex_inpat
3367 && (*s != 'N'
3368 || s[1] != '{'
4d68ffa0 3369 || regcurly(s + 1, FALSE)))
ff3f963a 3370 {
4d73d076 3371 *d++ = '\\';
cc74c5bd
ST
3372 goto default_action;
3373 }
02aa26ce 3374
79072805 3375 switch (*s) {
02aa26ce
NT
3376
3377 /* quoted - in transliterations */
79072805 3378 case '-':
3280af22 3379 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
3380 *d++ = *s++;
3381 continue;
3382 }
3383 /* FALL THROUGH */
3384 default:
11b8faa4 3385 {
15861f94 3386 if ((isALPHANUMERIC(*s)))
a2a5de95
NC
3387 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
3388 "Unrecognized escape \\%c passed through",
3389 *s);
11b8faa4 3390 /* default action is to copy the quoted character */
f9a63242 3391 goto default_action;
11b8faa4 3392 }
02aa26ce 3393
632403cc 3394 /* eg. \132 indicates the octal constant 0132 */
79072805
LW
3395 case '0': case '1': case '2': case '3':
3396 case '4': case '5': case '6': case '7':
ba210ebe 3397 {
5e0a247b 3398 I32 flags = PERL_SCAN_SILENT_ILLDIGIT;
53305cf1 3399 STRLEN len = 3;
06972766 3400 uv = grok_oct(s, &len, &flags, NULL);
ba210ebe 3401 s += len;
5e0a247b
KW
3402 if (len < 3 && s < send && isDIGIT(*s)
3403 && ckWARN(WARN_MISC))
3404 {
3405 Perl_warner(aTHX_ packWARN(WARN_MISC),
3406 "%s", form_short_octal_warning(s, len));
3407 }
ba210ebe 3408 }
012bcf8d 3409 goto NUM_ESCAPE_INSERT;
02aa26ce 3410
f0a2b745
KW
3411 /* eg. \o{24} indicates the octal constant \024 */
3412 case 'o':
3413 {
454155d9 3414 const char* error;
f0a2b745 3415
00ce5563 3416 bool valid = grok_bslash_o(&s, &uv, &error,
80f4111b
KW
3417 TRUE, /* Output warning */
3418 FALSE, /* Not strict */
17896a85
KW
3419 TRUE, /* Output warnings for
3420 non-portables */
80f4111b 3421 UTF);
454155d9 3422 if (! valid) {
f0a2b745
KW
3423 yyerror(error);
3424 continue;
3425 }
3426 goto NUM_ESCAPE_INSERT;
3427 }
3428
77a135fe 3429 /* eg. \x24 indicates the hex constant 0x24 */
79072805 3430 case 'x':
a0481293 3431 {
a0481293 3432 const char* error;
355860ce 3433
00ce5563 3434 bool valid = grok_bslash_x(&s, &uv, &error,
80f4111b
KW
3435 TRUE, /* Output warning */
3436 FALSE, /* Not strict */
17896a85
KW
3437 TRUE, /* Output warnings for
3438 non-portables */
80f4111b 3439 UTF);
a0481293
KW
3440 if (! valid) {
3441 yyerror(error);
355860ce 3442 continue;
ba210ebe 3443 }
012bcf8d
GS
3444 }
3445
3446 NUM_ESCAPE_INSERT:
ff3f963a
KW
3447 /* Insert oct or hex escaped character. There will always be
3448 * enough room in sv since such escapes will be longer than any
3449 * UTF-8 sequence they can end up as, except if they force us
3450 * to recode the rest of the string into utf8 */
ba7cea30 3451
06972766
KW
3452 /* Here uv is the ordinal of the next character being added */
3453 if (!NATIVE_IS_INVARIANT(uv)) {
9aa983d2 3454 if (!has_utf8 && uv > 255) {
77a135fe
KW
3455 /* Might need to recode whatever we have accumulated so
3456 * far if it contains any chars variant in utf8 or
3457 * utf-ebcdic. */
3458
3459 SvCUR_set(sv, d - SvPVX_const(sv));
3460 SvPOK_on(sv);
3461 *d = '\0';
77a135fe 3462 /* See Note on sizing above. */
7bf79863
KW
3463 sv_utf8_upgrade_flags_grow(sv,
3464 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3465 UNISKIP(uv) + (STRLEN)(send - s) + 1);
77a135fe
KW
3466 d = SvPVX(sv) + SvCUR(sv);
3467 has_utf8 = TRUE;
012bcf8d
GS
3468 }
3469
77a135fe 3470 if (has_utf8) {
c80e42f3 3471 d = (char*)uvchr_to_utf8((U8*)d, uv);
f9a63242
JH
3472 if (PL_lex_inwhat == OP_TRANS &&
3473 PL_sublex_info.sub_op) {
3474 PL_sublex_info.sub_op->op_private |=
3475 (PL_lex_repl ? OPpTRANS_FROM_UTF
3476 : OPpTRANS_TO_UTF);
f9a63242 3477 }
e294cc5d
JH
3478#ifdef EBCDIC
3479 if (uv > 255 && !dorange)
3480 native_range = FALSE;
3481#endif
012bcf8d 3482 }
a0ed51b3 3483 else {
012bcf8d 3484 *d++ = (char)uv;
a0ed51b3 3485 }
012bcf8d
GS
3486 }
3487 else {
c4d5f83a 3488 *d++ = (char) uv;
a0ed51b3 3489 }
79072805 3490 continue;
02aa26ce 3491
4a2d328f 3492 case 'N':
ff3f963a
KW
3493 /* In a non-pattern \N must be a named character, like \N{LATIN
3494 * SMALL LETTER A} or \N{U+0041}. For patterns, it also can
3495 * mean to match a non-newline. For non-patterns, named
3496 * characters are converted to their string equivalents. In
3497 * patterns, named characters are not converted to their
3498 * ultimate forms for the same reasons that other escapes
3499 * aren't. Instead, they are converted to the \N{U+...} form
3500 * to get the value from the charnames that is in effect right
3501 * now, while preserving the fact that it was a named character
3502 * so that the regex compiler knows this */
3503
ff3f963a
KW
3504 /* The structure of this section of code (besides checking for
3505 * errors and upgrading to utf8) is:
3506 * Further disambiguate between the two meanings of \N, and if
3507 * not a charname, go process it elsewhere
0a96133f
KW
3508 * If of form \N{U+...}, pass it through if a pattern;
3509 * otherwise convert to utf8
3510 * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3511 * pattern; otherwise convert to utf8 */
ff3f963a
KW
3512
3513 /* Here, s points to the 'N'; the test below is guaranteed to
3514 * succeed if we are being called on a pattern as we already
3515 * know from a test above that the next character is a '{'.
3516 * On a non-pattern \N must mean 'named sequence, which
3517 * requires braces */
3518 s++;
3519 if (*s != '{') {
3520 yyerror("Missing braces on \\N{}");
3521 continue;
3522 }
3523 s++;
3524
0a96133f 3525 /* If there is no matching '}', it is an error. */
ff3f963a
KW
3526 if (! (e = strchr(s, '}'))) {
3527 if (! PL_lex_inpat) {
5777a3f7 3528 yyerror("Missing right brace on \\N{}");
0a96133f
KW
3529 } else {
3530 yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
dbc0d4f2 3531 }
0a96133f 3532 continue;
ff3f963a 3533 }
cddc7ef4 3534
ff3f963a 3535 /* Here it looks like a named character */
cddc7ef4 3536
ff3f963a
KW
3537 if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3538 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3539 | PERL_SCAN_DISALLOW_PREFIX;
3540 STRLEN len;
3541
3542 /* For \N{U+...}, the '...' is a unicode value even on
3543 * EBCDIC machines */
3544 s += 2; /* Skip to next char after the 'U+' */
3545 len = e - s;
3546 uv = grok_hex(s, &len, &flags, NULL);
3547 if (len == 0 || len != (STRLEN)(e - s)) {
3548 yyerror("Invalid hexadecimal number in \\N{U+...}");
3549 s = e + 1;
3550 continue;
3551 }
3552
3553 if (PL_lex_inpat) {
3554
e2a7e165
KW
3555 /* On non-EBCDIC platforms, pass through to the regex
3556 * compiler unchanged. The reason we evaluated the
3557 * number above is to make sure there wasn't a syntax
3558 * error. But on EBCDIC we convert to native so
3559 * downstream code can continue to assume it's native
3560 */
ff3f963a 3561 s -= 5; /* Include the '\N{U+' */
e2a7e165
KW
3562#ifdef EBCDIC
3563 d += my_snprintf(d, e - s + 1 + 1, /* includes the }
3564 and the \0 */
3565 "\\N{U+%X}",
3566 (unsigned int) UNI_TO_NATIVE(uv));
3567#else
ff3f963a
KW
3568 Copy(s, d, e - s + 1, char); /* 1 = include the } */
3569 d += e - s + 1;
e2a7e165 3570#endif
ff3f963a
KW
3571 }
3572 else { /* Not a pattern: convert the hex to string */
3573
3574 /* If destination is not in utf8, unconditionally
3575 * recode it to be so. This is because \N{} implies
3576 * Unicode semantics, and scalars have to be in utf8
3577 * to guarantee those semantics */
3578 if (! has_utf8) {
3579 SvCUR_set(sv, d - SvPVX_const(sv));
3580 SvPOK_on(sv);
3581 *d = '\0';
3582 /* See Note on sizing above. */
3583 sv_utf8_upgrade_flags_grow(
3584 sv,
3585 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3586 UNISKIP(uv) + (STRLEN)(send - e) + 1);
3587 d = SvPVX(sv) + SvCUR(sv);
3588 has_utf8 = TRUE;
3589 }
3590
a46469e6 3591 /* Add the (Unicode) code point to the output. */
ff3f963a 3592 if (UNI_IS_INVARIANT(uv)) {
a46469e6 3593 *d++ = (char) LATIN1_TO_NATIVE(uv);
ff3f963a 3594 }
a46469e6
KW
3595 else {
3596 d = (char*) uvoffuni_to_utf8_flags((U8*)d, uv, 0);
3597 }
ff3f963a
KW
3598 }
3599 }
6f613c73
KW
3600 else /* Here is \N{NAME} but not \N{U+...}. */
3601 if ((res = get_and_check_backslash_N_name(s, e)))
3602 {
3603 STRLEN len;
3604 const char *str = SvPV_const(res, len);
3605 if (PL_lex_inpat) {
ff3f963a
KW
3606
3607 if (! len) { /* The name resolved to an empty string */
3608 Copy("\\N{}", d, 4, char);
3609 d += 4;
3610 }
3611 else {
3612 /* In order to not lose information for the regex
3613 * compiler, pass the result in the specially made
3614 * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3615 * the code points in hex of each character
3616 * returned by charnames */
3617
3618 const char *str_end = str + len;
3b721c4f 3619 const STRLEN off = d - SvPVX_const(sv);
94ca1619
KW
3620
3621 if (! SvUTF8(res)) {
3622 /* For the non-UTF-8 case, we can determine the
3623 * exact length needed without having to parse
3624 * through the string. Each character takes up
3625 * 2 hex digits plus either a trailing dot or
3626 * the "}" */
3627 d = off + SvGROW(sv, off
3628 + 3 * len
3629 + 6 /* For the "\N{U+", and
3630 trailing NUL */
3631 + (STRLEN)(send - e));
3632 Copy("\\N{U+", d, 5, char);
3633 d += 5;
3634 while (str < str_end) {
3635 char hex_string[4];
3636 my_snprintf(hex_string, sizeof(hex_string),
3637 "%02X.", (U8) *str);
3638 Copy(hex_string, d, 3, char);
3639 d += 3;
3640 str++;
3641 }
3642 d--; /* We will overwrite below the final
3643 dot with a right brace */
3644 }
3645 else {
1953db30
KW
3646 STRLEN char_length; /* cur char's byte length */
3647
3648 /* and the number of bytes after this is
3649 * translated into hex digits */
3650 STRLEN output_length;
3651
3652 /* 2 hex per byte; 2 chars for '\N'; 2 chars
3653 * for max('U+', '.'); and 1 for NUL */
3654 char hex_string[2 * UTF8_MAXBYTES + 5];
3655
3656 /* Get the first character of the result. */
a46469e6 3657 U32 uv = utf8n_to_uvchr((U8 *) str,
1953db30
KW
3658 len,
3659 &char_length,
3660 UTF8_ALLOW_ANYUV);
3661 /* Convert first code point to hex, including
a46469e6 3662 * the boiler plate before it. */
1953db30
KW
3663 output_length =
3664 my_snprintf(hex_string, sizeof(hex_string),
a46469e6
KW
3665 "\\N{U+%X",
3666 (unsigned int) uv);
1953db30
KW
3667
3668 /* Make sure there is enough space to hold it */
3669 d = off + SvGROW(sv, off
3670 + output_length
3671 + (STRLEN)(send - e)
3672 + 2); /* '}' + NUL */
3673 /* And output it */
3674 Copy(hex_string, d, output_length, char);
3675 d += output_length;
3676
3677 /* For each subsequent character, append dot and