| Commit | Line | Data |
|---|
| e807ee50 |
1 | /* stolen_chunk_of_toke.c - from perl 5.8.8 toke.c |
| 2 | * |
| 3 | * Copyright (C) 1991, 1992, 1993, 1994, 1995, 1996, 1997, 1998, 1999, |
| 4 | * 2000, 2001, 2002, 2003, 2004, 2005, 2006, by Larry Wall and others |
| 5 | * |
| 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. |
| 8 | * |
| 9 | */ |
| 10 | |
| 11 | /* |
| 12 | * "It all comes from here, the stench and the peril." --Frodo |
| 13 | */ |
| 14 | |
| 15 | /* |
| 16 | * this is all blatantly stolen. I sincerely hopes it doesn't fuck anything |
| 17 | * up but if it does blame me (Matt S Trout), not the poor original authors |
| 18 | */ |
| 19 | |
| 71511651 |
20 | /* the following #defines are stolen from assorted headers, not toke.c (mst) */ |
| e807ee50 |
21 | |
| a25db2dc |
22 | #define skipspace(a) S_skipspace(aTHX_ a, 0) |
| 23 | #define peekspace(a) S_skipspace(aTHX_ a, 1) |
| 24 | #define skipspace_force(a) S_skipspace(aTHX_ a, 2) |
| 78160085 |
25 | #define incline(a) S_incline(aTHX_ a) |
| 26 | #define filter_gets(a,b,c) S_filter_gets(aTHX_ a,b,c) |
| 27 | #define scan_str(a,b,c) S_scan_str(aTHX_ a,b,c) |
| 28 | #define scan_word(a,b,c,d,e) S_scan_word(aTHX_ a,b,c,d,e) |
| 92997e69 |
29 | #define scan_ident(a,b,c,d,e) S_scan_ident(aTHX_ a,b,c,d,e) |
| 78160085 |
30 | |
| 31 | STATIC void S_incline(pTHX_ char *s); |
| a25db2dc |
32 | STATIC char* S_skipspace(pTHX_ char *s, int incline); |
| 78160085 |
33 | STATIC char * S_filter_gets(pTHX_ SV *sv, PerlIO *fp, STRLEN append); |
| 34 | STATIC char* S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims); |
| 35 | STATIC char* S_scan_word(pTHX_ char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp); |
| 36 | |
| e807ee50 |
37 | #define DPTR2FPTR(t,p) ((t)PTR2nat(p)) /* data pointer to function pointer */ |
| 38 | #define FPTR2DPTR(t,p) ((t)PTR2nat(p)) /* function pointer to data pointer */ |
| c630715a |
39 | #define PTR2nat(p) (PTRV)(p) /* pointer to integer of PTRSIZE */ |
| 14ab3472 |
40 | |
| 41 | /* conditionalise these two because as of 5.9.5 we already get them from |
| 42 | the headers (mst) */ |
| 43 | #ifndef Newx |
| e807ee50 |
44 | #define Newx(v,n,t) (v = (MEM_WRAP_CHECK_(n,t) (t*)safemalloc((MEM_SIZE)((n)*sizeof(t))))) |
| 14ab3472 |
45 | #endif |
| 46 | #ifndef SvPVX_const |
| 0ba8c7aa |
47 | #define SvPVX_const(sv) ((const char*) (0 + SvPVX(sv))) |
| 14ab3472 |
48 | #endif |
| 60d774aa |
49 | #ifndef MEM_WRAP_CHECK_ |
| 50 | #define MEM_WRAP_CHECK_(n,t) MEM_WRAP_CHECK(n,t), |
| 51 | #endif |
| 14ab3472 |
52 | |
| 0ba8c7aa |
53 | #define SvPV_renew(sv,n) \ |
| 54 | STMT_START { SvLEN_set(sv, n); \ |
| 55 | SvPV_set((sv), (MEM_WRAP_CHECK_(n,char) \ |
| 56 | (char*)saferealloc((Malloc_t)SvPVX(sv), \ |
| 57 | (MEM_SIZE)((n))))); \ |
| 58 | } STMT_END |
| e807ee50 |
59 | |
| d6e7537a |
60 | #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) |
| 61 | |
| c630715a |
62 | /* On MacOS, respect nonbreaking spaces */ |
| 63 | #ifdef MACOS_TRADITIONAL |
| 64 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t') |
| 65 | #else |
| 66 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t') |
| 67 | #endif |
| e807ee50 |
68 | |
| 2688337b |
69 | /* |
| 70 | * Normally, during compile time, PL_curcop == &PL_compiling is true. However, |
| 71 | * Devel::Declare makes the interpreter call back to perl during compile time, |
| 72 | * which temporarily enters runtime. Then perl space calls various functions |
| 73 | * from this file, which are designed to work during compile time. They all |
| 74 | * happen to operate on PL_curcop, not PL_compiling. That doesn't make a |
| 75 | * difference in the core, but it does for Devel::Declare, which operates at |
| 76 | * runtime, but still wants to mangle the things that are about to be compiled. |
| 77 | * That's why we define our own PL_curcop and make it point to PL_compiling |
| 78 | * here. |
| 79 | */ |
| df718437 |
80 | #undef PL_curcop |
| 81 | #define PL_curcop (&PL_compiling) |
| 2688337b |
82 | |
| 0ba8c7aa |
83 | #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) |
| 84 | |
| e807ee50 |
85 | #define LEX_NORMAL 10 /* normal code (ie not within "...") */ |
| 86 | #define LEX_INTERPNORMAL 9 /* code within a string, eg "$foo[$x+1]" */ |
| 87 | #define LEX_INTERPCASEMOD 8 /* expecting a \U, \Q or \E etc */ |
| 88 | #define LEX_INTERPPUSH 7 /* starting a new sublex parse level */ |
| 89 | #define LEX_INTERPSTART 6 /* expecting the start of a $var */ |
| 90 | |
| 91 | /* at end of code, eg "$x" followed by: */ |
| 92 | #define LEX_INTERPEND 5 /* ... eg not one of [, { or -> */ |
| 93 | #define LEX_INTERPENDMAYBE 4 /* ... eg one of [, { or -> */ |
| 94 | |
| 95 | #define LEX_INTERPCONCAT 3 /* expecting anything, eg at start of |
| 96 | string or after \E, $foo, etc */ |
| 97 | #define LEX_INTERPCONST 2 /* NOT USED */ |
| 98 | #define LEX_FORMLINE 1 /* expecting a format line */ |
| 99 | #define LEX_KNOWNEXT 0 /* next token known; just return it */ |
| 100 | |
| 14ab3472 |
101 | /* and these two are my own madness (mst) */ |
| 71511651 |
102 | |
| 103 | #if PERL_REVISION == 5 && PERL_VERSION == 8 && PERL_SUBVERSION >= 8 |
| 104 | #define PERL_5_8_8_PLUS |
| 105 | #endif |
| 106 | |
| e38f5a74 |
107 | #if PERL_REVISION == 5 && PERL_VERSION > 8 |
| 108 | #define PERL_5_9_PLUS |
| 109 | #endif |
| 110 | |
| 1479b0b1 |
111 | #if !defined(PERL_5_9_PLUS) && defined(PERL_IMPLICIT_CONTEXT) |
| ef91f92b |
112 | /* These two are not exported from the core on Windows. With 5.9+ |
| 113 | it's not an issue, because they're part of the PL_parser structure, |
| 114 | which is exported. On multiplicity/thread builds we can work |
| 115 | around the lack of export by this formulation, where we provide |
| 116 | a substitute implementation of the unexported accessor functions. |
| 117 | On single-interpreter builds we can't, because access is directly |
| 118 | via symbols that are not exported. */ |
| 119 | # define Perl_Ilinestart_ptr my_Ilinestart_ptr |
| 120 | char **my_Ilinestart_ptr(pTHX) { return &(aTHX->Ilinestart); } |
| 121 | # define Perl_Isublex_info_ptr my_Isublex_info_ptr |
| 122 | static SUBLEXINFO *my_Isublex_info_ptr(pTHX) { return &(aTHX->Isublex_info); } |
| 123 | #endif |
| 124 | |
| e38f5a74 |
125 | #ifdef PERL_5_9_PLUS |
| 14ab3472 |
126 | /* 5.9+ moves a bunch of things to a PL_parser struct so we need to |
| 127 | declare the backcompat macros for things to still work (mst) */ |
| 128 | |
| 129 | /* XXX temporary backwards compatibility */ |
| 130 | #define PL_lex_brackets (PL_parser->lex_brackets) |
| 131 | #define PL_lex_brackstack (PL_parser->lex_brackstack) |
| 132 | #define PL_lex_casemods (PL_parser->lex_casemods) |
| 133 | #define PL_lex_casestack (PL_parser->lex_casestack) |
| 134 | #define PL_lex_defer (PL_parser->lex_defer) |
| 135 | #define PL_lex_dojoin (PL_parser->lex_dojoin) |
| 136 | #define PL_lex_expect (PL_parser->lex_expect) |
| 137 | #define PL_lex_formbrack (PL_parser->lex_formbrack) |
| 138 | #define PL_lex_inpat (PL_parser->lex_inpat) |
| 139 | #define PL_lex_inwhat (PL_parser->lex_inwhat) |
| 140 | #define PL_lex_op (PL_parser->lex_op) |
| 141 | #define PL_lex_repl (PL_parser->lex_repl) |
| 142 | #define PL_lex_starts (PL_parser->lex_starts) |
| 143 | #define PL_lex_stuff (PL_parser->lex_stuff) |
| 144 | #define PL_multi_start (PL_parser->multi_start) |
| 145 | #define PL_multi_open (PL_parser->multi_open) |
| 146 | #define PL_multi_close (PL_parser->multi_close) |
| 147 | #define PL_pending_ident (PL_parser->pending_ident) |
| 148 | #define PL_preambled (PL_parser->preambled) |
| 149 | #define PL_sublex_info (PL_parser->sublex_info) |
| 150 | #define PL_linestr (PL_parser->linestr) |
| 151 | #define PL_sublex_info (PL_parser->sublex_info) |
| 152 | #define PL_linestr (PL_parser->linestr) |
| 153 | #define PL_expect (PL_parser->expect) |
| 154 | #define PL_copline (PL_parser->copline) |
| 155 | #define PL_bufptr (PL_parser->bufptr) |
| 156 | #define PL_oldbufptr (PL_parser->oldbufptr) |
| 157 | #define PL_oldoldbufptr (PL_parser->oldoldbufptr) |
| 158 | #define PL_linestart (PL_parser->linestart) |
| 159 | #define PL_bufend (PL_parser->bufend) |
| 160 | #define PL_last_uni (PL_parser->last_uni) |
| 161 | #define PL_last_lop (PL_parser->last_lop) |
| 162 | #define PL_last_lop_op (PL_parser->last_lop_op) |
| 163 | #define PL_lex_state (PL_parser->lex_state) |
| 164 | #define PL_rsfp (PL_parser->rsfp) |
| 165 | #define PL_rsfp_filters (PL_parser->rsfp_filters) |
| 166 | #define PL_in_my (PL_parser->in_my) |
| 167 | #define PL_in_my_stash (PL_parser->in_my_stash) |
| 168 | #define PL_tokenbuf (PL_parser->tokenbuf) |
| 169 | #define PL_multi_end (PL_parser->multi_end) |
| 170 | #define PL_error_count (PL_parser->error_count) |
| 52eaf5fa |
171 | #define PL_nexttoke (PL_parser->nexttoke) |
| 172 | /* these are from the non-PERL_MAD path but I don't -think- I need |
| 14ab3472 |
173 | the PERL_MAD stuff since my code isn't really populating things (mst) */ |
| 52eaf5fa |
174 | # ifdef PERL_MAD |
| 175 | # define PL_curforce (PL_parser->curforce) |
| 176 | # define PL_lasttoke (PL_parser->lasttoke) |
| 177 | # else |
| 14ab3472 |
178 | # define PL_nexttype (PL_parser->nexttype) |
| 179 | # define PL_nextval (PL_parser->nextval) |
| 52eaf5fa |
180 | # endif |
| 181 | /* end of backcompat macros from 5.9 toke.c (mst) */ |
| 78160085 |
182 | #endif |
| 14ab3472 |
183 | |
| 184 | /* when ccflags include -DDEBUGGING we need this for earlier 5.8 perls */ |
| 185 | #ifndef SvPV_nolen_const |
| 186 | #define SvPV_nolen_const SvPV_nolen |
| 187 | #endif |
| 188 | |
| 3f61a25a |
189 | /* Name changed in 5.17; use new name in our code. Apparently we're meant |
| 190 | to use something else instead, but no non-underscored way to achieve |
| 191 | this is apparent. */ |
| 192 | |
| 193 | #ifndef _is_utf8_mark |
| 194 | #define _is_utf8_mark is_utf8_mark |
| 195 | #endif |
| 196 | |
| 197 | /* utf8_to_uvchr_buf() not defined in earlier perls, but less-capable |
| 198 | * substitute is available */ |
| 199 | |
| 200 | #ifndef utf8_to_uvchr_buf |
| 201 | #define utf8_to_uvchr_buf(s, e, lp) ((e), utf8_to_uvchr(s, lp)) |
| 202 | #endif |
| 203 | |
| 71511651 |
204 | /* and now we're back to the toke.c stuff again (mst) */ |
| 205 | |
| e807ee50 |
206 | static const char ident_too_long[] = |
| 207 | "Identifier too long"; |
| 208 | static const char c_without_g[] = |
| 209 | "Use of /c modifier is meaningless without /g"; |
| 210 | static const char c_in_subst[] = |
| 211 | "Use of /c modifier is meaningless in s///"; |
| 212 | |
| 213 | #ifdef USE_UTF8_SCRIPTS |
| 214 | # define UTF (!IN_BYTES) |
| 215 | #else |
| 216 | # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8)) |
| 217 | #endif |
| 218 | |
| 219 | /* Invoke the idxth filter function for the current rsfp. */ |
| 220 | /* maxlen 0 = read one text line */ |
| 221 | I32 |
| 222 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
| 223 | { |
| 224 | filter_t funcp; |
| 225 | SV *datasv = NULL; |
| 226 | |
| 227 | if (!PL_rsfp_filters) |
| 228 | return -1; |
| 229 | if (idx > AvFILLp(PL_rsfp_filters)) { /* Any more filters? */ |
| 230 | /* Provide a default input filter to make life easy. */ |
| 231 | /* Note that we append to the line. This is handy. */ |
| 232 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
| 233 | "filter_read %d: from rsfp\n", idx)); |
| 234 | if (maxlen) { |
| 235 | /* Want a block */ |
| 236 | int len ; |
| 237 | const int old_len = SvCUR(buf_sv); |
| 238 | |
| 239 | /* ensure buf_sv is large enough */ |
| 240 | SvGROW(buf_sv, (STRLEN)(old_len + maxlen)) ; |
| 241 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
| 242 | if (PerlIO_error(PL_rsfp)) |
| 243 | return -1; /* error */ |
| 244 | else |
| 245 | return 0 ; /* end of file */ |
| 246 | } |
| 247 | SvCUR_set(buf_sv, old_len + len) ; |
| 248 | } else { |
| 249 | /* Want a line */ |
| 250 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
| 251 | if (PerlIO_error(PL_rsfp)) |
| 252 | return -1; /* error */ |
| 253 | else |
| 254 | return 0 ; /* end of file */ |
| 255 | } |
| 256 | } |
| 257 | return SvCUR(buf_sv); |
| 258 | } |
| 259 | /* Skip this filter slot if filter has been deleted */ |
| 260 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef) { |
| 261 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
| 262 | "filter_read %d: skipped (filter deleted)\n", |
| 263 | idx)); |
| 264 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
| 265 | } |
| 266 | /* Get function pointer hidden within datasv */ |
| 267 | funcp = DPTR2FPTR(filter_t, IoANY(datasv)); |
| 268 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
| 269 | "filter_read %d: via function %p (%s)\n", |
| 270 | idx, datasv, SvPV_nolen_const(datasv))); |
| 271 | /* Call function. The function is expected to */ |
| 272 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ |
| 273 | /* Return: <0:error, =0:eof, >0:not eof */ |
| 274 | return (*funcp)(aTHX_ idx, buf_sv, maxlen); |
| 275 | } |
| 276 | |
| 277 | STATIC char * |
| 278 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
| 279 | { |
| 280 | #ifdef PERL_CR_FILTER |
| 281 | if (!PL_rsfp_filters) { |
| 282 | filter_add(S_cr_textfilter,NULL); |
| 283 | } |
| 284 | #endif |
| 285 | if (PL_rsfp_filters) { |
| 286 | if (!append) |
| 287 | SvCUR_set(sv, 0); /* start with empty line */ |
| 288 | if (FILTER_READ(0, sv, 0) > 0) |
| 289 | return ( SvPVX(sv) ) ; |
| 290 | else |
| 291 | return Nullch ; |
| 292 | } |
| 293 | else |
| 294 | return (sv_gets(sv, fp, append)); |
| 295 | } |
| 296 | |
| 297 | /* |
| 298 | * S_skipspace |
| 299 | * Called to gobble the appropriate amount and type of whitespace. |
| 300 | * Skips comments as well. |
| 301 | */ |
| 302 | |
| 303 | STATIC char * |
| a25db2dc |
304 | S_skipspace(pTHX_ register char *s, int incline) |
| e807ee50 |
305 | { |
| 306 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
| 307 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
| 308 | s++; |
| 309 | return s; |
| 310 | } |
| 311 | for (;;) { |
| 312 | STRLEN prevlen; |
| 313 | SSize_t oldprevlen, oldoldprevlen; |
| 314 | SSize_t oldloplen = 0, oldunilen = 0; |
| 315 | while (s < PL_bufend && isSPACE(*s)) { |
| ec25cea7 |
316 | if (*s++ == '\n' && ((incline == 2) || (PL_in_eval && !PL_rsfp && !incline))) |
| e807ee50 |
317 | incline(s); |
| 318 | } |
| 319 | |
| 320 | /* comment */ |
| 321 | if (s < PL_bufend && *s == '#') { |
| 322 | while (s < PL_bufend && *s != '\n') |
| 323 | s++; |
| 324 | if (s < PL_bufend) { |
| 325 | s++; |
| a25db2dc |
326 | if (PL_in_eval && !PL_rsfp && !incline) { |
| e807ee50 |
327 | incline(s); |
| 328 | continue; |
| 329 | } |
| 330 | } |
| 331 | } |
| 332 | |
| a25db2dc |
333 | /* also skip leading whitespace on the beginning of a line before deciding |
| 334 | * whether or not to recharge the linestr. --rafl |
| 335 | */ |
| 336 | while (s < PL_bufend && isSPACE(*s)) { |
| 337 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp && !incline) |
| 338 | incline(s); |
| 339 | } |
| 340 | |
| e807ee50 |
341 | /* only continue to recharge the buffer if we're at the end |
| 342 | * of the buffer, we're not reading from a source filter, and |
| 343 | * we're in normal lexing mode |
| 344 | */ |
| 345 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
| 346 | PL_lex_state == LEX_FORMLINE) |
| 347 | return s; |
| 348 | |
| 349 | /* try to recharge the buffer */ |
| 350 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
| 351 | (prevlen = SvCUR(PL_linestr)))) == Nullch) |
| 352 | { |
| 353 | /* end of file. Add on the -p or -n magic */ |
| 354 | if (PL_minus_p) { |
| 355 | sv_setpv(PL_linestr, |
| 356 | ";}continue{print or die qq(-p destination: $!\\n);}"); |
| 357 | PL_minus_n = PL_minus_p = 0; |
| 358 | } |
| 359 | else if (PL_minus_n) { |
| 360 | sv_setpvn(PL_linestr, ";}", 2); |
| 361 | PL_minus_n = 0; |
| 362 | } |
| 363 | else |
| 364 | sv_setpvn(PL_linestr,";", 1); |
| 365 | |
| 366 | /* reset variables for next time we lex */ |
| 367 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
| 368 | = SvPVX(PL_linestr); |
| 369 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 370 | PL_last_lop = PL_last_uni = Nullch; |
| 371 | |
| 35fc8ada |
372 | /* In perl versions previous to p4-rawid: //depot/perl@32954 -P |
| 373 | * preprocessors were supported here. We don't support -P at all, even |
| 374 | * on perls that support it, and use the following chunk from blead |
| 375 | * perl. (rafl) |
| 376 | */ |
| 377 | |
| 378 | /* Close the filehandle. Could be from |
| e807ee50 |
379 | * STDIN, or a regular file. If we were reading code from |
| 380 | * STDIN (because the commandline held no -e or filename) |
| 381 | * then we don't close it, we reset it so the code can |
| 382 | * read from STDIN too. |
| 383 | */ |
| 384 | |
| 35fc8ada |
385 | if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
| e807ee50 |
386 | PerlIO_clearerr(PL_rsfp); |
| 387 | else |
| 388 | (void)PerlIO_close(PL_rsfp); |
| 389 | PL_rsfp = Nullfp; |
| 390 | return s; |
| 391 | } |
| 392 | |
| 393 | /* not at end of file, so we only read another line */ |
| 394 | /* make corresponding updates to old pointers, for yyerror() */ |
| 395 | oldprevlen = PL_oldbufptr - PL_bufend; |
| 396 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; |
| 397 | if (PL_last_uni) |
| 398 | oldunilen = PL_last_uni - PL_bufend; |
| 399 | if (PL_last_lop) |
| 400 | oldloplen = PL_last_lop - PL_bufend; |
| 401 | PL_linestart = PL_bufptr = s + prevlen; |
| 402 | PL_bufend = s + SvCUR(PL_linestr); |
| 403 | s = PL_bufptr; |
| 404 | PL_oldbufptr = s + oldprevlen; |
| 405 | PL_oldoldbufptr = s + oldoldprevlen; |
| 406 | if (PL_last_uni) |
| 407 | PL_last_uni = s + oldunilen; |
| 408 | if (PL_last_lop) |
| 409 | PL_last_lop = s + oldloplen; |
| a25db2dc |
410 | if (!incline) |
| 411 | incline(s); |
| e807ee50 |
412 | |
| 413 | /* debugger active and we're not compiling the debugger code, |
| 414 | * so store the line into the debugger's array of lines |
| 415 | */ |
| 416 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| ec25cea7 |
417 | AV *fileav = CopFILEAV(PL_curcop); |
| 418 | if (fileav) { |
| 419 | SV * const sv = NEWSV(85,0); |
| 420 | sv_upgrade(sv, SVt_PVMG); |
| 421 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
| 422 | (void)SvIOK_on(sv); |
| 423 | SvIV_set(sv, 0); |
| 424 | av_store(fileav,(I32)CopLINE(PL_curcop),sv); |
| 425 | } |
| e807ee50 |
426 | } |
| 427 | } |
| 428 | } |
| 429 | |
| 430 | STATIC char * |
| 431 | S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp) |
| 432 | { |
| 433 | register char *d = dest; |
| 434 | register char * const e = d + destlen - 3; /* two-character token, ending NUL */ |
| 435 | for (;;) { |
| 436 | if (d >= e) |
| 437 | Perl_croak(aTHX_ ident_too_long); |
| 438 | if (isALNUM(*s)) /* UTF handled below */ |
| 439 | *d++ = *s++; |
| 440 | else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) { |
| 441 | *d++ = ':'; |
| 442 | *d++ = ':'; |
| 443 | s++; |
| 444 | } |
| 445 | else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') { |
| 446 | *d++ = *s++; |
| 447 | *d++ = *s++; |
| 448 | } |
| 449 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
| 450 | char *t = s + UTF8SKIP(s); |
| 3f61a25a |
451 | while (UTF8_IS_CONTINUED(*t) && _is_utf8_mark((U8*)t)) |
| e807ee50 |
452 | t += UTF8SKIP(t); |
| 453 | if (d + (t - s) > e) |
| 454 | Perl_croak(aTHX_ ident_too_long); |
| 455 | Copy(s, d, t - s, char); |
| 456 | d += t - s; |
| 457 | s = t; |
| 458 | } |
| 459 | else { |
| 460 | *d = '\0'; |
| 461 | *slp = d - dest; |
| 462 | return s; |
| 463 | } |
| 464 | } |
| 465 | } |
| 466 | |
| 467 | /* |
| 468 | * S_incline |
| 469 | * This subroutine has nothing to do with tilting, whether at windmills |
| 470 | * or pinball tables. Its name is short for "increment line". It |
| 471 | * increments the current line number in CopLINE(PL_curcop) and checks |
| 472 | * to see whether the line starts with a comment of the form |
| 473 | * # line 500 "foo.pm" |
| 474 | * If so, it sets the current line number and file to the values in the comment. |
| 475 | */ |
| 476 | |
| 477 | STATIC void |
| 78160085 |
478 | S_incline(pTHX_ char *s) |
| e807ee50 |
479 | { |
| 480 | char *t; |
| 78160085 |
481 | char *n; |
| 482 | char *e; |
| e807ee50 |
483 | char ch; |
| 484 | |
| 485 | CopLINE_inc(PL_curcop); |
| 486 | if (*s++ != '#') |
| 487 | return; |
| 488 | while (SPACE_OR_TAB(*s)) s++; |
| 489 | if (strnEQ(s, "line", 4)) |
| 490 | s += 4; |
| 491 | else |
| 492 | return; |
| 493 | if (SPACE_OR_TAB(*s)) |
| 494 | s++; |
| 495 | else |
| 496 | return; |
| 497 | while (SPACE_OR_TAB(*s)) s++; |
| 498 | if (!isDIGIT(*s)) |
| 499 | return; |
| 500 | n = s; |
| 501 | while (isDIGIT(*s)) |
| 502 | s++; |
| 503 | while (SPACE_OR_TAB(*s)) |
| 504 | s++; |
| 505 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
| 506 | s++; |
| 507 | e = t + 1; |
| 508 | } |
| 509 | else { |
| 78160085 |
510 | for (t = s; !isSPACE(*t); t++) ; |
| e807ee50 |
511 | e = t; |
| 512 | } |
| 513 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
| 514 | e++; |
| 515 | if (*e != '\n' && *e != '\0') |
| 516 | return; /* false alarm */ |
| 517 | |
| 518 | ch = *t; |
| 519 | *t = '\0'; |
| 520 | if (t - s > 0) { |
| 71511651 |
521 | /* this chunk was added to S_incline during 5.8.8. I don't know why but I don't |
| 522 | honestly care since I probably want to be bug-compatible anyway (mst) */ |
| 523 | |
| 524 | /* ... my kingdom for a perl parser in perl ... (mst) */ |
| 525 | |
| 526 | #ifdef PERL_5_8_8_PLUS |
| e807ee50 |
527 | #ifndef USE_ITHREADS |
| 528 | const char *cf = CopFILE(PL_curcop); |
| 529 | if (cf && strlen(cf) > 7 && strnEQ(cf, "(eval ", 6)) { |
| 530 | /* must copy *{"::_<(eval N)[oldfilename:L]"} |
| 531 | * to *{"::_<newfilename"} */ |
| 532 | char smallbuf[256], smallbuf2[256]; |
| 533 | char *tmpbuf, *tmpbuf2; |
| 534 | GV **gvp, *gv2; |
| 535 | STRLEN tmplen = strlen(cf); |
| 536 | STRLEN tmplen2 = strlen(s); |
| 537 | if (tmplen + 3 < sizeof smallbuf) |
| 538 | tmpbuf = smallbuf; |
| 539 | else |
| 540 | Newx(tmpbuf, tmplen + 3, char); |
| 541 | if (tmplen2 + 3 < sizeof smallbuf2) |
| 542 | tmpbuf2 = smallbuf2; |
| 543 | else |
| 544 | Newx(tmpbuf2, tmplen2 + 3, char); |
| 545 | tmpbuf[0] = tmpbuf2[0] = '_'; |
| 546 | tmpbuf[1] = tmpbuf2[1] = '<'; |
| 547 | memcpy(tmpbuf + 2, cf, ++tmplen); |
| 548 | memcpy(tmpbuf2 + 2, s, ++tmplen2); |
| 549 | ++tmplen; ++tmplen2; |
| 550 | gvp = (GV**)hv_fetch(PL_defstash, tmpbuf, tmplen, FALSE); |
| 551 | if (gvp) { |
| 552 | gv2 = *(GV**)hv_fetch(PL_defstash, tmpbuf2, tmplen2, TRUE); |
| 553 | if (!isGV(gv2)) |
| 554 | gv_init(gv2, PL_defstash, tmpbuf2, tmplen2, FALSE); |
| 555 | /* adjust ${"::_<newfilename"} to store the new file name */ |
| 556 | GvSV(gv2) = newSVpvn(tmpbuf2 + 2, tmplen2 - 2); |
| 557 | GvHV(gv2) = (HV*)SvREFCNT_inc(GvHV(*gvp)); |
| 558 | GvAV(gv2) = (AV*)SvREFCNT_inc(GvAV(*gvp)); |
| 559 | } |
| 560 | if (tmpbuf != smallbuf) Safefree(tmpbuf); |
| 561 | if (tmpbuf2 != smallbuf2) Safefree(tmpbuf2); |
| 562 | } |
| 563 | #endif |
| 71511651 |
564 | #endif |
| 565 | /* second endif closes out the "are we 5.8.(8+)" conditional */ |
| e807ee50 |
566 | CopFILE_free(PL_curcop); |
| 567 | CopFILE_set(PL_curcop, s); |
| 568 | } |
| 569 | *t = ch; |
| 570 | CopLINE_set(PL_curcop, atoi(n)-1); |
| 571 | } |
| 0ba8c7aa |
572 | |
| 573 | /* scan_str |
| 574 | takes: start position in buffer |
| 575 | keep_quoted preserve \ on the embedded delimiter(s) |
| 576 | keep_delims preserve the delimiters around the string |
| 577 | returns: position to continue reading from buffer |
| 578 | side-effects: multi_start, multi_close, lex_repl or lex_stuff, and |
| 579 | updates the read buffer. |
| 580 | |
| 581 | This subroutine pulls a string out of the input. It is called for: |
| 582 | q single quotes q(literal text) |
| 583 | ' single quotes 'literal text' |
| 584 | qq double quotes qq(interpolate $here please) |
| 585 | " double quotes "interpolate $here please" |
| 586 | qx backticks qx(/bin/ls -l) |
| 587 | ` backticks `/bin/ls -l` |
| 588 | qw quote words @EXPORT_OK = qw( func() $spam ) |
| 589 | m// regexp match m/this/ |
| 590 | s/// regexp substitute s/this/that/ |
| 591 | tr/// string transliterate tr/this/that/ |
| 592 | y/// string transliterate y/this/that/ |
| 593 | ($*@) sub prototypes sub foo ($) |
| 594 | (stuff) sub attr parameters sub foo : attr(stuff) |
| 595 | <> readline or globs <FOO>, <>, <$fh>, or <*.c> |
| 596 | |
| 597 | In most of these cases (all but <>, patterns and transliterate) |
| 598 | yylex() calls scan_str(). m// makes yylex() call scan_pat() which |
| 599 | calls scan_str(). s/// makes yylex() call scan_subst() which calls |
| 600 | scan_str(). tr/// and y/// make yylex() call scan_trans() which |
| 601 | calls scan_str(). |
| 602 | |
| 603 | It skips whitespace before the string starts, and treats the first |
| 604 | character as the delimiter. If the delimiter is one of ([{< then |
| 605 | the corresponding "close" character )]}> is used as the closing |
| 606 | delimiter. It allows quoting of delimiters, and if the string has |
| 607 | balanced delimiters ([{<>}]) it allows nesting. |
| 608 | |
| 609 | On success, the SV with the resulting string is put into lex_stuff or, |
| 610 | if that is already non-NULL, into lex_repl. The second case occurs only |
| 611 | when parsing the RHS of the special constructs s/// and tr/// (y///). |
| 612 | For convenience, the terminating delimiter character is stuffed into |
| 613 | SvIVX of the SV. |
| 614 | */ |
| 615 | |
| 616 | STATIC char * |
| 617 | S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims) |
| 618 | { |
| 619 | SV *sv; /* scalar value: string */ |
| 620 | char *tmps; /* temp string, used for delimiter matching */ |
| 621 | register char *s = start; /* current position in the buffer */ |
| 622 | register char term; /* terminating character */ |
| 623 | register char *to; /* current position in the sv's data */ |
| 624 | I32 brackets = 1; /* bracket nesting level */ |
| 625 | bool has_utf8 = FALSE; /* is there any utf8 content? */ |
| 626 | I32 termcode; /* terminating char. code */ |
| 71511651 |
627 | /* 5.8.7+ uses UTF8_MAXBYTES but also its utf8.h defs _MAXLEN to it so |
| 628 | I'm reasonably hopeful this won't destroy anything (mst) */ |
| 629 | U8 termstr[UTF8_MAXLEN]; /* terminating string */ |
| 0ba8c7aa |
630 | STRLEN termlen; /* length of terminating string */ |
| 631 | char *last = NULL; /* last position for nesting bracket */ |
| 632 | |
| 633 | /* skip space before the delimiter */ |
| 634 | if (isSPACE(*s)) |
| 635 | s = skipspace(s); |
| 636 | |
| 637 | /* mark where we are, in case we need to report errors */ |
| 638 | CLINE; |
| 639 | |
| 640 | /* after skipping whitespace, the next character is the terminator */ |
| 641 | term = *s; |
| 642 | if (!UTF) { |
| 643 | termcode = termstr[0] = term; |
| 644 | termlen = 1; |
| 645 | } |
| 646 | else { |
| 3f61a25a |
647 | termcode = utf8_to_uvchr_buf((U8*)s, PL_bufend, &termlen); |
| 0ba8c7aa |
648 | Copy(s, termstr, termlen, U8); |
| 649 | if (!UTF8_IS_INVARIANT(term)) |
| 650 | has_utf8 = TRUE; |
| 651 | } |
| 652 | |
| 653 | /* mark where we are */ |
| 654 | PL_multi_start = CopLINE(PL_curcop); |
| 655 | PL_multi_open = term; |
| 656 | |
| 657 | /* find corresponding closing delimiter */ |
| 658 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) |
| 659 | termcode = termstr[0] = term = tmps[5]; |
| 660 | |
| 661 | PL_multi_close = term; |
| 662 | |
| 663 | /* create a new SV to hold the contents. 87 is leak category, I'm |
| 664 | assuming. 79 is the SV's initial length. What a random number. */ |
| 665 | sv = NEWSV(87,79); |
| 666 | sv_upgrade(sv, SVt_PVIV); |
| 667 | SvIV_set(sv, termcode); |
| 668 | (void)SvPOK_only(sv); /* validate pointer */ |
| 669 | |
| 670 | /* move past delimiter and try to read a complete string */ |
| 671 | if (keep_delims) |
| 672 | sv_catpvn(sv, s, termlen); |
| 673 | s += termlen; |
| 674 | for (;;) { |
| 675 | if (PL_encoding && !UTF) { |
| 676 | bool cont = TRUE; |
| 677 | |
| 678 | while (cont) { |
| 679 | int offset = s - SvPVX_const(PL_linestr); |
| 680 | const bool found = sv_cat_decode(sv, PL_encoding, PL_linestr, |
| 681 | &offset, (char*)termstr, termlen); |
| 682 | const char *ns = SvPVX_const(PL_linestr) + offset; |
| 683 | char *svlast = SvEND(sv) - 1; |
| 684 | |
| 685 | for (; s < ns; s++) { |
| 686 | if (*s == '\n' && !PL_rsfp) |
| 687 | CopLINE_inc(PL_curcop); |
| 688 | } |
| 689 | if (!found) |
| 690 | goto read_more_line; |
| 691 | else { |
| 692 | /* handle quoted delimiters */ |
| 693 | if (SvCUR(sv) > 1 && *(svlast-1) == '\\') { |
| 694 | const char *t; |
| 695 | for (t = svlast-2; t >= SvPVX_const(sv) && *t == '\\';) |
| 696 | t--; |
| 697 | if ((svlast-1 - t) % 2) { |
| 698 | if (!keep_quoted) { |
| 699 | *(svlast-1) = term; |
| 700 | *svlast = '\0'; |
| 701 | SvCUR_set(sv, SvCUR(sv) - 1); |
| 702 | } |
| 703 | continue; |
| 704 | } |
| 705 | } |
| 706 | if (PL_multi_open == PL_multi_close) { |
| 707 | cont = FALSE; |
| 708 | } |
| 709 | else { |
| 710 | const char *t; |
| 711 | char *w; |
| 712 | if (!last) |
| 713 | last = SvPVX(sv); |
| 714 | for (t = w = last; t < svlast; w++, t++) { |
| 715 | /* At here, all closes are "was quoted" one, |
| 716 | so we don't check PL_multi_close. */ |
| 717 | if (*t == '\\') { |
| 718 | if (!keep_quoted && *(t+1) == PL_multi_open) |
| 719 | t++; |
| 720 | else |
| 721 | *w++ = *t++; |
| 722 | } |
| 723 | else if (*t == PL_multi_open) |
| 724 | brackets++; |
| 725 | |
| 726 | *w = *t; |
| 727 | } |
| 728 | if (w < t) { |
| 729 | *w++ = term; |
| 730 | *w = '\0'; |
| 731 | SvCUR_set(sv, w - SvPVX_const(sv)); |
| 732 | } |
| 733 | last = w; |
| 734 | if (--brackets <= 0) |
| 735 | cont = FALSE; |
| 736 | } |
| 737 | } |
| 738 | } |
| 739 | if (!keep_delims) { |
| 740 | SvCUR_set(sv, SvCUR(sv) - 1); |
| 741 | *SvEND(sv) = '\0'; |
| 742 | } |
| 743 | break; |
| 744 | } |
| 745 | |
| 746 | /* extend sv if need be */ |
| 747 | SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1); |
| 748 | /* set 'to' to the next character in the sv's string */ |
| 749 | to = SvPVX(sv)+SvCUR(sv); |
| 750 | |
| 751 | /* if open delimiter is the close delimiter read unbridle */ |
| 752 | if (PL_multi_open == PL_multi_close) { |
| 753 | for (; s < PL_bufend; s++,to++) { |
| 754 | /* embedded newlines increment the current line number */ |
| 755 | if (*s == '\n' && !PL_rsfp) |
| 756 | CopLINE_inc(PL_curcop); |
| 757 | /* handle quoted delimiters */ |
| 758 | if (*s == '\\' && s+1 < PL_bufend && term != '\\') { |
| 759 | if (!keep_quoted && s[1] == term) |
| 760 | s++; |
| 761 | /* any other quotes are simply copied straight through */ |
| 762 | else |
| 763 | *to++ = *s++; |
| 764 | } |
| 765 | /* terminate when run out of buffer (the for() condition), or |
| 766 | have found the terminator */ |
| 767 | else if (*s == term) { |
| 768 | if (termlen == 1) |
| 769 | break; |
| 770 | if (s+termlen <= PL_bufend && memEQ(s, (char*)termstr, termlen)) |
| 771 | break; |
| 772 | } |
| 773 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
| 774 | has_utf8 = TRUE; |
| 775 | *to = *s; |
| 776 | } |
| 777 | } |
| 778 | |
| 779 | /* if the terminator isn't the same as the start character (e.g., |
| 780 | matched brackets), we have to allow more in the quoting, and |
| 781 | be prepared for nested brackets. |
| 782 | */ |
| 783 | else { |
| 784 | /* read until we run out of string, or we find the terminator */ |
| 785 | for (; s < PL_bufend; s++,to++) { |
| 786 | /* embedded newlines increment the line count */ |
| 787 | if (*s == '\n' && !PL_rsfp) |
| 788 | CopLINE_inc(PL_curcop); |
| 789 | /* backslashes can escape the open or closing characters */ |
| 790 | if (*s == '\\' && s+1 < PL_bufend) { |
| 791 | if (!keep_quoted && |
| 792 | ((s[1] == PL_multi_open) || (s[1] == PL_multi_close))) |
| 793 | s++; |
| 794 | else |
| 795 | *to++ = *s++; |
| 796 | } |
| 797 | /* allow nested opens and closes */ |
| 798 | else if (*s == PL_multi_close && --brackets <= 0) |
| 799 | break; |
| 800 | else if (*s == PL_multi_open) |
| 801 | brackets++; |
| 802 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
| 803 | has_utf8 = TRUE; |
| 804 | *to = *s; |
| 805 | } |
| 806 | } |
| 807 | /* terminate the copied string and update the sv's end-of-string */ |
| 808 | *to = '\0'; |
| 809 | SvCUR_set(sv, to - SvPVX_const(sv)); |
| 810 | |
| 811 | /* |
| 812 | * this next chunk reads more into the buffer if we're not done yet |
| 813 | */ |
| 814 | |
| 815 | if (s < PL_bufend) |
| 816 | break; /* handle case where we are done yet :-) */ |
| 817 | |
| 818 | #ifndef PERL_STRICT_CR |
| 819 | if (to - SvPVX_const(sv) >= 2) { |
| 820 | if ((to[-2] == '\r' && to[-1] == '\n') || |
| 821 | (to[-2] == '\n' && to[-1] == '\r')) |
| 822 | { |
| 823 | to[-2] = '\n'; |
| 824 | to--; |
| 825 | SvCUR_set(sv, to - SvPVX_const(sv)); |
| 826 | } |
| 827 | else if (to[-1] == '\r') |
| 828 | to[-1] = '\n'; |
| 829 | } |
| 830 | else if (to - SvPVX_const(sv) == 1 && to[-1] == '\r') |
| 831 | to[-1] = '\n'; |
| 832 | #endif |
| 833 | |
| 834 | read_more_line: |
| 835 | /* if we're out of file, or a read fails, bail and reset the current |
| 836 | line marker so we can report where the unterminated string began |
| 837 | */ |
| 838 | if (!PL_rsfp || |
| 839 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { |
| 840 | sv_free(sv); |
| 841 | CopLINE_set(PL_curcop, (line_t)PL_multi_start); |
| 842 | return Nullch; |
| 843 | } |
| 844 | /* we read a line, so increment our line counter */ |
| 845 | CopLINE_inc(PL_curcop); |
| 846 | |
| 847 | /* update debugger info */ |
| 848 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| ec25cea7 |
849 | AV *fileav = CopFILEAV(PL_curcop); |
| 850 | if (fileav) { |
| 851 | SV *sv = NEWSV(88,0); |
| 852 | sv_upgrade(sv, SVt_PVMG); |
| 853 | sv_setsv(sv,PL_linestr); |
| 854 | (void)SvIOK_on(sv); |
| 855 | SvIV_set(sv, 0); |
| 856 | av_store(fileav, (I32)CopLINE(PL_curcop), sv); |
| 857 | } |
| 0ba8c7aa |
858 | } |
| 859 | |
| 860 | /* having changed the buffer, we must update PL_bufend */ |
| 861 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 862 | PL_last_lop = PL_last_uni = Nullch; |
| 863 | } |
| 864 | |
| 865 | /* at this point, we have successfully read the delimited string */ |
| 866 | |
| 867 | if (!PL_encoding || UTF) { |
| 868 | if (keep_delims) |
| 869 | sv_catpvn(sv, s, termlen); |
| 870 | s += termlen; |
| 871 | } |
| 872 | if (has_utf8 || PL_encoding) |
| 873 | SvUTF8_on(sv); |
| 874 | |
| 875 | PL_multi_end = CopLINE(PL_curcop); |
| 876 | |
| 877 | /* if we allocated too much space, give some back */ |
| 878 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
| 879 | SvLEN_set(sv, SvCUR(sv) + 1); |
| 71511651 |
880 | /* 5.8.8 uses SvPV_renew, no prior version actually has the damn thing (mst) */ |
| 881 | #ifdef PERL_5_8_8_PLUS |
| 0ba8c7aa |
882 | SvPV_renew(sv, SvLEN(sv)); |
| 71511651 |
883 | #else |
| 884 | Renew(SvPVX(sv), SvLEN(sv), char); |
| 885 | #endif |
| 0ba8c7aa |
886 | } |
| 887 | |
| 888 | /* decide whether this is the first or second quoted string we've read |
| 889 | for this op |
| 890 | */ |
| 891 | |
| 892 | if (PL_lex_stuff) |
| 893 | PL_lex_repl = sv; |
| 894 | else |
| 895 | PL_lex_stuff = sv; |
| 896 | return s; |
| 897 | } |
| 898 | |
| 92997e69 |
899 | #define XFAKEBRACK 128 |
| 900 | |
| 901 | STATIC char * |
| 902 | S_scan_ident(pTHX_ register char *s, register const char *send, char *dest, STRLEN destlen, I32 ck_uni) |
| 903 | { |
| 904 | register char *d; |
| 905 | register char *e; |
| 906 | char *bracket = Nullch; |
| 907 | char funny = *s++; |
| 908 | |
| 909 | if (isSPACE(*s)) |
| 910 | s = skipspace(s); |
| 911 | d = dest; |
| 912 | e = d + destlen - 3; /* two-character token, ending NUL */ |
| 913 | if (isDIGIT(*s)) { |
| 914 | while (isDIGIT(*s)) { |
| 915 | if (d >= e) |
| 916 | Perl_croak(aTHX_ ident_too_long); |
| 917 | *d++ = *s++; |
| 918 | } |
| 919 | } |
| 920 | else { |
| 921 | for (;;) { |
| 922 | if (d >= e) |
| 923 | Perl_croak(aTHX_ ident_too_long); |
| 924 | if (isALNUM(*s)) /* UTF handled below */ |
| 925 | *d++ = *s++; |
| 926 | else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) { |
| 927 | *d++ = ':'; |
| 928 | *d++ = ':'; |
| 929 | s++; |
| 930 | } |
| 931 | else if (*s == ':' && s[1] == ':') { |
| 932 | *d++ = *s++; |
| 933 | *d++ = *s++; |
| 934 | } |
| 935 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
| 936 | char *t = s + UTF8SKIP(s); |
| 3f61a25a |
937 | while (UTF8_IS_CONTINUED(*t) && _is_utf8_mark((U8*)t)) |
| 92997e69 |
938 | t += UTF8SKIP(t); |
| 939 | if (d + (t - s) > e) |
| 940 | Perl_croak(aTHX_ ident_too_long); |
| 941 | Copy(s, d, t - s, char); |
| 942 | d += t - s; |
| 943 | s = t; |
| 944 | } |
| 945 | else |
| 946 | break; |
| 947 | } |
| 948 | } |
| 949 | *d = '\0'; |
| 950 | d = dest; |
| 951 | if (*d) { |
| 952 | if (PL_lex_state != LEX_NORMAL) |
| 953 | PL_lex_state = LEX_INTERPENDMAYBE; |
| 954 | return s; |
| 955 | } |
| 956 | if (*s == '$' && s[1] && |
| 957 | (isALNUM_lazy_if(s+1,UTF) || s[1] == '$' || s[1] == '{' || strnEQ(s+1,"::",2)) ) |
| 958 | { |
| 959 | return s; |
| 960 | } |
| 961 | if (*s == '{') { |
| 962 | bracket = s; |
| 963 | s++; |
| ec25cea7 |
964 | } else if (ck_uni) { |
| 965 | /* we always call this with ck_uni == 0, so no need for check_uni() */ |
| 966 | /* check_uni(); */ |
| 92997e69 |
967 | } |
| 92997e69 |
968 | if (s < send) |
| 969 | *d = *s++; |
| 970 | d[1] = '\0'; |
| 971 | if (*d == '^' && *s && isCONTROLVAR(*s)) { |
| 972 | *d = toCTRL(*s); |
| 973 | s++; |
| 974 | } |
| 975 | if (bracket) { |
| 976 | if (isSPACE(s[-1])) { |
| 977 | while (s < send) { |
| 978 | const char ch = *s++; |
| 979 | if (!SPACE_OR_TAB(ch)) { |
| 980 | *d = ch; |
| 981 | break; |
| 982 | } |
| 983 | } |
| 984 | } |
| 985 | if (isIDFIRST_lazy_if(d,UTF)) { |
| 986 | d++; |
| 987 | if (UTF) { |
| 988 | e = s; |
| 989 | while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') { |
| 990 | e += UTF8SKIP(e); |
| 3f61a25a |
991 | while (e < send && UTF8_IS_CONTINUED(*e) && _is_utf8_mark((U8*)e)) |
| 92997e69 |
992 | e += UTF8SKIP(e); |
| 993 | } |
| 994 | Copy(s, d, e - s, char); |
| 995 | d += e - s; |
| 996 | s = e; |
| 997 | } |
| 998 | else { |
| 999 | while ((isALNUM(*s) || *s == ':') && d < e) |
| 1000 | *d++ = *s++; |
| 1001 | if (d >= e) |
| 1002 | Perl_croak(aTHX_ ident_too_long); |
| 1003 | } |
| 1004 | *d = '\0'; |
| 1005 | while (s < send && SPACE_OR_TAB(*s)) s++; |
| 1006 | if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) { |
| d6e7537a |
1007 | /* we don't want perl to guess what is meant. the keyword |
| 1008 | * parser decides that later. (rafl) |
| 1009 | */ |
| 1010 | /* |
| 92997e69 |
1011 | if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) { |
| 1012 | const char *brack = *s == '[' ? "[...]" : "{...}"; |
| 1013 | Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS), |
| 1014 | "Ambiguous use of %c{%s%s} resolved to %c%s%s", |
| 1015 | funny, dest, brack, funny, dest, brack); |
| 1016 | } |
| d6e7537a |
1017 | */ |
| 92997e69 |
1018 | bracket++; |
| 1019 | PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK); |
| 1020 | return s; |
| 1021 | } |
| 1022 | } |
| 1023 | /* Handle extended ${^Foo} variables |
| 1024 | * 1999-02-27 mjd-perl-patch@plover.com */ |
| 1025 | else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */ |
| 1026 | && isALNUM(*s)) |
| 1027 | { |
| 1028 | d++; |
| 1029 | while (isALNUM(*s) && d < e) { |
| 1030 | *d++ = *s++; |
| 1031 | } |
| 1032 | if (d >= e) |
| 1033 | Perl_croak(aTHX_ ident_too_long); |
| 1034 | *d = '\0'; |
| 1035 | } |
| 1036 | if (*s == '}') { |
| 1037 | s++; |
| 1038 | if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) { |
| 1039 | PL_lex_state = LEX_INTERPEND; |
| 1040 | PL_expect = XREF; |
| 1041 | } |
| 1042 | if (funny == '#') |
| 1043 | funny = '@'; |
| d6e7537a |
1044 | /* we don't want perl to guess what is meant. the keyword |
| 1045 | * parser decides that later. (rafl) |
| 1046 | */ |
| 1047 | /* |
| 92997e69 |
1048 | if (PL_lex_state == LEX_NORMAL) { |
| 1049 | if (ckWARN(WARN_AMBIGUOUS) && |
| 1050 | (keyword(dest, d - dest) || get_cv(dest, FALSE))) |
| 1051 | { |
| 1052 | Perl_warner(aTHX_ packWARN(WARN_AMBIGUOUS), |
| 1053 | "Ambiguous use of %c{%s} resolved to %c%s", |
| 1054 | funny, dest, funny, dest); |
| 1055 | } |
| 1056 | } |
| d6e7537a |
1057 | */ |
| 92997e69 |
1058 | } |
| 1059 | else { |
| 1060 | s = bracket; /* let the parser handle it */ |
| 1061 | *dest = '\0'; |
| 1062 | } |
| 1063 | } |
| d6e7537a |
1064 | /* don't intuit. we really just want the string. (rafl) */ |
| 1065 | /* |
| 92997e69 |
1066 | else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s)) |
| 1067 | PL_lex_state = LEX_INTERPEND; |
| d6e7537a |
1068 | */ |
| 92997e69 |
1069 | return s; |
| 1070 | } |
projects / p5sagit/Devel-Declare.git / blame