This is a live mirror of the Perl 5 development currently hosted at https://github.com/perl/perl5
correct comment about how strings are tokenised
[perl5.git] / toke.c
CommitLineData
a0d0e21e 1/* toke.c
a687059c 2 *
1129b882
NC
3 * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, 2000,
4 * 2001, 2002, 2003, 2004, 2005, 2006, 2007, 2008 by Larry Wall and others
a687059c 5 *
d48672a2
LW
6 * You may distribute under the terms of either the GNU General Public
7 * License or the Artistic License, as specified in the README file.
378cc40b 8 *
a0d0e21e
LW
9 */
10
11/*
4ac71550
TC
12 * 'It all comes from here, the stench and the peril.' --Frodo
13 *
14 * [p.719 of _The Lord of the Rings_, IV/ix: "Shelob's Lair"]
378cc40b
LW
15 */
16
9cbb5ea2
GS
17/*
18 * This file is the lexer for Perl. It's closely linked to the
4e553d73 19 * parser, perly.y.
ffb4593c
NT
20 *
21 * The main routine is yylex(), which returns the next token.
22 */
23
f0e67a1d
Z
24/*
25=head1 Lexer interface
26
27This is the lower layer of the Perl parser, managing characters and tokens.
28
29=for apidoc AmU|yy_parser *|PL_parser
30
31Pointer to a structure encapsulating the state of the parsing operation
32currently in progress. The pointer can be locally changed to perform
33a nested parse without interfering with the state of an outer parse.
34Individual members of C<PL_parser> have their own documentation.
35
36=cut
37*/
38
378cc40b 39#include "EXTERN.h"
864dbfa3 40#define PERL_IN_TOKE_C
378cc40b 41#include "perl.h"
04e98a4d 42#include "dquote_static.c"
378cc40b 43
eb0d8d16
NC
44#define new_constant(a,b,c,d,e,f,g) \
45 S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g)
46
6154021b 47#define pl_yylval (PL_parser->yylval)
d3b6f988 48
199e78b7
DM
49/* XXX temporary backwards compatibility */
50#define PL_lex_brackets (PL_parser->lex_brackets)
78cdf107
Z
51#define PL_lex_allbrackets (PL_parser->lex_allbrackets)
52#define PL_lex_fakeeof (PL_parser->lex_fakeeof)
199e78b7
DM
53#define PL_lex_brackstack (PL_parser->lex_brackstack)
54#define PL_lex_casemods (PL_parser->lex_casemods)
55#define PL_lex_casestack (PL_parser->lex_casestack)
56#define PL_lex_defer (PL_parser->lex_defer)
57#define PL_lex_dojoin (PL_parser->lex_dojoin)
58#define PL_lex_expect (PL_parser->lex_expect)
59#define PL_lex_formbrack (PL_parser->lex_formbrack)
60#define PL_lex_inpat (PL_parser->lex_inpat)
61#define PL_lex_inwhat (PL_parser->lex_inwhat)
62#define PL_lex_op (PL_parser->lex_op)
63#define PL_lex_repl (PL_parser->lex_repl)
64#define PL_lex_starts (PL_parser->lex_starts)
65#define PL_lex_stuff (PL_parser->lex_stuff)
66#define PL_multi_start (PL_parser->multi_start)
67#define PL_multi_open (PL_parser->multi_open)
68#define PL_multi_close (PL_parser->multi_close)
69#define PL_pending_ident (PL_parser->pending_ident)
70#define PL_preambled (PL_parser->preambled)
71#define PL_sublex_info (PL_parser->sublex_info)
bdc0bf6f 72#define PL_linestr (PL_parser->linestr)
c2598295
DM
73#define PL_expect (PL_parser->expect)
74#define PL_copline (PL_parser->copline)
f06b5848
DM
75#define PL_bufptr (PL_parser->bufptr)
76#define PL_oldbufptr (PL_parser->oldbufptr)
77#define PL_oldoldbufptr (PL_parser->oldoldbufptr)
78#define PL_linestart (PL_parser->linestart)
79#define PL_bufend (PL_parser->bufend)
80#define PL_last_uni (PL_parser->last_uni)
81#define PL_last_lop (PL_parser->last_lop)
82#define PL_last_lop_op (PL_parser->last_lop_op)
bc177e6b 83#define PL_lex_state (PL_parser->lex_state)
2f9285f8 84#define PL_rsfp (PL_parser->rsfp)
5486870f 85#define PL_rsfp_filters (PL_parser->rsfp_filters)
12bd6ede
DM
86#define PL_in_my (PL_parser->in_my)
87#define PL_in_my_stash (PL_parser->in_my_stash)
14047fc9 88#define PL_tokenbuf (PL_parser->tokenbuf)
670a9cb2 89#define PL_multi_end (PL_parser->multi_end)
13765c85 90#define PL_error_count (PL_parser->error_count)
199e78b7
DM
91
92#ifdef PERL_MAD
93# define PL_endwhite (PL_parser->endwhite)
94# define PL_faketokens (PL_parser->faketokens)
95# define PL_lasttoke (PL_parser->lasttoke)
96# define PL_nextwhite (PL_parser->nextwhite)
97# define PL_realtokenstart (PL_parser->realtokenstart)
98# define PL_skipwhite (PL_parser->skipwhite)
99# define PL_thisclose (PL_parser->thisclose)
100# define PL_thismad (PL_parser->thismad)
101# define PL_thisopen (PL_parser->thisopen)
102# define PL_thisstuff (PL_parser->thisstuff)
103# define PL_thistoken (PL_parser->thistoken)
104# define PL_thiswhite (PL_parser->thiswhite)
fb205e7a
DM
105# define PL_thiswhite (PL_parser->thiswhite)
106# define PL_nexttoke (PL_parser->nexttoke)
107# define PL_curforce (PL_parser->curforce)
108#else
109# define PL_nexttoke (PL_parser->nexttoke)
110# define PL_nexttype (PL_parser->nexttype)
111# define PL_nextval (PL_parser->nextval)
199e78b7
DM
112#endif
113
16173588
NC
114/* This can't be done with embed.fnc, because struct yy_parser contains a
115 member named pending_ident, which clashes with the generated #define */
3cbf51f5
DM
116static int
117S_pending_ident(pTHX);
199e78b7 118
0bd48802 119static const char ident_too_long[] = "Identifier too long";
8903cb82 120
29595ff2 121#ifdef PERL_MAD
29595ff2 122# define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; }
cd81e915 123# define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val
9ded7720 124#else
5db06880 125# define CURMAD(slot,sv)
9ded7720 126# define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke]
29595ff2
NC
127#endif
128
a7aaec61
Z
129#define XENUMMASK 0x3f
130#define XFAKEEOF 0x40
131#define XFAKEBRACK 0x80
9059aa12 132
39e02b42
JH
133#ifdef USE_UTF8_SCRIPTS
134# define UTF (!IN_BYTES)
2b9d42f0 135#else
802a15e9 136# define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || ( !(PL_parser->lex_flags & LEX_IGNORE_UTF8_HINTS) && (PL_hints & HINT_UTF8)))
2b9d42f0 137#endif
a0ed51b3 138
b1fc3636
CJ
139/* The maximum number of characters preceding the unrecognized one to display */
140#define UNRECOGNIZED_PRECEDE_COUNT 10
141
61f0cdd9 142/* In variables named $^X, these are the legal values for X.
2b92dfce
GS
143 * 1999-02-27 mjd-perl-patch@plover.com */
144#define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x)))
145
bf4acbe4 146#define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t')
bf4acbe4 147
ffb4593c
NT
148/* LEX_* are values for PL_lex_state, the state of the lexer.
149 * They are arranged oddly so that the guard on the switch statement
79072805
LW
150 * can get by with a single comparison (if the compiler is smart enough).
151 */
152
fb73857a
PP
153/* #define LEX_NOTPARSING 11 is done in perl.h. */
154
b6007c36
DM
155#define LEX_NORMAL 10 /* normal code (ie not within "...") */
156#define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */
157#define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */
158#define LEX_INTERPPUSH 7 /* starting a new sublex parse level */
159#define LEX_INTERPSTART 6 /* expecting the start of a $var */
160
161 /* at end of code, eg "$x" followed by: */
162#define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */
163#define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */
164
165#define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of
166 string or after \E, $foo, etc */
167#define LEX_INTERPCONST 2 /* NOT USED */
168#define LEX_FORMLINE 1 /* expecting a format line */
169#define LEX_KNOWNEXT 0 /* next token known; just return it */
170
79072805 171
bbf60fe6 172#ifdef DEBUGGING
27da23d5 173static const char* const lex_state_names[] = {
bbf60fe6
DM
174 "KNOWNEXT",
175 "FORMLINE",
176 "INTERPCONST",
177 "INTERPCONCAT",
178 "INTERPENDMAYBE",
179 "INTERPEND",
180 "INTERPSTART",
181 "INTERPPUSH",
182 "INTERPCASEMOD",
183 "INTERPNORMAL",
184 "NORMAL"
185};
186#endif
187
79072805
LW
188#ifdef ff_next
189#undef ff_next
d48672a2
LW
190#endif
191
79072805 192#include "keywords.h"
fe14fcc3 193
ffb4593c
NT
194/* CLINE is a macro that ensures PL_copline has a sane value */
195
ae986130
LW
196#ifdef CLINE
197#undef CLINE
198#endif
57843af0 199#define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline))
3280af22 200
5db06880 201#ifdef PERL_MAD
29595ff2
NC
202# define SKIPSPACE0(s) skipspace0(s)
203# define SKIPSPACE1(s) skipspace1(s)
204# define SKIPSPACE2(s,tsv) skipspace2(s,&tsv)
205# define PEEKSPACE(s) skipspace2(s,0)
206#else
207# define SKIPSPACE0(s) skipspace(s)
208# define SKIPSPACE1(s) skipspace(s)
209# define SKIPSPACE2(s,tsv) skipspace(s)
210# define PEEKSPACE(s) skipspace(s)
211#endif
212
ffb4593c
NT
213/*
214 * Convenience functions to return different tokens and prime the
9cbb5ea2 215 * lexer for the next token. They all take an argument.
ffb4593c
NT
216 *
217 * TOKEN : generic token (used for '(', DOLSHARP, etc)
218 * OPERATOR : generic operator
219 * AOPERATOR : assignment operator
220 * PREBLOCK : beginning the block after an if, while, foreach, ...
221 * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref)
222 * PREREF : *EXPR where EXPR is not a simple identifier
223 * TERM : expression term
224 * LOOPX : loop exiting command (goto, last, dump, etc)
225 * FTST : file test operator
226 * FUN0 : zero-argument function
7eb971ee 227 * FUN0OP : zero-argument function, with its op created in this file
2d2e263d 228 * FUN1 : not used, except for not, which isn't a UNIOP
ffb4593c
NT
229 * BOop : bitwise or or xor
230 * BAop : bitwise and
231 * SHop : shift operator
232 * PWop : power operator
9cbb5ea2 233 * PMop : pattern-matching operator
ffb4593c
NT
234 * Aop : addition-level operator
235 * Mop : multiplication-level operator
236 * Eop : equality-testing operator
e5edeb50 237 * Rop : relational operator <= != gt
ffb4593c
NT
238 *
239 * Also see LOP and lop() below.
240 */
241
998054bd 242#ifdef DEBUGGING /* Serve -DT. */
704d4215 243# define REPORT(retval) tokereport((I32)retval, &pl_yylval)
998054bd 244#else
bbf60fe6 245# define REPORT(retval) (retval)
998054bd
SC
246#endif
247
bbf60fe6
DM
248#define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval))
249#define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval))
250#define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval)))
251#define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval))
252#define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval))
253#define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval))
254#define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval))
6154021b
RGS
255#define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX))
256#define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP))
257#define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0))
7eb971ee 258#define FUN0OP(f) return (pl_yylval.opval=f, CLINE, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0OP))
6154021b
RGS
259#define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1))
260#define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP)))
261#define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP)))
262#define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP)))
263#define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP)))
264#define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP))
265#define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP)))
266#define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP)))
267#define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP))
268#define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP))
2f3197b3 269
a687059c
LW
270/* This bit of chicanery makes a unary function followed by
271 * a parenthesis into a function with one argument, highest precedence.
6f33ba73
RGS
272 * The UNIDOR macro is for unary functions that can be followed by the //
273 * operator (such as C<shift // 0>).
a687059c 274 */
376fcdbf 275#define UNI2(f,x) { \
6154021b 276 pl_yylval.ival = f; \
376fcdbf
AL
277 PL_expect = x; \
278 PL_bufptr = s; \
279 PL_last_uni = PL_oldbufptr; \
280 PL_last_lop_op = f; \
281 if (*s == '(') \
282 return REPORT( (int)FUNC1 ); \
29595ff2 283 s = PEEKSPACE(s); \
376fcdbf
AL
284 return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \
285 }
6f33ba73
RGS
286#define UNI(f) UNI2(f,XTERM)
287#define UNIDOR(f) UNI2(f,XTERMORDORDOR)
b5fb7ce3
FC
288#define UNIPROTO(f,optional) { \
289 if (optional) PL_last_uni = PL_oldbufptr; \
22393538
MH
290 OPERATOR(f); \
291 }
a687059c 292
376fcdbf 293#define UNIBRACK(f) { \
6154021b 294 pl_yylval.ival = f; \
376fcdbf
AL
295 PL_bufptr = s; \
296 PL_last_uni = PL_oldbufptr; \
297 if (*s == '(') \
298 return REPORT( (int)FUNC1 ); \
29595ff2 299 s = PEEKSPACE(s); \
376fcdbf
AL
300 return REPORT( (*s == '(') ? (int)FUNC1 : (int)UNIOP ); \
301 }
79072805 302
9f68db38 303/* grandfather return to old style */
78cdf107
Z
304#define OLDLOP(f) \
305 do { \
306 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) \
307 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; \
308 pl_yylval.ival = (f); \
309 PL_expect = XTERM; \
310 PL_bufptr = s; \
311 return (int)LSTOP; \
312 } while(0)
79072805 313
8fa7f367
JH
314#ifdef DEBUGGING
315
6154021b 316/* how to interpret the pl_yylval associated with the token */
bbf60fe6
DM
317enum token_type {
318 TOKENTYPE_NONE,
319 TOKENTYPE_IVAL,
6154021b 320 TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */
bbf60fe6
DM
321 TOKENTYPE_PVAL,
322 TOKENTYPE_OPVAL,
323 TOKENTYPE_GVVAL
324};
325
6d4a66ac
NC
326static struct debug_tokens {
327 const int token;
328 enum token_type type;
329 const char *name;
330} const debug_tokens[] =
9041c2e3 331{
bbf60fe6
DM
332 { ADDOP, TOKENTYPE_OPNUM, "ADDOP" },
333 { ANDAND, TOKENTYPE_NONE, "ANDAND" },
334 { ANDOP, TOKENTYPE_NONE, "ANDOP" },
335 { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" },
336 { ARROW, TOKENTYPE_NONE, "ARROW" },
337 { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" },
338 { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" },
339 { BITOROP, TOKENTYPE_OPNUM, "BITOROP" },
340 { COLONATTR, TOKENTYPE_NONE, "COLONATTR" },
341 { CONTINUE, TOKENTYPE_NONE, "CONTINUE" },
0d863452 342 { DEFAULT, TOKENTYPE_NONE, "DEFAULT" },
bbf60fe6
DM
343 { DO, TOKENTYPE_NONE, "DO" },
344 { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" },
345 { DORDOR, TOKENTYPE_NONE, "DORDOR" },
346 { DOROP, TOKENTYPE_OPNUM, "DOROP" },
347 { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" },
348 { ELSE, TOKENTYPE_NONE, "ELSE" },
349 { ELSIF, TOKENTYPE_IVAL, "ELSIF" },
350 { EQOP, TOKENTYPE_OPNUM, "EQOP" },
351 { FOR, TOKENTYPE_IVAL, "FOR" },
352 { FORMAT, TOKENTYPE_NONE, "FORMAT" },
353 { FUNC, TOKENTYPE_OPNUM, "FUNC" },
354 { FUNC0, TOKENTYPE_OPNUM, "FUNC0" },
7eb971ee 355 { FUNC0OP, TOKENTYPE_OPVAL, "FUNC0OP" },
bbf60fe6
DM
356 { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" },
357 { FUNC1, TOKENTYPE_OPNUM, "FUNC1" },
358 { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" },
0d863452 359 { GIVEN, TOKENTYPE_IVAL, "GIVEN" },
bbf60fe6
DM
360 { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" },
361 { IF, TOKENTYPE_IVAL, "IF" },
5db1eb8d 362 { LABEL, TOKENTYPE_OPVAL, "LABEL" },
bbf60fe6
DM
363 { LOCAL, TOKENTYPE_IVAL, "LOCAL" },
364 { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" },
365 { LSTOP, TOKENTYPE_OPNUM, "LSTOP" },
366 { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" },
367 { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" },
368 { METHOD, TOKENTYPE_OPVAL, "METHOD" },
369 { MULOP, TOKENTYPE_OPNUM, "MULOP" },
370 { MY, TOKENTYPE_IVAL, "MY" },
371 { MYSUB, TOKENTYPE_NONE, "MYSUB" },
372 { NOAMP, TOKENTYPE_NONE, "NOAMP" },
373 { NOTOP, TOKENTYPE_NONE, "NOTOP" },
374 { OROP, TOKENTYPE_IVAL, "OROP" },
375 { OROR, TOKENTYPE_NONE, "OROR" },
376 { PACKAGE, TOKENTYPE_NONE, "PACKAGE" },
88e1f1a2
JV
377 { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" },
378 { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" },
bbf60fe6
DM
379 { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" },
380 { POSTDEC, TOKENTYPE_NONE, "POSTDEC" },
381 { POSTINC, TOKENTYPE_NONE, "POSTINC" },
382 { POWOP, TOKENTYPE_OPNUM, "POWOP" },
383 { PREDEC, TOKENTYPE_NONE, "PREDEC" },
384 { PREINC, TOKENTYPE_NONE, "PREINC" },
385 { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" },
386 { REFGEN, TOKENTYPE_NONE, "REFGEN" },
387 { RELOP, TOKENTYPE_OPNUM, "RELOP" },
388 { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" },
389 { SUB, TOKENTYPE_NONE, "SUB" },
390 { THING, TOKENTYPE_OPVAL, "THING" },
391 { UMINUS, TOKENTYPE_NONE, "UMINUS" },
392 { UNIOP, TOKENTYPE_OPNUM, "UNIOP" },
393 { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" },
394 { UNLESS, TOKENTYPE_IVAL, "UNLESS" },
395 { UNTIL, TOKENTYPE_IVAL, "UNTIL" },
396 { USE, TOKENTYPE_IVAL, "USE" },
0d863452 397 { WHEN, TOKENTYPE_IVAL, "WHEN" },
bbf60fe6
DM
398 { WHILE, TOKENTYPE_IVAL, "WHILE" },
399 { WORD, TOKENTYPE_OPVAL, "WORD" },
be25f609 400 { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" },
c35e046a 401 { 0, TOKENTYPE_NONE, NULL }
bbf60fe6
DM
402};
403
6154021b 404/* dump the returned token in rv, plus any optional arg in pl_yylval */
998054bd 405
bbf60fe6 406STATIC int
704d4215 407S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp)
bbf60fe6 408{
97aff369 409 dVAR;
7918f24d
NC
410
411 PERL_ARGS_ASSERT_TOKEREPORT;
412
bbf60fe6 413 if (DEBUG_T_TEST) {
bd61b366 414 const char *name = NULL;
bbf60fe6 415 enum token_type type = TOKENTYPE_NONE;
f54cb97a 416 const struct debug_tokens *p;
396482e1 417 SV* const report = newSVpvs("<== ");
bbf60fe6 418
f54cb97a 419 for (p = debug_tokens; p->token; p++) {
bbf60fe6
DM
420 if (p->token == (int)rv) {
421 name = p->name;
422 type = p->type;
423 break;
424 }
425 }
426 if (name)
54667de8 427 Perl_sv_catpv(aTHX_ report, name);
bbf60fe6
DM
428 else if ((char)rv > ' ' && (char)rv < '~')
429 Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv);
430 else if (!rv)
396482e1 431 sv_catpvs(report, "EOF");
bbf60fe6
DM
432 else
433 Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv);
434 switch (type) {
435 case TOKENTYPE_NONE:
436 case TOKENTYPE_GVVAL: /* doesn't appear to be used */
437 break;
438 case TOKENTYPE_IVAL:
704d4215 439 Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival);
bbf60fe6
DM
440 break;
441 case TOKENTYPE_OPNUM:
442 Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)",
704d4215 443 PL_op_name[lvalp->ival]);
bbf60fe6
DM
444 break;
445 case TOKENTYPE_PVAL:
704d4215 446 Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval);
bbf60fe6
DM
447 break;
448 case TOKENTYPE_OPVAL:
704d4215 449 if (lvalp->opval) {
401441c0 450 Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)",
704d4215
GG
451 PL_op_name[lvalp->opval->op_type]);
452 if (lvalp->opval->op_type == OP_CONST) {
b6007c36 453 Perl_sv_catpvf(aTHX_ report, " %s",
704d4215 454 SvPEEK(cSVOPx_sv(lvalp->opval)));
b6007c36
DM
455 }
456
457 }
401441c0 458 else
396482e1 459 sv_catpvs(report, "(opval=null)");
bbf60fe6
DM
460 break;
461 }
b6007c36 462 PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report));
bbf60fe6
DM
463 };
464 return (int)rv;
998054bd
SC
465}
466
b6007c36
DM
467
468/* print the buffer with suitable escapes */
469
470STATIC void
15f169a1 471S_printbuf(pTHX_ const char *const fmt, const char *const s)
b6007c36 472{
396482e1 473 SV* const tmp = newSVpvs("");
7918f24d
NC
474
475 PERL_ARGS_ASSERT_PRINTBUF;
476
b6007c36
DM
477 PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60));
478 SvREFCNT_dec(tmp);
479}
480
8fa7f367
JH
481#endif
482
8290c323
NC
483static int
484S_deprecate_commaless_var_list(pTHX) {
485 PL_expect = XTERM;
486 deprecate("comma-less variable list");
487 return REPORT(','); /* grandfather non-comma-format format */
488}
489
ffb4593c
NT
490/*
491 * S_ao
492 *
c963b151
BD
493 * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR
494 * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN
ffb4593c
NT
495 */
496
76e3520e 497STATIC int
cea2e8a9 498S_ao(pTHX_ int toketype)
a0d0e21e 499{
97aff369 500 dVAR;
3280af22
NIS
501 if (*PL_bufptr == '=') {
502 PL_bufptr++;
a0d0e21e 503 if (toketype == ANDAND)
6154021b 504 pl_yylval.ival = OP_ANDASSIGN;
a0d0e21e 505 else if (toketype == OROR)
6154021b 506 pl_yylval.ival = OP_ORASSIGN;
c963b151 507 else if (toketype == DORDOR)
6154021b 508 pl_yylval.ival = OP_DORASSIGN;
a0d0e21e
LW
509 toketype = ASSIGNOP;
510 }
511 return toketype;
512}
513
ffb4593c
NT
514/*
515 * S_no_op
516 * When Perl expects an operator and finds something else, no_op
517 * prints the warning. It always prints "<something> found where
518 * operator expected. It prints "Missing semicolon on previous line?"
519 * if the surprise occurs at the start of the line. "do you need to
520 * predeclare ..." is printed out for code like "sub bar; foo bar $x"
521 * where the compiler doesn't know if foo is a method call or a function.
522 * It prints "Missing operator before end of line" if there's nothing
523 * after the missing operator, or "... before <...>" if there is something
524 * after the missing operator.
525 */
526
76e3520e 527STATIC void
15f169a1 528S_no_op(pTHX_ const char *const what, char *s)
463ee0b2 529{
97aff369 530 dVAR;
9d4ba2ae
AL
531 char * const oldbp = PL_bufptr;
532 const bool is_first = (PL_oldbufptr == PL_linestart);
68dc0745 533
7918f24d
NC
534 PERL_ARGS_ASSERT_NO_OP;
535
1189a94a
GS
536 if (!s)
537 s = oldbp;
07c798fb 538 else
1189a94a 539 PL_bufptr = s;
734ab321 540 yywarn(Perl_form(aTHX_ "%s found where operator expected", what), UTF ? SVf_UTF8 : 0);
56da5a46
RGS
541 if (ckWARN_d(WARN_SYNTAX)) {
542 if (is_first)
543 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
544 "\t(Missing semicolon on previous line?)\n");
545 else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) {
f54cb97a 546 const char *t;
734ab321
BF
547 for (t = PL_oldoldbufptr; (isALNUM_lazy_if(t,UTF) || *t == ':');
548 t += UTF ? UTF8SKIP(t) : 1)
c35e046a 549 NOOP;
56da5a46
RGS
550 if (t < PL_bufptr && isSPACE(*t))
551 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
734ab321
BF
552 "\t(Do you need to predeclare %"SVf"?)\n",
553 SVfARG(newSVpvn_flags(PL_oldoldbufptr, (STRLEN)(t - PL_oldoldbufptr),
554 SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
56da5a46
RGS
555 }
556 else {
557 assert(s >= oldbp);
558 Perl_warner(aTHX_ packWARN(WARN_SYNTAX),
734ab321
BF
559 "\t(Missing operator before %"SVf"?)\n",
560 SVfARG(newSVpvn_flags(oldbp, (STRLEN)(s - oldbp),
561 SVs_TEMP | (UTF ? SVf_UTF8 : 0))));
56da5a46 562 }
07c798fb 563 }
3280af22 564 PL_bufptr = oldbp;
8990e307
LW
565}
566
ffb4593c
NT
567/*
568 * S_missingterm
569 * Complain about missing quote/regexp/heredoc terminator.
d4c19fe8 570 * If it's called with NULL then it cauterizes the line buffer.
ffb4593c
NT
571 * If we're in a delimited string and the delimiter is a control
572 * character, it's reformatted into a two-char sequence like ^C.
573 * This is fatal.
574 */
575
76e3520e 576STATIC void
cea2e8a9 577S_missingterm(pTHX_ char *s)
8990e307 578{
97aff369 579 dVAR;
8990e307
LW
580 char tmpbuf[3];
581 char q;
582 if (s) {
9d4ba2ae 583 char * const nl = strrchr(s,'\n');
d2719217 584 if (nl)
8990e307
LW
585 *nl = '\0';
586 }
463559e7 587 else if (isCNTRL(PL_multi_close)) {
8990e307 588 *tmpbuf = '^';
585ec06d 589 tmpbuf[1] = (char)toCTRL(PL_multi_close);
8990e307
LW
590 tmpbuf[2] = '\0';
591 s = tmpbuf;
592 }
593 else {
eb160463 594 *tmpbuf = (char)PL_multi_close;
8990e307
LW
595 tmpbuf[1] = '\0';
596 s = tmpbuf;
597 }
598 q = strchr(s,'"') ? '\'' : '"';
cea2e8a9 599 Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q);
463ee0b2 600}
79072805 601
dd0ac2b9
FC
602#include "feature.h"
603
0d863452 604/*
0d863452
RH
605 * Check whether the named feature is enabled.
606 */
26ea9e12 607bool
3fff3427 608Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen)
0d863452 609{
97aff369 610 dVAR;
4a731d7b 611 char he_name[8 + MAX_FEATURE_LEN] = "feature_";
7918f24d
NC
612
613 PERL_ARGS_ASSERT_FEATURE_IS_ENABLED;
ca4d40c4
FC
614
615 assert(CURRENT_FEATURE_BUNDLE == FEATURE_BUNDLE_CUSTOM);
7918f24d 616
26ea9e12
NC
617 if (namelen > MAX_FEATURE_LEN)
618 return FALSE;
3fff3427 619 memcpy(&he_name[8], name, namelen);
7d69d4a6 620
c8ca97b0
NC
621 return cBOOL(cop_hints_fetch_pvn(PL_curcop, he_name, 8 + namelen, 0,
622 REFCOUNTED_HE_EXISTS));
0d863452
RH
623}
624
ffb4593c 625/*
9cbb5ea2
GS
626 * experimental text filters for win32 carriage-returns, utf16-to-utf8 and
627 * utf16-to-utf8-reversed.
ffb4593c
NT
628 */
629
c39cd008
GS
630#ifdef PERL_CR_FILTER
631static void
632strip_return(SV *sv)
633{
95a20fc0 634 register const char *s = SvPVX_const(sv);
9d4ba2ae 635 register const char * const e = s + SvCUR(sv);
7918f24d
NC
636
637 PERL_ARGS_ASSERT_STRIP_RETURN;
638
c39cd008
GS
639 /* outer loop optimized to do nothing if there are no CR-LFs */
640 while (s < e) {
641 if (*s++ == '\r' && *s == '\n') {
642 /* hit a CR-LF, need to copy the rest */
643 register char *d = s - 1;
644 *d++ = *s++;
645 while (s < e) {
646 if (*s == '\r' && s[1] == '\n')
647 s++;
648 *d++ = *s++;
649 }
650 SvCUR(sv) -= s - d;
651 return;
652 }
653 }
654}
a868473f 655
76e3520e 656STATIC I32
c39cd008 657S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen)
a868473f 658{
f54cb97a 659 const I32 count = FILTER_READ(idx+1, sv, maxlen);
c39cd008
GS
660 if (count > 0 && !maxlen)
661 strip_return(sv);
662 return count;
a868473f
NIS
663}
664#endif
665
ffb4593c 666/*
8eaa0acf
Z
667=for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags
668
669Creates and initialises a new lexer/parser state object, supplying
670a context in which to lex and parse from a new source of Perl code.
671A pointer to the new state object is placed in L</PL_parser>. An entry
672is made on the save stack so that upon unwinding the new state object
673will be destroyed and the former value of L</PL_parser> will be restored.
674Nothing else need be done to clean up the parsing context.
675
676The code to be parsed comes from I<line> and I<rsfp>. I<line>, if
677non-null, provides a string (in SV form) containing code to be parsed.
678A copy of the string is made, so subsequent modification of I<line>
679does not affect parsing. I<rsfp>, if non-null, provides an input stream
680from which code will be read to be parsed. If both are non-null, the
681code in I<line> comes first and must consist of complete lines of input,
682and I<rsfp> supplies the remainder of the source.
683
e368b3bd
FC
684The I<flags> parameter is reserved for future use. Currently it is only
685used by perl internally, so extensions should always pass zero.
8eaa0acf
Z
686
687=cut
688*/
ffb4593c 689
27fcb6ee 690/* LEX_START_SAME_FILTER indicates that this is not a new file, so it
87606032
NC
691 can share filters with the current parser.
692 LEX_START_DONT_CLOSE indicates that the file handle wasn't opened by the
693 caller, hence isn't owned by the parser, so shouldn't be closed on parser
694 destruction. This is used to handle the case of defaulting to reading the
695 script from the standard input because no filename was given on the command
696 line (without getting confused by situation where STDIN has been closed, so
697 the script handle is opened on fd 0) */
27fcb6ee 698
a0d0e21e 699void
8eaa0acf 700Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags)
79072805 701{
97aff369 702 dVAR;
6ef55633 703 const char *s = NULL;
5486870f 704 yy_parser *parser, *oparser;
60d63348 705 if (flags && flags & ~LEX_START_FLAGS)
8eaa0acf 706 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start");
acdf0a21
DM
707
708 /* create and initialise a parser */
709
199e78b7 710 Newxz(parser, 1, yy_parser);
5486870f 711 parser->old_parser = oparser = PL_parser;
acdf0a21
DM
712 PL_parser = parser;
713
28ac2b49
Z
714 parser->stack = NULL;
715 parser->ps = NULL;
716 parser->stack_size = 0;
acdf0a21 717
e3abe207
DM
718 /* on scope exit, free this parser and restore any outer one */
719 SAVEPARSER(parser);
7c4baf47 720 parser->saved_curcop = PL_curcop;
e3abe207 721
acdf0a21 722 /* initialise lexer state */
8990e307 723
fb205e7a
DM
724#ifdef PERL_MAD
725 parser->curforce = -1;
726#else
727 parser->nexttoke = 0;
728#endif
ca4cfd28 729 parser->error_count = oparser ? oparser->error_count : 0;
c2598295 730 parser->copline = NOLINE;
5afb0a62 731 parser->lex_state = LEX_NORMAL;
c2598295 732 parser->expect = XSTATE;
2f9285f8 733 parser->rsfp = rsfp;
27fcb6ee
FC
734 parser->rsfp_filters =
735 !(flags & LEX_START_SAME_FILTER) || !oparser
d3cd8e11
FC
736 ? NULL
737 : MUTABLE_AV(SvREFCNT_inc(
738 oparser->rsfp_filters
739 ? oparser->rsfp_filters
740 : (oparser->rsfp_filters = newAV())
741 ));
2f9285f8 742
199e78b7
DM
743 Newx(parser->lex_brackstack, 120, char);
744 Newx(parser->lex_casestack, 12, char);
745 *parser->lex_casestack = '\0';
02b34bbe 746
10efb74f 747 if (line) {
0528fd32 748 STRLEN len;
10efb74f 749 s = SvPV_const(line, len);
0abcdfa4
FC
750 parser->linestr = flags & LEX_START_COPIED
751 ? SvREFCNT_inc_simple_NN(line)
752 : newSVpvn_flags(s, len, SvUTF8(line));
37555a72 753 if (!len || s[len-1] != ';')
bdc0bf6f 754 sv_catpvs(parser->linestr, "\n;");
0abcdfa4
FC
755 } else {
756 parser->linestr = newSVpvs("\n;");
8990e307 757 }
f06b5848
DM
758 parser->oldoldbufptr =
759 parser->oldbufptr =
760 parser->bufptr =
761 parser->linestart = SvPVX(parser->linestr);
762 parser->bufend = parser->bufptr + SvCUR(parser->linestr);
763 parser->last_lop = parser->last_uni = NULL;
87606032
NC
764 parser->lex_flags = flags & (LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES
765 |LEX_DONT_CLOSE_RSFP);
737c24fc 766
60d63348 767 parser->in_pod = parser->filtered = 0;
79072805 768}
a687059c 769
e3abe207
DM
770
771/* delete a parser object */
772
773void
774Perl_parser_free(pTHX_ const yy_parser *parser)
775{
7918f24d
NC
776 PERL_ARGS_ASSERT_PARSER_FREE;
777
7c4baf47 778 PL_curcop = parser->saved_curcop;
bdc0bf6f
DM
779 SvREFCNT_dec(parser->linestr);
780
87606032 781 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
2f9285f8 782 PerlIO_clearerr(parser->rsfp);
799361c3
SH
783 else if (parser->rsfp && (!parser->old_parser ||
784 (parser->old_parser && parser->rsfp != parser->old_parser->rsfp)))
2f9285f8 785 PerlIO_close(parser->rsfp);
5486870f 786 SvREFCNT_dec(parser->rsfp_filters);
2f9285f8 787
e3abe207
DM
788 Safefree(parser->lex_brackstack);
789 Safefree(parser->lex_casestack);
790 PL_parser = parser->old_parser;
791 Safefree(parser);
792}
793
794
ffb4593c 795/*
f0e67a1d
Z
796=for apidoc AmxU|SV *|PL_parser-E<gt>linestr
797
798Buffer scalar containing the chunk currently under consideration of the
799text currently being lexed. This is always a plain string scalar (for
800which C<SvPOK> is true). It is not intended to be used as a scalar by
801normal scalar means; instead refer to the buffer directly by the pointer
802variables described below.
803
804The lexer maintains various C<char*> pointers to things in the
805C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever
806reallocated, all of these pointers must be updated. Don't attempt to
807do this manually, but rather use L</lex_grow_linestr> if you need to
808reallocate the buffer.
809
810The content of the text chunk in the buffer is commonly exactly one
811complete line of input, up to and including a newline terminator,
812but there are situations where it is otherwise. The octets of the
813buffer may be intended to be interpreted as either UTF-8 or Latin-1.
814The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8>
815flag on this scalar, which may disagree with it.
816
817For direct examination of the buffer, the variable
818L</PL_parser-E<gt>bufend> points to the end of the buffer. The current
819lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use
820of these pointers is usually preferable to examination of the scalar
821through normal scalar means.
822
823=for apidoc AmxU|char *|PL_parser-E<gt>bufend
824
825Direct pointer to the end of the chunk of text currently being lexed, the
826end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr)
827+ SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is
828always located at the end of the buffer, and does not count as part of
829the buffer's contents.
830
831=for apidoc AmxU|char *|PL_parser-E<gt>bufptr
832
833Points to the current position of lexing inside the lexer buffer.
834Characters around this point may be freely examined, within
835the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and
836L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be
837interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>.
838
839Lexing code (whether in the Perl core or not) moves this pointer past
840the characters that it consumes. It is also expected to perform some
841bookkeeping whenever a newline character is consumed. This movement
842can be more conveniently performed by the function L</lex_read_to>,
843which handles newlines appropriately.
844
845Interpretation of the buffer's octets can be abstracted out by
846using the slightly higher-level functions L</lex_peek_unichar> and
847L</lex_read_unichar>.
848
849=for apidoc AmxU|char *|PL_parser-E<gt>linestart
850
851Points to the start of the current line inside the lexer buffer.
852This is useful for indicating at which column an error occurred, and
853not much else. This must be updated by any lexing code that consumes
854a newline; the function L</lex_read_to> handles this detail.
855
856=cut
857*/
858
859/*
860=for apidoc Amx|bool|lex_bufutf8
861
862Indicates whether the octets in the lexer buffer
863(L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding
864of Unicode characters. If not, they should be interpreted as Latin-1
865characters. This is analogous to the C<SvUTF8> flag for scalars.
866
867In UTF-8 mode, it is not guaranteed that the lexer buffer actually
868contains valid UTF-8. Lexing code must be robust in the face of invalid
869encoding.
870
871The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar
872is significant, but not the whole story regarding the input character
873encoding. Normally, when a file is being read, the scalar contains octets
874and its C<SvUTF8> flag is off, but the octets should be interpreted as
875UTF-8 if the C<use utf8> pragma is in effect. During a string eval,
876however, the scalar may have the C<SvUTF8> flag on, and in this case its
877octets should be interpreted as UTF-8 unless the C<use bytes> pragma
878is in effect. This logic may change in the future; use this function
879instead of implementing the logic yourself.
880
881=cut
882*/
883
884bool
885Perl_lex_bufutf8(pTHX)
886{
887 return UTF;
888}
889
890/*
891=for apidoc Amx|char *|lex_grow_linestr|STRLEN len
892
893Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate
894at least I<len> octets (including terminating NUL). Returns a
895pointer to the reallocated buffer. This is necessary before making
896any direct modification of the buffer that would increase its length.
897L</lex_stuff_pvn> provides a more convenient way to insert text into
898the buffer.
899
900Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>;
901this function updates all of the lexer's variables that point directly
902into the buffer.
903
904=cut
905*/
906
907char *
908Perl_lex_grow_linestr(pTHX_ STRLEN len)
909{
910 SV *linestr;
911 char *buf;
912 STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
913 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
914 linestr = PL_parser->linestr;
915 buf = SvPVX(linestr);
916 if (len <= SvLEN(linestr))
917 return buf;
918 bufend_pos = PL_parser->bufend - buf;
919 bufptr_pos = PL_parser->bufptr - buf;
920 oldbufptr_pos = PL_parser->oldbufptr - buf;
921 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
922 linestart_pos = PL_parser->linestart - buf;
923 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
924 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
925 buf = sv_grow(linestr, len);
926 PL_parser->bufend = buf + bufend_pos;
927 PL_parser->bufptr = buf + bufptr_pos;
928 PL_parser->oldbufptr = buf + oldbufptr_pos;
929 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
930 PL_parser->linestart = buf + linestart_pos;
931 if (PL_parser->last_uni)
932 PL_parser->last_uni = buf + last_uni_pos;
933 if (PL_parser->last_lop)
934 PL_parser->last_lop = buf + last_lop_pos;
935 return buf;
936}
937
938/*
83aa740e 939=for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags
f0e67a1d
Z
940
941Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
942immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
943reallocating the buffer if necessary. This means that lexing code that
944runs later will see the characters as if they had appeared in the input.
945It is not recommended to do this as part of normal parsing, and most
946uses of this facility run the risk of the inserted characters being
947interpreted in an unintended manner.
948
949The string to be inserted is represented by I<len> octets starting
950at I<pv>. These octets are interpreted as either UTF-8 or Latin-1,
951according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>.
952The characters are recoded for the lexer buffer, according to how the
953buffer is currently being interpreted (L</lex_bufutf8>). If a string
9dcc53ea 954to be inserted is available as a Perl scalar, the L</lex_stuff_sv>
f0e67a1d
Z
955function is more convenient.
956
957=cut
958*/
959
960void
83aa740e 961Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags)
f0e67a1d 962{
749123ff 963 dVAR;
f0e67a1d
Z
964 char *bufptr;
965 PERL_ARGS_ASSERT_LEX_STUFF_PVN;
966 if (flags & ~(LEX_STUFF_UTF8))
967 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn");
968 if (UTF) {
969 if (flags & LEX_STUFF_UTF8) {
970 goto plain_copy;
971 } else {
972 STRLEN highhalf = 0;
83aa740e 973 const char *p, *e = pv+len;
f0e67a1d
Z
974 for (p = pv; p != e; p++)
975 highhalf += !!(((U8)*p) & 0x80);
976 if (!highhalf)
977 goto plain_copy;
978 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf);
979 bufptr = PL_parser->bufptr;
980 Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
981 SvCUR_set(PL_parser->linestr,
982 SvCUR(PL_parser->linestr) + len+highhalf);
f0e67a1d
Z
983 PL_parser->bufend += len+highhalf;
984 for (p = pv; p != e; p++) {
985 U8 c = (U8)*p;
986 if (c & 0x80) {
987 *bufptr++ = (char)(0xc0 | (c >> 6));
988 *bufptr++ = (char)(0x80 | (c & 0x3f));
989 } else {
990 *bufptr++ = (char)c;
991 }
992 }
993 }
994 } else {
995 if (flags & LEX_STUFF_UTF8) {
996 STRLEN highhalf = 0;
83aa740e 997 const char *p, *e = pv+len;
f0e67a1d
Z
998 for (p = pv; p != e; p++) {
999 U8 c = (U8)*p;
1000 if (c >= 0xc4) {
1001 Perl_croak(aTHX_ "Lexing code attempted to stuff "
1002 "non-Latin-1 character into Latin-1 input");
1003 } else if (c >= 0xc2 && p+1 != e &&
1004 (((U8)p[1]) & 0xc0) == 0x80) {
1005 p++;
1006 highhalf++;
1007 } else if (c >= 0x80) {
1008 /* malformed UTF-8 */
1009 ENTER;
1010 SAVESPTR(PL_warnhook);
1011 PL_warnhook = PERL_WARNHOOK_FATAL;
1012 utf8n_to_uvuni((U8*)p, e-p, NULL, 0);
1013 LEAVE;
1014 }
1015 }
1016 if (!highhalf)
1017 goto plain_copy;
1018 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf);
1019 bufptr = PL_parser->bufptr;
1020 Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char);
255fdf19
Z
1021 SvCUR_set(PL_parser->linestr,
1022 SvCUR(PL_parser->linestr) + len-highhalf);
f0e67a1d
Z
1023 PL_parser->bufend += len-highhalf;
1024 for (p = pv; p != e; p++) {
1025 U8 c = (U8)*p;
1026 if (c & 0x80) {
1027 *bufptr++ = (char)(((c & 0x3) << 6) | (p[1] & 0x3f));
1028 p++;
1029 } else {
1030 *bufptr++ = (char)c;
1031 }
1032 }
1033 } else {
1034 plain_copy:
1035 lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len);
1036 bufptr = PL_parser->bufptr;
1037 Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char);
255fdf19 1038 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len);
f0e67a1d
Z
1039 PL_parser->bufend += len;
1040 Copy(pv, bufptr, len, char);
1041 }
1042 }
1043}
1044
1045/*
9dcc53ea
Z
1046=for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags
1047
1048Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1049immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1050reallocating the buffer if necessary. This means that lexing code that
1051runs later will see the characters as if they had appeared in the input.
1052It is not recommended to do this as part of normal parsing, and most
1053uses of this facility run the risk of the inserted characters being
1054interpreted in an unintended manner.
1055
1056The string to be inserted is represented by octets starting at I<pv>
1057and continuing to the first nul. These octets are interpreted as either
1058UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set
1059in I<flags>. The characters are recoded for the lexer buffer, according
1060to how the buffer is currently being interpreted (L</lex_bufutf8>).
1061If it is not convenient to nul-terminate a string to be inserted, the
1062L</lex_stuff_pvn> function is more appropriate.
1063
1064=cut
1065*/
1066
1067void
1068Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags)
1069{
1070 PERL_ARGS_ASSERT_LEX_STUFF_PV;
1071 lex_stuff_pvn(pv, strlen(pv), flags);
1072}
1073
1074/*
f0e67a1d
Z
1075=for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags
1076
1077Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>),
1078immediately after the current lexing point (L</PL_parser-E<gt>bufptr>),
1079reallocating the buffer if necessary. This means that lexing code that
1080runs later will see the characters as if they had appeared in the input.
1081It is not recommended to do this as part of normal parsing, and most
1082uses of this facility run the risk of the inserted characters being
1083interpreted in an unintended manner.
1084
1085The string to be inserted is the string value of I<sv>. The characters
1086are recoded for the lexer buffer, according to how the buffer is currently
9dcc53ea 1087being interpreted (L</lex_bufutf8>). If a string to be inserted is
f0e67a1d
Z
1088not already a Perl scalar, the L</lex_stuff_pvn> function avoids the
1089need to construct a scalar.
1090
1091=cut
1092*/
1093
1094void
1095Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags)
1096{
1097 char *pv;
1098 STRLEN len;
1099 PERL_ARGS_ASSERT_LEX_STUFF_SV;
1100 if (flags)
1101 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv");
1102 pv = SvPV(sv, len);
1103 lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0));
1104}
1105
1106/*
1107=for apidoc Amx|void|lex_unstuff|char *ptr
1108
1109Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to
1110I<ptr>. Text following I<ptr> will be moved, and the buffer shortened.
1111This hides the discarded text from any lexing code that runs later,
1112as if the text had never appeared.
1113
1114This is not the normal way to consume lexed text. For that, use
1115L</lex_read_to>.
1116
1117=cut
1118*/
1119
1120void
1121Perl_lex_unstuff(pTHX_ char *ptr)
1122{
1123 char *buf, *bufend;
1124 STRLEN unstuff_len;
1125 PERL_ARGS_ASSERT_LEX_UNSTUFF;
1126 buf = PL_parser->bufptr;
1127 if (ptr < buf)
1128 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1129 if (ptr == buf)
1130 return;
1131 bufend = PL_parser->bufend;
1132 if (ptr > bufend)
1133 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff");
1134 unstuff_len = ptr - buf;
1135 Move(ptr, buf, bufend+1-ptr, char);
1136 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len);
1137 PL_parser->bufend = bufend - unstuff_len;
1138}
1139
1140/*
1141=for apidoc Amx|void|lex_read_to|char *ptr
1142
1143Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up
1144to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>,
1145performing the correct bookkeeping whenever a newline character is passed.
1146This is the normal way to consume lexed text.
1147
1148Interpretation of the buffer's octets can be abstracted out by
1149using the slightly higher-level functions L</lex_peek_unichar> and
1150L</lex_read_unichar>.
1151
1152=cut
1153*/
1154
1155void
1156Perl_lex_read_to(pTHX_ char *ptr)
1157{
1158 char *s;
1159 PERL_ARGS_ASSERT_LEX_READ_TO;
1160 s = PL_parser->bufptr;
1161 if (ptr < s || ptr > PL_parser->bufend)
1162 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to");
1163 for (; s != ptr; s++)
1164 if (*s == '\n') {
1165 CopLINE_inc(PL_curcop);
1166 PL_parser->linestart = s+1;
1167 }
1168 PL_parser->bufptr = ptr;
1169}
1170
1171/*
1172=for apidoc Amx|void|lex_discard_to|char *ptr
1173
1174Discards the first part of the L</PL_parser-E<gt>linestr> buffer,
1175up to I<ptr>. The remaining content of the buffer will be moved, and
1176all pointers into the buffer updated appropriately. I<ptr> must not
1177be later in the buffer than the position of L</PL_parser-E<gt>bufptr>:
1178it is not permitted to discard text that has yet to be lexed.
1179
1180Normally it is not necessarily to do this directly, because it suffices to
1181use the implicit discarding behaviour of L</lex_next_chunk> and things
1182based on it. However, if a token stretches across multiple lines,
1f317c95 1183and the lexing code has kept multiple lines of text in the buffer for
f0e67a1d
Z
1184that purpose, then after completion of the token it would be wise to
1185explicitly discard the now-unneeded earlier lines, to avoid future
1186multi-line tokens growing the buffer without bound.
1187
1188=cut
1189*/
1190
1191void
1192Perl_lex_discard_to(pTHX_ char *ptr)
1193{
1194 char *buf;
1195 STRLEN discard_len;
1196 PERL_ARGS_ASSERT_LEX_DISCARD_TO;
1197 buf = SvPVX(PL_parser->linestr);
1198 if (ptr < buf)
1199 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1200 if (ptr == buf)
1201 return;
1202 if (ptr > PL_parser->bufptr)
1203 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to");
1204 discard_len = ptr - buf;
1205 if (PL_parser->oldbufptr < ptr)
1206 PL_parser->oldbufptr = ptr;
1207 if (PL_parser->oldoldbufptr < ptr)
1208 PL_parser->oldoldbufptr = ptr;
1209 if (PL_parser->last_uni && PL_parser->last_uni < ptr)
1210 PL_parser->last_uni = NULL;
1211 if (PL_parser->last_lop && PL_parser->last_lop < ptr)
1212 PL_parser->last_lop = NULL;
1213 Move(ptr, buf, PL_parser->bufend+1-ptr, char);
1214 SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len);
1215 PL_parser->bufend -= discard_len;
1216 PL_parser->bufptr -= discard_len;
1217 PL_parser->oldbufptr -= discard_len;
1218 PL_parser->oldoldbufptr -= discard_len;
1219 if (PL_parser->last_uni)
1220 PL_parser->last_uni -= discard_len;
1221 if (PL_parser->last_lop)
1222 PL_parser->last_lop -= discard_len;
1223}
1224
1225/*
1226=for apidoc Amx|bool|lex_next_chunk|U32 flags
1227
1228Reads in the next chunk of text to be lexed, appending it to
1229L</PL_parser-E<gt>linestr>. This should be called when lexing code has
1230looked to the end of the current chunk and wants to know more. It is
1231usual, but not necessary, for lexing to have consumed the entirety of
1232the current chunk at this time.
1233
1234If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current
1235chunk (i.e., the current chunk has been entirely consumed), normally the
1236current chunk will be discarded at the same time that the new chunk is
1237read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk
1238will not be discarded. If the current chunk has not been entirely
1239consumed, then it will not be discarded regardless of the flag.
1240
1241Returns true if some new text was added to the buffer, or false if the
1242buffer has reached the end of the input text.
1243
1244=cut
1245*/
1246
1247#define LEX_FAKE_EOF 0x80000000
1248
1249bool
1250Perl_lex_next_chunk(pTHX_ U32 flags)
1251{
1252 SV *linestr;
1253 char *buf;
1254 STRLEN old_bufend_pos, new_bufend_pos;
1255 STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos;
1256 STRLEN linestart_pos, last_uni_pos, last_lop_pos;
17cc9359 1257 bool got_some_for_debugger = 0;
f0e67a1d
Z
1258 bool got_some;
1259 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF))
1260 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk");
f0e67a1d
Z
1261 linestr = PL_parser->linestr;
1262 buf = SvPVX(linestr);
1263 if (!(flags & LEX_KEEP_PREVIOUS) &&
1264 PL_parser->bufptr == PL_parser->bufend) {
1265 old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0;
1266 linestart_pos = 0;
1267 if (PL_parser->last_uni != PL_parser->bufend)
1268 PL_parser->last_uni = NULL;
1269 if (PL_parser->last_lop != PL_parser->bufend)
1270 PL_parser->last_lop = NULL;
1271 last_uni_pos = last_lop_pos = 0;
1272 *buf = 0;
1273 SvCUR(linestr) = 0;
1274 } else {
1275 old_bufend_pos = PL_parser->bufend - buf;
1276 bufptr_pos = PL_parser->bufptr - buf;
1277 oldbufptr_pos = PL_parser->oldbufptr - buf;
1278 oldoldbufptr_pos = PL_parser->oldoldbufptr - buf;
1279 linestart_pos = PL_parser->linestart - buf;
1280 last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0;
1281 last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0;
1282 }
1283 if (flags & LEX_FAKE_EOF) {
1284 goto eof;
60d63348 1285 } else if (!PL_parser->rsfp && !PL_parser->filtered) {
f0e67a1d
Z
1286 got_some = 0;
1287 } else if (filter_gets(linestr, old_bufend_pos)) {
1288 got_some = 1;
17cc9359 1289 got_some_for_debugger = 1;
f0e67a1d 1290 } else {
580561a3
Z
1291 if (!SvPOK(linestr)) /* can get undefined by filter_gets */
1292 sv_setpvs(linestr, "");
f0e67a1d
Z
1293 eof:
1294 /* End of real input. Close filehandle (unless it was STDIN),
1295 * then add implicit termination.
1296 */
87606032 1297 if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP)
f0e67a1d
Z
1298 PerlIO_clearerr(PL_parser->rsfp);
1299 else if (PL_parser->rsfp)
1300 (void)PerlIO_close(PL_parser->rsfp);
1301 PL_parser->rsfp = NULL;
60d63348 1302 PL_parser->in_pod = PL_parser->filtered = 0;
f0e67a1d
Z
1303#ifdef PERL_MAD
1304 if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n))
1305 PL_faketokens = 1;
1306#endif
1307 if (!PL_in_eval && PL_minus_p) {
1308 sv_catpvs(linestr,
1309 /*{*/";}continue{print or die qq(-p destination: $!\\n);}");
1310 PL_minus_n = PL_minus_p = 0;
1311 } else if (!PL_in_eval && PL_minus_n) {
1312 sv_catpvs(linestr, /*{*/";}");
1313 PL_minus_n = 0;
1314 } else
1315 sv_catpvs(linestr, ";");
1316 got_some = 1;
1317 }
1318 buf = SvPVX(linestr);
1319 new_bufend_pos = SvCUR(linestr);
1320 PL_parser->bufend = buf + new_bufend_pos;
1321 PL_parser->bufptr = buf + bufptr_pos;
1322 PL_parser->oldbufptr = buf + oldbufptr_pos;
1323 PL_parser->oldoldbufptr = buf + oldoldbufptr_pos;
1324 PL_parser->linestart = buf + linestart_pos;
1325 if (PL_parser->last_uni)
1326 PL_parser->last_uni = buf + last_uni_pos;
1327 if (PL_parser->last_lop)
1328 PL_parser->last_lop = buf + last_lop_pos;
17cc9359 1329 if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) &&
f0e67a1d
Z
1330 PL_curstash != PL_debstash) {
1331 /* debugger active and we're not compiling the debugger code,
1332 * so store the line into the debugger's array of lines
1333 */
1334 update_debugger_info(NULL, buf+old_bufend_pos,
1335 new_bufend_pos-old_bufend_pos);
1336 }
1337 return got_some;
1338}
1339
1340/*
1341=for apidoc Amx|I32|lex_peek_unichar|U32 flags
1342
1343Looks ahead one (Unicode) character in the text currently being lexed.
1344Returns the codepoint (unsigned integer value) of the next character,
1345or -1 if lexing has reached the end of the input text. To consume the
1346peeked character, use L</lex_read_unichar>.
1347
1348If the next character is in (or extends into) the next chunk of input
1349text, the next chunk will be read in. Normally the current chunk will be
1350discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1351then the current chunk will not be discarded.
1352
1353If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1354is encountered, an exception is generated.
1355
1356=cut
1357*/
1358
1359I32
1360Perl_lex_peek_unichar(pTHX_ U32 flags)
1361{
749123ff 1362 dVAR;
f0e67a1d
Z
1363 char *s, *bufend;
1364 if (flags & ~(LEX_KEEP_PREVIOUS))
1365 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar");
1366 s = PL_parser->bufptr;
1367 bufend = PL_parser->bufend;
1368 if (UTF) {
1369 U8 head;
1370 I32 unichar;
1371 STRLEN len, retlen;
1372 if (s == bufend) {
1373 if (!lex_next_chunk(flags))
1374 return -1;
1375 s = PL_parser->bufptr;
1376 bufend = PL_parser->bufend;
1377 }
1378 head = (U8)*s;
1379 if (!(head & 0x80))
1380 return head;
1381 if (head & 0x40) {
1382 len = PL_utf8skip[head];
1383 while ((STRLEN)(bufend-s) < len) {
1384 if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS))
1385 break;
1386 s = PL_parser->bufptr;
1387 bufend = PL_parser->bufend;
1388 }
1389 }
1390 unichar = utf8n_to_uvuni((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY);
1391 if (retlen == (STRLEN)-1) {
1392 /* malformed UTF-8 */
1393 ENTER;
1394 SAVESPTR(PL_warnhook);
1395 PL_warnhook = PERL_WARNHOOK_FATAL;
1396 utf8n_to_uvuni((U8*)s, bufend-s, NULL, 0);
1397 LEAVE;
1398 }
1399 return unichar;
1400 } else {
1401 if (s == bufend) {
1402 if (!lex_next_chunk(flags))
1403 return -1;
1404 s = PL_parser->bufptr;
1405 }
1406 return (U8)*s;
1407 }
1408}
1409
1410/*
1411=for apidoc Amx|I32|lex_read_unichar|U32 flags
1412
1413Reads the next (Unicode) character in the text currently being lexed.
1414Returns the codepoint (unsigned integer value) of the character read,
1415and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1
1416if lexing has reached the end of the input text. To non-destructively
1417examine the next character, use L</lex_peek_unichar> instead.
1418
1419If the next character is in (or extends into) the next chunk of input
1420text, the next chunk will be read in. Normally the current chunk will be
1421discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS>
1422then the current chunk will not be discarded.
1423
1424If the input is being interpreted as UTF-8 and a UTF-8 encoding error
1425is encountered, an exception is generated.
1426
1427=cut
1428*/
1429
1430I32
1431Perl_lex_read_unichar(pTHX_ U32 flags)
1432{
1433 I32 c;
1434 if (flags & ~(LEX_KEEP_PREVIOUS))
1435 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar");
1436 c = lex_peek_unichar(flags);
1437 if (c != -1) {
1438 if (c == '\n')
1439 CopLINE_inc(PL_curcop);
d9018cbe
EB
1440 if (UTF)
1441 PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr);
1442 else
1443 ++(PL_parser->bufptr);
f0e67a1d
Z
1444 }
1445 return c;
1446}
1447
1448/*
1449=for apidoc Amx|void|lex_read_space|U32 flags
1450
1451Reads optional spaces, in Perl style, in the text currently being
1452lexed. The spaces may include ordinary whitespace characters and
1453Perl-style comments. C<#line> directives are processed if encountered.
1454L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points
1455at a non-space character (or the end of the input text).
1456
1457If spaces extend into the next chunk of input text, the next chunk will
1458be read in. Normally the current chunk will be discarded at the same
1459time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current
1460chunk will not be discarded.
1461
1462=cut
1463*/
1464
f0998909
Z
1465#define LEX_NO_NEXT_CHUNK 0x80000000
1466
f0e67a1d
Z
1467void
1468Perl_lex_read_space(pTHX_ U32 flags)
1469{
1470 char *s, *bufend;
1471 bool need_incline = 0;
f0998909 1472 if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK))
f0e67a1d
Z
1473 Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space");
1474#ifdef PERL_MAD
1475 if (PL_skipwhite) {
1476 sv_free(PL_skipwhite);
1477 PL_skipwhite = NULL;
1478 }
1479 if (PL_madskills)
1480 PL_skipwhite = newSVpvs("");
1481#endif /* PERL_MAD */
1482 s = PL_parser->bufptr;
1483 bufend = PL_parser->bufend;
1484 while (1) {
1485 char c = *s;
1486 if (c == '#') {
1487 do {
1488 c = *++s;
1489 } while (!(c == '\n' || (c == 0 && s == bufend)));
1490 } else if (c == '\n') {
1491 s++;
1492 PL_parser->linestart = s;
1493 if (s == bufend)
1494 need_incline = 1;
1495 else
1496 incline(s);
1497 } else if (isSPACE(c)) {
1498 s++;
1499 } else if (c == 0 && s == bufend) {
1500 bool got_more;
1501#ifdef PERL_MAD
1502 if (PL_madskills)
1503 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1504#endif /* PERL_MAD */
f0998909
Z
1505 if (flags & LEX_NO_NEXT_CHUNK)
1506 break;
f0e67a1d
Z
1507 PL_parser->bufptr = s;
1508 CopLINE_inc(PL_curcop);
1509 got_more = lex_next_chunk(flags);
1510 CopLINE_dec(PL_curcop);
1511 s = PL_parser->bufptr;
1512 bufend = PL_parser->bufend;
1513 if (!got_more)
1514 break;
1515 if (need_incline && PL_parser->rsfp) {
1516 incline(s);
1517 need_incline = 0;
1518 }
1519 } else {
1520 break;
1521 }
1522 }
1523#ifdef PERL_MAD
1524 if (PL_madskills)
1525 sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr);
1526#endif /* PERL_MAD */
1527 PL_parser->bufptr = s;
1528}
1529
1530/*
ffb4593c
NT
1531 * S_incline
1532 * This subroutine has nothing to do with tilting, whether at windmills
1533 * or pinball tables. Its name is short for "increment line". It
57843af0 1534 * increments the current line number in CopLINE(PL_curcop) and checks
ffb4593c 1535 * to see whether the line starts with a comment of the form
9cbb5ea2
GS
1536 * # line 500 "foo.pm"
1537 * If so, it sets the current line number and file to the values in the comment.
ffb4593c
NT
1538 */
1539
76e3520e 1540STATIC void
d9095cec 1541S_incline(pTHX_ const char *s)
463ee0b2 1542{
97aff369 1543 dVAR;
d9095cec
NC
1544 const char *t;
1545 const char *n;
1546 const char *e;
8818d409 1547 line_t line_num;
463ee0b2 1548
7918f24d
NC
1549 PERL_ARGS_ASSERT_INCLINE;
1550
57843af0 1551 CopLINE_inc(PL_curcop);
463ee0b2
LW
1552 if (*s++ != '#')
1553 return;
d4c19fe8
AL
1554 while (SPACE_OR_TAB(*s))
1555 s++;
73659bf1
GS
1556 if (strnEQ(s, "line", 4))
1557 s += 4;
1558 else
1559 return;
084592ab 1560 if (SPACE_OR_TAB(*s))
73659bf1 1561 s++;
4e553d73 1562 else
73659bf1 1563 return;
d4c19fe8
AL
1564 while (SPACE_OR_TAB(*s))
1565 s++;
463ee0b2
LW
1566 if (!isDIGIT(*s))
1567 return;
d4c19fe8 1568
463ee0b2
LW
1569 n = s;
1570 while (isDIGIT(*s))
1571 s++;
07714eb4 1572 if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0')
26b6dc3f 1573 return;
bf4acbe4 1574 while (SPACE_OR_TAB(*s))
463ee0b2 1575 s++;
73659bf1 1576 if (*s == '"' && (t = strchr(s+1, '"'))) {
463ee0b2 1577 s++;
73659bf1
GS
1578 e = t + 1;
1579 }
463ee0b2 1580 else {
c35e046a
AL
1581 t = s;
1582 while (!isSPACE(*t))
1583 t++;
73659bf1 1584 e = t;
463ee0b2 1585 }
bf4acbe4 1586 while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f')
73659bf1
GS
1587 e++;
1588 if (*e != '\n' && *e != '\0')
1589 return; /* false alarm */
1590
8818d409
FC
1591 line_num = atoi(n)-1;
1592
f4dd75d9 1593 if (t - s > 0) {
d9095cec 1594 const STRLEN len = t - s;
19bad673
NC
1595 SV *const temp_sv = CopFILESV(PL_curcop);
1596 const char *cf;
1597 STRLEN tmplen;
1598
1599 if (temp_sv) {
1600 cf = SvPVX(temp_sv);
1601 tmplen = SvCUR(temp_sv);
1602 } else {
1603 cf = NULL;
1604 tmplen = 0;
1605 }
1606
d1299d44 1607 if (!PL_rsfp && !PL_parser->filtered) {
e66cf94c
RGS
1608 /* must copy *{"::_<(eval N)[oldfilename:L]"}
1609 * to *{"::_<newfilename"} */
44867030
NC
1610 /* However, the long form of evals is only turned on by the
1611 debugger - usually they're "(eval %lu)" */
1612 char smallbuf[128];
1613 char *tmpbuf;
1614 GV **gvp;
d9095cec 1615 STRLEN tmplen2 = len;
798b63bc 1616 if (tmplen + 2 <= sizeof smallbuf)
e66cf94c
RGS
1617 tmpbuf = smallbuf;
1618 else
2ae0db35 1619 Newx(tmpbuf, tmplen + 2, char);
44867030
NC
1620 tmpbuf[0] = '_';
1621 tmpbuf[1] = '<';
2ae0db35 1622 memcpy(tmpbuf + 2, cf, tmplen);
44867030 1623 tmplen += 2;
8a5ee598
RGS
1624 gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE);
1625 if (gvp) {
44867030
NC
1626 char *tmpbuf2;
1627 GV *gv2;
1628
1629 if (tmplen2 + 2 <= sizeof smallbuf)
1630 tmpbuf2 = smallbuf;
1631 else
1632 Newx(tmpbuf2, tmplen2 + 2, char);
1633
1634 if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) {
1635 /* Either they malloc'd it, or we malloc'd it,
1636 so no prefix is present in ours. */
1637 tmpbuf2[0] = '_';
1638 tmpbuf2[1] = '<';
1639 }
1640
1641 memcpy(tmpbuf2 + 2, s, tmplen2);
1642 tmplen2 += 2;
1643
8a5ee598 1644 gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE);
e5527e4b 1645 if (!isGV(gv2)) {
8a5ee598 1646 gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE);
e5527e4b
RGS
1647 /* adjust ${"::_<newfilename"} to store the new file name */
1648 GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2);
8818d409
FC
1649 /* The line number may differ. If that is the case,
1650 alias the saved lines that are in the array.
1651 Otherwise alias the whole array. */
1652 if (CopLINE(PL_curcop) == line_num) {
1653 GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp)));
1654 GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp)));
1655 }
1656 else if (GvAV(*gvp)) {
1657 AV * const av = GvAV(*gvp);
1658 const I32 start = CopLINE(PL_curcop)+1;
1659 I32 items = AvFILLp(av) - start;
1660 if (items > 0) {
1661 AV * const av2 = GvAVn(gv2);
1662 SV **svp = AvARRAY(av) + start;
1663 I32 l = (I32)line_num+1;
1664 while (items--)
1665 av_store(av2, l++, SvREFCNT_inc(*svp++));
1666 }
1667 }
e5527e4b 1668 }
44867030
NC
1669
1670 if (tmpbuf2 != smallbuf) Safefree(tmpbuf2);
8a5ee598 1671 }
e66cf94c 1672 if (tmpbuf != smallbuf) Safefree(tmpbuf);
e66cf94c 1673 }
05ec9bb3 1674 CopFILE_free(PL_curcop);
d9095cec 1675 CopFILE_setn(PL_curcop, s, len);
f4dd75d9 1676 }
8818d409 1677 CopLINE_set(PL_curcop, line_num);
463ee0b2
LW
1678}
1679
29595ff2 1680#ifdef PERL_MAD
cd81e915 1681/* skip space before PL_thistoken */
29595ff2
NC
1682
1683STATIC char *
1684S_skipspace0(pTHX_ register char *s)
1685{
7918f24d
NC
1686 PERL_ARGS_ASSERT_SKIPSPACE0;
1687
29595ff2
NC
1688 s = skipspace(s);
1689 if (!PL_madskills)
1690 return s;
cd81e915
NC
1691 if (PL_skipwhite) {
1692 if (!PL_thiswhite)
6b29d1f5 1693 PL_thiswhite = newSVpvs("");
cd81e915
NC
1694 sv_catsv(PL_thiswhite, PL_skipwhite);
1695 sv_free(PL_skipwhite);
1696 PL_skipwhite = 0;
1697 }
1698 PL_realtokenstart = s - SvPVX(PL_linestr);
29595ff2
NC
1699 return s;
1700}
1701
cd81e915 1702/* skip space after PL_thistoken */
29595ff2
NC
1703
1704STATIC char *
1705S_skipspace1(pTHX_ register char *s)
1706{
d4c19fe8 1707 const char *start = s;
29595ff2
NC
1708 I32 startoff = start - SvPVX(PL_linestr);
1709
7918f24d
NC
1710 PERL_ARGS_ASSERT_SKIPSPACE1;
1711
29595ff2
NC
1712 s = skipspace(s);
1713 if (!PL_madskills)
1714 return s;
1715 start = SvPVX(PL_linestr) + startoff;
cd81e915 1716 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1717 const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1718 PL_thistoken = newSVpvn(tstart, start - tstart);
1719 }
1720 PL_realtokenstart = -1;
1721 if (PL_skipwhite) {
1722 if (!PL_nextwhite)
6b29d1f5 1723 PL_nextwhite = newSVpvs("");
cd81e915
NC
1724 sv_catsv(PL_nextwhite, PL_skipwhite);
1725 sv_free(PL_skipwhite);
1726 PL_skipwhite = 0;
29595ff2
NC
1727 }
1728 return s;
1729}
1730
1731STATIC char *
1732S_skipspace2(pTHX_ register char *s, SV **svp)
1733{
c35e046a
AL
1734 char *start;
1735 const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr);
1736 const I32 startoff = s - SvPVX(PL_linestr);
1737
7918f24d
NC
1738 PERL_ARGS_ASSERT_SKIPSPACE2;
1739
29595ff2
NC
1740 s = skipspace(s);
1741 PL_bufptr = SvPVX(PL_linestr) + bufptroff;
1742 if (!PL_madskills || !svp)
1743 return s;
1744 start = SvPVX(PL_linestr) + startoff;
cd81e915 1745 if (!PL_thistoken && PL_realtokenstart >= 0) {
d4c19fe8 1746 char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart;
cd81e915
NC
1747 PL_thistoken = newSVpvn(tstart, start - tstart);
1748 PL_realtokenstart = -1;
29595ff2 1749 }
cd81e915 1750 if (PL_skipwhite) {
29595ff2 1751 if (!*svp)
6b29d1f5 1752 *svp = newSVpvs("");
cd81e915
NC
1753 sv_setsv(*svp, PL_skipwhite);
1754 sv_free(PL_skipwhite);
1755 PL_skipwhite = 0;
29595ff2
NC
1756 }
1757
1758 return s;
1759}
1760#endif
1761
80a702cd 1762STATIC void
15f169a1 1763S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len)
80a702cd
RGS
1764{
1765 AV *av = CopFILEAVx(PL_curcop);
1766 if (av) {
b9f83d2f 1767 SV * const sv = newSV_type(SVt_PVMG);
5fa550fb
NC
1768 if (orig_sv)
1769 sv_setsv(sv, orig_sv);
1770 else
1771 sv_setpvn(sv, buf, len);
80a702cd
RGS
1772 (void)SvIOK_on(sv);
1773 SvIV_set(sv, 0);
1774 av_store(av, (I32)CopLINE(PL_curcop), sv);
1775 }
1776}
1777
ffb4593c
NT
1778/*
1779 * S_skipspace
1780 * Called to gobble the appropriate amount and type of whitespace.
1781 * Skips comments as well.
1782 */
1783
76e3520e 1784STATIC char *
cea2e8a9 1785S_skipspace(pTHX_ register char *s)
a687059c 1786{
5db06880 1787#ifdef PERL_MAD
f0e67a1d
Z
1788 char *start = s;
1789#endif /* PERL_MAD */
7918f24d 1790 PERL_ARGS_ASSERT_SKIPSPACE;
f0e67a1d 1791#ifdef PERL_MAD
cd81e915
NC
1792 if (PL_skipwhite) {
1793 sv_free(PL_skipwhite);
f0e67a1d 1794 PL_skipwhite = NULL;
5db06880 1795 }
f0e67a1d 1796#endif /* PERL_MAD */
3280af22 1797 if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) {
bf4acbe4 1798 while (s < PL_bufend && SPACE_OR_TAB(*s))
463ee0b2 1799 s++;
f0e67a1d
Z
1800 } else {
1801 STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr);
1802 PL_bufptr = s;
f0998909
Z
1803 lex_read_space(LEX_KEEP_PREVIOUS |
1804 (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ?
1805 LEX_NO_NEXT_CHUNK : 0));
3280af22 1806 s = PL_bufptr;
f0e67a1d
Z
1807 PL_bufptr = SvPVX(PL_linestr) + bufptr_pos;
1808 if (PL_linestart > PL_bufptr)
1809 PL_bufptr = PL_linestart;
1810 return s;
463ee0b2 1811 }
5db06880 1812#ifdef PERL_MAD
f0e67a1d
Z
1813 if (PL_madskills)
1814 PL_skipwhite = newSVpvn(start, s-start);
1815#endif /* PERL_MAD */
5db06880 1816 return s;
a687059c 1817}
378cc40b 1818
ffb4593c
NT
1819/*
1820 * S_check_uni
1821 * Check the unary operators to ensure there's no ambiguity in how they're
1822 * used. An ambiguous piece of code would be:
1823 * rand + 5
1824 * This doesn't mean rand() + 5. Because rand() is a unary operator,
1825 * the +5 is its argument.
1826 */
1827
76e3520e 1828STATIC void
cea2e8a9 1829S_check_uni(pTHX)
ba106d47 1830{
97aff369 1831 dVAR;
d4c19fe8
AL
1832 const char *s;
1833 const char *t;
2f3197b3 1834
3280af22 1835 if (PL_oldoldbufptr != PL_last_uni)
2f3197b3 1836 return;
3280af22
NIS
1837 while (isSPACE(*PL_last_uni))
1838 PL_last_uni++;
c35e046a
AL
1839 s = PL_last_uni;
1840 while (isALNUM_lazy_if(s,UTF) || *s == '-')
1841 s++;
3280af22 1842 if ((t = strchr(s, '(')) && t < PL_bufptr)
a0d0e21e 1843 return;
6136c704 1844
9b387841
NC
1845 Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS),
1846 "Warning: Use of \"%.*s\" without parentheses is ambiguous",
1847 (int)(s - PL_last_uni), PL_last_uni);
2f3197b3
LW
1848}
1849
ffb4593c
NT
1850/*
1851 * LOP : macro to build a list operator. Its behaviour has been replaced
1852 * with a subroutine, S_lop() for which LOP is just another name.
1853 */
1854
a0d0e21e
LW
1855#define LOP(f,x) return lop(f,x,s)
1856
ffb4593c
NT
1857/*
1858 * S_lop
1859 * Build a list operator (or something that might be one). The rules:
1860 * - if we have a next token, then it's a list operator [why?]
1861 * - if the next thing is an opening paren, then it's a function
1862 * - else it's a list operator
1863 */
1864
76e3520e 1865STATIC I32
a0be28da 1866S_lop(pTHX_ I32 f, int x, char *s)
ffed7fef 1867{
97aff369 1868 dVAR;
7918f24d
NC
1869
1870 PERL_ARGS_ASSERT_LOP;
1871
6154021b 1872 pl_yylval.ival = f;
35c8bce7 1873 CLINE;
3280af22
NIS
1874 PL_expect = x;
1875 PL_bufptr = s;
1876 PL_last_lop = PL_oldbufptr;
eb160463 1877 PL_last_lop_op = (OPCODE)f;
5db06880
NC
1878#ifdef PERL_MAD
1879 if (PL_lasttoke)
78cdf107 1880 goto lstop;
5db06880 1881#else
3280af22 1882 if (PL_nexttoke)
78cdf107 1883 goto lstop;
5db06880 1884#endif
79072805 1885 if (*s == '(')
bbf60fe6 1886 return REPORT(FUNC);
29595ff2 1887 s = PEEKSPACE(s);
79072805 1888 if (*s == '(')
bbf60fe6 1889 return REPORT(FUNC);
78cdf107
Z
1890 else {
1891 lstop:
1892 if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC)
1893 PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC;
bbf60fe6 1894 return REPORT(LSTOP);
78cdf107 1895 }
79072805
LW
1896}
1897
5db06880
NC
1898#ifdef PERL_MAD
1899 /*
1900 * S_start_force
1901 * Sets up for an eventual force_next(). start_force(0) basically does
1902 * an unshift, while start_force(-1) does a push. yylex removes items
1903 * on the "pop" end.
1904 */
1905
1906STATIC void
1907S_start_force(pTHX_ int where)
1908{
1909 int i;
1910
cd81e915 1911 if (where < 0) /* so people can duplicate start_force(PL_curforce) */
5db06880 1912 where = PL_lasttoke;
cd81e915
NC
1913 assert(PL_curforce < 0 || PL_curforce == where);
1914 if (PL_curforce != where) {
5db06880
NC
1915 for (i = PL_lasttoke; i > where; --i) {
1916 PL_nexttoke[i] = PL_nexttoke[i-1];
1917 }
1918 PL_lasttoke++;
1919 }
cd81e915 1920 if (PL_curforce < 0) /* in case of duplicate start_force() */
5db06880 1921 Zero(&PL_nexttoke[where], 1, NEXTTOKE);
cd81e915
NC
1922 PL_curforce = where;
1923 if (PL_nextwhite) {
5db06880 1924 if (PL_madskills)
6b29d1f5 1925 curmad('^', newSVpvs(""));
cd81e915 1926 CURMAD('_', PL_nextwhite);
5db06880
NC
1927 }
1928}
1929
1930STATIC void
1931S_curmad(pTHX_ char slot, SV *sv)
1932{
1933 MADPROP **where;
1934
1935 if (!sv)
1936 return;
cd81e915
NC
1937 if (PL_curforce < 0)
1938 where = &PL_thismad;
5db06880 1939 else
cd81e915 1940 where = &PL_nexttoke[PL_curforce].next_mad;
5db06880 1941
cd81e915 1942 if (PL_faketokens)
76f68e9b 1943 sv_setpvs(sv, "");
5db06880
NC
1944 else {
1945 if (!IN_BYTES) {
1946 if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv)))
1947 SvUTF8_on(sv);
1948 else if (PL_encoding) {
1949 sv_recode_to_utf8(sv, PL_encoding);
1950 }
1951 }
1952 }
1953
1954 /* keep a slot open for the head of the list? */
1955 if (slot != '_' && *where && (*where)->mad_key == '^') {
1956 (*where)->mad_key = slot;
daba3364 1957 sv_free(MUTABLE_SV(((*where)->mad_val)));
5db06880
NC
1958 (*where)->mad_val = (void*)sv;
1959 }
1960 else
1961 addmad(newMADsv(slot, sv), where, 0);
1962}
1963#else
b3f24c00
MHM
1964# define start_force(where) NOOP
1965# define curmad(slot, sv) NOOP
5db06880
NC
1966#endif
1967
ffb4593c
NT
1968/*
1969 * S_force_next
9cbb5ea2 1970 * When the lexer realizes it knows the next token (for instance,
ffb4593c 1971 * it is reordering tokens for the parser) then it can call S_force_next
9cbb5ea2 1972 * to know what token to return the next time the lexer is called. Caller
5db06880
NC
1973 * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD),
1974 * and possibly PL_expect to ensure the lexer handles the token correctly.
ffb4593c
NT
1975 */
1976
4e553d73 1977STATIC void
cea2e8a9 1978S_force_next(pTHX_ I32 type)
79072805 1979{
97aff369 1980 dVAR;
704d4215
GG
1981#ifdef DEBUGGING
1982 if (DEBUG_T_TEST) {
1983 PerlIO_printf(Perl_debug_log, "### forced token:\n");
f05d7009 1984 tokereport(type, &NEXTVAL_NEXTTOKE);
704d4215
GG
1985 }
1986#endif
5db06880 1987#ifdef PERL_MAD
cd81e915 1988 if (PL_curforce < 0)
5db06880 1989 start_force(PL_lasttoke);
cd81e915 1990 PL_nexttoke[PL_curforce].next_type = type;
5db06880
NC
1991 if (PL_lex_state != LEX_KNOWNEXT)
1992 PL_lex_defer = PL_lex_state;
1993 PL_lex_state = LEX_KNOWNEXT;
1994 PL_lex_expect = PL_expect;
cd81e915 1995 PL_curforce = -1;
5db06880 1996#else
3280af22
NIS
1997 PL_nexttype[PL_nexttoke] = type;
1998 PL_nexttoke++;
1999 if (PL_lex_state != LEX_KNOWNEXT) {
2000 PL_lex_defer = PL_lex_state;
2001 PL_lex_expect = PL_expect;
2002 PL_lex_state = LEX_KNOWNEXT;
79072805 2003 }
5db06880 2004#endif
79072805
LW
2005}
2006
28ac2b49
Z
2007void
2008Perl_yyunlex(pTHX)
2009{
a7aaec61
Z
2010 int yyc = PL_parser->yychar;
2011 if (yyc != YYEMPTY) {
2012 if (yyc) {
2013 start_force(-1);
2014 NEXTVAL_NEXTTOKE = PL_parser->yylval;
2015 if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) {
78cdf107 2016 PL_lex_allbrackets--;
a7aaec61 2017 PL_lex_brackets--;
78cdf107
Z
2018 yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16);
2019 } else if (yyc == '('/*)*/) {
2020 PL_lex_allbrackets--;
2021 yyc |= (2<<24);
a7aaec61
Z
2022 }
2023 force_next(yyc);
2024 }
28ac2b49
Z
2025 PL_parser->yychar = YYEMPTY;
2026 }
2027}
2028
d0a148a6 2029STATIC SV *
15f169a1 2030S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len)
d0a148a6 2031{
97aff369 2032 dVAR;
740cce10 2033 SV * const sv = newSVpvn_utf8(start, len,
eaf7a4d2
CS
2034 !IN_BYTES
2035 && UTF
2036 && !is_ascii_string((const U8*)start, len)
740cce10 2037 && is_utf8_string((const U8*)start, len));
d0a148a6
NC
2038 return sv;
2039}
2040
ffb4593c
NT
2041/*
2042 * S_force_word
2043 * When the lexer knows the next thing is a word (for instance, it has
2044 * just seen -> and it knows that the next char is a word char, then
02b34bbe
DM
2045 * it calls S_force_word to stick the next word into the PL_nexttoke/val
2046 * lookahead.
ffb4593c
NT
2047 *
2048 * Arguments:
b1b65b59 2049 * char *start : buffer position (must be within PL_linestr)
02b34bbe 2050 * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD)
ffb4593c
NT
2051 * int check_keyword : if true, Perl checks to make sure the word isn't
2052 * a keyword (do this if the word is a label, e.g. goto FOO)
2053 * int allow_pack : if true, : characters will also be allowed (require,
2054 * use, etc. do this)
9cbb5ea2 2055 * int allow_initial_tick : used by the "sub" lexer only.
ffb4593c
NT
2056 */
2057
76e3520e 2058STATIC char *
cea2e8a9 2059S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick)
79072805 2060{
97aff369 2061 dVAR;
463ee0b2
LW
2062 register char *s;
2063 STRLEN len;
4e553d73 2064
7918f24d
NC
2065 PERL_ARGS_ASSERT_FORCE_WORD;
2066
29595ff2 2067 start = SKIPSPACE1(start);
463ee0b2 2068 s = start;
7e2040f0 2069 if (isIDFIRST_lazy_if(s,UTF) ||
a0d0e21e 2070 (allow_pack && *s == ':') ||
15f0808c 2071 (allow_initial_tick && *s == '\'') )
a0d0e21e 2072 {
3280af22 2073 s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len);
5458a98a 2074 if (check_keyword && keyword(PL_tokenbuf, len, 0))
463ee0b2 2075 return start;
cd81e915 2076 start_force(PL_curforce);
5db06880
NC
2077 if (PL_madskills)
2078 curmad('X', newSVpvn(start,s-start));
463ee0b2 2079 if (token == METHOD) {
29595ff2 2080 s = SKIPSPACE1(s);
463ee0b2 2081 if (*s == '(')
3280af22 2082 PL_expect = XTERM;
463ee0b2 2083 else {
3280af22 2084 PL_expect = XOPERATOR;
463ee0b2 2085 }
79072805 2086 }
e74e6b3d 2087 if (PL_madskills)
63575281 2088 curmad('g', newSVpvs( "forced" ));
9ded7720 2089 NEXTVAL_NEXTTOKE.opval
d0a148a6
NC
2090 = (OP*)newSVOP(OP_CONST,0,
2091 S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len));
9ded7720 2092 NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE;
79072805
LW
2093 force_next(token);
2094 }
2095 return s;
2096}
2097
ffb4593c
NT
2098/*
2099 * S_force_ident
9cbb5ea2 2100 * Called when the lexer wants $foo *foo &foo etc, but the program
ffb4593c
NT
2101 * text only contains the "foo" portion. The first argument is a pointer
2102 * to the "foo", and the second argument is the type symbol to prefix.
2103 * Forces the next token to be a "WORD".
9cbb5ea2 2104 * Creates the symbol if it didn't already exist (via gv_fetchpv()).
ffb4593c
NT
2105 */
2106
76e3520e 2107STATIC void
bfed75c6 2108S_force_ident(pTHX_ register const char *s, int kind)
79072805 2109{
97aff369 2110 dVAR;
7918f24d
NC
2111
2112 PERL_ARGS_ASSERT_FORCE_IDENT;
2113
c35e046a 2114 if (*s) {
90e5519e 2115 const STRLEN len = strlen(s);
728847b1
BF
2116 OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len,
2117 UTF ? SVf_UTF8 : 0));
cd81e915 2118 start_force(PL_curforce);
9ded7720 2119 NEXTVAL_NEXTTOKE.opval = o;
79072805 2120 force_next(WORD);
748a9306 2121 if (kind) {
11343788 2122 o->op_private = OPpCONST_ENTERED;
55497cff
PP
2123 /* XXX see note in pp_entereval() for why we forgo typo
2124 warnings if the symbol must be introduced in an eval.
2125 GSAR 96-10-12 */
90e5519e 2126 gv_fetchpvn_flags(s, len,
728847b1
BF
2127 (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL)
2128 : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ),
90e5519e
NC
2129 kind == '$' ? SVt_PV :
2130 kind == '@' ? SVt_PVAV :
2131 kind == '%' ? SVt_PVHV :
a0d0e21e 2132 SVt_PVGV
90e5519e 2133 );
748a9306 2134 }
79072805
LW
2135 }
2136}
2137
1571675a
GS
2138NV
2139Perl_str_to_version(pTHX_ SV *sv)
2140{
2141 NV retval = 0.0;
2142 NV nshift = 1.0;
2143 STRLEN len;
cfd0369c 2144 const char *start = SvPV_const(sv,len);
9d4ba2ae 2145 const char * const end = start + len;
504618e9 2146 const bool utf = SvUTF8(sv) ? TRUE : FALSE;
7918f24d
NC
2147
2148 PERL_ARGS_ASSERT_STR_TO_VERSION;
2149
1571675a 2150 while (start < end) {
ba210ebe 2151 STRLEN skip;
1571675a
GS
2152 UV n;
2153 if (utf)
9041c2e3 2154 n = utf8n_to_uvchr((U8*)start, len, &skip, 0);
1571675a
GS
2155 else {
2156 n = *(U8*)start;
2157 skip = 1;
2158 }
2159 retval += ((NV)n)/nshift;
2160 start += skip;
2161 nshift *= 1000;
2162 }
2163 return retval;
2164}
2165
4e553d73 2166/*
ffb4593c
NT
2167 * S_force_version
2168 * Forces the next token to be a version number.
e759cc13
RGS
2169 * If the next token appears to be an invalid version number, (e.g. "v2b"),
2170 * and if "guessing" is TRUE, then no new token is created (and the caller
2171 * must use an alternative parsing method).
ffb4593c
NT
2172 */
2173
76e3520e 2174STATIC char *
e759cc13 2175S_force_version(pTHX_ char *s, int guessing)
89bfa8cd 2176{
97aff369 2177 dVAR;
5f66b61c 2178 OP *version = NULL;
44dcb63b 2179 char *d;
5db06880
NC
2180#ifdef PERL_MAD
2181 I32 startoff = s - SvPVX(PL_linestr);
2182#endif
89bfa8cd 2183
7918f24d
NC
2184 PERL_ARGS_ASSERT_FORCE_VERSION;
2185
29595ff2 2186 s = SKIPSPACE1(s);
89bfa8cd 2187
44dcb63b 2188 d = s;
dd629d5b 2189 if (*d == 'v')
44dcb63b 2190 d++;
44dcb63b 2191 if (isDIGIT(*d)) {
e759cc13
RGS
2192 while (isDIGIT(*d) || *d == '_' || *d == '.')
2193 d++;
5db06880
NC
2194#ifdef PERL_MAD
2195 if (PL_madskills) {
cd81e915 2196 start_force(PL_curforce);
5db06880
NC
2197 curmad('X', newSVpvn(s,d-s));
2198 }
2199#endif
4e4da3ac 2200 if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) {
dd629d5b 2201 SV *ver;
8d08d9ba 2202#ifdef USE_LOCALE_NUMERIC
909d3787
KW
2203 char *loc = savepv(setlocale(LC_NUMERIC, NULL));
2204 setlocale(LC_NUMERIC, "C");
8d08d9ba 2205#endif
6154021b 2206 s = scan_num(s, &pl_yylval);
8d08d9ba
DG
2207#ifdef USE_LOCALE_NUMERIC
2208 setlocale(LC_NUMERIC, loc);
909d3787 2209 Safefree(loc);
8d08d9ba 2210#endif
6154021b 2211 version = pl_yylval.opval;
dd629d5b
GS
2212 ver = cSVOPx(version)->op_sv;
2213 if (SvPOK(ver) && !SvNIOK(ver)) {
862a34c6 2214 SvUPGRADE(ver, SVt_PVNV);
9d6ce603 2215 SvNV_set(ver, str_to_version(ver));
1571675a 2216 SvNOK_on(ver); /* hint that it is a version */
44dcb63b 2217 }
89bfa8cd 2218 }
5db06880
NC
2219 else if (guessing) {
2220#ifdef PERL_MAD
2221 if (PL_madskills) {
cd81e915
NC
2222 sv_free(PL_nextwhite); /* let next token collect whitespace */
2223 PL_nextwhite = 0;
5db06880
NC
2224 s = SvPVX(PL_linestr) + startoff;
2225 }
2226#endif
e759cc13 2227 return s;
5db06880 2228 }
89bfa8cd
PP
2229 }
2230
5db06880
NC
2231#ifdef PERL_MAD
2232 if (PL_madskills && !version) {
cd81e915
NC
2233 sv_free(PL_nextwhite); /* let next token collect whitespace */
2234 PL_nextwhite = 0;
5db06880
NC
2235 s = SvPVX(PL_linestr) + startoff;
2236 }
2237#endif
89bfa8cd 2238 /* NOTE: The parser sees the package name and the VERSION swapped */
cd81e915 2239 start_force(PL_curforce);
9ded7720 2240 NEXTVAL_NEXTTOKE.opval = version;
4e553d73 2241 force_next(WORD);
89bfa8cd 2242
e759cc13 2243 return s;
89bfa8cd
PP
2244}
2245
ffb4593c 2246/*
91152fc1
DG
2247 * S_force_strict_version
2248 * Forces the next token to be a version number using strict syntax rules.
2249 */
2250
2251STATIC char *
2252S_force_strict_version(pTHX_ char *s)
2253{
2254 dVAR;
2255 OP *version = NULL;
2256#ifdef PERL_MAD
2257 I32 startoff = s - SvPVX(PL_linestr);
2258#endif
2259 const char *errstr = NULL;
2260
2261 PERL_ARGS_ASSERT_FORCE_STRICT_VERSION;
2262
2263 while (isSPACE(*s)) /* leading whitespace */
2264 s++;
2265
2266 if (is_STRICT_VERSION(s,&errstr)) {
2267 SV *ver = newSV(0);
2268 s = (char *)scan_version(s, ver, 0);
2269 version = newSVOP(OP_CONST, 0, ver);
2270 }
4e4da3ac
Z
2271 else if ( (*s != ';' && *s != '{' && *s != '}' ) &&
2272 (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' )))
2273 {
91152fc1
DG
2274 PL_bufptr = s;
2275 if (errstr)
2276 yyerror(errstr); /* version required */
2277 return s;
2278 }
2279
2280#ifdef PERL_MAD
2281 if (PL_madskills && !version) {
2282 sv_free(PL_nextwhite); /* let next token collect whitespace */
2283 PL_nextwhite = 0;
2284 s = SvPVX(PL_linestr) + startoff;
2285 }
2286#endif
2287 /* NOTE: The parser sees the package name and the VERSION swapped */
2288 start_force(PL_curforce);
2289 NEXTVAL_NEXTTOKE.opval = version;
2290 force_next(WORD);
2291
2292 return s;
2293}
2294
2295/*
ffb4593c
NT
2296 * S_tokeq
2297 * Tokenize a quoted string passed in as an SV. It finds the next
2298 * chunk, up to end of string or a backslash. It may make a new
2299 * SV containing that chunk (if HINT_NEW_STRING is on). It also
2300 * turns \\ into \.
2301 */
2302
76e3520e 2303STATIC SV *
cea2e8a9 2304S_tokeq(pTHX_ SV *sv)
79072805 2305{
97aff369 2306 dVAR;
79072805
LW
2307 register char *s;
2308 register char *send;
2309 register char *d;
b3ac6de7
IZ
2310 STRLEN len = 0;
2311 SV *pv = sv;
79072805 2312
7918f24d
NC
2313 PERL_ARGS_ASSERT_TOKEQ;
2314
79072805 2315 if (!SvLEN(sv))
b3ac6de7 2316 goto finish;
79072805 2317
a0d0e21e 2318 s = SvPV_force(sv, len);
21a311ee 2319 if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1)
b3ac6de7 2320 goto finish;
463ee0b2 2321 send = s + len;
dcb21ed6
NC
2322 /* This is relying on the SV being "well formed" with a trailing '\0' */
2323 while (s < send && !(*s == '\\' && s[1] == '\\'))
79072805
LW
2324 s++;
2325 if (s == send)
b3ac6de7 2326 goto finish;
79072805 2327 d = s;
be4731d2 2328 if ( PL_hints & HINT_NEW_STRING ) {
59cd0e26 2329 pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv));
be4731d2 2330 }
79072805
LW
2331 while (s < send) {
2332 if (*s == '\\') {
a0d0e21e 2333 if (s + 1 < send && (s[1] == '\\'))
79072805
LW
2334 s++; /* all that, just for this */
2335 }
2336 *d++ = *s++;
2337 }
2338 *d = '\0';
95a20fc0 2339 SvCUR_set(sv, d - SvPVX_const(sv));
b3ac6de7 2340 finish:
3280af22 2341 if ( PL_hints & HINT_NEW_STRING )
eb0d8d16 2342 return new_constant(NULL, 0, "q", sv, pv, "q", 1);
79072805
LW
2343 return sv;
2344}
2345
ffb4593c
NT
2346/*
2347 * Now come three functions related to double-quote context,
2348 * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when
2349 * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They
2350 * interact with PL_lex_state, and create fake ( ... ) argument lists
2351 * to handle functions and concatenation.
ecd24171
DM
2352 * For example,
2353 * "foo\lbar"
2354 * is tokenised as
2355 * stringify ( const[foo] concat lcfirst ( const[bar] ) )
ffb4593c
NT
2356 */
2357
2358/*
2359 * S_sublex_start
6154021b 2360 * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST).
ffb4593c
NT
2361 *
2362 * Pattern matching will set PL_lex_op to the pattern-matching op to
6154021b 2363 * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise).
ffb4593c
NT
2364 *
2365 * OP_CONST and OP_READLINE are easy--just make the new op and return.
2366 *
2367 * Everything else becomes a FUNC.
2368 *
2369 * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we
2370 * had an OP_CONST or OP_READLINE). This just sets us up for a
2371 * call to S_sublex_push().
2372 */
2373
76e3520e 2374STATIC I32
cea2e8a9 2375S_sublex_start(pTHX)
79072805 2376{
97aff369 2377 dVAR;
6154021b 2378 register const I32 op_type = pl_yylval.ival;
79072805
LW
2379
2380 if (op_type == OP_NULL) {
6154021b 2381 pl_yylval.opval = PL_lex_op;
5f66b61c 2382 PL_lex_op = NULL;
79072805
LW
2383 return THING;
2384 }
2385 if (op_type == OP_CONST || op_type == OP_READLINE) {
3280af22 2386 SV *sv = tokeq(PL_lex_stuff);
b3ac6de7
IZ
2387
2388 if (SvTYPE(sv) == SVt_PVIV) {
2389 /* Overloaded constants, nothing fancy: Convert to SVt_PV: */
2390 STRLEN len;
96a5add6 2391 const char * const p = SvPV_const(sv, len);
740cce10 2392 SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv));
b3ac6de7
IZ
2393 SvREFCNT_dec(sv);
2394 sv = nsv;
4e553d73 2395 }
6154021b 2396 pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv);
a0714e2c 2397 PL_lex_stuff = NULL;
6f33ba73
RGS
2398 /* Allow <FH> // "foo" */
2399 if (op_type == OP_READLINE)
2400 PL_expect = XTERMORDORDOR;
79072805
LW
2401 return THING;
2402 }
e3f73d4e
RGS
2403 else if (op_type == OP_BACKTICK && PL_lex_op) {
2404 /* readpipe() vas overriden */
2405 cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff);
6154021b 2406 pl_yylval.opval = PL_lex_op;
9b201d7d 2407 PL_lex_op = NULL;
e3f73d4e
RGS
2408 PL_lex_stuff = NULL;
2409 return THING;
2410 }
79072805 2411
3280af22 2412 PL_sublex_info.super_state = PL_lex_state;
eac04b2e 2413 PL_sublex_info.sub_inwhat = (U16)op_type;
3280af22
NIS
2414 PL_sublex_info.sub_op = PL_lex_op;
2415 PL_lex_state = LEX_INTERPPUSH;
55497cff 2416
3280af22
NIS
2417 PL_expect = XTERM;
2418 if (PL_lex_op) {
6154021b 2419 pl_yylval.opval = PL_lex_op;
5f66b61c 2420 PL_lex_op = NULL;
55497cff
PP
2421 return PMFUNC;
2422 }
2423 else
2424 return FUNC;
2425}
2426
ffb4593c
NT
2427/*
2428 * S_sublex_push
2429 * Create a new scope to save the lexing state. The scope will be
2430 * ended in S_sublex_done. Returns a '(', starting the function arguments
2431 * to the uc, lc, etc. found before.
2432 * Sets PL_lex_state to LEX_INTERPCONCAT.
2433 */
2434
76e3520e 2435STATIC I32
cea2e8a9 2436S_sublex_push(pTHX)
55497cff 2437{
27da23d5 2438 dVAR;
f46d017c 2439 ENTER;
55497cff 2440
3280af22 2441 PL_lex_state = PL_sublex_info.super_state;
651b5b28 2442 SAVEBOOL(PL_lex_dojoin);
3280af22 2443 SAVEI32(PL_lex_brackets);
78cdf107
Z
2444 SAVEI32(PL_lex_allbrackets);
2445 SAVEI8(PL_lex_fakeeof);
3280af22
NIS
2446 SAVEI32(PL_lex_casemods);
2447 SAVEI32(PL_lex_starts);
651b5b28 2448 SAVEI8(PL_lex_state);
7766f137 2449 SAVEVPTR(PL_lex_inpat);
98246f1e 2450 SAVEI16(PL_lex_inwhat);
57843af0 2451 SAVECOPLINE(PL_curcop);
3280af22 2452 SAVEPPTR(PL_bufptr);
8452ff4b 2453 SAVEPPTR(PL_bufend);
3280af22
NIS
2454 SAVEPPTR(PL_oldbufptr);
2455 SAVEPPTR(PL_oldoldbufptr);
207e3d1a
JH
2456 SAVEPPTR(PL_last_lop);
2457 SAVEPPTR(PL_last_uni);
3280af22
NIS
2458 SAVEPPTR(PL_linestart);
2459 SAVESPTR(PL_linestr);
8edd5f42
RGS
2460 SAVEGENERICPV(PL_lex_brackstack);
2461 SAVEGENERICPV(PL_lex_casestack);
3280af22
NIS
2462
2463 PL_linestr = PL_lex_stuff;
a0714e2c 2464 PL_lex_stuff = NULL;
3280af22 2465
9cbb5ea2
GS
2466 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart
2467 = SvPVX(PL_linestr);
3280af22 2468 PL_bufend += SvCUR(PL_linestr);
bd61b366 2469 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2470 SAVEFREESV(PL_linestr);
2471
2472 PL_lex_dojoin = FALSE;
2473 PL_lex_brackets = 0;
78cdf107
Z
2474 PL_lex_allbrackets = 0;
2475 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
a02a5408
JC
2476 Newx(PL_lex_brackstack, 120, char);
2477 Newx(PL_lex_casestack, 12, char);
3280af22
NIS
2478 PL_lex_casemods = 0;
2479 *PL_lex_casestack = '\0';
2480 PL_lex_starts = 0;
2481 PL_lex_state = LEX_INTERPCONCAT;
eb160463 2482 CopLINE_set(PL_curcop, (line_t)PL_multi_start);
3280af22
NIS
2483
2484 PL_lex_inwhat = PL_sublex_info.sub_inwhat;
bb16bae8 2485 if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS;
3280af22
NIS
2486 if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST)
2487 PL_lex_inpat = PL_sublex_info.sub_op;
79072805 2488 else
5f66b61c 2489 PL_lex_inpat = NULL;
79072805 2490
55497cff 2491 return '(';
79072805
LW
2492}
2493
ffb4593c
NT
2494/*
2495 * S_sublex_done
2496 * Restores lexer state after a S_sublex_push.
2497 */
2498
76e3520e 2499STATIC I32
cea2e8a9 2500S_sublex_done(pTHX)
79072805 2501{
27da23d5 2502 dVAR;
3280af22 2503 if (!PL_lex_starts++) {
396482e1 2504 SV * const sv = newSVpvs("");
9aa983d2
JH
2505 if (SvUTF8(PL_linestr))
2506 SvUTF8_on(sv);
3280af22 2507 PL_expect = XOPERATOR;
6154021b 2508 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
79072805
LW
2509 return THING;
2510 }
2511
3280af22
NIS
2512 if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */
2513 PL_lex_state = LEX_INTERPCASEMOD;
cea2e8a9 2514 return yylex();
79072805
LW
2515 }
2516
ffb4593c 2517 /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */
bb16bae8 2518 assert(PL_lex_inwhat != OP_TRANSR);
3280af22
NIS
2519 if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) {
2520 PL_linestr = PL_lex_repl;
2521 PL_lex_inpat = 0;
2522 PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr);
2523 PL_bufend += SvCUR(PL_linestr);
bd61b366 2524 PL_last_lop = PL_last_uni = NULL;
3280af22
NIS
2525 SAVEFREESV(PL_linestr);
2526 PL_lex_dojoin = FALSE;
2527 PL_lex_brackets = 0;
78cdf107
Z
2528 PL_lex_allbrackets = 0;
2529 PL_lex_fakeeof = LEX_FAKEEOF_NEVER;
3280af22
NIS
2530 PL_lex_casemods = 0;
2531 *PL_lex_casestack = '\0';
2532 PL_lex_starts = 0;
25da4f38 2533 if (SvEVALED(PL_lex_repl)) {
3280af22
NIS
2534 PL_lex_state = LEX_INTERPNORMAL;
2535 PL_lex_starts++;
e9fa98b2
HS
2536 /* we don't clear PL_lex_repl here, so that we can check later
2537 whether this is an evalled subst; that means we rely on the
2538 logic to ensure sublex_done() is called again only via the
2539 branch (in yylex()) that clears PL_lex_repl, else we'll loop */
79072805 2540 }
e9fa98b2 2541 else {
3280af22 2542 PL_lex_state = LEX_INTERPCONCAT;
a0714e2c 2543 PL_lex_repl = NULL;
e9fa98b2 2544 }
79072805 2545 return ',';
ffed7fef
LW
2546 }
2547 else {
5db06880
NC
2548#ifdef PERL_MAD
2549 if (PL_madskills) {
cd81e915
NC
2550 if (PL_thiswhite) {
2551 if (!PL_endwhite)
6b29d1f5 2552 PL_endwhite = newSVpvs("");
cd81e915
NC
2553 sv_catsv(PL_endwhite, PL_thiswhite);
2554 PL_thiswhite = 0;
2555 }
2556 if (PL_thistoken)
76f68e9b 2557 sv_setpvs(PL_thistoken,"");
5db06880 2558 else
cd81e915 2559 PL_realtokenstart = -1;
5db06880
NC
2560 }
2561#endif
f46d017c 2562 LEAVE;
3280af22
NIS
2563 PL_bufend = SvPVX(PL_linestr);
2564 PL_bufend += SvCUR(PL_linestr);
2565 PL_expect = XOPERATOR;
09bef843 2566 PL_sublex_info.sub_inwhat = 0;
79072805 2567 return ')';
ffed7fef
LW
2568 }
2569}
2570
02aa26ce
NT
2571/*
2572 scan_const
2573
2574 Extracts a pattern, double-quoted string, or transliteration. This
2575 is terrifying code.
2576
94def140 2577 It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's
3280af22 2578 processing a pattern (PL_lex_inpat is true), a transliteration
94def140 2579 (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string.
02aa26ce 2580
94def140
ST
2581 Returns a pointer to the character scanned up to. If this is
2582 advanced from the start pointer supplied (i.e. if anything was
9b599b2a 2583 successfully parsed), will leave an OP for the substring scanned
6154021b 2584 in pl_yylval. Caller must intuit reason for not parsing further
9b599b2a
GS
2585 by looking at the next characters herself.
2586
02aa26ce
NT
2587 In patterns:
2588 backslashes:
ff3f963a 2589 constants: \N{NAME} only
02aa26ce
NT
2590 case and quoting: \U \Q \E
2591 stops on @ and $, but not for $ as tail anchor
2592
2593 In transliterations:
2594 characters are VERY literal, except for - not at the start or end
94def140
ST
2595 of the string, which indicates a range. If the range is in bytes,
2596 scan_const expands the range to the full set of intermediate
2597 characters. If the range is in utf8, the hyphen is replaced with
2598 a certain range mark which will be handled by pmtrans() in op.c.
02aa26ce
NT
2599
2600 In double-quoted strings:
2601 backslashes:
2602 double-quoted style: \r and \n
ff3f963a 2603 constants: \x31, etc.
94def140 2604 deprecated backrefs: \1 (in substitution replacements)
02aa26ce
NT
2605 case and quoting: \U \Q \E
2606 stops on @ and $
2607
2608 scan_const does *not* construct ops to handle interpolated strings.
2609 It stops processing as soon as it finds an embedded $ or @ variable
2610 and leaves it to the caller to work out what's going on.
2611
94def140
ST
2612 embedded arrays (whether in pattern or not) could be:
2613 @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-.
2614
2615 $ in double-quoted strings must be the symbol of an embedded scalar.
02aa26ce
NT
2616
2617 $ in pattern could be $foo or could be tail anchor. Assumption:
2618 it's a tail anchor if $ is the last thing in the string, or if it's
94def140 2619 followed by one of "()| \r\n\t"
02aa26ce
NT
2620
2621 \1 (backreferences) are turned into $1
2622
2623 The structure of the code is
2624 while (there's a character to process) {
94def140
ST
2625 handle transliteration ranges
2626 skip regexp comments /(?#comment)/ and codes /(?{code})/
2627 skip #-initiated comments in //x patterns
2628 check for embedded arrays
02aa26ce
NT
2629 check for embedded scalars
2630 if (backslash) {
94def140 2631 deprecate \1 in substitution replacements
02aa26ce
NT
2632 handle string-changing backslashes \l \U \Q \E, etc.
2633 switch (what was escaped) {
94def140 2634 handle \- in a transliteration (becomes a literal -)
ff3f963a 2635 if a pattern and not \N{, go treat as regular character
94def140
ST
2636 handle \132 (octal characters)
2637 handle \x15 and \x{1234} (hex characters)
ff3f963a 2638 handle \N{name} (named characters, also \N{3,5} in a pattern)
94def140
ST
2639 handle \cV (control characters)
2640 handle printf-style backslashes (\f, \r, \n, etc)
02aa26ce 2641 } (end switch)
77a135fe 2642 continue
02aa26ce 2643 } (end if backslash)
77a135fe 2644 handle regular character
02aa26ce 2645 } (end while character to read)
4e553d73 2646
02aa26ce
NT
2647*/
2648
76e3520e 2649STATIC char *
cea2e8a9 2650S_scan_const(pTHX_ char *start)
79072805 2651{
97aff369 2652 dVAR;
3280af22 2653 register char *send = PL_bufend; /* end of the constant */
77a135fe
KW
2654 SV *sv = newSV(send - start); /* sv for the constant. See
2655 note below on sizing. */
02aa26ce
NT
2656 register char *s = start; /* start of the constant */
2657 register char *d = SvPVX(sv); /* destination for copies */
2658 bool dorange = FALSE; /* are we in a translit range? */
c2e66d9e 2659 bool didrange = FALSE; /* did we just finish a range? */
b953e60c
KW
2660 bool has_utf8 = FALSE; /* Output constant is UTF8 */
2661 bool this_utf8 = cBOOL(UTF); /* Is the source string assumed
77a135fe
KW
2662 to be UTF8? But, this can
2663 show as true when the source
2664 isn't utf8, as for example
2665 when it is entirely composed
2666 of hex constants */
2667
2668 /* Note on sizing: The scanned constant is placed into sv, which is
2669 * initialized by newSV() assuming one byte of output for every byte of
2670 * input. This routine expects newSV() to allocate an extra byte for a
2671 * trailing NUL, which this routine will append if it gets to the end of
2672 * the input. There may be more bytes of input than output (eg., \N{LATIN
2673 * CAPITAL LETTER A}), or more output than input if the constant ends up
2674 * recoded to utf8, but each time a construct is found that might increase
2675 * the needed size, SvGROW() is called. Its size parameter each time is
2676 * based on the best guess estimate at the time, namely the length used so
2677 * far, plus the length the current construct will occupy, plus room for
2678 * the trailing NUL, plus one byte for every input byte still unscanned */
2679
012bcf8d 2680 UV uv;
4c3a8340
ST
2681#ifdef EBCDIC
2682 UV literal_endpoint = 0;
e294cc5d 2683 bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */
4c3a8340 2684#endif
012bcf8d 2685
7918f24d
NC
2686 PERL_ARGS_ASSERT_SCAN_CONST;
2687
bb16bae8 2688 assert(PL_lex_inwhat != OP_TRANSR);
2b9d42f0
NIS
2689 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
2690 /* If we are doing a trans and we know we want UTF8 set expectation */
2691 has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF);
2692 this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
2693 }
2694
2695
79072805 2696 while (s < send || dorange) {
ff3f963a 2697
02aa26ce 2698 /* get transliterations out of the way (they're most literal) */
3280af22 2699 if (PL_lex_inwhat == OP_TRANS) {
02aa26ce 2700 /* expand a range A-Z to the full set of characters. AIE! */
79072805 2701 if (dorange) {
1ba5c669
JH
2702 I32 i; /* current expanded character */
2703 I32 min; /* first character in range */
2704 I32 max; /* last character in range */
02aa26ce 2705
e294cc5d
JH
2706#ifdef EBCDIC
2707 UV uvmax = 0;
2708#endif
2709
2710 if (has_utf8
2711#ifdef EBCDIC
2712 && !native_range
2713#endif
2714 ) {
9d4ba2ae 2715 char * const c = (char*)utf8_hop((U8*)d, -1);
8973db79
JH
2716 char *e = d++;
2717 while (e-- > c)
2718 *(e + 1) = *e;
25716404 2719 *c = (char)UTF_TO_NATIVE(0xff);
8973db79
JH
2720 /* mark the range as done, and continue */
2721 dorange = FALSE;
2722 didrange = TRUE;
2723 continue;
2724 }
2b9d42f0 2725
95a20fc0 2726 i = d - SvPVX_const(sv); /* remember current offset */
e294cc5d
JH
2727#ifdef EBCDIC
2728 SvGROW(sv,
2729 SvLEN(sv) + (has_utf8 ?
2730 (512 - UTF_CONTINUATION_MARK +
2731 UNISKIP(0x100))
2732 : 256));
2733 /* How many two-byte within 0..255: 128 in UTF-8,
2734 * 96 in UTF-8-mod. */
2735#else
9cbb5ea2 2736 SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */
e294cc5d 2737#endif
9cbb5ea2 2738 d = SvPVX(sv) + i; /* refresh d after realloc */
e294cc5d
JH
2739#ifdef EBCDIC
2740 if (has_utf8) {
2741 int j;
2742 for (j = 0; j <= 1; j++) {
2743 char * const c = (char*)utf8_hop((U8*)d, -1);
2744 const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0);
2745 if (j)
2746 min = (U8)uv;
2747 else if (uv < 256)
2748 max = (U8)uv;
2749 else {
2750 max = (U8)0xff; /* only to \xff */
2751 uvmax = uv; /* \x{100} to uvmax */
2752 }
2753 d = c; /* eat endpoint chars */
2754 }
2755 }
2756 else {
2757#endif
2758 d -= 2; /* eat the first char and the - */
2759 min = (U8)*d; /* first char in range */
2760 max = (U8)d[1]; /* last char in range */
2761#ifdef EBCDIC
2762 }
2763#endif
8ada0baa 2764
c2e66d9e 2765 if (min > max) {
01ec43d0 2766 Perl_croak(aTHX_
d1573ac7 2767 "Invalid range \"%c-%c\" in transliteration operator",
1ba5c669 2768 (char)min, (char)max);
c2e66d9e
GS
2769 }
2770
c7f1f016 2771#ifdef EBCDIC
4c3a8340
ST
2772 if (literal_endpoint == 2 &&
2773 ((isLOWER(min) && isLOWER(max)) ||
2774 (isUPPER(min) && isUPPER(max)))) {
8ada0baa
JH
2775 if (isLOWER(min)) {
2776 for (i = min; i <= max; i++)
2777 if (isLOWER(i))
db42d148 2778 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
2779 } else {
2780 for (i = min; i <= max; i++)
2781 if (isUPPER(i))
db42d148 2782 *d++ = NATIVE_TO_NEED(has_utf8,i);
8ada0baa
JH
2783 }
2784 }
2785 else
2786#endif
2787 for (i = min; i <= max; i++)
e294cc5d
JH
2788#ifdef EBCDIC
2789 if (has_utf8) {
2790 const U8 ch = (U8)NATIVE_TO_UTF(i);
2791 if (UNI_IS_INVARIANT(ch))
2792 *d++ = (U8)i;
2793 else {
2794 *d++ = (U8)UTF8_EIGHT_BIT_HI(ch);
2795 *d++ = (U8)UTF8_EIGHT_BIT_LO(ch);
2796 }
2797 }
2798 else
2799#endif
2800 *d++ = (char)i;
2801
2802#ifdef EBCDIC
2803 if (uvmax) {
2804 d = (char*)uvchr_to_utf8((U8*)d, 0x100);
2805 if (uvmax > 0x101)
2806 *d++ = (char)UTF_TO_NATIVE(0xff);
2807 if (uvmax > 0x100)
2808 d = (char*)uvchr_to_utf8((U8*)d, uvmax);
2809 }
2810#endif
02aa26ce
NT
2811
2812 /* mark the range as done, and continue */
79072805 2813 dorange = FALSE;
01ec43d0 2814 didrange = TRUE;
4c3a8340
ST
2815#ifdef EBCDIC
2816 literal_endpoint = 0;
2817#endif
79072805 2818 continue;
4e553d73 2819 }
02aa26ce
NT
2820
2821 /* range begins (ignore - as first or last char) */
79072805 2822 else if (*s == '-' && s+1 < send && s != start) {
4e553d73 2823 if (didrange) {
1fafa243 2824 Perl_croak(aTHX_ "Ambiguous range in transliteration operator");
01ec43d0 2825 }
e294cc5d
JH
2826 if (has_utf8
2827#ifdef EBCDIC
2828 && !native_range
2829#endif
2830 ) {
25716404 2831 *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */
a0ed51b3
LW
2832 s++;
2833 continue;
2834 }
79072805
LW
2835 dorange = TRUE;
2836 s++;
01ec43d0
GS
2837 }
2838 else {
2839 didrange = FALSE;
4c3a8340
ST
2840#ifdef EBCDIC
2841 literal_endpoint = 0;
e294cc5d 2842 native_range = TRUE;
4c3a8340 2843#endif
01ec43d0 2844 }
79072805 2845 }
02aa26ce
NT
2846
2847 /* if we get here, we're not doing a transliteration */
2848
0f5d15d6
IZ
2849 /* skip for regexp comments /(?#comment)/ and code /(?{code})/,
2850 except for the last char, which will be done separately. */
3280af22 2851 else if (*s == '(' && PL_lex_inpat && s[1] == '?') {
cc6b7395 2852 if (s[2] == '#') {
e994fd66 2853 while (s+1 < send && *s != ')')
db42d148 2854 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
155aba94
GS
2855 }
2856 else if (s[2] == '{' /* This should match regcomp.c */
67edc0c9 2857 || (s[2] == '?' && s[3] == '{'))
155aba94 2858 {
cc6b7395 2859 I32 count = 1;
0f5d15d6 2860 char *regparse = s + (s[2] == '{' ? 3 : 4);
cc6b7395
IZ
2861 char c;
2862
d9f97599
GS
2863 while (count && (c = *regparse)) {
2864 if (c == '\\' && regparse[1])
2865 regparse++;
4e553d73 2866 else if (c == '{')
cc6b7395 2867 count++;
4e553d73 2868 else if (c == '}')
cc6b7395 2869 count--;
d9f97599 2870 regparse++;
cc6b7395 2871 }
e994fd66 2872 if (*regparse != ')')
5bdf89e7 2873 regparse--; /* Leave one char for continuation. */
0f5d15d6 2874 while (s < regparse)
db42d148 2875 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
cc6b7395 2876 }
748a9306 2877 }
02aa26ce
NT
2878
2879 /* likewise skip #-initiated comments in //x patterns */
3280af22 2880 else if (*s == '#' && PL_lex_inpat &&
73134a2e 2881 ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) {
748a9306 2882 while (s+1 < send && *s != '\n')
db42d148 2883 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
748a9306 2884 }
02aa26ce 2885
5d1d4326 2886 /* check for embedded arrays
da6eedaa 2887 (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-)
5d1d4326 2888 */
1749ea0d
ST
2889 else if (*s == '@' && s[1]) {
2890 if (isALNUM_lazy_if(s+1,UTF))
2891 break;
2892 if (strchr(":'{$", s[1]))
2893 break;
2894 if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-'))
2895 break; /* in regexp, neither @+ nor @- are interpolated */
2896 }
02aa26ce
NT
2897
2898 /* check for embedded scalars. only stop if we're sure it's a
2899 variable.
2900 */
79072805 2901 else if (*s == '$') {
3280af22 2902 if (!PL_lex_inpat) /* not a regexp, so $ must be var */
79072805 2903 break;
77772344 2904 if (s + 1 < send && !strchr("()| \r\n\t", s[1])) {
a2a5de95
NC
2905 if (s[1] == '\\') {
2906 Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS),
2907 "Possible unintended interpolation of $\\ in regex");
77772344 2908 }
79072805 2909 break; /* in regexp, $ might be tail anchor */
77772344 2910 }
79072805 2911 }
02aa26ce 2912
2b9d42f0
NIS
2913 /* End of else if chain - OP_TRANS rejoin rest */
2914
02aa26ce 2915 /* backslashes */
79072805 2916 if (*s == '\\' && s+1 < send) {
ff3f963a
KW
2917 char* e; /* Can be used for ending '}', etc. */
2918
79072805 2919 s++;
02aa26ce 2920
7d0fc23c
KW
2921 /* warn on \1 - \9 in substitution replacements, but note that \11
2922 * is an octal; and \19 is \1 followed by '9' */
3280af22 2923 if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat &&
a0d0e21e 2924 isDIGIT(*s) && *s != '0' && !isDIGIT(s[1]))
79072805 2925 {
a2a5de95 2926 Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s);
79072805
LW
2927 *--s = '$';
2928 break;
2929 }
02aa26ce
NT
2930
2931 /* string-change backslash escapes */
838f2281 2932 if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) {
79072805
LW
2933 --s;
2934 break;
2935 }
ff3f963a
KW
2936 /* In a pattern, process \N, but skip any other backslash escapes.
2937 * This is because we don't want to translate an escape sequence
2938 * into a meta symbol and have the regex compiler use the meta
2939 * symbol meaning, e.g. \x{2E} would be confused with a dot. But
2940 * in spite of this, we do have to process \N here while the proper
2941 * charnames handler is in scope. See bugs #56444 and #62056.
2942 * There is a complication because \N in a pattern may also stand
2943 * for 'match a non-nl', and not mean a charname, in which case its
2944 * processing should be deferred to the regex compiler. To be a
2945 * charname it must be followed immediately by a '{', and not look
2946 * like \N followed by a curly quantifier, i.e., not something like
2947 * \N{3,}. regcurly returns a boolean indicating if it is a legal
2948 * quantifier */
2949 else if (PL_lex_inpat
2950 && (*s != 'N'
2951 || s[1] != '{'
2952 || regcurly(s + 1)))
2953 {
cc74c5bd
ST
2954 *d++ = NATIVE_TO_NEED(has_utf8,'\\');
2955 goto default_action;
2956 }
02aa26ce 2957
79072805 2958 switch (*s) {
02aa26ce
NT
2959
2960 /* quoted - in transliterations */
79072805 2961 case '-':
3280af22 2962 if (PL_lex_inwhat == OP_TRANS) {
79072805
LW
2963 *d++ = *s++;
2964 continue;
2965 }
2966 /* FALL THROUGH */
2967 default:
11b8faa4 2968 {
a2a5de95
NC
2969 if ((isALPHA(*s) || isDIGIT(*s)))
2970 Perl_ck_warner(aTHX_ packWARN(WARN_MISC),
2971 "Unrecognized escape \\%c passed through",
2972 *s);
11b8faa4 2973 /* default action is to copy the quoted character */
f9a63242 2974 goto default_action;
11b8faa4 2975 }
02aa26ce 2976
632403cc 2977 /* eg. \132 indicates the octal constant 0132 */
79072805
LW
2978 case '0': case '1': case '2': case '3':
2979 case '4': case '5': case '6': case '7':
ba210ebe 2980 {
53305cf1
NC
2981 I32 flags = 0;
2982 STRLEN len = 3;
77a135fe 2983 uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL));
ba210ebe
JH
2984 s += len;
2985 }
012bcf8d 2986 goto NUM_ESCAPE_INSERT;
02aa26ce 2987
f0a2b745
KW
2988 /* eg. \o{24} indicates the octal constant \024 */
2989 case 'o':
2990 {
2991 STRLEN len;
454155d9 2992 const char* error;
f0a2b745 2993
454155d9 2994 bool valid = grok_bslash_o(s, &uv, &len, &error, 1);
f0a2b745 2995 s += len;
454155d9 2996 if (! valid) {
f0a2b745
KW
2997 yyerror(error);
2998 continue;
2999 }
3000 goto NUM_ESCAPE_INSERT;
3001 }
3002
77a135fe 3003 /* eg. \x24 indicates the hex constant 0x24 */
79072805 3004 case 'x':
a0ed51b3
LW
3005 ++s;
3006 if (*s == '{') {
9d4ba2ae 3007 char* const e = strchr(s, '}');
a4c04bdc
NC
3008 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES |
3009 PERL_SCAN_DISALLOW_PREFIX;
53305cf1 3010 STRLEN len;
355860ce 3011
53305cf1 3012 ++s;
adaeee49 3013 if (!e) {
a0ed51b3 3014 yyerror("Missing right brace on \\x{}");
355860ce 3015 continue;
ba210ebe 3016 }
53305cf1 3017 len = e - s;
77a135fe 3018 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
ba210ebe 3019 s = e + 1;
a0ed51b3
LW
3020 }
3021 else {
ba210ebe 3022 {
53305cf1 3023 STRLEN len = 2;
a4c04bdc 3024 I32 flags = PERL_SCAN_DISALLOW_PREFIX;
77a135fe 3025 uv = NATIVE_TO_UNI(grok_hex(s, &len, &flags, NULL));
ba210ebe
JH
3026 s += len;
3027 }
012bcf8d
GS
3028 }
3029
3030 NUM_ESCAPE_INSERT:
ff3f963a
KW
3031 /* Insert oct or hex escaped character. There will always be
3032 * enough room in sv since such escapes will be longer than any
3033 * UTF-8 sequence they can end up as, except if they force us
3034 * to recode the rest of the string into utf8 */
ba7cea30 3035
77a135fe 3036 /* Here uv is the ordinal of the next character being added in
ff3f963a 3037 * unicode (converted from native). */
77a135fe 3038 if (!UNI_IS_INVARIANT(uv)) {
9aa983d2 3039 if (!has_utf8 && uv > 255) {
77a135fe
KW
3040 /* Might need to recode whatever we have accumulated so
3041 * far if it contains any chars variant in utf8 or
3042 * utf-ebcdic. */
3043
3044 SvCUR_set(sv, d - SvPVX_const(sv));
3045 SvPOK_on(sv);
3046 *d = '\0';
77a135fe 3047 /* See Note on sizing above. */
7bf79863
KW
3048 sv_utf8_upgrade_flags_grow(sv,
3049 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3050 UNISKIP(uv) + (STRLEN)(send - s) + 1);
77a135fe
KW
3051 d = SvPVX(sv) + SvCUR(sv);
3052 has_utf8 = TRUE;
012bcf8d
GS
3053 }
3054
77a135fe
KW
3055 if (has_utf8) {
3056 d = (char*)uvuni_to_utf8((U8*)d, uv);
f9a63242
JH
3057 if (PL_lex_inwhat == OP_TRANS &&
3058 PL_sublex_info.sub_op) {
3059 PL_sublex_info.sub_op->op_private |=
3060 (PL_lex_repl ? OPpTRANS_FROM_UTF
3061 : OPpTRANS_TO_UTF);
f9a63242 3062 }
e294cc5d
JH
3063#ifdef EBCDIC
3064 if (uv > 255 && !dorange)
3065 native_range = FALSE;
3066#endif
012bcf8d 3067 }
a0ed51b3 3068 else {
012bcf8d 3069 *d++ = (char)uv;
a0ed51b3 3070 }
012bcf8d
GS
3071 }
3072 else {
c4d5f83a 3073 *d++ = (char) uv;
a0ed51b3 3074 }
79072805 3075 continue;
02aa26ce 3076
4a2d328f 3077 case 'N':
ff3f963a
KW
3078 /* In a non-pattern \N must be a named character, like \N{LATIN
3079 * SMALL LETTER A} or \N{U+0041}. For patterns, it also can
3080 * mean to match a non-newline. For non-patterns, named
3081 * characters are converted to their string equivalents. In
3082 * patterns, named characters are not converted to their
3083 * ultimate forms for the same reasons that other escapes
3084 * aren't. Instead, they are converted to the \N{U+...} form
3085 * to get the value from the charnames that is in effect right
3086 * now, while preserving the fact that it was a named character
3087 * so that the regex compiler knows this */
3088
3089 /* This section of code doesn't generally use the
3090 * NATIVE_TO_NEED() macro to transform the input. I (khw) did
3091 * a close examination of this macro and determined it is a
3092 * no-op except on utfebcdic variant characters. Every
3093 * character generated by this that would normally need to be
3094 * enclosed by this macro is invariant, so the macro is not
7538f724
KW
3095 * needed, and would complicate use of copy(). XXX There are
3096 * other parts of this file where the macro is used
3097 * inconsistently, but are saved by it being a no-op */
ff3f963a
KW
3098
3099 /* The structure of this section of code (besides checking for
3100 * errors and upgrading to utf8) is:
3101 * Further disambiguate between the two meanings of \N, and if
3102 * not a charname, go process it elsewhere
0a96133f
KW
3103 * If of form \N{U+...}, pass it through if a pattern;
3104 * otherwise convert to utf8
3105 * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a
3106 * pattern; otherwise convert to utf8 */
ff3f963a
KW
3107
3108 /* Here, s points to the 'N'; the test below is guaranteed to
3109 * succeed if we are being called on a pattern as we already
3110 * know from a test above that the next character is a '{'.
3111 * On a non-pattern \N must mean 'named sequence, which
3112 * requires braces */
3113 s++;
3114 if (*s != '{') {
3115 yyerror("Missing braces on \\N{}");
3116 continue;
3117 }
3118 s++;
3119
0a96133f 3120 /* If there is no matching '}', it is an error. */
ff3f963a
KW
3121 if (! (e = strchr(s, '}'))) {
3122 if (! PL_lex_inpat) {
5777a3f7 3123 yyerror("Missing right brace on \\N{}");
0a96133f
KW
3124 } else {
3125 yyerror("Missing right brace on \\N{} or unescaped left brace after \\N.");
dbc0d4f2 3126 }
0a96133f 3127 continue;
ff3f963a 3128 }
cddc7ef4 3129
ff3f963a 3130 /* Here it looks like a named character */
cddc7ef4 3131
ff3f963a
KW
3132 if (PL_lex_inpat) {
3133
3134 /* XXX This block is temporary code. \N{} implies that the
3135 * pattern is to have Unicode semantics, and therefore
3136 * currently has to be encoded in utf8. By putting it in
3137 * utf8 now, we save a whole pass in the regular expression
3138 * compiler. Once that code is changed so Unicode
3139 * semantics doesn't necessarily have to be in utf8, this
da3a4baf
KW
3140 * block should be removed. However, the code that parses
3141 * the output of this would have to be changed to not
3142 * necessarily expect utf8 */
ff3f963a 3143 if (!has_utf8) {
77a135fe 3144 SvCUR_set(sv, d - SvPVX_const(sv));
f08d6ad9 3145 SvPOK_on(sv);
e4f3eed8 3146 *d = '\0';
77a135fe 3147 /* See Note on sizing above. */
7bf79863 3148 sv_utf8_upgrade_flags_grow(sv,
ff3f963a
KW
3149 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3150 /* 5 = '\N{' + cur char + NUL */
3151 (STRLEN)(send - s) + 5);
f08d6ad9 3152 d = SvPVX(sv) + SvCUR(sv);
89491803 3153 has_utf8 = TRUE;
ff3f963a
KW
3154 }
3155 }
423cee85 3156
ff3f963a
KW
3157 if (*s == 'U' && s[1] == '+') { /* \N{U+...} */
3158 I32 flags = PERL_SCAN_ALLOW_UNDERSCORES
3159 | PERL_SCAN_DISALLOW_PREFIX;
3160 STRLEN len;
3161
3162 /* For \N{U+...}, the '...' is a unicode value even on
3163 * EBCDIC machines */
3164 s += 2; /* Skip to next char after the 'U+' */
3165 len = e - s;
3166 uv = grok_hex(s, &len, &flags, NULL);
3167 if (len == 0 || len != (STRLEN)(e - s)) {
3168 yyerror("Invalid hexadecimal number in \\N{U+...}");
3169 s = e + 1;
3170 continue;
3171 }
3172
3173 if (PL_lex_inpat) {
3174
e2a7e165
KW
3175 /* On non-EBCDIC platforms, pass through to the regex
3176 * compiler unchanged. The reason we evaluated the
3177 * number above is to make sure there wasn't a syntax
3178 * error. But on EBCDIC we convert to native so
3179 * downstream code can continue to assume it's native
3180 */
ff3f963a 3181 s -= 5; /* Include the '\N{U+' */
e2a7e165
KW
3182#ifdef EBCDIC
3183 d += my_snprintf(d, e - s + 1 + 1, /* includes the }
3184 and the \0 */
3185 "\\N{U+%X}",
3186 (unsigned int) UNI_TO_NATIVE(uv));
3187#else
ff3f963a
KW
3188 Copy(s, d, e - s + 1, char); /* 1 = include the } */
3189 d += e - s + 1;
e2a7e165 3190#endif
ff3f963a
KW
3191 }
3192 else { /* Not a pattern: convert the hex to string */
3193
3194 /* If destination is not in utf8, unconditionally
3195 * recode it to be so. This is because \N{} implies
3196 * Unicode semantics, and scalars have to be in utf8
3197 * to guarantee those semantics */
3198 if (! has_utf8) {
3199 SvCUR_set(sv, d - SvPVX_const(sv));
3200 SvPOK_on(sv);
3201 *d = '\0';
3202 /* See Note on sizing above. */
3203 sv_utf8_upgrade_flags_grow(
3204 sv,
3205 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3206 UNISKIP(uv) + (STRLEN)(send - e) + 1);
3207 d = SvPVX(sv) + SvCUR(sv);
3208 has_utf8 = TRUE;
3209 }
3210
3211 /* Add the string to the output */
3212 if (UNI_IS_INVARIANT(uv)) {
3213 *d++ = (char) uv;
3214 }
3215 else d = (char*)uvuni_to_utf8((U8*)d, uv);
3216 }
3217 }
3218 else { /* Here is \N{NAME} but not \N{U+...}. */
3219
3220 SV *res; /* result from charnames */
3221 const char *str; /* the string in 'res' */
3222 STRLEN len; /* its length */
3223
3224 /* Get the value for NAME */
3225 res = newSVpvn(s, e - s);
3226 res = new_constant( NULL, 0, "charnames",
3227 /* includes all of: \N{...} */
3228 res, NULL, s - 3, e - s + 4 );
3229
3230 /* Most likely res will be in utf8 already since the
3231 * standard charnames uses pack U, but a custom translator
3232 * can leave it otherwise, so make sure. XXX This can be
3233 * revisited to not have charnames use utf8 for characters
3234 * that don't need it when regexes don't have to be in utf8
3235 * for Unicode semantics. If doing so, remember EBCDIC */
3236 sv_utf8_upgrade(res);
3237 str = SvPV_const(res, len);
3238
3239 /* Don't accept malformed input */
3240 if (! is_utf8_string((U8 *) str, len)) {
3241 yyerror("Malformed UTF-8 returned by \\N");
3242 }
3243 else if (PL_lex_inpat) {
3244
3245 if (! len) { /* The name resolved to an empty string */
3246 Copy("\\N{}", d, 4, char);
3247 d += 4;
3248 }
3249 else {
3250 /* In order to not lose information for the regex
3251 * compiler, pass the result in the specially made
3252 * syntax: \N{U+c1.c2.c3...}, where c1 etc. are
3253 * the code points in hex of each character
3254 * returned by charnames */
3255
3256 const char *str_end = str + len;
3257 STRLEN char_length; /* cur char's byte length */
3258 STRLEN output_length; /* and the number of bytes
3259 after this is translated
3260 into hex digits */
3261 const STRLEN off = d - SvPVX_const(sv);
3262
3263 /* 2 hex per byte; 2 chars for '\N'; 2 chars for
3264 * max('U+', '.'); and 1 for NUL */
3265 char hex_string[2 * UTF8_MAXBYTES + 5];
3266
3267 /* Get the first character of the result. */
3268 U32 uv = utf8n_to_uvuni((U8 *) str,
3269 len,
3270 &char_length,
3271 UTF8_ALLOW_ANYUV);
3272
3273 /* The call to is_utf8_string() above hopefully
3274 * guarantees that there won't be an error. But
3275 * it's easy here to make sure. The function just
3276 * above warns and returns 0 if invalid utf8, but
3277 * it can also return 0 if the input is validly a
3278 * NUL. Disambiguate */
3279 if (uv == 0 && NATIVE_TO_ASCII(*str) != '\0') {
3280 uv = UNICODE_REPLACEMENT;
3281 }
3282
3283 /* Convert first code point to hex, including the
e2a7e165
KW
3284 * boiler plate before it. For all these, we
3285 * convert to native format so that downstream code
3286 * can continue to assume the input is native */
78c35590 3287 output_length =
3353de27 3288 my_snprintf(hex_string, sizeof(hex_string),
e2a7e165
KW
3289 "\\N{U+%X",
3290 (unsigned int) UNI_TO_NATIVE(uv));
ff3f963a
KW
3291
3292 /* Make sure there is enough space to hold it */
3293 d = off + SvGROW(sv, off
3294 + output_length
3295 + (STRLEN)(send - e)
3296 + 2); /* '}' + NUL */
3297 /* And output it */
3298 Copy(hex_string, d, output_length, char);
3299 d += output_length;
3300
3301 /* For each subsequent character, append dot and
3302 * its ordinal in hex */
3303 while ((str += char_length) < str_end) {
3304 const STRLEN off = d - SvPVX_const(sv);
3305 U32 uv = utf8n_to_uvuni((U8 *) str,
3306 str_end - str,
3307 &char_length,
3308 UTF8_ALLOW_ANYUV);
3309 if (uv == 0 && NATIVE_TO_ASCII(*str) != '\0') {
3310 uv = UNICODE_REPLACEMENT;
3311 }
3312
78c35590 3313 output_length =
3353de27 3314 my_snprintf(hex_string, sizeof(hex_string),
e2a7e165
KW
3315 ".%X",
3316 (unsigned int) UNI_TO_NATIVE(uv));
ff3f963a
KW
3317
3318 d = off + SvGROW(sv, off
3319 + output_length
3320 + (STRLEN)(send - e)
3321 + 2); /* '}' + NUL */
3322 Copy(hex_string, d, output_length, char);
3323 d += output_length;
3324 }
3325
3326 *d++ = '}'; /* Done. Add the trailing brace */
3327 }
3328 }
3329 else { /* Here, not in a pattern. Convert the name to a
3330 * string. */
3331
3332 /* If destination is not in utf8, unconditionally
3333 * recode it to be so. This is because \N{} implies
3334 * Unicode semantics, and scalars have to be in utf8
3335 * to guarantee those semantics */
3336 if (! has_utf8) {
3337 SvCUR_set(sv, d - SvPVX_const(sv));
3338 SvPOK_on(sv);
3339 *d = '\0';
3340 /* See Note on sizing above. */
3341 sv_utf8_upgrade_flags_grow(sv,
3342 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3343 len + (STRLEN)(send - s) + 1);
3344 d = SvPVX(sv) + SvCUR(sv);
3345 has_utf8 = TRUE;
3346 } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */
3347
3348 /* See Note on sizing above. (NOTE: SvCUR() is not
3349 * set correctly here). */
3350 const STRLEN off = d - SvPVX_const(sv);
3351 d = off + SvGROW(sv, off + len + (STRLEN)(send - s) + 1);
3352 }
3353 Copy(str, d, len, char);
3354 d += len;
423cee85 3355 }
423cee85 3356 SvREFCNT_dec(res);
cb233ae3
KW
3357
3358 /* Deprecate non-approved name syntax */
3359 if (ckWARN_d(WARN_DEPRECATED)) {
3360 bool problematic = FALSE;
3361 char* i = s;
3362
3363 /* For non-ut8 input, look to see that the first
3364 * character is an alpha, then loop through the rest
3365 * checking that each is a continuation */
3366 if (! this_utf8) {
3367 if (! isALPHAU(*i)) problematic = TRUE;
3368 else for (i = s + 1; i < e; i++) {
3369 if (isCHARNAME_CONT(*i)) continue;
3370 problematic = TRUE;
3371 break;
3372 }
3373 }
3374 else {
3375 /* Similarly for utf8. For invariants can check
3376 * directly. We accept anything above the latin1
3377 * range because it is immaterial to Perl if it is
3378 * correct or not, and is expensive to check. But
3379 * it is fairly easy in the latin1 range to convert
3380 * the variants into a single character and check
3381 * those */
3382 if (UTF8_IS_INVARIANT(*i)) {
3383 if (! isALPHAU(*i)) problematic = TRUE;
3384 } else if (UTF8_IS_DOWNGRADEABLE_START(*i)) {
81c14aa2 3385 if (! isALPHAU(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*i,
cb233ae3
KW
3386 *(i+1)))))
3387 {
3388 problematic = TRUE;
3389 }
3390 }
3391 if (! problematic) for (i = s + UTF8SKIP(s);
3392 i < e;
3393 i+= UTF8SKIP(i))
3394 {
3395 if (UTF8_IS_INVARIANT(*i)) {
3396 if (isCHARNAME_CONT(*i)) continue;
3397 } else if (! UTF8_IS_DOWNGRADEABLE_START(*i)) {
3398 continue;
3399 } else if (isCHARNAME_CONT(
3400 UNI_TO_NATIVE(
81c14aa2 3401 TWO_BYTE_UTF8_TO_UNI(*i, *(i+1)))))
cb233ae3
KW
3402 {
3403 continue;
3404 }
3405 problematic = TRUE;
3406 break;
3407 }
3408 }
3409 if (problematic) {
6e1bad6c
KW
3410 /* The e-i passed to the final %.*s makes sure that
3411 * should the trailing NUL be missing that this
3412 * print won't run off the end of the string */
cb233ae3 3413 Perl_warner(aTHX_ packWARN(WARN_DEPRECATED),
b00fc8d4
NC
3414 "Deprecated character in \\N{...}; marked by <-- HERE in \\N{%.*s<-- HERE %.*s",
3415 (int)(i - s + 1), s, (int)(e - i), i + 1);
cb233ae3
KW
3416 }
3417 }
3418 } /* End \N{NAME} */
ff3f963a
KW
3419#ifdef EBCDIC
3420 if (!dorange)
3421 native_range = FALSE; /* \N{} is defined to be Unicode */
3422#endif
3423 s = e + 1; /* Point to just after the '}' */
423cee85
JH
3424 continue;
3425
02aa26ce 3426 /* \c is a control character */
79072805
LW
3427 case 'c':
3428 s++;
961ce445 3429 if (s < send) {
17a3df4c 3430 *d++ = grok_bslash_c(*s++, has_utf8, 1);
ba210ebe 3431 }
961ce445
RGS
3432 else {
3433 yyerror("Missing control char name in \\c");
3434 }
79072805 3435 continue;
02aa26ce
NT
3436
3437 /* printf-style backslashes, formfeeds, newlines, etc */
79072805 3438 case 'b':
db42d148 3439 *d++ = NATIVE_TO_NEED(has_utf8,'\b');
79072805
LW
3440 break;
3441 case 'n':
db42d148 3442 *d++ = NATIVE_TO_NEED(has_utf8,'\n');
79072805
LW
3443 break;
3444 case 'r':
db42d148 3445 *d++ = NATIVE_TO_NEED(has_utf8,'\r');
79072805
LW
3446 break;
3447 case 'f':
db42d148 3448 *d++ = NATIVE_TO_NEED(has_utf8,'\f');
79072805
LW
3449 break;
3450 case 't':
db42d148 3451 *d++ = NATIVE_TO_NEED(has_utf8,'\t');
79072805 3452 break;
34a3fe2a 3453 case 'e':
db42d148 3454 *d++ = ASCII_TO_NEED(has_utf8,'\033');
34a3fe2a
PP
3455 break;
3456 case 'a':
db42d148 3457 *d++ = ASCII_TO_NEED(has_utf8,'\007');
79072805 3458 break;
02aa26ce
NT
3459 } /* end switch */
3460
79072805
LW
3461 s++;
3462 continue;
02aa26ce 3463 } /* end if (backslash) */
4c3a8340
ST
3464#ifdef EBCDIC
3465 else
3466 literal_endpoint++;
3467#endif
02aa26ce 3468
f9a63242 3469 default_action:
77a135fe
KW
3470 /* If we started with encoded form, or already know we want it,
3471 then encode the next character */
3472 if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) {
2b9d42f0 3473 STRLEN len = 1;
77a135fe
KW
3474
3475
3476 /* One might think that it is wasted effort in the case of the
3477 * source being utf8 (this_utf8 == TRUE) to take the next character
3478 * in the source, convert it to an unsigned value, and then convert
3479 * it back again. But the source has not been validated here. The
3480 * routine that does the conversion checks for errors like
3481 * malformed utf8 */
3482
5f66b61c
AL
3483 const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s);
3484 const STRLEN need = UNISKIP(NATIVE_TO_UNI(nextuv));
77a135fe
KW
3485 if (!has_utf8) {
3486 SvCUR_set(sv, d - SvPVX_const(sv));
3487 SvPOK_on(sv);
3488 *d = '\0';
77a135fe 3489 /* See Note on sizing above. */
7bf79863
KW
3490 sv_utf8_upgrade_flags_grow(sv,
3491 SV_GMAGIC|SV_FORCE_UTF8_UPGRADE,
3492 need + (STRLEN)(send - s) + 1);
77a135fe
KW
3493 d = SvPVX(sv) + SvCUR(sv);
3494 has_utf8 = TRUE;
3495 } else if (need > len) {
3496 /* encoded value larger than old, may need extra space (NOTE:
3497 * SvCUR() is not set correctly here). See Note on sizing
3498 * above. */
9d4ba2ae 3499 const STRLEN off = d - SvPVX_const(sv);
77a135fe 3500 d = SvGROW(sv, off + need + (STRLEN)(send - s) + 1) + off;
2b9d42f0 3501 }
77a135fe
KW
3502 s += len;
3503
5f66b61c 3504 d = (char*)uvchr_to_utf8((U8*)d, nextuv);
e294cc5d
JH
3505#ifdef EBCDIC
3506 if (uv > 255 && !dorange)
3507 native_range = FALSE;
3508#endif
2b9d42f0
NIS
3509 }
3510 else {
3511 *d++ = NATIVE_TO_NEED(has_utf8,*s++);
3512 }
02aa26ce
NT
3513 } /* while loop to process each character */
3514
3515 /* terminate the string and set up the sv */
79072805 3516 *d = '\0';
95a20fc0 3517 SvCUR_set(sv, d - SvPVX_const(sv));
2b9d42f0 3518 if (SvCUR(sv) >= SvLEN(sv))
5637ef5b
NC
3519 Perl_croak(aTHX_ "panic: constant overflowed allocated space, %"UVuf
3520 " >= %"UVuf, (UV)SvCUR(sv), (UV)SvLEN(sv));
2b9d42f0 3521
79072805 3522 SvPOK_on(sv);
9f4817db 3523 if (PL_encoding && !has_utf8) {
d0063567
DK
3524 sv_recode_to_utf8(sv, PL_encoding);
3525 if (SvUTF8(sv))
3526 has_utf8 = TRUE;
9f4817db 3527 }
2b9d42f0 3528 if (has_utf8) {
7e2040f0 3529 SvUTF8_on(sv);
2b9d42f0 3530 if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) {
d0063567 3531 PL_sublex_info.sub_op->op_private |=
2b9d42f0
NIS
3532 (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF);
3533 }
3534 }
79072805 3535
02aa26ce 3536 /* shrink the sv if we allocated more than we used */
79072805 3537 if (SvCUR(sv) + 5 < SvLEN(sv)) {
1da4ca5f 3538 SvPV_shrink_to_cur(sv);
79072805 3539 }
02aa26ce 3540
6154021b 3541 /* return the substring (via pl_yylval) only if we parsed anything */
3280af22 3542 if (s > PL_bufptr) {
eb0d8d16
NC
3543 if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) {
3544 const char *const key = PL_lex_inpat ? "qr" : "q";
3545 const STRLEN keylen = PL_lex_inpat ? 2 : 1;
3546 const char *type;
3547 STRLEN typelen;
3548
3549 if (PL_lex_inwhat == OP_TRANS) {
3550 type = "tr";
3551 typelen = 2;
3552 } else if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) {
3553 type = "s";
3554 typelen = 1;
3555 } else {
3556 type = "qq";
3557 typelen = 2;
3558 }
3559
3560 sv = S_new_constant(aTHX_ start, s - start, key, keylen, sv, NULL,
3561 type, typelen);
3562 }
6154021b 3563 pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv);
b3ac6de7 3564 } else
8990e307 3565 SvREFCNT_dec(sv);
79072805
LW
3566 return s;
3567}
3568
ffb4593c
NT
3569/* S_intuit_more
3570 * Returns TRUE if there's more to the expression (e.g., a subscript),
3571 * FALSE otherwise.
ffb4593c
NT
3572 *
3573 * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/
3574 *
3575 * ->[ and ->{ return TRUE
3576 * { and [ outside a pattern are always subscripts, so return TRUE
3577 * if we're outside a pattern and it's not { or [, then return FALSE
3578 * if we're in a pattern and the first char is a {
3579 * {4,5} (any digits around the comma) returns FALSE
3580 * if we're in a pattern and the first char is a [
3581 * [] returns FALSE
3582 * [SOMETHING] has a funky algorithm to decide whether it's a
3583 * character class or not. It has to deal with things like
3584 * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/
3585 * anything else returns TRUE
3586 */
3587
9cbb5ea2
GS
3588/* This is the one truly awful dwimmer necessary to conflate C and sed. */
3589
76e3520e 3590STATIC int
cea2e8a9 3591S_intuit_more(pTHX_ register char *s)
79072805 3592{
97aff369 3593 dVAR;
7918f24d
NC
3594
3595 PERL_ARGS_ASSERT_INTUIT_MORE;
3596
3280af22 3597 if (PL_lex_brackets)
79072805
LW
3598 return TRUE;
3599 if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{'))
3600 return TRUE;
3601 if (*s != '{' && *s != '[')
3602 return FALSE;
3280af22 3603 if (!PL_lex_inpat)
79072805
LW
3604 return TRUE;
3605
3606 /* In a pattern, so maybe we have {n,m}. */
3607 if (*s == '{') {
b3155d95 3608 if (regcurly(s)) {
79072805 3609 return FALSE;
b3155d95 3610 }
79072805 3611 return TRUE;
79072805
LW
3612 }
3613
3614 /* On the other hand, maybe we have a character class */
3615
3616 s++;
3617 if (*s == ']' || *s == '^')
3618 return FALSE;
3619 else {
ffb4593c 3620 /* this is terrifying, and it works */
79072805
LW
3621 int weight = 2; /* let's weigh the evidence */
3622 char seen[256];
f27ffc4a 3623 unsigned char un_char = 255, last_un_char;
9d4ba2ae 3624 const char * const send = strchr(s,']');
3280af22 3625 char tmpbuf[sizeof PL_tokenbuf * 4];
79072805
LW
3626
3627 if (!send) /* has to be an expression */
3628 return TRUE;
3629
3630 Zero(seen,256,char);
3631 if (*s == '$')
3632 weight -= 3;
3633 else if (isDIGIT(*s)) {
3634 if (s[1] != ']') {
3635 if (isDIGIT(s[1]) && s[2] == ']')
3636 weight -= 10;
3637 }
3638 else
3639 weight -= 100;
3640 }
3641 for (; s < send; s++) {
3642 last_un_char = un_char;
3643 un_char = (unsigned char)*s;
3644 switch (*s) {
3645 case '@':
3646 case '&':
3647 case '$':
3648 weight -= seen[un_char] * 10;
7e2040f0 3649 if (isALNUM_lazy_if(s+1,UTF)) {
90e5519e 3650 int len;
8903cb82 3651 scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE);
90e5519e 3652 len = (int)strlen(tmpbuf);
6fbd0d97
BF
3653 if (len > 1 && gv_fetchpvn_flags(tmpbuf, len,
3654 UTF ? SVf_UTF8 : 0, SVt_PV))
79072805
LW
3655 weight -= 100;
3656 else
3657 weight -= 10;
3658 }
3659 else if (*s == '$' && s[1] &&
93a17b20
LW
3660 strchr("[#!%*<>()-=",s[1])) {
3661 if (/*{*/ strchr("])} =",s[2]))
79072805
LW
3662 weight -= 10;
3663 else
3664 weight -= 1;
3665 }
3666 break;
3667 case '\\':
3668 un_char = 254;
3669 if (s[1]) {
93a17b20 3670 if (strchr("wds]",s[1]))
79072805 3671 weight += 100;
10edeb5d 3672 else if (seen[(U8)'\''] || seen[(U8)'"'])
79072805 3673 weight += 1;
93a17b20 3674 else if (strchr("rnftbxcav",s[1]))
79072805
LW
3675 weight += 40;
3676 else if (isDIGIT(s[1])) {
3677 weight += 40;
3678 while (s[1] && isDIGIT(s[1]))
3679 s++;
3680 }
3681 }
3682 else
3683 weight += 100;
3684 break;
3685 case '-':
3686 if (s[1] == '\\')
3687 weight += 50;
93a17b20 3688 if (strchr("aA01! ",last_un_char))
79072805 3689 weight += 30;
93a17b20 3690 if (strchr("zZ79~",s[1]))
79072805 3691 weight += 30;
f27ffc4a
GS
3692 if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$'))
3693 weight -= 5; /* cope with negative subscript */
79072805
LW
3694 break;
3695 default:
3792a11b
NC
3696 if (!isALNUM(last_un_char)
3697 && !(last_un_char == '$' || last_un_char == '@'
3698 || last_un_char == '&')
3699 && isALPHA(*s) && s[1] && isALPHA(s[1])) {
79072805
LW
3700 char *d = tmpbuf;
3701 while (isALPHA(*s))
3702 *d++ = *s++;
3703 *d = '\0';
5458a98a 3704 if (keyword(tmpbuf, d - tmpbuf, 0))
79072805
LW
3705 weight -= 150;
3706 }
3707 if (un_char == last_un_char + 1)
3708 weight += 5;
3709 weight -= seen[un_char];
3710 break;
3711 }
3712 seen[un_char]++;
3713 }
3714 if (weight >= 0) /* probably a character class */
3715 return FALSE;
3716 }
3717
3718 return TRUE;
3719}
ffed7fef 3720
ffb4593c
NT
3721/*
3722 * S_intuit_method
3723 *
3724 * Does all the checking to disambiguate
3725 * foo bar
3726 * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise
3727 * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args).
3728 *
3729 * First argument is the stuff after the first token, e.g. "bar".
3730 *
3731 * Not a method if bar is a filehandle.
3732 * Not a method if foo is a subroutine prototyped to take a filehandle.
3733 * Not a method if it's really "Foo $bar"
3734 * Method if it's "foo $bar"
3735 * Not a method if it's really "print foo $bar"
3736 * Method if it's really "foo package::" (interpreted as package->foo)
8f8cf39c 3737 * Not a method if bar is known to be a subroutine ("sub bar; foo bar")
3cb0bbe5 3738 * Not a method if bar is a filehandle or package, but is quoted with
ffb4593c
NT
3739 * =>
3740 */
3741
76e3520e 3742STATIC int
62d55b22 3743S_intuit_method(pTHX_ char *start, GV *gv, CV *cv)
a0d0e21e 3744{
97aff369 3745 dVAR;
a0d0e21e 3746 char *s = start + (*start == '$');
3280af22 3747 char tmpbuf[sizeof PL_tokenbuf];
a0d0e21e
LW
3748 STRLEN len;
3749 GV* indirgv;
5db06880
NC
3750#ifdef PERL_MAD
3751 int soff;
3752#endif
a0d0e21e 3753
7918f24d
NC
3754 PERL_ARGS_ASSERT_INTUIT_METHOD;
3755
a0d0e21e 3756 if (gv) {
62d55b22 3757 if (SvTYPE(gv) == SVt_PVGV && GvIO(gv))
a0d0e21e 3758 return 0;
62d55b22
NC
3759 if (cv) {
3760 if (SvPOK(cv)) {
8fa6a409 3761 const char *proto = CvPROTO(cv);
62d55b22
NC