Commit | Line | Data |
---|---|---|
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 | ||
27 | This is the lower layer of the Perl parser, managing characters and tokens. | |
28 | ||
29 | =for apidoc AmU|yy_parser *|PL_parser | |
30 | ||
31 | Pointer to a structure encapsulating the state of the parsing operation | |
32 | currently in progress. The pointer can be locally changed to perform | |
33 | a nested parse without interfering with the state of an outer parse. | |
34 | Individual members of C<PL_parser> have their own documentation. | |
35 | ||
36 | =cut | |
37 | */ | |
38 | ||
378cc40b | 39 | #include "EXTERN.h" |
864dbfa3 | 40 | #define PERL_IN_TOKE_C |
378cc40b | 41 | #include "perl.h" |
04e98a4d | 42 | #include "dquote_static.c" |
378cc40b | 43 | |
eb0d8d16 NC |
44 | #define new_constant(a,b,c,d,e,f,g) \ |
45 | S_new_constant(aTHX_ a,b,STR_WITH_LEN(c),d,e,f, g) | |
46 | ||
6154021b | 47 | #define pl_yylval (PL_parser->yylval) |
d3b6f988 | 48 | |
199e78b7 DM |
49 | /* XXX temporary backwards compatibility */ |
50 | #define PL_lex_brackets (PL_parser->lex_brackets) | |
78cdf107 Z |
51 | #define PL_lex_allbrackets (PL_parser->lex_allbrackets) |
52 | #define PL_lex_fakeeof (PL_parser->lex_fakeeof) | |
199e78b7 DM |
53 | #define PL_lex_brackstack (PL_parser->lex_brackstack) |
54 | #define PL_lex_casemods (PL_parser->lex_casemods) | |
55 | #define PL_lex_casestack (PL_parser->lex_casestack) | |
56 | #define PL_lex_defer (PL_parser->lex_defer) | |
57 | #define PL_lex_dojoin (PL_parser->lex_dojoin) | |
58 | #define PL_lex_expect (PL_parser->lex_expect) | |
59 | #define PL_lex_formbrack (PL_parser->lex_formbrack) | |
60 | #define PL_lex_inpat (PL_parser->lex_inpat) | |
61 | #define PL_lex_inwhat (PL_parser->lex_inwhat) | |
62 | #define PL_lex_op (PL_parser->lex_op) | |
63 | #define PL_lex_repl (PL_parser->lex_repl) | |
64 | #define PL_lex_starts (PL_parser->lex_starts) | |
65 | #define PL_lex_stuff (PL_parser->lex_stuff) | |
66 | #define PL_multi_start (PL_parser->multi_start) | |
67 | #define PL_multi_open (PL_parser->multi_open) | |
68 | #define PL_multi_close (PL_parser->multi_close) | |
199e78b7 DM |
69 | #define PL_preambled (PL_parser->preambled) |
70 | #define PL_sublex_info (PL_parser->sublex_info) | |
bdc0bf6f | 71 | #define PL_linestr (PL_parser->linestr) |
c2598295 DM |
72 | #define PL_expect (PL_parser->expect) |
73 | #define PL_copline (PL_parser->copline) | |
f06b5848 DM |
74 | #define PL_bufptr (PL_parser->bufptr) |
75 | #define PL_oldbufptr (PL_parser->oldbufptr) | |
76 | #define PL_oldoldbufptr (PL_parser->oldoldbufptr) | |
77 | #define PL_linestart (PL_parser->linestart) | |
78 | #define PL_bufend (PL_parser->bufend) | |
79 | #define PL_last_uni (PL_parser->last_uni) | |
80 | #define PL_last_lop (PL_parser->last_lop) | |
81 | #define PL_last_lop_op (PL_parser->last_lop_op) | |
bc177e6b | 82 | #define PL_lex_state (PL_parser->lex_state) |
2f9285f8 | 83 | #define PL_rsfp (PL_parser->rsfp) |
5486870f | 84 | #define PL_rsfp_filters (PL_parser->rsfp_filters) |
12bd6ede DM |
85 | #define PL_in_my (PL_parser->in_my) |
86 | #define PL_in_my_stash (PL_parser->in_my_stash) | |
14047fc9 | 87 | #define PL_tokenbuf (PL_parser->tokenbuf) |
670a9cb2 | 88 | #define PL_multi_end (PL_parser->multi_end) |
13765c85 | 89 | #define PL_error_count (PL_parser->error_count) |
199e78b7 DM |
90 | |
91 | #ifdef PERL_MAD | |
92 | # define PL_endwhite (PL_parser->endwhite) | |
93 | # define PL_faketokens (PL_parser->faketokens) | |
94 | # define PL_lasttoke (PL_parser->lasttoke) | |
95 | # define PL_nextwhite (PL_parser->nextwhite) | |
96 | # define PL_realtokenstart (PL_parser->realtokenstart) | |
97 | # define PL_skipwhite (PL_parser->skipwhite) | |
98 | # define PL_thisclose (PL_parser->thisclose) | |
99 | # define PL_thismad (PL_parser->thismad) | |
100 | # define PL_thisopen (PL_parser->thisopen) | |
101 | # define PL_thisstuff (PL_parser->thisstuff) | |
102 | # define PL_thistoken (PL_parser->thistoken) | |
103 | # define PL_thiswhite (PL_parser->thiswhite) | |
fb205e7a DM |
104 | # define PL_thiswhite (PL_parser->thiswhite) |
105 | # define PL_nexttoke (PL_parser->nexttoke) | |
106 | # define PL_curforce (PL_parser->curforce) | |
107 | #else | |
108 | # define PL_nexttoke (PL_parser->nexttoke) | |
109 | # define PL_nexttype (PL_parser->nexttype) | |
110 | # define PL_nextval (PL_parser->nextval) | |
199e78b7 DM |
111 | #endif |
112 | ||
a1894d81 | 113 | static const char* const ident_too_long = "Identifier too long"; |
8903cb82 | 114 | |
29595ff2 | 115 | #ifdef PERL_MAD |
29595ff2 | 116 | # define CURMAD(slot,sv) if (PL_madskills) { curmad(slot,sv); sv = 0; } |
cd81e915 | 117 | # define NEXTVAL_NEXTTOKE PL_nexttoke[PL_curforce].next_val |
9ded7720 | 118 | #else |
5db06880 | 119 | # define CURMAD(slot,sv) |
9ded7720 | 120 | # define NEXTVAL_NEXTTOKE PL_nextval[PL_nexttoke] |
29595ff2 NC |
121 | #endif |
122 | ||
a7aaec61 Z |
123 | #define XENUMMASK 0x3f |
124 | #define XFAKEEOF 0x40 | |
125 | #define XFAKEBRACK 0x80 | |
9059aa12 | 126 | |
39e02b42 JH |
127 | #ifdef USE_UTF8_SCRIPTS |
128 | # define UTF (!IN_BYTES) | |
2b9d42f0 | 129 | #else |
802a15e9 | 130 | # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || ( !(PL_parser->lex_flags & LEX_IGNORE_UTF8_HINTS) && (PL_hints & HINT_UTF8))) |
2b9d42f0 | 131 | #endif |
a0ed51b3 | 132 | |
b1fc3636 CJ |
133 | /* The maximum number of characters preceding the unrecognized one to display */ |
134 | #define UNRECOGNIZED_PRECEDE_COUNT 10 | |
135 | ||
61f0cdd9 | 136 | /* In variables named $^X, these are the legal values for X. |
2b92dfce GS |
137 | * 1999-02-27 mjd-perl-patch@plover.com */ |
138 | #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) | |
139 | ||
14bd96d0 | 140 | #define SPACE_OR_TAB(c) isBLANK_A(c) |
bf4acbe4 | 141 | |
ffb4593c NT |
142 | /* LEX_* are values for PL_lex_state, the state of the lexer. |
143 | * They are arranged oddly so that the guard on the switch statement | |
79072805 | 144 | * can get by with a single comparison (if the compiler is smart enough). |
9da1dd8f DM |
145 | * |
146 | * These values refer to the various states within a sublex parse, | |
147 | * i.e. within a double quotish string | |
79072805 LW |
148 | */ |
149 | ||
fb73857a | 150 | /* #define LEX_NOTPARSING 11 is done in perl.h. */ |
151 | ||
b6007c36 DM |
152 | #define LEX_NORMAL 10 /* normal code (ie not within "...") */ |
153 | #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */ | |
154 | #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */ | |
155 | #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */ | |
156 | #define LEX_INTERPSTART 6 /* expecting the start of a $var */ | |
157 | ||
158 | /* at end of code, eg "$x" followed by: */ | |
159 | #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */ | |
160 | #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */ | |
161 | ||
162 | #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of | |
163 | string or after \E, $foo, etc */ | |
164 | #define LEX_INTERPCONST 2 /* NOT USED */ | |
165 | #define LEX_FORMLINE 1 /* expecting a format line */ | |
166 | #define LEX_KNOWNEXT 0 /* next token known; just return it */ | |
167 | ||
79072805 | 168 | |
bbf60fe6 | 169 | #ifdef DEBUGGING |
27da23d5 | 170 | static const char* const lex_state_names[] = { |
bbf60fe6 DM |
171 | "KNOWNEXT", |
172 | "FORMLINE", | |
173 | "INTERPCONST", | |
174 | "INTERPCONCAT", | |
175 | "INTERPENDMAYBE", | |
176 | "INTERPEND", | |
177 | "INTERPSTART", | |
178 | "INTERPPUSH", | |
179 | "INTERPCASEMOD", | |
180 | "INTERPNORMAL", | |
181 | "NORMAL" | |
182 | }; | |
183 | #endif | |
184 | ||
79072805 LW |
185 | #ifdef ff_next |
186 | #undef ff_next | |
d48672a2 LW |
187 | #endif |
188 | ||
79072805 | 189 | #include "keywords.h" |
fe14fcc3 | 190 | |
ffb4593c NT |
191 | /* CLINE is a macro that ensures PL_copline has a sane value */ |
192 | ||
ae986130 LW |
193 | #ifdef CLINE |
194 | #undef CLINE | |
195 | #endif | |
57843af0 | 196 | #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) |
3280af22 | 197 | |
5db06880 | 198 | #ifdef PERL_MAD |
29595ff2 NC |
199 | # define SKIPSPACE0(s) skipspace0(s) |
200 | # define SKIPSPACE1(s) skipspace1(s) | |
201 | # define SKIPSPACE2(s,tsv) skipspace2(s,&tsv) | |
202 | # define PEEKSPACE(s) skipspace2(s,0) | |
203 | #else | |
204 | # define SKIPSPACE0(s) skipspace(s) | |
205 | # define SKIPSPACE1(s) skipspace(s) | |
206 | # define SKIPSPACE2(s,tsv) skipspace(s) | |
207 | # define PEEKSPACE(s) skipspace(s) | |
208 | #endif | |
209 | ||
ffb4593c NT |
210 | /* |
211 | * Convenience functions to return different tokens and prime the | |
9cbb5ea2 | 212 | * lexer for the next token. They all take an argument. |
ffb4593c NT |
213 | * |
214 | * TOKEN : generic token (used for '(', DOLSHARP, etc) | |
215 | * OPERATOR : generic operator | |
216 | * AOPERATOR : assignment operator | |
217 | * PREBLOCK : beginning the block after an if, while, foreach, ... | |
218 | * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref) | |
219 | * PREREF : *EXPR where EXPR is not a simple identifier | |
220 | * TERM : expression term | |
221 | * LOOPX : loop exiting command (goto, last, dump, etc) | |
222 | * FTST : file test operator | |
223 | * FUN0 : zero-argument function | |
7eb971ee | 224 | * FUN0OP : zero-argument function, with its op created in this file |
2d2e263d | 225 | * FUN1 : not used, except for not, which isn't a UNIOP |
ffb4593c NT |
226 | * BOop : bitwise or or xor |
227 | * BAop : bitwise and | |
228 | * SHop : shift operator | |
229 | * PWop : power operator | |
9cbb5ea2 | 230 | * PMop : pattern-matching operator |
ffb4593c NT |
231 | * Aop : addition-level operator |
232 | * Mop : multiplication-level operator | |
233 | * Eop : equality-testing operator | |
e5edeb50 | 234 | * Rop : relational operator <= != gt |
ffb4593c NT |
235 | * |
236 | * Also see LOP and lop() below. | |
237 | */ | |
238 | ||
998054bd | 239 | #ifdef DEBUGGING /* Serve -DT. */ |
704d4215 | 240 | # define REPORT(retval) tokereport((I32)retval, &pl_yylval) |
998054bd | 241 | #else |
bbf60fe6 | 242 | # define REPORT(retval) (retval) |
998054bd SC |
243 | #endif |
244 | ||
bbf60fe6 DM |
245 | #define TOKEN(retval) return ( PL_bufptr = s, REPORT(retval)) |
246 | #define OPERATOR(retval) return (PL_expect = XTERM, PL_bufptr = s, REPORT(retval)) | |
247 | #define AOPERATOR(retval) return ao((PL_expect = XTERM, PL_bufptr = s, REPORT(retval))) | |
248 | #define PREBLOCK(retval) return (PL_expect = XBLOCK,PL_bufptr = s, REPORT(retval)) | |
249 | #define PRETERMBLOCK(retval) return (PL_expect = XTERMBLOCK,PL_bufptr = s, REPORT(retval)) | |
250 | #define PREREF(retval) return (PL_expect = XREF,PL_bufptr = s, REPORT(retval)) | |
251 | #define TERM(retval) return (CLINE, PL_expect = XOPERATOR, PL_bufptr = s, REPORT(retval)) | |
6154021b RGS |
252 | #define LOOPX(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)LOOPEX)) |
253 | #define FTST(f) return (pl_yylval.ival=f, PL_expect=XTERMORDORDOR, PL_bufptr=s, REPORT((int)UNIOP)) | |
254 | #define FUN0(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0)) | |
7eb971ee | 255 | #define FUN0OP(f) return (pl_yylval.opval=f, CLINE, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC0OP)) |
6154021b RGS |
256 | #define FUN1(f) return (pl_yylval.ival=f, PL_expect=XOPERATOR, PL_bufptr=s, REPORT((int)FUNC1)) |
257 | #define BOop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITOROP))) | |
258 | #define BAop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)BITANDOP))) | |
259 | #define SHop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)SHIFTOP))) | |
260 | #define PWop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)POWOP))) | |
261 | #define PMop(f) return(pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MATCHOP)) | |
262 | #define Aop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)ADDOP))) | |
263 | #define Mop(f) return ao((pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)MULOP))) | |
264 | #define Eop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)EQOP)) | |
265 | #define Rop(f) return (pl_yylval.ival=f, PL_expect=XTERM, PL_bufptr=s, REPORT((int)RELOP)) | |
2f3197b3 | 266 | |
a687059c LW |
267 | /* This bit of chicanery makes a unary function followed by |
268 | * a parenthesis into a function with one argument, highest precedence. | |
6f33ba73 RGS |
269 | * The UNIDOR macro is for unary functions that can be followed by the // |
270 | * operator (such as C<shift // 0>). | |
a687059c | 271 | */ |
d68ce4ac | 272 | #define UNI3(f,x,have_x) { \ |
6154021b | 273 | pl_yylval.ival = f; \ |
d68ce4ac | 274 | if (have_x) PL_expect = x; \ |
376fcdbf AL |
275 | PL_bufptr = s; \ |
276 | PL_last_uni = PL_oldbufptr; \ | |
277 | PL_last_lop_op = f; \ | |
278 | if (*s == '(') \ | |
279 | return REPORT( (int)FUNC1 ); \ | |
29595ff2 | 280 | s = PEEKSPACE(s); \ |
376fcdbf AL |
281 | return REPORT( *s=='(' ? (int)FUNC1 : (int)UNIOP ); \ |
282 | } | |
d68ce4ac FC |
283 | #define UNI(f) UNI3(f,XTERM,1) |
284 | #define UNIDOR(f) UNI3(f,XTERMORDORDOR,1) | |
b5fb7ce3 FC |
285 | #define UNIPROTO(f,optional) { \ |
286 | if (optional) PL_last_uni = PL_oldbufptr; \ | |
22393538 MH |
287 | OPERATOR(f); \ |
288 | } | |
a687059c | 289 | |
d68ce4ac | 290 | #define UNIBRACK(f) UNI3(f,0,0) |
79072805 | 291 | |
9f68db38 | 292 | /* grandfather return to old style */ |
78cdf107 Z |
293 | #define OLDLOP(f) \ |
294 | do { \ | |
295 | if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) \ | |
296 | PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; \ | |
297 | pl_yylval.ival = (f); \ | |
298 | PL_expect = XTERM; \ | |
299 | PL_bufptr = s; \ | |
300 | return (int)LSTOP; \ | |
301 | } while(0) | |
79072805 | 302 | |
83944c01 FC |
303 | #define COPLINE_INC_WITH_HERELINES \ |
304 | STMT_START { \ | |
305 | CopLINE_inc(PL_curcop); \ | |
d794b522 FC |
306 | if (PL_parser->lex_shared->herelines) \ |
307 | CopLINE(PL_curcop) += PL_parser->lex_shared->herelines, \ | |
308 | PL_parser->lex_shared->herelines = 0; \ | |
83944c01 FC |
309 | } STMT_END |
310 | ||
311 | ||
8fa7f367 JH |
312 | #ifdef DEBUGGING |
313 | ||
6154021b | 314 | /* how to interpret the pl_yylval associated with the token */ |
bbf60fe6 DM |
315 | enum token_type { |
316 | TOKENTYPE_NONE, | |
317 | TOKENTYPE_IVAL, | |
6154021b | 318 | TOKENTYPE_OPNUM, /* pl_yylval.ival contains an opcode number */ |
bbf60fe6 | 319 | TOKENTYPE_PVAL, |
aeaef349 | 320 | TOKENTYPE_OPVAL |
bbf60fe6 DM |
321 | }; |
322 | ||
6d4a66ac NC |
323 | static struct debug_tokens { |
324 | const int token; | |
325 | enum token_type type; | |
326 | const char *name; | |
327 | } const debug_tokens[] = | |
9041c2e3 | 328 | { |
bbf60fe6 DM |
329 | { ADDOP, TOKENTYPE_OPNUM, "ADDOP" }, |
330 | { ANDAND, TOKENTYPE_NONE, "ANDAND" }, | |
331 | { ANDOP, TOKENTYPE_NONE, "ANDOP" }, | |
332 | { ANONSUB, TOKENTYPE_IVAL, "ANONSUB" }, | |
333 | { ARROW, TOKENTYPE_NONE, "ARROW" }, | |
334 | { ASSIGNOP, TOKENTYPE_OPNUM, "ASSIGNOP" }, | |
335 | { BITANDOP, TOKENTYPE_OPNUM, "BITANDOP" }, | |
336 | { BITOROP, TOKENTYPE_OPNUM, "BITOROP" }, | |
337 | { COLONATTR, TOKENTYPE_NONE, "COLONATTR" }, | |
338 | { CONTINUE, TOKENTYPE_NONE, "CONTINUE" }, | |
0d863452 | 339 | { DEFAULT, TOKENTYPE_NONE, "DEFAULT" }, |
bbf60fe6 DM |
340 | { DO, TOKENTYPE_NONE, "DO" }, |
341 | { DOLSHARP, TOKENTYPE_NONE, "DOLSHARP" }, | |
342 | { DORDOR, TOKENTYPE_NONE, "DORDOR" }, | |
343 | { DOROP, TOKENTYPE_OPNUM, "DOROP" }, | |
344 | { DOTDOT, TOKENTYPE_IVAL, "DOTDOT" }, | |
345 | { ELSE, TOKENTYPE_NONE, "ELSE" }, | |
346 | { ELSIF, TOKENTYPE_IVAL, "ELSIF" }, | |
347 | { EQOP, TOKENTYPE_OPNUM, "EQOP" }, | |
348 | { FOR, TOKENTYPE_IVAL, "FOR" }, | |
349 | { FORMAT, TOKENTYPE_NONE, "FORMAT" }, | |
705fe0e5 FC |
350 | { FORMLBRACK, TOKENTYPE_NONE, "FORMLBRACK" }, |
351 | { FORMRBRACK, TOKENTYPE_NONE, "FORMRBRACK" }, | |
bbf60fe6 DM |
352 | { FUNC, TOKENTYPE_OPNUM, "FUNC" }, |
353 | { FUNC0, TOKENTYPE_OPNUM, "FUNC0" }, | |
7eb971ee | 354 | { FUNC0OP, TOKENTYPE_OPVAL, "FUNC0OP" }, |
bbf60fe6 DM |
355 | { FUNC0SUB, TOKENTYPE_OPVAL, "FUNC0SUB" }, |
356 | { FUNC1, TOKENTYPE_OPNUM, "FUNC1" }, | |
357 | { FUNCMETH, TOKENTYPE_OPVAL, "FUNCMETH" }, | |
0d863452 | 358 | { GIVEN, TOKENTYPE_IVAL, "GIVEN" }, |
bbf60fe6 DM |
359 | { HASHBRACK, TOKENTYPE_NONE, "HASHBRACK" }, |
360 | { IF, TOKENTYPE_IVAL, "IF" }, | |
5504e6cf | 361 | { LABEL, TOKENTYPE_PVAL, "LABEL" }, |
bbf60fe6 DM |
362 | { LOCAL, TOKENTYPE_IVAL, "LOCAL" }, |
363 | { LOOPEX, TOKENTYPE_OPNUM, "LOOPEX" }, | |
364 | { LSTOP, TOKENTYPE_OPNUM, "LSTOP" }, | |
365 | { LSTOPSUB, TOKENTYPE_OPVAL, "LSTOPSUB" }, | |
366 | { MATCHOP, TOKENTYPE_OPNUM, "MATCHOP" }, | |
367 | { METHOD, TOKENTYPE_OPVAL, "METHOD" }, | |
368 | { MULOP, TOKENTYPE_OPNUM, "MULOP" }, | |
369 | { MY, TOKENTYPE_IVAL, "MY" }, | |
bbf60fe6 DM |
370 | { NOAMP, TOKENTYPE_NONE, "NOAMP" }, |
371 | { NOTOP, TOKENTYPE_NONE, "NOTOP" }, | |
372 | { OROP, TOKENTYPE_IVAL, "OROP" }, | |
373 | { OROR, TOKENTYPE_NONE, "OROR" }, | |
374 | { PACKAGE, TOKENTYPE_NONE, "PACKAGE" }, | |
f3f204dc | 375 | { PEG, TOKENTYPE_NONE, "PEG" }, |
88e1f1a2 JV |
376 | { PLUGEXPR, TOKENTYPE_OPVAL, "PLUGEXPR" }, |
377 | { PLUGSTMT, TOKENTYPE_OPVAL, "PLUGSTMT" }, | |
bbf60fe6 DM |
378 | { PMFUNC, TOKENTYPE_OPVAL, "PMFUNC" }, |
379 | { POSTDEC, TOKENTYPE_NONE, "POSTDEC" }, | |
380 | { POSTINC, TOKENTYPE_NONE, "POSTINC" }, | |
381 | { POWOP, TOKENTYPE_OPNUM, "POWOP" }, | |
382 | { PREDEC, TOKENTYPE_NONE, "PREDEC" }, | |
383 | { PREINC, TOKENTYPE_NONE, "PREINC" }, | |
384 | { PRIVATEREF, TOKENTYPE_OPVAL, "PRIVATEREF" }, | |
f3f204dc | 385 | { QWLIST, TOKENTYPE_OPVAL, "QWLIST" }, |
bbf60fe6 DM |
386 | { REFGEN, TOKENTYPE_NONE, "REFGEN" }, |
387 | { RELOP, TOKENTYPE_OPNUM, "RELOP" }, | |
f3f204dc | 388 | { REQUIRE, TOKENTYPE_NONE, "REQUIRE" }, |
bbf60fe6 DM |
389 | { SHIFTOP, TOKENTYPE_OPNUM, "SHIFTOP" }, |
390 | { SUB, TOKENTYPE_NONE, "SUB" }, | |
391 | { THING, TOKENTYPE_OPVAL, "THING" }, | |
392 | { UMINUS, TOKENTYPE_NONE, "UMINUS" }, | |
393 | { UNIOP, TOKENTYPE_OPNUM, "UNIOP" }, | |
394 | { UNIOPSUB, TOKENTYPE_OPVAL, "UNIOPSUB" }, | |
395 | { UNLESS, TOKENTYPE_IVAL, "UNLESS" }, | |
396 | { UNTIL, TOKENTYPE_IVAL, "UNTIL" }, | |
397 | { USE, TOKENTYPE_IVAL, "USE" }, | |
0d863452 | 398 | { WHEN, TOKENTYPE_IVAL, "WHEN" }, |
bbf60fe6 DM |
399 | { WHILE, TOKENTYPE_IVAL, "WHILE" }, |
400 | { WORD, TOKENTYPE_OPVAL, "WORD" }, | |
be25f609 | 401 | { YADAYADA, TOKENTYPE_IVAL, "YADAYADA" }, |
c35e046a | 402 | { 0, TOKENTYPE_NONE, NULL } |
bbf60fe6 DM |
403 | }; |
404 | ||
6154021b | 405 | /* dump the returned token in rv, plus any optional arg in pl_yylval */ |
998054bd | 406 | |
bbf60fe6 | 407 | STATIC int |
704d4215 | 408 | S_tokereport(pTHX_ I32 rv, const YYSTYPE* lvalp) |
bbf60fe6 | 409 | { |
97aff369 | 410 | dVAR; |
7918f24d NC |
411 | |
412 | PERL_ARGS_ASSERT_TOKEREPORT; | |
413 | ||
bbf60fe6 | 414 | if (DEBUG_T_TEST) { |
bd61b366 | 415 | const char *name = NULL; |
bbf60fe6 | 416 | enum token_type type = TOKENTYPE_NONE; |
f54cb97a | 417 | const struct debug_tokens *p; |
396482e1 | 418 | SV* const report = newSVpvs("<== "); |
bbf60fe6 | 419 | |
f54cb97a | 420 | for (p = debug_tokens; p->token; p++) { |
bbf60fe6 DM |
421 | if (p->token == (int)rv) { |
422 | name = p->name; | |
423 | type = p->type; | |
424 | break; | |
425 | } | |
426 | } | |
427 | if (name) | |
54667de8 | 428 | Perl_sv_catpv(aTHX_ report, name); |
74736ae6 | 429 | else if ((char)rv > ' ' && (char)rv <= '~') |
4ebc7986 | 430 | { |
bbf60fe6 | 431 | Perl_sv_catpvf(aTHX_ report, "'%c'", (char)rv); |
4ebc7986 FC |
432 | if ((char)rv == 'p') |
433 | sv_catpvs(report, " (pending identifier)"); | |
434 | } | |
bbf60fe6 | 435 | else if (!rv) |
396482e1 | 436 | sv_catpvs(report, "EOF"); |
bbf60fe6 DM |
437 | else |
438 | Perl_sv_catpvf(aTHX_ report, "?? %"IVdf, (IV)rv); | |
439 | switch (type) { | |
440 | case TOKENTYPE_NONE: | |
bbf60fe6 DM |
441 | break; |
442 | case TOKENTYPE_IVAL: | |
704d4215 | 443 | Perl_sv_catpvf(aTHX_ report, "(ival=%"IVdf")", (IV)lvalp->ival); |
bbf60fe6 DM |
444 | break; |
445 | case TOKENTYPE_OPNUM: | |
446 | Perl_sv_catpvf(aTHX_ report, "(ival=op_%s)", | |
704d4215 | 447 | PL_op_name[lvalp->ival]); |
bbf60fe6 DM |
448 | break; |
449 | case TOKENTYPE_PVAL: | |
704d4215 | 450 | Perl_sv_catpvf(aTHX_ report, "(pval=\"%s\")", lvalp->pval); |
bbf60fe6 DM |
451 | break; |
452 | case TOKENTYPE_OPVAL: | |
704d4215 | 453 | if (lvalp->opval) { |
401441c0 | 454 | Perl_sv_catpvf(aTHX_ report, "(opval=op_%s)", |
704d4215 GG |
455 | PL_op_name[lvalp->opval->op_type]); |
456 | if (lvalp->opval->op_type == OP_CONST) { | |
b6007c36 | 457 | Perl_sv_catpvf(aTHX_ report, " %s", |
704d4215 | 458 | SvPEEK(cSVOPx_sv(lvalp->opval))); |
b6007c36 DM |
459 | } |
460 | ||
461 | } | |
401441c0 | 462 | else |
396482e1 | 463 | sv_catpvs(report, "(opval=null)"); |
bbf60fe6 DM |
464 | break; |
465 | } | |
b6007c36 | 466 | PerlIO_printf(Perl_debug_log, "### %s\n\n", SvPV_nolen_const(report)); |
bbf60fe6 DM |
467 | }; |
468 | return (int)rv; | |
998054bd SC |
469 | } |
470 | ||
b6007c36 DM |
471 | |
472 | /* print the buffer with suitable escapes */ | |
473 | ||
474 | STATIC void | |
15f169a1 | 475 | S_printbuf(pTHX_ const char *const fmt, const char *const s) |
b6007c36 | 476 | { |
396482e1 | 477 | SV* const tmp = newSVpvs(""); |
7918f24d NC |
478 | |
479 | PERL_ARGS_ASSERT_PRINTBUF; | |
480 | ||
b6007c36 DM |
481 | PerlIO_printf(Perl_debug_log, fmt, pv_display(tmp, s, strlen(s), 0, 60)); |
482 | SvREFCNT_dec(tmp); | |
483 | } | |
484 | ||
8fa7f367 JH |
485 | #endif |
486 | ||
8290c323 NC |
487 | static int |
488 | S_deprecate_commaless_var_list(pTHX) { | |
489 | PL_expect = XTERM; | |
490 | deprecate("comma-less variable list"); | |
491 | return REPORT(','); /* grandfather non-comma-format format */ | |
492 | } | |
493 | ||
ffb4593c NT |
494 | /* |
495 | * S_ao | |
496 | * | |
c963b151 BD |
497 | * This subroutine detects &&=, ||=, and //= and turns an ANDAND, OROR or DORDOR |
498 | * into an OP_ANDASSIGN, OP_ORASSIGN, or OP_DORASSIGN | |
ffb4593c NT |
499 | */ |
500 | ||
76e3520e | 501 | STATIC int |
cea2e8a9 | 502 | S_ao(pTHX_ int toketype) |
a0d0e21e | 503 | { |
97aff369 | 504 | dVAR; |
3280af22 NIS |
505 | if (*PL_bufptr == '=') { |
506 | PL_bufptr++; | |
a0d0e21e | 507 | if (toketype == ANDAND) |
6154021b | 508 | pl_yylval.ival = OP_ANDASSIGN; |
a0d0e21e | 509 | else if (toketype == OROR) |
6154021b | 510 | pl_yylval.ival = OP_ORASSIGN; |
c963b151 | 511 | else if (toketype == DORDOR) |
6154021b | 512 | pl_yylval.ival = OP_DORASSIGN; |
a0d0e21e LW |
513 | toketype = ASSIGNOP; |
514 | } | |
515 | return toketype; | |
516 | } | |
517 | ||
ffb4593c NT |
518 | /* |
519 | * S_no_op | |
520 | * When Perl expects an operator and finds something else, no_op | |
521 | * prints the warning. It always prints "<something> found where | |
522 | * operator expected. It prints "Missing semicolon on previous line?" | |
523 | * if the surprise occurs at the start of the line. "do you need to | |
524 | * predeclare ..." is printed out for code like "sub bar; foo bar $x" | |
525 | * where the compiler doesn't know if foo is a method call or a function. | |
526 | * It prints "Missing operator before end of line" if there's nothing | |
527 | * after the missing operator, or "... before <...>" if there is something | |
528 | * after the missing operator. | |
529 | */ | |
530 | ||
76e3520e | 531 | STATIC void |
15f169a1 | 532 | S_no_op(pTHX_ const char *const what, char *s) |
463ee0b2 | 533 | { |
97aff369 | 534 | dVAR; |
9d4ba2ae AL |
535 | char * const oldbp = PL_bufptr; |
536 | const bool is_first = (PL_oldbufptr == PL_linestart); | |
68dc0745 | 537 | |
7918f24d NC |
538 | PERL_ARGS_ASSERT_NO_OP; |
539 | ||
1189a94a GS |
540 | if (!s) |
541 | s = oldbp; | |
07c798fb | 542 | else |
1189a94a | 543 | PL_bufptr = s; |
734ab321 | 544 | yywarn(Perl_form(aTHX_ "%s found where operator expected", what), UTF ? SVf_UTF8 : 0); |
56da5a46 RGS |
545 | if (ckWARN_d(WARN_SYNTAX)) { |
546 | if (is_first) | |
547 | Perl_warner(aTHX_ packWARN(WARN_SYNTAX), | |
548 | "\t(Missing semicolon on previous line?)\n"); | |
549 | else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) { | |
f54cb97a | 550 | const char *t; |
8a2bca12 | 551 | for (t = PL_oldoldbufptr; (isWORDCHAR_lazy_if(t,UTF) || *t == ':'); |
734ab321 | 552 | t += UTF ? UTF8SKIP(t) : 1) |
c35e046a | 553 | NOOP; |
56da5a46 RGS |
554 | if (t < PL_bufptr && isSPACE(*t)) |
555 | Perl_warner(aTHX_ packWARN(WARN_SYNTAX), | |
734ab321 BF |
556 | "\t(Do you need to predeclare %"SVf"?)\n", |
557 | SVfARG(newSVpvn_flags(PL_oldoldbufptr, (STRLEN)(t - PL_oldoldbufptr), | |
558 | SVs_TEMP | (UTF ? SVf_UTF8 : 0)))); | |
56da5a46 RGS |
559 | } |
560 | else { | |
561 | assert(s >= oldbp); | |
562 | Perl_warner(aTHX_ packWARN(WARN_SYNTAX), | |
734ab321 BF |
563 | "\t(Missing operator before %"SVf"?)\n", |
564 | SVfARG(newSVpvn_flags(oldbp, (STRLEN)(s - oldbp), | |
565 | SVs_TEMP | (UTF ? SVf_UTF8 : 0)))); | |
56da5a46 | 566 | } |
07c798fb | 567 | } |
3280af22 | 568 | PL_bufptr = oldbp; |
8990e307 LW |
569 | } |
570 | ||
ffb4593c NT |
571 | /* |
572 | * S_missingterm | |
573 | * Complain about missing quote/regexp/heredoc terminator. | |
d4c19fe8 | 574 | * If it's called with NULL then it cauterizes the line buffer. |
ffb4593c NT |
575 | * If we're in a delimited string and the delimiter is a control |
576 | * character, it's reformatted into a two-char sequence like ^C. | |
577 | * This is fatal. | |
578 | */ | |
579 | ||
76e3520e | 580 | STATIC void |
cea2e8a9 | 581 | S_missingterm(pTHX_ char *s) |
8990e307 | 582 | { |
97aff369 | 583 | dVAR; |
8990e307 LW |
584 | char tmpbuf[3]; |
585 | char q; | |
586 | if (s) { | |
9d4ba2ae | 587 | char * const nl = strrchr(s,'\n'); |
d2719217 | 588 | if (nl) |
8990e307 LW |
589 | *nl = '\0'; |
590 | } | |
463559e7 | 591 | else if (isCNTRL(PL_multi_close)) { |
8990e307 | 592 | *tmpbuf = '^'; |
585ec06d | 593 | tmpbuf[1] = (char)toCTRL(PL_multi_close); |
8990e307 LW |
594 | tmpbuf[2] = '\0'; |
595 | s = tmpbuf; | |
596 | } | |
597 | else { | |
eb160463 | 598 | *tmpbuf = (char)PL_multi_close; |
8990e307 LW |
599 | tmpbuf[1] = '\0'; |
600 | s = tmpbuf; | |
601 | } | |
602 | q = strchr(s,'"') ? '\'' : '"'; | |
cea2e8a9 | 603 | Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q); |
463ee0b2 | 604 | } |
79072805 | 605 | |
dd0ac2b9 FC |
606 | #include "feature.h" |
607 | ||
0d863452 | 608 | /* |
0d863452 RH |
609 | * Check whether the named feature is enabled. |
610 | */ | |
26ea9e12 | 611 | bool |
3fff3427 | 612 | Perl_feature_is_enabled(pTHX_ const char *const name, STRLEN namelen) |
0d863452 | 613 | { |
97aff369 | 614 | dVAR; |
4a731d7b | 615 | char he_name[8 + MAX_FEATURE_LEN] = "feature_"; |
7918f24d NC |
616 | |
617 | PERL_ARGS_ASSERT_FEATURE_IS_ENABLED; | |
ca4d40c4 FC |
618 | |
619 | assert(CURRENT_FEATURE_BUNDLE == FEATURE_BUNDLE_CUSTOM); | |
7918f24d | 620 | |
26ea9e12 NC |
621 | if (namelen > MAX_FEATURE_LEN) |
622 | return FALSE; | |
3fff3427 | 623 | memcpy(&he_name[8], name, namelen); |
7d69d4a6 | 624 | |
c8ca97b0 NC |
625 | return cBOOL(cop_hints_fetch_pvn(PL_curcop, he_name, 8 + namelen, 0, |
626 | REFCOUNTED_HE_EXISTS)); | |
0d863452 RH |
627 | } |
628 | ||
ffb4593c | 629 | /* |
9cbb5ea2 GS |
630 | * experimental text filters for win32 carriage-returns, utf16-to-utf8 and |
631 | * utf16-to-utf8-reversed. | |
ffb4593c NT |
632 | */ |
633 | ||
c39cd008 GS |
634 | #ifdef PERL_CR_FILTER |
635 | static void | |
636 | strip_return(SV *sv) | |
637 | { | |
eb578fdb KW |
638 | const char *s = SvPVX_const(sv); |
639 | const char * const e = s + SvCUR(sv); | |
7918f24d NC |
640 | |
641 | PERL_ARGS_ASSERT_STRIP_RETURN; | |
642 | ||
c39cd008 GS |
643 | /* outer loop optimized to do nothing if there are no CR-LFs */ |
644 | while (s < e) { | |
645 | if (*s++ == '\r' && *s == '\n') { | |
646 | /* hit a CR-LF, need to copy the rest */ | |
eb578fdb | 647 | char *d = s - 1; |
c39cd008 GS |
648 | *d++ = *s++; |
649 | while (s < e) { | |
650 | if (*s == '\r' && s[1] == '\n') | |
651 | s++; | |
652 | *d++ = *s++; | |
653 | } | |
654 | SvCUR(sv) -= s - d; | |
655 | return; | |
656 | } | |
657 | } | |
658 | } | |
a868473f | 659 | |
76e3520e | 660 | STATIC I32 |
c39cd008 | 661 | S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
a868473f | 662 | { |
f54cb97a | 663 | const I32 count = FILTER_READ(idx+1, sv, maxlen); |
c39cd008 GS |
664 | if (count > 0 && !maxlen) |
665 | strip_return(sv); | |
666 | return count; | |
a868473f NIS |
667 | } |
668 | #endif | |
669 | ||
ffb4593c | 670 | /* |
8eaa0acf Z |
671 | =for apidoc Amx|void|lex_start|SV *line|PerlIO *rsfp|U32 flags |
672 | ||
673 | Creates and initialises a new lexer/parser state object, supplying | |
674 | a context in which to lex and parse from a new source of Perl code. | |
675 | A pointer to the new state object is placed in L</PL_parser>. An entry | |
676 | is made on the save stack so that upon unwinding the new state object | |
677 | will be destroyed and the former value of L</PL_parser> will be restored. | |
678 | Nothing else need be done to clean up the parsing context. | |
679 | ||
680 | The code to be parsed comes from I<line> and I<rsfp>. I<line>, if | |
681 | non-null, provides a string (in SV form) containing code to be parsed. | |
682 | A copy of the string is made, so subsequent modification of I<line> | |
683 | does not affect parsing. I<rsfp>, if non-null, provides an input stream | |
684 | from which code will be read to be parsed. If both are non-null, the | |
685 | code in I<line> comes first and must consist of complete lines of input, | |
686 | and I<rsfp> supplies the remainder of the source. | |
687 | ||
e368b3bd FC |
688 | The I<flags> parameter is reserved for future use. Currently it is only |
689 | used by perl internally, so extensions should always pass zero. | |
8eaa0acf Z |
690 | |
691 | =cut | |
692 | */ | |
ffb4593c | 693 | |
27fcb6ee | 694 | /* LEX_START_SAME_FILTER indicates that this is not a new file, so it |
87606032 NC |
695 | can share filters with the current parser. |
696 | LEX_START_DONT_CLOSE indicates that the file handle wasn't opened by the | |
697 | caller, hence isn't owned by the parser, so shouldn't be closed on parser | |
698 | destruction. This is used to handle the case of defaulting to reading the | |
699 | script from the standard input because no filename was given on the command | |
700 | line (without getting confused by situation where STDIN has been closed, so | |
701 | the script handle is opened on fd 0) */ | |
27fcb6ee | 702 | |
a0d0e21e | 703 | void |
8eaa0acf | 704 | Perl_lex_start(pTHX_ SV *line, PerlIO *rsfp, U32 flags) |
79072805 | 705 | { |
97aff369 | 706 | dVAR; |
6ef55633 | 707 | const char *s = NULL; |
5486870f | 708 | yy_parser *parser, *oparser; |
60d63348 | 709 | if (flags && flags & ~LEX_START_FLAGS) |
8eaa0acf | 710 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_start"); |
acdf0a21 DM |
711 | |
712 | /* create and initialise a parser */ | |
713 | ||
199e78b7 | 714 | Newxz(parser, 1, yy_parser); |
5486870f | 715 | parser->old_parser = oparser = PL_parser; |
acdf0a21 DM |
716 | PL_parser = parser; |
717 | ||
28ac2b49 Z |
718 | parser->stack = NULL; |
719 | parser->ps = NULL; | |
720 | parser->stack_size = 0; | |
acdf0a21 | 721 | |
e3abe207 DM |
722 | /* on scope exit, free this parser and restore any outer one */ |
723 | SAVEPARSER(parser); | |
7c4baf47 | 724 | parser->saved_curcop = PL_curcop; |
e3abe207 | 725 | |
acdf0a21 | 726 | /* initialise lexer state */ |
8990e307 | 727 | |
fb205e7a DM |
728 | #ifdef PERL_MAD |
729 | parser->curforce = -1; | |
730 | #else | |
731 | parser->nexttoke = 0; | |
732 | #endif | |
ca4cfd28 | 733 | parser->error_count = oparser ? oparser->error_count : 0; |
c2598295 | 734 | parser->copline = NOLINE; |
5afb0a62 | 735 | parser->lex_state = LEX_NORMAL; |
c2598295 | 736 | parser->expect = XSTATE; |
2f9285f8 | 737 | parser->rsfp = rsfp; |
27fcb6ee FC |
738 | parser->rsfp_filters = |
739 | !(flags & LEX_START_SAME_FILTER) || !oparser | |
d3cd8e11 FC |
740 | ? NULL |
741 | : MUTABLE_AV(SvREFCNT_inc( | |
742 | oparser->rsfp_filters | |
743 | ? oparser->rsfp_filters | |
744 | : (oparser->rsfp_filters = newAV()) | |
745 | )); | |
2f9285f8 | 746 | |
199e78b7 DM |
747 | Newx(parser->lex_brackstack, 120, char); |
748 | Newx(parser->lex_casestack, 12, char); | |
749 | *parser->lex_casestack = '\0'; | |
d794b522 | 750 | Newxz(parser->lex_shared, 1, LEXSHARED); |
02b34bbe | 751 | |
10efb74f | 752 | if (line) { |
0528fd32 | 753 | STRLEN len; |
10efb74f | 754 | s = SvPV_const(line, len); |
0abcdfa4 FC |
755 | parser->linestr = flags & LEX_START_COPIED |
756 | ? SvREFCNT_inc_simple_NN(line) | |
757 | : newSVpvn_flags(s, len, SvUTF8(line)); | |
11076590 | 758 | sv_catpvs(parser->linestr, "\n;"); |
0abcdfa4 FC |
759 | } else { |
760 | parser->linestr = newSVpvs("\n;"); | |
8990e307 | 761 | } |
f06b5848 DM |
762 | parser->oldoldbufptr = |
763 | parser->oldbufptr = | |
764 | parser->bufptr = | |
765 | parser->linestart = SvPVX(parser->linestr); | |
766 | parser->bufend = parser->bufptr + SvCUR(parser->linestr); | |
767 | parser->last_lop = parser->last_uni = NULL; | |
87606032 NC |
768 | parser->lex_flags = flags & (LEX_IGNORE_UTF8_HINTS|LEX_EVALBYTES |
769 | |LEX_DONT_CLOSE_RSFP); | |
737c24fc | 770 | |
60d63348 | 771 | parser->in_pod = parser->filtered = 0; |
79072805 | 772 | } |
a687059c | 773 | |
e3abe207 DM |
774 | |
775 | /* delete a parser object */ | |
776 | ||
777 | void | |
778 | Perl_parser_free(pTHX_ const yy_parser *parser) | |
779 | { | |
7918f24d NC |
780 | PERL_ARGS_ASSERT_PARSER_FREE; |
781 | ||
7c4baf47 | 782 | PL_curcop = parser->saved_curcop; |
bdc0bf6f DM |
783 | SvREFCNT_dec(parser->linestr); |
784 | ||
87606032 | 785 | if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP) |
2f9285f8 | 786 | PerlIO_clearerr(parser->rsfp); |
799361c3 SH |
787 | else if (parser->rsfp && (!parser->old_parser || |
788 | (parser->old_parser && parser->rsfp != parser->old_parser->rsfp))) | |
2f9285f8 | 789 | PerlIO_close(parser->rsfp); |
5486870f | 790 | SvREFCNT_dec(parser->rsfp_filters); |
10002bc1 FC |
791 | SvREFCNT_dec(parser->lex_stuff); |
792 | SvREFCNT_dec(parser->sublex_info.repl); | |
3ac7ff8f FC |
793 | |
794 | Safefree(parser->lex_brackstack); | |
795 | Safefree(parser->lex_casestack); | |
796 | Safefree(parser->lex_shared); | |
797 | PL_parser = parser->old_parser; | |
798 | Safefree(parser); | |
799 | } | |
800 | ||
801 | void | |
802 | Perl_parser_free_nexttoke_ops(pTHX_ yy_parser *parser, OPSLAB *slab) | |
803 | { | |
804 | #ifdef PERL_MAD | |
805 | I32 nexttoke = parser->lasttoke; | |
806 | #else | |
807 | I32 nexttoke = parser->nexttoke; | |
808 | #endif | |
809 | PERL_ARGS_ASSERT_PARSER_FREE_NEXTTOKE_OPS; | |
3ce3dcd9 FC |
810 | while (nexttoke--) { |
811 | #ifdef PERL_MAD | |
812 | if (S_is_opval_token(parser->nexttoke[nexttoke].next_type | |
3ac7ff8f FC |
813 | & 0xffff) |
814 | && parser->nexttoke[nexttoke].next_val.opval | |
815 | && parser->nexttoke[nexttoke].next_val.opval->op_slabbed | |
816 | && OpSLAB(parser->nexttoke[nexttoke].next_val.opval) == slab) { | |
817 | op_free(parser->nexttoke[nexttoke].next_val.opval); | |
818 | parser->nexttoke[nexttoke].next_val.opval = NULL; | |
819 | } | |
3ce3dcd9 | 820 | #else |
3ac7ff8f FC |
821 | if (S_is_opval_token(parser->nexttype[nexttoke] & 0xffff) |
822 | && parser->nextval[nexttoke].opval | |
823 | && parser->nextval[nexttoke].opval->op_slabbed | |
824 | && OpSLAB(parser->nextval[nexttoke].opval) == slab) { | |
3ce3dcd9 | 825 | op_free(parser->nextval[nexttoke].opval); |
3ac7ff8f FC |
826 | parser->nextval[nexttoke].opval = NULL; |
827 | } | |
3ce3dcd9 FC |
828 | #endif |
829 | } | |
e3abe207 DM |
830 | } |
831 | ||
832 | ||
ffb4593c | 833 | /* |
f0e67a1d Z |
834 | =for apidoc AmxU|SV *|PL_parser-E<gt>linestr |
835 | ||
836 | Buffer scalar containing the chunk currently under consideration of the | |
837 | text currently being lexed. This is always a plain string scalar (for | |
838 | which C<SvPOK> is true). It is not intended to be used as a scalar by | |
839 | normal scalar means; instead refer to the buffer directly by the pointer | |
840 | variables described below. | |
841 | ||
842 | The lexer maintains various C<char*> pointers to things in the | |
843 | C<PL_parser-E<gt>linestr> buffer. If C<PL_parser-E<gt>linestr> is ever | |
844 | reallocated, all of these pointers must be updated. Don't attempt to | |
845 | do this manually, but rather use L</lex_grow_linestr> if you need to | |
846 | reallocate the buffer. | |
847 | ||
848 | The content of the text chunk in the buffer is commonly exactly one | |
849 | complete line of input, up to and including a newline terminator, | |
850 | but there are situations where it is otherwise. The octets of the | |
851 | buffer may be intended to be interpreted as either UTF-8 or Latin-1. | |
852 | The function L</lex_bufutf8> tells you which. Do not use the C<SvUTF8> | |
853 | flag on this scalar, which may disagree with it. | |
854 | ||
855 | For direct examination of the buffer, the variable | |
856 | L</PL_parser-E<gt>bufend> points to the end of the buffer. The current | |
857 | lexing position is pointed to by L</PL_parser-E<gt>bufptr>. Direct use | |
858 | of these pointers is usually preferable to examination of the scalar | |
859 | through normal scalar means. | |
860 | ||
861 | =for apidoc AmxU|char *|PL_parser-E<gt>bufend | |
862 | ||
863 | Direct pointer to the end of the chunk of text currently being lexed, the | |
864 | end of the lexer buffer. This is equal to C<SvPVX(PL_parser-E<gt>linestr) | |
865 | + SvCUR(PL_parser-E<gt>linestr)>. A NUL character (zero octet) is | |
866 | always located at the end of the buffer, and does not count as part of | |
867 | the buffer's contents. | |
868 | ||
869 | =for apidoc AmxU|char *|PL_parser-E<gt>bufptr | |
870 | ||
871 | Points to the current position of lexing inside the lexer buffer. | |
872 | Characters around this point may be freely examined, within | |
873 | the range delimited by C<SvPVX(L</PL_parser-E<gt>linestr>)> and | |
874 | L</PL_parser-E<gt>bufend>. The octets of the buffer may be intended to be | |
875 | interpreted as either UTF-8 or Latin-1, as indicated by L</lex_bufutf8>. | |
876 | ||
877 | Lexing code (whether in the Perl core or not) moves this pointer past | |
878 | the characters that it consumes. It is also expected to perform some | |
879 | bookkeeping whenever a newline character is consumed. This movement | |
880 | can be more conveniently performed by the function L</lex_read_to>, | |
881 | which handles newlines appropriately. | |
882 | ||
883 | Interpretation of the buffer's octets can be abstracted out by | |
884 | using the slightly higher-level functions L</lex_peek_unichar> and | |
885 | L</lex_read_unichar>. | |
886 | ||
887 | =for apidoc AmxU|char *|PL_parser-E<gt>linestart | |
888 | ||
889 | Points to the start of the current line inside the lexer buffer. | |
890 | This is useful for indicating at which column an error occurred, and | |
891 | not much else. This must be updated by any lexing code that consumes | |
892 | a newline; the function L</lex_read_to> handles this detail. | |
893 | ||
894 | =cut | |
895 | */ | |
896 | ||
897 | /* | |
898 | =for apidoc Amx|bool|lex_bufutf8 | |
899 | ||
900 | Indicates whether the octets in the lexer buffer | |
901 | (L</PL_parser-E<gt>linestr>) should be interpreted as the UTF-8 encoding | |
902 | of Unicode characters. If not, they should be interpreted as Latin-1 | |
903 | characters. This is analogous to the C<SvUTF8> flag for scalars. | |
904 | ||
905 | In UTF-8 mode, it is not guaranteed that the lexer buffer actually | |
906 | contains valid UTF-8. Lexing code must be robust in the face of invalid | |
907 | encoding. | |
908 | ||
909 | The actual C<SvUTF8> flag of the L</PL_parser-E<gt>linestr> scalar | |
910 | is significant, but not the whole story regarding the input character | |
911 | encoding. Normally, when a file is being read, the scalar contains octets | |
912 | and its C<SvUTF8> flag is off, but the octets should be interpreted as | |
913 | UTF-8 if the C<use utf8> pragma is in effect. During a string eval, | |
914 | however, the scalar may have the C<SvUTF8> flag on, and in this case its | |
915 | octets should be interpreted as UTF-8 unless the C<use bytes> pragma | |
916 | is in effect. This logic may change in the future; use this function | |
917 | instead of implementing the logic yourself. | |
918 | ||
919 | =cut | |
920 | */ | |
921 | ||
922 | bool | |
923 | Perl_lex_bufutf8(pTHX) | |
924 | { | |
925 | return UTF; | |
926 | } | |
927 | ||
928 | /* | |
929 | =for apidoc Amx|char *|lex_grow_linestr|STRLEN len | |
930 | ||
931 | Reallocates the lexer buffer (L</PL_parser-E<gt>linestr>) to accommodate | |
932 | at least I<len> octets (including terminating NUL). Returns a | |
933 | pointer to the reallocated buffer. This is necessary before making | |
934 | any direct modification of the buffer that would increase its length. | |
935 | L</lex_stuff_pvn> provides a more convenient way to insert text into | |
936 | the buffer. | |
937 | ||
938 | Do not use C<SvGROW> or C<sv_grow> directly on C<PL_parser-E<gt>linestr>; | |
939 | this function updates all of the lexer's variables that point directly | |
940 | into the buffer. | |
941 | ||
942 | =cut | |
943 | */ | |
944 | ||
945 | char * | |
946 | Perl_lex_grow_linestr(pTHX_ STRLEN len) | |
947 | { | |
948 | SV *linestr; | |
949 | char *buf; | |
950 | STRLEN bufend_pos, bufptr_pos, oldbufptr_pos, oldoldbufptr_pos; | |
c7641931 | 951 | STRLEN linestart_pos, last_uni_pos, last_lop_pos, re_eval_start_pos; |
f0e67a1d Z |
952 | linestr = PL_parser->linestr; |
953 | buf = SvPVX(linestr); | |
954 | if (len <= SvLEN(linestr)) | |
955 | return buf; | |
956 | bufend_pos = PL_parser->bufend - buf; | |
957 | bufptr_pos = PL_parser->bufptr - buf; | |
958 | oldbufptr_pos = PL_parser->oldbufptr - buf; | |
959 | oldoldbufptr_pos = PL_parser->oldoldbufptr - buf; | |
960 | linestart_pos = PL_parser->linestart - buf; | |
961 | last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0; | |
962 | last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0; | |
3328ab5a FC |
963 | re_eval_start_pos = PL_parser->lex_shared->re_eval_start ? |
964 | PL_parser->lex_shared->re_eval_start - buf : 0; | |
c7641931 | 965 | |
f0e67a1d | 966 | buf = sv_grow(linestr, len); |
c7641931 | 967 | |
f0e67a1d Z |
968 | PL_parser->bufend = buf + bufend_pos; |
969 | PL_parser->bufptr = buf + bufptr_pos; | |
970 | PL_parser->oldbufptr = buf + oldbufptr_pos; | |
971 | PL_parser->oldoldbufptr = buf + oldoldbufptr_pos; | |
972 | PL_parser->linestart = buf + linestart_pos; | |
973 | if (PL_parser->last_uni) | |
974 | PL_parser->last_uni = buf + last_uni_pos; | |
975 | if (PL_parser->last_lop) | |
976 | PL_parser->last_lop = buf + last_lop_pos; | |
3328ab5a FC |
977 | if (PL_parser->lex_shared->re_eval_start) |
978 | PL_parser->lex_shared->re_eval_start = buf + re_eval_start_pos; | |
f0e67a1d Z |
979 | return buf; |
980 | } | |
981 | ||
982 | /* | |
83aa740e | 983 | =for apidoc Amx|void|lex_stuff_pvn|const char *pv|STRLEN len|U32 flags |
f0e67a1d Z |
984 | |
985 | Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>), | |
986 | immediately after the current lexing point (L</PL_parser-E<gt>bufptr>), | |
987 | reallocating the buffer if necessary. This means that lexing code that | |
988 | runs later will see the characters as if they had appeared in the input. | |
989 | It is not recommended to do this as part of normal parsing, and most | |
990 | uses of this facility run the risk of the inserted characters being | |
991 | interpreted in an unintended manner. | |
992 | ||
993 | The string to be inserted is represented by I<len> octets starting | |
994 | at I<pv>. These octets are interpreted as either UTF-8 or Latin-1, | |
995 | according to whether the C<LEX_STUFF_UTF8> flag is set in I<flags>. | |
996 | The characters are recoded for the lexer buffer, according to how the | |
997 | buffer is currently being interpreted (L</lex_bufutf8>). If a string | |
9dcc53ea | 998 | to be inserted is available as a Perl scalar, the L</lex_stuff_sv> |
f0e67a1d Z |
999 | function is more convenient. |
1000 | ||
1001 | =cut | |
1002 | */ | |
1003 | ||
1004 | void | |
83aa740e | 1005 | Perl_lex_stuff_pvn(pTHX_ const char *pv, STRLEN len, U32 flags) |
f0e67a1d | 1006 | { |
749123ff | 1007 | dVAR; |
f0e67a1d Z |
1008 | char *bufptr; |
1009 | PERL_ARGS_ASSERT_LEX_STUFF_PVN; | |
1010 | if (flags & ~(LEX_STUFF_UTF8)) | |
1011 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_pvn"); | |
1012 | if (UTF) { | |
1013 | if (flags & LEX_STUFF_UTF8) { | |
1014 | goto plain_copy; | |
1015 | } else { | |
54d004e8 | 1016 | STRLEN highhalf = 0; /* Count of variants */ |
83aa740e | 1017 | const char *p, *e = pv+len; |
54d004e8 KW |
1018 | for (p = pv; p != e; p++) { |
1019 | if (! UTF8_IS_INVARIANT(*p)) { | |
1020 | highhalf++; | |
1021 | } | |
1022 | } | |
f0e67a1d Z |
1023 | if (!highhalf) |
1024 | goto plain_copy; | |
1025 | lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len+highhalf); | |
1026 | bufptr = PL_parser->bufptr; | |
1027 | Move(bufptr, bufptr+len+highhalf, PL_parser->bufend+1-bufptr, char); | |
255fdf19 Z |
1028 | SvCUR_set(PL_parser->linestr, |
1029 | SvCUR(PL_parser->linestr) + len+highhalf); | |
f0e67a1d Z |
1030 | PL_parser->bufend += len+highhalf; |
1031 | for (p = pv; p != e; p++) { | |
1032 | U8 c = (U8)*p; | |
54d004e8 KW |
1033 | if (! UTF8_IS_INVARIANT(c)) { |
1034 | *bufptr++ = UTF8_TWO_BYTE_HI(c); | |
1035 | *bufptr++ = UTF8_TWO_BYTE_LO(c); | |
f0e67a1d Z |
1036 | } else { |
1037 | *bufptr++ = (char)c; | |
1038 | } | |
1039 | } | |
1040 | } | |
1041 | } else { | |
1042 | if (flags & LEX_STUFF_UTF8) { | |
1043 | STRLEN highhalf = 0; | |
83aa740e | 1044 | const char *p, *e = pv+len; |
f0e67a1d Z |
1045 | for (p = pv; p != e; p++) { |
1046 | U8 c = (U8)*p; | |
54d004e8 | 1047 | if (UTF8_IS_ABOVE_LATIN1(c)) { |
f0e67a1d Z |
1048 | Perl_croak(aTHX_ "Lexing code attempted to stuff " |
1049 | "non-Latin-1 character into Latin-1 input"); | |
54d004e8 | 1050 | } else if (UTF8_IS_NEXT_CHAR_DOWNGRADEABLE(p, e)) { |
f0e67a1d Z |
1051 | p++; |
1052 | highhalf++; | |
54d004e8 | 1053 | } else if (! UTF8_IS_INVARIANT(c)) { |
f0e67a1d Z |
1054 | /* malformed UTF-8 */ |
1055 | ENTER; | |
1056 | SAVESPTR(PL_warnhook); | |
1057 | PL_warnhook = PERL_WARNHOOK_FATAL; | |
1058 | utf8n_to_uvuni((U8*)p, e-p, NULL, 0); | |
1059 | LEAVE; | |
1060 | } | |
1061 | } | |
1062 | if (!highhalf) | |
1063 | goto plain_copy; | |
1064 | lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len-highhalf); | |
1065 | bufptr = PL_parser->bufptr; | |
1066 | Move(bufptr, bufptr+len-highhalf, PL_parser->bufend+1-bufptr, char); | |
255fdf19 Z |
1067 | SvCUR_set(PL_parser->linestr, |
1068 | SvCUR(PL_parser->linestr) + len-highhalf); | |
f0e67a1d | 1069 | PL_parser->bufend += len-highhalf; |
54d004e8 KW |
1070 | p = pv; |
1071 | while (p < e) { | |
1072 | if (UTF8_IS_INVARIANT(*p)) { | |
1073 | *bufptr++ = *p; | |
1074 | p++; | |
f0e67a1d | 1075 | } |
54d004e8 KW |
1076 | else { |
1077 | assert(p < e -1 ); | |
1078 | *bufptr++ = TWO_BYTE_UTF8_TO_UNI(*p, *(p+1)); | |
1079 | p += 2; | |
1080 | } | |
f0e67a1d Z |
1081 | } |
1082 | } else { | |
54d004e8 | 1083 | plain_copy: |
f0e67a1d Z |
1084 | lex_grow_linestr(SvCUR(PL_parser->linestr)+1+len); |
1085 | bufptr = PL_parser->bufptr; | |
1086 | Move(bufptr, bufptr+len, PL_parser->bufend+1-bufptr, char); | |
255fdf19 | 1087 | SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) + len); |
f0e67a1d Z |
1088 | PL_parser->bufend += len; |
1089 | Copy(pv, bufptr, len, char); | |
1090 | } | |
1091 | } | |
1092 | } | |
1093 | ||
1094 | /* | |
9dcc53ea Z |
1095 | =for apidoc Amx|void|lex_stuff_pv|const char *pv|U32 flags |
1096 | ||
1097 | Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>), | |
1098 | immediately after the current lexing point (L</PL_parser-E<gt>bufptr>), | |
1099 | reallocating the buffer if necessary. This means that lexing code that | |
1100 | runs later will see the characters as if they had appeared in the input. | |
1101 | It is not recommended to do this as part of normal parsing, and most | |
1102 | uses of this facility run the risk of the inserted characters being | |
1103 | interpreted in an unintended manner. | |
1104 | ||
1105 | The string to be inserted is represented by octets starting at I<pv> | |
1106 | and continuing to the first nul. These octets are interpreted as either | |
1107 | UTF-8 or Latin-1, according to whether the C<LEX_STUFF_UTF8> flag is set | |
1108 | in I<flags>. The characters are recoded for the lexer buffer, according | |
1109 | to how the buffer is currently being interpreted (L</lex_bufutf8>). | |
1110 | If it is not convenient to nul-terminate a string to be inserted, the | |
1111 | L</lex_stuff_pvn> function is more appropriate. | |
1112 | ||
1113 | =cut | |
1114 | */ | |
1115 | ||
1116 | void | |
1117 | Perl_lex_stuff_pv(pTHX_ const char *pv, U32 flags) | |
1118 | { | |
1119 | PERL_ARGS_ASSERT_LEX_STUFF_PV; | |
1120 | lex_stuff_pvn(pv, strlen(pv), flags); | |
1121 | } | |
1122 | ||
1123 | /* | |
f0e67a1d Z |
1124 | =for apidoc Amx|void|lex_stuff_sv|SV *sv|U32 flags |
1125 | ||
1126 | Insert characters into the lexer buffer (L</PL_parser-E<gt>linestr>), | |
1127 | immediately after the current lexing point (L</PL_parser-E<gt>bufptr>), | |
1128 | reallocating the buffer if necessary. This means that lexing code that | |
1129 | runs later will see the characters as if they had appeared in the input. | |
1130 | It is not recommended to do this as part of normal parsing, and most | |
1131 | uses of this facility run the risk of the inserted characters being | |
1132 | interpreted in an unintended manner. | |
1133 | ||
1134 | The string to be inserted is the string value of I<sv>. The characters | |
1135 | are recoded for the lexer buffer, according to how the buffer is currently | |
9dcc53ea | 1136 | being interpreted (L</lex_bufutf8>). If a string to be inserted is |
f0e67a1d Z |
1137 | not already a Perl scalar, the L</lex_stuff_pvn> function avoids the |
1138 | need to construct a scalar. | |
1139 | ||
1140 | =cut | |
1141 | */ | |
1142 | ||
1143 | void | |
1144 | Perl_lex_stuff_sv(pTHX_ SV *sv, U32 flags) | |
1145 | { | |
1146 | char *pv; | |
1147 | STRLEN len; | |
1148 | PERL_ARGS_ASSERT_LEX_STUFF_SV; | |
1149 | if (flags) | |
1150 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_stuff_sv"); | |
1151 | pv = SvPV(sv, len); | |
1152 | lex_stuff_pvn(pv, len, flags | (SvUTF8(sv) ? LEX_STUFF_UTF8 : 0)); | |
1153 | } | |
1154 | ||
1155 | /* | |
1156 | =for apidoc Amx|void|lex_unstuff|char *ptr | |
1157 | ||
1158 | Discards text about to be lexed, from L</PL_parser-E<gt>bufptr> up to | |
1159 | I<ptr>. Text following I<ptr> will be moved, and the buffer shortened. | |
1160 | This hides the discarded text from any lexing code that runs later, | |
1161 | as if the text had never appeared. | |
1162 | ||
1163 | This is not the normal way to consume lexed text. For that, use | |
1164 | L</lex_read_to>. | |
1165 | ||
1166 | =cut | |
1167 | */ | |
1168 | ||
1169 | void | |
1170 | Perl_lex_unstuff(pTHX_ char *ptr) | |
1171 | { | |
1172 | char *buf, *bufend; | |
1173 | STRLEN unstuff_len; | |
1174 | PERL_ARGS_ASSERT_LEX_UNSTUFF; | |
1175 | buf = PL_parser->bufptr; | |
1176 | if (ptr < buf) | |
1177 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff"); | |
1178 | if (ptr == buf) | |
1179 | return; | |
1180 | bufend = PL_parser->bufend; | |
1181 | if (ptr > bufend) | |
1182 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_unstuff"); | |
1183 | unstuff_len = ptr - buf; | |
1184 | Move(ptr, buf, bufend+1-ptr, char); | |
1185 | SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - unstuff_len); | |
1186 | PL_parser->bufend = bufend - unstuff_len; | |
1187 | } | |
1188 | ||
1189 | /* | |
1190 | =for apidoc Amx|void|lex_read_to|char *ptr | |
1191 | ||
1192 | Consume text in the lexer buffer, from L</PL_parser-E<gt>bufptr> up | |
1193 | to I<ptr>. This advances L</PL_parser-E<gt>bufptr> to match I<ptr>, | |
1194 | performing the correct bookkeeping whenever a newline character is passed. | |
1195 | This is the normal way to consume lexed text. | |
1196 | ||
1197 | Interpretation of the buffer's octets can be abstracted out by | |
1198 | using the slightly higher-level functions L</lex_peek_unichar> and | |
1199 | L</lex_read_unichar>. | |
1200 | ||
1201 | =cut | |
1202 | */ | |
1203 | ||
1204 | void | |
1205 | Perl_lex_read_to(pTHX_ char *ptr) | |
1206 | { | |
1207 | char *s; | |
1208 | PERL_ARGS_ASSERT_LEX_READ_TO; | |
1209 | s = PL_parser->bufptr; | |
1210 | if (ptr < s || ptr > PL_parser->bufend) | |
1211 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_to"); | |
1212 | for (; s != ptr; s++) | |
1213 | if (*s == '\n') { | |
83944c01 | 1214 | COPLINE_INC_WITH_HERELINES; |
f0e67a1d Z |
1215 | PL_parser->linestart = s+1; |
1216 | } | |
1217 | PL_parser->bufptr = ptr; | |
1218 | } | |
1219 | ||
1220 | /* | |
1221 | =for apidoc Amx|void|lex_discard_to|char *ptr | |
1222 | ||
1223 | Discards the first part of the L</PL_parser-E<gt>linestr> buffer, | |
1224 | up to I<ptr>. The remaining content of the buffer will be moved, and | |
1225 | all pointers into the buffer updated appropriately. I<ptr> must not | |
1226 | be later in the buffer than the position of L</PL_parser-E<gt>bufptr>: | |
1227 | it is not permitted to discard text that has yet to be lexed. | |
1228 | ||
1229 | Normally it is not necessarily to do this directly, because it suffices to | |
1230 | use the implicit discarding behaviour of L</lex_next_chunk> and things | |
1231 | based on it. However, if a token stretches across multiple lines, | |
1f317c95 | 1232 | and the lexing code has kept multiple lines of text in the buffer for |
f0e67a1d Z |
1233 | that purpose, then after completion of the token it would be wise to |
1234 | explicitly discard the now-unneeded earlier lines, to avoid future | |
1235 | multi-line tokens growing the buffer without bound. | |
1236 | ||
1237 | =cut | |
1238 | */ | |
1239 | ||
1240 | void | |
1241 | Perl_lex_discard_to(pTHX_ char *ptr) | |
1242 | { | |
1243 | char *buf; | |
1244 | STRLEN discard_len; | |
1245 | PERL_ARGS_ASSERT_LEX_DISCARD_TO; | |
1246 | buf = SvPVX(PL_parser->linestr); | |
1247 | if (ptr < buf) | |
1248 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to"); | |
1249 | if (ptr == buf) | |
1250 | return; | |
1251 | if (ptr > PL_parser->bufptr) | |
1252 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_discard_to"); | |
1253 | discard_len = ptr - buf; | |
1254 | if (PL_parser->oldbufptr < ptr) | |
1255 | PL_parser->oldbufptr = ptr; | |
1256 | if (PL_parser->oldoldbufptr < ptr) | |
1257 | PL_parser->oldoldbufptr = ptr; | |
1258 | if (PL_parser->last_uni && PL_parser->last_uni < ptr) | |
1259 | PL_parser->last_uni = NULL; | |
1260 | if (PL_parser->last_lop && PL_parser->last_lop < ptr) | |
1261 | PL_parser->last_lop = NULL; | |
1262 | Move(ptr, buf, PL_parser->bufend+1-ptr, char); | |
1263 | SvCUR_set(PL_parser->linestr, SvCUR(PL_parser->linestr) - discard_len); | |
1264 | PL_parser->bufend -= discard_len; | |
1265 | PL_parser->bufptr -= discard_len; | |
1266 | PL_parser->oldbufptr -= discard_len; | |
1267 | PL_parser->oldoldbufptr -= discard_len; | |
1268 | if (PL_parser->last_uni) | |
1269 | PL_parser->last_uni -= discard_len; | |
1270 | if (PL_parser->last_lop) | |
1271 | PL_parser->last_lop -= discard_len; | |
1272 | } | |
1273 | ||
1274 | /* | |
1275 | =for apidoc Amx|bool|lex_next_chunk|U32 flags | |
1276 | ||
1277 | Reads in the next chunk of text to be lexed, appending it to | |
1278 | L</PL_parser-E<gt>linestr>. This should be called when lexing code has | |
1279 | looked to the end of the current chunk and wants to know more. It is | |
1280 | usual, but not necessary, for lexing to have consumed the entirety of | |
1281 | the current chunk at this time. | |
1282 | ||
1283 | If L</PL_parser-E<gt>bufptr> is pointing to the very end of the current | |
1284 | chunk (i.e., the current chunk has been entirely consumed), normally the | |
1285 | current chunk will be discarded at the same time that the new chunk is | |
1286 | read in. If I<flags> includes C<LEX_KEEP_PREVIOUS>, the current chunk | |
1287 | will not be discarded. If the current chunk has not been entirely | |
1288 | consumed, then it will not be discarded regardless of the flag. | |
1289 | ||
1290 | Returns true if some new text was added to the buffer, or false if the | |
1291 | buffer has reached the end of the input text. | |
1292 | ||
1293 | =cut | |
1294 | */ | |
1295 | ||
1296 | #define LEX_FAKE_EOF 0x80000000 | |
112d1284 | 1297 | #define LEX_NO_TERM 0x40000000 |
f0e67a1d Z |
1298 | |
1299 | bool | |
1300 | Perl_lex_next_chunk(pTHX_ U32 flags) | |
1301 | { | |
1302 | SV *linestr; | |
1303 | char *buf; | |
1304 | STRLEN old_bufend_pos, new_bufend_pos; | |
1305 | STRLEN bufptr_pos, oldbufptr_pos, oldoldbufptr_pos; | |
1306 | STRLEN linestart_pos, last_uni_pos, last_lop_pos; | |
17cc9359 | 1307 | bool got_some_for_debugger = 0; |
f0e67a1d | 1308 | bool got_some; |
112d1284 | 1309 | if (flags & ~(LEX_KEEP_PREVIOUS|LEX_FAKE_EOF|LEX_NO_TERM)) |
f0e67a1d | 1310 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_next_chunk"); |
f0e67a1d Z |
1311 | linestr = PL_parser->linestr; |
1312 | buf = SvPVX(linestr); | |
1313 | if (!(flags & LEX_KEEP_PREVIOUS) && | |
1314 | PL_parser->bufptr == PL_parser->bufend) { | |
1315 | old_bufend_pos = bufptr_pos = oldbufptr_pos = oldoldbufptr_pos = 0; | |
1316 | linestart_pos = 0; | |
1317 | if (PL_parser->last_uni != PL_parser->bufend) | |
1318 | PL_parser->last_uni = NULL; | |
1319 | if (PL_parser->last_lop != PL_parser->bufend) | |
1320 | PL_parser->last_lop = NULL; | |
1321 | last_uni_pos = last_lop_pos = 0; | |
1322 | *buf = 0; | |
1323 | SvCUR(linestr) = 0; | |
1324 | } else { | |
1325 | old_bufend_pos = PL_parser->bufend - buf; | |
1326 | bufptr_pos = PL_parser->bufptr - buf; | |
1327 | oldbufptr_pos = PL_parser->oldbufptr - buf; | |
1328 | oldoldbufptr_pos = PL_parser->oldoldbufptr - buf; | |
1329 | linestart_pos = PL_parser->linestart - buf; | |
1330 | last_uni_pos = PL_parser->last_uni ? PL_parser->last_uni - buf : 0; | |
1331 | last_lop_pos = PL_parser->last_lop ? PL_parser->last_lop - buf : 0; | |
1332 | } | |
1333 | if (flags & LEX_FAKE_EOF) { | |
1334 | goto eof; | |
60d63348 | 1335 | } else if (!PL_parser->rsfp && !PL_parser->filtered) { |
f0e67a1d Z |
1336 | got_some = 0; |
1337 | } else if (filter_gets(linestr, old_bufend_pos)) { | |
1338 | got_some = 1; | |
17cc9359 | 1339 | got_some_for_debugger = 1; |
112d1284 FC |
1340 | } else if (flags & LEX_NO_TERM) { |
1341 | got_some = 0; | |
f0e67a1d | 1342 | } else { |
580561a3 Z |
1343 | if (!SvPOK(linestr)) /* can get undefined by filter_gets */ |
1344 | sv_setpvs(linestr, ""); | |
f0e67a1d Z |
1345 | eof: |
1346 | /* End of real input. Close filehandle (unless it was STDIN), | |
1347 | * then add implicit termination. | |
1348 | */ | |
87606032 | 1349 | if (PL_parser->lex_flags & LEX_DONT_CLOSE_RSFP) |
f0e67a1d Z |
1350 | PerlIO_clearerr(PL_parser->rsfp); |
1351 | else if (PL_parser->rsfp) | |
1352 | (void)PerlIO_close(PL_parser->rsfp); | |
1353 | PL_parser->rsfp = NULL; | |
60d63348 | 1354 | PL_parser->in_pod = PL_parser->filtered = 0; |
f0e67a1d Z |
1355 | #ifdef PERL_MAD |
1356 | if (PL_madskills && !PL_in_eval && (PL_minus_p || PL_minus_n)) | |
1357 | PL_faketokens = 1; | |
1358 | #endif | |
1359 | if (!PL_in_eval && PL_minus_p) { | |
1360 | sv_catpvs(linestr, | |
1361 | /*{*/";}continue{print or die qq(-p destination: $!\\n);}"); | |
1362 | PL_minus_n = PL_minus_p = 0; | |
1363 | } else if (!PL_in_eval && PL_minus_n) { | |
1364 | sv_catpvs(linestr, /*{*/";}"); | |
1365 | PL_minus_n = 0; | |
1366 | } else | |
1367 | sv_catpvs(linestr, ";"); | |
1368 | got_some = 1; | |
1369 | } | |
1370 | buf = SvPVX(linestr); | |
1371 | new_bufend_pos = SvCUR(linestr); | |
1372 | PL_parser->bufend = buf + new_bufend_pos; | |
1373 | PL_parser->bufptr = buf + bufptr_pos; | |
1374 | PL_parser->oldbufptr = buf + oldbufptr_pos; | |
1375 | PL_parser->oldoldbufptr = buf + oldoldbufptr_pos; | |
1376 | PL_parser->linestart = buf + linestart_pos; | |
1377 | if (PL_parser->last_uni) | |
1378 | PL_parser->last_uni = buf + last_uni_pos; | |
1379 | if (PL_parser->last_lop) | |
1380 | PL_parser->last_lop = buf + last_lop_pos; | |
17cc9359 | 1381 | if (got_some_for_debugger && (PERLDB_LINE || PERLDB_SAVESRC) && |
f0e67a1d Z |
1382 | PL_curstash != PL_debstash) { |
1383 | /* debugger active and we're not compiling the debugger code, | |
1384 | * so store the line into the debugger's array of lines | |
1385 | */ | |
1386 | update_debugger_info(NULL, buf+old_bufend_pos, | |
1387 | new_bufend_pos-old_bufend_pos); | |
1388 | } | |
1389 | return got_some; | |
1390 | } | |
1391 | ||
1392 | /* | |
1393 | =for apidoc Amx|I32|lex_peek_unichar|U32 flags | |
1394 | ||
1395 | Looks ahead one (Unicode) character in the text currently being lexed. | |
1396 | Returns the codepoint (unsigned integer value) of the next character, | |
1397 | or -1 if lexing has reached the end of the input text. To consume the | |
1398 | peeked character, use L</lex_read_unichar>. | |
1399 | ||
1400 | If the next character is in (or extends into) the next chunk of input | |
1401 | text, the next chunk will be read in. Normally the current chunk will be | |
1402 | discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> | |
1403 | then the current chunk will not be discarded. | |
1404 | ||
1405 | If the input is being interpreted as UTF-8 and a UTF-8 encoding error | |
1406 | is encountered, an exception is generated. | |
1407 | ||
1408 | =cut | |
1409 | */ | |
1410 | ||
1411 | I32 | |
1412 | Perl_lex_peek_unichar(pTHX_ U32 flags) | |
1413 | { | |
749123ff | 1414 | dVAR; |
f0e67a1d Z |
1415 | char *s, *bufend; |
1416 | if (flags & ~(LEX_KEEP_PREVIOUS)) | |
1417 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_peek_unichar"); | |
1418 | s = PL_parser->bufptr; | |
1419 | bufend = PL_parser->bufend; | |
1420 | if (UTF) { | |
1421 | U8 head; | |
1422 | I32 unichar; | |
1423 | STRLEN len, retlen; | |
1424 | if (s == bufend) { | |
1425 | if (!lex_next_chunk(flags)) | |
1426 | return -1; | |
1427 | s = PL_parser->bufptr; | |
1428 | bufend = PL_parser->bufend; | |
1429 | } | |
1430 | head = (U8)*s; | |
54d004e8 | 1431 | if (UTF8_IS_INVARIANT(head)) |
f0e67a1d | 1432 | return head; |
54d004e8 KW |
1433 | if (UTF8_IS_START(head)) { |
1434 | len = UTF8SKIP(&head); | |
f0e67a1d Z |
1435 | while ((STRLEN)(bufend-s) < len) { |
1436 | if (!lex_next_chunk(flags | LEX_KEEP_PREVIOUS)) | |
1437 | break; | |
1438 | s = PL_parser->bufptr; | |
1439 | bufend = PL_parser->bufend; | |
1440 | } | |
1441 | } | |
1442 | unichar = utf8n_to_uvuni((U8*)s, bufend-s, &retlen, UTF8_CHECK_ONLY); | |
1443 | if (retlen == (STRLEN)-1) { | |
1444 | /* malformed UTF-8 */ | |
1445 | ENTER; | |
1446 | SAVESPTR(PL_warnhook); | |
1447 | PL_warnhook = PERL_WARNHOOK_FATAL; | |
1448 | utf8n_to_uvuni((U8*)s, bufend-s, NULL, 0); | |
1449 | LEAVE; | |
1450 | } | |
1451 | return unichar; | |
1452 | } else { | |
1453 | if (s == bufend) { | |
1454 | if (!lex_next_chunk(flags)) | |
1455 | return -1; | |
1456 | s = PL_parser->bufptr; | |
1457 | } | |
1458 | return (U8)*s; | |
1459 | } | |
1460 | } | |
1461 | ||
1462 | /* | |
1463 | =for apidoc Amx|I32|lex_read_unichar|U32 flags | |
1464 | ||
1465 | Reads the next (Unicode) character in the text currently being lexed. | |
1466 | Returns the codepoint (unsigned integer value) of the character read, | |
1467 | and moves L</PL_parser-E<gt>bufptr> past the character, or returns -1 | |
1468 | if lexing has reached the end of the input text. To non-destructively | |
1469 | examine the next character, use L</lex_peek_unichar> instead. | |
1470 | ||
1471 | If the next character is in (or extends into) the next chunk of input | |
1472 | text, the next chunk will be read in. Normally the current chunk will be | |
1473 | discarded at the same time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> | |
1474 | then the current chunk will not be discarded. | |
1475 | ||
1476 | If the input is being interpreted as UTF-8 and a UTF-8 encoding error | |
1477 | is encountered, an exception is generated. | |
1478 | ||
1479 | =cut | |
1480 | */ | |
1481 | ||
1482 | I32 | |
1483 | Perl_lex_read_unichar(pTHX_ U32 flags) | |
1484 | { | |
1485 | I32 c; | |
1486 | if (flags & ~(LEX_KEEP_PREVIOUS)) | |
1487 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_unichar"); | |
1488 | c = lex_peek_unichar(flags); | |
1489 | if (c != -1) { | |
1490 | if (c == '\n') | |
83944c01 | 1491 | COPLINE_INC_WITH_HERELINES; |
d9018cbe EB |
1492 | if (UTF) |
1493 | PL_parser->bufptr += UTF8SKIP(PL_parser->bufptr); | |
1494 | else | |
1495 | ++(PL_parser->bufptr); | |
f0e67a1d Z |
1496 | } |
1497 | return c; | |
1498 | } | |
1499 | ||
1500 | /* | |
1501 | =for apidoc Amx|void|lex_read_space|U32 flags | |
1502 | ||
1503 | Reads optional spaces, in Perl style, in the text currently being | |
1504 | lexed. The spaces may include ordinary whitespace characters and | |
1505 | Perl-style comments. C<#line> directives are processed if encountered. | |
1506 | L</PL_parser-E<gt>bufptr> is moved past the spaces, so that it points | |
1507 | at a non-space character (or the end of the input text). | |
1508 | ||
1509 | If spaces extend into the next chunk of input text, the next chunk will | |
1510 | be read in. Normally the current chunk will be discarded at the same | |
1511 | time, but if I<flags> includes C<LEX_KEEP_PREVIOUS> then the current | |
1512 | chunk will not be discarded. | |
1513 | ||
1514 | =cut | |
1515 | */ | |
1516 | ||
f0998909 Z |
1517 | #define LEX_NO_NEXT_CHUNK 0x80000000 |
1518 | ||
f0e67a1d Z |
1519 | void |
1520 | Perl_lex_read_space(pTHX_ U32 flags) | |
1521 | { | |
1522 | char *s, *bufend; | |
1523 | bool need_incline = 0; | |
f0998909 | 1524 | if (flags & ~(LEX_KEEP_PREVIOUS|LEX_NO_NEXT_CHUNK)) |
f0e67a1d Z |
1525 | Perl_croak(aTHX_ "Lexing code internal error (%s)", "lex_read_space"); |
1526 | #ifdef PERL_MAD | |
1527 | if (PL_skipwhite) { | |
1528 | sv_free(PL_skipwhite); | |
1529 | PL_skipwhite = NULL; | |
1530 | } | |
1531 | if (PL_madskills) | |
1532 | PL_skipwhite = newSVpvs(""); | |
1533 | #endif /* PERL_MAD */ | |
1534 | s = PL_parser->bufptr; | |
1535 | bufend = PL_parser->bufend; | |
1536 | while (1) { | |
1537 | char c = *s; | |
1538 | if (c == '#') { | |
1539 | do { | |
1540 | c = *++s; | |
1541 | } while (!(c == '\n' || (c == 0 && s == bufend))); | |
1542 | } else if (c == '\n') { | |
1543 | s++; | |
1544 | PL_parser->linestart = s; | |
1545 | if (s == bufend) | |
1546 | need_incline = 1; | |
1547 | else | |
1548 | incline(s); | |
1549 | } else if (isSPACE(c)) { | |
1550 | s++; | |
1551 | } else if (c == 0 && s == bufend) { | |
1552 | bool got_more; | |
1553 | #ifdef PERL_MAD | |
1554 | if (PL_madskills) | |
1555 | sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr); | |
1556 | #endif /* PERL_MAD */ | |
f0998909 Z |
1557 | if (flags & LEX_NO_NEXT_CHUNK) |
1558 | break; | |
f0e67a1d | 1559 | PL_parser->bufptr = s; |
83944c01 | 1560 | COPLINE_INC_WITH_HERELINES; |
f0e67a1d Z |
1561 | got_more = lex_next_chunk(flags); |
1562 | CopLINE_dec(PL_curcop); | |
1563 | s = PL_parser->bufptr; | |
1564 | bufend = PL_parser->bufend; | |
1565 | if (!got_more) | |
1566 | break; | |
1567 | if (need_incline && PL_parser->rsfp) { | |
1568 | incline(s); | |
1569 | need_incline = 0; | |
1570 | } | |
1571 | } else { | |
1572 | break; | |
1573 | } | |
1574 | } | |
1575 | #ifdef PERL_MAD | |
1576 | if (PL_madskills) | |
1577 | sv_catpvn(PL_skipwhite, PL_parser->bufptr, s-PL_parser->bufptr); | |
1578 | #endif /* PERL_MAD */ | |
1579 | PL_parser->bufptr = s; | |
1580 | } | |
1581 | ||
1582 | /* | |
ffb4593c NT |
1583 | * S_incline |
1584 | * This subroutine has nothing to do with tilting, whether at windmills | |
1585 | * or pinball tables. Its name is short for "increment line". It | |
57843af0 | 1586 | * increments the current line number in CopLINE(PL_curcop) and checks |
ffb4593c | 1587 | * to see whether the line starts with a comment of the form |
9cbb5ea2 GS |
1588 | * # line 500 "foo.pm" |
1589 | * If so, it sets the current line number and file to the values in the comment. | |
ffb4593c NT |
1590 | */ |
1591 | ||
76e3520e | 1592 | STATIC void |
d9095cec | 1593 | S_incline(pTHX_ const char *s) |
463ee0b2 | 1594 | { |
97aff369 | 1595 | dVAR; |
d9095cec NC |
1596 | const char *t; |
1597 | const char *n; | |
1598 | const char *e; | |
8818d409 | 1599 | line_t line_num; |
463ee0b2 | 1600 | |
7918f24d NC |
1601 | PERL_ARGS_ASSERT_INCLINE; |
1602 | ||
83944c01 | 1603 | COPLINE_INC_WITH_HERELINES; |
451f421f FC |
1604 | if (!PL_rsfp && !PL_parser->filtered && PL_lex_state == LEX_NORMAL |
1605 | && s+1 == PL_bufend && *s == ';') { | |
1606 | /* fake newline in string eval */ | |
1607 | CopLINE_dec(PL_curcop); | |
1608 | return; | |
1609 | } | |
463ee0b2 LW |
1610 | if (*s++ != '#') |
1611 | return; | |
d4c19fe8 AL |
1612 | while (SPACE_OR_TAB(*s)) |
1613 | s++; | |
73659bf1 GS |
1614 | if (strnEQ(s, "line", 4)) |
1615 | s += 4; | |
1616 | else | |
1617 | return; | |
084592ab | 1618 | if (SPACE_OR_TAB(*s)) |
73659bf1 | 1619 | s++; |
4e553d73 | 1620 | else |
73659bf1 | 1621 | return; |
d4c19fe8 AL |
1622 | while (SPACE_OR_TAB(*s)) |
1623 | s++; | |
463ee0b2 LW |
1624 | if (!isDIGIT(*s)) |
1625 | return; | |
d4c19fe8 | 1626 | |
463ee0b2 LW |
1627 | n = s; |
1628 | while (isDIGIT(*s)) | |
1629 | s++; | |
07714eb4 | 1630 | if (!SPACE_OR_TAB(*s) && *s != '\r' && *s != '\n' && *s != '\0') |
26b6dc3f | 1631 | return; |
bf4acbe4 | 1632 | while (SPACE_OR_TAB(*s)) |
463ee0b2 | 1633 | s++; |
73659bf1 | 1634 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
463ee0b2 | 1635 | s++; |
73659bf1 GS |
1636 | e = t + 1; |
1637 | } | |
463ee0b2 | 1638 | else { |
c35e046a AL |
1639 | t = s; |
1640 | while (!isSPACE(*t)) | |
1641 | t++; | |
73659bf1 | 1642 | e = t; |
463ee0b2 | 1643 | } |
bf4acbe4 | 1644 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
73659bf1 GS |
1645 | e++; |
1646 | if (*e != '\n' && *e != '\0') | |
1647 | return; /* false alarm */ | |
1648 | ||
8818d409 FC |
1649 | line_num = atoi(n)-1; |
1650 | ||
f4dd75d9 | 1651 | if (t - s > 0) { |
d9095cec | 1652 | const STRLEN len = t - s; |
19bad673 NC |
1653 | SV *const temp_sv = CopFILESV(PL_curcop); |
1654 | const char *cf; | |
1655 | STRLEN tmplen; | |
1656 | ||
1657 | if (temp_sv) { | |
1658 | cf = SvPVX(temp_sv); | |
1659 | tmplen = SvCUR(temp_sv); | |
1660 | } else { | |
1661 | cf = NULL; | |
1662 | tmplen = 0; | |
1663 | } | |
1664 | ||
d1299d44 | 1665 | if (!PL_rsfp && !PL_parser->filtered) { |
e66cf94c RGS |
1666 | /* must copy *{"::_<(eval N)[oldfilename:L]"} |
1667 | * to *{"::_<newfilename"} */ | |
44867030 NC |
1668 | /* However, the long form of evals is only turned on by the |
1669 | debugger - usually they're "(eval %lu)" */ | |
1670 | char smallbuf[128]; | |
1671 | char *tmpbuf; | |
1672 | GV **gvp; | |
d9095cec | 1673 | STRLEN tmplen2 = len; |
798b63bc | 1674 | if (tmplen + 2 <= sizeof smallbuf) |
e66cf94c RGS |
1675 | tmpbuf = smallbuf; |
1676 | else | |
2ae0db35 | 1677 | Newx(tmpbuf, tmplen + 2, char); |
44867030 NC |
1678 | tmpbuf[0] = '_'; |
1679 | tmpbuf[1] = '<'; | |
2ae0db35 | 1680 | memcpy(tmpbuf + 2, cf, tmplen); |
44867030 | 1681 | tmplen += 2; |
8a5ee598 RGS |
1682 | gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE); |
1683 | if (gvp) { | |
44867030 NC |
1684 | char *tmpbuf2; |
1685 | GV *gv2; | |
1686 | ||
1687 | if (tmplen2 + 2 <= sizeof smallbuf) | |
1688 | tmpbuf2 = smallbuf; | |
1689 | else | |
1690 | Newx(tmpbuf2, tmplen2 + 2, char); | |
1691 | ||
1692 | if (tmpbuf2 != smallbuf || tmpbuf != smallbuf) { | |
1693 | /* Either they malloc'd it, or we malloc'd it, | |
1694 | so no prefix is present in ours. */ | |
1695 | tmpbuf2[0] = '_'; | |
1696 | tmpbuf2[1] = '<'; | |
1697 | } | |
1698 | ||
1699 | memcpy(tmpbuf2 + 2, s, tmplen2); | |
1700 | tmplen2 += 2; | |
1701 | ||
8a5ee598 | 1702 | gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE); |
e5527e4b | 1703 | if (!isGV(gv2)) { |
8a5ee598 | 1704 | gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE); |
e5527e4b RGS |
1705 | /* adjust ${"::_<newfilename"} to store the new file name */ |
1706 | GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2); | |
8818d409 FC |
1707 | /* The line number may differ. If that is the case, |
1708 | alias the saved lines that are in the array. | |
1709 | Otherwise alias the whole array. */ | |
1710 | if (CopLINE(PL_curcop) == line_num) { | |
1711 | GvHV(gv2) = MUTABLE_HV(SvREFCNT_inc(GvHV(*gvp))); | |
1712 | GvAV(gv2) = MUTABLE_AV(SvREFCNT_inc(GvAV(*gvp))); | |
1713 | } | |
1714 | else if (GvAV(*gvp)) { | |
1715 | AV * const av = GvAV(*gvp); | |
1716 | const I32 start = CopLINE(PL_curcop)+1; | |
1717 | I32 items = AvFILLp(av) - start; | |
1718 | if (items > 0) { | |
1719 | AV * const av2 = GvAVn(gv2); | |
1720 | SV **svp = AvARRAY(av) + start; | |
1721 | I32 l = (I32)line_num+1; | |
1722 | while (items--) | |
1723 | av_store(av2, l++, SvREFCNT_inc(*svp++)); | |
1724 | } | |
1725 | } | |
e5527e4b | 1726 | } |
44867030 NC |
1727 | |
1728 | if (tmpbuf2 != smallbuf) Safefree(tmpbuf2); | |
8a5ee598 | 1729 | } |
e66cf94c | 1730 | if (tmpbuf != smallbuf) Safefree(tmpbuf); |
e66cf94c | 1731 | } |
05ec9bb3 | 1732 | CopFILE_free(PL_curcop); |
d9095cec | 1733 | CopFILE_setn(PL_curcop, s, len); |
f4dd75d9 | 1734 | } |
8818d409 | 1735 | CopLINE_set(PL_curcop, line_num); |
463ee0b2 LW |
1736 | } |
1737 | ||
29595ff2 | 1738 | #ifdef PERL_MAD |
cd81e915 | 1739 | /* skip space before PL_thistoken */ |
29595ff2 NC |
1740 | |
1741 | STATIC char * | |
5aaab254 | 1742 | S_skipspace0(pTHX_ char *s) |
29595ff2 | 1743 | { |
7918f24d NC |
1744 | PERL_ARGS_ASSERT_SKIPSPACE0; |
1745 | ||
29595ff2 NC |
1746 | s = skipspace(s); |
1747 | if (!PL_madskills) | |
1748 | return s; | |
cd81e915 NC |
1749 | if (PL_skipwhite) { |
1750 | if (!PL_thiswhite) | |
6b29d1f5 | 1751 | PL_thiswhite = newSVpvs(""); |
cd81e915 NC |
1752 | sv_catsv(PL_thiswhite, PL_skipwhite); |
1753 | sv_free(PL_skipwhite); | |
1754 | PL_skipwhite = 0; | |
1755 | } | |
1756 | PL_realtokenstart = s - SvPVX(PL_linestr); | |
29595ff2 NC |
1757 | return s; |
1758 | } | |
1759 | ||
cd81e915 | 1760 | /* skip space after PL_thistoken */ |
29595ff2 NC |
1761 | |
1762 | STATIC char * | |
5aaab254 | 1763 | S_skipspace1(pTHX_ char *s) |
29595ff2 | 1764 | { |
d4c19fe8 | 1765 | const char *start = s; |
29595ff2 NC |
1766 | I32 startoff = start - SvPVX(PL_linestr); |
1767 | ||
7918f24d NC |
1768 | PERL_ARGS_ASSERT_SKIPSPACE1; |
1769 | ||
29595ff2 NC |
1770 | s = skipspace(s); |
1771 | if (!PL_madskills) | |
1772 | return s; | |
1773 | start = SvPVX(PL_linestr) + startoff; | |
cd81e915 | 1774 | if (!PL_thistoken && PL_realtokenstart >= 0) { |
d4c19fe8 | 1775 | const char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart; |
cd81e915 NC |
1776 | PL_thistoken = newSVpvn(tstart, start - tstart); |
1777 | } | |
1778 | PL_realtokenstart = -1; | |
1779 | if (PL_skipwhite) { | |
1780 | if (!PL_nextwhite) | |
6b29d1f5 | 1781 | PL_nextwhite = newSVpvs(""); |
cd81e915 NC |
1782 | sv_catsv(PL_nextwhite, PL_skipwhite); |
1783 | sv_free(PL_skipwhite); | |
1784 | PL_skipwhite = 0; | |
29595ff2 NC |
1785 | } |
1786 | return s; | |
1787 | } | |
1788 | ||
1789 | STATIC char * | |
5aaab254 | 1790 | S_skipspace2(pTHX_ char *s, SV **svp) |
29595ff2 | 1791 | { |
c35e046a AL |
1792 | char *start; |
1793 | const I32 bufptroff = PL_bufptr - SvPVX(PL_linestr); | |
1794 | const I32 startoff = s - SvPVX(PL_linestr); | |
1795 | ||
7918f24d NC |
1796 | PERL_ARGS_ASSERT_SKIPSPACE2; |
1797 | ||
29595ff2 NC |
1798 | s = skipspace(s); |
1799 | PL_bufptr = SvPVX(PL_linestr) + bufptroff; | |
1800 | if (!PL_madskills || !svp) | |
1801 | return s; | |
1802 | start = SvPVX(PL_linestr) + startoff; | |
cd81e915 | 1803 | if (!PL_thistoken && PL_realtokenstart >= 0) { |
d4c19fe8 | 1804 | char * const tstart = SvPVX(PL_linestr) + PL_realtokenstart; |
cd81e915 NC |
1805 | PL_thistoken = newSVpvn(tstart, start - tstart); |
1806 | PL_realtokenstart = -1; | |
29595ff2 | 1807 | } |
cd81e915 | 1808 | if (PL_skipwhite) { |
29595ff2 | 1809 | if (!*svp) |
6b29d1f5 | 1810 | *svp = newSVpvs(""); |
cd81e915 NC |
1811 | sv_setsv(*svp, PL_skipwhite); |
1812 | sv_free(PL_skipwhite); | |
1813 | PL_skipwhite = 0; | |
29595ff2 NC |
1814 | } |
1815 | ||
1816 | return s; | |
1817 | } | |
1818 | #endif | |
1819 | ||
80a702cd | 1820 | STATIC void |
15f169a1 | 1821 | S_update_debugger_info(pTHX_ SV *orig_sv, const char *const buf, STRLEN len) |
80a702cd RGS |
1822 | { |
1823 | AV *av = CopFILEAVx(PL_curcop); | |
1824 | if (av) { | |
b9f83d2f | 1825 | SV * const sv = newSV_type(SVt_PVMG); |
5fa550fb NC |
1826 | if (orig_sv) |
1827 | sv_setsv(sv, orig_sv); | |
1828 | else | |
1829 | sv_setpvn(sv, buf, len); | |
80a702cd RGS |
1830 | (void)SvIOK_on(sv); |
1831 | SvIV_set(sv, 0); | |
1832 | av_store(av, (I32)CopLINE(PL_curcop), sv); | |
1833 | } | |
1834 | } | |
1835 | ||
ffb4593c NT |
1836 | /* |
1837 | * S_skipspace | |
1838 | * Called to gobble the appropriate amount and type of whitespace. | |
1839 | * Skips comments as well. | |
1840 | */ | |
1841 | ||
76e3520e | 1842 | STATIC char * |
5aaab254 | 1843 | S_skipspace(pTHX_ char *s) |
a687059c | 1844 | { |
5db06880 | 1845 | #ifdef PERL_MAD |
f0e67a1d Z |
1846 | char *start = s; |
1847 | #endif /* PERL_MAD */ | |
7918f24d | 1848 | PERL_ARGS_ASSERT_SKIPSPACE; |
f0e67a1d | 1849 | #ifdef PERL_MAD |
cd81e915 NC |
1850 | if (PL_skipwhite) { |
1851 | sv_free(PL_skipwhite); | |
f0e67a1d | 1852 | PL_skipwhite = NULL; |
5db06880 | 1853 | } |
f0e67a1d | 1854 | #endif /* PERL_MAD */ |
3280af22 | 1855 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
bf4acbe4 | 1856 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
463ee0b2 | 1857 | s++; |
f0e67a1d Z |
1858 | } else { |
1859 | STRLEN bufptr_pos = PL_bufptr - SvPVX(PL_linestr); | |
1860 | PL_bufptr = s; | |
f0998909 Z |
1861 | lex_read_space(LEX_KEEP_PREVIOUS | |
1862 | (PL_sublex_info.sub_inwhat || PL_lex_state == LEX_FORMLINE ? | |
1863 | LEX_NO_NEXT_CHUNK : 0)); | |
3280af22 | 1864 | s = PL_bufptr; |
f0e67a1d Z |
1865 | PL_bufptr = SvPVX(PL_linestr) + bufptr_pos; |
1866 | if (PL_linestart > PL_bufptr) | |
1867 | PL_bufptr = PL_linestart; | |
1868 | return s; | |
463ee0b2 | 1869 | } |
5db06880 | 1870 | #ifdef PERL_MAD |
f0e67a1d Z |
1871 | if (PL_madskills) |
1872 | PL_skipwhite = newSVpvn(start, s-start); | |
1873 | #endif /* PERL_MAD */ | |
5db06880 | 1874 | return s; |
a687059c | 1875 | } |
378cc40b | 1876 | |
ffb4593c NT |
1877 | /* |
1878 | * S_check_uni | |
1879 | * Check the unary operators to ensure there's no ambiguity in how they're | |
1880 | * used. An ambiguous piece of code would be: | |
1881 | * rand + 5 | |
1882 | * This doesn't mean rand() + 5. Because rand() is a unary operator, | |
1883 | * the +5 is its argument. | |
1884 | */ | |
1885 | ||
76e3520e | 1886 | STATIC void |
cea2e8a9 | 1887 | S_check_uni(pTHX) |
ba106d47 | 1888 | { |
97aff369 | 1889 | dVAR; |
d4c19fe8 AL |
1890 | const char *s; |
1891 | const char *t; | |
2f3197b3 | 1892 | |
3280af22 | 1893 | if (PL_oldoldbufptr != PL_last_uni) |
2f3197b3 | 1894 | return; |
3280af22 NIS |
1895 | while (isSPACE(*PL_last_uni)) |
1896 | PL_last_uni++; | |
c35e046a | 1897 | s = PL_last_uni; |
8a2bca12 | 1898 | while (isWORDCHAR_lazy_if(s,UTF) || *s == '-') |
c35e046a | 1899 | s++; |
3280af22 | 1900 | if ((t = strchr(s, '(')) && t < PL_bufptr) |
a0d0e21e | 1901 | return; |
6136c704 | 1902 | |
9b387841 NC |
1903 | Perl_ck_warner_d(aTHX_ packWARN(WARN_AMBIGUOUS), |
1904 | "Warning: Use of \"%.*s\" without parentheses is ambiguous", | |
1905 | (int)(s - PL_last_uni), PL_last_uni); | |
2f3197b3 LW |
1906 | } |
1907 | ||
ffb4593c NT |
1908 | /* |
1909 | * LOP : macro to build a list operator. Its behaviour has been replaced | |
1910 | * with a subroutine, S_lop() for which LOP is just another name. | |
1911 | */ | |
1912 | ||
a0d0e21e LW |
1913 | #define LOP(f,x) return lop(f,x,s) |
1914 | ||
ffb4593c NT |
1915 | /* |
1916 | * S_lop | |
1917 | * Build a list operator (or something that might be one). The rules: | |
1918 | * - if we have a next token, then it's a list operator [why?] | |
1919 | * - if the next thing is an opening paren, then it's a function | |
1920 | * - else it's a list operator | |
1921 | */ | |
1922 | ||
76e3520e | 1923 | STATIC I32 |
a0be28da | 1924 | S_lop(pTHX_ I32 f, int x, char *s) |
ffed7fef | 1925 | { |
97aff369 | 1926 | dVAR; |
7918f24d NC |
1927 | |
1928 | PERL_ARGS_ASSERT_LOP; | |
1929 | ||
6154021b | 1930 | pl_yylval.ival = f; |
35c8bce7 | 1931 | CLINE; |
3280af22 NIS |
1932 | PL_expect = x; |
1933 | PL_bufptr = s; | |
1934 | PL_last_lop = PL_oldbufptr; | |
eb160463 | 1935 | PL_last_lop_op = (OPCODE)f; |
5db06880 NC |
1936 | #ifdef PERL_MAD |
1937 | if (PL_lasttoke) | |
78cdf107 | 1938 | goto lstop; |
5db06880 | 1939 | #else |
3280af22 | 1940 | if (PL_nexttoke) |
78cdf107 | 1941 | goto lstop; |
5db06880 | 1942 | #endif |
79072805 | 1943 | if (*s == '(') |
bbf60fe6 | 1944 | return REPORT(FUNC); |
29595ff2 | 1945 | s = PEEKSPACE(s); |
79072805 | 1946 | if (*s == '(') |
bbf60fe6 | 1947 | return REPORT(FUNC); |
78cdf107 Z |
1948 | else { |
1949 | lstop: | |
1950 | if (!PL_lex_allbrackets && PL_lex_fakeeof > LEX_FAKEEOF_LOWLOGIC) | |
1951 | PL_lex_fakeeof = LEX_FAKEEOF_LOWLOGIC; | |
bbf60fe6 | 1952 | return REPORT(LSTOP); |
78cdf107 | 1953 | } |
79072805 LW |
1954 | } |
1955 | ||
5db06880 NC |
1956 | #ifdef PERL_MAD |
1957 | /* | |
1958 | * S_start_force | |
1959 | * Sets up for an eventual force_next(). start_force(0) basically does | |
1960 | * an unshift, while start_force(-1) does a push. yylex removes items | |
1961 | * on the "pop" end. | |
1962 | */ | |
1963 | ||
1964 | STATIC void | |
1965 | S_start_force(pTHX_ int where) | |
1966 | { | |
1967 | int i; | |
1968 | ||
cd81e915 | 1969 | if (where < 0) /* so people can duplicate start_force(PL_curforce) */ |
5db06880 | 1970 | where = PL_lasttoke; |
cd81e915 NC |
1971 | assert(PL_curforce < 0 || PL_curforce == where); |
1972 | if (PL_curforce != where) { | |
5db06880 NC |
1973 | for (i = PL_lasttoke; i > where; --i) { |
1974 | PL_nexttoke[i] = PL_nexttoke[i-1]; | |
1975 | } | |
1976 | PL_lasttoke++; | |
1977 | } | |
cd81e915 | 1978 | if (PL_curforce < 0) /* in case of duplicate start_force() */ |
5db06880 | 1979 | Zero(&PL_nexttoke[where], 1, NEXTTOKE); |
cd81e915 NC |
1980 | PL_curforce = where; |
1981 | if (PL_nextwhite) { | |
5db06880 | 1982 | if (PL_madskills) |
6b29d1f5 | 1983 | curmad('^', newSVpvs("")); |
cd81e915 | 1984 | CURMAD('_', PL_nextwhite); |
5db06880 NC |
1985 | } |
1986 | } | |
1987 | ||
1988 | STATIC void | |
1989 | S_curmad(pTHX_ char slot, SV *sv) | |
1990 | { | |
1991 | MADPROP **where; | |
1992 | ||
1993 | if (!sv) | |
1994 | return; | |
cd81e915 NC |
1995 | if (PL_curforce < 0) |
1996 | where = &PL_thismad; | |
5db06880 | 1997 | else |
cd81e915 | 1998 | where = &PL_nexttoke[PL_curforce].next_mad; |
5db06880 | 1999 | |
cd81e915 | 2000 | if (PL_faketokens) |
76f68e9b | 2001 | sv_setpvs(sv, ""); |
5db06880 NC |
2002 | else { |
2003 | if (!IN_BYTES) { | |
2004 | if (UTF && is_utf8_string((U8*)SvPVX(sv), SvCUR(sv))) | |
2005 | SvUTF8_on(sv); | |
2006 | else if (PL_encoding) { | |
2007 | sv_recode_to_utf8(sv, PL_encoding); | |
2008 | } | |
2009 | } | |
2010 | } | |
2011 | ||
2012 | /* keep a slot open for the head of the list? */ | |
2013 | if (slot != '_' && *where && (*where)->mad_key == '^') { | |
2014 | (*where)->mad_key = slot; | |
daba3364 | 2015 | sv_free(MUTABLE_SV(((*where)->mad_val))); |
5db06880 NC |
2016 | (*where)->mad_val = (void*)sv; |
2017 | } | |
2018 | else | |
2019 | addmad(newMADsv(slot, sv), where, 0); | |
2020 | } | |
2021 | #else | |
b3f24c00 MHM |
2022 | # define start_force(where) NOOP |
2023 | # define curmad(slot, sv) NOOP | |
5db06880 NC |
2024 | #endif |
2025 | ||
ffb4593c NT |
2026 | /* |
2027 | * S_force_next | |
9cbb5ea2 | 2028 | * When the lexer realizes it knows the next token (for instance, |
ffb4593c | 2029 | * it is reordering tokens for the parser) then it can call S_force_next |
9cbb5ea2 | 2030 | * to know what token to return the next time the lexer is called. Caller |
5db06880 NC |
2031 | * will need to set PL_nextval[] (or PL_nexttoke[].next_val with PERL_MAD), |
2032 | * and possibly PL_expect to ensure the lexer handles the token correctly. | |
ffb4593c NT |
2033 | */ |
2034 | ||
4e553d73 | 2035 | STATIC void |
cea2e8a9 | 2036 | S_force_next(pTHX_ I32 type) |
79072805 | 2037 | { |
97aff369 | 2038 | dVAR; |
704d4215 GG |
2039 | #ifdef DEBUGGING |
2040 | if (DEBUG_T_TEST) { | |
2041 | PerlIO_printf(Perl_debug_log, "### forced token:\n"); | |
f05d7009 | 2042 | tokereport(type, &NEXTVAL_NEXTTOKE); |
704d4215 GG |
2043 | } |
2044 | #endif | |
5db06880 | 2045 | #ifdef PERL_MAD |
cd81e915 | 2046 | if (PL_curforce < 0) |
5db06880 | 2047 | start_force(PL_lasttoke); |
cd81e915 | 2048 | PL_nexttoke[PL_curforce].next_type = type; |
5db06880 NC |
2049 | if (PL_lex_state != LEX_KNOWNEXT) |
2050 | PL_lex_defer = PL_lex_state; | |
2051 | PL_lex_state = LEX_KNOWNEXT; | |
2052 | PL_lex_expect = PL_expect; | |
cd81e915 | 2053 | PL_curforce = -1; |
5db06880 | 2054 | #else |
3280af22 NIS |
2055 | PL_nexttype[PL_nexttoke] = type; |
2056 | PL_nexttoke++; | |
2057 | if (PL_lex_state != LEX_KNOWNEXT) { | |
2058 | PL_lex_defer = PL_lex_state; | |
2059 | PL_lex_expect = PL_expect; | |
2060 | PL_lex_state = LEX_KNOWNEXT; | |
79072805 | 2061 | } |
5db06880 | 2062 | #endif |
79072805 LW |
2063 | } |
2064 | ||
28ac2b49 Z |
2065 | void |
2066 | Perl_yyunlex(pTHX) | |
2067 | { | |
a7aaec61 Z |
2068 | int yyc = PL_parser->yychar; |
2069 | if (yyc != YYEMPTY) { | |
2070 | if (yyc) { | |
2071 | start_force(-1); | |
2072 | NEXTVAL_NEXTTOKE = PL_parser->yylval; | |
2073 | if (yyc == '{'/*}*/ || yyc == HASHBRACK || yyc == '['/*]*/) { | |
78cdf107 | 2074 | PL_lex_allbrackets--; |
a7aaec61 | 2075 | PL_lex_brackets--; |
78cdf107 Z |
2076 | yyc |= (3<<24) | (PL_lex_brackstack[PL_lex_brackets] << 16); |
2077 | } else if (yyc == '('/*)*/) { | |
2078 | PL_lex_allbrackets--; | |
2079 | yyc |= (2<<24); | |
a7aaec61 Z |
2080 | } |
2081 | force_next(yyc); | |
2082 | } | |
28ac2b49 Z |
2083 | PL_parser->yychar = YYEMPTY; |
2084 | } | |
2085 | } | |
2086 | ||
d0a148a6 | 2087 | STATIC SV * |
15f169a1 | 2088 | S_newSV_maybe_utf8(pTHX_ const char *const start, STRLEN len) |
d0a148a6 | 2089 | { |
97aff369 | 2090 | dVAR; |
740cce10 | 2091 | SV * const sv = newSVpvn_utf8(start, len, |
eaf7a4d2 CS |
2092 | !IN_BYTES |
2093 | && UTF | |
2094 | && !is_ascii_string((const U8*)start, len) | |
740cce10 | 2095 | && is_utf8_string((const U8*)start, len)); |
d0a148a6 NC |
2096 | return sv; |
2097 | } | |
2098 | ||
ffb4593c NT |
2099 | /* |
2100 | * S_force_word | |
2101 | * When the lexer knows the next thing is a word (for instance, it has | |
2102 | * just seen -> and it knows that the next char is a word char, then | |
02b34bbe DM |
2103 | * it calls S_force_word to stick the next word into the PL_nexttoke/val |
2104 | * lookahead. | |
ffb4593c NT |
2105 | * |
2106 | * Arguments: | |
b1b65b59 | 2107 | * char *start : buffer position (must be within PL_linestr) |
02b34bbe | 2108 | * int token : PL_next* will be this type of bare word (e.g., METHOD,WORD) |
ffb4593c NT |
2109 | * int check_keyword : if true, Perl checks to make sure the word isn't |
2110 | * a keyword (do this if the word is a label, e.g. goto FOO) | |
2111 | * int allow_pack : if true, : characters will also be allowed (require, | |
2112 | * use, etc. do this) | |
9cbb5ea2 | 2113 | * int allow_initial_tick : used by the "sub" lexer only. |
ffb4593c NT |
2114 | */ |
2115 | ||
76e3520e | 2116 | STATIC char * |
5aaab254 | 2117 | S_force_word(pTHX_ char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick) |
79072805 | 2118 | { |
97aff369 | 2119 | dVAR; |
eb578fdb | 2120 | char *s; |
463ee0b2 | 2121 | STRLEN len; |
4e553d73 | 2122 | |
7918f24d NC |
2123 | PERL_ARGS_ASSERT_FORCE_WORD; |
2124 | ||
29595ff2 | 2125 | start = SKIPSPACE1(start); |
463ee0b2 | 2126 | s = start; |
7e2040f0 | 2127 | if (isIDFIRST_lazy_if(s,UTF) || |
a0d0e21e | 2128 | (allow_pack && *s == ':') || |
15f0808c | 2129 | (allow_initial_tick && *s == '\'') ) |
a0d0e21e | 2130 | { |
3280af22 | 2131 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len); |
5458a98a | 2132 | if (check_keyword && keyword(PL_tokenbuf, len, 0)) |
463ee0b2 | 2133 | return start; |
cd81e915 | 2134 | start_force(PL_curforce); |
5db06880 NC |
2135 | if (PL_madskills) |
2136 | curmad('X', newSVpvn(start,s-start)); | |
463ee0b2 | 2137 | if (token == METHOD) { |
29595ff2 | 2138 | s = SKIPSPACE1(s); |
463ee0b2 | 2139 | if (*s == '(') |
3280af22 | 2140 | PL_expect = XTERM; |
463ee0b2 | 2141 | else { |
3280af22 | 2142 | PL_expect = XOPERATOR; |
463ee0b2 | 2143 | } |
79072805 | 2144 | } |
e74e6b3d | 2145 | if (PL_madskills) |
63575281 | 2146 | curmad('g', newSVpvs( "forced" )); |
9ded7720 | 2147 | NEXTVAL_NEXTTOKE.opval |
d0a148a6 NC |
2148 | = (OP*)newSVOP(OP_CONST,0, |
2149 | S_newSV_maybe_utf8(aTHX_ PL_tokenbuf, len)); | |
9ded7720 | 2150 | NEXTVAL_NEXTTOKE.opval->op_private |= OPpCONST_BARE; |
79072805 LW |
2151 | force_next(token); |
2152 | } | |
2153 | return s; | |
2154 | } | |
2155 | ||
ffb4593c NT |
2156 | /* |
2157 | * S_force_ident | |
9cbb5ea2 | 2158 | * Called when the lexer wants $foo *foo &foo etc, but the program |
ffb4593c NT |
2159 | * text only contains the "foo" portion. The first argument is a pointer |
2160 | * to the "foo", and the second argument is the type symbol to prefix. | |
2161 | * Forces the next token to be a "WORD". | |
9cbb5ea2 | 2162 | * Creates the symbol if it didn't already exist (via gv_fetchpv()). |
ffb4593c NT |
2163 | */ |
2164 | ||
76e3520e | 2165 | STATIC void |
5aaab254 | 2166 | S_force_ident(pTHX_ const char *s, int kind) |
79072805 | 2167 | { |
97aff369 | 2168 | dVAR; |
7918f24d NC |
2169 | |
2170 | PERL_ARGS_ASSERT_FORCE_IDENT; | |
2171 | ||
c9b48522 DD |
2172 | if (s[0]) { |
2173 | const STRLEN len = s[1] ? strlen(s) : 1; /* s = "\"" see yylex */ | |
728847b1 BF |
2174 | OP* const o = (OP*)newSVOP(OP_CONST, 0, newSVpvn_flags(s, len, |
2175 | UTF ? SVf_UTF8 : 0)); | |
cd81e915 | 2176 | start_force(PL_curforce); |
9ded7720 | 2177 | NEXTVAL_NEXTTOKE.opval = o; |
79072805 | 2178 | force_next(WORD); |
748a9306 | 2179 | if (kind) { |
11343788 | 2180 | o->op_private = OPpCONST_ENTERED; |
55497cff | 2181 | /* XXX see note in pp_entereval() for why we forgo typo |
2182 | warnings if the symbol must be introduced in an eval. | |
2183 | GSAR 96-10-12 */ | |
90e5519e | 2184 | gv_fetchpvn_flags(s, len, |
728847b1 BF |
2185 | (PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) |
2186 | : GV_ADD) | ( UTF ? SVf_UTF8 : 0 ), | |
90e5519e NC |
2187 | kind == '$' ? SVt_PV : |
2188 | kind == '@' ? SVt_PVAV : | |
2189 | kind == '%' ? SVt_PVHV : | |
a0d0e21e | 2190 | SVt_PVGV |
90e5519e | 2191 | ); |
748a9306 | 2192 | } |
79072805 LW |
2193 | } |
2194 | } | |
2195 | ||
3f33d153 FC |
2196 | static void |
2197 | S_force_ident_maybe_lex(pTHX_ char pit) | |
2198 | { | |
2199 | start_force(PL_curforce); | |
2200 | NEXTVAL_NEXTTOKE.ival = pit; | |
2201 | force_next('p'); | |
2202 | } | |
2203 | ||
1571675a GS |
2204 | NV |
2205 | Perl_str_to_version(pTHX_ SV *sv) | |
2206 | { | |
2207 | NV retval = 0.0; | |
2208 | NV nshift = 1.0; | |
2209 | STRLEN len; | |
cfd0369c | 2210 | const char *start = SvPV_const(sv,len); |
9d4ba2ae | 2211 | const char * const end = start + len; |
504618e9 | 2212 | const bool utf = SvUTF8(sv) ? TRUE : FALSE; |
7918f24d NC |
2213 | |
2214 | PERL_ARGS_ASSERT_STR_TO_VERSION; | |
2215 | ||
1571675a | 2216 | while (start < end) { |
ba210ebe | 2217 | STRLEN skip; |
1571675a GS |
2218 | UV n; |
2219 | if (utf) | |
9041c2e3 | 2220 | n = utf8n_to_uvchr((U8*)start, len, &skip, 0); |
1571675a GS |
2221 | else { |
2222 | n = *(U8*)start; | |
2223 | skip = 1; | |
2224 | } | |
2225 | retval += ((NV)n)/nshift; | |
2226 | start += skip; | |
2227 | nshift *= 1000; | |
2228 | } | |
2229 | return retval; | |
2230 | } | |
2231 | ||
4e553d73 | 2232 | /* |
ffb4593c NT |
2233 | * S_force_version |
2234 | * Forces the next token to be a version number. | |
e759cc13 RGS |
2235 | * If the next token appears to be an invalid version number, (e.g. "v2b"), |
2236 | * and if "guessing" is TRUE, then no new token is created (and the caller | |
2237 | * must use an alternative parsing method). | |
ffb4593c NT |
2238 | */ |
2239 | ||
76e3520e | 2240 | STATIC char * |
e759cc13 | 2241 | S_force_version(pTHX_ char *s, int guessing) |
89bfa8cd | 2242 | { |
97aff369 | 2243 | dVAR; |
5f66b61c | 2244 | OP *version = NULL; |
44dcb63b | 2245 | char *d; |
5db06880 NC |
2246 | #ifdef PERL_MAD |
2247 | I32 startoff = s - SvPVX(PL_linestr); | |
2248 | #endif | |
89bfa8cd | 2249 | |
7918f24d NC |
2250 | PERL_ARGS_ASSERT_FORCE_VERSION; |
2251 | ||
29595ff2 | 2252 | s = SKIPSPACE1(s); |
89bfa8cd | 2253 | |
44dcb63b | 2254 | d = s; |
dd629d5b | 2255 | if (*d == 'v') |
44dcb63b | 2256 | d++; |
44dcb63b | 2257 | if (isDIGIT(*d)) { |
e759cc13 RGS |
2258 | while (isDIGIT(*d) || *d == '_' || *d == '.') |
2259 | d++; | |
5db06880 NC |
2260 | #ifdef PERL_MAD |
2261 | if (PL_madskills) { | |
cd81e915 | 2262 | start_force(PL_curforce); |
5db06880 NC |
2263 | curmad('X', newSVpvn(s,d-s)); |
2264 | } | |
2265 | #endif | |
4e4da3ac | 2266 | if (*d == ';' || isSPACE(*d) || *d == '{' || *d == '}' || !*d) { |
dd629d5b | 2267 | SV *ver; |
8d08d9ba | 2268 | #ifdef USE_LOCALE_NUMERIC |
909d3787 KW |
2269 | char *loc = savepv(setlocale(LC_NUMERIC, NULL)); |
2270 | setlocale(LC_NUMERIC, "C"); | |
8d08d9ba | 2271 | #endif |
6154021b | 2272 | s = scan_num(s, &pl_yylval); |
8d08d9ba DG |
2273 | #ifdef USE_LOCALE_NUMERIC |
2274 | setlocale(LC_NUMERIC, loc); | |
909d3787 | 2275 | Safefree(loc); |
8d08d9ba | 2276 | #endif |
6154021b | 2277 | version = pl_yylval.opval; |
dd629d5b GS |
2278 | ver = cSVOPx(version)->op_sv; |
2279 | if (SvPOK(ver) && !SvNIOK(ver)) { | |
862a34c6 | 2280 | SvUPGRADE(ver, SVt_PVNV); |
9d6ce603 | 2281 | SvNV_set(ver, str_to_version(ver)); |
1571675a | 2282 | SvNOK_on(ver); /* hint that it is a version */ |
44dcb63b | 2283 | } |
89bfa8cd | 2284 | } |
5db06880 NC |
2285 | else if (guessing) { |
2286 | #ifdef PERL_MAD | |
2287 | if (PL_madskills) { | |
cd81e915 NC |
2288 | sv_free(PL_nextwhite); /* let next token collect whitespace */ |
2289 | PL_nextwhite = 0; | |
5db06880 NC |
2290 | s = SvPVX(PL_linestr) + startoff; |
2291 | } | |
2292 | #endif | |
e759cc13 | 2293 | return s; |
5db06880 | 2294 | } |
89bfa8cd | 2295 | } |
2296 | ||
5db06880 NC |
2297 | #ifdef PERL_MAD |
2298 | if (PL_madskills && !version) { | |
cd81e915 NC |
2299 | sv_free(PL_nextwhite); /* let next token collect whitespace */ |
2300 | PL_nextwhite = 0; | |
5db06880 NC |
2301 | s = SvPVX(PL_linestr) + startoff; |
2302 | } | |
2303 | #endif | |
89bfa8cd | 2304 | /* NOTE: The parser sees the package name and the VERSION swapped */ |
cd81e915 | 2305 | start_force(PL_curforce); |
9ded7720 | 2306 | NEXTVAL_NEXTTOKE.opval = version; |
4e553d73 | 2307 | force_next(WORD); |
89bfa8cd | 2308 | |
e759cc13 | 2309 | return s; |
89bfa8cd | 2310 | } |
2311 | ||
ffb4593c | 2312 | /* |
91152fc1 DG |
2313 | * S_force_strict_version |
2314 | * Forces the next token to be a version number using strict syntax rules. | |
2315 | */ | |
2316 | ||
2317 | STATIC char * | |
2318 | S_force_strict_version(pTHX_ char *s) | |
2319 | { | |
2320 | dVAR; | |
2321 | OP *version = NULL; | |
2322 | #ifdef PERL_MAD | |
2323 | I32 startoff = s - SvPVX(PL_linestr); | |
2324 | #endif | |
2325 | const char *errstr = NULL; | |
2326 | ||
2327 | PERL_ARGS_ASSERT_FORCE_STRICT_VERSION; | |
2328 | ||
2329 | while (isSPACE(*s)) /* leading whitespace */ | |
2330 | s++; | |
2331 | ||
2332 | if (is_STRICT_VERSION(s,&errstr)) { | |
2333 | SV *ver = newSV(0); | |
2334 | s = (char *)scan_version(s, ver, 0); | |
2335 | version = newSVOP(OP_CONST, 0, ver); | |
2336 | } | |
4e4da3ac Z |
2337 | else if ( (*s != ';' && *s != '{' && *s != '}' ) && |
2338 | (s = SKIPSPACE1(s), (*s != ';' && *s != '{' && *s != '}' ))) | |
2339 | { | |
91152fc1 DG |
2340 | PL_bufptr = s; |
2341 | if (errstr) | |
2342 | yyerror(errstr); /* version required */ | |
2343 | return s; | |
2344 | } | |
2345 | ||
2346 | #ifdef PERL_MAD | |
2347 | if (PL_madskills && !version) { | |
2348 | sv_free(PL_nextwhite); /* let next token collect whitespace */ | |
2349 | PL_nextwhite = 0; | |
2350 | s = SvPVX(PL_linestr) + startoff; | |
2351 | } | |
2352 | #endif | |
2353 | /* NOTE: The parser sees the package name and the VERSION swapped */ | |
2354 | start_force(PL_curforce); | |
2355 | NEXTVAL_NEXTTOKE.opval = version; | |
2356 | force_next(WORD); | |
2357 | ||
2358 | return s; | |
2359 | } | |
2360 | ||
2361 | /* | |
ffb4593c NT |
2362 | * S_tokeq |
2363 | * Tokenize a quoted string passed in as an SV. It finds the next | |
2364 | * chunk, up to end of string or a backslash. It may make a new | |
2365 | * SV containing that chunk (if HINT_NEW_STRING is on). It also | |
2366 | * turns \\ into \. | |
2367 | */ | |
2368 | ||
76e3520e | 2369 | STATIC SV * |
cea2e8a9 | 2370 | S_tokeq(pTHX_ SV *sv) |
79072805 | 2371 | { |
97aff369 | 2372 | dVAR; |
eb578fdb KW |
2373 | char *s; |
2374 | char *send; | |
2375 | char *d; | |
b3ac6de7 IZ |
2376 | STRLEN len = 0; |
2377 | SV *pv = sv; | |
79072805 | 2378 | |
7918f24d NC |
2379 | PERL_ARGS_ASSERT_TOKEQ; |
2380 | ||
79072805 | 2381 | if (!SvLEN(sv)) |
b3ac6de7 | 2382 | goto finish; |
79072805 | 2383 | |
a0d0e21e | 2384 | s = SvPV_force(sv, len); |
21a311ee | 2385 | if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1) |
b3ac6de7 | 2386 | goto finish; |
463ee0b2 | 2387 | send = s + len; |
dcb21ed6 NC |
2388 | /* This is relying on the SV being "well formed" with a trailing '\0' */ |
2389 | while (s < send && !(*s == '\\' && s[1] == '\\')) | |
79072805 LW |
2390 | s++; |
2391 | if (s == send) | |
b3ac6de7 | 2392 | goto finish; |
79072805 | 2393 | d = s; |
be4731d2 | 2394 | if ( PL_hints & HINT_NEW_STRING ) { |
59cd0e26 | 2395 | pv = newSVpvn_flags(SvPVX_const(pv), len, SVs_TEMP | SvUTF8(sv)); |
be4731d2 | 2396 | } |
79072805 LW |
2397 | while (s < send) { |
2398 | if (*s == '\\') { | |
a0d0e21e | 2399 | if (s + 1 < send && (s[1] == '\\')) |
79072805 LW |
2400 | s++; /* all that, just for this */ |
2401 | } | |
2402 | *d++ = *s++; | |
2403 | } | |
2404 | *d = '\0'; | |
95a20fc0 | 2405 | SvCUR_set(sv, d - SvPVX_const(sv)); |
b3ac6de7 | 2406 | finish: |
3280af22 | 2407 | if ( PL_hints & HINT_NEW_STRING ) |
eb0d8d16 | 2408 | return new_constant(NULL, 0, "q", sv, pv, "q", 1); |
79072805 LW |
2409 | return sv; |
2410 | } | |
2411 | ||
ffb4593c NT |
2412 | /* |
2413 | * Now come three functions related to double-quote context, | |
2414 | * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when | |
2415 | * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They | |
2416 | * interact with PL_lex_state, and create fake ( ... ) argument lists | |
2417 | * to handle functions and concatenation. | |
ecd24171 DM |
2418 | * For example, |
2419 | * "foo\lbar" | |
2420 | * is tokenised as | |
2421 | * stringify ( const[foo] concat lcfirst ( const[bar] ) ) | |
ffb4593c NT |
2422 | */ |
2423 | ||
2424 | /* | |
2425 | * S_sublex_start | |
6154021b | 2426 | * Assumes that pl_yylval.ival is the op we're creating (e.g. OP_LCFIRST). |
ffb4593c NT |
2427 | * |
2428 | * Pattern matching will set PL_lex_op to the pattern-matching op to | |
6154021b | 2429 | * make (we return THING if pl_yylval.ival is OP_NULL, PMFUNC otherwise). |
ffb4593c NT |
2430 | * |
2431 | * OP_CONST and OP_READLINE are easy--just make the new op and return. | |
2432 | * | |
2433 | * Everything else becomes a FUNC. | |
2434 | * | |
2435 | * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we | |
2436 | * had an OP_CONST or OP_READLINE). This just sets us up for a | |
2437 | * call to S_sublex_push(). | |
2438 | */ | |
2439 | ||
76e3520e | 2440 | STATIC I32 |
cea2e8a9 | 2441 | S_sublex_start(pTHX) |
79072805 | 2442 | { |
97aff369 | 2443 | dVAR; |
eb578fdb | 2444 | const I32 op_type = pl_yylval.ival; |
79072805 LW |
2445 | |
2446 | if (op_type == OP_NULL) { | |
6154021b | 2447 | pl_yylval.opval = PL_lex_op; |
5f66b61c | 2448 | PL_lex_op = NULL; |
79072805 LW |
2449 | return THING; |
2450 | } | |
2451 | if (op_type == OP_CONST || op_type == OP_READLINE) { | |
3280af22 | 2452 | SV *sv = tokeq(PL_lex_stuff); |
b3ac6de7 IZ |
2453 | |
2454 | if (SvTYPE(sv) == SVt_PVIV) { | |
2455 | /* Overloaded constants, nothing fancy: Convert to SVt_PV: */ | |
2456 | STRLEN len; | |
96a5add6 | 2457 | const char * const p = SvPV_const(sv, len); |
740cce10 | 2458 | SV * const nsv = newSVpvn_flags(p, len, SvUTF8(sv)); |
b3ac6de7 IZ |
2459 | SvREFCNT_dec(sv); |
2460 | sv = nsv; | |
4e553d73 | 2461 | } |
6154021b | 2462 | pl_yylval.opval = (OP*)newSVOP(op_type, 0, sv); |
a0714e2c | 2463 | PL_lex_stuff = NULL; |
6f33ba73 RGS |
2464 | /* Allow <FH> // "foo" */ |
2465 | if (op_type == OP_READLINE) | |
2466 | PL_expect = XTERMORDORDOR; | |
79072805 LW |
2467 | return THING; |
2468 | } | |
e3f73d4e RGS |
2469 | else if (op_type == OP_BACKTICK && PL_lex_op) { |
2470 | /* readpipe() vas overriden */ | |
2471 | cSVOPx(cLISTOPx(cUNOPx(PL_lex_op)->op_first)->op_first->op_sibling)->op_sv = tokeq(PL_lex_stuff); | |
6154021b | 2472 | pl_yylval.opval = PL_lex_op; |
9b201d7d | 2473 | PL_lex_op = NULL; |
e3f73d4e RGS |
2474 | PL_lex_stuff = NULL; |
2475 | return THING; | |
2476 | } | |
79072805 | 2477 | |
3280af22 | 2478 | PL_sublex_info.super_state = PL_lex_state; |
eac04b2e | 2479 | PL_sublex_info.sub_inwhat = (U16)op_type; |
3280af22 NIS |
2480 | PL_sublex_info.sub_op = PL_lex_op; |
2481 | PL_lex_state = LEX_INTERPPUSH; | |
55497cff | 2482 | |
3280af22 NIS |
2483 | PL_expect = XTERM; |
2484 | if (PL_lex_op) { | |
6154021b | 2485 | pl_yylval.opval = PL_lex_op; |
5f66b61c | 2486 | PL_lex_op = NULL; |
55497cff | 2487 | return PMFUNC; |
2488 | } | |
2489 | else | |
2490 | return FUNC; | |
2491 | } | |
2492 | ||
ffb4593c NT |
2493 | /* |
2494 | * S_sublex_push | |
2495 | * Create a new scope to save the lexing state. The scope will be | |
2496 | * ended in S_sublex_done. Returns a '(', starting the function arguments | |
2497 | * to the uc, lc, etc. found before. | |
2498 | * Sets PL_lex_state to LEX_INTERPCONCAT. | |
2499 | */ | |
2500 | ||
76e3520e | 2501 | STATIC I32 |
cea2e8a9 | 2502 | S_sublex_push(pTHX) |
55497cff | 2503 | { |
27da23d5 | 2504 | dVAR; |
78a635de | 2505 | LEXSHARED *shared; |
f46d017c | 2506 | ENTER; |
55497cff | 2507 | |
3280af22 | 2508 | PL_lex_state = PL_sublex_info.super_state; |
651b5b28 | 2509 | SAVEBOOL(PL_lex_dojoin); |
3280af22 | 2510 | SAVEI32(PL_lex_brackets); |
78cdf107 | 2511 | SAVEI32(PL_lex_allbrackets); |
b27dce25 | 2512 | SAVEI32(PL_lex_formbrack); |
78cdf107 | 2513 | SAVEI8(PL_lex_fakeeof); |
3280af22 NIS |
2514 | SAVEI32(PL_lex_casemods); |
2515 | SAVEI32(PL_lex_starts); | |
651b5b28 | 2516 | SAVEI8(PL_lex_state); |
7cc34111 | 2517 | SAVESPTR(PL_lex_repl); |
7766f137 | 2518 | SAVEVPTR(PL_lex_inpat); |
98246f1e | 2519 | SAVEI16(PL_lex_inwhat); |
57843af0 | 2520 | SAVECOPLINE(PL_curcop); |
3280af22 | 2521 | SAVEPPTR(PL_bufptr); |
8452ff4b | 2522 | SAVEPPTR(PL_bufend); |
3280af22 NIS |
2523 | SAVEPPTR(PL_oldbufptr); |
2524 | SAVEPPTR(PL_oldoldbufptr); | |
207e3d1a JH |
2525 | SAVEPPTR(PL_last_lop); |
2526 | SAVEPPTR(PL_last_uni); | |
3280af22 NIS |
2527 | SAVEPPTR(PL_linestart); |
2528 | SAVESPTR(PL_linestr); | |
8edd5f42 RGS |
2529 | SAVEGENERICPV(PL_lex_brackstack); |
2530 | SAVEGENERICPV(PL_lex_casestack); | |
78a635de | 2531 | SAVEGENERICPV(PL_parser->lex_shared); |
3a54fd60 | 2532 | SAVEBOOL(PL_parser->lex_re_reparsing); |
3280af22 | 2533 | |
99bd9d90 | 2534 | /* The here-doc parser needs to be able to peek into outer lexing |
60f40a38 FC |
2535 | scopes to find the body of the here-doc. So we put PL_linestr and |
2536 | PL_bufptr into lex_shared, to ‘share’ those values. | |
99bd9d90 | 2537 | */ |
60f40a38 FC |
2538 | PL_parser->lex_shared->ls_linestr = PL_linestr; |
2539 | PL_parser->lex_shared->ls_bufptr = PL_bufptr; | |
99bd9d90 | 2540 | |
3280af22 | 2541 | PL_linestr = PL_lex_stuff; |
7cc34111 | 2542 | PL_lex_repl = PL_sublex_info.repl; |
a0714e2c | 2543 | PL_lex_stuff = NULL; |
7cc34111 | 2544 | PL_sublex_info.repl = NULL; |
3280af22 | 2545 | |
9cbb5ea2 GS |
2546 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart |
2547 | = SvPVX(PL_linestr); | |
3280af22 | 2548 | PL_bufend += SvCUR(PL_linestr); |
bd61b366 | 2549 | PL_last_lop = PL_last_uni = NULL; |
3280af22 | 2550 | SAVEFREESV(PL_linestr); |
4dc843bc | 2551 | if (PL_lex_repl) SAVEFREESV(PL_lex_repl); |
3280af22 NIS |
2552 | |
2553 | PL_lex_dojoin = FALSE; | |
b27dce25 | 2554 | PL_lex_brackets = PL_lex_formbrack = 0; |
78cdf107 Z |
2555 | PL_lex_allbrackets = 0; |
2556 | PL_lex_fakeeof = LEX_FAKEEOF_NEVER; | |
a02a5408 JC |
2557 | Newx(PL_lex_brackstack, 120, char); |
2558 | Newx(PL_lex_casestack, 12, char); | |
3280af22 NIS |
2559 | PL_lex_casemods = 0; |
2560 | *PL_lex_casestack = '\0'; | |
2561 | PL_lex_starts = 0; | |
2562 | PL_lex_state = LEX_INTERPCONCAT; | |
eb160463 | 2563 | CopLINE_set(PL_curcop, (line_t)PL_multi_start); |
78a635de FC |
2564 | |
2565 | Newxz(shared, 1, LEXSHARED); | |
2566 | shared->ls_prev = PL_parser->lex_shared; | |
2567 | PL_parser->lex_shared = shared; | |
3280af22 NIS |
2568 | |
2569 | PL_lex_inwhat = PL_sublex_info.sub_inwhat; | |
bb16bae8 | 2570 | if (PL_lex_inwhat == OP_TRANSR) PL_lex_inwhat = OP_TRANS; |
3280af22 NIS |
2571 | if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST) |
2572 | PL_lex_inpat = PL_sublex_info.sub_op; | |
79072805 | 2573 | else |
5f66b61c | 2574 | PL_lex_inpat = NULL; |
79072805 | 2575 | |
3a54fd60 DM |
2576 | PL_parser->lex_re_reparsing = cBOOL(PL_in_eval & EVAL_RE_REPARSING); |
2577 | PL_in_eval &= ~EVAL_RE_REPARSING; | |
2578 | ||
55497cff | 2579 | return '('; |
79072805 LW |
2580 | } |
2581 | ||
ffb4593c NT |
2582 | /* |
2583 | * S_sublex_done | |
2584 | * Restores lexer state after a S_sublex_push. | |
2585 | */ | |
2586 | ||
76e3520e | 2587 | STATIC I32 |
cea2e8a9 | 2588 | S_sublex_done(pTHX) |
79072805 | 2589 | { |
27da23d5 | 2590 | dVAR; |
3280af22 | 2591 | if (!PL_lex_starts++) { |
396482e1 | 2592 | SV * const sv = newSVpvs(""); |
9aa983d2 JH |
2593 | if (SvUTF8(PL_linestr)) |
2594 | SvUTF8_on(sv); | |
3280af22 | 2595 | PL_expect = XOPERATOR; |
6154021b | 2596 | pl_yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
79072805 LW |
2597 | return THING; |
2598 | } | |
2599 | ||
3280af22 NIS |
2600 | if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */ |
2601 | PL_lex_state = LEX_INTERPCASEMOD; | |
cea2e8a9 | 2602 | return yylex(); |
79072805 LW |
2603 | } |
2604 | ||
ffb4593c | 2605 | /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */ |
bb16bae8 | 2606 | assert(PL_lex_inwhat != OP_TRANSR); |
3280af22 NIS |
2607 | if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) { |
2608 | PL_linestr = PL_lex_repl; | |
2609 | PL_lex_inpat = 0; | |
2610 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr); | |
2611 | PL_bufend += SvCUR(PL_linestr); | |
bd61b366 | 2612 | PL_last_lop = PL_last_uni = NULL; |
3280af22 NIS |
2613 | PL_lex_dojoin = FALSE; |
2614 | PL_lex_brackets = 0; | |
78cdf107 Z |
2615 | PL_lex_allbrackets = 0; |
2616 | PL_lex_fakeeof = LEX_FAKEEOF_NEVER; | |
3280af22 NIS |
2617 | PL_lex_casemods = 0; |
2618 | *PL_lex_casestack = '\0'; | |
2619 | PL_lex_starts = 0; | |
25da4f38 | 2620 | if (SvEVALED(PL_lex_repl)) { |
3280af22 NIS |
2621 | PL_lex_state = LEX_INTERPNORMAL; |
2622 | PL_lex_starts++; | |
e9fa98b2 HS |
2623 | /* we don't clear PL_lex_repl here, so that we can check later |
2624 | whether this is an evalled subst; that means we rely on the | |
2625 | logic to ensure sublex_done() is called again only via the | |
2626 | branch (in yylex()) that clears PL_lex_repl, else we'll loop */ | |
79072805 | 2627 | } |
e9fa98b2 | 2628 | else { |
3280af22 | 2629 | PL_lex_state = LEX_INTERPCONCAT; |
a0714e2c | 2630 | PL_lex_repl = NULL; |
e9fa98b2 | 2631 | } |
79072805 | 2632 | return ','; |
ffed7fef LW |
2633 | } |
2634 | else { | |
5db06880 NC |
2635 | #ifdef PERL_MAD |
2636 | if (PL_madskills) { | |
cd81e915 NC |
2637 | if (PL_thiswhite) { |
2638 | if (!PL_endwhite) | |
6b29d1f5 | 2639 | PL_endwhite = newSVpvs(""); |
cd81e915 NC |
2640 | sv_catsv(PL_endwhite, PL_thiswhite); |
2641 | PL_thiswhite = 0; | |
2642 | } | |
2643 | if (PL_thistoken) | |
76f68e9b | 2644 | sv_setpvs(PL_thistoken,""); |
5db06880 | 2645 | else |
cd81e915 | 2646 | PL_realtokenstart = -1; |
5db06880 NC |
2647 | } |
2648 | #endif | |
f46d017c | 2649 | LEAVE; |
3280af22 NIS |
2650 | PL_bufend = SvPVX(PL_linestr); |
2651 | PL_bufend += SvCUR(PL_linestr); | |
2652 | PL_expect = XOPERATOR; | |
09bef843 | 2653 | PL_sublex_info.sub_inwhat = 0; |
79072805 | 2654 | return ')'; |
ffed7fef LW |
2655 | } |
2656 | } | |
2657 | ||
6f613c73 KW |
2658 | PERL_STATIC_INLINE SV* |
2659 | S_get_and_check_backslash_N_name(pTHX_ const char* s, const char* const e) | |
2660 | { | |
140b12ad KW |
2661 | /* <s> points to first character of interior of \N{}, <e> to one beyond the |
2662 | * interior, hence to the "}". Finds what the name resolves to, returning | |
2663 | * an SV* containing it; NULL if no valid one found */ | |
2664 | ||
dd2b1b72 | 2665 | SV* res = newSVpvn_flags(s, e - s, UTF ? SVf_UTF8 : 0); |
6f613c73 | 2666 | |
0c415a79 KW |
2667 | HV * table; |
2668 | SV **cvp; | |
2669 | SV *cv; | |
2670 | SV *rv; | |
2671 | HV *stash; | |
2672 | const U8* first_bad_char_loc; | |
2673 | const char* backslash_ptr = s - 3; /* Points to the <\> of \N{... */ | |
2674 | ||
6f613c73 KW |
2675 | PERL_ARGS_ASSERT_GET_AND_CHECK_BACKSLASH_N_NAME; |
2676 | ||
107160e2 KW |
2677 | if (UTF && ! is_utf8_string_loc((U8 *) backslash_ptr, |
2678 | e - backslash_ptr, | |
2679 | &first_bad_char_loc)) | |
2680 | { | |
2681 | /* If warnings are on, this will print a more detailed analysis of what | |
2682 | * is wrong than the error message below */ | |
2683 | utf8n_to_uvuni(first_bad_char_loc, | |
2684 | e - ((char *) first_bad_char_loc), | |
2685 | NULL, 0); | |
2686 | ||
2687 | /* We deliberately don't try to print the malformed character, which | |
2688 | * might not print very well; it also may be just the first of many | |
2689 | * malformations, so don't print what comes after it */ | |
2690 | yyerror(Perl_form(aTHX_ | |
2691 | "Malformed UTF-8 character immediately after '%.*s'", | |
2692 | (int) (first_bad_char_loc - (U8 *) backslash_ptr), backslash_ptr)); | |
2693 | return NULL; | |
2694 | } | |
2695 | ||
2696 | res = new_constant( NULL, 0, "charnames", res, NULL, backslash_ptr, | |
2697 | /* include the <}> */ | |
2698 | e - backslash_ptr + 1); | |
6f613c73 | 2699 | if (! SvPOK(res)) { |
b6407c49 | 2700 | SvREFCNT_dec_NN(res); |
6f613c73 KW |
2701 | return NULL; |
2702 | } | |
2703 | ||
0c415a79 KW |
2704 | /* See if the charnames handler is the Perl core's, and if so, we can skip |
2705 | * the validation needed for a user-supplied one, as Perl's does its own | |
2706 | * validation. */ | |
2707 | table = GvHV(PL_hintgv); /* ^H */ | |
2708 | cvp = hv_fetchs(table, "charnames", FALSE); | |
67a057d6 FC |
2709 | if (cvp && (cv = *cvp) && SvROK(cv) && ((rv = SvRV(cv)) != NULL) |
2710 | && SvTYPE(rv) == SVt_PVCV && ((stash = CvSTASH(rv)) != NULL)) | |
0c415a79 KW |
2711 | { |
2712 | const char * const name = HvNAME(stash); | |
2713 | if strEQ(name, "_charnames") { | |
2714 | return res; | |
2715 | } | |
2716 | } | |
2717 | ||
bde9e88d KW |
2718 | /* Here, it isn't Perl's charname handler. We can't rely on a |
2719 | * user-supplied handler to validate the input name. For non-ut8 input, | |
2720 | * look to see that the first character is legal. Then loop through the | |
2721 | * rest checking that each is a continuation */ | |
6f613c73 | 2722 | |
b6ba1137 KW |
2723 | /* This code needs to be sync'ed with a regex in _charnames.pm which does |
2724 | * the same thing */ | |
2725 | ||
b6ba1137 | 2726 | if (! UTF) { |
bde9e88d | 2727 | if (! isALPHAU(*s)) { |
b6ba1137 KW |
2728 | goto bad_charname; |
2729 | } | |
bde9e88d KW |
2730 | s++; |
2731 | while (s < e) { | |
2732 | if (! isCHARNAME_CONT(*s)) { | |
b6ba1137 KW |
2733 | goto bad_charname; |
2734 | } | |
3036c853 | 2735 | if (*s == ' ' && *(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) { |
94ec3a20 FC |
2736 | Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), |
2737 | "A sequence of multiple spaces in a charnames " | |
2738 | "alias definition is deprecated"); | |
bd299e29 | 2739 | } |
bde9e88d | 2740 | s++; |
b6ba1137 | 2741 | } |
3036c853 | 2742 | if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) { |
94ec3a20 FC |
2743 | Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), |
2744 | "Trailing white-space in a charnames alias " | |
2745 | "definition is deprecated"); | |
bd299e29 | 2746 | } |
b6ba1137 KW |
2747 | } |
2748 | else { | |
bde9e88d KW |
2749 | /* Similarly for utf8. For invariants can check directly; for other |
2750 | * Latin1, can calculate their code point and check; otherwise use a | |
2751 | * swash */ | |
2752 | if (UTF8_IS_INVARIANT(*s)) { | |
2753 | if (! isALPHAU(*s)) { | |
140b12ad KW |
2754 | goto bad_charname; |
2755 | } | |
bde9e88d KW |
2756 | s++; |
2757 | } else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { | |
2758 | if (! isALPHAU(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s, *(s+1))))) { | |
b6ba1137 | 2759 | goto bad_charname; |
6f613c73 | 2760 | } |
bde9e88d | 2761 | s += 2; |
6f613c73 | 2762 | } |
bde9e88d KW |
2763 | else { |
2764 | if (! PL_utf8_charname_begin) { | |
2765 | U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST; | |
2766 | PL_utf8_charname_begin = _core_swash_init("utf8", | |
2767 | "_Perl_Charname_Begin", | |
2768 | &PL_sv_undef, | |
2769 | 1, 0, NULL, &flags); | |
2770 | } | |
2771 | if (! swash_fetch(PL_utf8_charname_begin, (U8 *) s, TRUE)) { | |
2772 | goto bad_charname; | |
2773 | } | |
2774 | s += UTF8SKIP(s); | |
2775 | } | |
2776 | ||
2777 | while (s < e) { | |
2778 | if (UTF8_IS_INVARIANT(*s)) { | |
2779 | if (! isCHARNAME_CONT(*s)) { | |
2780 | goto bad_charname; | |
2781 | } | |
3036c853 FC |
2782 | if (*s == ' ' && *(s-1) == ' ' |
2783 | && ckWARN_d(WARN_DEPRECATED)) { | |
94ec3a20 FC |
2784 | Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), |
2785 | "A sequence of multiple spaces in a charnam" | |
2786 | "es alias definition is deprecated"); | |
bd299e29 | 2787 | } |
bde9e88d KW |
2788 | s++; |
2789 | } | |
2790 | else if (UTF8_IS_DOWNGRADEABLE_START(*s)) { | |
2791 | if (! isCHARNAME_CONT(UNI_TO_NATIVE(TWO_BYTE_UTF8_TO_UNI(*s, | |
2792 | *(s+1))))) | |
2793 | { | |
2794 | goto bad_charname; | |
2795 | } | |
2796 | s += 2; | |
2797 | } | |
2798 | else { | |
2799 | if (! PL_utf8_charname_continue) { | |
2800 | U8 flags = _CORE_SWASH_INIT_ACCEPT_INVLIST; | |
2801 | PL_utf8_charname_continue = _core_swash_init("utf8", | |
2802 | "_Perl_Charname_Continue", | |
2803 | &PL_sv_undef, | |
2804 | 1, 0, NULL, &flags); | |
2805 | } | |
2806 | if (! swash_fetch(PL_utf8_charname_continue, (U8 *) s, TRUE)) { | |
2807 | goto bad_charname; | |
2808 | } | |
2809 | s += UTF8SKIP(s); | |
6f613c73 KW |
2810 | } |
2811 | } | |
3036c853 | 2812 | if (*(s-1) == ' ' && ckWARN_d(WARN_DEPRECATED)) { |
94ec3a20 FC |
2813 | Perl_warner(aTHX_ packWARN(WARN_DEPRECATED), |
2814 | "Trailing white-space in a charnames alias " | |
2815 | "definition is deprecated"); | |
bd299e29 | 2816 | } |
6f613c73 KW |
2817 | } |
2818 | ||
94ca1619 | 2819 | if (SvUTF8(res)) { /* Don't accept malformed input */ |
bde9e88d KW |
2820 | const U8* first_bad_char_loc; |
2821 | STRLEN len; | |
2822 | const char* const str = SvPV_const(res, len); | |
2823 | if (! is_utf8_string_loc((U8 *) str, len, &first_bad_char_loc)) { | |
2824 | /* If warnings are on, this will print a more detailed analysis of | |
2825 | * what is wrong than the error message below */ | |
2826 | utf8n_to_uvuni(first_bad_char_loc, | |
2827 | (char *) first_bad_char_loc - str, | |
2828 | NULL, 0); | |
2829 | ||
2830 | /* We deliberately don't try to print the malformed character, | |
2831 | * which might not print very well; it also may be just the first | |
2832 | * of many malformations, so don't print what comes after it */ | |
2833 | yyerror_pv( | |
2834 | Perl_form(aTHX_ | |
2835 | "Malformed UTF-8 returned by %.*s immediately after '%.*s'", | |
2836 | (int) (e - backslash_ptr + 1), backslash_ptr, | |
2837 | (int) ((char *) first_bad_char_loc - str), str | |
2838 | ), | |
2839 | SVf_UTF8); | |
2840 | return NULL; | |
2841 | } | |
2842 | } | |
140b12ad | 2843 | |
bde9e88d | 2844 | return res; |
140b12ad | 2845 | |
bde9e88d KW |
2846 | bad_charname: { |
2847 | int bad_char_size = ((UTF) ? UTF8SKIP(s) : 1); | |
2848 | ||
2849 | /* The final %.*s makes sure that should the trailing NUL be missing | |
2850 | * that this print won't run off the end of the string */ | |
2851 | yyerror_pv( | |
2852 | Perl_form(aTHX_ | |
2853 | "Invalid character in \\N{...}; marked by <-- HERE in %.*s<-- HERE %.*s", | |
2854 | (int)(s - backslash_ptr + bad_char_size), backslash_ptr, | |
2855 | (int)(e - s + bad_char_size), s + bad_char_size | |
2856 | ), | |
2857 | UTF ? SVf_UTF8 : 0); | |
2858 | return NULL; | |
2859 | } | |
6f613c73 KW |
2860 | } |
2861 | ||
02aa26ce NT |
2862 | /* |
2863 | scan_const | |
2864 | ||
9da1dd8f DM |
2865 | Extracts the next constant part of a pattern, double-quoted string, |
2866 | or transliteration. This is terrifying code. | |
2867 | ||
2868 | For example, in parsing the double-quoted string "ab\x63$d", it would | |
2869 | stop at the '$' and return an OP_CONST containing 'abc'. | |
02aa26ce | 2870 | |
94def140 | 2871 | It looks at PL_lex_inwhat and PL_lex_inpat to find out whether it's |
3280af22 | 2872 | processing a pattern (PL_lex_inpat is true), a transliteration |
94def140 | 2873 | (PL_lex_inwhat == OP_TRANS is true), or a double-quoted string. |
02aa26ce | 2874 | |
94def140 TS |
2875 | Returns a pointer to the character scanned up to. If this is |
2876 | advanced from the start pointer supplied (i.e. if anything was | |
9da1dd8f | 2877 | successfully parsed), will leave an OP_CONST for the substring scanned |
6154021b | 2878 | in pl_yylval. Caller must intuit reason for not parsing further |
9b599b2a GS |
2879 | by looking at the next characters herself. |
2880 | ||
02aa26ce | 2881 | In patterns: |
9da1dd8f | 2882 | expand: |
537124e4 KW |
2883 | \N{FOO} => \N{U+hex_for_character_FOO} |
2884 | (if FOO expands to multiple characters, expands to \N{U+xx.XX.yy ...}) | |
9da1dd8f DM |
2885 | |
2886 | pass through: | |
2887 | all other \-char, including \N and \N{ apart from \N{ABC} | |
2888 | ||
2889 | stops on: | |
2890 | @ and $ where it appears to be a var, but not for $ as tail anchor | |
2891 | \l \L \u \U \Q \E | |
2892 | (?{ or (??{ | |
2893 | ||
02aa26ce NT |
2894 | |
2895 | In transliterations: | |
2896 | characters are VERY literal, except for - not at the start or end | |
94def140 TS |
2897 | of the string, which indicates a range. If the range is in bytes, |
2898 | scan_const expands the range to the full set of intermediate | |
2899 | characters. If the range is in utf8, the hyphen is replaced with | |
2900 | a certain range mark which will be handled by pmtrans() in op.c. | |
02aa26ce NT |
2901 | |
2902 | In double-quoted strings: | |
2903 | backslashes: | |
2904 | double-quoted style: \r and \n | |
ff3f963a | 2905 | constants: \x31, etc. |
94def140 | 2906 | deprecated backrefs: \1 (in substitution replacements) |
02aa26ce NT |
2907 | case and quoting: \U \Q \E |
2908 | stops on @ and $ | |
2909 | ||
2910 | scan_const does *not* construct ops to handle interpolated strings. | |
2911 | It stops processing as soon as it finds an embedded $ or @ variable | |
2912 | and leaves it to the caller to work out what's going on. | |
2913 | ||
94def140 TS |
2914 | embedded arrays (whether in pattern or not) could be: |
2915 | @foo, @::foo, @'foo, @{foo}, @$foo, @+, @-. | |
2916 | ||
2917 | $ in double-quoted strings must be the symbol of an embedded scalar. | |
02aa26ce NT |
2918 | |
2919 | $ in pattern could be $foo or could be tail anchor. Assumption: | |
2920 | it's a tail anchor if $ is the last thing in the string, or if it's | |
94def140 | 2921 | followed by one of "()| \r\n\t" |
02aa26ce | 2922 | |
9da1dd8f | 2923 | \1 (backreferences) are turned into $1 in substitutions |
02aa26ce NT |
2924 | |
2925 | The structure of the code is | |
2926 | while (there's a character to process) { | |
94def140 TS |
2927 | handle transliteration ranges |
2928 | skip regexp comments /(?#comment)/ and codes /(?{code})/ | |
2929 | skip #-initiated comments in //x patterns | |
2930 | check for embedded arrays | |
02aa26ce NT |
2931 | check for embedded scalars |
2932 | if (backslash) { | |
94def140 | 2933 | deprecate \1 in substitution replacements |
02aa26ce NT |
2934 | handle string-changing backslashes \l \U \Q \E, etc. |
2935 | switch (what was escaped) { | |
94def140 | 2936 | handle \- in a transliteration (becomes a literal -) |
ff3f963a | 2937 | if a pattern and not \N{, go treat as regular character |
94def140 TS |
2938 | handle \132 (octal characters) |
2939 | handle \x15 and \x{1234} (hex characters) | |
ff3f963a | 2940 | handle \N{name} (named characters, also \N{3,5} in a pattern) |
94def140 TS |
2941 | handle \cV (control characters) |
2942 | handle printf-style backslashes (\f, \r, \n, etc) | |
02aa26ce | 2943 | } (end switch) |
77a135fe | 2944 | continue |
02aa26ce | 2945 | } (end if backslash) |
77a135fe | 2946 | handle regular character |
02aa26ce | 2947 | } (end while character to read) |
4e553d73 | 2948 | |
02aa26ce NT |
2949 | */ |
2950 | ||
76e3520e | 2951 | STATIC char * |
cea2e8a9 | 2952 | S_scan_const(pTHX_ char *start) |
79072805 | 2953 | { |
97aff369 | 2954 | dVAR; |
eb578fdb | 2955 | char *send = PL_bufend; /* end of the constant */ |
77a135fe KW |
2956 | SV *sv = newSV(send - start); /* sv for the constant. See |
2957 | note below on sizing. */ | |
eb578fdb KW |
2958 | char *s = start; /* start of the constant */ |
2959 | char *d = SvPVX(sv); /* destination for copies */ | |
02aa26ce | 2960 | bool dorange = FALSE; /* are we in a translit range? */ |
c2e66d9e | 2961 | bool didrange = FALSE; /* did we just finish a range? */ |
2866decb | 2962 | bool in_charclass = FALSE; /* within /[...]/ */ |
b953e60c KW |
2963 | bool has_utf8 = FALSE; /* Output constant is UTF8 */ |
2964 | bool this_utf8 = cBOOL(UTF); /* Is the source string assumed | |
77a135fe KW |
2965 | to be UTF8? But, this can |
2966 | show as true when the source | |
2967 | isn't utf8, as for example | |
2968 | when it is entirely composed | |
2969 | of hex constants */ | |
6f613c73 | 2970 | SV *res; /* result from charnames */ |
77a135fe KW |
2971 | |
2972 | /* Note on sizing: The scanned constant is placed into sv, which is | |
2973 | * initialized by newSV() assuming one byte of output for every byte of | |
2974 | * input. This routine expects newSV() to allocate an extra byte for a | |
2975 | * trailing NUL, which this routine will append if it gets to the end of | |
2976 | * the input. There may be more bytes of input than output (eg., \N{LATIN | |
2977 | * CAPITAL LETTER A}), or more output than input if the constant ends up | |
2978 | * recoded to utf8, but each time a construct is found that might increase | |
2979 | * the needed size, SvGROW() is called. Its size parameter each time is | |
2980 | * based on the best guess estimate at the time, namely the length used so | |
2981 | * far, plus the length the current construct will occupy, plus room for | |
2982 | * the trailing NUL, plus one byte for every input byte still unscanned */ | |
2983 | ||
c3320c2a KW |
2984 | UV uv = UV_MAX; /* Initialize to weird value to try to catch any uses |
2985 | before set */ | |
4c3a8340 TS |
2986 | #ifdef EBCDIC |
2987 | UV literal_endpoint = 0; | |
e294cc5d | 2988 | bool native_range = TRUE; /* turned to FALSE if the first endpoint is Unicode. */ |
4c3a8340 | 2989 | #endif |
012bcf8d | 2990 | |
7918f24d NC |
2991 | PERL_ARGS_ASSERT_SCAN_CONST; |
2992 | ||
bb16bae8 | 2993 | assert(PL_lex_inwhat != OP_TRANSR); |
2b9d42f0 NIS |
2994 | if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) { |
2995 | /* If we are doing a trans and we know we want UTF8 set expectation */ | |
2996 | has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF); | |
2997 | this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF); | |
2998 | } | |
2999 | ||
b899e89d FC |
3000 | /* Protect sv from errors and fatal warnings. */ |
3001 | ENTER_with_name("scan_const"); | |
3002 | SAVEFREESV(sv); | |
2b9d42f0 | 3003 | |
79072805 | 3004 | while (s < send || dorange) { |
ff3f963a | 3005 | |
02aa26ce | 3006 | /* get transliterations out of the way (they're most literal) */ |
3280af22 | 3007 | if (PL_lex_inwhat == OP_TRANS) { |
02aa26ce | 3008 | /* expand a range A-Z to the full set of characters. AIE! */ |
79072805 | 3009 | if (dorange) { |
1ba5c669 JH |
3010 | I32 i; /* current expanded character */ |
3011 | I32 min; /* first character in range */ | |
3012 | I32 max; /* last character in range */ | |
02aa26ce | 3013 | |
e294cc5d JH |
3014 | #ifdef EBCDIC |
3015 | UV uvmax = 0; | |
3016 | #endif | |
3017 | ||
3018 | if (has_utf8 | |
3019 | #ifdef EBCDIC | |
3020 | && !native_range | |
3021 | #endif | |
1953db30 | 3022 | ) { |
9d4ba2ae | 3023 | char * const c = (char*)utf8_hop((U8*)d, -1); |
8973db79 JH |
3024 | char *e = d++; |
3025 | while (e-- > c) | |
3026 | *(e + 1) = *e; | |
25716404 | 3027 | *c = (char)UTF_TO_NATIVE(0xff); |
8973db79 JH |
3028 | /* mark the range as done, and continue */ |
3029 | dorange = FALSE; | |
3030 | didrange = TRUE; | |
3031 | continue; | |
3032 | } | |
2b9d42f0 | 3033 | |
95a20fc0 | 3034 | i = d - SvPVX_const(sv); /* remember current offset */ |
e294cc5d JH |
3035 | #ifdef EBCDIC |
3036 | SvGROW(sv, | |
3037 | SvLEN(sv) + (has_utf8 ? | |
3038 | (512 - UTF_CONTINUATION_MARK + | |
3039 | UNISKIP(0x100)) | |
3040 | : 256)); | |
3041 | /* How many two-byte within 0..255: 128 in UTF-8, | |
3042 | * 96 in UTF-8-mod. */ | |
3043 | #else | |
9cbb5ea2 | 3044 | SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */ |
e294cc5d | 3045 | #endif |
9cbb5ea2 | 3046 | d = SvPVX(sv) + i; /* refresh d after realloc */ |
e294cc5d JH |
3047 | #ifdef EBCDIC |
3048 | if (has_utf8) { | |
3049 | int j; | |
3050 | for (j = 0; j <= 1; j++) { | |
3051 | char * const c = (char*)utf8_hop((U8*)d, -1); | |
3052 | const UV uv = utf8n_to_uvchr((U8*)c, d - c, NULL, 0); | |
3053 | if (j) | |
3054 | min = (U8)uv; | |
3055 | else if (uv < 256) | |
3056 | max = (U8)uv; | |
3057 | else { | |
3058 | max = (U8)0xff; /* only to \xff */ | |
3059 | uvmax = uv; /* \x{100} to uvmax */ | |
3060 | } | |
3061 | d = c; /* eat endpoint chars */ | |
3062 | } | |
3063 | } | |
3064 | else { | |
3065 | #endif | |
3066 | d -= 2; /* eat the first char and the - */ | |
3067 | min = (U8)*d; /* first char in range */ | |
3068 | max = (U8)d[1]; /* last char in range */ | |
3069 | #ifdef EBCDIC | |
3070 | } | |
3071 | #endif | |
8ada0baa | 3072 | |
c2e66d9e | 3073 | if (min > max) { |
01ec43d0 | 3074 | Perl_croak(aTHX_ |
d1573ac7 | 3075 | "Invalid range \"%c-%c\" in transliteration operator", |
1ba5c669 | 3076 | (char)min, (char)max); |
c2e66d9e GS |
3077 | } |
3078 | ||
c7f1f016 | 3079 | #ifdef EBCDIC |
4c3a8340 TS |
3080 | if (literal_endpoint == 2 && |
3081 | ((isLOWER(min) && isLOWER(max)) || | |
3082 | (isUPPER(min) && isUPPER(max)))) { | |
8ada0baa JH |
3083 | if (isLOWER(min)) { |
3084 | for (i = min; i <= max; i++) | |
3085 | if (isLOWER(i)) | |
db42d148 | 3086 | *d++ = NATIVE_TO_NEED(has_utf8,i); |
8ada0baa JH |
3087 | } else { |
3088 | for (i = min; i <= max; i++) | |
3089 | if (isUPPER(i)) | |
db42d148 | 3090 | *d++ = NATIVE_TO_NEED(has_utf8,i); |
8ada0baa JH |
3091 | } |
3092 | } | |
3093 | else | |
3094 | #endif | |
3095 | for (i = min; i <= max; i++) | |
e294cc5d JH |
3096 | #ifdef EBCDIC |
3097 | if (has_utf8) { | |
3098 | const U8 ch = (U8)NATIVE_TO_UTF(i); | |
3099 | if (UNI_IS_INVARIANT(ch)) | |
3100 | *d++ = (U8)i; | |
3101 | else { | |
3102 | *d++ = (U8)UTF8_EIGHT_BIT_HI(ch); | |
3103 | *d++ = (U8)UTF8_EIGHT_BIT_LO(ch); | |
3104 | } | |
3105 | } | |
3106 | else | |
3107 | #endif | |
3108 | *d++ = (char)i; | |
3109 | ||
3110 | #ifdef EBCDIC | |
3111 | if (uvmax) { | |
3112 | d = (char*)uvchr_to_utf8((U8*)d, 0x100); | |
3113 | if (uvmax > 0x101) | |
3114 | *d++ = (char)UTF_TO_NATIVE(0xff); | |
3115 | if (uvmax > 0x100) | |
3116 | d = (char*)uvchr_to_utf8((U8*)d, uvmax); | |
3117 | } | |
3118 | #endif | |
02aa26ce NT |
3119 | |
3120 | /* mark the range as done, and continue */ | |
79072805 | 3121 | dorange = FALSE; |
01ec43d0 | 3122 | didrange = TRUE; |
4c3a8340 TS |
3123 | #ifdef EBCDIC |
3124 | literal_endpoint = 0; | |
3125 | #endif | |
79072805 | 3126 | continue; |
4e553d73 | 3127 | } |
02aa26ce NT |
3128 | |
3129 | /* range begins (ignore - as first or last char) */ | |
79072805 | 3130 | else if (*s == '-' && s+1 < send && s != start) { |
4e553d73 | 3131 | if (didrange) { |
1fafa243 | 3132 | Perl_croak(aTHX_ "Ambiguous range in transliteration operator"); |
01ec43d0 | 3133 | } |
e294cc5d JH |
3134 | if (has_utf8 |
3135 | #ifdef EBCDIC | |
3136 | && !native_range | |
3137 | #endif | |
3138 | ) { | |
25716404 | 3139 | *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */ |
a0ed51b3 LW |
3140 | s++; |
3141 | continue; | |
3142 | } | |
79072805 LW |
3143 | dorange = TRUE; |
3144 | s++; | |
01ec43d0 GS |
3145 | } |
3146 | else { | |
3147 | didrange = FALSE; | |
4c3a8340 TS |
3148 | #ifdef EBCDIC |
3149 | literal_endpoint = 0; | |
e294cc5d | 3150 | native_range = TRUE; |
4c3a8340 | 3151 | #endif |
01ec43d0 | 3152 | } |
79072805 | 3153 | } |
02aa26ce NT |
3154 | |
3155 | /* if we get here, we're not doing a transliteration */ | |
3156 | ||
e4a2df84 DM |
3157 | else if (*s == '[' && PL_lex_inpat && !in_charclass) { |
3158 | char *s1 = s-1; | |
3159 | int esc = 0; | |
3160 | while (s1 >= start && *s1-- == '\\') | |
3161 | esc = !esc; | |
3162 | if (!esc) | |
3163 | in_charclass = TRUE; | |
3164 | } | |
2866decb | 3165 | |
e4a2df84 DM |
3166 | else if (*s == ']' && PL_lex_inpat && in_charclass) { |
3167 | char *s1 = s-1; | |
3168 | int esc = 0; | |
3169 | while (s1 >= start && *s1-- == '\\') | |
3170 | esc = !esc; | |
3171 | if (!esc) | |
3172 | in_charclass = FALSE; | |
3173 | } | |
2866decb | 3174 | |
9da1dd8f DM |
3175 | /* skip for regexp comments /(?#comment)/, except for the last |
3176 | * char, which will be done separately. | |
3177 | * Stop on (?{..}) and friends */ | |
3178 | ||
3280af22 | 3179 | else if (*s == '(' && PL_lex_inpat && s[1] == '?') { |
cc6b7395 | 3180 | if (s[2] == '#') { |
e994fd66 | 3181 | while (s+1 < send && *s != ')') |
db42d148 | 3182 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
155aba94 | 3183 | } |
2866decb | 3184 | else if (!PL_lex_casemods && !in_charclass && |
d3cec5e5 DM |
3185 | ( s[2] == '{' /* This should match regcomp.c */ |
3186 | || (s[2] == '?' && s[3] == '{'))) | |
155aba94 | 3187 | { |
9da1dd8f | 3188 | break; |
cc6b7395 | 3189 | } |
748a9306 | 3190 | } |
02aa26ce NT |
3191 | |
3192 | /* likewise skip #-initiated comments in //x patterns */ | |
3280af22 | 3193 | else if (*s == '#' && PL_lex_inpat && |
73134a2e | 3194 | ((PMOP*)PL_lex_inpat)->op_pmflags & RXf_PMf_EXTENDED) { |
748a9306 | 3195 | while (s+1 < send && *s != '\n') |
db42d148 | 3196 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
748a9306 | 3197 | } |
02aa26ce | 3198 | |
9da1dd8f DM |
3199 | /* no further processing of single-quoted regex */ |
3200 | else if (PL_lex_inpat && SvIVX(PL_linestr) == '\'') | |
3201 | goto default_action; | |
3202 | ||
5d1d4326 | 3203 | /* check for embedded arrays |
da6eedaa | 3204 | (@foo, @::foo, @'foo, @{foo}, @$foo, @+, @-) |
5d1d4326 | 3205 | */ |
1749ea0d | 3206 | else if (*s == '@' && s[1]) { |
8a2bca12 | 3207 | if (isWORDCHAR_lazy_if(s+1,UTF)) |
1749ea0d TS |
3208 | break; |
3209 | if (strchr(":'{$", s[1])) | |
3210 | break; | |
3211 | if (!PL_lex_inpat && (s[1] == '+' || s[1] == '-')) | |
3212 | break; /* in regexp, neither @+ nor @- are interpolated */ | |
3213 | } | |
02aa26ce NT |
3214 | |
3215 | /* check for embedded scalars. only stop if we're sure it's a | |
3216 | variable. | |
3217 | */ | |
79072805 | 3218 | else if (*s == '$') { |
3280af22 | 3219 | if (!PL_lex_inpat) /* not a regexp, so $ must be var */ |
79072805 | 3220 | break; |
77772344 | 3221 | if (s + 1 < send && !strchr("()| \r\n\t", s[1])) { |
a2a5de95 NC |
3222 | if (s[1] == '\\') { |
3223 | Perl_ck_warner(aTHX_ packWARN(WARN_AMBIGUOUS), | |
3224 | "Possible unintended interpolation of $\\ in regex"); | |
77772344 | 3225 | } |
79072805 | 3226 | break; /* in regexp, $ might be tail anchor */ |
77772344 | 3227 | } |
79072805 | 3228 | } |
02aa26ce | 3229 | |
2b9d42f0 NIS |
3230 | /* End of else if chain - OP_TRANS rejoin rest */ |
3231 | ||
02aa26ce | 3232 | /* backslashes */ |
79072805 | 3233 | if (*s == '\\' && s+1 < send) { |
ff3f963a KW |
3234 | char* e; /* Can be used for ending '}', etc. */ |
3235 | ||
79072805 | 3236 | s++; |
02aa26ce | 3237 | |
7d0fc23c KW |
3238 | /* warn on \1 - \9 in substitution replacements, but note that \11 |
3239 | * is an octal; and \19 is \1 followed by '9' */ | |
3280af22 | 3240 | if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat && |
a0d0e21e | 3241 | isDIGIT(*s) && *s != '0' && !isDIGIT(s[1])) |
79072805 | 3242 | { |
a2a5de95 | 3243 | Perl_ck_warner(aTHX_ packWARN(WARN_SYNTAX), "\\%c better written as $%c", *s, *s); |
79072805 LW |
3244 | *--s = '$'; |
3245 | break; | |
3246 | } | |
02aa26ce NT |
3247 | |
3248 | /* string-change backslash escapes */ | |
838f2281 | 3249 | if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQF", *s)) { |
79072805 LW |
3250 | --s; |
3251 | break; | |
3252 | } | |
ff3f963a KW |
3253 | /* In a pattern, process \N, but skip any other backslash escapes. |
3254 | * This is because we don't want to translate an escape sequence | |
3255 | * into a meta symbol and have the regex compiler use the meta | |
3256 | * symbol meaning, e.g. \x{2E} would be confused with a dot. But | |
3257 | * in spite of this, we do have to process \N here while the proper | |
3258 | * charnames handler is in scope. See bugs #56444 and #62056. | |
3259 | * There is a complication because \N in a pattern may also stand | |
3260 | * for 'match a non-nl', and not mean a charname, in which case its | |
3261 | * processing should be deferred to the regex compiler. To be a | |
3262 | * charname it must be followed immediately by a '{', and not look | |
3263 | * like \N followed by a curly quantifier, i.e., not something like | |
3264 | * \N{3,}. regcurly returns a boolean indicating if it is a legal | |
3265 | * quantifier */ | |
3266 | else if (PL_lex_inpat | |
3267 | && (*s != 'N' | |
3268 | || s[1] != '{' | |
4d68ffa0 | 3269 | || regcurly(s + 1, FALSE))) |
ff3f963a | 3270 | { |
cc74c5bd TS |
3271 | *d++ = NATIVE_TO_NEED(has_utf8,'\\'); |
3272 | goto default_action; | |
3273 | } | |
02aa26ce | 3274 | |
79072805 | 3275 | switch (*s) { |
02aa26ce NT |
3276 | |
3277 | /* quoted - in transliterations */ | |
79072805 | 3278 | case '-': |
3280af22 | 3279 | if (PL_lex_inwhat == OP_TRANS) { |
79072805 LW |
3280 | *d++ = *s++; |
3281 | continue; | |
3282 | } | |
3283 | /* FALL THROUGH */ | |
3284 | default: | |
11b8faa4 | 3285 | { |
15861f94 | 3286 | if ((isALPHANUMERIC(*s))) |
a2a5de95 NC |
3287 | Perl_ck_warner(aTHX_ packWARN(WARN_MISC), |
3288 | "Unrecognized escape \\%c passed through", | |
3289 | *s); | |
11b8faa4 | 3290 | /* default action is to copy the quoted character */ |
f9a63242 | 3291 | goto default_action; |
11b8faa4 | 3292 | } |
02aa26ce | 3293 | |
632403cc | 3294 | /* eg. \132 indicates the octal constant 0132 */ |
79072805 LW |
3295 | case '0': case '1': case '2': case '3': |
3296 | case '4': case '5': case '6': case '7': | |
ba210ebe | 3297 | { |
5e0a247b | 3298 | I32 flags = PERL_SCAN_SILENT_ILLDIGIT; |
53305cf1 | 3299 | STRLEN len = 3; |
77a135fe | 3300 | uv = NATIVE_TO_UNI(grok_oct(s, &len, &flags, NULL)); |
ba210ebe | 3301 | s += len; |
5e0a247b KW |
3302 | if (len < 3 && s < send && isDIGIT(*s) |
3303 | && ckWARN(WARN_MISC)) | |
3304 | { | |
3305 | Perl_warner(aTHX_ packWARN(WARN_MISC), | |
3306 | "%s", form_short_octal_warning(s, len)); | |
3307 | } | |
ba210ebe | 3308 | } |
012bcf8d | 3309 | goto NUM_ESCAPE_INSERT; |
02aa26ce | 3310 | |
f0a2b745 KW |
3311 | /* eg. \o{24} indicates the octal constant \024 */ |
3312 | case 'o': | |
3313 | { | |
454155d9 | 3314 | const char* error; |
f0a2b745 | 3315 | |
00ce5563 | 3316 | bool valid = grok_bslash_o(&s, &uv, &error, |
80f4111b KW |
3317 | TRUE, /* Output warning */ |
3318 | FALSE, /* Not strict */ | |
17896a85 KW |
3319 | TRUE, /* Output warnings for |
3320 | non-portables */ | |
80f4111b | 3321 | UTF); |
454155d9 | 3322 | if (! valid) { |
f0a2b745 KW |
3323 | yyerror(error); |
3324 | continue; | |
3325 | } | |
3326 | goto NUM_ESCAPE_INSERT; | |
3327 | } | |
3328 | ||
77a135fe | 3329 | /* eg. \x24 indicates the hex constant 0x24 */ |
79072805 | 3330 | case 'x': |
a0481293 | 3331 | { |
a0481293 | 3332 | const char* error; |
355860ce | 3333 | |
00ce5563 | 3334 | bool valid = grok_bslash_x(&s, &uv, &error, |
80f4111b KW |
3335 | TRUE, /* Output warning */ |
3336 | FALSE, /* Not strict */ | |
17896a85 KW |
3337 | TRUE, /* Output warnings for |
3338 | non-portables */ | |
80f4111b | 3339 | UTF); |
a0481293 KW |
3340 | if (! valid) { |
3341 | yyerror(error); | |
355860ce | 3342 | continue; |
ba210ebe | 3343 | } |
012bcf8d GS |
3344 | } |
3345 | ||
3346 | NUM_ESCAPE_INSERT: | |
ff3f963a KW |
3347 | /* Insert oct or hex escaped character. There will always be |
3348 | * enough room in sv since such escapes will be longer than any | |
3349 | * UTF-8 sequence they can end up as, except if they force us | |
3350 | * to recode the rest of the string into utf8 */ | |
ba7cea30 | 3351 | |
77a135fe | 3352 | /* Here uv is the ordinal of the next character being added in |
ff3f963a | 3353 | * unicode (converted from native). */ |
77a135fe | 3354 | if (!UNI_IS_INVARIANT(uv)) { |
9aa983d2 | 3355 | if (!has_utf8 && uv > 255) { |
77a135fe KW |
3356 | /* Might need to recode whatever we have accumulated so |
3357 | * far if it contains any chars variant in utf8 or | |
3358 | * utf-ebcdic. */ | |
3359 | ||
3360 | SvCUR_set(sv, d - SvPVX_const(sv)); | |
3361 | SvPOK_on(sv); | |
3362 | *d = '\0'; | |
77a135fe | 3363 | /* See Note on sizing above. */ |
7bf79863 KW |
3364 | sv_utf8_upgrade_flags_grow(sv, |
3365 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
3366 | UNISKIP(uv) + (STRLEN)(send - s) + 1); | |
77a135fe KW |
3367 | d = SvPVX(sv) + SvCUR(sv); |
3368 | has_utf8 = TRUE; | |
012bcf8d GS |
3369 | } |
3370 | ||
77a135fe KW |
3371 | if (has_utf8) { |
3372 | d = (char*)uvuni_to_utf8((U8*)d, uv); | |
f9a63242 JH |
3373 | if (PL_lex_inwhat == OP_TRANS && |
3374 | PL_sublex_info.sub_op) { | |
3375 | PL_sublex_info.sub_op->op_private |= | |
3376 | (PL_lex_repl ? OPpTRANS_FROM_UTF | |
3377 | : OPpTRANS_TO_UTF); | |
f9a63242 | 3378 | } |
e294cc5d JH |
3379 | #ifdef EBCDIC |
3380 | if (uv > 255 && !dorange) | |
3381 | native_range = FALSE; | |
3382 | #endif | |
012bcf8d | 3383 | } |
a0ed51b3 | 3384 | else { |
012bcf8d | 3385 | *d++ = (char)uv; |
a0ed51b3 | 3386 | } |
012bcf8d GS |
3387 | } |
3388 | else { | |
c4d5f83a | 3389 | *d++ = (char) uv; |
a0ed51b3 | 3390 | } |
79072805 | 3391 | continue; |
02aa26ce | 3392 | |
4a2d328f | 3393 | case 'N': |
ff3f963a KW |
3394 | /* In a non-pattern \N must be a named character, like \N{LATIN |
3395 | * SMALL LETTER A} or \N{U+0041}. For patterns, it also can | |
3396 | * mean to match a non-newline. For non-patterns, named | |
3397 | * characters are converted to their string equivalents. In | |
3398 | * patterns, named characters are not converted to their | |
3399 | * ultimate forms for the same reasons that other escapes | |
3400 | * aren't. Instead, they are converted to the \N{U+...} form | |
3401 | * to get the value from the charnames that is in effect right | |
3402 | * now, while preserving the fact that it was a named character | |
3403 | * so that the regex compiler knows this */ | |
3404 | ||
3405 | /* This section of code doesn't generally use the | |
3406 | * NATIVE_TO_NEED() macro to transform the input. I (khw) did | |
3407 | * a close examination of this macro and determined it is a | |
3408 | * no-op except on utfebcdic variant characters. Every | |
3409 | * character generated by this that would normally need to be | |
3410 | * enclosed by this macro is invariant, so the macro is not | |
7538f724 KW |
3411 | * needed, and would complicate use of copy(). XXX There are |
3412 | * other parts of this file where the macro is used | |
3413 | * inconsistently, but are saved by it being a no-op */ | |
ff3f963a KW |
3414 | |
3415 | /* The structure of this section of code (besides checking for | |
3416 | * errors and upgrading to utf8) is: | |
3417 | * Further disambiguate between the two meanings of \N, and if | |
3418 | * not a charname, go process it elsewhere | |
0a96133f KW |
3419 | * If of form \N{U+...}, pass it through if a pattern; |
3420 | * otherwise convert to utf8 | |
3421 | * Otherwise must be \N{NAME}: convert to \N{U+c1.c2...} if a | |
3422 | * pattern; otherwise convert to utf8 */ | |
ff3f963a KW |
3423 | |
3424 | /* Here, s points to the 'N'; the test below is guaranteed to | |
3425 | * succeed if we are being called on a pattern as we already | |
3426 | * know from a test above that the next character is a '{'. | |
3427 | * On a non-pattern \N must mean 'named sequence, which | |
3428 | * requires braces */ | |
3429 | s++; | |
3430 | if (*s != '{') { | |
3431 | yyerror("Missing braces on \\N{}"); | |
3432 | continue; | |
3433 | } | |
3434 | s++; | |
3435 | ||
0a96133f | 3436 | /* If there is no matching '}', it is an error. */ |
ff3f963a KW |
3437 | if (! (e = strchr(s, '}'))) { |
3438 | if (! PL_lex_inpat) { | |
5777a3f7 | 3439 | yyerror("Missing right brace on \\N{}"); |
0a96133f KW |
3440 | } else { |
3441 | yyerror("Missing right brace on \\N{} or unescaped left brace after \\N."); | |
dbc0d4f2 | 3442 | } |
0a96133f | 3443 | continue; |
ff3f963a | 3444 | } |
cddc7ef4 | 3445 | |
ff3f963a | 3446 | /* Here it looks like a named character */ |
cddc7ef4 | 3447 | |
ff3f963a KW |
3448 | if (*s == 'U' && s[1] == '+') { /* \N{U+...} */ |
3449 | I32 flags = PERL_SCAN_ALLOW_UNDERSCORES | |
3450 | | PERL_SCAN_DISALLOW_PREFIX; | |
3451 | STRLEN len; | |
3452 | ||
3453 | /* For \N{U+...}, the '...' is a unicode value even on | |
3454 | * EBCDIC machines */ | |
3455 | s += 2; /* Skip to next char after the 'U+' */ | |
3456 | len = e - s; | |
3457 | uv = grok_hex(s, &len, &flags, NULL); | |
3458 | if (len == 0 || len != (STRLEN)(e - s)) { | |
3459 | yyerror("Invalid hexadecimal number in \\N{U+...}"); | |
3460 | s = e + 1; | |
3461 | continue; | |
3462 | } | |
3463 | ||
3464 | if (PL_lex_inpat) { | |
3465 | ||
e2a7e165 KW |
3466 | /* On non-EBCDIC platforms, pass through to the regex |
3467 | * compiler unchanged. The reason we evaluated the | |
3468 | * number above is to make sure there wasn't a syntax | |
3469 | * error. But on EBCDIC we convert to native so | |
3470 | * downstream code can continue to assume it's native | |
3471 | */ | |
ff3f963a | 3472 | s -= 5; /* Include the '\N{U+' */ |
e2a7e165 KW |
3473 | #ifdef EBCDIC |
3474 | d += my_snprintf(d, e - s + 1 + 1, /* includes the } | |
3475 | and the \0 */ | |
3476 | "\\N{U+%X}", | |
3477 | (unsigned int) UNI_TO_NATIVE(uv)); | |
3478 | #else | |
ff3f963a KW |
3479 | Copy(s, d, e - s + 1, char); /* 1 = include the } */ |
3480 | d += e - s + 1; | |
e2a7e165 | 3481 | #endif |
ff3f963a KW |
3482 | } |
3483 | else { /* Not a pattern: convert the hex to string */ | |
3484 | ||
3485 | /* If destination is not in utf8, unconditionally | |
3486 | * recode it to be so. This is because \N{} implies | |
3487 | * Unicode semantics, and scalars have to be in utf8 | |
3488 | * to guarantee those semantics */ | |
3489 | if (! has_utf8) { | |
3490 | SvCUR_set(sv, d - SvPVX_const(sv)); | |
3491 | SvPOK_on(sv); | |
3492 | *d = '\0'; | |
3493 | /* See Note on sizing above. */ | |
3494 | sv_utf8_upgrade_flags_grow( | |
3495 | sv, | |
3496 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
3497 | UNISKIP(uv) + (STRLEN)(send - e) + 1); | |
3498 | d = SvPVX(sv) + SvCUR(sv); | |
3499 | has_utf8 = TRUE; | |
3500 | } | |
3501 | ||
3502 | /* Add the string to the output */ | |
3503 | if (UNI_IS_INVARIANT(uv)) { | |
3504 | *d++ = (char) uv; | |
3505 | } | |
3506 | else d = (char*)uvuni_to_utf8((U8*)d, uv); | |
3507 | } | |
3508 | } | |
6f613c73 KW |
3509 | else /* Here is \N{NAME} but not \N{U+...}. */ |
3510 | if ((res = get_and_check_backslash_N_name(s, e))) | |
3511 | { | |
3512 | STRLEN len; | |
3513 | const char *str = SvPV_const(res, len); | |
3514 | if (PL_lex_inpat) { | |
ff3f963a KW |
3515 | |
3516 | if (! len) { /* The name resolved to an empty string */ | |
3517 | Copy("\\N{}", d, 4, char); | |
3518 | d += 4; | |
3519 | } | |
3520 | else { | |
3521 | /* In order to not lose information for the regex | |
3522 | * compiler, pass the result in the specially made | |
3523 | * syntax: \N{U+c1.c2.c3...}, where c1 etc. are | |
3524 | * the code points in hex of each character | |
3525 | * returned by charnames */ | |
3526 | ||
3527 | const char *str_end = str + len; | |
3b721c4f | 3528 | const STRLEN off = d - SvPVX_const(sv); |
94ca1619 KW |
3529 | |
3530 | if (! SvUTF8(res)) { | |
3531 | /* For the non-UTF-8 case, we can determine the | |
3532 | * exact length needed without having to parse | |
3533 | * through the string. Each character takes up | |
3534 | * 2 hex digits plus either a trailing dot or | |
3535 | * the "}" */ | |
3536 | d = off + SvGROW(sv, off | |
3537 | + 3 * len | |
3538 | + 6 /* For the "\N{U+", and | |
3539 | trailing NUL */ | |
3540 | + (STRLEN)(send - e)); | |
3541 | Copy("\\N{U+", d, 5, char); | |
3542 | d += 5; | |
3543 | while (str < str_end) { | |
3544 | char hex_string[4]; | |
3545 | my_snprintf(hex_string, sizeof(hex_string), | |
3546 | "%02X.", (U8) *str); | |
3547 | Copy(hex_string, d, 3, char); | |
3548 | d += 3; | |
3549 | str++; | |
3550 | } | |
3551 | d--; /* We will overwrite below the final | |
3552 | dot with a right brace */ | |
3553 | } | |
3554 | else { | |
1953db30 KW |
3555 | STRLEN char_length; /* cur char's byte length */ |
3556 | ||
3557 | /* and the number of bytes after this is | |
3558 | * translated into hex digits */ | |
3559 | STRLEN output_length; | |
3560 | ||
3561 | /* 2 hex per byte; 2 chars for '\N'; 2 chars | |
3562 | * for max('U+', '.'); and 1 for NUL */ | |
3563 | char hex_string[2 * UTF8_MAXBYTES + 5]; | |
3564 | ||
3565 | /* Get the first character of the result. */ | |
3566 | U32 uv = utf8n_to_uvuni((U8 *) str, | |
3567 | len, | |
3568 | &char_length, | |
3569 | UTF8_ALLOW_ANYUV); | |
3570 | /* Convert first code point to hex, including | |
3571 | * the boiler plate before it. For all these, | |
3572 | * we convert to native format so that | |
3573 | * downstream code can continue to assume the | |
3574 | * input is native */ | |
3575 | output_length = | |
3576 | my_snprintf(hex_string, sizeof(hex_string), | |
3577 | "\\N{U+%X", | |
3578 | (unsigned int) UNI_TO_NATIVE(uv)); | |
3579 | ||
3580 | /* Make sure there is enough space to hold it */ | |
3581 | d = off + SvGROW(sv, off | |
3582 | + output_length | |
3583 | + (STRLEN)(send - e) | |
3584 | + 2); /* '}' + NUL */ | |
3585 | /* And output it */ | |
3586 | Copy(hex_string, d, output_length, char); | |
3587 | d += output_length; | |
3588 | ||
3589 | /* For each subsequent character, append dot and | |
3590 | * its ordinal in hex */ | |
3591 | while ((str += char_length) < str_end) { | |
3592 | const STRLEN off = d - SvPVX_const(sv); | |
3593 | U32 uv = utf8n_to_uvuni((U8 *) str, | |
3594 | str_end - str, | |
3595 | &char_length, | |
3596 | UTF8_ALLOW_ANYUV); | |
3597 | output_length = | |
3598 | my_snprintf(hex_string, | |
3599 | sizeof(hex_string), | |
3600 | ".%X", | |
3601 | (unsigned int) UNI_TO_NATIVE(uv)); | |
3602 | ||
3603 | d = off + SvGROW(sv, off | |
3604 | + output_length | |
3605 | + (STRLEN)(send - e) | |
3606 | + 2); /* '}' + NUL */ | |
3607 | Copy(hex_string, d, output_length, char); | |
3608 | d += output_length; | |
3609 | } | |
94ca1619 | 3610 | } |
ff3f963a KW |
3611 | |
3612 | *d++ = '}'; /* Done. Add the trailing brace */ | |
3613 | } | |
3614 | } | |
3615 | else { /* Here, not in a pattern. Convert the name to a | |
3616 | * string. */ | |
3617 | ||
3618 | /* If destination is not in utf8, unconditionally | |
3619 | * recode it to be so. This is because \N{} implies | |
3620 | * Unicode semantics, and scalars have to be in utf8 | |
3621 | * to guarantee those semantics */ | |
3622 | if (! has_utf8) { | |
3623 | SvCUR_set(sv, d - SvPVX_const(sv)); | |
3624 | SvPOK_on(sv); | |
3625 | *d = '\0'; | |
3626 | /* See Note on sizing above. */ | |
3627 | sv_utf8_upgrade_flags_grow(sv, | |
3628 | SV_GMAGIC|SV_FORCE_UTF8_UPGRADE, | |
3629 | len + (STRLEN)(send - s) + 1); | |
3630 | d = SvPVX(sv) + SvCUR(sv); | |
3631 | has_utf8 = TRUE; | |
3632 | } else if (len > (STRLEN)(e - s + 4)) { /* I _guess_ 4 is \N{} --jhi */ | |
3633 | ||
3634 | /* See Note on sizing above. (NOTE: SvCUR() is not | |
3635 | * set correctly here). */ | |
3636 | const STRLEN off = d - SvPVX_const(sv); | |
3637 | d = off + SvGROW(sv, off + len + (STRLEN)(send - s) + 1); | |
3638 | } | |
3639 | Copy(str, d, len, char); | |
3640 | d += len; | |
423cee85 | 3641 | } |
6f613c73 | 3642 | |
423cee85 | 3643 | SvREFCNT_dec(res); |
cb233ae3 | 3644 | |
cb233ae3 | 3645 | } /* End \N{NAME} */ |
ff3f963a KW |
3646 | #ifdef EBCDIC |
3647 | if (!dorange) | |
3648 | native_range = FALSE; /* \N{} is defined to be Unicode */ | |
3649 | #endif | |
3650 | s = e + 1; /* Point to just after the '}' */ | |
423cee85 JH |
3651 | continue; |
3652 | ||
02aa26ce | 3653 | /* \c is a control character */ |
79072805 LW |
3654 | case 'c': |
3655 | s++; | |
961ce445 | 3656 | if (s < send) { |
17a3df4c | 3657 | *d++ = grok_bslash_c(*s++, has_utf8, 1); |
ba210ebe | 3658 | } |
961ce445 RGS |
3659 | else { |
3660 | yyerror("Missing control char name in \\c"); | |
3661 | } | |
79072805 | 3662 | continue; |
02aa26ce NT |
3663 | |
3664 | /* printf-style backslashes, formfeeds, newlines, etc */ | |
79072805 | 3665 | case 'b': |
db42d148 | 3666 | *d++ = NATIVE_TO_NEED(has_utf8,'\b'); |
79072805 LW |
3667 | break; |
3668 | case 'n': | |
db42d148 | 3669 | *d++ = NATIVE_TO_NEED(has_utf8,'\n'); |
79072805 LW |
3670 | break; |
3671 | case 'r': | |
db42d148 | 3672 | *d++ = NATIVE_TO_NEED(has_utf8,'\r'); |
79072805 LW |
3673 | break; |
3674 | case 'f': | |
db42d148 | 3675 | *d++ = NATIVE_TO_NEED(has_utf8,'\f'); |
79072805 LW |
3676 | break; |
3677 | case 't': | |
db42d148 | 3678 | *d++ = NATIVE_TO_NEED(has_utf8,'\t'); |
79072805 | 3679 | break; |
34a3fe2a | 3680 | case 'e': |
db42d148 | 3681 | *d++ = ASCII_TO_NEED(has_utf8,'\033'); |
34a3fe2a PP |
3682 | break; |
3683 | case 'a': | |
db42d148 | 3684 | *d++ = ASCII_TO_NEED(has_utf8,'\007'); |
79072805 | 3685 | break; |
02aa26ce NT |
3686 | } /* end switch */ |
3687 | ||
79072805 LW |
3688 | s++; |
3689 | continue; | |
02aa26ce | 3690 | } /* end if (backslash) */ |
4c3a8340 TS |
3691 | #ifdef EBCDIC |
3692 | else | |
3693 | literal_endpoint++; | |
3694 | #endif | |
02aa26ce | 3695 | |
f9a63242 | 3696 | default_action: |
77a135fe KW |
3697 | /* If we started with encoded form, or already know we want it, |
3698 | then encode the next character */ | |
3699 | if (! NATIVE_IS_INVARIANT((U8)(*s)) && (this_utf8 || has_utf8)) { | |
2b9d42f0 | 3700 | STRLEN len = 1; |
77a135fe KW |
3701 | |
3702 | ||
3703 | /* One might think that it is wasted effort in the case of the | |
3704 | * source being utf8 (this_utf8 == TRUE) to take the next character | |
3705 | * in the source, convert it to an unsigned value, and then convert | |
3706 | * it back again. But the source has not been validated here. The | |
3707 | * routine that does the conversion checks for errors like | |
3708 | * malformed utf8 */ | |
3709 | ||
5f66b61c AL |
3710 | const UV nextuv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s); |