| 1 | /* toke.c |
| 2 | * |
| 3 | * Copyright (c) 1991-2001, Larry Wall |
| 4 | * |
| 5 | * You may distribute under the terms of either the GNU General Public |
| 6 | * License or the Artistic License, as specified in the README file. |
| 7 | * |
| 8 | */ |
| 9 | |
| 10 | /* |
| 11 | * "It all comes from here, the stench and the peril." --Frodo |
| 12 | */ |
| 13 | |
| 14 | /* |
| 15 | * This file is the lexer for Perl. It's closely linked to the |
| 16 | * parser, perly.y. |
| 17 | * |
| 18 | * The main routine is yylex(), which returns the next token. |
| 19 | */ |
| 20 | |
| 21 | #include "EXTERN.h" |
| 22 | #define PERL_IN_TOKE_C |
| 23 | #include "perl.h" |
| 24 | |
| 25 | #define yychar PL_yychar |
| 26 | #define yylval PL_yylval |
| 27 | |
| 28 | static char ident_too_long[] = "Identifier too long"; |
| 29 | |
| 30 | static void restore_rsfp(pTHX_ void *f); |
| 31 | #ifndef PERL_NO_UTF16_FILTER |
| 32 | static I32 utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen); |
| 33 | static I32 utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen); |
| 34 | #endif |
| 35 | |
| 36 | #define XFAKEBRACK 128 |
| 37 | #define XENUMMASK 127 |
| 38 | |
| 39 | #ifdef USE_UTF8_SCRIPTS |
| 40 | # define UTF (!IN_BYTES) |
| 41 | #else |
| 42 | # ifdef EBCDIC /* For now 'use utf8' does not affect tokenizer on EBCDIC */ |
| 43 | # define UTF (PL_linestr && DO_UTF8(PL_linestr)) |
| 44 | # else |
| 45 | # define UTF ((PL_linestr && DO_UTF8(PL_linestr)) || (PL_hints & HINT_UTF8)) |
| 46 | # endif |
| 47 | #endif |
| 48 | |
| 49 | /* In variables named $^X, these are the legal values for X. |
| 50 | * 1999-02-27 mjd-perl-patch@plover.com */ |
| 51 | #define isCONTROLVAR(x) (isUPPER(x) || strchr("[\\]^_?", (x))) |
| 52 | |
| 53 | /* On MacOS, respect nonbreaking spaces */ |
| 54 | #ifdef MACOS_TRADITIONAL |
| 55 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\312'||(c)=='\t') |
| 56 | #else |
| 57 | #define SPACE_OR_TAB(c) ((c)==' '||(c)=='\t') |
| 58 | #endif |
| 59 | |
| 60 | /* LEX_* are values for PL_lex_state, the state of the lexer. |
| 61 | * They are arranged oddly so that the guard on the switch statement |
| 62 | * can get by with a single comparison (if the compiler is smart enough). |
| 63 | */ |
| 64 | |
| 65 | /* #define LEX_NOTPARSING 11 is done in perl.h. */ |
| 66 | |
| 67 | #define LEX_NORMAL 10 |
| 68 | #define LEX_INTERPNORMAL 9 |
| 69 | #define LEX_INTERPCASEMOD 8 |
| 70 | #define LEX_INTERPPUSH 7 |
| 71 | #define LEX_INTERPSTART 6 |
| 72 | #define LEX_INTERPEND 5 |
| 73 | #define LEX_INTERPENDMAYBE 4 |
| 74 | #define LEX_INTERPCONCAT 3 |
| 75 | #define LEX_INTERPCONST 2 |
| 76 | #define LEX_FORMLINE 1 |
| 77 | #define LEX_KNOWNEXT 0 |
| 78 | |
| 79 | #ifdef ff_next |
| 80 | #undef ff_next |
| 81 | #endif |
| 82 | |
| 83 | #ifdef USE_PURE_BISON |
| 84 | # ifndef YYMAXLEVEL |
| 85 | # define YYMAXLEVEL 100 |
| 86 | # endif |
| 87 | YYSTYPE* yylval_pointer[YYMAXLEVEL]; |
| 88 | int* yychar_pointer[YYMAXLEVEL]; |
| 89 | int yyactlevel = -1; |
| 90 | # undef yylval |
| 91 | # undef yychar |
| 92 | # define yylval (*yylval_pointer[yyactlevel]) |
| 93 | # define yychar (*yychar_pointer[yyactlevel]) |
| 94 | # define PERL_YYLEX_PARAM yylval_pointer[yyactlevel],yychar_pointer[yyactlevel] |
| 95 | # undef yylex |
| 96 | # define yylex() Perl_yylex_r(aTHX_ yylval_pointer[yyactlevel],yychar_pointer[yyactlevel]) |
| 97 | #endif |
| 98 | |
| 99 | #include "keywords.h" |
| 100 | |
| 101 | /* CLINE is a macro that ensures PL_copline has a sane value */ |
| 102 | |
| 103 | #ifdef CLINE |
| 104 | #undef CLINE |
| 105 | #endif |
| 106 | #define CLINE (PL_copline = (CopLINE(PL_curcop) < PL_copline ? CopLINE(PL_curcop) : PL_copline)) |
| 107 | |
| 108 | /* |
| 109 | * Convenience functions to return different tokens and prime the |
| 110 | * lexer for the next token. They all take an argument. |
| 111 | * |
| 112 | * TOKEN : generic token (used for '(', DOLSHARP, etc) |
| 113 | * OPERATOR : generic operator |
| 114 | * AOPERATOR : assignment operator |
| 115 | * PREBLOCK : beginning the block after an if, while, foreach, ... |
| 116 | * PRETERMBLOCK : beginning a non-code-defining {} block (eg, hash ref) |
| 117 | * PREREF : *EXPR where EXPR is not a simple identifier |
| 118 | * TERM : expression term |
| 119 | * LOOPX : loop exiting command (goto, last, dump, etc) |
| 120 | * FTST : file test operator |
| 121 | * FUN0 : zero-argument function |
| 122 | * FUN1 : not used, except for not, which isn't a UNIOP |
| 123 | * BOop : bitwise or or xor |
| 124 | * BAop : bitwise and |
| 125 | * SHop : shift operator |
| 126 | * PWop : power operator |
| 127 | * PMop : pattern-matching operator |
| 128 | * Aop : addition-level operator |
| 129 | * Mop : multiplication-level operator |
| 130 | * Eop : equality-testing operator |
| 131 | * Rop : relational operator <= != gt |
| 132 | * |
| 133 | * Also see LOP and lop() below. |
| 134 | */ |
| 135 | |
| 136 | /* Note that REPORT() and REPORT2() will be expressions that supply |
| 137 | * their own trailing comma, not suitable for statements as such. */ |
| 138 | #ifdef DEBUGGING /* Serve -DT. */ |
| 139 | # define REPORT(x,retval) tokereport(x,s,(int)retval), |
| 140 | # define REPORT2(x,retval) tokereport(x,s, yylval.ival), |
| 141 | #else |
| 142 | # define REPORT(x,retval) |
| 143 | # define REPORT2(x,retval) |
| 144 | #endif |
| 145 | |
| 146 | #define TOKEN(retval) return (REPORT2("token",retval) PL_bufptr = s,(int)retval) |
| 147 | #define OPERATOR(retval) return (REPORT2("operator",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval) |
| 148 | #define AOPERATOR(retval) return ao((REPORT2("aop",retval) PL_expect = XTERM, PL_bufptr = s,(int)retval)) |
| 149 | #define PREBLOCK(retval) return (REPORT2("preblock",retval) PL_expect = XBLOCK,PL_bufptr = s,(int)retval) |
| 150 | #define PRETERMBLOCK(retval) return (REPORT2("pretermblock",retval) PL_expect = XTERMBLOCK,PL_bufptr = s,(int)retval) |
| 151 | #define PREREF(retval) return (REPORT2("preref",retval) PL_expect = XREF,PL_bufptr = s,(int)retval) |
| 152 | #define TERM(retval) return (CLINE, REPORT2("term",retval) PL_expect = XOPERATOR, PL_bufptr = s,(int)retval) |
| 153 | #define LOOPX(f) return(yylval.ival=f, REPORT("loopx",f) PL_expect = XTERM,PL_bufptr = s,(int)LOOPEX) |
| 154 | #define FTST(f) return(yylval.ival=f, REPORT("ftst",f) PL_expect = XTERM,PL_bufptr = s,(int)UNIOP) |
| 155 | #define FUN0(f) return(yylval.ival = f, REPORT("fun0",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC0) |
| 156 | #define FUN1(f) return(yylval.ival = f, REPORT("fun1",f) PL_expect = XOPERATOR,PL_bufptr = s,(int)FUNC1) |
| 157 | #define BOop(f) return ao((yylval.ival=f, REPORT("bitorop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITOROP)) |
| 158 | #define BAop(f) return ao((yylval.ival=f, REPORT("bitandop",f) PL_expect = XTERM,PL_bufptr = s,(int)BITANDOP)) |
| 159 | #define SHop(f) return ao((yylval.ival=f, REPORT("shiftop",f) PL_expect = XTERM,PL_bufptr = s,(int)SHIFTOP)) |
| 160 | #define PWop(f) return ao((yylval.ival=f, REPORT("powop",f) PL_expect = XTERM,PL_bufptr = s,(int)POWOP)) |
| 161 | #define PMop(f) return(yylval.ival=f, REPORT("matchop",f) PL_expect = XTERM,PL_bufptr = s,(int)MATCHOP) |
| 162 | #define Aop(f) return ao((yylval.ival=f, REPORT("add",f) PL_expect = XTERM,PL_bufptr = s,(int)ADDOP)) |
| 163 | #define Mop(f) return ao((yylval.ival=f, REPORT("mul",f) PL_expect = XTERM,PL_bufptr = s,(int)MULOP)) |
| 164 | #define Eop(f) return(yylval.ival=f, REPORT("eq",f) PL_expect = XTERM,PL_bufptr = s,(int)EQOP) |
| 165 | #define Rop(f) return(yylval.ival=f, REPORT("rel",f) PL_expect = XTERM,PL_bufptr = s,(int)RELOP) |
| 166 | |
| 167 | /* This bit of chicanery makes a unary function followed by |
| 168 | * a parenthesis into a function with one argument, highest precedence. |
| 169 | */ |
| 170 | #define UNI(f) return(yylval.ival = f, \ |
| 171 | REPORT("uni",f) \ |
| 172 | PL_expect = XTERM, \ |
| 173 | PL_bufptr = s, \ |
| 174 | PL_last_uni = PL_oldbufptr, \ |
| 175 | PL_last_lop_op = f, \ |
| 176 | (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) ) |
| 177 | |
| 178 | #define UNIBRACK(f) return(yylval.ival = f, \ |
| 179 | REPORT("uni",f) \ |
| 180 | PL_bufptr = s, \ |
| 181 | PL_last_uni = PL_oldbufptr, \ |
| 182 | (*s == '(' || (s = skipspace(s), *s == '(') ? (int)FUNC1 : (int)UNIOP) ) |
| 183 | |
| 184 | /* grandfather return to old style */ |
| 185 | #define OLDLOP(f) return(yylval.ival=f,PL_expect = XTERM,PL_bufptr = s,(int)LSTOP) |
| 186 | |
| 187 | #ifdef DEBUGGING |
| 188 | |
| 189 | STATIC void |
| 190 | S_tokereport(pTHX_ char *thing, char* s, I32 rv) |
| 191 | { |
| 192 | DEBUG_T({ |
| 193 | SV* report = newSVpv(thing, 0); |
| 194 | Perl_sv_catpvf(aTHX_ report, ":line %d:%"IVdf":", CopLINE(PL_curcop), |
| 195 | (IV)rv); |
| 196 | |
| 197 | if (s - PL_bufptr > 0) |
| 198 | sv_catpvn(report, PL_bufptr, s - PL_bufptr); |
| 199 | else { |
| 200 | if (PL_oldbufptr && *PL_oldbufptr) |
| 201 | sv_catpv(report, PL_tokenbuf); |
| 202 | } |
| 203 | PerlIO_printf(Perl_debug_log, "### %s\n", SvPV_nolen(report)); |
| 204 | }); |
| 205 | } |
| 206 | |
| 207 | #endif |
| 208 | |
| 209 | /* |
| 210 | * S_ao |
| 211 | * |
| 212 | * This subroutine detects &&= and ||= and turns an ANDAND or OROR |
| 213 | * into an OP_ANDASSIGN or OP_ORASSIGN |
| 214 | */ |
| 215 | |
| 216 | STATIC int |
| 217 | S_ao(pTHX_ int toketype) |
| 218 | { |
| 219 | if (*PL_bufptr == '=') { |
| 220 | PL_bufptr++; |
| 221 | if (toketype == ANDAND) |
| 222 | yylval.ival = OP_ANDASSIGN; |
| 223 | else if (toketype == OROR) |
| 224 | yylval.ival = OP_ORASSIGN; |
| 225 | toketype = ASSIGNOP; |
| 226 | } |
| 227 | return toketype; |
| 228 | } |
| 229 | |
| 230 | /* |
| 231 | * S_no_op |
| 232 | * When Perl expects an operator and finds something else, no_op |
| 233 | * prints the warning. It always prints "<something> found where |
| 234 | * operator expected. It prints "Missing semicolon on previous line?" |
| 235 | * if the surprise occurs at the start of the line. "do you need to |
| 236 | * predeclare ..." is printed out for code like "sub bar; foo bar $x" |
| 237 | * where the compiler doesn't know if foo is a method call or a function. |
| 238 | * It prints "Missing operator before end of line" if there's nothing |
| 239 | * after the missing operator, or "... before <...>" if there is something |
| 240 | * after the missing operator. |
| 241 | */ |
| 242 | |
| 243 | STATIC void |
| 244 | S_no_op(pTHX_ char *what, char *s) |
| 245 | { |
| 246 | char *oldbp = PL_bufptr; |
| 247 | bool is_first = (PL_oldbufptr == PL_linestart); |
| 248 | |
| 249 | if (!s) |
| 250 | s = oldbp; |
| 251 | else |
| 252 | PL_bufptr = s; |
| 253 | yywarn(Perl_form(aTHX_ "%s found where operator expected", what)); |
| 254 | if (is_first) |
| 255 | Perl_warn(aTHX_ "\t(Missing semicolon on previous line?)\n"); |
| 256 | else if (PL_oldoldbufptr && isIDFIRST_lazy_if(PL_oldoldbufptr,UTF)) { |
| 257 | char *t; |
| 258 | for (t = PL_oldoldbufptr; *t && (isALNUM_lazy_if(t,UTF) || *t == ':'); t++) ; |
| 259 | if (t < PL_bufptr && isSPACE(*t)) |
| 260 | Perl_warn(aTHX_ "\t(Do you need to predeclare %.*s?)\n", |
| 261 | t - PL_oldoldbufptr, PL_oldoldbufptr); |
| 262 | } |
| 263 | else { |
| 264 | assert(s >= oldbp); |
| 265 | Perl_warn(aTHX_ "\t(Missing operator before %.*s?)\n", s - oldbp, oldbp); |
| 266 | } |
| 267 | PL_bufptr = oldbp; |
| 268 | } |
| 269 | |
| 270 | /* |
| 271 | * S_missingterm |
| 272 | * Complain about missing quote/regexp/heredoc terminator. |
| 273 | * If it's called with (char *)NULL then it cauterizes the line buffer. |
| 274 | * If we're in a delimited string and the delimiter is a control |
| 275 | * character, it's reformatted into a two-char sequence like ^C. |
| 276 | * This is fatal. |
| 277 | */ |
| 278 | |
| 279 | STATIC void |
| 280 | S_missingterm(pTHX_ char *s) |
| 281 | { |
| 282 | char tmpbuf[3]; |
| 283 | char q; |
| 284 | if (s) { |
| 285 | char *nl = strrchr(s,'\n'); |
| 286 | if (nl) |
| 287 | *nl = '\0'; |
| 288 | } |
| 289 | else if ( |
| 290 | #ifdef EBCDIC |
| 291 | iscntrl(PL_multi_close) |
| 292 | #else |
| 293 | PL_multi_close < 32 || PL_multi_close == 127 |
| 294 | #endif |
| 295 | ) { |
| 296 | *tmpbuf = '^'; |
| 297 | tmpbuf[1] = toCTRL(PL_multi_close); |
| 298 | s = "\\n"; |
| 299 | tmpbuf[2] = '\0'; |
| 300 | s = tmpbuf; |
| 301 | } |
| 302 | else { |
| 303 | *tmpbuf = PL_multi_close; |
| 304 | tmpbuf[1] = '\0'; |
| 305 | s = tmpbuf; |
| 306 | } |
| 307 | q = strchr(s,'"') ? '\'' : '"'; |
| 308 | Perl_croak(aTHX_ "Can't find string terminator %c%s%c anywhere before EOF",q,s,q); |
| 309 | } |
| 310 | |
| 311 | /* |
| 312 | * Perl_deprecate |
| 313 | */ |
| 314 | |
| 315 | void |
| 316 | Perl_deprecate(pTHX_ char *s) |
| 317 | { |
| 318 | if (ckWARN(WARN_DEPRECATED)) |
| 319 | Perl_warner(aTHX_ WARN_DEPRECATED, "Use of %s is deprecated", s); |
| 320 | } |
| 321 | |
| 322 | /* |
| 323 | * depcom |
| 324 | * Deprecate a comma-less variable list. |
| 325 | */ |
| 326 | |
| 327 | STATIC void |
| 328 | S_depcom(pTHX) |
| 329 | { |
| 330 | deprecate("comma-less variable list"); |
| 331 | } |
| 332 | |
| 333 | /* |
| 334 | * experimental text filters for win32 carriage-returns, utf16-to-utf8 and |
| 335 | * utf16-to-utf8-reversed. |
| 336 | */ |
| 337 | |
| 338 | #ifdef PERL_CR_FILTER |
| 339 | static void |
| 340 | strip_return(SV *sv) |
| 341 | { |
| 342 | register char *s = SvPVX(sv); |
| 343 | register char *e = s + SvCUR(sv); |
| 344 | /* outer loop optimized to do nothing if there are no CR-LFs */ |
| 345 | while (s < e) { |
| 346 | if (*s++ == '\r' && *s == '\n') { |
| 347 | /* hit a CR-LF, need to copy the rest */ |
| 348 | register char *d = s - 1; |
| 349 | *d++ = *s++; |
| 350 | while (s < e) { |
| 351 | if (*s == '\r' && s[1] == '\n') |
| 352 | s++; |
| 353 | *d++ = *s++; |
| 354 | } |
| 355 | SvCUR(sv) -= s - d; |
| 356 | return; |
| 357 | } |
| 358 | } |
| 359 | } |
| 360 | |
| 361 | STATIC I32 |
| 362 | S_cr_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
| 363 | { |
| 364 | I32 count = FILTER_READ(idx+1, sv, maxlen); |
| 365 | if (count > 0 && !maxlen) |
| 366 | strip_return(sv); |
| 367 | return count; |
| 368 | } |
| 369 | #endif |
| 370 | |
| 371 | /* |
| 372 | * Perl_lex_start |
| 373 | * Initialize variables. Uses the Perl save_stack to save its state (for |
| 374 | * recursive calls to the parser). |
| 375 | */ |
| 376 | |
| 377 | void |
| 378 | Perl_lex_start(pTHX_ SV *line) |
| 379 | { |
| 380 | char *s; |
| 381 | STRLEN len; |
| 382 | |
| 383 | SAVEI32(PL_lex_dojoin); |
| 384 | SAVEI32(PL_lex_brackets); |
| 385 | SAVEI32(PL_lex_casemods); |
| 386 | SAVEI32(PL_lex_starts); |
| 387 | SAVEI32(PL_lex_state); |
| 388 | SAVEVPTR(PL_lex_inpat); |
| 389 | SAVEI32(PL_lex_inwhat); |
| 390 | if (PL_lex_state == LEX_KNOWNEXT) { |
| 391 | I32 toke = PL_nexttoke; |
| 392 | while (--toke >= 0) { |
| 393 | SAVEI32(PL_nexttype[toke]); |
| 394 | SAVEVPTR(PL_nextval[toke]); |
| 395 | } |
| 396 | SAVEI32(PL_nexttoke); |
| 397 | } |
| 398 | SAVECOPLINE(PL_curcop); |
| 399 | SAVEPPTR(PL_bufptr); |
| 400 | SAVEPPTR(PL_bufend); |
| 401 | SAVEPPTR(PL_oldbufptr); |
| 402 | SAVEPPTR(PL_oldoldbufptr); |
| 403 | SAVEPPTR(PL_last_lop); |
| 404 | SAVEPPTR(PL_last_uni); |
| 405 | SAVEPPTR(PL_linestart); |
| 406 | SAVESPTR(PL_linestr); |
| 407 | SAVEPPTR(PL_lex_brackstack); |
| 408 | SAVEPPTR(PL_lex_casestack); |
| 409 | SAVEDESTRUCTOR_X(restore_rsfp, PL_rsfp); |
| 410 | SAVESPTR(PL_lex_stuff); |
| 411 | SAVEI32(PL_lex_defer); |
| 412 | SAVEI32(PL_sublex_info.sub_inwhat); |
| 413 | SAVESPTR(PL_lex_repl); |
| 414 | SAVEINT(PL_expect); |
| 415 | SAVEINT(PL_lex_expect); |
| 416 | |
| 417 | PL_lex_state = LEX_NORMAL; |
| 418 | PL_lex_defer = 0; |
| 419 | PL_expect = XSTATE; |
| 420 | PL_lex_brackets = 0; |
| 421 | New(899, PL_lex_brackstack, 120, char); |
| 422 | New(899, PL_lex_casestack, 12, char); |
| 423 | SAVEFREEPV(PL_lex_brackstack); |
| 424 | SAVEFREEPV(PL_lex_casestack); |
| 425 | PL_lex_casemods = 0; |
| 426 | *PL_lex_casestack = '\0'; |
| 427 | PL_lex_dojoin = 0; |
| 428 | PL_lex_starts = 0; |
| 429 | PL_lex_stuff = Nullsv; |
| 430 | PL_lex_repl = Nullsv; |
| 431 | PL_lex_inpat = 0; |
| 432 | PL_nexttoke = 0; |
| 433 | PL_lex_inwhat = 0; |
| 434 | PL_sublex_info.sub_inwhat = 0; |
| 435 | PL_linestr = line; |
| 436 | if (SvREADONLY(PL_linestr)) |
| 437 | PL_linestr = sv_2mortal(newSVsv(PL_linestr)); |
| 438 | s = SvPV(PL_linestr, len); |
| 439 | if (len && s[len-1] != ';') { |
| 440 | if (!(SvFLAGS(PL_linestr) & SVs_TEMP)) |
| 441 | PL_linestr = sv_2mortal(newSVsv(PL_linestr)); |
| 442 | sv_catpvn(PL_linestr, "\n;", 2); |
| 443 | } |
| 444 | SvTEMP_off(PL_linestr); |
| 445 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr); |
| 446 | PL_bufend = PL_bufptr + SvCUR(PL_linestr); |
| 447 | PL_last_lop = PL_last_uni = Nullch; |
| 448 | PL_rsfp = 0; |
| 449 | } |
| 450 | |
| 451 | /* |
| 452 | * Perl_lex_end |
| 453 | * Finalizer for lexing operations. Must be called when the parser is |
| 454 | * done with the lexer. |
| 455 | */ |
| 456 | |
| 457 | void |
| 458 | Perl_lex_end(pTHX) |
| 459 | { |
| 460 | PL_doextract = FALSE; |
| 461 | } |
| 462 | |
| 463 | /* |
| 464 | * S_incline |
| 465 | * This subroutine has nothing to do with tilting, whether at windmills |
| 466 | * or pinball tables. Its name is short for "increment line". It |
| 467 | * increments the current line number in CopLINE(PL_curcop) and checks |
| 468 | * to see whether the line starts with a comment of the form |
| 469 | * # line 500 "foo.pm" |
| 470 | * If so, it sets the current line number and file to the values in the comment. |
| 471 | */ |
| 472 | |
| 473 | STATIC void |
| 474 | S_incline(pTHX_ char *s) |
| 475 | { |
| 476 | char *t; |
| 477 | char *n; |
| 478 | char *e; |
| 479 | char ch; |
| 480 | |
| 481 | CopLINE_inc(PL_curcop); |
| 482 | if (*s++ != '#') |
| 483 | return; |
| 484 | while (SPACE_OR_TAB(*s)) s++; |
| 485 | if (strnEQ(s, "line", 4)) |
| 486 | s += 4; |
| 487 | else |
| 488 | return; |
| 489 | if (SPACE_OR_TAB(*s)) |
| 490 | s++; |
| 491 | else |
| 492 | return; |
| 493 | while (SPACE_OR_TAB(*s)) s++; |
| 494 | if (!isDIGIT(*s)) |
| 495 | return; |
| 496 | n = s; |
| 497 | while (isDIGIT(*s)) |
| 498 | s++; |
| 499 | while (SPACE_OR_TAB(*s)) |
| 500 | s++; |
| 501 | if (*s == '"' && (t = strchr(s+1, '"'))) { |
| 502 | s++; |
| 503 | e = t + 1; |
| 504 | } |
| 505 | else { |
| 506 | for (t = s; !isSPACE(*t); t++) ; |
| 507 | e = t; |
| 508 | } |
| 509 | while (SPACE_OR_TAB(*e) || *e == '\r' || *e == '\f') |
| 510 | e++; |
| 511 | if (*e != '\n' && *e != '\0') |
| 512 | return; /* false alarm */ |
| 513 | |
| 514 | ch = *t; |
| 515 | *t = '\0'; |
| 516 | if (t - s > 0) { |
| 517 | #ifdef USE_ITHREADS |
| 518 | Safefree(CopFILE(PL_curcop)); |
| 519 | #else |
| 520 | SvREFCNT_dec(CopFILEGV(PL_curcop)); |
| 521 | #endif |
| 522 | CopFILE_set(PL_curcop, s); |
| 523 | } |
| 524 | *t = ch; |
| 525 | CopLINE_set(PL_curcop, atoi(n)-1); |
| 526 | } |
| 527 | |
| 528 | /* |
| 529 | * S_skipspace |
| 530 | * Called to gobble the appropriate amount and type of whitespace. |
| 531 | * Skips comments as well. |
| 532 | */ |
| 533 | |
| 534 | STATIC char * |
| 535 | S_skipspace(pTHX_ register char *s) |
| 536 | { |
| 537 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
| 538 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
| 539 | s++; |
| 540 | return s; |
| 541 | } |
| 542 | for (;;) { |
| 543 | STRLEN prevlen; |
| 544 | SSize_t oldprevlen, oldoldprevlen; |
| 545 | SSize_t oldloplen = 0, oldunilen = 0; |
| 546 | while (s < PL_bufend && isSPACE(*s)) { |
| 547 | if (*s++ == '\n' && PL_in_eval && !PL_rsfp) |
| 548 | incline(s); |
| 549 | } |
| 550 | |
| 551 | /* comment */ |
| 552 | if (s < PL_bufend && *s == '#') { |
| 553 | while (s < PL_bufend && *s != '\n') |
| 554 | s++; |
| 555 | if (s < PL_bufend) { |
| 556 | s++; |
| 557 | if (PL_in_eval && !PL_rsfp) { |
| 558 | incline(s); |
| 559 | continue; |
| 560 | } |
| 561 | } |
| 562 | } |
| 563 | |
| 564 | /* only continue to recharge the buffer if we're at the end |
| 565 | * of the buffer, we're not reading from a source filter, and |
| 566 | * we're in normal lexing mode |
| 567 | */ |
| 568 | if (s < PL_bufend || !PL_rsfp || PL_sublex_info.sub_inwhat || |
| 569 | PL_lex_state == LEX_FORMLINE) |
| 570 | return s; |
| 571 | |
| 572 | /* try to recharge the buffer */ |
| 573 | if ((s = filter_gets(PL_linestr, PL_rsfp, |
| 574 | (prevlen = SvCUR(PL_linestr)))) == Nullch) |
| 575 | { |
| 576 | /* end of file. Add on the -p or -n magic */ |
| 577 | if (PL_minus_n || PL_minus_p) { |
| 578 | sv_setpv(PL_linestr,PL_minus_p ? |
| 579 | ";}continue{print or die qq(-p destination: $!\\n)" : |
| 580 | ""); |
| 581 | sv_catpv(PL_linestr,";}"); |
| 582 | PL_minus_n = PL_minus_p = 0; |
| 583 | } |
| 584 | else |
| 585 | sv_setpv(PL_linestr,";"); |
| 586 | |
| 587 | /* reset variables for next time we lex */ |
| 588 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart |
| 589 | = SvPVX(PL_linestr); |
| 590 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 591 | PL_last_lop = PL_last_uni = Nullch; |
| 592 | |
| 593 | /* Close the filehandle. Could be from -P preprocessor, |
| 594 | * STDIN, or a regular file. If we were reading code from |
| 595 | * STDIN (because the commandline held no -e or filename) |
| 596 | * then we don't close it, we reset it so the code can |
| 597 | * read from STDIN too. |
| 598 | */ |
| 599 | |
| 600 | if (PL_preprocess && !PL_in_eval) |
| 601 | (void)PerlProc_pclose(PL_rsfp); |
| 602 | else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
| 603 | PerlIO_clearerr(PL_rsfp); |
| 604 | else |
| 605 | (void)PerlIO_close(PL_rsfp); |
| 606 | PL_rsfp = Nullfp; |
| 607 | return s; |
| 608 | } |
| 609 | |
| 610 | /* not at end of file, so we only read another line */ |
| 611 | /* make corresponding updates to old pointers, for yyerror() */ |
| 612 | oldprevlen = PL_oldbufptr - PL_bufend; |
| 613 | oldoldprevlen = PL_oldoldbufptr - PL_bufend; |
| 614 | if (PL_last_uni) |
| 615 | oldunilen = PL_last_uni - PL_bufend; |
| 616 | if (PL_last_lop) |
| 617 | oldloplen = PL_last_lop - PL_bufend; |
| 618 | PL_linestart = PL_bufptr = s + prevlen; |
| 619 | PL_bufend = s + SvCUR(PL_linestr); |
| 620 | s = PL_bufptr; |
| 621 | PL_oldbufptr = s + oldprevlen; |
| 622 | PL_oldoldbufptr = s + oldoldprevlen; |
| 623 | if (PL_last_uni) |
| 624 | PL_last_uni = s + oldunilen; |
| 625 | if (PL_last_lop) |
| 626 | PL_last_lop = s + oldloplen; |
| 627 | incline(s); |
| 628 | |
| 629 | /* debugger active and we're not compiling the debugger code, |
| 630 | * so store the line into the debugger's array of lines |
| 631 | */ |
| 632 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| 633 | SV *sv = NEWSV(85,0); |
| 634 | |
| 635 | sv_upgrade(sv, SVt_PVMG); |
| 636 | sv_setpvn(sv,PL_bufptr,PL_bufend-PL_bufptr); |
| 637 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
| 638 | } |
| 639 | } |
| 640 | } |
| 641 | |
| 642 | /* |
| 643 | * S_check_uni |
| 644 | * Check the unary operators to ensure there's no ambiguity in how they're |
| 645 | * used. An ambiguous piece of code would be: |
| 646 | * rand + 5 |
| 647 | * This doesn't mean rand() + 5. Because rand() is a unary operator, |
| 648 | * the +5 is its argument. |
| 649 | */ |
| 650 | |
| 651 | STATIC void |
| 652 | S_check_uni(pTHX) |
| 653 | { |
| 654 | char *s; |
| 655 | char *t; |
| 656 | |
| 657 | if (PL_oldoldbufptr != PL_last_uni) |
| 658 | return; |
| 659 | while (isSPACE(*PL_last_uni)) |
| 660 | PL_last_uni++; |
| 661 | for (s = PL_last_uni; isALNUM_lazy_if(s,UTF) || *s == '-'; s++) ; |
| 662 | if ((t = strchr(s, '(')) && t < PL_bufptr) |
| 663 | return; |
| 664 | if (ckWARN_d(WARN_AMBIGUOUS)){ |
| 665 | char ch = *s; |
| 666 | *s = '\0'; |
| 667 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 668 | "Warning: Use of \"%s\" without parens is ambiguous", |
| 669 | PL_last_uni); |
| 670 | *s = ch; |
| 671 | } |
| 672 | } |
| 673 | |
| 674 | /* workaround to replace the UNI() macro with a function. Only the |
| 675 | * hints/uts.sh file mentions this. Other comments elsewhere in the |
| 676 | * source indicate Microport Unix might need it too. |
| 677 | */ |
| 678 | |
| 679 | #ifdef CRIPPLED_CC |
| 680 | |
| 681 | #undef UNI |
| 682 | #define UNI(f) return uni(f,s) |
| 683 | |
| 684 | STATIC int |
| 685 | S_uni(pTHX_ I32 f, char *s) |
| 686 | { |
| 687 | yylval.ival = f; |
| 688 | PL_expect = XTERM; |
| 689 | PL_bufptr = s; |
| 690 | PL_last_uni = PL_oldbufptr; |
| 691 | PL_last_lop_op = f; |
| 692 | if (*s == '(') |
| 693 | return FUNC1; |
| 694 | s = skipspace(s); |
| 695 | if (*s == '(') |
| 696 | return FUNC1; |
| 697 | else |
| 698 | return UNIOP; |
| 699 | } |
| 700 | |
| 701 | #endif /* CRIPPLED_CC */ |
| 702 | |
| 703 | /* |
| 704 | * LOP : macro to build a list operator. Its behaviour has been replaced |
| 705 | * with a subroutine, S_lop() for which LOP is just another name. |
| 706 | */ |
| 707 | |
| 708 | #define LOP(f,x) return lop(f,x,s) |
| 709 | |
| 710 | /* |
| 711 | * S_lop |
| 712 | * Build a list operator (or something that might be one). The rules: |
| 713 | * - if we have a next token, then it's a list operator [why?] |
| 714 | * - if the next thing is an opening paren, then it's a function |
| 715 | * - else it's a list operator |
| 716 | */ |
| 717 | |
| 718 | STATIC I32 |
| 719 | S_lop(pTHX_ I32 f, int x, char *s) |
| 720 | { |
| 721 | yylval.ival = f; |
| 722 | CLINE; |
| 723 | REPORT("lop", f) |
| 724 | PL_expect = x; |
| 725 | PL_bufptr = s; |
| 726 | PL_last_lop = PL_oldbufptr; |
| 727 | PL_last_lop_op = f; |
| 728 | if (PL_nexttoke) |
| 729 | return LSTOP; |
| 730 | if (*s == '(') |
| 731 | return FUNC; |
| 732 | s = skipspace(s); |
| 733 | if (*s == '(') |
| 734 | return FUNC; |
| 735 | else |
| 736 | return LSTOP; |
| 737 | } |
| 738 | |
| 739 | /* |
| 740 | * S_force_next |
| 741 | * When the lexer realizes it knows the next token (for instance, |
| 742 | * it is reordering tokens for the parser) then it can call S_force_next |
| 743 | * to know what token to return the next time the lexer is called. Caller |
| 744 | * will need to set PL_nextval[], and possibly PL_expect to ensure the lexer |
| 745 | * handles the token correctly. |
| 746 | */ |
| 747 | |
| 748 | STATIC void |
| 749 | S_force_next(pTHX_ I32 type) |
| 750 | { |
| 751 | PL_nexttype[PL_nexttoke] = type; |
| 752 | PL_nexttoke++; |
| 753 | if (PL_lex_state != LEX_KNOWNEXT) { |
| 754 | PL_lex_defer = PL_lex_state; |
| 755 | PL_lex_expect = PL_expect; |
| 756 | PL_lex_state = LEX_KNOWNEXT; |
| 757 | } |
| 758 | } |
| 759 | |
| 760 | /* |
| 761 | * S_force_word |
| 762 | * When the lexer knows the next thing is a word (for instance, it has |
| 763 | * just seen -> and it knows that the next char is a word char, then |
| 764 | * it calls S_force_word to stick the next word into the PL_next lookahead. |
| 765 | * |
| 766 | * Arguments: |
| 767 | * char *start : buffer position (must be within PL_linestr) |
| 768 | * int token : PL_next will be this type of bare word (e.g., METHOD,WORD) |
| 769 | * int check_keyword : if true, Perl checks to make sure the word isn't |
| 770 | * a keyword (do this if the word is a label, e.g. goto FOO) |
| 771 | * int allow_pack : if true, : characters will also be allowed (require, |
| 772 | * use, etc. do this) |
| 773 | * int allow_initial_tick : used by the "sub" lexer only. |
| 774 | */ |
| 775 | |
| 776 | STATIC char * |
| 777 | S_force_word(pTHX_ register char *start, int token, int check_keyword, int allow_pack, int allow_initial_tick) |
| 778 | { |
| 779 | register char *s; |
| 780 | STRLEN len; |
| 781 | |
| 782 | start = skipspace(start); |
| 783 | s = start; |
| 784 | if (isIDFIRST_lazy_if(s,UTF) || |
| 785 | (allow_pack && *s == ':') || |
| 786 | (allow_initial_tick && *s == '\'') ) |
| 787 | { |
| 788 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, allow_pack, &len); |
| 789 | if (check_keyword && keyword(PL_tokenbuf, len)) |
| 790 | return start; |
| 791 | if (token == METHOD) { |
| 792 | s = skipspace(s); |
| 793 | if (*s == '(') |
| 794 | PL_expect = XTERM; |
| 795 | else { |
| 796 | PL_expect = XOPERATOR; |
| 797 | } |
| 798 | } |
| 799 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST,0, newSVpv(PL_tokenbuf,0)); |
| 800 | PL_nextval[PL_nexttoke].opval->op_private |= OPpCONST_BARE; |
| 801 | force_next(token); |
| 802 | } |
| 803 | return s; |
| 804 | } |
| 805 | |
| 806 | /* |
| 807 | * S_force_ident |
| 808 | * Called when the lexer wants $foo *foo &foo etc, but the program |
| 809 | * text only contains the "foo" portion. The first argument is a pointer |
| 810 | * to the "foo", and the second argument is the type symbol to prefix. |
| 811 | * Forces the next token to be a "WORD". |
| 812 | * Creates the symbol if it didn't already exist (via gv_fetchpv()). |
| 813 | */ |
| 814 | |
| 815 | STATIC void |
| 816 | S_force_ident(pTHX_ register char *s, int kind) |
| 817 | { |
| 818 | if (s && *s) { |
| 819 | OP* o = (OP*)newSVOP(OP_CONST, 0, newSVpv(s,0)); |
| 820 | PL_nextval[PL_nexttoke].opval = o; |
| 821 | force_next(WORD); |
| 822 | if (kind) { |
| 823 | o->op_private = OPpCONST_ENTERED; |
| 824 | /* XXX see note in pp_entereval() for why we forgo typo |
| 825 | warnings if the symbol must be introduced in an eval. |
| 826 | GSAR 96-10-12 */ |
| 827 | gv_fetchpv(s, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE, |
| 828 | kind == '$' ? SVt_PV : |
| 829 | kind == '@' ? SVt_PVAV : |
| 830 | kind == '%' ? SVt_PVHV : |
| 831 | SVt_PVGV |
| 832 | ); |
| 833 | } |
| 834 | } |
| 835 | } |
| 836 | |
| 837 | NV |
| 838 | Perl_str_to_version(pTHX_ SV *sv) |
| 839 | { |
| 840 | NV retval = 0.0; |
| 841 | NV nshift = 1.0; |
| 842 | STRLEN len; |
| 843 | char *start = SvPVx(sv,len); |
| 844 | bool utf = SvUTF8(sv) ? TRUE : FALSE; |
| 845 | char *end = start + len; |
| 846 | while (start < end) { |
| 847 | STRLEN skip; |
| 848 | UV n; |
| 849 | if (utf) |
| 850 | n = utf8n_to_uvchr((U8*)start, len, &skip, 0); |
| 851 | else { |
| 852 | n = *(U8*)start; |
| 853 | skip = 1; |
| 854 | } |
| 855 | retval += ((NV)n)/nshift; |
| 856 | start += skip; |
| 857 | nshift *= 1000; |
| 858 | } |
| 859 | return retval; |
| 860 | } |
| 861 | |
| 862 | /* |
| 863 | * S_force_version |
| 864 | * Forces the next token to be a version number. |
| 865 | * If the next token appears to be an invalid version number, (e.g. "v2b"), |
| 866 | * and if "guessing" is TRUE, then no new token is created (and the caller |
| 867 | * must use an alternative parsing method). |
| 868 | */ |
| 869 | |
| 870 | STATIC char * |
| 871 | S_force_version(pTHX_ char *s, int guessing) |
| 872 | { |
| 873 | OP *version = Nullop; |
| 874 | char *d; |
| 875 | |
| 876 | s = skipspace(s); |
| 877 | |
| 878 | d = s; |
| 879 | if (*d == 'v') |
| 880 | d++; |
| 881 | if (isDIGIT(*d)) { |
| 882 | while (isDIGIT(*d) || *d == '_' || *d == '.') |
| 883 | d++; |
| 884 | if (*d == ';' || isSPACE(*d) || *d == '}' || !*d) { |
| 885 | SV *ver; |
| 886 | s = scan_num(s, &yylval); |
| 887 | version = yylval.opval; |
| 888 | ver = cSVOPx(version)->op_sv; |
| 889 | if (SvPOK(ver) && !SvNIOK(ver)) { |
| 890 | (void)SvUPGRADE(ver, SVt_PVNV); |
| 891 | SvNVX(ver) = str_to_version(ver); |
| 892 | SvNOK_on(ver); /* hint that it is a version */ |
| 893 | } |
| 894 | } |
| 895 | else if (guessing) |
| 896 | return s; |
| 897 | } |
| 898 | |
| 899 | /* NOTE: The parser sees the package name and the VERSION swapped */ |
| 900 | PL_nextval[PL_nexttoke].opval = version; |
| 901 | force_next(WORD); |
| 902 | |
| 903 | return s; |
| 904 | } |
| 905 | |
| 906 | /* |
| 907 | * S_tokeq |
| 908 | * Tokenize a quoted string passed in as an SV. It finds the next |
| 909 | * chunk, up to end of string or a backslash. It may make a new |
| 910 | * SV containing that chunk (if HINT_NEW_STRING is on). It also |
| 911 | * turns \\ into \. |
| 912 | */ |
| 913 | |
| 914 | STATIC SV * |
| 915 | S_tokeq(pTHX_ SV *sv) |
| 916 | { |
| 917 | register char *s; |
| 918 | register char *send; |
| 919 | register char *d; |
| 920 | STRLEN len = 0; |
| 921 | SV *pv = sv; |
| 922 | |
| 923 | if (!SvLEN(sv)) |
| 924 | goto finish; |
| 925 | |
| 926 | s = SvPV_force(sv, len); |
| 927 | if (SvTYPE(sv) >= SVt_PVIV && SvIVX(sv) == -1) |
| 928 | goto finish; |
| 929 | send = s + len; |
| 930 | while (s < send && *s != '\\') |
| 931 | s++; |
| 932 | if (s == send) |
| 933 | goto finish; |
| 934 | d = s; |
| 935 | if ( PL_hints & HINT_NEW_STRING ) { |
| 936 | pv = sv_2mortal(newSVpvn(SvPVX(pv), len)); |
| 937 | if (SvUTF8(sv)) |
| 938 | SvUTF8_on(pv); |
| 939 | } |
| 940 | while (s < send) { |
| 941 | if (*s == '\\') { |
| 942 | if (s + 1 < send && (s[1] == '\\')) |
| 943 | s++; /* all that, just for this */ |
| 944 | } |
| 945 | *d++ = *s++; |
| 946 | } |
| 947 | *d = '\0'; |
| 948 | SvCUR_set(sv, d - SvPVX(sv)); |
| 949 | finish: |
| 950 | if ( PL_hints & HINT_NEW_STRING ) |
| 951 | return new_constant(NULL, 0, "q", sv, pv, "q"); |
| 952 | return sv; |
| 953 | } |
| 954 | |
| 955 | /* |
| 956 | * Now come three functions related to double-quote context, |
| 957 | * S_sublex_start, S_sublex_push, and S_sublex_done. They're used when |
| 958 | * converting things like "\u\Lgnat" into ucfirst(lc("gnat")). They |
| 959 | * interact with PL_lex_state, and create fake ( ... ) argument lists |
| 960 | * to handle functions and concatenation. |
| 961 | * They assume that whoever calls them will be setting up a fake |
| 962 | * join call, because each subthing puts a ',' after it. This lets |
| 963 | * "lower \luPpEr" |
| 964 | * become |
| 965 | * join($, , 'lower ', lcfirst( 'uPpEr', ) ,) |
| 966 | * |
| 967 | * (I'm not sure whether the spurious commas at the end of lcfirst's |
| 968 | * arguments and join's arguments are created or not). |
| 969 | */ |
| 970 | |
| 971 | /* |
| 972 | * S_sublex_start |
| 973 | * Assumes that yylval.ival is the op we're creating (e.g. OP_LCFIRST). |
| 974 | * |
| 975 | * Pattern matching will set PL_lex_op to the pattern-matching op to |
| 976 | * make (we return THING if yylval.ival is OP_NULL, PMFUNC otherwise). |
| 977 | * |
| 978 | * OP_CONST and OP_READLINE are easy--just make the new op and return. |
| 979 | * |
| 980 | * Everything else becomes a FUNC. |
| 981 | * |
| 982 | * Sets PL_lex_state to LEX_INTERPPUSH unless (ival was OP_NULL or we |
| 983 | * had an OP_CONST or OP_READLINE). This just sets us up for a |
| 984 | * call to S_sublex_push(). |
| 985 | */ |
| 986 | |
| 987 | STATIC I32 |
| 988 | S_sublex_start(pTHX) |
| 989 | { |
| 990 | register I32 op_type = yylval.ival; |
| 991 | |
| 992 | if (op_type == OP_NULL) { |
| 993 | yylval.opval = PL_lex_op; |
| 994 | PL_lex_op = Nullop; |
| 995 | return THING; |
| 996 | } |
| 997 | if (op_type == OP_CONST || op_type == OP_READLINE) { |
| 998 | SV *sv = tokeq(PL_lex_stuff); |
| 999 | |
| 1000 | if (SvTYPE(sv) == SVt_PVIV) { |
| 1001 | /* Overloaded constants, nothing fancy: Convert to SVt_PV: */ |
| 1002 | STRLEN len; |
| 1003 | char *p; |
| 1004 | SV *nsv; |
| 1005 | |
| 1006 | p = SvPV(sv, len); |
| 1007 | nsv = newSVpvn(p, len); |
| 1008 | if (SvUTF8(sv)) |
| 1009 | SvUTF8_on(nsv); |
| 1010 | SvREFCNT_dec(sv); |
| 1011 | sv = nsv; |
| 1012 | } |
| 1013 | yylval.opval = (OP*)newSVOP(op_type, 0, sv); |
| 1014 | PL_lex_stuff = Nullsv; |
| 1015 | return THING; |
| 1016 | } |
| 1017 | |
| 1018 | PL_sublex_info.super_state = PL_lex_state; |
| 1019 | PL_sublex_info.sub_inwhat = op_type; |
| 1020 | PL_sublex_info.sub_op = PL_lex_op; |
| 1021 | PL_lex_state = LEX_INTERPPUSH; |
| 1022 | |
| 1023 | PL_expect = XTERM; |
| 1024 | if (PL_lex_op) { |
| 1025 | yylval.opval = PL_lex_op; |
| 1026 | PL_lex_op = Nullop; |
| 1027 | return PMFUNC; |
| 1028 | } |
| 1029 | else |
| 1030 | return FUNC; |
| 1031 | } |
| 1032 | |
| 1033 | /* |
| 1034 | * S_sublex_push |
| 1035 | * Create a new scope to save the lexing state. The scope will be |
| 1036 | * ended in S_sublex_done. Returns a '(', starting the function arguments |
| 1037 | * to the uc, lc, etc. found before. |
| 1038 | * Sets PL_lex_state to LEX_INTERPCONCAT. |
| 1039 | */ |
| 1040 | |
| 1041 | STATIC I32 |
| 1042 | S_sublex_push(pTHX) |
| 1043 | { |
| 1044 | ENTER; |
| 1045 | |
| 1046 | PL_lex_state = PL_sublex_info.super_state; |
| 1047 | SAVEI32(PL_lex_dojoin); |
| 1048 | SAVEI32(PL_lex_brackets); |
| 1049 | SAVEI32(PL_lex_casemods); |
| 1050 | SAVEI32(PL_lex_starts); |
| 1051 | SAVEI32(PL_lex_state); |
| 1052 | SAVEVPTR(PL_lex_inpat); |
| 1053 | SAVEI32(PL_lex_inwhat); |
| 1054 | SAVECOPLINE(PL_curcop); |
| 1055 | SAVEPPTR(PL_bufptr); |
| 1056 | SAVEPPTR(PL_bufend); |
| 1057 | SAVEPPTR(PL_oldbufptr); |
| 1058 | SAVEPPTR(PL_oldoldbufptr); |
| 1059 | SAVEPPTR(PL_last_lop); |
| 1060 | SAVEPPTR(PL_last_uni); |
| 1061 | SAVEPPTR(PL_linestart); |
| 1062 | SAVESPTR(PL_linestr); |
| 1063 | SAVEPPTR(PL_lex_brackstack); |
| 1064 | SAVEPPTR(PL_lex_casestack); |
| 1065 | |
| 1066 | PL_linestr = PL_lex_stuff; |
| 1067 | PL_lex_stuff = Nullsv; |
| 1068 | |
| 1069 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart |
| 1070 | = SvPVX(PL_linestr); |
| 1071 | PL_bufend += SvCUR(PL_linestr); |
| 1072 | PL_last_lop = PL_last_uni = Nullch; |
| 1073 | SAVEFREESV(PL_linestr); |
| 1074 | |
| 1075 | PL_lex_dojoin = FALSE; |
| 1076 | PL_lex_brackets = 0; |
| 1077 | New(899, PL_lex_brackstack, 120, char); |
| 1078 | New(899, PL_lex_casestack, 12, char); |
| 1079 | SAVEFREEPV(PL_lex_brackstack); |
| 1080 | SAVEFREEPV(PL_lex_casestack); |
| 1081 | PL_lex_casemods = 0; |
| 1082 | *PL_lex_casestack = '\0'; |
| 1083 | PL_lex_starts = 0; |
| 1084 | PL_lex_state = LEX_INTERPCONCAT; |
| 1085 | CopLINE_set(PL_curcop, PL_multi_start); |
| 1086 | |
| 1087 | PL_lex_inwhat = PL_sublex_info.sub_inwhat; |
| 1088 | if (PL_lex_inwhat == OP_MATCH || PL_lex_inwhat == OP_QR || PL_lex_inwhat == OP_SUBST) |
| 1089 | PL_lex_inpat = PL_sublex_info.sub_op; |
| 1090 | else |
| 1091 | PL_lex_inpat = Nullop; |
| 1092 | |
| 1093 | return '('; |
| 1094 | } |
| 1095 | |
| 1096 | /* |
| 1097 | * S_sublex_done |
| 1098 | * Restores lexer state after a S_sublex_push. |
| 1099 | */ |
| 1100 | |
| 1101 | STATIC I32 |
| 1102 | S_sublex_done(pTHX) |
| 1103 | { |
| 1104 | if (!PL_lex_starts++) { |
| 1105 | SV *sv = newSVpvn("",0); |
| 1106 | if (SvUTF8(PL_linestr)) |
| 1107 | SvUTF8_on(sv); |
| 1108 | PL_expect = XOPERATOR; |
| 1109 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
| 1110 | return THING; |
| 1111 | } |
| 1112 | |
| 1113 | if (PL_lex_casemods) { /* oops, we've got some unbalanced parens */ |
| 1114 | PL_lex_state = LEX_INTERPCASEMOD; |
| 1115 | return yylex(); |
| 1116 | } |
| 1117 | |
| 1118 | /* Is there a right-hand side to take care of? (s//RHS/ or tr//RHS/) */ |
| 1119 | if (PL_lex_repl && (PL_lex_inwhat == OP_SUBST || PL_lex_inwhat == OP_TRANS)) { |
| 1120 | PL_linestr = PL_lex_repl; |
| 1121 | PL_lex_inpat = 0; |
| 1122 | PL_bufend = PL_bufptr = PL_oldbufptr = PL_oldoldbufptr = PL_linestart = SvPVX(PL_linestr); |
| 1123 | PL_bufend += SvCUR(PL_linestr); |
| 1124 | PL_last_lop = PL_last_uni = Nullch; |
| 1125 | SAVEFREESV(PL_linestr); |
| 1126 | PL_lex_dojoin = FALSE; |
| 1127 | PL_lex_brackets = 0; |
| 1128 | PL_lex_casemods = 0; |
| 1129 | *PL_lex_casestack = '\0'; |
| 1130 | PL_lex_starts = 0; |
| 1131 | if (SvEVALED(PL_lex_repl)) { |
| 1132 | PL_lex_state = LEX_INTERPNORMAL; |
| 1133 | PL_lex_starts++; |
| 1134 | /* we don't clear PL_lex_repl here, so that we can check later |
| 1135 | whether this is an evalled subst; that means we rely on the |
| 1136 | logic to ensure sublex_done() is called again only via the |
| 1137 | branch (in yylex()) that clears PL_lex_repl, else we'll loop */ |
| 1138 | } |
| 1139 | else { |
| 1140 | PL_lex_state = LEX_INTERPCONCAT; |
| 1141 | PL_lex_repl = Nullsv; |
| 1142 | } |
| 1143 | return ','; |
| 1144 | } |
| 1145 | else { |
| 1146 | LEAVE; |
| 1147 | PL_bufend = SvPVX(PL_linestr); |
| 1148 | PL_bufend += SvCUR(PL_linestr); |
| 1149 | PL_expect = XOPERATOR; |
| 1150 | PL_sublex_info.sub_inwhat = 0; |
| 1151 | return ')'; |
| 1152 | } |
| 1153 | } |
| 1154 | |
| 1155 | /* |
| 1156 | scan_const |
| 1157 | |
| 1158 | Extracts a pattern, double-quoted string, or transliteration. This |
| 1159 | is terrifying code. |
| 1160 | |
| 1161 | It looks at lex_inwhat and PL_lex_inpat to find out whether it's |
| 1162 | processing a pattern (PL_lex_inpat is true), a transliteration |
| 1163 | (lex_inwhat & OP_TRANS is true), or a double-quoted string. |
| 1164 | |
| 1165 | Returns a pointer to the character scanned up to. Iff this is |
| 1166 | advanced from the start pointer supplied (ie if anything was |
| 1167 | successfully parsed), will leave an OP for the substring scanned |
| 1168 | in yylval. Caller must intuit reason for not parsing further |
| 1169 | by looking at the next characters herself. |
| 1170 | |
| 1171 | In patterns: |
| 1172 | backslashes: |
| 1173 | double-quoted style: \r and \n |
| 1174 | regexp special ones: \D \s |
| 1175 | constants: \x3 |
| 1176 | backrefs: \1 (deprecated in substitution replacements) |
| 1177 | case and quoting: \U \Q \E |
| 1178 | stops on @ and $, but not for $ as tail anchor |
| 1179 | |
| 1180 | In transliterations: |
| 1181 | characters are VERY literal, except for - not at the start or end |
| 1182 | of the string, which indicates a range. scan_const expands the |
| 1183 | range to the full set of intermediate characters. |
| 1184 | |
| 1185 | In double-quoted strings: |
| 1186 | backslashes: |
| 1187 | double-quoted style: \r and \n |
| 1188 | constants: \x3 |
| 1189 | backrefs: \1 (deprecated) |
| 1190 | case and quoting: \U \Q \E |
| 1191 | stops on @ and $ |
| 1192 | |
| 1193 | scan_const does *not* construct ops to handle interpolated strings. |
| 1194 | It stops processing as soon as it finds an embedded $ or @ variable |
| 1195 | and leaves it to the caller to work out what's going on. |
| 1196 | |
| 1197 | @ in pattern could be: @foo, @{foo}, @$foo, @'foo, @:foo. |
| 1198 | |
| 1199 | $ in pattern could be $foo or could be tail anchor. Assumption: |
| 1200 | it's a tail anchor if $ is the last thing in the string, or if it's |
| 1201 | followed by one of ")| \n\t" |
| 1202 | |
| 1203 | \1 (backreferences) are turned into $1 |
| 1204 | |
| 1205 | The structure of the code is |
| 1206 | while (there's a character to process) { |
| 1207 | handle transliteration ranges |
| 1208 | skip regexp comments |
| 1209 | skip # initiated comments in //x patterns |
| 1210 | check for embedded @foo |
| 1211 | check for embedded scalars |
| 1212 | if (backslash) { |
| 1213 | leave intact backslashes from leave (below) |
| 1214 | deprecate \1 in strings and sub replacements |
| 1215 | handle string-changing backslashes \l \U \Q \E, etc. |
| 1216 | switch (what was escaped) { |
| 1217 | handle - in a transliteration (becomes a literal -) |
| 1218 | handle \132 octal characters |
| 1219 | handle 0x15 hex characters |
| 1220 | handle \cV (control V) |
| 1221 | handle printf backslashes (\f, \r, \n, etc) |
| 1222 | } (end switch) |
| 1223 | } (end if backslash) |
| 1224 | } (end while character to read) |
| 1225 | |
| 1226 | */ |
| 1227 | |
| 1228 | STATIC char * |
| 1229 | S_scan_const(pTHX_ char *start) |
| 1230 | { |
| 1231 | register char *send = PL_bufend; /* end of the constant */ |
| 1232 | SV *sv = NEWSV(93, send - start); /* sv for the constant */ |
| 1233 | register char *s = start; /* start of the constant */ |
| 1234 | register char *d = SvPVX(sv); /* destination for copies */ |
| 1235 | bool dorange = FALSE; /* are we in a translit range? */ |
| 1236 | bool didrange = FALSE; /* did we just finish a range? */ |
| 1237 | I32 has_utf8 = FALSE; /* Output constant is UTF8 */ |
| 1238 | I32 this_utf8 = UTF; /* The source string is assumed to be UTF8 */ |
| 1239 | UV uv; |
| 1240 | |
| 1241 | const char *leaveit = /* set of acceptably-backslashed characters */ |
| 1242 | PL_lex_inpat |
| 1243 | ? "\\.^$@AGZdDwWsSbBpPXC+*?|()-nrtfeaxcz0123456789[{]} \t\n\r\f\v#" |
| 1244 | : ""; |
| 1245 | |
| 1246 | if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) { |
| 1247 | /* If we are doing a trans and we know we want UTF8 set expectation */ |
| 1248 | has_utf8 = PL_sublex_info.sub_op->op_private & (OPpTRANS_FROM_UTF|OPpTRANS_TO_UTF); |
| 1249 | this_utf8 = PL_sublex_info.sub_op->op_private & (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF); |
| 1250 | } |
| 1251 | |
| 1252 | |
| 1253 | while (s < send || dorange) { |
| 1254 | /* get transliterations out of the way (they're most literal) */ |
| 1255 | if (PL_lex_inwhat == OP_TRANS) { |
| 1256 | /* expand a range A-Z to the full set of characters. AIE! */ |
| 1257 | if (dorange) { |
| 1258 | I32 i; /* current expanded character */ |
| 1259 | I32 min; /* first character in range */ |
| 1260 | I32 max; /* last character in range */ |
| 1261 | |
| 1262 | if (has_utf8) { |
| 1263 | char *c = (char*)utf8_hop((U8*)d, -1); |
| 1264 | char *e = d++; |
| 1265 | while (e-- > c) |
| 1266 | *(e + 1) = *e; |
| 1267 | *c = (char)UTF_TO_NATIVE(0xff); |
| 1268 | /* mark the range as done, and continue */ |
| 1269 | dorange = FALSE; |
| 1270 | didrange = TRUE; |
| 1271 | continue; |
| 1272 | } |
| 1273 | |
| 1274 | i = d - SvPVX(sv); /* remember current offset */ |
| 1275 | SvGROW(sv, SvLEN(sv) + 256); /* never more than 256 chars in a range */ |
| 1276 | d = SvPVX(sv) + i; /* refresh d after realloc */ |
| 1277 | d -= 2; /* eat the first char and the - */ |
| 1278 | |
| 1279 | min = (U8)*d; /* first char in range */ |
| 1280 | max = (U8)d[1]; /* last char in range */ |
| 1281 | |
| 1282 | if (min > max) { |
| 1283 | Perl_croak(aTHX_ |
| 1284 | "Invalid [] range \"%c-%c\" in transliteration operator", |
| 1285 | (char)min, (char)max); |
| 1286 | } |
| 1287 | |
| 1288 | #ifdef EBCDIC |
| 1289 | if ((isLOWER(min) && isLOWER(max)) || |
| 1290 | (isUPPER(min) && isUPPER(max))) { |
| 1291 | if (isLOWER(min)) { |
| 1292 | for (i = min; i <= max; i++) |
| 1293 | if (isLOWER(i)) |
| 1294 | *d++ = NATIVE_TO_NEED(has_utf8,i); |
| 1295 | } else { |
| 1296 | for (i = min; i <= max; i++) |
| 1297 | if (isUPPER(i)) |
| 1298 | *d++ = NATIVE_TO_NEED(has_utf8,i); |
| 1299 | } |
| 1300 | } |
| 1301 | else |
| 1302 | #endif |
| 1303 | for (i = min; i <= max; i++) |
| 1304 | *d++ = i; |
| 1305 | |
| 1306 | /* mark the range as done, and continue */ |
| 1307 | dorange = FALSE; |
| 1308 | didrange = TRUE; |
| 1309 | continue; |
| 1310 | } |
| 1311 | |
| 1312 | /* range begins (ignore - as first or last char) */ |
| 1313 | else if (*s == '-' && s+1 < send && s != start) { |
| 1314 | if (didrange) { |
| 1315 | Perl_croak(aTHX_ "Ambiguous range in transliteration operator"); |
| 1316 | } |
| 1317 | if (has_utf8) { |
| 1318 | *d++ = (char)UTF_TO_NATIVE(0xff); /* use illegal utf8 byte--see pmtrans */ |
| 1319 | s++; |
| 1320 | continue; |
| 1321 | } |
| 1322 | dorange = TRUE; |
| 1323 | s++; |
| 1324 | } |
| 1325 | else { |
| 1326 | didrange = FALSE; |
| 1327 | } |
| 1328 | } |
| 1329 | |
| 1330 | /* if we get here, we're not doing a transliteration */ |
| 1331 | |
| 1332 | /* skip for regexp comments /(?#comment)/ and code /(?{code})/, |
| 1333 | except for the last char, which will be done separately. */ |
| 1334 | else if (*s == '(' && PL_lex_inpat && s[1] == '?') { |
| 1335 | if (s[2] == '#') { |
| 1336 | while (s < send && *s != ')') |
| 1337 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
| 1338 | } |
| 1339 | else if (s[2] == '{' /* This should match regcomp.c */ |
| 1340 | || ((s[2] == 'p' || s[2] == '?') && s[3] == '{')) |
| 1341 | { |
| 1342 | I32 count = 1; |
| 1343 | char *regparse = s + (s[2] == '{' ? 3 : 4); |
| 1344 | char c; |
| 1345 | |
| 1346 | while (count && (c = *regparse)) { |
| 1347 | if (c == '\\' && regparse[1]) |
| 1348 | regparse++; |
| 1349 | else if (c == '{') |
| 1350 | count++; |
| 1351 | else if (c == '}') |
| 1352 | count--; |
| 1353 | regparse++; |
| 1354 | } |
| 1355 | if (*regparse != ')') { |
| 1356 | regparse--; /* Leave one char for continuation. */ |
| 1357 | yyerror("Sequence (?{...}) not terminated or not {}-balanced"); |
| 1358 | } |
| 1359 | while (s < regparse) |
| 1360 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
| 1361 | } |
| 1362 | } |
| 1363 | |
| 1364 | /* likewise skip #-initiated comments in //x patterns */ |
| 1365 | else if (*s == '#' && PL_lex_inpat && |
| 1366 | ((PMOP*)PL_lex_inpat)->op_pmflags & PMf_EXTENDED) { |
| 1367 | while (s+1 < send && *s != '\n') |
| 1368 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
| 1369 | } |
| 1370 | |
| 1371 | /* check for embedded arrays |
| 1372 | (@foo, @:foo, @'foo, @{foo}, @$foo, @+, @-) |
| 1373 | */ |
| 1374 | else if (*s == '@' && s[1] |
| 1375 | && (isALNUM_lazy_if(s+1,UTF) || strchr(":'{$+-", s[1]))) |
| 1376 | break; |
| 1377 | |
| 1378 | /* check for embedded scalars. only stop if we're sure it's a |
| 1379 | variable. |
| 1380 | */ |
| 1381 | else if (*s == '$') { |
| 1382 | if (!PL_lex_inpat) /* not a regexp, so $ must be var */ |
| 1383 | break; |
| 1384 | if (s + 1 < send && !strchr("()| \r\n\t", s[1])) |
| 1385 | break; /* in regexp, $ might be tail anchor */ |
| 1386 | } |
| 1387 | |
| 1388 | /* End of else if chain - OP_TRANS rejoin rest */ |
| 1389 | |
| 1390 | /* backslashes */ |
| 1391 | if (*s == '\\' && s+1 < send) { |
| 1392 | s++; |
| 1393 | |
| 1394 | /* some backslashes we leave behind */ |
| 1395 | if (*leaveit && *s && strchr(leaveit, *s)) { |
| 1396 | *d++ = NATIVE_TO_NEED(has_utf8,'\\'); |
| 1397 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
| 1398 | continue; |
| 1399 | } |
| 1400 | |
| 1401 | /* deprecate \1 in strings and substitution replacements */ |
| 1402 | if (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat && |
| 1403 | isDIGIT(*s) && *s != '0' && !isDIGIT(s[1])) |
| 1404 | { |
| 1405 | if (ckWARN(WARN_SYNTAX)) |
| 1406 | Perl_warner(aTHX_ WARN_SYNTAX, "\\%c better written as $%c", *s, *s); |
| 1407 | *--s = '$'; |
| 1408 | break; |
| 1409 | } |
| 1410 | |
| 1411 | /* string-change backslash escapes */ |
| 1412 | if (PL_lex_inwhat != OP_TRANS && *s && strchr("lLuUEQ", *s)) { |
| 1413 | --s; |
| 1414 | break; |
| 1415 | } |
| 1416 | |
| 1417 | /* if we get here, it's either a quoted -, or a digit */ |
| 1418 | switch (*s) { |
| 1419 | |
| 1420 | /* quoted - in transliterations */ |
| 1421 | case '-': |
| 1422 | if (PL_lex_inwhat == OP_TRANS) { |
| 1423 | *d++ = *s++; |
| 1424 | continue; |
| 1425 | } |
| 1426 | /* FALL THROUGH */ |
| 1427 | default: |
| 1428 | { |
| 1429 | if (ckWARN(WARN_MISC) && isALNUM(*s)) |
| 1430 | Perl_warner(aTHX_ WARN_MISC, |
| 1431 | "Unrecognized escape \\%c passed through", |
| 1432 | *s); |
| 1433 | /* default action is to copy the quoted character */ |
| 1434 | goto default_action; |
| 1435 | } |
| 1436 | |
| 1437 | /* \132 indicates an octal constant */ |
| 1438 | case '0': case '1': case '2': case '3': |
| 1439 | case '4': case '5': case '6': case '7': |
| 1440 | { |
| 1441 | I32 flags = 0; |
| 1442 | STRLEN len = 3; |
| 1443 | uv = grok_oct(s, &len, &flags, NULL); |
| 1444 | s += len; |
| 1445 | } |
| 1446 | goto NUM_ESCAPE_INSERT; |
| 1447 | |
| 1448 | /* \x24 indicates a hex constant */ |
| 1449 | case 'x': |
| 1450 | ++s; |
| 1451 | if (*s == '{') { |
| 1452 | char* e = strchr(s, '}'); |
| 1453 | I32 flags = PERL_SCAN_ALLOW_UNDERSCORES | |
| 1454 | PERL_SCAN_DISALLOW_PREFIX; |
| 1455 | STRLEN len; |
| 1456 | |
| 1457 | ++s; |
| 1458 | if (!e) { |
| 1459 | yyerror("Missing right brace on \\x{}"); |
| 1460 | continue; |
| 1461 | } |
| 1462 | len = e - s; |
| 1463 | uv = grok_hex(s, &len, &flags, NULL); |
| 1464 | s = e + 1; |
| 1465 | } |
| 1466 | else { |
| 1467 | { |
| 1468 | STRLEN len = 2; |
| 1469 | I32 flags = PERL_SCAN_DISALLOW_PREFIX; |
| 1470 | uv = grok_hex(s, &len, &flags, NULL); |
| 1471 | s += len; |
| 1472 | } |
| 1473 | } |
| 1474 | |
| 1475 | NUM_ESCAPE_INSERT: |
| 1476 | /* Insert oct or hex escaped character. |
| 1477 | * There will always enough room in sv since such |
| 1478 | * escapes will be longer than any UTF-8 sequence |
| 1479 | * they can end up as. */ |
| 1480 | |
| 1481 | /* We need to map to chars to ASCII before doing the tests |
| 1482 | to cover EBCDIC |
| 1483 | */ |
| 1484 | if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(uv))) { |
| 1485 | if (!has_utf8 && uv > 255) { |
| 1486 | /* Might need to recode whatever we have |
| 1487 | * accumulated so far if it contains any |
| 1488 | * hibit chars. |
| 1489 | * |
| 1490 | * (Can't we keep track of that and avoid |
| 1491 | * this rescan? --jhi) |
| 1492 | */ |
| 1493 | int hicount = 0; |
| 1494 | U8 *c; |
| 1495 | for (c = (U8 *) SvPVX(sv); c < (U8 *)d; c++) { |
| 1496 | if (!NATIVE_IS_INVARIANT(*c)) { |
| 1497 | hicount++; |
| 1498 | } |
| 1499 | } |
| 1500 | if (hicount) { |
| 1501 | STRLEN offset = d - SvPVX(sv); |
| 1502 | U8 *src, *dst; |
| 1503 | d = SvGROW(sv, SvLEN(sv) + hicount + 1) + offset; |
| 1504 | src = (U8 *)d - 1; |
| 1505 | dst = src+hicount; |
| 1506 | d += hicount; |
| 1507 | while (src >= (U8 *)SvPVX(sv)) { |
| 1508 | if (!NATIVE_IS_INVARIANT(*src)) { |
| 1509 | U8 ch = NATIVE_TO_ASCII(*src); |
| 1510 | *dst-- = UTF8_EIGHT_BIT_LO(ch); |
| 1511 | *dst-- = UTF8_EIGHT_BIT_HI(ch); |
| 1512 | } |
| 1513 | else { |
| 1514 | *dst-- = *src; |
| 1515 | } |
| 1516 | src--; |
| 1517 | } |
| 1518 | } |
| 1519 | } |
| 1520 | |
| 1521 | if (has_utf8 || uv > 255) { |
| 1522 | d = (char*)uvchr_to_utf8((U8*)d, uv); |
| 1523 | has_utf8 = TRUE; |
| 1524 | if (PL_lex_inwhat == OP_TRANS && |
| 1525 | PL_sublex_info.sub_op) { |
| 1526 | PL_sublex_info.sub_op->op_private |= |
| 1527 | (PL_lex_repl ? OPpTRANS_FROM_UTF |
| 1528 | : OPpTRANS_TO_UTF); |
| 1529 | } |
| 1530 | } |
| 1531 | else { |
| 1532 | *d++ = (char)uv; |
| 1533 | } |
| 1534 | } |
| 1535 | else { |
| 1536 | *d++ = (char) uv; |
| 1537 | } |
| 1538 | continue; |
| 1539 | |
| 1540 | /* \N{latin small letter a} is a named character */ |
| 1541 | case 'N': |
| 1542 | ++s; |
| 1543 | if (*s == '{') { |
| 1544 | char* e = strchr(s, '}'); |
| 1545 | SV *res; |
| 1546 | STRLEN len; |
| 1547 | char *str; |
| 1548 | |
| 1549 | if (!e) { |
| 1550 | yyerror("Missing right brace on \\N{}"); |
| 1551 | e = s - 1; |
| 1552 | goto cont_scan; |
| 1553 | } |
| 1554 | res = newSVpvn(s + 1, e - s - 1); |
| 1555 | res = new_constant( Nullch, 0, "charnames", |
| 1556 | res, Nullsv, "\\N{...}" ); |
| 1557 | if (has_utf8) |
| 1558 | sv_utf8_upgrade(res); |
| 1559 | str = SvPV(res,len); |
| 1560 | if (!has_utf8 && SvUTF8(res)) { |
| 1561 | char *ostart = SvPVX(sv); |
| 1562 | SvCUR_set(sv, d - ostart); |
| 1563 | SvPOK_on(sv); |
| 1564 | *d = '\0'; |
| 1565 | sv_utf8_upgrade(sv); |
| 1566 | /* this just broke our allocation above... */ |
| 1567 | SvGROW(sv, send - start); |
| 1568 | d = SvPVX(sv) + SvCUR(sv); |
| 1569 | has_utf8 = TRUE; |
| 1570 | } |
| 1571 | if (len > e - s + 4) { |
| 1572 | char *odest = SvPVX(sv); |
| 1573 | |
| 1574 | SvGROW(sv, (SvLEN(sv) + len - (e - s + 4))); |
| 1575 | d = SvPVX(sv) + (d - odest); |
| 1576 | } |
| 1577 | Copy(str, d, len, char); |
| 1578 | d += len; |
| 1579 | SvREFCNT_dec(res); |
| 1580 | cont_scan: |
| 1581 | s = e + 1; |
| 1582 | } |
| 1583 | else |
| 1584 | yyerror("Missing braces on \\N{}"); |
| 1585 | continue; |
| 1586 | |
| 1587 | /* \c is a control character */ |
| 1588 | case 'c': |
| 1589 | s++; |
| 1590 | { |
| 1591 | U8 c = *s++; |
| 1592 | #ifdef EBCDIC |
| 1593 | if (isLOWER(c)) |
| 1594 | c = toUPPER(c); |
| 1595 | #endif |
| 1596 | *d++ = NATIVE_TO_NEED(has_utf8,toCTRL(c)); |
| 1597 | } |
| 1598 | continue; |
| 1599 | |
| 1600 | /* printf-style backslashes, formfeeds, newlines, etc */ |
| 1601 | case 'b': |
| 1602 | *d++ = NATIVE_TO_NEED(has_utf8,'\b'); |
| 1603 | break; |
| 1604 | case 'n': |
| 1605 | *d++ = NATIVE_TO_NEED(has_utf8,'\n'); |
| 1606 | break; |
| 1607 | case 'r': |
| 1608 | *d++ = NATIVE_TO_NEED(has_utf8,'\r'); |
| 1609 | break; |
| 1610 | case 'f': |
| 1611 | *d++ = NATIVE_TO_NEED(has_utf8,'\f'); |
| 1612 | break; |
| 1613 | case 't': |
| 1614 | *d++ = NATIVE_TO_NEED(has_utf8,'\t'); |
| 1615 | break; |
| 1616 | case 'e': |
| 1617 | *d++ = ASCII_TO_NEED(has_utf8,'\033'); |
| 1618 | break; |
| 1619 | case 'a': |
| 1620 | *d++ = ASCII_TO_NEED(has_utf8,'\007'); |
| 1621 | break; |
| 1622 | } /* end switch */ |
| 1623 | |
| 1624 | s++; |
| 1625 | continue; |
| 1626 | } /* end if (backslash) */ |
| 1627 | |
| 1628 | default_action: |
| 1629 | /* If we started with encoded form, or already know we want it |
| 1630 | and then encode the next character */ |
| 1631 | if ((has_utf8 || this_utf8) && !NATIVE_IS_INVARIANT((U8)(*s))) { |
| 1632 | STRLEN len = 1; |
| 1633 | UV uv = (this_utf8) ? utf8n_to_uvchr((U8*)s, send - s, &len, 0) : (UV) ((U8) *s); |
| 1634 | STRLEN need = UNISKIP(NATIVE_TO_UNI(uv)); |
| 1635 | s += len; |
| 1636 | if (need > len) { |
| 1637 | /* encoded value larger than old, need extra space (NOTE: SvCUR() not set here) */ |
| 1638 | STRLEN off = d - SvPVX(sv); |
| 1639 | d = SvGROW(sv, SvLEN(sv) + (need-len)) + off; |
| 1640 | } |
| 1641 | d = (char*)uvchr_to_utf8((U8*)d, uv); |
| 1642 | has_utf8 = TRUE; |
| 1643 | } |
| 1644 | else { |
| 1645 | *d++ = NATIVE_TO_NEED(has_utf8,*s++); |
| 1646 | } |
| 1647 | } /* while loop to process each character */ |
| 1648 | |
| 1649 | /* terminate the string and set up the sv */ |
| 1650 | *d = '\0'; |
| 1651 | SvCUR_set(sv, d - SvPVX(sv)); |
| 1652 | if (SvCUR(sv) >= SvLEN(sv)) |
| 1653 | Perl_croak(aTHX_ "panic: constant overflowed allocated space"); |
| 1654 | |
| 1655 | SvPOK_on(sv); |
| 1656 | if (has_utf8) { |
| 1657 | SvUTF8_on(sv); |
| 1658 | if (PL_lex_inwhat == OP_TRANS && PL_sublex_info.sub_op) { |
| 1659 | PL_sublex_info.sub_op->op_private |= |
| 1660 | (PL_lex_repl ? OPpTRANS_FROM_UTF : OPpTRANS_TO_UTF); |
| 1661 | } |
| 1662 | } |
| 1663 | |
| 1664 | /* shrink the sv if we allocated more than we used */ |
| 1665 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
| 1666 | SvLEN_set(sv, SvCUR(sv) + 1); |
| 1667 | Renew(SvPVX(sv), SvLEN(sv), char); |
| 1668 | } |
| 1669 | |
| 1670 | /* return the substring (via yylval) only if we parsed anything */ |
| 1671 | if (s > PL_bufptr) { |
| 1672 | if ( PL_hints & ( PL_lex_inpat ? HINT_NEW_RE : HINT_NEW_STRING ) ) |
| 1673 | sv = new_constant(start, s - start, (PL_lex_inpat ? "qr" : "q"), |
| 1674 | sv, Nullsv, |
| 1675 | ( PL_lex_inwhat == OP_TRANS |
| 1676 | ? "tr" |
| 1677 | : ( (PL_lex_inwhat == OP_SUBST && !PL_lex_inpat) |
| 1678 | ? "s" |
| 1679 | : "qq"))); |
| 1680 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
| 1681 | } else |
| 1682 | SvREFCNT_dec(sv); |
| 1683 | return s; |
| 1684 | } |
| 1685 | |
| 1686 | /* S_intuit_more |
| 1687 | * Returns TRUE if there's more to the expression (e.g., a subscript), |
| 1688 | * FALSE otherwise. |
| 1689 | * |
| 1690 | * It deals with "$foo[3]" and /$foo[3]/ and /$foo[0123456789$]+/ |
| 1691 | * |
| 1692 | * ->[ and ->{ return TRUE |
| 1693 | * { and [ outside a pattern are always subscripts, so return TRUE |
| 1694 | * if we're outside a pattern and it's not { or [, then return FALSE |
| 1695 | * if we're in a pattern and the first char is a { |
| 1696 | * {4,5} (any digits around the comma) returns FALSE |
| 1697 | * if we're in a pattern and the first char is a [ |
| 1698 | * [] returns FALSE |
| 1699 | * [SOMETHING] has a funky algorithm to decide whether it's a |
| 1700 | * character class or not. It has to deal with things like |
| 1701 | * /$foo[-3]/ and /$foo[$bar]/ as well as /$foo[$\d]+/ |
| 1702 | * anything else returns TRUE |
| 1703 | */ |
| 1704 | |
| 1705 | /* This is the one truly awful dwimmer necessary to conflate C and sed. */ |
| 1706 | |
| 1707 | STATIC int |
| 1708 | S_intuit_more(pTHX_ register char *s) |
| 1709 | { |
| 1710 | if (PL_lex_brackets) |
| 1711 | return TRUE; |
| 1712 | if (*s == '-' && s[1] == '>' && (s[2] == '[' || s[2] == '{')) |
| 1713 | return TRUE; |
| 1714 | if (*s != '{' && *s != '[') |
| 1715 | return FALSE; |
| 1716 | if (!PL_lex_inpat) |
| 1717 | return TRUE; |
| 1718 | |
| 1719 | /* In a pattern, so maybe we have {n,m}. */ |
| 1720 | if (*s == '{') { |
| 1721 | s++; |
| 1722 | if (!isDIGIT(*s)) |
| 1723 | return TRUE; |
| 1724 | while (isDIGIT(*s)) |
| 1725 | s++; |
| 1726 | if (*s == ',') |
| 1727 | s++; |
| 1728 | while (isDIGIT(*s)) |
| 1729 | s++; |
| 1730 | if (*s == '}') |
| 1731 | return FALSE; |
| 1732 | return TRUE; |
| 1733 | |
| 1734 | } |
| 1735 | |
| 1736 | /* On the other hand, maybe we have a character class */ |
| 1737 | |
| 1738 | s++; |
| 1739 | if (*s == ']' || *s == '^') |
| 1740 | return FALSE; |
| 1741 | else { |
| 1742 | /* this is terrifying, and it works */ |
| 1743 | int weight = 2; /* let's weigh the evidence */ |
| 1744 | char seen[256]; |
| 1745 | unsigned char un_char = 255, last_un_char; |
| 1746 | char *send = strchr(s,']'); |
| 1747 | char tmpbuf[sizeof PL_tokenbuf * 4]; |
| 1748 | |
| 1749 | if (!send) /* has to be an expression */ |
| 1750 | return TRUE; |
| 1751 | |
| 1752 | Zero(seen,256,char); |
| 1753 | if (*s == '$') |
| 1754 | weight -= 3; |
| 1755 | else if (isDIGIT(*s)) { |
| 1756 | if (s[1] != ']') { |
| 1757 | if (isDIGIT(s[1]) && s[2] == ']') |
| 1758 | weight -= 10; |
| 1759 | } |
| 1760 | else |
| 1761 | weight -= 100; |
| 1762 | } |
| 1763 | for (; s < send; s++) { |
| 1764 | last_un_char = un_char; |
| 1765 | un_char = (unsigned char)*s; |
| 1766 | switch (*s) { |
| 1767 | case '@': |
| 1768 | case '&': |
| 1769 | case '$': |
| 1770 | weight -= seen[un_char] * 10; |
| 1771 | if (isALNUM_lazy_if(s+1,UTF)) { |
| 1772 | scan_ident(s, send, tmpbuf, sizeof tmpbuf, FALSE); |
| 1773 | if ((int)strlen(tmpbuf) > 1 && gv_fetchpv(tmpbuf,FALSE, SVt_PV)) |
| 1774 | weight -= 100; |
| 1775 | else |
| 1776 | weight -= 10; |
| 1777 | } |
| 1778 | else if (*s == '$' && s[1] && |
| 1779 | strchr("[#!%*<>()-=",s[1])) { |
| 1780 | if (/*{*/ strchr("])} =",s[2])) |
| 1781 | weight -= 10; |
| 1782 | else |
| 1783 | weight -= 1; |
| 1784 | } |
| 1785 | break; |
| 1786 | case '\\': |
| 1787 | un_char = 254; |
| 1788 | if (s[1]) { |
| 1789 | if (strchr("wds]",s[1])) |
| 1790 | weight += 100; |
| 1791 | else if (seen['\''] || seen['"']) |
| 1792 | weight += 1; |
| 1793 | else if (strchr("rnftbxcav",s[1])) |
| 1794 | weight += 40; |
| 1795 | else if (isDIGIT(s[1])) { |
| 1796 | weight += 40; |
| 1797 | while (s[1] && isDIGIT(s[1])) |
| 1798 | s++; |
| 1799 | } |
| 1800 | } |
| 1801 | else |
| 1802 | weight += 100; |
| 1803 | break; |
| 1804 | case '-': |
| 1805 | if (s[1] == '\\') |
| 1806 | weight += 50; |
| 1807 | if (strchr("aA01! ",last_un_char)) |
| 1808 | weight += 30; |
| 1809 | if (strchr("zZ79~",s[1])) |
| 1810 | weight += 30; |
| 1811 | if (last_un_char == 255 && (isDIGIT(s[1]) || s[1] == '$')) |
| 1812 | weight -= 5; /* cope with negative subscript */ |
| 1813 | break; |
| 1814 | default: |
| 1815 | if (!isALNUM(last_un_char) && !strchr("$@&",last_un_char) && |
| 1816 | isALPHA(*s) && s[1] && isALPHA(s[1])) { |
| 1817 | char *d = tmpbuf; |
| 1818 | while (isALPHA(*s)) |
| 1819 | *d++ = *s++; |
| 1820 | *d = '\0'; |
| 1821 | if (keyword(tmpbuf, d - tmpbuf)) |
| 1822 | weight -= 150; |
| 1823 | } |
| 1824 | if (un_char == last_un_char + 1) |
| 1825 | weight += 5; |
| 1826 | weight -= seen[un_char]; |
| 1827 | break; |
| 1828 | } |
| 1829 | seen[un_char]++; |
| 1830 | } |
| 1831 | if (weight >= 0) /* probably a character class */ |
| 1832 | return FALSE; |
| 1833 | } |
| 1834 | |
| 1835 | return TRUE; |
| 1836 | } |
| 1837 | |
| 1838 | /* |
| 1839 | * S_intuit_method |
| 1840 | * |
| 1841 | * Does all the checking to disambiguate |
| 1842 | * foo bar |
| 1843 | * between foo(bar) and bar->foo. Returns 0 if not a method, otherwise |
| 1844 | * FUNCMETH (bar->foo(args)) or METHOD (bar->foo args). |
| 1845 | * |
| 1846 | * First argument is the stuff after the first token, e.g. "bar". |
| 1847 | * |
| 1848 | * Not a method if bar is a filehandle. |
| 1849 | * Not a method if foo is a subroutine prototyped to take a filehandle. |
| 1850 | * Not a method if it's really "Foo $bar" |
| 1851 | * Method if it's "foo $bar" |
| 1852 | * Not a method if it's really "print foo $bar" |
| 1853 | * Method if it's really "foo package::" (interpreted as package->foo) |
| 1854 | * Not a method if bar is known to be a subroutne ("sub bar; foo bar") |
| 1855 | * Not a method if bar is a filehandle or package, but is quoted with |
| 1856 | * => |
| 1857 | */ |
| 1858 | |
| 1859 | STATIC int |
| 1860 | S_intuit_method(pTHX_ char *start, GV *gv) |
| 1861 | { |
| 1862 | char *s = start + (*start == '$'); |
| 1863 | char tmpbuf[sizeof PL_tokenbuf]; |
| 1864 | STRLEN len; |
| 1865 | GV* indirgv; |
| 1866 | |
| 1867 | if (gv) { |
| 1868 | CV *cv; |
| 1869 | if (GvIO(gv)) |
| 1870 | return 0; |
| 1871 | if ((cv = GvCVu(gv))) { |
| 1872 | char *proto = SvPVX(cv); |
| 1873 | if (proto) { |
| 1874 | if (*proto == ';') |
| 1875 | proto++; |
| 1876 | if (*proto == '*') |
| 1877 | return 0; |
| 1878 | } |
| 1879 | } else |
| 1880 | gv = 0; |
| 1881 | } |
| 1882 | s = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
| 1883 | /* start is the beginning of the possible filehandle/object, |
| 1884 | * and s is the end of it |
| 1885 | * tmpbuf is a copy of it |
| 1886 | */ |
| 1887 | |
| 1888 | if (*start == '$') { |
| 1889 | if (gv || PL_last_lop_op == OP_PRINT || isUPPER(*PL_tokenbuf)) |
| 1890 | return 0; |
| 1891 | s = skipspace(s); |
| 1892 | PL_bufptr = start; |
| 1893 | PL_expect = XREF; |
| 1894 | return *s == '(' ? FUNCMETH : METHOD; |
| 1895 | } |
| 1896 | if (!keyword(tmpbuf, len)) { |
| 1897 | if (len > 2 && tmpbuf[len - 2] == ':' && tmpbuf[len - 1] == ':') { |
| 1898 | len -= 2; |
| 1899 | tmpbuf[len] = '\0'; |
| 1900 | goto bare_package; |
| 1901 | } |
| 1902 | indirgv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV); |
| 1903 | if (indirgv && GvCVu(indirgv)) |
| 1904 | return 0; |
| 1905 | /* filehandle or package name makes it a method */ |
| 1906 | if (!gv || GvIO(indirgv) || gv_stashpvn(tmpbuf, len, FALSE)) { |
| 1907 | s = skipspace(s); |
| 1908 | if ((PL_bufend - s) >= 2 && *s == '=' && *(s+1) == '>') |
| 1909 | return 0; /* no assumptions -- "=>" quotes bearword */ |
| 1910 | bare_package: |
| 1911 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, |
| 1912 | newSVpvn(tmpbuf,len)); |
| 1913 | PL_nextval[PL_nexttoke].opval->op_private = OPpCONST_BARE; |
| 1914 | PL_expect = XTERM; |
| 1915 | force_next(WORD); |
| 1916 | PL_bufptr = s; |
| 1917 | return *s == '(' ? FUNCMETH : METHOD; |
| 1918 | } |
| 1919 | } |
| 1920 | return 0; |
| 1921 | } |
| 1922 | |
| 1923 | /* |
| 1924 | * S_incl_perldb |
| 1925 | * Return a string of Perl code to load the debugger. If PERL5DB |
| 1926 | * is set, it will return the contents of that, otherwise a |
| 1927 | * compile-time require of perl5db.pl. |
| 1928 | */ |
| 1929 | |
| 1930 | STATIC char* |
| 1931 | S_incl_perldb(pTHX) |
| 1932 | { |
| 1933 | if (PL_perldb) { |
| 1934 | char *pdb = PerlEnv_getenv("PERL5DB"); |
| 1935 | |
| 1936 | if (pdb) |
| 1937 | return pdb; |
| 1938 | SETERRNO(0,SS$_NORMAL); |
| 1939 | return "BEGIN { require 'perl5db.pl' }"; |
| 1940 | } |
| 1941 | return ""; |
| 1942 | } |
| 1943 | |
| 1944 | |
| 1945 | /* Encoded script support. filter_add() effectively inserts a |
| 1946 | * 'pre-processing' function into the current source input stream. |
| 1947 | * Note that the filter function only applies to the current source file |
| 1948 | * (e.g., it will not affect files 'require'd or 'use'd by this one). |
| 1949 | * |
| 1950 | * The datasv parameter (which may be NULL) can be used to pass |
| 1951 | * private data to this instance of the filter. The filter function |
| 1952 | * can recover the SV using the FILTER_DATA macro and use it to |
| 1953 | * store private buffers and state information. |
| 1954 | * |
| 1955 | * The supplied datasv parameter is upgraded to a PVIO type |
| 1956 | * and the IoDIRP/IoANY field is used to store the function pointer, |
| 1957 | * and IOf_FAKE_DIRP is enabled on datasv to mark this as such. |
| 1958 | * Note that IoTOP_NAME, IoFMT_NAME, IoBOTTOM_NAME, if set for |
| 1959 | * private use must be set using malloc'd pointers. |
| 1960 | */ |
| 1961 | |
| 1962 | SV * |
| 1963 | Perl_filter_add(pTHX_ filter_t funcp, SV *datasv) |
| 1964 | { |
| 1965 | if (!funcp) |
| 1966 | return Nullsv; |
| 1967 | |
| 1968 | if (!PL_rsfp_filters) |
| 1969 | PL_rsfp_filters = newAV(); |
| 1970 | if (!datasv) |
| 1971 | datasv = NEWSV(255,0); |
| 1972 | if (!SvUPGRADE(datasv, SVt_PVIO)) |
| 1973 | Perl_die(aTHX_ "Can't upgrade filter_add data to SVt_PVIO"); |
| 1974 | IoANY(datasv) = (void *)funcp; /* stash funcp into spare field */ |
| 1975 | IoFLAGS(datasv) |= IOf_FAKE_DIRP; |
| 1976 | DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_add func %p (%s)\n", |
| 1977 | funcp, SvPV_nolen(datasv))); |
| 1978 | av_unshift(PL_rsfp_filters, 1); |
| 1979 | av_store(PL_rsfp_filters, 0, datasv) ; |
| 1980 | return(datasv); |
| 1981 | } |
| 1982 | |
| 1983 | |
| 1984 | /* Delete most recently added instance of this filter function. */ |
| 1985 | void |
| 1986 | Perl_filter_del(pTHX_ filter_t funcp) |
| 1987 | { |
| 1988 | SV *datasv; |
| 1989 | DEBUG_P(PerlIO_printf(Perl_debug_log, "filter_del func %p", funcp)); |
| 1990 | if (!PL_rsfp_filters || AvFILLp(PL_rsfp_filters)<0) |
| 1991 | return; |
| 1992 | /* if filter is on top of stack (usual case) just pop it off */ |
| 1993 | datasv = FILTER_DATA(AvFILLp(PL_rsfp_filters)); |
| 1994 | if (IoANY(datasv) == (void *)funcp) { |
| 1995 | IoFLAGS(datasv) &= ~IOf_FAKE_DIRP; |
| 1996 | IoANY(datasv) = (void *)NULL; |
| 1997 | sv_free(av_pop(PL_rsfp_filters)); |
| 1998 | |
| 1999 | return; |
| 2000 | } |
| 2001 | /* we need to search for the correct entry and clear it */ |
| 2002 | Perl_die(aTHX_ "filter_del can only delete in reverse order (currently)"); |
| 2003 | } |
| 2004 | |
| 2005 | |
| 2006 | /* Invoke the n'th filter function for the current rsfp. */ |
| 2007 | I32 |
| 2008 | Perl_filter_read(pTHX_ int idx, SV *buf_sv, int maxlen) |
| 2009 | |
| 2010 | |
| 2011 | /* 0 = read one text line */ |
| 2012 | { |
| 2013 | filter_t funcp; |
| 2014 | SV *datasv = NULL; |
| 2015 | |
| 2016 | if (!PL_rsfp_filters) |
| 2017 | return -1; |
| 2018 | if (idx > AvFILLp(PL_rsfp_filters)){ /* Any more filters? */ |
| 2019 | /* Provide a default input filter to make life easy. */ |
| 2020 | /* Note that we append to the line. This is handy. */ |
| 2021 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
| 2022 | "filter_read %d: from rsfp\n", idx)); |
| 2023 | if (maxlen) { |
| 2024 | /* Want a block */ |
| 2025 | int len ; |
| 2026 | int old_len = SvCUR(buf_sv) ; |
| 2027 | |
| 2028 | /* ensure buf_sv is large enough */ |
| 2029 | SvGROW(buf_sv, old_len + maxlen) ; |
| 2030 | if ((len = PerlIO_read(PL_rsfp, SvPVX(buf_sv) + old_len, maxlen)) <= 0){ |
| 2031 | if (PerlIO_error(PL_rsfp)) |
| 2032 | return -1; /* error */ |
| 2033 | else |
| 2034 | return 0 ; /* end of file */ |
| 2035 | } |
| 2036 | SvCUR_set(buf_sv, old_len + len) ; |
| 2037 | } else { |
| 2038 | /* Want a line */ |
| 2039 | if (sv_gets(buf_sv, PL_rsfp, SvCUR(buf_sv)) == NULL) { |
| 2040 | if (PerlIO_error(PL_rsfp)) |
| 2041 | return -1; /* error */ |
| 2042 | else |
| 2043 | return 0 ; /* end of file */ |
| 2044 | } |
| 2045 | } |
| 2046 | return SvCUR(buf_sv); |
| 2047 | } |
| 2048 | /* Skip this filter slot if filter has been deleted */ |
| 2049 | if ( (datasv = FILTER_DATA(idx)) == &PL_sv_undef){ |
| 2050 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
| 2051 | "filter_read %d: skipped (filter deleted)\n", |
| 2052 | idx)); |
| 2053 | return FILTER_READ(idx+1, buf_sv, maxlen); /* recurse */ |
| 2054 | } |
| 2055 | /* Get function pointer hidden within datasv */ |
| 2056 | funcp = (filter_t)IoANY(datasv); |
| 2057 | DEBUG_P(PerlIO_printf(Perl_debug_log, |
| 2058 | "filter_read %d: via function %p (%s)\n", |
| 2059 | idx, funcp, SvPV_nolen(datasv))); |
| 2060 | /* Call function. The function is expected to */ |
| 2061 | /* call "FILTER_READ(idx+1, buf_sv)" first. */ |
| 2062 | /* Return: <0:error, =0:eof, >0:not eof */ |
| 2063 | return (*funcp)(aTHX_ idx, buf_sv, maxlen); |
| 2064 | } |
| 2065 | |
| 2066 | STATIC char * |
| 2067 | S_filter_gets(pTHX_ register SV *sv, register PerlIO *fp, STRLEN append) |
| 2068 | { |
| 2069 | #ifdef PERL_CR_FILTER |
| 2070 | if (!PL_rsfp_filters) { |
| 2071 | filter_add(S_cr_textfilter,NULL); |
| 2072 | } |
| 2073 | #endif |
| 2074 | if (PL_rsfp_filters) { |
| 2075 | |
| 2076 | if (!append) |
| 2077 | SvCUR_set(sv, 0); /* start with empty line */ |
| 2078 | if (FILTER_READ(0, sv, 0) > 0) |
| 2079 | return ( SvPVX(sv) ) ; |
| 2080 | else |
| 2081 | return Nullch ; |
| 2082 | } |
| 2083 | else |
| 2084 | return (sv_gets(sv, fp, append)); |
| 2085 | } |
| 2086 | |
| 2087 | STATIC HV * |
| 2088 | S_find_in_my_stash(pTHX_ char *pkgname, I32 len) |
| 2089 | { |
| 2090 | GV *gv; |
| 2091 | |
| 2092 | if (len == 11 && *pkgname == '_' && strEQ(pkgname, "__PACKAGE__")) |
| 2093 | return PL_curstash; |
| 2094 | |
| 2095 | if (len > 2 && |
| 2096 | (pkgname[len - 2] == ':' && pkgname[len - 1] == ':') && |
| 2097 | (gv = gv_fetchpv(pkgname, FALSE, SVt_PVHV))) |
| 2098 | { |
| 2099 | return GvHV(gv); /* Foo:: */ |
| 2100 | } |
| 2101 | |
| 2102 | /* use constant CLASS => 'MyClass' */ |
| 2103 | if ((gv = gv_fetchpv(pkgname, FALSE, SVt_PVCV))) { |
| 2104 | SV *sv; |
| 2105 | if (GvCV(gv) && (sv = cv_const_sv(GvCV(gv)))) { |
| 2106 | pkgname = SvPV_nolen(sv); |
| 2107 | } |
| 2108 | } |
| 2109 | |
| 2110 | return gv_stashpv(pkgname, FALSE); |
| 2111 | } |
| 2112 | |
| 2113 | #ifdef DEBUGGING |
| 2114 | static char* exp_name[] = |
| 2115 | { "OPERATOR", "TERM", "REF", "STATE", "BLOCK", "ATTRBLOCK", |
| 2116 | "ATTRTERM", "TERMBLOCK" |
| 2117 | }; |
| 2118 | #endif |
| 2119 | |
| 2120 | /* |
| 2121 | yylex |
| 2122 | |
| 2123 | Works out what to call the token just pulled out of the input |
| 2124 | stream. The yacc parser takes care of taking the ops we return and |
| 2125 | stitching them into a tree. |
| 2126 | |
| 2127 | Returns: |
| 2128 | PRIVATEREF |
| 2129 | |
| 2130 | Structure: |
| 2131 | if read an identifier |
| 2132 | if we're in a my declaration |
| 2133 | croak if they tried to say my($foo::bar) |
| 2134 | build the ops for a my() declaration |
| 2135 | if it's an access to a my() variable |
| 2136 | are we in a sort block? |
| 2137 | croak if my($a); $a <=> $b |
| 2138 | build ops for access to a my() variable |
| 2139 | if in a dq string, and they've said @foo and we can't find @foo |
| 2140 | croak |
| 2141 | build ops for a bareword |
| 2142 | if we already built the token before, use it. |
| 2143 | */ |
| 2144 | |
| 2145 | #ifdef USE_PURE_BISON |
| 2146 | int |
| 2147 | Perl_yylex_r(pTHX_ YYSTYPE *lvalp, int *lcharp) |
| 2148 | { |
| 2149 | int r; |
| 2150 | |
| 2151 | yyactlevel++; |
| 2152 | yylval_pointer[yyactlevel] = lvalp; |
| 2153 | yychar_pointer[yyactlevel] = lcharp; |
| 2154 | if (yyactlevel >= YYMAXLEVEL) |
| 2155 | Perl_croak(aTHX_ "panic: YYMAXLEVEL"); |
| 2156 | |
| 2157 | r = Perl_yylex(aTHX); |
| 2158 | |
| 2159 | if (yyactlevel > 0) |
| 2160 | yyactlevel--; |
| 2161 | |
| 2162 | return r; |
| 2163 | } |
| 2164 | #endif |
| 2165 | |
| 2166 | #ifdef __SC__ |
| 2167 | #pragma segment Perl_yylex |
| 2168 | #endif |
| 2169 | int |
| 2170 | Perl_yylex(pTHX) |
| 2171 | { |
| 2172 | register char *s; |
| 2173 | register char *d; |
| 2174 | register I32 tmp; |
| 2175 | STRLEN len; |
| 2176 | GV *gv = Nullgv; |
| 2177 | GV **gvp = 0; |
| 2178 | bool bof = FALSE; |
| 2179 | |
| 2180 | /* check if there's an identifier for us to look at */ |
| 2181 | if (PL_pending_ident) |
| 2182 | return S_pending_ident(aTHX); |
| 2183 | |
| 2184 | /* no identifier pending identification */ |
| 2185 | |
| 2186 | switch (PL_lex_state) { |
| 2187 | #ifdef COMMENTARY |
| 2188 | case LEX_NORMAL: /* Some compilers will produce faster */ |
| 2189 | case LEX_INTERPNORMAL: /* code if we comment these out. */ |
| 2190 | break; |
| 2191 | #endif |
| 2192 | |
| 2193 | /* when we've already built the next token, just pull it out of the queue */ |
| 2194 | case LEX_KNOWNEXT: |
| 2195 | PL_nexttoke--; |
| 2196 | yylval = PL_nextval[PL_nexttoke]; |
| 2197 | if (!PL_nexttoke) { |
| 2198 | PL_lex_state = PL_lex_defer; |
| 2199 | PL_expect = PL_lex_expect; |
| 2200 | PL_lex_defer = LEX_NORMAL; |
| 2201 | } |
| 2202 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
| 2203 | "### Next token after '%s' was known, type %"IVdf"\n", PL_bufptr, |
| 2204 | (IV)PL_nexttype[PL_nexttoke]); }); |
| 2205 | |
| 2206 | return(PL_nexttype[PL_nexttoke]); |
| 2207 | |
| 2208 | /* interpolated case modifiers like \L \U, including \Q and \E. |
| 2209 | when we get here, PL_bufptr is at the \ |
| 2210 | */ |
| 2211 | case LEX_INTERPCASEMOD: |
| 2212 | #ifdef DEBUGGING |
| 2213 | if (PL_bufptr != PL_bufend && *PL_bufptr != '\\') |
| 2214 | Perl_croak(aTHX_ "panic: INTERPCASEMOD"); |
| 2215 | #endif |
| 2216 | /* handle \E or end of string */ |
| 2217 | if (PL_bufptr == PL_bufend || PL_bufptr[1] == 'E') { |
| 2218 | char oldmod; |
| 2219 | |
| 2220 | /* if at a \E */ |
| 2221 | if (PL_lex_casemods) { |
| 2222 | oldmod = PL_lex_casestack[--PL_lex_casemods]; |
| 2223 | PL_lex_casestack[PL_lex_casemods] = '\0'; |
| 2224 | |
| 2225 | if (PL_bufptr != PL_bufend && strchr("LUQ", oldmod)) { |
| 2226 | PL_bufptr += 2; |
| 2227 | PL_lex_state = LEX_INTERPCONCAT; |
| 2228 | } |
| 2229 | return ')'; |
| 2230 | } |
| 2231 | if (PL_bufptr != PL_bufend) |
| 2232 | PL_bufptr += 2; |
| 2233 | PL_lex_state = LEX_INTERPCONCAT; |
| 2234 | return yylex(); |
| 2235 | } |
| 2236 | else { |
| 2237 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
| 2238 | "### Saw case modifier at '%s'\n", PL_bufptr); }); |
| 2239 | s = PL_bufptr + 1; |
| 2240 | if (strnEQ(s, "L\\u", 3) || strnEQ(s, "U\\l", 3)) |
| 2241 | tmp = *s, *s = s[2], s[2] = tmp; /* misordered... */ |
| 2242 | if (strchr("LU", *s) && |
| 2243 | (strchr(PL_lex_casestack, 'L') || strchr(PL_lex_casestack, 'U'))) |
| 2244 | { |
| 2245 | PL_lex_casestack[--PL_lex_casemods] = '\0'; |
| 2246 | return ')'; |
| 2247 | } |
| 2248 | if (PL_lex_casemods > 10) { |
| 2249 | char* newlb = Renew(PL_lex_casestack, PL_lex_casemods + 2, char); |
| 2250 | if (newlb != PL_lex_casestack) { |
| 2251 | SAVEFREEPV(newlb); |
| 2252 | PL_lex_casestack = newlb; |
| 2253 | } |
| 2254 | } |
| 2255 | PL_lex_casestack[PL_lex_casemods++] = *s; |
| 2256 | PL_lex_casestack[PL_lex_casemods] = '\0'; |
| 2257 | PL_lex_state = LEX_INTERPCONCAT; |
| 2258 | PL_nextval[PL_nexttoke].ival = 0; |
| 2259 | force_next('('); |
| 2260 | if (*s == 'l') |
| 2261 | PL_nextval[PL_nexttoke].ival = OP_LCFIRST; |
| 2262 | else if (*s == 'u') |
| 2263 | PL_nextval[PL_nexttoke].ival = OP_UCFIRST; |
| 2264 | else if (*s == 'L') |
| 2265 | PL_nextval[PL_nexttoke].ival = OP_LC; |
| 2266 | else if (*s == 'U') |
| 2267 | PL_nextval[PL_nexttoke].ival = OP_UC; |
| 2268 | else if (*s == 'Q') |
| 2269 | PL_nextval[PL_nexttoke].ival = OP_QUOTEMETA; |
| 2270 | else |
| 2271 | Perl_croak(aTHX_ "panic: yylex"); |
| 2272 | PL_bufptr = s + 1; |
| 2273 | force_next(FUNC); |
| 2274 | if (PL_lex_starts) { |
| 2275 | s = PL_bufptr; |
| 2276 | PL_lex_starts = 0; |
| 2277 | Aop(OP_CONCAT); |
| 2278 | } |
| 2279 | else |
| 2280 | return yylex(); |
| 2281 | } |
| 2282 | |
| 2283 | case LEX_INTERPPUSH: |
| 2284 | return sublex_push(); |
| 2285 | |
| 2286 | case LEX_INTERPSTART: |
| 2287 | if (PL_bufptr == PL_bufend) |
| 2288 | return sublex_done(); |
| 2289 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
| 2290 | "### Interpolated variable at '%s'\n", PL_bufptr); }); |
| 2291 | PL_expect = XTERM; |
| 2292 | PL_lex_dojoin = (*PL_bufptr == '@'); |
| 2293 | PL_lex_state = LEX_INTERPNORMAL; |
| 2294 | if (PL_lex_dojoin) { |
| 2295 | PL_nextval[PL_nexttoke].ival = 0; |
| 2296 | force_next(','); |
| 2297 | #ifdef USE_5005THREADS |
| 2298 | PL_nextval[PL_nexttoke].opval = newOP(OP_THREADSV, 0); |
| 2299 | PL_nextval[PL_nexttoke].opval->op_targ = find_threadsv("\""); |
| 2300 | force_next(PRIVATEREF); |
| 2301 | #else |
| 2302 | force_ident("\"", '$'); |
| 2303 | #endif /* USE_5005THREADS */ |
| 2304 | PL_nextval[PL_nexttoke].ival = 0; |
| 2305 | force_next('$'); |
| 2306 | PL_nextval[PL_nexttoke].ival = 0; |
| 2307 | force_next('('); |
| 2308 | PL_nextval[PL_nexttoke].ival = OP_JOIN; /* emulate join($", ...) */ |
| 2309 | force_next(FUNC); |
| 2310 | } |
| 2311 | if (PL_lex_starts++) { |
| 2312 | s = PL_bufptr; |
| 2313 | Aop(OP_CONCAT); |
| 2314 | } |
| 2315 | return yylex(); |
| 2316 | |
| 2317 | case LEX_INTERPENDMAYBE: |
| 2318 | if (intuit_more(PL_bufptr)) { |
| 2319 | PL_lex_state = LEX_INTERPNORMAL; /* false alarm, more expr */ |
| 2320 | break; |
| 2321 | } |
| 2322 | /* FALL THROUGH */ |
| 2323 | |
| 2324 | case LEX_INTERPEND: |
| 2325 | if (PL_lex_dojoin) { |
| 2326 | PL_lex_dojoin = FALSE; |
| 2327 | PL_lex_state = LEX_INTERPCONCAT; |
| 2328 | return ')'; |
| 2329 | } |
| 2330 | if (PL_lex_inwhat == OP_SUBST && PL_linestr == PL_lex_repl |
| 2331 | && SvEVALED(PL_lex_repl)) |
| 2332 | { |
| 2333 | if (PL_bufptr != PL_bufend) |
| 2334 | Perl_croak(aTHX_ "Bad evalled substitution pattern"); |
| 2335 | PL_lex_repl = Nullsv; |
| 2336 | } |
| 2337 | /* FALLTHROUGH */ |
| 2338 | case LEX_INTERPCONCAT: |
| 2339 | #ifdef DEBUGGING |
| 2340 | if (PL_lex_brackets) |
| 2341 | Perl_croak(aTHX_ "panic: INTERPCONCAT"); |
| 2342 | #endif |
| 2343 | if (PL_bufptr == PL_bufend) |
| 2344 | return sublex_done(); |
| 2345 | |
| 2346 | if (SvIVX(PL_linestr) == '\'') { |
| 2347 | SV *sv = newSVsv(PL_linestr); |
| 2348 | if (!PL_lex_inpat) |
| 2349 | sv = tokeq(sv); |
| 2350 | else if ( PL_hints & HINT_NEW_RE ) |
| 2351 | sv = new_constant(NULL, 0, "qr", sv, sv, "q"); |
| 2352 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
| 2353 | s = PL_bufend; |
| 2354 | } |
| 2355 | else { |
| 2356 | s = scan_const(PL_bufptr); |
| 2357 | if (*s == '\\') |
| 2358 | PL_lex_state = LEX_INTERPCASEMOD; |
| 2359 | else |
| 2360 | PL_lex_state = LEX_INTERPSTART; |
| 2361 | } |
| 2362 | |
| 2363 | if (s != PL_bufptr) { |
| 2364 | PL_nextval[PL_nexttoke] = yylval; |
| 2365 | PL_expect = XTERM; |
| 2366 | force_next(THING); |
| 2367 | if (PL_lex_starts++) |
| 2368 | Aop(OP_CONCAT); |
| 2369 | else { |
| 2370 | PL_bufptr = s; |
| 2371 | return yylex(); |
| 2372 | } |
| 2373 | } |
| 2374 | |
| 2375 | return yylex(); |
| 2376 | case LEX_FORMLINE: |
| 2377 | PL_lex_state = LEX_NORMAL; |
| 2378 | s = scan_formline(PL_bufptr); |
| 2379 | if (!PL_lex_formbrack) |
| 2380 | goto rightbracket; |
| 2381 | OPERATOR(';'); |
| 2382 | } |
| 2383 | |
| 2384 | s = PL_bufptr; |
| 2385 | PL_oldoldbufptr = PL_oldbufptr; |
| 2386 | PL_oldbufptr = s; |
| 2387 | DEBUG_T( { |
| 2388 | PerlIO_printf(Perl_debug_log, "### Tokener expecting %s at %s\n", |
| 2389 | exp_name[PL_expect], s); |
| 2390 | } ); |
| 2391 | |
| 2392 | retry: |
| 2393 | switch (*s) { |
| 2394 | default: |
| 2395 | if (isIDFIRST_lazy_if(s,UTF)) |
| 2396 | goto keylookup; |
| 2397 | Perl_croak(aTHX_ "Unrecognized character \\x%02X", *s & 255); |
| 2398 | case 4: |
| 2399 | case 26: |
| 2400 | goto fake_eof; /* emulate EOF on ^D or ^Z */ |
| 2401 | case 0: |
| 2402 | if (!PL_rsfp) { |
| 2403 | PL_last_uni = 0; |
| 2404 | PL_last_lop = 0; |
| 2405 | if (PL_lex_brackets) |
| 2406 | yyerror("Missing right curly or square bracket"); |
| 2407 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 2408 | "### Tokener got EOF\n"); |
| 2409 | } ); |
| 2410 | TOKEN(0); |
| 2411 | } |
| 2412 | if (s++ < PL_bufend) |
| 2413 | goto retry; /* ignore stray nulls */ |
| 2414 | PL_last_uni = 0; |
| 2415 | PL_last_lop = 0; |
| 2416 | if (!PL_in_eval && !PL_preambled) { |
| 2417 | PL_preambled = TRUE; |
| 2418 | sv_setpv(PL_linestr,incl_perldb()); |
| 2419 | if (SvCUR(PL_linestr)) |
| 2420 | sv_catpv(PL_linestr,";"); |
| 2421 | if (PL_preambleav){ |
| 2422 | while(AvFILLp(PL_preambleav) >= 0) { |
| 2423 | SV *tmpsv = av_shift(PL_preambleav); |
| 2424 | sv_catsv(PL_linestr, tmpsv); |
| 2425 | sv_catpv(PL_linestr, ";"); |
| 2426 | sv_free(tmpsv); |
| 2427 | } |
| 2428 | sv_free((SV*)PL_preambleav); |
| 2429 | PL_preambleav = NULL; |
| 2430 | } |
| 2431 | if (PL_minus_n || PL_minus_p) { |
| 2432 | sv_catpv(PL_linestr, "LINE: while (<>) {"); |
| 2433 | if (PL_minus_l) |
| 2434 | sv_catpv(PL_linestr,"chomp;"); |
| 2435 | if (PL_minus_a) { |
| 2436 | if (PL_minus_F) { |
| 2437 | if (strchr("/'\"", *PL_splitstr) |
| 2438 | && strchr(PL_splitstr + 1, *PL_splitstr)) |
| 2439 | Perl_sv_catpvf(aTHX_ PL_linestr, "@F=split(%s);", PL_splitstr); |
| 2440 | else { |
| 2441 | char delim; |
| 2442 | s = "'~#\200\1'"; /* surely one char is unused...*/ |
| 2443 | while (s[1] && strchr(PL_splitstr, *s)) s++; |
| 2444 | delim = *s; |
| 2445 | Perl_sv_catpvf(aTHX_ PL_linestr, "our @F=split(%s%c", |
| 2446 | "q" + (delim == '\''), delim); |
| 2447 | for (s = PL_splitstr; *s; s++) { |
| 2448 | if (*s == '\\') |
| 2449 | sv_catpvn(PL_linestr, "\\", 1); |
| 2450 | sv_catpvn(PL_linestr, s, 1); |
| 2451 | } |
| 2452 | Perl_sv_catpvf(aTHX_ PL_linestr, "%c);", delim); |
| 2453 | } |
| 2454 | } |
| 2455 | else |
| 2456 | sv_catpv(PL_linestr,"our @F=split(' ');"); |
| 2457 | } |
| 2458 | } |
| 2459 | sv_catpv(PL_linestr, "\n"); |
| 2460 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); |
| 2461 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 2462 | PL_last_lop = PL_last_uni = Nullch; |
| 2463 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| 2464 | SV *sv = NEWSV(85,0); |
| 2465 | |
| 2466 | sv_upgrade(sv, SVt_PVMG); |
| 2467 | sv_setsv(sv,PL_linestr); |
| 2468 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
| 2469 | } |
| 2470 | goto retry; |
| 2471 | } |
| 2472 | do { |
| 2473 | bof = PL_rsfp ? TRUE : FALSE; |
| 2474 | if ((s = filter_gets(PL_linestr, PL_rsfp, 0)) == Nullch) { |
| 2475 | fake_eof: |
| 2476 | if (PL_rsfp) { |
| 2477 | if (PL_preprocess && !PL_in_eval) |
| 2478 | (void)PerlProc_pclose(PL_rsfp); |
| 2479 | else if ((PerlIO *)PL_rsfp == PerlIO_stdin()) |
| 2480 | PerlIO_clearerr(PL_rsfp); |
| 2481 | else |
| 2482 | (void)PerlIO_close(PL_rsfp); |
| 2483 | PL_rsfp = Nullfp; |
| 2484 | PL_doextract = FALSE; |
| 2485 | } |
| 2486 | if (!PL_in_eval && (PL_minus_n || PL_minus_p)) { |
| 2487 | sv_setpv(PL_linestr,PL_minus_p ? ";}continue{print" : ""); |
| 2488 | sv_catpv(PL_linestr,";}"); |
| 2489 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); |
| 2490 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 2491 | PL_last_lop = PL_last_uni = Nullch; |
| 2492 | PL_minus_n = PL_minus_p = 0; |
| 2493 | goto retry; |
| 2494 | } |
| 2495 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); |
| 2496 | PL_last_lop = PL_last_uni = Nullch; |
| 2497 | sv_setpv(PL_linestr,""); |
| 2498 | TOKEN(';'); /* not infinite loop because rsfp is NULL now */ |
| 2499 | } |
| 2500 | /* if it looks like the start of a BOM, check if it in fact is */ |
| 2501 | else if (bof && (!*s || *(U8*)s == 0xEF || *(U8*)s >= 0xFE)) { |
| 2502 | #ifdef PERLIO_IS_STDIO |
| 2503 | # ifdef __GNU_LIBRARY__ |
| 2504 | # if __GNU_LIBRARY__ == 1 /* Linux glibc5 */ |
| 2505 | # define FTELL_FOR_PIPE_IS_BROKEN |
| 2506 | # endif |
| 2507 | # else |
| 2508 | # ifdef __GLIBC__ |
| 2509 | # if __GLIBC__ == 1 /* maybe some glibc5 release had it like this? */ |
| 2510 | # define FTELL_FOR_PIPE_IS_BROKEN |
| 2511 | # endif |
| 2512 | # endif |
| 2513 | # endif |
| 2514 | #endif |
| 2515 | #ifdef FTELL_FOR_PIPE_IS_BROKEN |
| 2516 | /* This loses the possibility to detect the bof |
| 2517 | * situation on perl -P when the libc5 is being used. |
| 2518 | * Workaround? Maybe attach some extra state to PL_rsfp? |
| 2519 | */ |
| 2520 | if (!PL_preprocess) |
| 2521 | bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr); |
| 2522 | #else |
| 2523 | bof = PerlIO_tell(PL_rsfp) == SvCUR(PL_linestr); |
| 2524 | #endif |
| 2525 | if (bof) { |
| 2526 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 2527 | s = swallow_bom((U8*)s); |
| 2528 | } |
| 2529 | } |
| 2530 | if (PL_doextract) { |
| 2531 | if (*s == '#' && s[1] == '!' && instr(s,"perl")) |
| 2532 | PL_doextract = FALSE; |
| 2533 | |
| 2534 | /* Incest with pod. */ |
| 2535 | if (*s == '=' && strnEQ(s, "=cut", 4)) { |
| 2536 | sv_setpv(PL_linestr, ""); |
| 2537 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); |
| 2538 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 2539 | PL_last_lop = PL_last_uni = Nullch; |
| 2540 | PL_doextract = FALSE; |
| 2541 | } |
| 2542 | } |
| 2543 | incline(s); |
| 2544 | } while (PL_doextract); |
| 2545 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = s; |
| 2546 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| 2547 | SV *sv = NEWSV(85,0); |
| 2548 | |
| 2549 | sv_upgrade(sv, SVt_PVMG); |
| 2550 | sv_setsv(sv,PL_linestr); |
| 2551 | av_store(CopFILEAV(PL_curcop),(I32)CopLINE(PL_curcop),sv); |
| 2552 | } |
| 2553 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 2554 | PL_last_lop = PL_last_uni = Nullch; |
| 2555 | if (CopLINE(PL_curcop) == 1) { |
| 2556 | while (s < PL_bufend && isSPACE(*s)) |
| 2557 | s++; |
| 2558 | if (*s == ':' && s[1] != ':') /* for csh execing sh scripts */ |
| 2559 | s++; |
| 2560 | d = Nullch; |
| 2561 | if (!PL_in_eval) { |
| 2562 | if (*s == '#' && *(s+1) == '!') |
| 2563 | d = s + 2; |
| 2564 | #ifdef ALTERNATE_SHEBANG |
| 2565 | else { |
| 2566 | static char as[] = ALTERNATE_SHEBANG; |
| 2567 | if (*s == as[0] && strnEQ(s, as, sizeof(as) - 1)) |
| 2568 | d = s + (sizeof(as) - 1); |
| 2569 | } |
| 2570 | #endif /* ALTERNATE_SHEBANG */ |
| 2571 | } |
| 2572 | if (d) { |
| 2573 | char *ipath; |
| 2574 | char *ipathend; |
| 2575 | |
| 2576 | while (isSPACE(*d)) |
| 2577 | d++; |
| 2578 | ipath = d; |
| 2579 | while (*d && !isSPACE(*d)) |
| 2580 | d++; |
| 2581 | ipathend = d; |
| 2582 | |
| 2583 | #ifdef ARG_ZERO_IS_SCRIPT |
| 2584 | if (ipathend > ipath) { |
| 2585 | /* |
| 2586 | * HP-UX (at least) sets argv[0] to the script name, |
| 2587 | * which makes $^X incorrect. And Digital UNIX and Linux, |
| 2588 | * at least, set argv[0] to the basename of the Perl |
| 2589 | * interpreter. So, having found "#!", we'll set it right. |
| 2590 | */ |
| 2591 | SV *x = GvSV(gv_fetchpv("\030", TRUE, SVt_PV)); |
| 2592 | assert(SvPOK(x) || SvGMAGICAL(x)); |
| 2593 | if (sv_eq(x, CopFILESV(PL_curcop))) { |
| 2594 | sv_setpvn(x, ipath, ipathend - ipath); |
| 2595 | SvSETMAGIC(x); |
| 2596 | } |
| 2597 | TAINT_NOT; /* $^X is always tainted, but that's OK */ |
| 2598 | } |
| 2599 | #endif /* ARG_ZERO_IS_SCRIPT */ |
| 2600 | |
| 2601 | /* |
| 2602 | * Look for options. |
| 2603 | */ |
| 2604 | d = instr(s,"perl -"); |
| 2605 | if (!d) { |
| 2606 | d = instr(s,"perl"); |
| 2607 | #if defined(DOSISH) |
| 2608 | /* avoid getting into infinite loops when shebang |
| 2609 | * line contains "Perl" rather than "perl" */ |
| 2610 | if (!d) { |
| 2611 | for (d = ipathend-4; d >= ipath; --d) { |
| 2612 | if ((*d == 'p' || *d == 'P') |
| 2613 | && !ibcmp(d, "perl", 4)) |
| 2614 | { |
| 2615 | break; |
| 2616 | } |
| 2617 | } |
| 2618 | if (d < ipath) |
| 2619 | d = Nullch; |
| 2620 | } |
| 2621 | #endif |
| 2622 | } |
| 2623 | #ifdef ALTERNATE_SHEBANG |
| 2624 | /* |
| 2625 | * If the ALTERNATE_SHEBANG on this system starts with a |
| 2626 | * character that can be part of a Perl expression, then if |
| 2627 | * we see it but not "perl", we're probably looking at the |
| 2628 | * start of Perl code, not a request to hand off to some |
| 2629 | * other interpreter. Similarly, if "perl" is there, but |
| 2630 | * not in the first 'word' of the line, we assume the line |
| 2631 | * contains the start of the Perl program. |
| 2632 | */ |
| 2633 | if (d && *s != '#') { |
| 2634 | char *c = ipath; |
| 2635 | while (*c && !strchr("; \t\r\n\f\v#", *c)) |
| 2636 | c++; |
| 2637 | if (c < d) |
| 2638 | d = Nullch; /* "perl" not in first word; ignore */ |
| 2639 | else |
| 2640 | *s = '#'; /* Don't try to parse shebang line */ |
| 2641 | } |
| 2642 | #endif /* ALTERNATE_SHEBANG */ |
| 2643 | #ifndef MACOS_TRADITIONAL |
| 2644 | if (!d && |
| 2645 | *s == '#' && |
| 2646 | ipathend > ipath && |
| 2647 | !PL_minus_c && |
| 2648 | !instr(s,"indir") && |
| 2649 | instr(PL_origargv[0],"perl")) |
| 2650 | { |
| 2651 | char **newargv; |
| 2652 | |
| 2653 | *ipathend = '\0'; |
| 2654 | s = ipathend + 1; |
| 2655 | while (s < PL_bufend && isSPACE(*s)) |
| 2656 | s++; |
| 2657 | if (s < PL_bufend) { |
| 2658 | Newz(899,newargv,PL_origargc+3,char*); |
| 2659 | newargv[1] = s; |
| 2660 | while (s < PL_bufend && !isSPACE(*s)) |
| 2661 | s++; |
| 2662 | *s = '\0'; |
| 2663 | Copy(PL_origargv+1, newargv+2, PL_origargc+1, char*); |
| 2664 | } |
| 2665 | else |
| 2666 | newargv = PL_origargv; |
| 2667 | newargv[0] = ipath; |
| 2668 | PerlProc_execv(ipath, EXEC_ARGV_CAST(newargv)); |
| 2669 | Perl_croak(aTHX_ "Can't exec %s", ipath); |
| 2670 | } |
| 2671 | #endif |
| 2672 | if (d) { |
| 2673 | U32 oldpdb = PL_perldb; |
| 2674 | bool oldn = PL_minus_n; |
| 2675 | bool oldp = PL_minus_p; |
| 2676 | |
| 2677 | while (*d && !isSPACE(*d)) d++; |
| 2678 | while (SPACE_OR_TAB(*d)) d++; |
| 2679 | |
| 2680 | if (*d++ == '-') { |
| 2681 | do { |
| 2682 | if (*d == 'M' || *d == 'm') { |
| 2683 | char *m = d; |
| 2684 | while (*d && !isSPACE(*d)) d++; |
| 2685 | Perl_croak(aTHX_ "Too late for \"-%.*s\" option", |
| 2686 | (int)(d - m), m); |
| 2687 | } |
| 2688 | d = moreswitches(d); |
| 2689 | } while (d); |
| 2690 | if ((PERLDB_LINE && !oldpdb) || |
| 2691 | ((PL_minus_n || PL_minus_p) && !(oldn || oldp))) |
| 2692 | /* if we have already added "LINE: while (<>) {", |
| 2693 | we must not do it again */ |
| 2694 | { |
| 2695 | sv_setpv(PL_linestr, ""); |
| 2696 | PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = SvPVX(PL_linestr); |
| 2697 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 2698 | PL_last_lop = PL_last_uni = Nullch; |
| 2699 | PL_preambled = FALSE; |
| 2700 | if (PERLDB_LINE) |
| 2701 | (void)gv_fetchfile(PL_origfilename); |
| 2702 | goto retry; |
| 2703 | } |
| 2704 | } |
| 2705 | } |
| 2706 | } |
| 2707 | } |
| 2708 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
| 2709 | PL_bufptr = s; |
| 2710 | PL_lex_state = LEX_FORMLINE; |
| 2711 | return yylex(); |
| 2712 | } |
| 2713 | goto retry; |
| 2714 | case '\r': |
| 2715 | #ifdef PERL_STRICT_CR |
| 2716 | Perl_warn(aTHX_ "Illegal character \\%03o (carriage return)", '\r'); |
| 2717 | Perl_croak(aTHX_ |
| 2718 | "\t(Maybe you didn't strip carriage returns after a network transfer?)\n"); |
| 2719 | #endif |
| 2720 | case ' ': case '\t': case '\f': case 013: |
| 2721 | #ifdef MACOS_TRADITIONAL |
| 2722 | case '\312': |
| 2723 | #endif |
| 2724 | s++; |
| 2725 | goto retry; |
| 2726 | case '#': |
| 2727 | case '\n': |
| 2728 | if (PL_lex_state != LEX_NORMAL || (PL_in_eval && !PL_rsfp)) { |
| 2729 | if (*s == '#' && s == PL_linestart && PL_in_eval && !PL_rsfp) { |
| 2730 | /* handle eval qq[#line 1 "foo"\n ...] */ |
| 2731 | CopLINE_dec(PL_curcop); |
| 2732 | incline(s); |
| 2733 | } |
| 2734 | d = PL_bufend; |
| 2735 | while (s < d && *s != '\n') |
| 2736 | s++; |
| 2737 | if (s < d) |
| 2738 | s++; |
| 2739 | else if (s > d) /* Found by Ilya: feed random input to Perl. */ |
| 2740 | Perl_croak(aTHX_ "panic: input overflow"); |
| 2741 | incline(s); |
| 2742 | if (PL_lex_formbrack && PL_lex_brackets <= PL_lex_formbrack) { |
| 2743 | PL_bufptr = s; |
| 2744 | PL_lex_state = LEX_FORMLINE; |
| 2745 | return yylex(); |
| 2746 | } |
| 2747 | } |
| 2748 | else { |
| 2749 | *s = '\0'; |
| 2750 | PL_bufend = s; |
| 2751 | } |
| 2752 | goto retry; |
| 2753 | case '-': |
| 2754 | if (s[1] && isALPHA(s[1]) && !isALNUM(s[2])) { |
| 2755 | I32 ftst = 0; |
| 2756 | |
| 2757 | s++; |
| 2758 | PL_bufptr = s; |
| 2759 | tmp = *s++; |
| 2760 | |
| 2761 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
| 2762 | s++; |
| 2763 | |
| 2764 | if (strnEQ(s,"=>",2)) { |
| 2765 | s = force_word(PL_bufptr,WORD,FALSE,FALSE,FALSE); |
| 2766 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 2767 | "### Saw unary minus before =>, forcing word '%s'\n", s); |
| 2768 | } ); |
| 2769 | OPERATOR('-'); /* unary minus */ |
| 2770 | } |
| 2771 | PL_last_uni = PL_oldbufptr; |
| 2772 | switch (tmp) { |
| 2773 | case 'r': ftst = OP_FTEREAD; break; |
| 2774 | case 'w': ftst = OP_FTEWRITE; break; |
| 2775 | case 'x': ftst = OP_FTEEXEC; break; |
| 2776 | case 'o': ftst = OP_FTEOWNED; break; |
| 2777 | case 'R': ftst = OP_FTRREAD; break; |
| 2778 | case 'W': ftst = OP_FTRWRITE; break; |
| 2779 | case 'X': ftst = OP_FTREXEC; break; |
| 2780 | case 'O': ftst = OP_FTROWNED; break; |
| 2781 | case 'e': ftst = OP_FTIS; break; |
| 2782 | case 'z': ftst = OP_FTZERO; break; |
| 2783 | case 's': ftst = OP_FTSIZE; break; |
| 2784 | case 'f': ftst = OP_FTFILE; break; |
| 2785 | case 'd': ftst = OP_FTDIR; break; |
| 2786 | case 'l': ftst = OP_FTLINK; break; |
| 2787 | case 'p': ftst = OP_FTPIPE; break; |
| 2788 | case 'S': ftst = OP_FTSOCK; break; |
| 2789 | case 'u': ftst = OP_FTSUID; break; |
| 2790 | case 'g': ftst = OP_FTSGID; break; |
| 2791 | case 'k': ftst = OP_FTSVTX; break; |
| 2792 | case 'b': ftst = OP_FTBLK; break; |
| 2793 | case 'c': ftst = OP_FTCHR; break; |
| 2794 | case 't': ftst = OP_FTTTY; break; |
| 2795 | case 'T': ftst = OP_FTTEXT; break; |
| 2796 | case 'B': ftst = OP_FTBINARY; break; |
| 2797 | case 'M': case 'A': case 'C': |
| 2798 | gv_fetchpv("\024",TRUE, SVt_PV); |
| 2799 | switch (tmp) { |
| 2800 | case 'M': ftst = OP_FTMTIME; break; |
| 2801 | case 'A': ftst = OP_FTATIME; break; |
| 2802 | case 'C': ftst = OP_FTCTIME; break; |
| 2803 | default: break; |
| 2804 | } |
| 2805 | break; |
| 2806 | default: |
| 2807 | break; |
| 2808 | } |
| 2809 | if (ftst) { |
| 2810 | PL_last_lop_op = ftst; |
| 2811 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 2812 | "### Saw file test %c\n", (int)ftst); |
| 2813 | } ); |
| 2814 | FTST(ftst); |
| 2815 | } |
| 2816 | else { |
| 2817 | /* Assume it was a minus followed by a one-letter named |
| 2818 | * subroutine call (or a -bareword), then. */ |
| 2819 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 2820 | "### %c looked like a file test but was not\n", |
| 2821 | (int)ftst); |
| 2822 | } ); |
| 2823 | s -= 2; |
| 2824 | } |
| 2825 | } |
| 2826 | tmp = *s++; |
| 2827 | if (*s == tmp) { |
| 2828 | s++; |
| 2829 | if (PL_expect == XOPERATOR) |
| 2830 | TERM(POSTDEC); |
| 2831 | else |
| 2832 | OPERATOR(PREDEC); |
| 2833 | } |
| 2834 | else if (*s == '>') { |
| 2835 | s++; |
| 2836 | s = skipspace(s); |
| 2837 | if (isIDFIRST_lazy_if(s,UTF)) { |
| 2838 | s = force_word(s,METHOD,FALSE,TRUE,FALSE); |
| 2839 | TOKEN(ARROW); |
| 2840 | } |
| 2841 | else if (*s == '$') |
| 2842 | OPERATOR(ARROW); |
| 2843 | else |
| 2844 | TERM(ARROW); |
| 2845 | } |
| 2846 | if (PL_expect == XOPERATOR) |
| 2847 | Aop(OP_SUBTRACT); |
| 2848 | else { |
| 2849 | if (isSPACE(*s) || !isSPACE(*PL_bufptr)) |
| 2850 | check_uni(); |
| 2851 | OPERATOR('-'); /* unary minus */ |
| 2852 | } |
| 2853 | |
| 2854 | case '+': |
| 2855 | tmp = *s++; |
| 2856 | if (*s == tmp) { |
| 2857 | s++; |
| 2858 | if (PL_expect == XOPERATOR) |
| 2859 | TERM(POSTINC); |
| 2860 | else |
| 2861 | OPERATOR(PREINC); |
| 2862 | } |
| 2863 | if (PL_expect == XOPERATOR) |
| 2864 | Aop(OP_ADD); |
| 2865 | else { |
| 2866 | if (isSPACE(*s) || !isSPACE(*PL_bufptr)) |
| 2867 | check_uni(); |
| 2868 | OPERATOR('+'); |
| 2869 | } |
| 2870 | |
| 2871 | case '*': |
| 2872 | if (PL_expect != XOPERATOR) { |
| 2873 | s = scan_ident(s, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE); |
| 2874 | PL_expect = XOPERATOR; |
| 2875 | force_ident(PL_tokenbuf, '*'); |
| 2876 | if (!*PL_tokenbuf) |
| 2877 | PREREF('*'); |
| 2878 | TERM('*'); |
| 2879 | } |
| 2880 | s++; |
| 2881 | if (*s == '*') { |
| 2882 | s++; |
| 2883 | PWop(OP_POW); |
| 2884 | } |
| 2885 | Mop(OP_MULTIPLY); |
| 2886 | |
| 2887 | case '%': |
| 2888 | if (PL_expect == XOPERATOR) { |
| 2889 | ++s; |
| 2890 | Mop(OP_MODULO); |
| 2891 | } |
| 2892 | PL_tokenbuf[0] = '%'; |
| 2893 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, TRUE); |
| 2894 | if (!PL_tokenbuf[1]) { |
| 2895 | if (s == PL_bufend) |
| 2896 | yyerror("Final % should be \\% or %name"); |
| 2897 | PREREF('%'); |
| 2898 | } |
| 2899 | PL_pending_ident = '%'; |
| 2900 | TERM('%'); |
| 2901 | |
| 2902 | case '^': |
| 2903 | s++; |
| 2904 | BOop(OP_BIT_XOR); |
| 2905 | case '[': |
| 2906 | PL_lex_brackets++; |
| 2907 | /* FALL THROUGH */ |
| 2908 | case '~': |
| 2909 | case ',': |
| 2910 | tmp = *s++; |
| 2911 | OPERATOR(tmp); |
| 2912 | case ':': |
| 2913 | if (s[1] == ':') { |
| 2914 | len = 0; |
| 2915 | goto just_a_word; |
| 2916 | } |
| 2917 | s++; |
| 2918 | switch (PL_expect) { |
| 2919 | OP *attrs; |
| 2920 | case XOPERATOR: |
| 2921 | if (!PL_in_my || PL_lex_state != LEX_NORMAL) |
| 2922 | break; |
| 2923 | PL_bufptr = s; /* update in case we back off */ |
| 2924 | goto grabattrs; |
| 2925 | case XATTRBLOCK: |
| 2926 | PL_expect = XBLOCK; |
| 2927 | goto grabattrs; |
| 2928 | case XATTRTERM: |
| 2929 | PL_expect = XTERMBLOCK; |
| 2930 | grabattrs: |
| 2931 | s = skipspace(s); |
| 2932 | attrs = Nullop; |
| 2933 | while (isIDFIRST_lazy_if(s,UTF)) { |
| 2934 | d = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); |
| 2935 | if (isLOWER(*s) && (tmp = keyword(PL_tokenbuf, len))) { |
| 2936 | if (tmp < 0) tmp = -tmp; |
| 2937 | switch (tmp) { |
| 2938 | case KEY_or: |
| 2939 | case KEY_and: |
| 2940 | case KEY_for: |
| 2941 | case KEY_unless: |
| 2942 | case KEY_if: |
| 2943 | case KEY_while: |
| 2944 | case KEY_until: |
| 2945 | goto got_attrs; |
| 2946 | default: |
| 2947 | break; |
| 2948 | } |
| 2949 | } |
| 2950 | if (*d == '(') { |
| 2951 | d = scan_str(d,TRUE,TRUE); |
| 2952 | if (!d) { |
| 2953 | /* MUST advance bufptr here to avoid bogus |
| 2954 | "at end of line" context messages from yyerror(). |
| 2955 | */ |
| 2956 | PL_bufptr = s + len; |
| 2957 | yyerror("Unterminated attribute parameter in attribute list"); |
| 2958 | if (attrs) |
| 2959 | op_free(attrs); |
| 2960 | return 0; /* EOF indicator */ |
| 2961 | } |
| 2962 | } |
| 2963 | if (PL_lex_stuff) { |
| 2964 | SV *sv = newSVpvn(s, len); |
| 2965 | sv_catsv(sv, PL_lex_stuff); |
| 2966 | attrs = append_elem(OP_LIST, attrs, |
| 2967 | newSVOP(OP_CONST, 0, sv)); |
| 2968 | SvREFCNT_dec(PL_lex_stuff); |
| 2969 | PL_lex_stuff = Nullsv; |
| 2970 | } |
| 2971 | else { |
| 2972 | if (!PL_in_my && len == 6 && strnEQ(s, "lvalue", len)) |
| 2973 | CvLVALUE_on(PL_compcv); |
| 2974 | else if (!PL_in_my && len == 6 && strnEQ(s, "locked", len)) |
| 2975 | CvLOCKED_on(PL_compcv); |
| 2976 | else if (!PL_in_my && len == 6 && strnEQ(s, "method", len)) |
| 2977 | CvMETHOD_on(PL_compcv); |
| 2978 | #ifdef USE_ITHREADS |
| 2979 | else if (PL_in_my == KEY_our && len == 6 && strnEQ(s, "unique", len)) |
| 2980 | GvUNIQUE_on(cGVOPx_gv(yylval.opval)); |
| 2981 | #endif |
| 2982 | /* After we've set the flags, it could be argued that |
| 2983 | we don't need to do the attributes.pm-based setting |
| 2984 | process, and shouldn't bother appending recognized |
| 2985 | flags. To experiment with that, uncomment the |
| 2986 | following "else": */ |
| 2987 | else |
| 2988 | attrs = append_elem(OP_LIST, attrs, |
| 2989 | newSVOP(OP_CONST, 0, |
| 2990 | newSVpvn(s, len))); |
| 2991 | } |
| 2992 | s = skipspace(d); |
| 2993 | if (*s == ':' && s[1] != ':') |
| 2994 | s = skipspace(s+1); |
| 2995 | else if (s == d) |
| 2996 | break; /* require real whitespace or :'s */ |
| 2997 | } |
| 2998 | tmp = (PL_expect == XOPERATOR ? '=' : '{'); /*'}(' for vi */ |
| 2999 | if (*s != ';' && *s != tmp && (tmp != '=' || *s != ')')) { |
| 3000 | char q = ((*s == '\'') ? '"' : '\''); |
| 3001 | /* If here for an expression, and parsed no attrs, back off. */ |
| 3002 | if (tmp == '=' && !attrs) { |
| 3003 | s = PL_bufptr; |
| 3004 | break; |
| 3005 | } |
| 3006 | /* MUST advance bufptr here to avoid bogus "at end of line" |
| 3007 | context messages from yyerror(). |
| 3008 | */ |
| 3009 | PL_bufptr = s; |
| 3010 | if (!*s) |
| 3011 | yyerror("Unterminated attribute list"); |
| 3012 | else |
| 3013 | yyerror(Perl_form(aTHX_ "Invalid separator character %c%c%c in attribute list", |
| 3014 | q, *s, q)); |
| 3015 | if (attrs) |
| 3016 | op_free(attrs); |
| 3017 | OPERATOR(':'); |
| 3018 | } |
| 3019 | got_attrs: |
| 3020 | if (attrs) { |
| 3021 | PL_nextval[PL_nexttoke].opval = attrs; |
| 3022 | force_next(THING); |
| 3023 | } |
| 3024 | TOKEN(COLONATTR); |
| 3025 | } |
| 3026 | OPERATOR(':'); |
| 3027 | case '(': |
| 3028 | s++; |
| 3029 | if (PL_last_lop == PL_oldoldbufptr || PL_last_uni == PL_oldoldbufptr) |
| 3030 | PL_oldbufptr = PL_oldoldbufptr; /* allow print(STDOUT 123) */ |
| 3031 | else |
| 3032 | PL_expect = XTERM; |
| 3033 | TOKEN('('); |
| 3034 | case ';': |
| 3035 | CLINE; |
| 3036 | tmp = *s++; |
| 3037 | OPERATOR(tmp); |
| 3038 | case ')': |
| 3039 | tmp = *s++; |
| 3040 | s = skipspace(s); |
| 3041 | if (*s == '{') |
| 3042 | PREBLOCK(tmp); |
| 3043 | TERM(tmp); |
| 3044 | case ']': |
| 3045 | s++; |
| 3046 | if (PL_lex_brackets <= 0) |
| 3047 | yyerror("Unmatched right square bracket"); |
| 3048 | else |
| 3049 | --PL_lex_brackets; |
| 3050 | if (PL_lex_state == LEX_INTERPNORMAL) { |
| 3051 | if (PL_lex_brackets == 0) { |
| 3052 | if (*s != '[' && *s != '{' && (*s != '-' || s[1] != '>')) |
| 3053 | PL_lex_state = LEX_INTERPEND; |
| 3054 | } |
| 3055 | } |
| 3056 | TERM(']'); |
| 3057 | case '{': |
| 3058 | leftbracket: |
| 3059 | s++; |
| 3060 | if (PL_lex_brackets > 100) { |
| 3061 | char* newlb = Renew(PL_lex_brackstack, PL_lex_brackets + 1, char); |
| 3062 | if (newlb != PL_lex_brackstack) { |
| 3063 | SAVEFREEPV(newlb); |
| 3064 | PL_lex_brackstack = newlb; |
| 3065 | } |
| 3066 | } |
| 3067 | switch (PL_expect) { |
| 3068 | case XTERM: |
| 3069 | if (PL_lex_formbrack) { |
| 3070 | s--; |
| 3071 | PRETERMBLOCK(DO); |
| 3072 | } |
| 3073 | if (PL_oldoldbufptr == PL_last_lop) |
| 3074 | PL_lex_brackstack[PL_lex_brackets++] = XTERM; |
| 3075 | else |
| 3076 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
| 3077 | OPERATOR(HASHBRACK); |
| 3078 | case XOPERATOR: |
| 3079 | while (s < PL_bufend && SPACE_OR_TAB(*s)) |
| 3080 | s++; |
| 3081 | d = s; |
| 3082 | PL_tokenbuf[0] = '\0'; |
| 3083 | if (d < PL_bufend && *d == '-') { |
| 3084 | PL_tokenbuf[0] = '-'; |
| 3085 | d++; |
| 3086 | while (d < PL_bufend && SPACE_OR_TAB(*d)) |
| 3087 | d++; |
| 3088 | } |
| 3089 | if (d < PL_bufend && isIDFIRST_lazy_if(d,UTF)) { |
| 3090 | d = scan_word(d, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, |
| 3091 | FALSE, &len); |
| 3092 | while (d < PL_bufend && SPACE_OR_TAB(*d)) |
| 3093 | d++; |
| 3094 | if (*d == '}') { |
| 3095 | char minus = (PL_tokenbuf[0] == '-'); |
| 3096 | s = force_word(s + minus, WORD, FALSE, TRUE, FALSE); |
| 3097 | if (minus) |
| 3098 | force_next('-'); |
| 3099 | } |
| 3100 | } |
| 3101 | /* FALL THROUGH */ |
| 3102 | case XATTRBLOCK: |
| 3103 | case XBLOCK: |
| 3104 | PL_lex_brackstack[PL_lex_brackets++] = XSTATE; |
| 3105 | PL_expect = XSTATE; |
| 3106 | break; |
| 3107 | case XATTRTERM: |
| 3108 | case XTERMBLOCK: |
| 3109 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
| 3110 | PL_expect = XSTATE; |
| 3111 | break; |
| 3112 | default: { |
| 3113 | char *t; |
| 3114 | if (PL_oldoldbufptr == PL_last_lop) |
| 3115 | PL_lex_brackstack[PL_lex_brackets++] = XTERM; |
| 3116 | else |
| 3117 | PL_lex_brackstack[PL_lex_brackets++] = XOPERATOR; |
| 3118 | s = skipspace(s); |
| 3119 | if (*s == '}') { |
| 3120 | if (PL_expect == XREF && PL_lex_state == LEX_INTERPNORMAL) { |
| 3121 | PL_expect = XTERM; |
| 3122 | /* This hack is to get the ${} in the message. */ |
| 3123 | PL_bufptr = s+1; |
| 3124 | yyerror("syntax error"); |
| 3125 | break; |
| 3126 | } |
| 3127 | OPERATOR(HASHBRACK); |
| 3128 | } |
| 3129 | /* This hack serves to disambiguate a pair of curlies |
| 3130 | * as being a block or an anon hash. Normally, expectation |
| 3131 | * determines that, but in cases where we're not in a |
| 3132 | * position to expect anything in particular (like inside |
| 3133 | * eval"") we have to resolve the ambiguity. This code |
| 3134 | * covers the case where the first term in the curlies is a |
| 3135 | * quoted string. Most other cases need to be explicitly |
| 3136 | * disambiguated by prepending a `+' before the opening |
| 3137 | * curly in order to force resolution as an anon hash. |
| 3138 | * |
| 3139 | * XXX should probably propagate the outer expectation |
| 3140 | * into eval"" to rely less on this hack, but that could |
| 3141 | * potentially break current behavior of eval"". |
| 3142 | * GSAR 97-07-21 |
| 3143 | */ |
| 3144 | t = s; |
| 3145 | if (*s == '\'' || *s == '"' || *s == '`') { |
| 3146 | /* common case: get past first string, handling escapes */ |
| 3147 | for (t++; t < PL_bufend && *t != *s;) |
| 3148 | if (*t++ == '\\' && (*t == '\\' || *t == *s)) |
| 3149 | t++; |
| 3150 | t++; |
| 3151 | } |
| 3152 | else if (*s == 'q') { |
| 3153 | if (++t < PL_bufend |
| 3154 | && (!isALNUM(*t) |
| 3155 | || ((*t == 'q' || *t == 'x') && ++t < PL_bufend |
| 3156 | && !isALNUM(*t)))) |
| 3157 | { |
| 3158 | char *tmps; |
| 3159 | char open, close, term; |
| 3160 | I32 brackets = 1; |
| 3161 | |
| 3162 | while (t < PL_bufend && isSPACE(*t)) |
| 3163 | t++; |
| 3164 | term = *t; |
| 3165 | open = term; |
| 3166 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) |
| 3167 | term = tmps[5]; |
| 3168 | close = term; |
| 3169 | if (open == close) |
| 3170 | for (t++; t < PL_bufend; t++) { |
| 3171 | if (*t == '\\' && t+1 < PL_bufend && open != '\\') |
| 3172 | t++; |
| 3173 | else if (*t == open) |
| 3174 | break; |
| 3175 | } |
| 3176 | else |
| 3177 | for (t++; t < PL_bufend; t++) { |
| 3178 | if (*t == '\\' && t+1 < PL_bufend) |
| 3179 | t++; |
| 3180 | else if (*t == close && --brackets <= 0) |
| 3181 | break; |
| 3182 | else if (*t == open) |
| 3183 | brackets++; |
| 3184 | } |
| 3185 | } |
| 3186 | t++; |
| 3187 | } |
| 3188 | else if (isALNUM_lazy_if(t,UTF)) { |
| 3189 | t += UTF8SKIP(t); |
| 3190 | while (t < PL_bufend && isALNUM_lazy_if(t,UTF)) |
| 3191 | t += UTF8SKIP(t); |
| 3192 | } |
| 3193 | while (t < PL_bufend && isSPACE(*t)) |
| 3194 | t++; |
| 3195 | /* if comma follows first term, call it an anon hash */ |
| 3196 | /* XXX it could be a comma expression with loop modifiers */ |
| 3197 | if (t < PL_bufend && ((*t == ',' && (*s == 'q' || !isLOWER(*s))) |
| 3198 | || (*t == '=' && t[1] == '>'))) |
| 3199 | OPERATOR(HASHBRACK); |
| 3200 | if (PL_expect == XREF) |
| 3201 | PL_expect = XTERM; |
| 3202 | else { |
| 3203 | PL_lex_brackstack[PL_lex_brackets-1] = XSTATE; |
| 3204 | PL_expect = XSTATE; |
| 3205 | } |
| 3206 | } |
| 3207 | break; |
| 3208 | } |
| 3209 | yylval.ival = CopLINE(PL_curcop); |
| 3210 | if (isSPACE(*s) || *s == '#') |
| 3211 | PL_copline = NOLINE; /* invalidate current command line number */ |
| 3212 | TOKEN('{'); |
| 3213 | case '}': |
| 3214 | rightbracket: |
| 3215 | s++; |
| 3216 | if (PL_lex_brackets <= 0) |
| 3217 | yyerror("Unmatched right curly bracket"); |
| 3218 | else |
| 3219 | PL_expect = (expectation)PL_lex_brackstack[--PL_lex_brackets]; |
| 3220 | if (PL_lex_brackets < PL_lex_formbrack && PL_lex_state != LEX_INTERPNORMAL) |
| 3221 | PL_lex_formbrack = 0; |
| 3222 | if (PL_lex_state == LEX_INTERPNORMAL) { |
| 3223 | if (PL_lex_brackets == 0) { |
| 3224 | if (PL_expect & XFAKEBRACK) { |
| 3225 | PL_expect &= XENUMMASK; |
| 3226 | PL_lex_state = LEX_INTERPEND; |
| 3227 | PL_bufptr = s; |
| 3228 | return yylex(); /* ignore fake brackets */ |
| 3229 | } |
| 3230 | if (*s == '-' && s[1] == '>') |
| 3231 | PL_lex_state = LEX_INTERPENDMAYBE; |
| 3232 | else if (*s != '[' && *s != '{') |
| 3233 | PL_lex_state = LEX_INTERPEND; |
| 3234 | } |
| 3235 | } |
| 3236 | if (PL_expect & XFAKEBRACK) { |
| 3237 | PL_expect &= XENUMMASK; |
| 3238 | PL_bufptr = s; |
| 3239 | return yylex(); /* ignore fake brackets */ |
| 3240 | } |
| 3241 | force_next('}'); |
| 3242 | TOKEN(';'); |
| 3243 | case '&': |
| 3244 | s++; |
| 3245 | tmp = *s++; |
| 3246 | if (tmp == '&') |
| 3247 | AOPERATOR(ANDAND); |
| 3248 | s--; |
| 3249 | if (PL_expect == XOPERATOR) { |
| 3250 | if (ckWARN(WARN_SEMICOLON) |
| 3251 | && isIDFIRST_lazy_if(s,UTF) && PL_bufptr == PL_linestart) |
| 3252 | { |
| 3253 | CopLINE_dec(PL_curcop); |
| 3254 | Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi); |
| 3255 | CopLINE_inc(PL_curcop); |
| 3256 | } |
| 3257 | BAop(OP_BIT_AND); |
| 3258 | } |
| 3259 | |
| 3260 | s = scan_ident(s - 1, PL_bufend, PL_tokenbuf, sizeof PL_tokenbuf, TRUE); |
| 3261 | if (*PL_tokenbuf) { |
| 3262 | PL_expect = XOPERATOR; |
| 3263 | force_ident(PL_tokenbuf, '&'); |
| 3264 | } |
| 3265 | else |
| 3266 | PREREF('&'); |
| 3267 | yylval.ival = (OPpENTERSUB_AMPER<<8); |
| 3268 | TERM('&'); |
| 3269 | |
| 3270 | case '|': |
| 3271 | s++; |
| 3272 | tmp = *s++; |
| 3273 | if (tmp == '|') |
| 3274 | AOPERATOR(OROR); |
| 3275 | s--; |
| 3276 | BOop(OP_BIT_OR); |
| 3277 | case '=': |
| 3278 | s++; |
| 3279 | tmp = *s++; |
| 3280 | if (tmp == '=') |
| 3281 | Eop(OP_EQ); |
| 3282 | if (tmp == '>') |
| 3283 | OPERATOR(','); |
| 3284 | if (tmp == '~') |
| 3285 | PMop(OP_MATCH); |
| 3286 | if (ckWARN(WARN_SYNTAX) && tmp && isSPACE(*s) && strchr("+-*/%.^&|<",tmp)) |
| 3287 | Perl_warner(aTHX_ WARN_SYNTAX, "Reversed %c= operator",(int)tmp); |
| 3288 | s--; |
| 3289 | if (PL_expect == XSTATE && isALPHA(tmp) && |
| 3290 | (s == PL_linestart+1 || s[-2] == '\n') ) |
| 3291 | { |
| 3292 | if (PL_in_eval && !PL_rsfp) { |
| 3293 | d = PL_bufend; |
| 3294 | while (s < d) { |
| 3295 | if (*s++ == '\n') { |
| 3296 | incline(s); |
| 3297 | if (strnEQ(s,"=cut",4)) { |
| 3298 | s = strchr(s,'\n'); |
| 3299 | if (s) |
| 3300 | s++; |
| 3301 | else |
| 3302 | s = d; |
| 3303 | incline(s); |
| 3304 | goto retry; |
| 3305 | } |
| 3306 | } |
| 3307 | } |
| 3308 | goto retry; |
| 3309 | } |
| 3310 | s = PL_bufend; |
| 3311 | PL_doextract = TRUE; |
| 3312 | goto retry; |
| 3313 | } |
| 3314 | if (PL_lex_brackets < PL_lex_formbrack) { |
| 3315 | char *t; |
| 3316 | #ifdef PERL_STRICT_CR |
| 3317 | for (t = s; SPACE_OR_TAB(*t); t++) ; |
| 3318 | #else |
| 3319 | for (t = s; SPACE_OR_TAB(*t) || *t == '\r'; t++) ; |
| 3320 | #endif |
| 3321 | if (*t == '\n' || *t == '#') { |
| 3322 | s--; |
| 3323 | PL_expect = XBLOCK; |
| 3324 | goto leftbracket; |
| 3325 | } |
| 3326 | } |
| 3327 | yylval.ival = 0; |
| 3328 | OPERATOR(ASSIGNOP); |
| 3329 | case '!': |
| 3330 | s++; |
| 3331 | tmp = *s++; |
| 3332 | if (tmp == '=') |
| 3333 | Eop(OP_NE); |
| 3334 | if (tmp == '~') |
| 3335 | PMop(OP_NOT); |
| 3336 | s--; |
| 3337 | OPERATOR('!'); |
| 3338 | case '<': |
| 3339 | if (PL_expect != XOPERATOR) { |
| 3340 | if (s[1] != '<' && !strchr(s,'>')) |
| 3341 | check_uni(); |
| 3342 | if (s[1] == '<') |
| 3343 | s = scan_heredoc(s); |
| 3344 | else |
| 3345 | s = scan_inputsymbol(s); |
| 3346 | TERM(sublex_start()); |
| 3347 | } |
| 3348 | s++; |
| 3349 | tmp = *s++; |
| 3350 | if (tmp == '<') |
| 3351 | SHop(OP_LEFT_SHIFT); |
| 3352 | if (tmp == '=') { |
| 3353 | tmp = *s++; |
| 3354 | if (tmp == '>') |
| 3355 | Eop(OP_NCMP); |
| 3356 | s--; |
| 3357 | Rop(OP_LE); |
| 3358 | } |
| 3359 | s--; |
| 3360 | Rop(OP_LT); |
| 3361 | case '>': |
| 3362 | s++; |
| 3363 | tmp = *s++; |
| 3364 | if (tmp == '>') |
| 3365 | SHop(OP_RIGHT_SHIFT); |
| 3366 | if (tmp == '=') |
| 3367 | Rop(OP_GE); |
| 3368 | s--; |
| 3369 | Rop(OP_GT); |
| 3370 | |
| 3371 | case '$': |
| 3372 | CLINE; |
| 3373 | |
| 3374 | if (PL_expect == XOPERATOR) { |
| 3375 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { |
| 3376 | PL_expect = XTERM; |
| 3377 | depcom(); |
| 3378 | return ','; /* grandfather non-comma-format format */ |
| 3379 | } |
| 3380 | } |
| 3381 | |
| 3382 | if (s[1] == '#' && (isIDFIRST_lazy_if(s+2,UTF) || strchr("{$:+-", s[2]))) { |
| 3383 | PL_tokenbuf[0] = '@'; |
| 3384 | s = scan_ident(s + 1, PL_bufend, PL_tokenbuf + 1, |
| 3385 | sizeof PL_tokenbuf - 1, FALSE); |
| 3386 | if (PL_expect == XOPERATOR) |
| 3387 | no_op("Array length", s); |
| 3388 | if (!PL_tokenbuf[1]) |
| 3389 | PREREF(DOLSHARP); |
| 3390 | PL_expect = XOPERATOR; |
| 3391 | PL_pending_ident = '#'; |
| 3392 | TOKEN(DOLSHARP); |
| 3393 | } |
| 3394 | |
| 3395 | PL_tokenbuf[0] = '$'; |
| 3396 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, |
| 3397 | sizeof PL_tokenbuf - 1, FALSE); |
| 3398 | if (PL_expect == XOPERATOR) |
| 3399 | no_op("Scalar", s); |
| 3400 | if (!PL_tokenbuf[1]) { |
| 3401 | if (s == PL_bufend) |
| 3402 | yyerror("Final $ should be \\$ or $name"); |
| 3403 | PREREF('$'); |
| 3404 | } |
| 3405 | |
| 3406 | /* This kludge not intended to be bulletproof. */ |
| 3407 | if (PL_tokenbuf[1] == '[' && !PL_tokenbuf[2]) { |
| 3408 | yylval.opval = newSVOP(OP_CONST, 0, |
| 3409 | newSViv(PL_compiling.cop_arybase)); |
| 3410 | yylval.opval->op_private = OPpCONST_ARYBASE; |
| 3411 | TERM(THING); |
| 3412 | } |
| 3413 | |
| 3414 | d = s; |
| 3415 | tmp = (I32)*s; |
| 3416 | if (PL_lex_state == LEX_NORMAL) |
| 3417 | s = skipspace(s); |
| 3418 | |
| 3419 | if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) { |
| 3420 | char *t; |
| 3421 | if (*s == '[') { |
| 3422 | PL_tokenbuf[0] = '@'; |
| 3423 | if (ckWARN(WARN_SYNTAX)) { |
| 3424 | for(t = s + 1; |
| 3425 | isSPACE(*t) || isALNUM_lazy_if(t,UTF) || *t == '$'; |
| 3426 | t++) ; |
| 3427 | if (*t++ == ',') { |
| 3428 | PL_bufptr = skipspace(PL_bufptr); |
| 3429 | while (t < PL_bufend && *t != ']') |
| 3430 | t++; |
| 3431 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 3432 | "Multidimensional syntax %.*s not supported", |
| 3433 | (t - PL_bufptr) + 1, PL_bufptr); |
| 3434 | } |
| 3435 | } |
| 3436 | } |
| 3437 | else if (*s == '{') { |
| 3438 | PL_tokenbuf[0] = '%'; |
| 3439 | if (ckWARN(WARN_SYNTAX) && strEQ(PL_tokenbuf+1, "SIG") && |
| 3440 | (t = strchr(s, '}')) && (t = strchr(t, '='))) |
| 3441 | { |
| 3442 | char tmpbuf[sizeof PL_tokenbuf]; |
| 3443 | STRLEN len; |
| 3444 | for (t++; isSPACE(*t); t++) ; |
| 3445 | if (isIDFIRST_lazy_if(t,UTF)) { |
| 3446 | t = scan_word(t, tmpbuf, sizeof tmpbuf, TRUE, &len); |
| 3447 | for (; isSPACE(*t); t++) ; |
| 3448 | if (*t == ';' && get_cv(tmpbuf, FALSE)) |
| 3449 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 3450 | "You need to quote \"%s\"", tmpbuf); |
| 3451 | } |
| 3452 | } |
| 3453 | } |
| 3454 | } |
| 3455 | |
| 3456 | PL_expect = XOPERATOR; |
| 3457 | if (PL_lex_state == LEX_NORMAL && isSPACE((char)tmp)) { |
| 3458 | bool islop = (PL_last_lop == PL_oldoldbufptr); |
| 3459 | if (!islop || PL_last_lop_op == OP_GREPSTART) |
| 3460 | PL_expect = XOPERATOR; |
| 3461 | else if (strchr("$@\"'`q", *s)) |
| 3462 | PL_expect = XTERM; /* e.g. print $fh "foo" */ |
| 3463 | else if (strchr("&*<%", *s) && isIDFIRST_lazy_if(s+1,UTF)) |
| 3464 | PL_expect = XTERM; /* e.g. print $fh &sub */ |
| 3465 | else if (isIDFIRST_lazy_if(s,UTF)) { |
| 3466 | char tmpbuf[sizeof PL_tokenbuf]; |
| 3467 | scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
| 3468 | if ((tmp = keyword(tmpbuf, len))) { |
| 3469 | /* binary operators exclude handle interpretations */ |
| 3470 | switch (tmp) { |
| 3471 | case -KEY_x: |
| 3472 | case -KEY_eq: |
| 3473 | case -KEY_ne: |
| 3474 | case -KEY_gt: |
| 3475 | case -KEY_lt: |
| 3476 | case -KEY_ge: |
| 3477 | case -KEY_le: |
| 3478 | case -KEY_cmp: |
| 3479 | break; |
| 3480 | default: |
| 3481 | PL_expect = XTERM; /* e.g. print $fh length() */ |
| 3482 | break; |
| 3483 | } |
| 3484 | } |
| 3485 | else { |
| 3486 | GV *gv = gv_fetchpv(tmpbuf, FALSE, SVt_PVCV); |
| 3487 | if (gv && GvCVu(gv)) |
| 3488 | PL_expect = XTERM; /* e.g. print $fh subr() */ |
| 3489 | } |
| 3490 | } |
| 3491 | else if (isDIGIT(*s)) |
| 3492 | PL_expect = XTERM; /* e.g. print $fh 3 */ |
| 3493 | else if (*s == '.' && isDIGIT(s[1])) |
| 3494 | PL_expect = XTERM; /* e.g. print $fh .3 */ |
| 3495 | else if (strchr("/?-+", *s) && !isSPACE(s[1]) && s[1] != '=') |
| 3496 | PL_expect = XTERM; /* e.g. print $fh -1 */ |
| 3497 | else if (*s == '<' && s[1] == '<' && !isSPACE(s[2]) && s[2] != '=') |
| 3498 | PL_expect = XTERM; /* print $fh <<"EOF" */ |
| 3499 | } |
| 3500 | PL_pending_ident = '$'; |
| 3501 | TOKEN('$'); |
| 3502 | |
| 3503 | case '@': |
| 3504 | if (PL_expect == XOPERATOR) |
| 3505 | no_op("Array", s); |
| 3506 | PL_tokenbuf[0] = '@'; |
| 3507 | s = scan_ident(s, PL_bufend, PL_tokenbuf + 1, sizeof PL_tokenbuf - 1, FALSE); |
| 3508 | if (!PL_tokenbuf[1]) { |
| 3509 | if (s == PL_bufend) |
| 3510 | yyerror("Final @ should be \\@ or @name"); |
| 3511 | PREREF('@'); |
| 3512 | } |
| 3513 | if (PL_lex_state == LEX_NORMAL) |
| 3514 | s = skipspace(s); |
| 3515 | if ((PL_expect != XREF || PL_oldoldbufptr == PL_last_lop) && intuit_more(s)) { |
| 3516 | if (*s == '{') |
| 3517 | PL_tokenbuf[0] = '%'; |
| 3518 | |
| 3519 | /* Warn about @ where they meant $. */ |
| 3520 | if (ckWARN(WARN_SYNTAX)) { |
| 3521 | if (*s == '[' || *s == '{') { |
| 3522 | char *t = s + 1; |
| 3523 | while (*t && (isALNUM_lazy_if(t,UTF) || strchr(" \t$#+-'\"", *t))) |
| 3524 | t++; |
| 3525 | if (*t == '}' || *t == ']') { |
| 3526 | t++; |
| 3527 | PL_bufptr = skipspace(PL_bufptr); |
| 3528 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 3529 | "Scalar value %.*s better written as $%.*s", |
| 3530 | t-PL_bufptr, PL_bufptr, t-PL_bufptr-1, PL_bufptr+1); |
| 3531 | } |
| 3532 | } |
| 3533 | } |
| 3534 | } |
| 3535 | PL_pending_ident = '@'; |
| 3536 | TERM('@'); |
| 3537 | |
| 3538 | case '/': /* may either be division or pattern */ |
| 3539 | case '?': /* may either be conditional or pattern */ |
| 3540 | if (PL_expect != XOPERATOR) { |
| 3541 | /* Disable warning on "study /blah/" */ |
| 3542 | if (PL_oldoldbufptr == PL_last_uni |
| 3543 | && (*PL_last_uni != 's' || s - PL_last_uni < 5 |
| 3544 | || memNE(PL_last_uni, "study", 5) |
| 3545 | || isALNUM_lazy_if(PL_last_uni+5,UTF))) |
| 3546 | check_uni(); |
| 3547 | s = scan_pat(s,OP_MATCH); |
| 3548 | TERM(sublex_start()); |
| 3549 | } |
| 3550 | tmp = *s++; |
| 3551 | if (tmp == '/') |
| 3552 | Mop(OP_DIVIDE); |
| 3553 | OPERATOR(tmp); |
| 3554 | |
| 3555 | case '.': |
| 3556 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack |
| 3557 | #ifdef PERL_STRICT_CR |
| 3558 | && s[1] == '\n' |
| 3559 | #else |
| 3560 | && (s[1] == '\n' || (s[1] == '\r' && s[2] == '\n')) |
| 3561 | #endif |
| 3562 | && (s == PL_linestart || s[-1] == '\n') ) |
| 3563 | { |
| 3564 | PL_lex_formbrack = 0; |
| 3565 | PL_expect = XSTATE; |
| 3566 | goto rightbracket; |
| 3567 | } |
| 3568 | if (PL_expect == XOPERATOR || !isDIGIT(s[1])) { |
| 3569 | tmp = *s++; |
| 3570 | if (*s == tmp) { |
| 3571 | s++; |
| 3572 | if (*s == tmp) { |
| 3573 | s++; |
| 3574 | yylval.ival = OPf_SPECIAL; |
| 3575 | } |
| 3576 | else |
| 3577 | yylval.ival = 0; |
| 3578 | OPERATOR(DOTDOT); |
| 3579 | } |
| 3580 | if (PL_expect != XOPERATOR) |
| 3581 | check_uni(); |
| 3582 | Aop(OP_CONCAT); |
| 3583 | } |
| 3584 | /* FALL THROUGH */ |
| 3585 | case '0': case '1': case '2': case '3': case '4': |
| 3586 | case '5': case '6': case '7': case '8': case '9': |
| 3587 | s = scan_num(s, &yylval); |
| 3588 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 3589 | "### Saw number in '%s'\n", s); |
| 3590 | } ); |
| 3591 | if (PL_expect == XOPERATOR) |
| 3592 | no_op("Number",s); |
| 3593 | TERM(THING); |
| 3594 | |
| 3595 | case '\'': |
| 3596 | s = scan_str(s,FALSE,FALSE); |
| 3597 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 3598 | "### Saw string before '%s'\n", s); |
| 3599 | } ); |
| 3600 | if (PL_expect == XOPERATOR) { |
| 3601 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { |
| 3602 | PL_expect = XTERM; |
| 3603 | depcom(); |
| 3604 | return ','; /* grandfather non-comma-format format */ |
| 3605 | } |
| 3606 | else |
| 3607 | no_op("String",s); |
| 3608 | } |
| 3609 | if (!s) |
| 3610 | missingterm((char*)0); |
| 3611 | yylval.ival = OP_CONST; |
| 3612 | TERM(sublex_start()); |
| 3613 | |
| 3614 | case '"': |
| 3615 | s = scan_str(s,FALSE,FALSE); |
| 3616 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 3617 | "### Saw string before '%s'\n", s); |
| 3618 | } ); |
| 3619 | if (PL_expect == XOPERATOR) { |
| 3620 | if (PL_lex_formbrack && PL_lex_brackets == PL_lex_formbrack) { |
| 3621 | PL_expect = XTERM; |
| 3622 | depcom(); |
| 3623 | return ','; /* grandfather non-comma-format format */ |
| 3624 | } |
| 3625 | else |
| 3626 | no_op("String",s); |
| 3627 | } |
| 3628 | if (!s) |
| 3629 | missingterm((char*)0); |
| 3630 | yylval.ival = OP_CONST; |
| 3631 | for (d = SvPV(PL_lex_stuff, len); len; len--, d++) { |
| 3632 | if (*d == '$' || *d == '@' || *d == '\\' || !UTF8_IS_INVARIANT((U8)*d)) { |
| 3633 | yylval.ival = OP_STRINGIFY; |
| 3634 | break; |
| 3635 | } |
| 3636 | } |
| 3637 | TERM(sublex_start()); |
| 3638 | |
| 3639 | case '`': |
| 3640 | s = scan_str(s,FALSE,FALSE); |
| 3641 | DEBUG_T( { PerlIO_printf(Perl_debug_log, |
| 3642 | "### Saw backtick string before '%s'\n", s); |
| 3643 | } ); |
| 3644 | if (PL_expect == XOPERATOR) |
| 3645 | no_op("Backticks",s); |
| 3646 | if (!s) |
| 3647 | missingterm((char*)0); |
| 3648 | yylval.ival = OP_BACKTICK; |
| 3649 | set_csh(); |
| 3650 | TERM(sublex_start()); |
| 3651 | |
| 3652 | case '\\': |
| 3653 | s++; |
| 3654 | if (ckWARN(WARN_SYNTAX) && PL_lex_inwhat && isDIGIT(*s)) |
| 3655 | Perl_warner(aTHX_ WARN_SYNTAX,"Can't use \\%c to mean $%c in expression", |
| 3656 | *s, *s); |
| 3657 | if (PL_expect == XOPERATOR) |
| 3658 | no_op("Backslash",s); |
| 3659 | OPERATOR(REFGEN); |
| 3660 | |
| 3661 | case 'v': |
| 3662 | if (isDIGIT(s[1]) && PL_expect != XOPERATOR) { |
| 3663 | char *start = s; |
| 3664 | start++; |
| 3665 | start++; |
| 3666 | while (isDIGIT(*start) || *start == '_') |
| 3667 | start++; |
| 3668 | if (*start == '.' && isDIGIT(start[1])) { |
| 3669 | s = scan_num(s, &yylval); |
| 3670 | TERM(THING); |
| 3671 | } |
| 3672 | /* avoid v123abc() or $h{v1}, allow C<print v10;> */ |
| 3673 | else if (!isALPHA(*start) && (PL_expect == XTERM || PL_expect == XREF || PL_expect == XSTATE)) { |
| 3674 | char c = *start; |
| 3675 | GV *gv; |
| 3676 | *start = '\0'; |
| 3677 | gv = gv_fetchpv(s, FALSE, SVt_PVCV); |
| 3678 | *start = c; |
| 3679 | if (!gv) { |
| 3680 | s = scan_num(s, &yylval); |
| 3681 | TERM(THING); |
| 3682 | } |
| 3683 | } |
| 3684 | } |
| 3685 | goto keylookup; |
| 3686 | case 'x': |
| 3687 | if (isDIGIT(s[1]) && PL_expect == XOPERATOR) { |
| 3688 | s++; |
| 3689 | Mop(OP_REPEAT); |
| 3690 | } |
| 3691 | goto keylookup; |
| 3692 | |
| 3693 | case '_': |
| 3694 | case 'a': case 'A': |
| 3695 | case 'b': case 'B': |
| 3696 | case 'c': case 'C': |
| 3697 | case 'd': case 'D': |
| 3698 | case 'e': case 'E': |
| 3699 | case 'f': case 'F': |
| 3700 | case 'g': case 'G': |
| 3701 | case 'h': case 'H': |
| 3702 | case 'i': case 'I': |
| 3703 | case 'j': case 'J': |
| 3704 | case 'k': case 'K': |
| 3705 | case 'l': case 'L': |
| 3706 | case 'm': case 'M': |
| 3707 | case 'n': case 'N': |
| 3708 | case 'o': case 'O': |
| 3709 | case 'p': case 'P': |
| 3710 | case 'q': case 'Q': |
| 3711 | case 'r': case 'R': |
| 3712 | case 's': case 'S': |
| 3713 | case 't': case 'T': |
| 3714 | case 'u': case 'U': |
| 3715 | case 'V': |
| 3716 | case 'w': case 'W': |
| 3717 | case 'X': |
| 3718 | case 'y': case 'Y': |
| 3719 | case 'z': case 'Z': |
| 3720 | |
| 3721 | keylookup: { |
| 3722 | gv = Nullgv; |
| 3723 | gvp = 0; |
| 3724 | |
| 3725 | PL_bufptr = s; |
| 3726 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); |
| 3727 | |
| 3728 | /* Some keywords can be followed by any delimiter, including ':' */ |
| 3729 | tmp = ((len == 1 && strchr("msyq", PL_tokenbuf[0])) || |
| 3730 | (len == 2 && ((PL_tokenbuf[0] == 't' && PL_tokenbuf[1] == 'r') || |
| 3731 | (PL_tokenbuf[0] == 'q' && |
| 3732 | strchr("qwxr", PL_tokenbuf[1]))))); |
| 3733 | |
| 3734 | /* x::* is just a word, unless x is "CORE" */ |
| 3735 | if (!tmp && *s == ':' && s[1] == ':' && strNE(PL_tokenbuf, "CORE")) |
| 3736 | goto just_a_word; |
| 3737 | |
| 3738 | d = s; |
| 3739 | while (d < PL_bufend && isSPACE(*d)) |
| 3740 | d++; /* no comments skipped here, or s### is misparsed */ |
| 3741 | |
| 3742 | /* Is this a label? */ |
| 3743 | if (!tmp && PL_expect == XSTATE |
| 3744 | && d < PL_bufend && *d == ':' && *(d + 1) != ':') { |
| 3745 | s = d + 1; |
| 3746 | yylval.pval = savepv(PL_tokenbuf); |
| 3747 | CLINE; |
| 3748 | TOKEN(LABEL); |
| 3749 | } |
| 3750 | |
| 3751 | /* Check for keywords */ |
| 3752 | tmp = keyword(PL_tokenbuf, len); |
| 3753 | |
| 3754 | /* Is this a word before a => operator? */ |
| 3755 | if (*d == '=' && d[1] == '>') { |
| 3756 | CLINE; |
| 3757 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf,0)); |
| 3758 | yylval.opval->op_private = OPpCONST_BARE; |
| 3759 | if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len)) |
| 3760 | SvUTF8_on(((SVOP*)yylval.opval)->op_sv); |
| 3761 | TERM(WORD); |
| 3762 | } |
| 3763 | |
| 3764 | if (tmp < 0) { /* second-class keyword? */ |
| 3765 | GV *ogv = Nullgv; /* override (winner) */ |
| 3766 | GV *hgv = Nullgv; /* hidden (loser) */ |
| 3767 | if (PL_expect != XOPERATOR && (*s != ':' || s[1] != ':')) { |
| 3768 | CV *cv; |
| 3769 | if ((gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV)) && |
| 3770 | (cv = GvCVu(gv))) |
| 3771 | { |
| 3772 | if (GvIMPORTED_CV(gv)) |
| 3773 | ogv = gv; |
| 3774 | else if (! CvMETHOD(cv)) |
| 3775 | hgv = gv; |
| 3776 | } |
| 3777 | if (!ogv && |
| 3778 | (gvp = (GV**)hv_fetch(PL_globalstash,PL_tokenbuf,len,FALSE)) && |
| 3779 | (gv = *gvp) != (GV*)&PL_sv_undef && |
| 3780 | GvCVu(gv) && GvIMPORTED_CV(gv)) |
| 3781 | { |
| 3782 | ogv = gv; |
| 3783 | } |
| 3784 | } |
| 3785 | if (ogv) { |
| 3786 | tmp = 0; /* overridden by import or by GLOBAL */ |
| 3787 | } |
| 3788 | else if (gv && !gvp |
| 3789 | && -tmp==KEY_lock /* XXX generalizable kludge */ |
| 3790 | && GvCVu(gv) |
| 3791 | && !hv_fetch(GvHVn(PL_incgv), "Thread.pm", 9, FALSE)) |
| 3792 | { |
| 3793 | tmp = 0; /* any sub overrides "weak" keyword */ |
| 3794 | } |
| 3795 | else { /* no override */ |
| 3796 | tmp = -tmp; |
| 3797 | gv = Nullgv; |
| 3798 | gvp = 0; |
| 3799 | if (ckWARN(WARN_AMBIGUOUS) && hgv |
| 3800 | && tmp != KEY_x && tmp != KEY_CORE) /* never ambiguous */ |
| 3801 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 3802 | "Ambiguous call resolved as CORE::%s(), %s", |
| 3803 | GvENAME(hgv), "qualify as such or use &"); |
| 3804 | } |
| 3805 | } |
| 3806 | |
| 3807 | reserved_word: |
| 3808 | switch (tmp) { |
| 3809 | |
| 3810 | default: /* not a keyword */ |
| 3811 | just_a_word: { |
| 3812 | SV *sv; |
| 3813 | int pkgname = 0; |
| 3814 | char lastchar = (PL_bufptr == PL_oldoldbufptr ? 0 : PL_bufptr[-1]); |
| 3815 | |
| 3816 | /* Get the rest if it looks like a package qualifier */ |
| 3817 | |
| 3818 | if (*s == '\'' || (*s == ':' && s[1] == ':')) { |
| 3819 | STRLEN morelen; |
| 3820 | s = scan_word(s, PL_tokenbuf + len, sizeof PL_tokenbuf - len, |
| 3821 | TRUE, &morelen); |
| 3822 | if (!morelen) |
| 3823 | Perl_croak(aTHX_ "Bad name after %s%s", PL_tokenbuf, |
| 3824 | *s == '\'' ? "'" : "::"); |
| 3825 | len += morelen; |
| 3826 | pkgname = 1; |
| 3827 | } |
| 3828 | |
| 3829 | if (PL_expect == XOPERATOR) { |
| 3830 | if (PL_bufptr == PL_linestart) { |
| 3831 | CopLINE_dec(PL_curcop); |
| 3832 | Perl_warner(aTHX_ WARN_SEMICOLON, PL_warn_nosemi); |
| 3833 | CopLINE_inc(PL_curcop); |
| 3834 | } |
| 3835 | else |
| 3836 | no_op("Bareword",s); |
| 3837 | } |
| 3838 | |
| 3839 | /* Look for a subroutine with this name in current package, |
| 3840 | unless name is "Foo::", in which case Foo is a bearword |
| 3841 | (and a package name). */ |
| 3842 | |
| 3843 | if (len > 2 && |
| 3844 | PL_tokenbuf[len - 2] == ':' && PL_tokenbuf[len - 1] == ':') |
| 3845 | { |
| 3846 | if (ckWARN(WARN_BAREWORD) && ! gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVHV)) |
| 3847 | Perl_warner(aTHX_ WARN_BAREWORD, |
| 3848 | "Bareword \"%s\" refers to nonexistent package", |
| 3849 | PL_tokenbuf); |
| 3850 | len -= 2; |
| 3851 | PL_tokenbuf[len] = '\0'; |
| 3852 | gv = Nullgv; |
| 3853 | gvp = 0; |
| 3854 | } |
| 3855 | else { |
| 3856 | len = 0; |
| 3857 | if (!gv) |
| 3858 | gv = gv_fetchpv(PL_tokenbuf, FALSE, SVt_PVCV); |
| 3859 | } |
| 3860 | |
| 3861 | /* if we saw a global override before, get the right name */ |
| 3862 | |
| 3863 | if (gvp) { |
| 3864 | sv = newSVpvn("CORE::GLOBAL::",14); |
| 3865 | sv_catpv(sv,PL_tokenbuf); |
| 3866 | } |
| 3867 | else |
| 3868 | sv = newSVpv(PL_tokenbuf,0); |
| 3869 | |
| 3870 | /* Presume this is going to be a bareword of some sort. */ |
| 3871 | |
| 3872 | CLINE; |
| 3873 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sv); |
| 3874 | yylval.opval->op_private = OPpCONST_BARE; |
| 3875 | |
| 3876 | /* And if "Foo::", then that's what it certainly is. */ |
| 3877 | |
| 3878 | if (len) |
| 3879 | goto safe_bareword; |
| 3880 | |
| 3881 | /* See if it's the indirect object for a list operator. */ |
| 3882 | |
| 3883 | if (PL_oldoldbufptr && |
| 3884 | PL_oldoldbufptr < PL_bufptr && |
| 3885 | (PL_oldoldbufptr == PL_last_lop |
| 3886 | || PL_oldoldbufptr == PL_last_uni) && |
| 3887 | /* NO SKIPSPACE BEFORE HERE! */ |
| 3888 | (PL_expect == XREF || |
| 3889 | ((PL_opargs[PL_last_lop_op] >> OASHIFT)& 7) == OA_FILEREF)) |
| 3890 | { |
| 3891 | bool immediate_paren = *s == '('; |
| 3892 | |
| 3893 | /* (Now we can afford to cross potential line boundary.) */ |
| 3894 | s = skipspace(s); |
| 3895 | |
| 3896 | /* Two barewords in a row may indicate method call. */ |
| 3897 | |
| 3898 | if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp=intuit_method(s,gv))) |
| 3899 | return tmp; |
| 3900 | |
| 3901 | /* If not a declared subroutine, it's an indirect object. */ |
| 3902 | /* (But it's an indir obj regardless for sort.) */ |
| 3903 | |
| 3904 | if ( !immediate_paren && (PL_last_lop_op == OP_SORT || |
| 3905 | ((!gv || !GvCVu(gv)) && |
| 3906 | (PL_last_lop_op != OP_MAPSTART && |
| 3907 | PL_last_lop_op != OP_GREPSTART)))) |
| 3908 | { |
| 3909 | PL_expect = (PL_last_lop == PL_oldoldbufptr) ? XTERM : XOPERATOR; |
| 3910 | goto bareword; |
| 3911 | } |
| 3912 | } |
| 3913 | |
| 3914 | PL_expect = XOPERATOR; |
| 3915 | s = skipspace(s); |
| 3916 | |
| 3917 | /* Is this a word before a => operator? */ |
| 3918 | if (*s == '=' && s[1] == '>' && !pkgname) { |
| 3919 | CLINE; |
| 3920 | sv_setpv(((SVOP*)yylval.opval)->op_sv, PL_tokenbuf); |
| 3921 | if (UTF && !IN_BYTES && is_utf8_string((U8*)PL_tokenbuf, len)) |
| 3922 | SvUTF8_on(((SVOP*)yylval.opval)->op_sv); |
| 3923 | TERM(WORD); |
| 3924 | } |
| 3925 | |
| 3926 | /* If followed by a paren, it's certainly a subroutine. */ |
| 3927 | if (*s == '(') { |
| 3928 | CLINE; |
| 3929 | if (gv && GvCVu(gv)) { |
| 3930 | for (d = s + 1; SPACE_OR_TAB(*d); d++) ; |
| 3931 | if (*d == ')' && (sv = cv_const_sv(GvCV(gv)))) { |
| 3932 | s = d + 1; |
| 3933 | goto its_constant; |
| 3934 | } |
| 3935 | } |
| 3936 | PL_nextval[PL_nexttoke].opval = yylval.opval; |
| 3937 | PL_expect = XOPERATOR; |
| 3938 | force_next(WORD); |
| 3939 | yylval.ival = 0; |
| 3940 | TOKEN('&'); |
| 3941 | } |
| 3942 | |
| 3943 | /* If followed by var or block, call it a method (unless sub) */ |
| 3944 | |
| 3945 | if ((*s == '$' || *s == '{') && (!gv || !GvCVu(gv))) { |
| 3946 | PL_last_lop = PL_oldbufptr; |
| 3947 | PL_last_lop_op = OP_METHOD; |
| 3948 | PREBLOCK(METHOD); |
| 3949 | } |
| 3950 | |
| 3951 | /* If followed by a bareword, see if it looks like indir obj. */ |
| 3952 | |
| 3953 | if ((isIDFIRST_lazy_if(s,UTF) || *s == '$') && (tmp = intuit_method(s,gv))) |
| 3954 | return tmp; |
| 3955 | |
| 3956 | /* Not a method, so call it a subroutine (if defined) */ |
| 3957 | |
| 3958 | if (gv && GvCVu(gv)) { |
| 3959 | CV* cv; |
| 3960 | if (lastchar == '-' && ckWARN_d(WARN_AMBIGUOUS)) |
| 3961 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 3962 | "Ambiguous use of -%s resolved as -&%s()", |
| 3963 | PL_tokenbuf, PL_tokenbuf); |
| 3964 | /* Check for a constant sub */ |
| 3965 | cv = GvCV(gv); |
| 3966 | if ((sv = cv_const_sv(cv))) { |
| 3967 | its_constant: |
| 3968 | SvREFCNT_dec(((SVOP*)yylval.opval)->op_sv); |
| 3969 | ((SVOP*)yylval.opval)->op_sv = SvREFCNT_inc(sv); |
| 3970 | yylval.opval->op_private = 0; |
| 3971 | TOKEN(WORD); |
| 3972 | } |
| 3973 | |
| 3974 | /* Resolve to GV now. */ |
| 3975 | op_free(yylval.opval); |
| 3976 | yylval.opval = newCVREF(0, newGVOP(OP_GV, 0, gv)); |
| 3977 | yylval.opval->op_private |= OPpENTERSUB_NOPAREN; |
| 3978 | PL_last_lop = PL_oldbufptr; |
| 3979 | PL_last_lop_op = OP_ENTERSUB; |
| 3980 | /* Is there a prototype? */ |
| 3981 | if (SvPOK(cv)) { |
| 3982 | STRLEN len; |
| 3983 | char *proto = SvPV((SV*)cv, len); |
| 3984 | if (!len) |
| 3985 | TERM(FUNC0SUB); |
| 3986 | if (strEQ(proto, "$")) |
| 3987 | OPERATOR(UNIOPSUB); |
| 3988 | if (*proto == '&' && *s == '{') { |
| 3989 | sv_setpv(PL_subname,"__ANON__"); |
| 3990 | PREBLOCK(LSTOPSUB); |
| 3991 | } |
| 3992 | } |
| 3993 | PL_nextval[PL_nexttoke].opval = yylval.opval; |
| 3994 | PL_expect = XTERM; |
| 3995 | force_next(WORD); |
| 3996 | TOKEN(NOAMP); |
| 3997 | } |
| 3998 | |
| 3999 | /* Call it a bare word */ |
| 4000 | |
| 4001 | if (PL_hints & HINT_STRICT_SUBS) |
| 4002 | yylval.opval->op_private |= OPpCONST_STRICT; |
| 4003 | else { |
| 4004 | bareword: |
| 4005 | if (ckWARN(WARN_RESERVED)) { |
| 4006 | if (lastchar != '-') { |
| 4007 | for (d = PL_tokenbuf; *d && isLOWER(*d); d++) ; |
| 4008 | if (!*d && strNE(PL_tokenbuf,"main")) |
| 4009 | Perl_warner(aTHX_ WARN_RESERVED, PL_warn_reserved, |
| 4010 | PL_tokenbuf); |
| 4011 | } |
| 4012 | } |
| 4013 | } |
| 4014 | |
| 4015 | safe_bareword: |
| 4016 | if (lastchar && strchr("*%&", lastchar) && ckWARN_d(WARN_AMBIGUOUS)) { |
| 4017 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 4018 | "Operator or semicolon missing before %c%s", |
| 4019 | lastchar, PL_tokenbuf); |
| 4020 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 4021 | "Ambiguous use of %c resolved as operator %c", |
| 4022 | lastchar, lastchar); |
| 4023 | } |
| 4024 | TOKEN(WORD); |
| 4025 | } |
| 4026 | |
| 4027 | case KEY___FILE__: |
| 4028 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, |
| 4029 | newSVpv(CopFILE(PL_curcop),0)); |
| 4030 | TERM(THING); |
| 4031 | |
| 4032 | case KEY___LINE__: |
| 4033 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, |
| 4034 | Perl_newSVpvf(aTHX_ "%"IVdf, (IV)CopLINE(PL_curcop))); |
| 4035 | TERM(THING); |
| 4036 | |
| 4037 | case KEY___PACKAGE__: |
| 4038 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, |
| 4039 | (PL_curstash |
| 4040 | ? newSVsv(PL_curstname) |
| 4041 | : &PL_sv_undef)); |
| 4042 | TERM(THING); |
| 4043 | |
| 4044 | case KEY___DATA__: |
| 4045 | case KEY___END__: { |
| 4046 | GV *gv; |
| 4047 | |
| 4048 | /*SUPPRESS 560*/ |
| 4049 | if (PL_rsfp && (!PL_in_eval || PL_tokenbuf[2] == 'D')) { |
| 4050 | char *pname = "main"; |
| 4051 | if (PL_tokenbuf[2] == 'D') |
| 4052 | pname = HvNAME(PL_curstash ? PL_curstash : PL_defstash); |
| 4053 | gv = gv_fetchpv(Perl_form(aTHX_ "%s::DATA", pname), TRUE, SVt_PVIO); |
| 4054 | GvMULTI_on(gv); |
| 4055 | if (!GvIO(gv)) |
| 4056 | GvIOp(gv) = newIO(); |
| 4057 | IoIFP(GvIOp(gv)) = PL_rsfp; |
| 4058 | #if defined(HAS_FCNTL) && defined(F_SETFD) |
| 4059 | { |
| 4060 | int fd = PerlIO_fileno(PL_rsfp); |
| 4061 | fcntl(fd,F_SETFD,fd >= 3); |
| 4062 | } |
| 4063 | #endif |
| 4064 | /* Mark this internal pseudo-handle as clean */ |
| 4065 | IoFLAGS(GvIOp(gv)) |= IOf_UNTAINT; |
| 4066 | if (PL_preprocess) |
| 4067 | IoTYPE(GvIOp(gv)) = IoTYPE_PIPE; |
| 4068 | else if ((PerlIO*)PL_rsfp == PerlIO_stdin()) |
| 4069 | IoTYPE(GvIOp(gv)) = IoTYPE_STD; |
| 4070 | else |
| 4071 | IoTYPE(GvIOp(gv)) = IoTYPE_RDONLY; |
| 4072 | #if defined(WIN32) && !defined(PERL_TEXTMODE_SCRIPTS) |
| 4073 | /* if the script was opened in binmode, we need to revert |
| 4074 | * it to text mode for compatibility; but only iff it has CRs |
| 4075 | * XXX this is a questionable hack at best. */ |
| 4076 | if (PL_bufend-PL_bufptr > 2 |
| 4077 | && PL_bufend[-1] == '\n' && PL_bufend[-2] == '\r') |
| 4078 | { |
| 4079 | Off_t loc = 0; |
| 4080 | if (IoTYPE(GvIOp(gv)) == IoTYPE_RDONLY) { |
| 4081 | loc = PerlIO_tell(PL_rsfp); |
| 4082 | (void)PerlIO_seek(PL_rsfp, 0L, 0); |
| 4083 | } |
| 4084 | #ifdef NETWARE |
| 4085 | if (PerlLIO_setmode(PL_rsfp, O_TEXT) != -1) { |
| 4086 | #else |
| 4087 | if (PerlLIO_setmode(PerlIO_fileno(PL_rsfp), O_TEXT) != -1) { |
| 4088 | #endif /* NETWARE */ |
| 4089 | #ifdef PERLIO_IS_STDIO /* really? */ |
| 4090 | # if defined(__BORLANDC__) |
| 4091 | /* XXX see note in do_binmode() */ |
| 4092 | ((FILE*)PL_rsfp)->flags &= ~_F_BIN; |
| 4093 | # endif |
| 4094 | #endif |
| 4095 | if (loc > 0) |
| 4096 | PerlIO_seek(PL_rsfp, loc, 0); |
| 4097 | } |
| 4098 | } |
| 4099 | #endif |
| 4100 | #ifdef PERLIO_LAYERS |
| 4101 | if (UTF && !IN_BYTES) |
| 4102 | PerlIO_apply_layers(aTHX_ PL_rsfp, NULL, ":utf8"); |
| 4103 | #endif |
| 4104 | PL_rsfp = Nullfp; |
| 4105 | } |
| 4106 | goto fake_eof; |
| 4107 | } |
| 4108 | |
| 4109 | case KEY_AUTOLOAD: |
| 4110 | case KEY_DESTROY: |
| 4111 | case KEY_BEGIN: |
| 4112 | case KEY_CHECK: |
| 4113 | case KEY_INIT: |
| 4114 | case KEY_END: |
| 4115 | if (PL_expect == XSTATE) { |
| 4116 | s = PL_bufptr; |
| 4117 | goto really_sub; |
| 4118 | } |
| 4119 | goto just_a_word; |
| 4120 | |
| 4121 | case KEY_CORE: |
| 4122 | if (*s == ':' && s[1] == ':') { |
| 4123 | s += 2; |
| 4124 | d = s; |
| 4125 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, FALSE, &len); |
| 4126 | if (!(tmp = keyword(PL_tokenbuf, len))) |
| 4127 | Perl_croak(aTHX_ "CORE::%s is not a keyword", PL_tokenbuf); |
| 4128 | if (tmp < 0) |
| 4129 | tmp = -tmp; |
| 4130 | goto reserved_word; |
| 4131 | } |
| 4132 | goto just_a_word; |
| 4133 | |
| 4134 | case KEY_abs: |
| 4135 | UNI(OP_ABS); |
| 4136 | |
| 4137 | case KEY_alarm: |
| 4138 | UNI(OP_ALARM); |
| 4139 | |
| 4140 | case KEY_accept: |
| 4141 | LOP(OP_ACCEPT,XTERM); |
| 4142 | |
| 4143 | case KEY_and: |
| 4144 | OPERATOR(ANDOP); |
| 4145 | |
| 4146 | case KEY_atan2: |
| 4147 | LOP(OP_ATAN2,XTERM); |
| 4148 | |
| 4149 | case KEY_bind: |
| 4150 | LOP(OP_BIND,XTERM); |
| 4151 | |
| 4152 | case KEY_binmode: |
| 4153 | LOP(OP_BINMODE,XTERM); |
| 4154 | |
| 4155 | case KEY_bless: |
| 4156 | LOP(OP_BLESS,XTERM); |
| 4157 | |
| 4158 | case KEY_chop: |
| 4159 | UNI(OP_CHOP); |
| 4160 | |
| 4161 | case KEY_continue: |
| 4162 | PREBLOCK(CONTINUE); |
| 4163 | |
| 4164 | case KEY_chdir: |
| 4165 | (void)gv_fetchpv("ENV",TRUE, SVt_PVHV); /* may use HOME */ |
| 4166 | UNI(OP_CHDIR); |
| 4167 | |
| 4168 | case KEY_close: |
| 4169 | UNI(OP_CLOSE); |
| 4170 | |
| 4171 | case KEY_closedir: |
| 4172 | UNI(OP_CLOSEDIR); |
| 4173 | |
| 4174 | case KEY_cmp: |
| 4175 | Eop(OP_SCMP); |
| 4176 | |
| 4177 | case KEY_caller: |
| 4178 | UNI(OP_CALLER); |
| 4179 | |
| 4180 | case KEY_crypt: |
| 4181 | #ifdef FCRYPT |
| 4182 | if (!PL_cryptseen) { |
| 4183 | PL_cryptseen = TRUE; |
| 4184 | init_des(); |
| 4185 | } |
| 4186 | #endif |
| 4187 | LOP(OP_CRYPT,XTERM); |
| 4188 | |
| 4189 | case KEY_chmod: |
| 4190 | LOP(OP_CHMOD,XTERM); |
| 4191 | |
| 4192 | case KEY_chown: |
| 4193 | LOP(OP_CHOWN,XTERM); |
| 4194 | |
| 4195 | case KEY_connect: |
| 4196 | LOP(OP_CONNECT,XTERM); |
| 4197 | |
| 4198 | case KEY_chr: |
| 4199 | UNI(OP_CHR); |
| 4200 | |
| 4201 | case KEY_cos: |
| 4202 | UNI(OP_COS); |
| 4203 | |
| 4204 | case KEY_chroot: |
| 4205 | UNI(OP_CHROOT); |
| 4206 | |
| 4207 | case KEY_do: |
| 4208 | s = skipspace(s); |
| 4209 | if (*s == '{') |
| 4210 | PRETERMBLOCK(DO); |
| 4211 | if (*s != '\'') |
| 4212 | s = force_word(s,WORD,TRUE,TRUE,FALSE); |
| 4213 | OPERATOR(DO); |
| 4214 | |
| 4215 | case KEY_die: |
| 4216 | PL_hints |= HINT_BLOCK_SCOPE; |
| 4217 | LOP(OP_DIE,XTERM); |
| 4218 | |
| 4219 | case KEY_defined: |
| 4220 | UNI(OP_DEFINED); |
| 4221 | |
| 4222 | case KEY_delete: |
| 4223 | UNI(OP_DELETE); |
| 4224 | |
| 4225 | case KEY_dbmopen: |
| 4226 | gv_fetchpv("AnyDBM_File::ISA", GV_ADDMULTI, SVt_PVAV); |
| 4227 | LOP(OP_DBMOPEN,XTERM); |
| 4228 | |
| 4229 | case KEY_dbmclose: |
| 4230 | UNI(OP_DBMCLOSE); |
| 4231 | |
| 4232 | case KEY_dump: |
| 4233 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
| 4234 | LOOPX(OP_DUMP); |
| 4235 | |
| 4236 | case KEY_else: |
| 4237 | PREBLOCK(ELSE); |
| 4238 | |
| 4239 | case KEY_elsif: |
| 4240 | yylval.ival = CopLINE(PL_curcop); |
| 4241 | OPERATOR(ELSIF); |
| 4242 | |
| 4243 | case KEY_eq: |
| 4244 | Eop(OP_SEQ); |
| 4245 | |
| 4246 | case KEY_exists: |
| 4247 | UNI(OP_EXISTS); |
| 4248 | |
| 4249 | case KEY_exit: |
| 4250 | UNI(OP_EXIT); |
| 4251 | |
| 4252 | case KEY_eval: |
| 4253 | s = skipspace(s); |
| 4254 | PL_expect = (*s == '{') ? XTERMBLOCK : XTERM; |
| 4255 | UNIBRACK(OP_ENTEREVAL); |
| 4256 | |
| 4257 | case KEY_eof: |
| 4258 | UNI(OP_EOF); |
| 4259 | |
| 4260 | case KEY_exp: |
| 4261 | UNI(OP_EXP); |
| 4262 | |
| 4263 | case KEY_each: |
| 4264 | UNI(OP_EACH); |
| 4265 | |
| 4266 | case KEY_exec: |
| 4267 | set_csh(); |
| 4268 | LOP(OP_EXEC,XREF); |
| 4269 | |
| 4270 | case KEY_endhostent: |
| 4271 | FUN0(OP_EHOSTENT); |
| 4272 | |
| 4273 | case KEY_endnetent: |
| 4274 | FUN0(OP_ENETENT); |
| 4275 | |
| 4276 | case KEY_endservent: |
| 4277 | FUN0(OP_ESERVENT); |
| 4278 | |
| 4279 | case KEY_endprotoent: |
| 4280 | FUN0(OP_EPROTOENT); |
| 4281 | |
| 4282 | case KEY_endpwent: |
| 4283 | FUN0(OP_EPWENT); |
| 4284 | |
| 4285 | case KEY_endgrent: |
| 4286 | FUN0(OP_EGRENT); |
| 4287 | |
| 4288 | case KEY_for: |
| 4289 | case KEY_foreach: |
| 4290 | yylval.ival = CopLINE(PL_curcop); |
| 4291 | s = skipspace(s); |
| 4292 | if (PL_expect == XSTATE && isIDFIRST_lazy_if(s,UTF)) { |
| 4293 | char *p = s; |
| 4294 | if ((PL_bufend - p) >= 3 && |
| 4295 | strnEQ(p, "my", 2) && isSPACE(*(p + 2))) |
| 4296 | p += 2; |
| 4297 | else if ((PL_bufend - p) >= 4 && |
| 4298 | strnEQ(p, "our", 3) && isSPACE(*(p + 3))) |
| 4299 | p += 3; |
| 4300 | p = skipspace(p); |
| 4301 | if (isIDFIRST_lazy_if(p,UTF)) { |
| 4302 | p = scan_ident(p, PL_bufend, |
| 4303 | PL_tokenbuf, sizeof PL_tokenbuf, TRUE); |
| 4304 | p = skipspace(p); |
| 4305 | } |
| 4306 | if (*p != '$') |
| 4307 | Perl_croak(aTHX_ "Missing $ on loop variable"); |
| 4308 | } |
| 4309 | OPERATOR(FOR); |
| 4310 | |
| 4311 | case KEY_formline: |
| 4312 | LOP(OP_FORMLINE,XTERM); |
| 4313 | |
| 4314 | case KEY_fork: |
| 4315 | FUN0(OP_FORK); |
| 4316 | |
| 4317 | case KEY_fcntl: |
| 4318 | LOP(OP_FCNTL,XTERM); |
| 4319 | |
| 4320 | case KEY_fileno: |
| 4321 | UNI(OP_FILENO); |
| 4322 | |
| 4323 | case KEY_flock: |
| 4324 | LOP(OP_FLOCK,XTERM); |
| 4325 | |
| 4326 | case KEY_gt: |
| 4327 | Rop(OP_SGT); |
| 4328 | |
| 4329 | case KEY_ge: |
| 4330 | Rop(OP_SGE); |
| 4331 | |
| 4332 | case KEY_grep: |
| 4333 | LOP(OP_GREPSTART, XREF); |
| 4334 | |
| 4335 | case KEY_goto: |
| 4336 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
| 4337 | LOOPX(OP_GOTO); |
| 4338 | |
| 4339 | case KEY_gmtime: |
| 4340 | UNI(OP_GMTIME); |
| 4341 | |
| 4342 | case KEY_getc: |
| 4343 | UNI(OP_GETC); |
| 4344 | |
| 4345 | case KEY_getppid: |
| 4346 | FUN0(OP_GETPPID); |
| 4347 | |
| 4348 | case KEY_getpgrp: |
| 4349 | UNI(OP_GETPGRP); |
| 4350 | |
| 4351 | case KEY_getpriority: |
| 4352 | LOP(OP_GETPRIORITY,XTERM); |
| 4353 | |
| 4354 | case KEY_getprotobyname: |
| 4355 | UNI(OP_GPBYNAME); |
| 4356 | |
| 4357 | case KEY_getprotobynumber: |
| 4358 | LOP(OP_GPBYNUMBER,XTERM); |
| 4359 | |
| 4360 | case KEY_getprotoent: |
| 4361 | FUN0(OP_GPROTOENT); |
| 4362 | |
| 4363 | case KEY_getpwent: |
| 4364 | FUN0(OP_GPWENT); |
| 4365 | |
| 4366 | case KEY_getpwnam: |
| 4367 | UNI(OP_GPWNAM); |
| 4368 | |
| 4369 | case KEY_getpwuid: |
| 4370 | UNI(OP_GPWUID); |
| 4371 | |
| 4372 | case KEY_getpeername: |
| 4373 | UNI(OP_GETPEERNAME); |
| 4374 | |
| 4375 | case KEY_gethostbyname: |
| 4376 | UNI(OP_GHBYNAME); |
| 4377 | |
| 4378 | case KEY_gethostbyaddr: |
| 4379 | LOP(OP_GHBYADDR,XTERM); |
| 4380 | |
| 4381 | case KEY_gethostent: |
| 4382 | FUN0(OP_GHOSTENT); |
| 4383 | |
| 4384 | case KEY_getnetbyname: |
| 4385 | UNI(OP_GNBYNAME); |
| 4386 | |
| 4387 | case KEY_getnetbyaddr: |
| 4388 | LOP(OP_GNBYADDR,XTERM); |
| 4389 | |
| 4390 | case KEY_getnetent: |
| 4391 | FUN0(OP_GNETENT); |
| 4392 | |
| 4393 | case KEY_getservbyname: |
| 4394 | LOP(OP_GSBYNAME,XTERM); |
| 4395 | |
| 4396 | case KEY_getservbyport: |
| 4397 | LOP(OP_GSBYPORT,XTERM); |
| 4398 | |
| 4399 | case KEY_getservent: |
| 4400 | FUN0(OP_GSERVENT); |
| 4401 | |
| 4402 | case KEY_getsockname: |
| 4403 | UNI(OP_GETSOCKNAME); |
| 4404 | |
| 4405 | case KEY_getsockopt: |
| 4406 | LOP(OP_GSOCKOPT,XTERM); |
| 4407 | |
| 4408 | case KEY_getgrent: |
| 4409 | FUN0(OP_GGRENT); |
| 4410 | |
| 4411 | case KEY_getgrnam: |
| 4412 | UNI(OP_GGRNAM); |
| 4413 | |
| 4414 | case KEY_getgrgid: |
| 4415 | UNI(OP_GGRGID); |
| 4416 | |
| 4417 | case KEY_getlogin: |
| 4418 | FUN0(OP_GETLOGIN); |
| 4419 | |
| 4420 | case KEY_glob: |
| 4421 | set_csh(); |
| 4422 | LOP(OP_GLOB,XTERM); |
| 4423 | |
| 4424 | case KEY_hex: |
| 4425 | UNI(OP_HEX); |
| 4426 | |
| 4427 | case KEY_if: |
| 4428 | yylval.ival = CopLINE(PL_curcop); |
| 4429 | OPERATOR(IF); |
| 4430 | |
| 4431 | case KEY_index: |
| 4432 | LOP(OP_INDEX,XTERM); |
| 4433 | |
| 4434 | case KEY_int: |
| 4435 | UNI(OP_INT); |
| 4436 | |
| 4437 | case KEY_ioctl: |
| 4438 | LOP(OP_IOCTL,XTERM); |
| 4439 | |
| 4440 | case KEY_join: |
| 4441 | LOP(OP_JOIN,XTERM); |
| 4442 | |
| 4443 | case KEY_keys: |
| 4444 | UNI(OP_KEYS); |
| 4445 | |
| 4446 | case KEY_kill: |
| 4447 | LOP(OP_KILL,XTERM); |
| 4448 | |
| 4449 | case KEY_last: |
| 4450 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
| 4451 | LOOPX(OP_LAST); |
| 4452 | |
| 4453 | case KEY_lc: |
| 4454 | UNI(OP_LC); |
| 4455 | |
| 4456 | case KEY_lcfirst: |
| 4457 | UNI(OP_LCFIRST); |
| 4458 | |
| 4459 | case KEY_local: |
| 4460 | yylval.ival = 0; |
| 4461 | OPERATOR(LOCAL); |
| 4462 | |
| 4463 | case KEY_length: |
| 4464 | UNI(OP_LENGTH); |
| 4465 | |
| 4466 | case KEY_lt: |
| 4467 | Rop(OP_SLT); |
| 4468 | |
| 4469 | case KEY_le: |
| 4470 | Rop(OP_SLE); |
| 4471 | |
| 4472 | case KEY_localtime: |
| 4473 | UNI(OP_LOCALTIME); |
| 4474 | |
| 4475 | case KEY_log: |
| 4476 | UNI(OP_LOG); |
| 4477 | |
| 4478 | case KEY_link: |
| 4479 | LOP(OP_LINK,XTERM); |
| 4480 | |
| 4481 | case KEY_listen: |
| 4482 | LOP(OP_LISTEN,XTERM); |
| 4483 | |
| 4484 | case KEY_lock: |
| 4485 | UNI(OP_LOCK); |
| 4486 | |
| 4487 | case KEY_lstat: |
| 4488 | UNI(OP_LSTAT); |
| 4489 | |
| 4490 | case KEY_m: |
| 4491 | s = scan_pat(s,OP_MATCH); |
| 4492 | TERM(sublex_start()); |
| 4493 | |
| 4494 | case KEY_map: |
| 4495 | LOP(OP_MAPSTART, XREF); |
| 4496 | |
| 4497 | case KEY_mkdir: |
| 4498 | LOP(OP_MKDIR,XTERM); |
| 4499 | |
| 4500 | case KEY_msgctl: |
| 4501 | LOP(OP_MSGCTL,XTERM); |
| 4502 | |
| 4503 | case KEY_msgget: |
| 4504 | LOP(OP_MSGGET,XTERM); |
| 4505 | |
| 4506 | case KEY_msgrcv: |
| 4507 | LOP(OP_MSGRCV,XTERM); |
| 4508 | |
| 4509 | case KEY_msgsnd: |
| 4510 | LOP(OP_MSGSND,XTERM); |
| 4511 | |
| 4512 | case KEY_our: |
| 4513 | case KEY_my: |
| 4514 | PL_in_my = tmp; |
| 4515 | s = skipspace(s); |
| 4516 | if (isIDFIRST_lazy_if(s,UTF)) { |
| 4517 | s = scan_word(s, PL_tokenbuf, sizeof PL_tokenbuf, TRUE, &len); |
| 4518 | if (len == 3 && strnEQ(PL_tokenbuf, "sub", 3)) |
| 4519 | goto really_sub; |
| 4520 | PL_in_my_stash = find_in_my_stash(PL_tokenbuf, len); |
| 4521 | if (!PL_in_my_stash) { |
| 4522 | char tmpbuf[1024]; |
| 4523 | PL_bufptr = s; |
| 4524 | sprintf(tmpbuf, "No such class %.1000s", PL_tokenbuf); |
| 4525 | yyerror(tmpbuf); |
| 4526 | } |
| 4527 | } |
| 4528 | yylval.ival = 1; |
| 4529 | OPERATOR(MY); |
| 4530 | |
| 4531 | case KEY_next: |
| 4532 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
| 4533 | LOOPX(OP_NEXT); |
| 4534 | |
| 4535 | case KEY_ne: |
| 4536 | Eop(OP_SNE); |
| 4537 | |
| 4538 | case KEY_no: |
| 4539 | if (PL_expect != XSTATE) |
| 4540 | yyerror("\"no\" not allowed in expression"); |
| 4541 | s = force_word(s,WORD,FALSE,TRUE,FALSE); |
| 4542 | s = force_version(s, FALSE); |
| 4543 | yylval.ival = 0; |
| 4544 | OPERATOR(USE); |
| 4545 | |
| 4546 | case KEY_not: |
| 4547 | if (*s == '(' || (s = skipspace(s), *s == '(')) |
| 4548 | FUN1(OP_NOT); |
| 4549 | else |
| 4550 | OPERATOR(NOTOP); |
| 4551 | |
| 4552 | case KEY_open: |
| 4553 | s = skipspace(s); |
| 4554 | if (isIDFIRST_lazy_if(s,UTF)) { |
| 4555 | char *t; |
| 4556 | for (d = s; isALNUM_lazy_if(d,UTF); d++) ; |
| 4557 | t = skipspace(d); |
| 4558 | if (strchr("|&*+-=!?:.", *t) && ckWARN_d(WARN_PRECEDENCE)) |
| 4559 | Perl_warner(aTHX_ WARN_PRECEDENCE, |
| 4560 | "Precedence problem: open %.*s should be open(%.*s)", |
| 4561 | d-s,s, d-s,s); |
| 4562 | } |
| 4563 | LOP(OP_OPEN,XTERM); |
| 4564 | |
| 4565 | case KEY_or: |
| 4566 | yylval.ival = OP_OR; |
| 4567 | OPERATOR(OROP); |
| 4568 | |
| 4569 | case KEY_ord: |
| 4570 | UNI(OP_ORD); |
| 4571 | |
| 4572 | case KEY_oct: |
| 4573 | UNI(OP_OCT); |
| 4574 | |
| 4575 | case KEY_opendir: |
| 4576 | LOP(OP_OPEN_DIR,XTERM); |
| 4577 | |
| 4578 | case KEY_print: |
| 4579 | checkcomma(s,PL_tokenbuf,"filehandle"); |
| 4580 | LOP(OP_PRINT,XREF); |
| 4581 | |
| 4582 | case KEY_printf: |
| 4583 | checkcomma(s,PL_tokenbuf,"filehandle"); |
| 4584 | LOP(OP_PRTF,XREF); |
| 4585 | |
| 4586 | case KEY_prototype: |
| 4587 | UNI(OP_PROTOTYPE); |
| 4588 | |
| 4589 | case KEY_push: |
| 4590 | LOP(OP_PUSH,XTERM); |
| 4591 | |
| 4592 | case KEY_pop: |
| 4593 | UNI(OP_POP); |
| 4594 | |
| 4595 | case KEY_pos: |
| 4596 | UNI(OP_POS); |
| 4597 | |
| 4598 | case KEY_pack: |
| 4599 | LOP(OP_PACK,XTERM); |
| 4600 | |
| 4601 | case KEY_package: |
| 4602 | s = force_word(s,WORD,FALSE,TRUE,FALSE); |
| 4603 | OPERATOR(PACKAGE); |
| 4604 | |
| 4605 | case KEY_pipe: |
| 4606 | LOP(OP_PIPE_OP,XTERM); |
| 4607 | |
| 4608 | case KEY_q: |
| 4609 | s = scan_str(s,FALSE,FALSE); |
| 4610 | if (!s) |
| 4611 | missingterm((char*)0); |
| 4612 | yylval.ival = OP_CONST; |
| 4613 | TERM(sublex_start()); |
| 4614 | |
| 4615 | case KEY_quotemeta: |
| 4616 | UNI(OP_QUOTEMETA); |
| 4617 | |
| 4618 | case KEY_qw: |
| 4619 | s = scan_str(s,FALSE,FALSE); |
| 4620 | if (!s) |
| 4621 | missingterm((char*)0); |
| 4622 | force_next(')'); |
| 4623 | if (SvCUR(PL_lex_stuff)) { |
| 4624 | OP *words = Nullop; |
| 4625 | int warned = 0; |
| 4626 | d = SvPV_force(PL_lex_stuff, len); |
| 4627 | while (len) { |
| 4628 | SV *sv; |
| 4629 | for (; isSPACE(*d) && len; --len, ++d) ; |
| 4630 | if (len) { |
| 4631 | char *b = d; |
| 4632 | if (!warned && ckWARN(WARN_QW)) { |
| 4633 | for (; !isSPACE(*d) && len; --len, ++d) { |
| 4634 | if (*d == ',') { |
| 4635 | Perl_warner(aTHX_ WARN_QW, |
| 4636 | "Possible attempt to separate words with commas"); |
| 4637 | ++warned; |
| 4638 | } |
| 4639 | else if (*d == '#') { |
| 4640 | Perl_warner(aTHX_ WARN_QW, |
| 4641 | "Possible attempt to put comments in qw() list"); |
| 4642 | ++warned; |
| 4643 | } |
| 4644 | } |
| 4645 | } |
| 4646 | else { |
| 4647 | for (; !isSPACE(*d) && len; --len, ++d) ; |
| 4648 | } |
| 4649 | sv = newSVpvn(b, d-b); |
| 4650 | if (DO_UTF8(PL_lex_stuff)) |
| 4651 | SvUTF8_on(sv); |
| 4652 | words = append_elem(OP_LIST, words, |
| 4653 | newSVOP(OP_CONST, 0, tokeq(sv))); |
| 4654 | } |
| 4655 | } |
| 4656 | if (words) { |
| 4657 | PL_nextval[PL_nexttoke].opval = words; |
| 4658 | force_next(THING); |
| 4659 | } |
| 4660 | } |
| 4661 | if (PL_lex_stuff) { |
| 4662 | SvREFCNT_dec(PL_lex_stuff); |
| 4663 | PL_lex_stuff = Nullsv; |
| 4664 | } |
| 4665 | PL_expect = XTERM; |
| 4666 | TOKEN('('); |
| 4667 | |
| 4668 | case KEY_qq: |
| 4669 | s = scan_str(s,FALSE,FALSE); |
| 4670 | if (!s) |
| 4671 | missingterm((char*)0); |
| 4672 | yylval.ival = OP_STRINGIFY; |
| 4673 | if (SvIVX(PL_lex_stuff) == '\'') |
| 4674 | SvIVX(PL_lex_stuff) = 0; /* qq'$foo' should intepolate */ |
| 4675 | TERM(sublex_start()); |
| 4676 | |
| 4677 | case KEY_qr: |
| 4678 | s = scan_pat(s,OP_QR); |
| 4679 | TERM(sublex_start()); |
| 4680 | |
| 4681 | case KEY_qx: |
| 4682 | s = scan_str(s,FALSE,FALSE); |
| 4683 | if (!s) |
| 4684 | missingterm((char*)0); |
| 4685 | yylval.ival = OP_BACKTICK; |
| 4686 | set_csh(); |
| 4687 | TERM(sublex_start()); |
| 4688 | |
| 4689 | case KEY_return: |
| 4690 | OLDLOP(OP_RETURN); |
| 4691 | |
| 4692 | case KEY_require: |
| 4693 | s = skipspace(s); |
| 4694 | if (isDIGIT(*s)) { |
| 4695 | s = force_version(s, FALSE); |
| 4696 | } |
| 4697 | else if (*s != 'v' || !isDIGIT(s[1]) |
| 4698 | || (s = force_version(s, TRUE), *s == 'v')) |
| 4699 | { |
| 4700 | *PL_tokenbuf = '\0'; |
| 4701 | s = force_word(s,WORD,TRUE,TRUE,FALSE); |
| 4702 | if (isIDFIRST_lazy_if(PL_tokenbuf,UTF)) |
| 4703 | gv_stashpvn(PL_tokenbuf, strlen(PL_tokenbuf), TRUE); |
| 4704 | else if (*s == '<') |
| 4705 | yyerror("<> should be quotes"); |
| 4706 | } |
| 4707 | UNI(OP_REQUIRE); |
| 4708 | |
| 4709 | case KEY_reset: |
| 4710 | UNI(OP_RESET); |
| 4711 | |
| 4712 | case KEY_redo: |
| 4713 | s = force_word(s,WORD,TRUE,FALSE,FALSE); |
| 4714 | LOOPX(OP_REDO); |
| 4715 | |
| 4716 | case KEY_rename: |
| 4717 | LOP(OP_RENAME,XTERM); |
| 4718 | |
| 4719 | case KEY_rand: |
| 4720 | UNI(OP_RAND); |
| 4721 | |
| 4722 | case KEY_rmdir: |
| 4723 | UNI(OP_RMDIR); |
| 4724 | |
| 4725 | case KEY_rindex: |
| 4726 | LOP(OP_RINDEX,XTERM); |
| 4727 | |
| 4728 | case KEY_read: |
| 4729 | LOP(OP_READ,XTERM); |
| 4730 | |
| 4731 | case KEY_readdir: |
| 4732 | UNI(OP_READDIR); |
| 4733 | |
| 4734 | case KEY_readline: |
| 4735 | set_csh(); |
| 4736 | UNI(OP_READLINE); |
| 4737 | |
| 4738 | case KEY_readpipe: |
| 4739 | set_csh(); |
| 4740 | UNI(OP_BACKTICK); |
| 4741 | |
| 4742 | case KEY_rewinddir: |
| 4743 | UNI(OP_REWINDDIR); |
| 4744 | |
| 4745 | case KEY_recv: |
| 4746 | LOP(OP_RECV,XTERM); |
| 4747 | |
| 4748 | case KEY_reverse: |
| 4749 | LOP(OP_REVERSE,XTERM); |
| 4750 | |
| 4751 | case KEY_readlink: |
| 4752 | UNI(OP_READLINK); |
| 4753 | |
| 4754 | case KEY_ref: |
| 4755 | UNI(OP_REF); |
| 4756 | |
| 4757 | case KEY_s: |
| 4758 | s = scan_subst(s); |
| 4759 | if (yylval.opval) |
| 4760 | TERM(sublex_start()); |
| 4761 | else |
| 4762 | TOKEN(1); /* force error */ |
| 4763 | |
| 4764 | case KEY_chomp: |
| 4765 | UNI(OP_CHOMP); |
| 4766 | |
| 4767 | case KEY_scalar: |
| 4768 | UNI(OP_SCALAR); |
| 4769 | |
| 4770 | case KEY_select: |
| 4771 | LOP(OP_SELECT,XTERM); |
| 4772 | |
| 4773 | case KEY_seek: |
| 4774 | LOP(OP_SEEK,XTERM); |
| 4775 | |
| 4776 | case KEY_semctl: |
| 4777 | LOP(OP_SEMCTL,XTERM); |
| 4778 | |
| 4779 | case KEY_semget: |
| 4780 | LOP(OP_SEMGET,XTERM); |
| 4781 | |
| 4782 | case KEY_semop: |
| 4783 | LOP(OP_SEMOP,XTERM); |
| 4784 | |
| 4785 | case KEY_send: |
| 4786 | LOP(OP_SEND,XTERM); |
| 4787 | |
| 4788 | case KEY_setpgrp: |
| 4789 | LOP(OP_SETPGRP,XTERM); |
| 4790 | |
| 4791 | case KEY_setpriority: |
| 4792 | LOP(OP_SETPRIORITY,XTERM); |
| 4793 | |
| 4794 | case KEY_sethostent: |
| 4795 | UNI(OP_SHOSTENT); |
| 4796 | |
| 4797 | case KEY_setnetent: |
| 4798 | UNI(OP_SNETENT); |
| 4799 | |
| 4800 | case KEY_setservent: |
| 4801 | UNI(OP_SSERVENT); |
| 4802 | |
| 4803 | case KEY_setprotoent: |
| 4804 | UNI(OP_SPROTOENT); |
| 4805 | |
| 4806 | case KEY_setpwent: |
| 4807 | FUN0(OP_SPWENT); |
| 4808 | |
| 4809 | case KEY_setgrent: |
| 4810 | FUN0(OP_SGRENT); |
| 4811 | |
| 4812 | case KEY_seekdir: |
| 4813 | LOP(OP_SEEKDIR,XTERM); |
| 4814 | |
| 4815 | case KEY_setsockopt: |
| 4816 | LOP(OP_SSOCKOPT,XTERM); |
| 4817 | |
| 4818 | case KEY_shift: |
| 4819 | UNI(OP_SHIFT); |
| 4820 | |
| 4821 | case KEY_shmctl: |
| 4822 | LOP(OP_SHMCTL,XTERM); |
| 4823 | |
| 4824 | case KEY_shmget: |
| 4825 | LOP(OP_SHMGET,XTERM); |
| 4826 | |
| 4827 | case KEY_shmread: |
| 4828 | LOP(OP_SHMREAD,XTERM); |
| 4829 | |
| 4830 | case KEY_shmwrite: |
| 4831 | LOP(OP_SHMWRITE,XTERM); |
| 4832 | |
| 4833 | case KEY_shutdown: |
| 4834 | LOP(OP_SHUTDOWN,XTERM); |
| 4835 | |
| 4836 | case KEY_sin: |
| 4837 | UNI(OP_SIN); |
| 4838 | |
| 4839 | case KEY_sleep: |
| 4840 | UNI(OP_SLEEP); |
| 4841 | |
| 4842 | case KEY_socket: |
| 4843 | LOP(OP_SOCKET,XTERM); |
| 4844 | |
| 4845 | case KEY_socketpair: |
| 4846 | LOP(OP_SOCKPAIR,XTERM); |
| 4847 | |
| 4848 | case KEY_sort: |
| 4849 | checkcomma(s,PL_tokenbuf,"subroutine name"); |
| 4850 | s = skipspace(s); |
| 4851 | if (*s == ';' || *s == ')') /* probably a close */ |
| 4852 | Perl_croak(aTHX_ "sort is now a reserved word"); |
| 4853 | PL_expect = XTERM; |
| 4854 | s = force_word(s,WORD,TRUE,TRUE,FALSE); |
| 4855 | LOP(OP_SORT,XREF); |
| 4856 | |
| 4857 | case KEY_split: |
| 4858 | LOP(OP_SPLIT,XTERM); |
| 4859 | |
| 4860 | case KEY_sprintf: |
| 4861 | LOP(OP_SPRINTF,XTERM); |
| 4862 | |
| 4863 | case KEY_splice: |
| 4864 | LOP(OP_SPLICE,XTERM); |
| 4865 | |
| 4866 | case KEY_sqrt: |
| 4867 | UNI(OP_SQRT); |
| 4868 | |
| 4869 | case KEY_srand: |
| 4870 | UNI(OP_SRAND); |
| 4871 | |
| 4872 | case KEY_stat: |
| 4873 | UNI(OP_STAT); |
| 4874 | |
| 4875 | case KEY_study: |
| 4876 | UNI(OP_STUDY); |
| 4877 | |
| 4878 | case KEY_substr: |
| 4879 | LOP(OP_SUBSTR,XTERM); |
| 4880 | |
| 4881 | case KEY_format: |
| 4882 | case KEY_sub: |
| 4883 | really_sub: |
| 4884 | { |
| 4885 | char tmpbuf[sizeof PL_tokenbuf]; |
| 4886 | SSize_t tboffset = 0; |
| 4887 | expectation attrful; |
| 4888 | bool have_name, have_proto; |
| 4889 | int key = tmp; |
| 4890 | |
| 4891 | s = skipspace(s); |
| 4892 | |
| 4893 | if (isIDFIRST_lazy_if(s,UTF) || *s == '\'' || |
| 4894 | (*s == ':' && s[1] == ':')) |
| 4895 | { |
| 4896 | PL_expect = XBLOCK; |
| 4897 | attrful = XATTRBLOCK; |
| 4898 | /* remember buffer pos'n for later force_word */ |
| 4899 | tboffset = s - PL_oldbufptr; |
| 4900 | d = scan_word(s, tmpbuf, sizeof tmpbuf, TRUE, &len); |
| 4901 | if (strchr(tmpbuf, ':')) |
| 4902 | sv_setpv(PL_subname, tmpbuf); |
| 4903 | else { |
| 4904 | sv_setsv(PL_subname,PL_curstname); |
| 4905 | sv_catpvn(PL_subname,"::",2); |
| 4906 | sv_catpvn(PL_subname,tmpbuf,len); |
| 4907 | } |
| 4908 | s = skipspace(d); |
| 4909 | have_name = TRUE; |
| 4910 | } |
| 4911 | else { |
| 4912 | if (key == KEY_my) |
| 4913 | Perl_croak(aTHX_ "Missing name in \"my sub\""); |
| 4914 | PL_expect = XTERMBLOCK; |
| 4915 | attrful = XATTRTERM; |
| 4916 | sv_setpv(PL_subname,"?"); |
| 4917 | have_name = FALSE; |
| 4918 | } |
| 4919 | |
| 4920 | if (key == KEY_format) { |
| 4921 | if (*s == '=') |
| 4922 | PL_lex_formbrack = PL_lex_brackets + 1; |
| 4923 | if (have_name) |
| 4924 | (void) force_word(PL_oldbufptr + tboffset, WORD, |
| 4925 | FALSE, TRUE, TRUE); |
| 4926 | OPERATOR(FORMAT); |
| 4927 | } |
| 4928 | |
| 4929 | /* Look for a prototype */ |
| 4930 | if (*s == '(') { |
| 4931 | char *p; |
| 4932 | |
| 4933 | s = scan_str(s,FALSE,FALSE); |
| 4934 | if (!s) |
| 4935 | Perl_croak(aTHX_ "Prototype not terminated"); |
| 4936 | /* strip spaces */ |
| 4937 | d = SvPVX(PL_lex_stuff); |
| 4938 | tmp = 0; |
| 4939 | for (p = d; *p; ++p) { |
| 4940 | if (!isSPACE(*p)) |
| 4941 | d[tmp++] = *p; |
| 4942 | } |
| 4943 | d[tmp] = '\0'; |
| 4944 | SvCUR(PL_lex_stuff) = tmp; |
| 4945 | have_proto = TRUE; |
| 4946 | |
| 4947 | s = skipspace(s); |
| 4948 | } |
| 4949 | else |
| 4950 | have_proto = FALSE; |
| 4951 | |
| 4952 | if (*s == ':' && s[1] != ':') |
| 4953 | PL_expect = attrful; |
| 4954 | |
| 4955 | if (have_proto) { |
| 4956 | PL_nextval[PL_nexttoke].opval = |
| 4957 | (OP*)newSVOP(OP_CONST, 0, PL_lex_stuff); |
| 4958 | PL_lex_stuff = Nullsv; |
| 4959 | force_next(THING); |
| 4960 | } |
| 4961 | if (!have_name) { |
| 4962 | sv_setpv(PL_subname,"__ANON__"); |
| 4963 | TOKEN(ANONSUB); |
| 4964 | } |
| 4965 | (void) force_word(PL_oldbufptr + tboffset, WORD, |
| 4966 | FALSE, TRUE, TRUE); |
| 4967 | if (key == KEY_my) |
| 4968 | TOKEN(MYSUB); |
| 4969 | TOKEN(SUB); |
| 4970 | } |
| 4971 | |
| 4972 | case KEY_system: |
| 4973 | set_csh(); |
| 4974 | LOP(OP_SYSTEM,XREF); |
| 4975 | |
| 4976 | case KEY_symlink: |
| 4977 | LOP(OP_SYMLINK,XTERM); |
| 4978 | |
| 4979 | case KEY_syscall: |
| 4980 | LOP(OP_SYSCALL,XTERM); |
| 4981 | |
| 4982 | case KEY_sysopen: |
| 4983 | LOP(OP_SYSOPEN,XTERM); |
| 4984 | |
| 4985 | case KEY_sysseek: |
| 4986 | LOP(OP_SYSSEEK,XTERM); |
| 4987 | |
| 4988 | case KEY_sysread: |
| 4989 | LOP(OP_SYSREAD,XTERM); |
| 4990 | |
| 4991 | case KEY_syswrite: |
| 4992 | LOP(OP_SYSWRITE,XTERM); |
| 4993 | |
| 4994 | case KEY_tr: |
| 4995 | s = scan_trans(s); |
| 4996 | TERM(sublex_start()); |
| 4997 | |
| 4998 | case KEY_tell: |
| 4999 | UNI(OP_TELL); |
| 5000 | |
| 5001 | case KEY_telldir: |
| 5002 | UNI(OP_TELLDIR); |
| 5003 | |
| 5004 | case KEY_tie: |
| 5005 | LOP(OP_TIE,XTERM); |
| 5006 | |
| 5007 | case KEY_tied: |
| 5008 | UNI(OP_TIED); |
| 5009 | |
| 5010 | case KEY_time: |
| 5011 | FUN0(OP_TIME); |
| 5012 | |
| 5013 | case KEY_times: |
| 5014 | FUN0(OP_TMS); |
| 5015 | |
| 5016 | case KEY_truncate: |
| 5017 | LOP(OP_TRUNCATE,XTERM); |
| 5018 | |
| 5019 | case KEY_uc: |
| 5020 | UNI(OP_UC); |
| 5021 | |
| 5022 | case KEY_ucfirst: |
| 5023 | UNI(OP_UCFIRST); |
| 5024 | |
| 5025 | case KEY_untie: |
| 5026 | UNI(OP_UNTIE); |
| 5027 | |
| 5028 | case KEY_until: |
| 5029 | yylval.ival = CopLINE(PL_curcop); |
| 5030 | OPERATOR(UNTIL); |
| 5031 | |
| 5032 | case KEY_unless: |
| 5033 | yylval.ival = CopLINE(PL_curcop); |
| 5034 | OPERATOR(UNLESS); |
| 5035 | |
| 5036 | case KEY_unlink: |
| 5037 | LOP(OP_UNLINK,XTERM); |
| 5038 | |
| 5039 | case KEY_undef: |
| 5040 | UNI(OP_UNDEF); |
| 5041 | |
| 5042 | case KEY_unpack: |
| 5043 | LOP(OP_UNPACK,XTERM); |
| 5044 | |
| 5045 | case KEY_utime: |
| 5046 | LOP(OP_UTIME,XTERM); |
| 5047 | |
| 5048 | case KEY_umask: |
| 5049 | UNI(OP_UMASK); |
| 5050 | |
| 5051 | case KEY_unshift: |
| 5052 | LOP(OP_UNSHIFT,XTERM); |
| 5053 | |
| 5054 | case KEY_use: |
| 5055 | if (PL_expect != XSTATE) |
| 5056 | yyerror("\"use\" not allowed in expression"); |
| 5057 | s = skipspace(s); |
| 5058 | if (isDIGIT(*s) || (*s == 'v' && isDIGIT(s[1]))) { |
| 5059 | s = force_version(s, TRUE); |
| 5060 | if (*s == ';' || (s = skipspace(s), *s == ';')) { |
| 5061 | PL_nextval[PL_nexttoke].opval = Nullop; |
| 5062 | force_next(WORD); |
| 5063 | } |
| 5064 | else if (*s == 'v') { |
| 5065 | s = force_word(s,WORD,FALSE,TRUE,FALSE); |
| 5066 | s = force_version(s, FALSE); |
| 5067 | } |
| 5068 | } |
| 5069 | else { |
| 5070 | s = force_word(s,WORD,FALSE,TRUE,FALSE); |
| 5071 | s = force_version(s, FALSE); |
| 5072 | } |
| 5073 | yylval.ival = 1; |
| 5074 | OPERATOR(USE); |
| 5075 | |
| 5076 | case KEY_values: |
| 5077 | UNI(OP_VALUES); |
| 5078 | |
| 5079 | case KEY_vec: |
| 5080 | LOP(OP_VEC,XTERM); |
| 5081 | |
| 5082 | case KEY_while: |
| 5083 | yylval.ival = CopLINE(PL_curcop); |
| 5084 | OPERATOR(WHILE); |
| 5085 | |
| 5086 | case KEY_warn: |
| 5087 | PL_hints |= HINT_BLOCK_SCOPE; |
| 5088 | LOP(OP_WARN,XTERM); |
| 5089 | |
| 5090 | case KEY_wait: |
| 5091 | FUN0(OP_WAIT); |
| 5092 | |
| 5093 | case KEY_waitpid: |
| 5094 | LOP(OP_WAITPID,XTERM); |
| 5095 | |
| 5096 | case KEY_wantarray: |
| 5097 | FUN0(OP_WANTARRAY); |
| 5098 | |
| 5099 | case KEY_write: |
| 5100 | #ifdef EBCDIC |
| 5101 | { |
| 5102 | static char ctl_l[2]; |
| 5103 | |
| 5104 | if (ctl_l[0] == '\0') |
| 5105 | ctl_l[0] = toCTRL('L'); |
| 5106 | gv_fetchpv(ctl_l,TRUE, SVt_PV); |
| 5107 | } |
| 5108 | #else |
| 5109 | gv_fetchpv("\f",TRUE, SVt_PV); /* Make sure $^L is defined */ |
| 5110 | #endif |
| 5111 | UNI(OP_ENTERWRITE); |
| 5112 | |
| 5113 | case KEY_x: |
| 5114 | if (PL_expect == XOPERATOR) |
| 5115 | Mop(OP_REPEAT); |
| 5116 | check_uni(); |
| 5117 | goto just_a_word; |
| 5118 | |
| 5119 | case KEY_xor: |
| 5120 | yylval.ival = OP_XOR; |
| 5121 | OPERATOR(OROP); |
| 5122 | |
| 5123 | case KEY_y: |
| 5124 | s = scan_trans(s); |
| 5125 | TERM(sublex_start()); |
| 5126 | } |
| 5127 | }} |
| 5128 | } |
| 5129 | #ifdef __SC__ |
| 5130 | #pragma segment Main |
| 5131 | #endif |
| 5132 | |
| 5133 | static int |
| 5134 | S_pending_ident(pTHX) |
| 5135 | { |
| 5136 | register char *d; |
| 5137 | register I32 tmp; |
| 5138 | /* pit holds the identifier we read and pending_ident is reset */ |
| 5139 | char pit = PL_pending_ident; |
| 5140 | PL_pending_ident = 0; |
| 5141 | |
| 5142 | DEBUG_T({ PerlIO_printf(Perl_debug_log, |
| 5143 | "### Tokener saw identifier '%s'\n", PL_tokenbuf); }); |
| 5144 | |
| 5145 | /* if we're in a my(), we can't allow dynamics here. |
| 5146 | $foo'bar has already been turned into $foo::bar, so |
| 5147 | just check for colons. |
| 5148 | |
| 5149 | if it's a legal name, the OP is a PADANY. |
| 5150 | */ |
| 5151 | if (PL_in_my) { |
| 5152 | if (PL_in_my == KEY_our) { /* "our" is merely analogous to "my" */ |
| 5153 | if (strchr(PL_tokenbuf,':')) |
| 5154 | yyerror(Perl_form(aTHX_ "No package name allowed for " |
| 5155 | "variable %s in \"our\"", |
| 5156 | PL_tokenbuf)); |
| 5157 | tmp = pad_allocmy(PL_tokenbuf); |
| 5158 | } |
| 5159 | else { |
| 5160 | if (strchr(PL_tokenbuf,':')) |
| 5161 | yyerror(Perl_form(aTHX_ PL_no_myglob,PL_tokenbuf)); |
| 5162 | |
| 5163 | yylval.opval = newOP(OP_PADANY, 0); |
| 5164 | yylval.opval->op_targ = pad_allocmy(PL_tokenbuf); |
| 5165 | return PRIVATEREF; |
| 5166 | } |
| 5167 | } |
| 5168 | |
| 5169 | /* |
| 5170 | build the ops for accesses to a my() variable. |
| 5171 | |
| 5172 | Deny my($a) or my($b) in a sort block, *if* $a or $b is |
| 5173 | then used in a comparison. This catches most, but not |
| 5174 | all cases. For instance, it catches |
| 5175 | sort { my($a); $a <=> $b } |
| 5176 | but not |
| 5177 | sort { my($a); $a < $b ? -1 : $a == $b ? 0 : 1; } |
| 5178 | (although why you'd do that is anyone's guess). |
| 5179 | */ |
| 5180 | |
| 5181 | if (!strchr(PL_tokenbuf,':')) { |
| 5182 | #ifdef USE_5005THREADS |
| 5183 | /* Check for single character per-thread SVs */ |
| 5184 | if (PL_tokenbuf[0] == '$' && PL_tokenbuf[2] == '\0' |
| 5185 | && !isALPHA(PL_tokenbuf[1]) /* Rule out obvious non-threadsvs */ |
| 5186 | && (tmp = find_threadsv(&PL_tokenbuf[1])) != NOT_IN_PAD) |
| 5187 | { |
| 5188 | yylval.opval = newOP(OP_THREADSV, 0); |
| 5189 | yylval.opval->op_targ = tmp; |
| 5190 | return PRIVATEREF; |
| 5191 | } |
| 5192 | #endif /* USE_5005THREADS */ |
| 5193 | if ((tmp = pad_findmy(PL_tokenbuf)) != NOT_IN_PAD) { |
| 5194 | SV *namesv = AvARRAY(PL_comppad_name)[tmp]; |
| 5195 | /* might be an "our" variable" */ |
| 5196 | if (SvFLAGS(namesv) & SVpad_OUR) { |
| 5197 | /* build ops for a bareword */ |
| 5198 | SV *sym = newSVpv(HvNAME(GvSTASH(namesv)),0); |
| 5199 | sv_catpvn(sym, "::", 2); |
| 5200 | sv_catpv(sym, PL_tokenbuf+1); |
| 5201 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, sym); |
| 5202 | yylval.opval->op_private = OPpCONST_ENTERED; |
| 5203 | gv_fetchpv(SvPVX(sym), |
| 5204 | (PL_in_eval |
| 5205 | ? (GV_ADDMULTI | GV_ADDINEVAL) |
| 5206 | : TRUE |
| 5207 | ), |
| 5208 | ((PL_tokenbuf[0] == '$') ? SVt_PV |
| 5209 | : (PL_tokenbuf[0] == '@') ? SVt_PVAV |
| 5210 | : SVt_PVHV)); |
| 5211 | return WORD; |
| 5212 | } |
| 5213 | |
| 5214 | /* if it's a sort block and they're naming $a or $b */ |
| 5215 | if (PL_last_lop_op == OP_SORT && |
| 5216 | PL_tokenbuf[0] == '$' && |
| 5217 | (PL_tokenbuf[1] == 'a' || PL_tokenbuf[1] == 'b') |
| 5218 | && !PL_tokenbuf[2]) |
| 5219 | { |
| 5220 | for (d = PL_in_eval ? PL_oldoldbufptr : PL_linestart; |
| 5221 | d < PL_bufend && *d != '\n'; |
| 5222 | d++) |
| 5223 | { |
| 5224 | if (strnEQ(d,"<=>",3) || strnEQ(d,"cmp",3)) { |
| 5225 | Perl_croak(aTHX_ "Can't use \"my %s\" in sort comparison", |
| 5226 | PL_tokenbuf); |
| 5227 | } |
| 5228 | } |
| 5229 | } |
| 5230 | |
| 5231 | yylval.opval = newOP(OP_PADANY, 0); |
| 5232 | yylval.opval->op_targ = tmp; |
| 5233 | return PRIVATEREF; |
| 5234 | } |
| 5235 | } |
| 5236 | |
| 5237 | /* |
| 5238 | Whine if they've said @foo in a doublequoted string, |
| 5239 | and @foo isn't a variable we can find in the symbol |
| 5240 | table. |
| 5241 | */ |
| 5242 | if (pit == '@' && PL_lex_state != LEX_NORMAL && !PL_lex_brackets) { |
| 5243 | GV *gv = gv_fetchpv(PL_tokenbuf+1, FALSE, SVt_PVAV); |
| 5244 | if ((!gv || ((PL_tokenbuf[0] == '@') ? !GvAV(gv) : !GvHV(gv))) |
| 5245 | && ckWARN(WARN_AMBIGUOUS)) |
| 5246 | { |
| 5247 | /* Downgraded from fatal to warning 20000522 mjd */ |
| 5248 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 5249 | "Possible unintended interpolation of %s in string", |
| 5250 | PL_tokenbuf); |
| 5251 | } |
| 5252 | } |
| 5253 | |
| 5254 | /* build ops for a bareword */ |
| 5255 | yylval.opval = (OP*)newSVOP(OP_CONST, 0, newSVpv(PL_tokenbuf+1, 0)); |
| 5256 | yylval.opval->op_private = OPpCONST_ENTERED; |
| 5257 | gv_fetchpv(PL_tokenbuf+1, PL_in_eval ? (GV_ADDMULTI | GV_ADDINEVAL) : TRUE, |
| 5258 | ((PL_tokenbuf[0] == '$') ? SVt_PV |
| 5259 | : (PL_tokenbuf[0] == '@') ? SVt_PVAV |
| 5260 | : SVt_PVHV)); |
| 5261 | return WORD; |
| 5262 | } |
| 5263 | |
| 5264 | I32 |
| 5265 | Perl_keyword(pTHX_ register char *d, I32 len) |
| 5266 | { |
| 5267 | switch (*d) { |
| 5268 | case '_': |
| 5269 | if (d[1] == '_') { |
| 5270 | if (strEQ(d,"__FILE__")) return -KEY___FILE__; |
| 5271 | if (strEQ(d,"__LINE__")) return -KEY___LINE__; |
| 5272 | if (strEQ(d,"__PACKAGE__")) return -KEY___PACKAGE__; |
| 5273 | if (strEQ(d,"__DATA__")) return KEY___DATA__; |
| 5274 | if (strEQ(d,"__END__")) return KEY___END__; |
| 5275 | } |
| 5276 | break; |
| 5277 | case 'A': |
| 5278 | if (strEQ(d,"AUTOLOAD")) return KEY_AUTOLOAD; |
| 5279 | break; |
| 5280 | case 'a': |
| 5281 | switch (len) { |
| 5282 | case 3: |
| 5283 | if (strEQ(d,"and")) return -KEY_and; |
| 5284 | if (strEQ(d,"abs")) return -KEY_abs; |
| 5285 | break; |
| 5286 | case 5: |
| 5287 | if (strEQ(d,"alarm")) return -KEY_alarm; |
| 5288 | if (strEQ(d,"atan2")) return -KEY_atan2; |
| 5289 | break; |
| 5290 | case 6: |
| 5291 | if (strEQ(d,"accept")) return -KEY_accept; |
| 5292 | break; |
| 5293 | } |
| 5294 | break; |
| 5295 | case 'B': |
| 5296 | if (strEQ(d,"BEGIN")) return KEY_BEGIN; |
| 5297 | break; |
| 5298 | case 'b': |
| 5299 | if (strEQ(d,"bless")) return -KEY_bless; |
| 5300 | if (strEQ(d,"bind")) return -KEY_bind; |
| 5301 | if (strEQ(d,"binmode")) return -KEY_binmode; |
| 5302 | break; |
| 5303 | case 'C': |
| 5304 | if (strEQ(d,"CORE")) return -KEY_CORE; |
| 5305 | if (strEQ(d,"CHECK")) return KEY_CHECK; |
| 5306 | break; |
| 5307 | case 'c': |
| 5308 | switch (len) { |
| 5309 | case 3: |
| 5310 | if (strEQ(d,"cmp")) return -KEY_cmp; |
| 5311 | if (strEQ(d,"chr")) return -KEY_chr; |
| 5312 | if (strEQ(d,"cos")) return -KEY_cos; |
| 5313 | break; |
| 5314 | case 4: |
| 5315 | if (strEQ(d,"chop")) return -KEY_chop; |
| 5316 | break; |
| 5317 | case 5: |
| 5318 | if (strEQ(d,"close")) return -KEY_close; |
| 5319 | if (strEQ(d,"chdir")) return -KEY_chdir; |
| 5320 | if (strEQ(d,"chomp")) return -KEY_chomp; |
| 5321 | if (strEQ(d,"chmod")) return -KEY_chmod; |
| 5322 | if (strEQ(d,"chown")) return -KEY_chown; |
| 5323 | if (strEQ(d,"crypt")) return -KEY_crypt; |
| 5324 | break; |
| 5325 | case 6: |
| 5326 | if (strEQ(d,"chroot")) return -KEY_chroot; |
| 5327 | if (strEQ(d,"caller")) return -KEY_caller; |
| 5328 | break; |
| 5329 | case 7: |
| 5330 | if (strEQ(d,"connect")) return -KEY_connect; |
| 5331 | break; |
| 5332 | case 8: |
| 5333 | if (strEQ(d,"closedir")) return -KEY_closedir; |
| 5334 | if (strEQ(d,"continue")) return -KEY_continue; |
| 5335 | break; |
| 5336 | } |
| 5337 | break; |
| 5338 | case 'D': |
| 5339 | if (strEQ(d,"DESTROY")) return KEY_DESTROY; |
| 5340 | break; |
| 5341 | case 'd': |
| 5342 | switch (len) { |
| 5343 | case 2: |
| 5344 | if (strEQ(d,"do")) return KEY_do; |
| 5345 | break; |
| 5346 | case 3: |
| 5347 | if (strEQ(d,"die")) return -KEY_die; |
| 5348 | break; |
| 5349 | case 4: |
| 5350 | if (strEQ(d,"dump")) return -KEY_dump; |
| 5351 | break; |
| 5352 | case 6: |
| 5353 | if (strEQ(d,"delete")) return KEY_delete; |
| 5354 | break; |
| 5355 | case 7: |
| 5356 | if (strEQ(d,"defined")) return KEY_defined; |
| 5357 | if (strEQ(d,"dbmopen")) return -KEY_dbmopen; |
| 5358 | break; |
| 5359 | case 8: |
| 5360 | if (strEQ(d,"dbmclose")) return -KEY_dbmclose; |
| 5361 | break; |
| 5362 | } |
| 5363 | break; |
| 5364 | case 'E': |
| 5365 | if (strEQ(d,"END")) return KEY_END; |
| 5366 | break; |
| 5367 | case 'e': |
| 5368 | switch (len) { |
| 5369 | case 2: |
| 5370 | if (strEQ(d,"eq")) return -KEY_eq; |
| 5371 | break; |
| 5372 | case 3: |
| 5373 | if (strEQ(d,"eof")) return -KEY_eof; |
| 5374 | if (strEQ(d,"exp")) return -KEY_exp; |
| 5375 | break; |
| 5376 | case 4: |
| 5377 | if (strEQ(d,"else")) return KEY_else; |
| 5378 | if (strEQ(d,"exit")) return -KEY_exit; |
| 5379 | if (strEQ(d,"eval")) return KEY_eval; |
| 5380 | if (strEQ(d,"exec")) return -KEY_exec; |
| 5381 | if (strEQ(d,"each")) return -KEY_each; |
| 5382 | break; |
| 5383 | case 5: |
| 5384 | if (strEQ(d,"elsif")) return KEY_elsif; |
| 5385 | break; |
| 5386 | case 6: |
| 5387 | if (strEQ(d,"exists")) return KEY_exists; |
| 5388 | if (strEQ(d,"elseif")) Perl_warn(aTHX_ "elseif should be elsif"); |
| 5389 | break; |
| 5390 | case 8: |
| 5391 | if (strEQ(d,"endgrent")) return -KEY_endgrent; |
| 5392 | if (strEQ(d,"endpwent")) return -KEY_endpwent; |
| 5393 | break; |
| 5394 | case 9: |
| 5395 | if (strEQ(d,"endnetent")) return -KEY_endnetent; |
| 5396 | break; |
| 5397 | case 10: |
| 5398 | if (strEQ(d,"endhostent")) return -KEY_endhostent; |
| 5399 | if (strEQ(d,"endservent")) return -KEY_endservent; |
| 5400 | break; |
| 5401 | case 11: |
| 5402 | if (strEQ(d,"endprotoent")) return -KEY_endprotoent; |
| 5403 | break; |
| 5404 | } |
| 5405 | break; |
| 5406 | case 'f': |
| 5407 | switch (len) { |
| 5408 | case 3: |
| 5409 | if (strEQ(d,"for")) return KEY_for; |
| 5410 | break; |
| 5411 | case 4: |
| 5412 | if (strEQ(d,"fork")) return -KEY_fork; |
| 5413 | break; |
| 5414 | case 5: |
| 5415 | if (strEQ(d,"fcntl")) return -KEY_fcntl; |
| 5416 | if (strEQ(d,"flock")) return -KEY_flock; |
| 5417 | break; |
| 5418 | case 6: |
| 5419 | if (strEQ(d,"format")) return KEY_format; |
| 5420 | if (strEQ(d,"fileno")) return -KEY_fileno; |
| 5421 | break; |
| 5422 | case 7: |
| 5423 | if (strEQ(d,"foreach")) return KEY_foreach; |
| 5424 | break; |
| 5425 | case 8: |
| 5426 | if (strEQ(d,"formline")) return -KEY_formline; |
| 5427 | break; |
| 5428 | } |
| 5429 | break; |
| 5430 | case 'g': |
| 5431 | if (strnEQ(d,"get",3)) { |
| 5432 | d += 3; |
| 5433 | if (*d == 'p') { |
| 5434 | switch (len) { |
| 5435 | case 7: |
| 5436 | if (strEQ(d,"ppid")) return -KEY_getppid; |
| 5437 | if (strEQ(d,"pgrp")) return -KEY_getpgrp; |
| 5438 | break; |
| 5439 | case 8: |
| 5440 | if (strEQ(d,"pwent")) return -KEY_getpwent; |
| 5441 | if (strEQ(d,"pwnam")) return -KEY_getpwnam; |
| 5442 | if (strEQ(d,"pwuid")) return -KEY_getpwuid; |
| 5443 | break; |
| 5444 | case 11: |
| 5445 | if (strEQ(d,"peername")) return -KEY_getpeername; |
| 5446 | if (strEQ(d,"protoent")) return -KEY_getprotoent; |
| 5447 | if (strEQ(d,"priority")) return -KEY_getpriority; |
| 5448 | break; |
| 5449 | case 14: |
| 5450 | if (strEQ(d,"protobyname")) return -KEY_getprotobyname; |
| 5451 | break; |
| 5452 | case 16: |
| 5453 | if (strEQ(d,"protobynumber"))return -KEY_getprotobynumber; |
| 5454 | break; |
| 5455 | } |
| 5456 | } |
| 5457 | else if (*d == 'h') { |
| 5458 | if (strEQ(d,"hostbyname")) return -KEY_gethostbyname; |
| 5459 | if (strEQ(d,"hostbyaddr")) return -KEY_gethostbyaddr; |
| 5460 | if (strEQ(d,"hostent")) return -KEY_gethostent; |
| 5461 | } |
| 5462 | else if (*d == 'n') { |
| 5463 | if (strEQ(d,"netbyname")) return -KEY_getnetbyname; |
| 5464 | if (strEQ(d,"netbyaddr")) return -KEY_getnetbyaddr; |
| 5465 | if (strEQ(d,"netent")) return -KEY_getnetent; |
| 5466 | } |
| 5467 | else if (*d == 's') { |
| 5468 | if (strEQ(d,"servbyname")) return -KEY_getservbyname; |
| 5469 | if (strEQ(d,"servbyport")) return -KEY_getservbyport; |
| 5470 | if (strEQ(d,"servent")) return -KEY_getservent; |
| 5471 | if (strEQ(d,"sockname")) return -KEY_getsockname; |
| 5472 | if (strEQ(d,"sockopt")) return -KEY_getsockopt; |
| 5473 | } |
| 5474 | else if (*d == 'g') { |
| 5475 | if (strEQ(d,"grent")) return -KEY_getgrent; |
| 5476 | if (strEQ(d,"grnam")) return -KEY_getgrnam; |
| 5477 | if (strEQ(d,"grgid")) return -KEY_getgrgid; |
| 5478 | } |
| 5479 | else if (*d == 'l') { |
| 5480 | if (strEQ(d,"login")) return -KEY_getlogin; |
| 5481 | } |
| 5482 | else if (strEQ(d,"c")) return -KEY_getc; |
| 5483 | break; |
| 5484 | } |
| 5485 | switch (len) { |
| 5486 | case 2: |
| 5487 | if (strEQ(d,"gt")) return -KEY_gt; |
| 5488 | if (strEQ(d,"ge")) return -KEY_ge; |
| 5489 | break; |
| 5490 | case 4: |
| 5491 | if (strEQ(d,"grep")) return KEY_grep; |
| 5492 | if (strEQ(d,"goto")) return KEY_goto; |
| 5493 | if (strEQ(d,"glob")) return KEY_glob; |
| 5494 | break; |
| 5495 | case 6: |
| 5496 | if (strEQ(d,"gmtime")) return -KEY_gmtime; |
| 5497 | break; |
| 5498 | } |
| 5499 | break; |
| 5500 | case 'h': |
| 5501 | if (strEQ(d,"hex")) return -KEY_hex; |
| 5502 | break; |
| 5503 | case 'I': |
| 5504 | if (strEQ(d,"INIT")) return KEY_INIT; |
| 5505 | break; |
| 5506 | case 'i': |
| 5507 | switch (len) { |
| 5508 | case 2: |
| 5509 | if (strEQ(d,"if")) return KEY_if; |
| 5510 | break; |
| 5511 | case 3: |
| 5512 | if (strEQ(d,"int")) return -KEY_int; |
| 5513 | break; |
| 5514 | case 5: |
| 5515 | if (strEQ(d,"index")) return -KEY_index; |
| 5516 | if (strEQ(d,"ioctl")) return -KEY_ioctl; |
| 5517 | break; |
| 5518 | } |
| 5519 | break; |
| 5520 | case 'j': |
| 5521 | if (strEQ(d,"join")) return -KEY_join; |
| 5522 | break; |
| 5523 | case 'k': |
| 5524 | if (len == 4) { |
| 5525 | if (strEQ(d,"keys")) return -KEY_keys; |
| 5526 | if (strEQ(d,"kill")) return -KEY_kill; |
| 5527 | } |
| 5528 | break; |
| 5529 | case 'l': |
| 5530 | switch (len) { |
| 5531 | case 2: |
| 5532 | if (strEQ(d,"lt")) return -KEY_lt; |
| 5533 | if (strEQ(d,"le")) return -KEY_le; |
| 5534 | if (strEQ(d,"lc")) return -KEY_lc; |
| 5535 | break; |
| 5536 | case 3: |
| 5537 | if (strEQ(d,"log")) return -KEY_log; |
| 5538 | break; |
| 5539 | case 4: |
| 5540 | if (strEQ(d,"last")) return KEY_last; |
| 5541 | if (strEQ(d,"link")) return -KEY_link; |
| 5542 | if (strEQ(d,"lock")) return -KEY_lock; |
| 5543 | break; |
| 5544 | case 5: |
| 5545 | if (strEQ(d,"local")) return KEY_local; |
| 5546 | if (strEQ(d,"lstat")) return -KEY_lstat; |
| 5547 | break; |
| 5548 | case 6: |
| 5549 | if (strEQ(d,"length")) return -KEY_length; |
| 5550 | if (strEQ(d,"listen")) return -KEY_listen; |
| 5551 | break; |
| 5552 | case 7: |
| 5553 | if (strEQ(d,"lcfirst")) return -KEY_lcfirst; |
| 5554 | break; |
| 5555 | case 9: |
| 5556 | if (strEQ(d,"localtime")) return -KEY_localtime; |
| 5557 | break; |
| 5558 | } |
| 5559 | break; |
| 5560 | case 'm': |
| 5561 | switch (len) { |
| 5562 | case 1: return KEY_m; |
| 5563 | case 2: |
| 5564 | if (strEQ(d,"my")) return KEY_my; |
| 5565 | break; |
| 5566 | case 3: |
| 5567 | if (strEQ(d,"map")) return KEY_map; |
| 5568 | break; |
| 5569 | case 5: |
| 5570 | if (strEQ(d,"mkdir")) return -KEY_mkdir; |
| 5571 | break; |
| 5572 | case 6: |
| 5573 | if (strEQ(d,"msgctl")) return -KEY_msgctl; |
| 5574 | if (strEQ(d,"msgget")) return -KEY_msgget; |
| 5575 | if (strEQ(d,"msgrcv")) return -KEY_msgrcv; |
| 5576 | if (strEQ(d,"msgsnd")) return -KEY_msgsnd; |
| 5577 | break; |
| 5578 | } |
| 5579 | break; |
| 5580 | case 'n': |
| 5581 | if (strEQ(d,"next")) return KEY_next; |
| 5582 | if (strEQ(d,"ne")) return -KEY_ne; |
| 5583 | if (strEQ(d,"not")) return -KEY_not; |
| 5584 | if (strEQ(d,"no")) return KEY_no; |
| 5585 | break; |
| 5586 | case 'o': |
| 5587 | switch (len) { |
| 5588 | case 2: |
| 5589 | if (strEQ(d,"or")) return -KEY_or; |
| 5590 | break; |
| 5591 | case 3: |
| 5592 | if (strEQ(d,"ord")) return -KEY_ord; |
| 5593 | if (strEQ(d,"oct")) return -KEY_oct; |
| 5594 | if (strEQ(d,"our")) return KEY_our; |
| 5595 | break; |
| 5596 | case 4: |
| 5597 | if (strEQ(d,"open")) return -KEY_open; |
| 5598 | break; |
| 5599 | case 7: |
| 5600 | if (strEQ(d,"opendir")) return -KEY_opendir; |
| 5601 | break; |
| 5602 | } |
| 5603 | break; |
| 5604 | case 'p': |
| 5605 | switch (len) { |
| 5606 | case 3: |
| 5607 | if (strEQ(d,"pop")) return -KEY_pop; |
| 5608 | if (strEQ(d,"pos")) return KEY_pos; |
| 5609 | break; |
| 5610 | case 4: |
| 5611 | if (strEQ(d,"push")) return -KEY_push; |
| 5612 | if (strEQ(d,"pack")) return -KEY_pack; |
| 5613 | if (strEQ(d,"pipe")) return -KEY_pipe; |
| 5614 | break; |
| 5615 | case 5: |
| 5616 | if (strEQ(d,"print")) return KEY_print; |
| 5617 | break; |
| 5618 | case 6: |
| 5619 | if (strEQ(d,"printf")) return KEY_printf; |
| 5620 | break; |
| 5621 | case 7: |
| 5622 | if (strEQ(d,"package")) return KEY_package; |
| 5623 | break; |
| 5624 | case 9: |
| 5625 | if (strEQ(d,"prototype")) return KEY_prototype; |
| 5626 | } |
| 5627 | break; |
| 5628 | case 'q': |
| 5629 | if (len <= 2) { |
| 5630 | if (strEQ(d,"q")) return KEY_q; |
| 5631 | if (strEQ(d,"qr")) return KEY_qr; |
| 5632 | if (strEQ(d,"qq")) return KEY_qq; |
| 5633 | if (strEQ(d,"qw")) return KEY_qw; |
| 5634 | if (strEQ(d,"qx")) return KEY_qx; |
| 5635 | } |
| 5636 | else if (strEQ(d,"quotemeta")) return -KEY_quotemeta; |
| 5637 | break; |
| 5638 | case 'r': |
| 5639 | switch (len) { |
| 5640 | case 3: |
| 5641 | if (strEQ(d,"ref")) return -KEY_ref; |
| 5642 | break; |
| 5643 | case 4: |
| 5644 | if (strEQ(d,"read")) return -KEY_read; |
| 5645 | if (strEQ(d,"rand")) return -KEY_rand; |
| 5646 | if (strEQ(d,"recv")) return -KEY_recv; |
| 5647 | if (strEQ(d,"redo")) return KEY_redo; |
| 5648 | break; |
| 5649 | case 5: |
| 5650 | if (strEQ(d,"rmdir")) return -KEY_rmdir; |
| 5651 | if (strEQ(d,"reset")) return -KEY_reset; |
| 5652 | break; |
| 5653 | case 6: |
| 5654 | if (strEQ(d,"return")) return KEY_return; |
| 5655 | if (strEQ(d,"rename")) return -KEY_rename; |
| 5656 | if (strEQ(d,"rindex")) return -KEY_rindex; |
| 5657 | break; |
| 5658 | case 7: |
| 5659 | if (strEQ(d,"require")) return KEY_require; |
| 5660 | if (strEQ(d,"reverse")) return -KEY_reverse; |
| 5661 | if (strEQ(d,"readdir")) return -KEY_readdir; |
| 5662 | break; |
| 5663 | case 8: |
| 5664 | if (strEQ(d,"readlink")) return -KEY_readlink; |
| 5665 | if (strEQ(d,"readline")) return -KEY_readline; |
| 5666 | if (strEQ(d,"readpipe")) return -KEY_readpipe; |
| 5667 | break; |
| 5668 | case 9: |
| 5669 | if (strEQ(d,"rewinddir")) return -KEY_rewinddir; |
| 5670 | break; |
| 5671 | } |
| 5672 | break; |
| 5673 | case 's': |
| 5674 | switch (d[1]) { |
| 5675 | case 0: return KEY_s; |
| 5676 | case 'c': |
| 5677 | if (strEQ(d,"scalar")) return KEY_scalar; |
| 5678 | break; |
| 5679 | case 'e': |
| 5680 | switch (len) { |
| 5681 | case 4: |
| 5682 | if (strEQ(d,"seek")) return -KEY_seek; |
| 5683 | if (strEQ(d,"send")) return -KEY_send; |
| 5684 | break; |
| 5685 | case 5: |
| 5686 | if (strEQ(d,"semop")) return -KEY_semop; |
| 5687 | break; |
| 5688 | case 6: |
| 5689 | if (strEQ(d,"select")) return -KEY_select; |
| 5690 | if (strEQ(d,"semctl")) return -KEY_semctl; |
| 5691 | if (strEQ(d,"semget")) return -KEY_semget; |
| 5692 | break; |
| 5693 | case 7: |
| 5694 | if (strEQ(d,"setpgrp")) return -KEY_setpgrp; |
| 5695 | if (strEQ(d,"seekdir")) return -KEY_seekdir; |
| 5696 | break; |
| 5697 | case 8: |
| 5698 | if (strEQ(d,"setpwent")) return -KEY_setpwent; |
| 5699 | if (strEQ(d,"setgrent")) return -KEY_setgrent; |
| 5700 | break; |
| 5701 | case 9: |
| 5702 | if (strEQ(d,"setnetent")) return -KEY_setnetent; |
| 5703 | break; |
| 5704 | case 10: |
| 5705 | if (strEQ(d,"setsockopt")) return -KEY_setsockopt; |
| 5706 | if (strEQ(d,"sethostent")) return -KEY_sethostent; |
| 5707 | if (strEQ(d,"setservent")) return -KEY_setservent; |
| 5708 | break; |
| 5709 | case 11: |
| 5710 | if (strEQ(d,"setpriority")) return -KEY_setpriority; |
| 5711 | if (strEQ(d,"setprotoent")) return -KEY_setprotoent; |
| 5712 | break; |
| 5713 | } |
| 5714 | break; |
| 5715 | case 'h': |
| 5716 | switch (len) { |
| 5717 | case 5: |
| 5718 | if (strEQ(d,"shift")) return -KEY_shift; |
| 5719 | break; |
| 5720 | case 6: |
| 5721 | if (strEQ(d,"shmctl")) return -KEY_shmctl; |
| 5722 | if (strEQ(d,"shmget")) return -KEY_shmget; |
| 5723 | break; |
| 5724 | case 7: |
| 5725 | if (strEQ(d,"shmread")) return -KEY_shmread; |
| 5726 | break; |
| 5727 | case 8: |
| 5728 | if (strEQ(d,"shmwrite")) return -KEY_shmwrite; |
| 5729 | if (strEQ(d,"shutdown")) return -KEY_shutdown; |
| 5730 | break; |
| 5731 | } |
| 5732 | break; |
| 5733 | case 'i': |
| 5734 | if (strEQ(d,"sin")) return -KEY_sin; |
| 5735 | break; |
| 5736 | case 'l': |
| 5737 | if (strEQ(d,"sleep")) return -KEY_sleep; |
| 5738 | break; |
| 5739 | case 'o': |
| 5740 | if (strEQ(d,"sort")) return KEY_sort; |
| 5741 | if (strEQ(d,"socket")) return -KEY_socket; |
| 5742 | if (strEQ(d,"socketpair")) return -KEY_socketpair; |
| 5743 | break; |
| 5744 | case 'p': |
| 5745 | if (strEQ(d,"split")) return KEY_split; |
| 5746 | if (strEQ(d,"sprintf")) return -KEY_sprintf; |
| 5747 | if (strEQ(d,"splice")) return -KEY_splice; |
| 5748 | break; |
| 5749 | case 'q': |
| 5750 | if (strEQ(d,"sqrt")) return -KEY_sqrt; |
| 5751 | break; |
| 5752 | case 'r': |
| 5753 | if (strEQ(d,"srand")) return -KEY_srand; |
| 5754 | break; |
| 5755 | case 't': |
| 5756 | if (strEQ(d,"stat")) return -KEY_stat; |
| 5757 | if (strEQ(d,"study")) return KEY_study; |
| 5758 | break; |
| 5759 | case 'u': |
| 5760 | if (strEQ(d,"substr")) return -KEY_substr; |
| 5761 | if (strEQ(d,"sub")) return KEY_sub; |
| 5762 | break; |
| 5763 | case 'y': |
| 5764 | switch (len) { |
| 5765 | case 6: |
| 5766 | if (strEQ(d,"system")) return -KEY_system; |
| 5767 | break; |
| 5768 | case 7: |
| 5769 | if (strEQ(d,"symlink")) return -KEY_symlink; |
| 5770 | if (strEQ(d,"syscall")) return -KEY_syscall; |
| 5771 | if (strEQ(d,"sysopen")) return -KEY_sysopen; |
| 5772 | if (strEQ(d,"sysread")) return -KEY_sysread; |
| 5773 | if (strEQ(d,"sysseek")) return -KEY_sysseek; |
| 5774 | break; |
| 5775 | case 8: |
| 5776 | if (strEQ(d,"syswrite")) return -KEY_syswrite; |
| 5777 | break; |
| 5778 | } |
| 5779 | break; |
| 5780 | } |
| 5781 | break; |
| 5782 | case 't': |
| 5783 | switch (len) { |
| 5784 | case 2: |
| 5785 | if (strEQ(d,"tr")) return KEY_tr; |
| 5786 | break; |
| 5787 | case 3: |
| 5788 | if (strEQ(d,"tie")) return KEY_tie; |
| 5789 | break; |
| 5790 | case 4: |
| 5791 | if (strEQ(d,"tell")) return -KEY_tell; |
| 5792 | if (strEQ(d,"tied")) return KEY_tied; |
| 5793 | if (strEQ(d,"time")) return -KEY_time; |
| 5794 | break; |
| 5795 | case 5: |
| 5796 | if (strEQ(d,"times")) return -KEY_times; |
| 5797 | break; |
| 5798 | case 7: |
| 5799 | if (strEQ(d,"telldir")) return -KEY_telldir; |
| 5800 | break; |
| 5801 | case 8: |
| 5802 | if (strEQ(d,"truncate")) return -KEY_truncate; |
| 5803 | break; |
| 5804 | } |
| 5805 | break; |
| 5806 | case 'u': |
| 5807 | switch (len) { |
| 5808 | case 2: |
| 5809 | if (strEQ(d,"uc")) return -KEY_uc; |
| 5810 | break; |
| 5811 | case 3: |
| 5812 | if (strEQ(d,"use")) return KEY_use; |
| 5813 | break; |
| 5814 | case 5: |
| 5815 | if (strEQ(d,"undef")) return KEY_undef; |
| 5816 | if (strEQ(d,"until")) return KEY_until; |
| 5817 | if (strEQ(d,"untie")) return KEY_untie; |
| 5818 | if (strEQ(d,"utime")) return -KEY_utime; |
| 5819 | if (strEQ(d,"umask")) return -KEY_umask; |
| 5820 | break; |
| 5821 | case 6: |
| 5822 | if (strEQ(d,"unless")) return KEY_unless; |
| 5823 | if (strEQ(d,"unpack")) return -KEY_unpack; |
| 5824 | if (strEQ(d,"unlink")) return -KEY_unlink; |
| 5825 | break; |
| 5826 | case 7: |
| 5827 | if (strEQ(d,"unshift")) return -KEY_unshift; |
| 5828 | if (strEQ(d,"ucfirst")) return -KEY_ucfirst; |
| 5829 | break; |
| 5830 | } |
| 5831 | break; |
| 5832 | case 'v': |
| 5833 | if (strEQ(d,"values")) return -KEY_values; |
| 5834 | if (strEQ(d,"vec")) return -KEY_vec; |
| 5835 | break; |
| 5836 | case 'w': |
| 5837 | switch (len) { |
| 5838 | case 4: |
| 5839 | if (strEQ(d,"warn")) return -KEY_warn; |
| 5840 | if (strEQ(d,"wait")) return -KEY_wait; |
| 5841 | break; |
| 5842 | case 5: |
| 5843 | if (strEQ(d,"while")) return KEY_while; |
| 5844 | if (strEQ(d,"write")) return -KEY_write; |
| 5845 | break; |
| 5846 | case 7: |
| 5847 | if (strEQ(d,"waitpid")) return -KEY_waitpid; |
| 5848 | break; |
| 5849 | case 9: |
| 5850 | if (strEQ(d,"wantarray")) return -KEY_wantarray; |
| 5851 | break; |
| 5852 | } |
| 5853 | break; |
| 5854 | case 'x': |
| 5855 | if (len == 1) return -KEY_x; |
| 5856 | if (strEQ(d,"xor")) return -KEY_xor; |
| 5857 | break; |
| 5858 | case 'y': |
| 5859 | if (len == 1) return KEY_y; |
| 5860 | break; |
| 5861 | case 'z': |
| 5862 | break; |
| 5863 | } |
| 5864 | return 0; |
| 5865 | } |
| 5866 | |
| 5867 | STATIC void |
| 5868 | S_checkcomma(pTHX_ register char *s, char *name, char *what) |
| 5869 | { |
| 5870 | char *w; |
| 5871 | |
| 5872 | if (*s == ' ' && s[1] == '(') { /* XXX gotta be a better way */ |
| 5873 | if (ckWARN(WARN_SYNTAX)) { |
| 5874 | int level = 1; |
| 5875 | for (w = s+2; *w && level; w++) { |
| 5876 | if (*w == '(') |
| 5877 | ++level; |
| 5878 | else if (*w == ')') |
| 5879 | --level; |
| 5880 | } |
| 5881 | if (*w) |
| 5882 | for (; *w && isSPACE(*w); w++) ; |
| 5883 | if (!*w || !strchr(";|})]oaiuw!=", *w)) /* an advisory hack only... */ |
| 5884 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 5885 | "%s (...) interpreted as function",name); |
| 5886 | } |
| 5887 | } |
| 5888 | while (s < PL_bufend && isSPACE(*s)) |
| 5889 | s++; |
| 5890 | if (*s == '(') |
| 5891 | s++; |
| 5892 | while (s < PL_bufend && isSPACE(*s)) |
| 5893 | s++; |
| 5894 | if (isIDFIRST_lazy_if(s,UTF)) { |
| 5895 | w = s++; |
| 5896 | while (isALNUM_lazy_if(s,UTF)) |
| 5897 | s++; |
| 5898 | while (s < PL_bufend && isSPACE(*s)) |
| 5899 | s++; |
| 5900 | if (*s == ',') { |
| 5901 | int kw; |
| 5902 | *s = '\0'; |
| 5903 | kw = keyword(w, s - w) || get_cv(w, FALSE) != 0; |
| 5904 | *s = ','; |
| 5905 | if (kw) |
| 5906 | return; |
| 5907 | Perl_croak(aTHX_ "No comma allowed after %s", what); |
| 5908 | } |
| 5909 | } |
| 5910 | } |
| 5911 | |
| 5912 | /* Either returns sv, or mortalizes sv and returns a new SV*. |
| 5913 | Best used as sv=new_constant(..., sv, ...). |
| 5914 | If s, pv are NULL, calls subroutine with one argument, |
| 5915 | and type is used with error messages only. */ |
| 5916 | |
| 5917 | STATIC SV * |
| 5918 | S_new_constant(pTHX_ char *s, STRLEN len, const char *key, SV *sv, SV *pv, |
| 5919 | const char *type) |
| 5920 | { |
| 5921 | dSP; |
| 5922 | HV *table = GvHV(PL_hintgv); /* ^H */ |
| 5923 | SV *res; |
| 5924 | SV **cvp; |
| 5925 | SV *cv, *typesv; |
| 5926 | const char *why1, *why2, *why3; |
| 5927 | |
| 5928 | if (!table || !(PL_hints & HINT_LOCALIZE_HH)) { |
| 5929 | SV *msg; |
| 5930 | |
| 5931 | why2 = strEQ(key,"charnames") |
| 5932 | ? "(possibly a missing \"use charnames ...\")" |
| 5933 | : ""; |
| 5934 | msg = Perl_newSVpvf(aTHX_ "Constant(%s) unknown: %s", |
| 5935 | (type ? type: "undef"), why2); |
| 5936 | |
| 5937 | /* This is convoluted and evil ("goto considered harmful") |
| 5938 | * but I do not understand the intricacies of all the different |
| 5939 | * failure modes of %^H in here. The goal here is to make |
| 5940 | * the most probable error message user-friendly. --jhi */ |
| 5941 | |
| 5942 | goto msgdone; |
| 5943 | |
| 5944 | report: |
| 5945 | msg = Perl_newSVpvf(aTHX_ "Constant(%s): %s%s%s", |
| 5946 | (type ? type: "undef"), why1, why2, why3); |
| 5947 | msgdone: |
| 5948 | yyerror(SvPVX(msg)); |
| 5949 | SvREFCNT_dec(msg); |
| 5950 | return sv; |
| 5951 | } |
| 5952 | cvp = hv_fetch(table, key, strlen(key), FALSE); |
| 5953 | if (!cvp || !SvOK(*cvp)) { |
| 5954 | why1 = "$^H{"; |
| 5955 | why2 = key; |
| 5956 | why3 = "} is not defined"; |
| 5957 | goto report; |
| 5958 | } |
| 5959 | sv_2mortal(sv); /* Parent created it permanently */ |
| 5960 | cv = *cvp; |
| 5961 | if (!pv && s) |
| 5962 | pv = sv_2mortal(newSVpvn(s, len)); |
| 5963 | if (type && pv) |
| 5964 | typesv = sv_2mortal(newSVpv(type, 0)); |
| 5965 | else |
| 5966 | typesv = &PL_sv_undef; |
| 5967 | |
| 5968 | PUSHSTACKi(PERLSI_OVERLOAD); |
| 5969 | ENTER ; |
| 5970 | SAVETMPS; |
| 5971 | |
| 5972 | PUSHMARK(SP) ; |
| 5973 | EXTEND(sp, 3); |
| 5974 | if (pv) |
| 5975 | PUSHs(pv); |
| 5976 | PUSHs(sv); |
| 5977 | if (pv) |
| 5978 | PUSHs(typesv); |
| 5979 | PUTBACK; |
| 5980 | call_sv(cv, G_SCALAR | ( PL_in_eval ? 0 : G_EVAL)); |
| 5981 | |
| 5982 | SPAGAIN ; |
| 5983 | |
| 5984 | /* Check the eval first */ |
| 5985 | if (!PL_in_eval && SvTRUE(ERRSV)) { |
| 5986 | STRLEN n_a; |
| 5987 | sv_catpv(ERRSV, "Propagated"); |
| 5988 | yyerror(SvPV(ERRSV, n_a)); /* Duplicates the message inside eval */ |
| 5989 | (void)POPs; |
| 5990 | res = SvREFCNT_inc(sv); |
| 5991 | } |
| 5992 | else { |
| 5993 | res = POPs; |
| 5994 | (void)SvREFCNT_inc(res); |
| 5995 | } |
| 5996 | |
| 5997 | PUTBACK ; |
| 5998 | FREETMPS ; |
| 5999 | LEAVE ; |
| 6000 | POPSTACK; |
| 6001 | |
| 6002 | if (!SvOK(res)) { |
| 6003 | why1 = "Call to &{$^H{"; |
| 6004 | why2 = key; |
| 6005 | why3 = "}} did not return a defined value"; |
| 6006 | sv = res; |
| 6007 | goto report; |
| 6008 | } |
| 6009 | |
| 6010 | return res; |
| 6011 | } |
| 6012 | |
| 6013 | STATIC char * |
| 6014 | S_scan_word(pTHX_ register char *s, char *dest, STRLEN destlen, int allow_package, STRLEN *slp) |
| 6015 | { |
| 6016 | register char *d = dest; |
| 6017 | register char *e = d + destlen - 3; /* two-character token, ending NUL */ |
| 6018 | for (;;) { |
| 6019 | if (d >= e) |
| 6020 | Perl_croak(aTHX_ ident_too_long); |
| 6021 | if (isALNUM(*s)) /* UTF handled below */ |
| 6022 | *d++ = *s++; |
| 6023 | else if (*s == '\'' && allow_package && isIDFIRST_lazy_if(s+1,UTF)) { |
| 6024 | *d++ = ':'; |
| 6025 | *d++ = ':'; |
| 6026 | s++; |
| 6027 | } |
| 6028 | else if (*s == ':' && s[1] == ':' && allow_package && s[2] != '$') { |
| 6029 | *d++ = *s++; |
| 6030 | *d++ = *s++; |
| 6031 | } |
| 6032 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
| 6033 | char *t = s + UTF8SKIP(s); |
| 6034 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
| 6035 | t += UTF8SKIP(t); |
| 6036 | if (d + (t - s) > e) |
| 6037 | Perl_croak(aTHX_ ident_too_long); |
| 6038 | Copy(s, d, t - s, char); |
| 6039 | d += t - s; |
| 6040 | s = t; |
| 6041 | } |
| 6042 | else { |
| 6043 | *d = '\0'; |
| 6044 | *slp = d - dest; |
| 6045 | return s; |
| 6046 | } |
| 6047 | } |
| 6048 | } |
| 6049 | |
| 6050 | STATIC char * |
| 6051 | S_scan_ident(pTHX_ register char *s, register char *send, char *dest, STRLEN destlen, I32 ck_uni) |
| 6052 | { |
| 6053 | register char *d; |
| 6054 | register char *e; |
| 6055 | char *bracket = 0; |
| 6056 | char funny = *s++; |
| 6057 | |
| 6058 | if (isSPACE(*s)) |
| 6059 | s = skipspace(s); |
| 6060 | d = dest; |
| 6061 | e = d + destlen - 3; /* two-character token, ending NUL */ |
| 6062 | if (isDIGIT(*s)) { |
| 6063 | while (isDIGIT(*s)) { |
| 6064 | if (d >= e) |
| 6065 | Perl_croak(aTHX_ ident_too_long); |
| 6066 | *d++ = *s++; |
| 6067 | } |
| 6068 | } |
| 6069 | else { |
| 6070 | for (;;) { |
| 6071 | if (d >= e) |
| 6072 | Perl_croak(aTHX_ ident_too_long); |
| 6073 | if (isALNUM(*s)) /* UTF handled below */ |
| 6074 | *d++ = *s++; |
| 6075 | else if (*s == '\'' && isIDFIRST_lazy_if(s+1,UTF)) { |
| 6076 | *d++ = ':'; |
| 6077 | *d++ = ':'; |
| 6078 | s++; |
| 6079 | } |
| 6080 | else if (*s == ':' && s[1] == ':') { |
| 6081 | *d++ = *s++; |
| 6082 | *d++ = *s++; |
| 6083 | } |
| 6084 | else if (UTF && UTF8_IS_START(*s) && isALNUM_utf8((U8*)s)) { |
| 6085 | char *t = s + UTF8SKIP(s); |
| 6086 | while (UTF8_IS_CONTINUED(*t) && is_utf8_mark((U8*)t)) |
| 6087 | t += UTF8SKIP(t); |
| 6088 | if (d + (t - s) > e) |
| 6089 | Perl_croak(aTHX_ ident_too_long); |
| 6090 | Copy(s, d, t - s, char); |
| 6091 | d += t - s; |
| 6092 | s = t; |
| 6093 | } |
| 6094 | else |
| 6095 | break; |
| 6096 | } |
| 6097 | } |
| 6098 | *d = '\0'; |
| 6099 | d = dest; |
| 6100 | if (*d) { |
| 6101 | if (PL_lex_state != LEX_NORMAL) |
| 6102 | PL_lex_state = LEX_INTERPENDMAYBE; |
| 6103 | return s; |
| 6104 | } |
| 6105 | if (*s == '$' && s[1] && |
| 6106 | (isALNUM_lazy_if(s+1,UTF) || strchr("${", s[1]) || strnEQ(s+1,"::",2)) ) |
| 6107 | { |
| 6108 | return s; |
| 6109 | } |
| 6110 | if (*s == '{') { |
| 6111 | bracket = s; |
| 6112 | s++; |
| 6113 | } |
| 6114 | else if (ck_uni) |
| 6115 | check_uni(); |
| 6116 | if (s < send) |
| 6117 | *d = *s++; |
| 6118 | d[1] = '\0'; |
| 6119 | if (*d == '^' && *s && isCONTROLVAR(*s)) { |
| 6120 | *d = toCTRL(*s); |
| 6121 | s++; |
| 6122 | } |
| 6123 | if (bracket) { |
| 6124 | if (isSPACE(s[-1])) { |
| 6125 | while (s < send) { |
| 6126 | char ch = *s++; |
| 6127 | if (!SPACE_OR_TAB(ch)) { |
| 6128 | *d = ch; |
| 6129 | break; |
| 6130 | } |
| 6131 | } |
| 6132 | } |
| 6133 | if (isIDFIRST_lazy_if(d,UTF)) { |
| 6134 | d++; |
| 6135 | if (UTF) { |
| 6136 | e = s; |
| 6137 | while ((e < send && isALNUM_lazy_if(e,UTF)) || *e == ':') { |
| 6138 | e += UTF8SKIP(e); |
| 6139 | while (e < send && UTF8_IS_CONTINUED(*e) && is_utf8_mark((U8*)e)) |
| 6140 | e += UTF8SKIP(e); |
| 6141 | } |
| 6142 | Copy(s, d, e - s, char); |
| 6143 | d += e - s; |
| 6144 | s = e; |
| 6145 | } |
| 6146 | else { |
| 6147 | while ((isALNUM(*s) || *s == ':') && d < e) |
| 6148 | *d++ = *s++; |
| 6149 | if (d >= e) |
| 6150 | Perl_croak(aTHX_ ident_too_long); |
| 6151 | } |
| 6152 | *d = '\0'; |
| 6153 | while (s < send && SPACE_OR_TAB(*s)) s++; |
| 6154 | if ((*s == '[' || (*s == '{' && strNE(dest, "sub")))) { |
| 6155 | if (ckWARN(WARN_AMBIGUOUS) && keyword(dest, d - dest)) { |
| 6156 | const char *brack = *s == '[' ? "[...]" : "{...}"; |
| 6157 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 6158 | "Ambiguous use of %c{%s%s} resolved to %c%s%s", |
| 6159 | funny, dest, brack, funny, dest, brack); |
| 6160 | } |
| 6161 | bracket++; |
| 6162 | PL_lex_brackstack[PL_lex_brackets++] = (char)(XOPERATOR | XFAKEBRACK); |
| 6163 | return s; |
| 6164 | } |
| 6165 | } |
| 6166 | /* Handle extended ${^Foo} variables |
| 6167 | * 1999-02-27 mjd-perl-patch@plover.com */ |
| 6168 | else if (!isALNUM(*d) && !isPRINT(*d) /* isCTRL(d) */ |
| 6169 | && isALNUM(*s)) |
| 6170 | { |
| 6171 | d++; |
| 6172 | while (isALNUM(*s) && d < e) { |
| 6173 | *d++ = *s++; |
| 6174 | } |
| 6175 | if (d >= e) |
| 6176 | Perl_croak(aTHX_ ident_too_long); |
| 6177 | *d = '\0'; |
| 6178 | } |
| 6179 | if (*s == '}') { |
| 6180 | s++; |
| 6181 | if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets) |
| 6182 | PL_lex_state = LEX_INTERPEND; |
| 6183 | if (funny == '#') |
| 6184 | funny = '@'; |
| 6185 | if (PL_lex_state == LEX_NORMAL) { |
| 6186 | if (ckWARN(WARN_AMBIGUOUS) && |
| 6187 | (keyword(dest, d - dest) || get_cv(dest, FALSE))) |
| 6188 | { |
| 6189 | Perl_warner(aTHX_ WARN_AMBIGUOUS, |
| 6190 | "Ambiguous use of %c{%s} resolved to %c%s", |
| 6191 | funny, dest, funny, dest); |
| 6192 | } |
| 6193 | } |
| 6194 | } |
| 6195 | else { |
| 6196 | s = bracket; /* let the parser handle it */ |
| 6197 | *dest = '\0'; |
| 6198 | } |
| 6199 | } |
| 6200 | else if (PL_lex_state == LEX_INTERPNORMAL && !PL_lex_brackets && !intuit_more(s)) |
| 6201 | PL_lex_state = LEX_INTERPEND; |
| 6202 | return s; |
| 6203 | } |
| 6204 | |
| 6205 | void |
| 6206 | Perl_pmflag(pTHX_ U16 *pmfl, int ch) |
| 6207 | { |
| 6208 | if (ch == 'i') |
| 6209 | *pmfl |= PMf_FOLD; |
| 6210 | else if (ch == 'g') |
| 6211 | *pmfl |= PMf_GLOBAL; |
| 6212 | else if (ch == 'c') |
| 6213 | *pmfl |= PMf_CONTINUE; |
| 6214 | else if (ch == 'o') |
| 6215 | *pmfl |= PMf_KEEP; |
| 6216 | else if (ch == 'm') |
| 6217 | *pmfl |= PMf_MULTILINE; |
| 6218 | else if (ch == 's') |
| 6219 | *pmfl |= PMf_SINGLELINE; |
| 6220 | else if (ch == 'x') |
| 6221 | *pmfl |= PMf_EXTENDED; |
| 6222 | } |
| 6223 | |
| 6224 | STATIC char * |
| 6225 | S_scan_pat(pTHX_ char *start, I32 type) |
| 6226 | { |
| 6227 | PMOP *pm; |
| 6228 | char *s; |
| 6229 | |
| 6230 | s = scan_str(start,FALSE,FALSE); |
| 6231 | if (!s) |
| 6232 | Perl_croak(aTHX_ "Search pattern not terminated"); |
| 6233 | |
| 6234 | pm = (PMOP*)newPMOP(type, 0); |
| 6235 | if (PL_multi_open == '?') |
| 6236 | pm->op_pmflags |= PMf_ONCE; |
| 6237 | if(type == OP_QR) { |
| 6238 | while (*s && strchr("iomsx", *s)) |
| 6239 | pmflag(&pm->op_pmflags,*s++); |
| 6240 | } |
| 6241 | else { |
| 6242 | while (*s && strchr("iogcmsx", *s)) |
| 6243 | pmflag(&pm->op_pmflags,*s++); |
| 6244 | } |
| 6245 | pm->op_pmpermflags = pm->op_pmflags; |
| 6246 | |
| 6247 | PL_lex_op = (OP*)pm; |
| 6248 | yylval.ival = OP_MATCH; |
| 6249 | return s; |
| 6250 | } |
| 6251 | |
| 6252 | STATIC char * |
| 6253 | S_scan_subst(pTHX_ char *start) |
| 6254 | { |
| 6255 | register char *s; |
| 6256 | register PMOP *pm; |
| 6257 | I32 first_start; |
| 6258 | I32 es = 0; |
| 6259 | |
| 6260 | yylval.ival = OP_NULL; |
| 6261 | |
| 6262 | s = scan_str(start,FALSE,FALSE); |
| 6263 | |
| 6264 | if (!s) |
| 6265 | Perl_croak(aTHX_ "Substitution pattern not terminated"); |
| 6266 | |
| 6267 | if (s[-1] == PL_multi_open) |
| 6268 | s--; |
| 6269 | |
| 6270 | first_start = PL_multi_start; |
| 6271 | s = scan_str(s,FALSE,FALSE); |
| 6272 | if (!s) { |
| 6273 | if (PL_lex_stuff) { |
| 6274 | SvREFCNT_dec(PL_lex_stuff); |
| 6275 | PL_lex_stuff = Nullsv; |
| 6276 | } |
| 6277 | Perl_croak(aTHX_ "Substitution replacement not terminated"); |
| 6278 | } |
| 6279 | PL_multi_start = first_start; /* so whole substitution is taken together */ |
| 6280 | |
| 6281 | pm = (PMOP*)newPMOP(OP_SUBST, 0); |
| 6282 | while (*s) { |
| 6283 | if (*s == 'e') { |
| 6284 | s++; |
| 6285 | es++; |
| 6286 | } |
| 6287 | else if (strchr("iogcmsx", *s)) |
| 6288 | pmflag(&pm->op_pmflags,*s++); |
| 6289 | else |
| 6290 | break; |
| 6291 | } |
| 6292 | |
| 6293 | if (es) { |
| 6294 | SV *repl; |
| 6295 | PL_sublex_info.super_bufptr = s; |
| 6296 | PL_sublex_info.super_bufend = PL_bufend; |
| 6297 | PL_multi_end = 0; |
| 6298 | pm->op_pmflags |= PMf_EVAL; |
| 6299 | repl = newSVpvn("",0); |
| 6300 | while (es-- > 0) |
| 6301 | sv_catpv(repl, es ? "eval " : "do "); |
| 6302 | sv_catpvn(repl, "{ ", 2); |
| 6303 | sv_catsv(repl, PL_lex_repl); |
| 6304 | sv_catpvn(repl, " };", 2); |
| 6305 | SvEVALED_on(repl); |
| 6306 | SvREFCNT_dec(PL_lex_repl); |
| 6307 | PL_lex_repl = repl; |
| 6308 | } |
| 6309 | |
| 6310 | pm->op_pmpermflags = pm->op_pmflags; |
| 6311 | PL_lex_op = (OP*)pm; |
| 6312 | yylval.ival = OP_SUBST; |
| 6313 | return s; |
| 6314 | } |
| 6315 | |
| 6316 | STATIC char * |
| 6317 | S_scan_trans(pTHX_ char *start) |
| 6318 | { |
| 6319 | register char* s; |
| 6320 | OP *o; |
| 6321 | short *tbl; |
| 6322 | I32 squash; |
| 6323 | I32 del; |
| 6324 | I32 complement; |
| 6325 | |
| 6326 | yylval.ival = OP_NULL; |
| 6327 | |
| 6328 | s = scan_str(start,FALSE,FALSE); |
| 6329 | if (!s) |
| 6330 | Perl_croak(aTHX_ "Transliteration pattern not terminated"); |
| 6331 | if (s[-1] == PL_multi_open) |
| 6332 | s--; |
| 6333 | |
| 6334 | s = scan_str(s,FALSE,FALSE); |
| 6335 | if (!s) { |
| 6336 | if (PL_lex_stuff) { |
| 6337 | SvREFCNT_dec(PL_lex_stuff); |
| 6338 | PL_lex_stuff = Nullsv; |
| 6339 | } |
| 6340 | Perl_croak(aTHX_ "Transliteration replacement not terminated"); |
| 6341 | } |
| 6342 | |
| 6343 | complement = del = squash = 0; |
| 6344 | while (strchr("cds", *s)) { |
| 6345 | if (*s == 'c') |
| 6346 | complement = OPpTRANS_COMPLEMENT; |
| 6347 | else if (*s == 'd') |
| 6348 | del = OPpTRANS_DELETE; |
| 6349 | else if (*s == 's') |
| 6350 | squash = OPpTRANS_SQUASH; |
| 6351 | s++; |
| 6352 | } |
| 6353 | |
| 6354 | New(803, tbl, complement&&!del?258:256, short); |
| 6355 | o = newPVOP(OP_TRANS, 0, (char*)tbl); |
| 6356 | o->op_private = del|squash|complement| |
| 6357 | (DO_UTF8(PL_lex_stuff)? OPpTRANS_FROM_UTF : 0)| |
| 6358 | (DO_UTF8(PL_lex_repl) ? OPpTRANS_TO_UTF : 0); |
| 6359 | |
| 6360 | PL_lex_op = o; |
| 6361 | yylval.ival = OP_TRANS; |
| 6362 | return s; |
| 6363 | } |
| 6364 | |
| 6365 | STATIC char * |
| 6366 | S_scan_heredoc(pTHX_ register char *s) |
| 6367 | { |
| 6368 | SV *herewas; |
| 6369 | I32 op_type = OP_SCALAR; |
| 6370 | I32 len; |
| 6371 | SV *tmpstr; |
| 6372 | char term; |
| 6373 | register char *d; |
| 6374 | register char *e; |
| 6375 | char *peek; |
| 6376 | int outer = (PL_rsfp && !(PL_lex_inwhat == OP_SCALAR)); |
| 6377 | |
| 6378 | s += 2; |
| 6379 | d = PL_tokenbuf; |
| 6380 | e = PL_tokenbuf + sizeof PL_tokenbuf - 1; |
| 6381 | if (!outer) |
| 6382 | *d++ = '\n'; |
| 6383 | for (peek = s; SPACE_OR_TAB(*peek); peek++) ; |
| 6384 | if (*peek && strchr("`'\"",*peek)) { |
| 6385 | s = peek; |
| 6386 | term = *s++; |
| 6387 | s = delimcpy(d, e, s, PL_bufend, term, &len); |
| 6388 | d += len; |
| 6389 | if (s < PL_bufend) |
| 6390 | s++; |
| 6391 | } |
| 6392 | else { |
| 6393 | if (*s == '\\') |
| 6394 | s++, term = '\''; |
| 6395 | else |
| 6396 | term = '"'; |
| 6397 | if (!isALNUM_lazy_if(s,UTF)) |
| 6398 | deprecate("bare << to mean <<\"\""); |
| 6399 | for (; isALNUM_lazy_if(s,UTF); s++) { |
| 6400 | if (d < e) |
| 6401 | *d++ = *s; |
| 6402 | } |
| 6403 | } |
| 6404 | if (d >= PL_tokenbuf + sizeof PL_tokenbuf - 1) |
| 6405 | Perl_croak(aTHX_ "Delimiter for here document is too long"); |
| 6406 | *d++ = '\n'; |
| 6407 | *d = '\0'; |
| 6408 | len = d - PL_tokenbuf; |
| 6409 | #ifndef PERL_STRICT_CR |
| 6410 | d = strchr(s, '\r'); |
| 6411 | if (d) { |
| 6412 | char *olds = s; |
| 6413 | s = d; |
| 6414 | while (s < PL_bufend) { |
| 6415 | if (*s == '\r') { |
| 6416 | *d++ = '\n'; |
| 6417 | if (*++s == '\n') |
| 6418 | s++; |
| 6419 | } |
| 6420 | else if (*s == '\n' && s[1] == '\r') { /* \015\013 on a mac? */ |
| 6421 | *d++ = *s++; |
| 6422 | s++; |
| 6423 | } |
| 6424 | else |
| 6425 | *d++ = *s++; |
| 6426 | } |
| 6427 | *d = '\0'; |
| 6428 | PL_bufend = d; |
| 6429 | SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr)); |
| 6430 | s = olds; |
| 6431 | } |
| 6432 | #endif |
| 6433 | d = "\n"; |
| 6434 | if (outer || !(d=ninstr(s,PL_bufend,d,d+1))) |
| 6435 | herewas = newSVpvn(s,PL_bufend-s); |
| 6436 | else |
| 6437 | s--, herewas = newSVpvn(s,d-s); |
| 6438 | s += SvCUR(herewas); |
| 6439 | |
| 6440 | tmpstr = NEWSV(87,79); |
| 6441 | sv_upgrade(tmpstr, SVt_PVIV); |
| 6442 | if (term == '\'') { |
| 6443 | op_type = OP_CONST; |
| 6444 | SvIVX(tmpstr) = -1; |
| 6445 | } |
| 6446 | else if (term == '`') { |
| 6447 | op_type = OP_BACKTICK; |
| 6448 | SvIVX(tmpstr) = '\\'; |
| 6449 | } |
| 6450 | |
| 6451 | CLINE; |
| 6452 | PL_multi_start = CopLINE(PL_curcop); |
| 6453 | PL_multi_open = PL_multi_close = '<'; |
| 6454 | term = *PL_tokenbuf; |
| 6455 | if (PL_lex_inwhat == OP_SUBST && PL_in_eval && !PL_rsfp) { |
| 6456 | char *bufptr = PL_sublex_info.super_bufptr; |
| 6457 | char *bufend = PL_sublex_info.super_bufend; |
| 6458 | char *olds = s - SvCUR(herewas); |
| 6459 | s = strchr(bufptr, '\n'); |
| 6460 | if (!s) |
| 6461 | s = bufend; |
| 6462 | d = s; |
| 6463 | while (s < bufend && |
| 6464 | (*s != term || memNE(s,PL_tokenbuf,len)) ) { |
| 6465 | if (*s++ == '\n') |
| 6466 | CopLINE_inc(PL_curcop); |
| 6467 | } |
| 6468 | if (s >= bufend) { |
| 6469 | CopLINE_set(PL_curcop, PL_multi_start); |
| 6470 | missingterm(PL_tokenbuf); |
| 6471 | } |
| 6472 | sv_setpvn(herewas,bufptr,d-bufptr+1); |
| 6473 | sv_setpvn(tmpstr,d+1,s-d); |
| 6474 | s += len - 1; |
| 6475 | sv_catpvn(herewas,s,bufend-s); |
| 6476 | (void)strcpy(bufptr,SvPVX(herewas)); |
| 6477 | |
| 6478 | s = olds; |
| 6479 | goto retval; |
| 6480 | } |
| 6481 | else if (!outer) { |
| 6482 | d = s; |
| 6483 | while (s < PL_bufend && |
| 6484 | (*s != term || memNE(s,PL_tokenbuf,len)) ) { |
| 6485 | if (*s++ == '\n') |
| 6486 | CopLINE_inc(PL_curcop); |
| 6487 | } |
| 6488 | if (s >= PL_bufend) { |
| 6489 | CopLINE_set(PL_curcop, PL_multi_start); |
| 6490 | missingterm(PL_tokenbuf); |
| 6491 | } |
| 6492 | sv_setpvn(tmpstr,d+1,s-d); |
| 6493 | s += len - 1; |
| 6494 | CopLINE_inc(PL_curcop); /* the preceding stmt passes a newline */ |
| 6495 | |
| 6496 | sv_catpvn(herewas,s,PL_bufend-s); |
| 6497 | sv_setsv(PL_linestr,herewas); |
| 6498 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = s = PL_linestart = SvPVX(PL_linestr); |
| 6499 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 6500 | PL_last_lop = PL_last_uni = Nullch; |
| 6501 | } |
| 6502 | else |
| 6503 | sv_setpvn(tmpstr,"",0); /* avoid "uninitialized" warning */ |
| 6504 | while (s >= PL_bufend) { /* multiple line string? */ |
| 6505 | if (!outer || |
| 6506 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { |
| 6507 | CopLINE_set(PL_curcop, PL_multi_start); |
| 6508 | missingterm(PL_tokenbuf); |
| 6509 | } |
| 6510 | CopLINE_inc(PL_curcop); |
| 6511 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 6512 | PL_last_lop = PL_last_uni = Nullch; |
| 6513 | #ifndef PERL_STRICT_CR |
| 6514 | if (PL_bufend - PL_linestart >= 2) { |
| 6515 | if ((PL_bufend[-2] == '\r' && PL_bufend[-1] == '\n') || |
| 6516 | (PL_bufend[-2] == '\n' && PL_bufend[-1] == '\r')) |
| 6517 | { |
| 6518 | PL_bufend[-2] = '\n'; |
| 6519 | PL_bufend--; |
| 6520 | SvCUR_set(PL_linestr, PL_bufend - SvPVX(PL_linestr)); |
| 6521 | } |
| 6522 | else if (PL_bufend[-1] == '\r') |
| 6523 | PL_bufend[-1] = '\n'; |
| 6524 | } |
| 6525 | else if (PL_bufend - PL_linestart == 1 && PL_bufend[-1] == '\r') |
| 6526 | PL_bufend[-1] = '\n'; |
| 6527 | #endif |
| 6528 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| 6529 | SV *sv = NEWSV(88,0); |
| 6530 | |
| 6531 | sv_upgrade(sv, SVt_PVMG); |
| 6532 | sv_setsv(sv,PL_linestr); |
| 6533 | av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop),sv); |
| 6534 | } |
| 6535 | if (*s == term && memEQ(s,PL_tokenbuf,len)) { |
| 6536 | s = PL_bufend - 1; |
| 6537 | *s = ' '; |
| 6538 | sv_catsv(PL_linestr,herewas); |
| 6539 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 6540 | } |
| 6541 | else { |
| 6542 | s = PL_bufend; |
| 6543 | sv_catsv(tmpstr,PL_linestr); |
| 6544 | } |
| 6545 | } |
| 6546 | s++; |
| 6547 | retval: |
| 6548 | PL_multi_end = CopLINE(PL_curcop); |
| 6549 | if (SvCUR(tmpstr) + 5 < SvLEN(tmpstr)) { |
| 6550 | SvLEN_set(tmpstr, SvCUR(tmpstr) + 1); |
| 6551 | Renew(SvPVX(tmpstr), SvLEN(tmpstr), char); |
| 6552 | } |
| 6553 | SvREFCNT_dec(herewas); |
| 6554 | if (UTF && !IN_BYTES && is_utf8_string((U8*)SvPVX(tmpstr), SvCUR(tmpstr))) |
| 6555 | SvUTF8_on(tmpstr); |
| 6556 | PL_lex_stuff = tmpstr; |
| 6557 | yylval.ival = op_type; |
| 6558 | return s; |
| 6559 | } |
| 6560 | |
| 6561 | /* scan_inputsymbol |
| 6562 | takes: current position in input buffer |
| 6563 | returns: new position in input buffer |
| 6564 | side-effects: yylval and lex_op are set. |
| 6565 | |
| 6566 | This code handles: |
| 6567 | |
| 6568 | <> read from ARGV |
| 6569 | <FH> read from filehandle |
| 6570 | <pkg::FH> read from package qualified filehandle |
| 6571 | <pkg'FH> read from package qualified filehandle |
| 6572 | <$fh> read from filehandle in $fh |
| 6573 | <*.h> filename glob |
| 6574 | |
| 6575 | */ |
| 6576 | |
| 6577 | STATIC char * |
| 6578 | S_scan_inputsymbol(pTHX_ char *start) |
| 6579 | { |
| 6580 | register char *s = start; /* current position in buffer */ |
| 6581 | register char *d; |
| 6582 | register char *e; |
| 6583 | char *end; |
| 6584 | I32 len; |
| 6585 | |
| 6586 | d = PL_tokenbuf; /* start of temp holding space */ |
| 6587 | e = PL_tokenbuf + sizeof PL_tokenbuf; /* end of temp holding space */ |
| 6588 | end = strchr(s, '\n'); |
| 6589 | if (!end) |
| 6590 | end = PL_bufend; |
| 6591 | s = delimcpy(d, e, s + 1, end, '>', &len); /* extract until > */ |
| 6592 | |
| 6593 | /* die if we didn't have space for the contents of the <>, |
| 6594 | or if it didn't end, or if we see a newline |
| 6595 | */ |
| 6596 | |
| 6597 | if (len >= sizeof PL_tokenbuf) |
| 6598 | Perl_croak(aTHX_ "Excessively long <> operator"); |
| 6599 | if (s >= end) |
| 6600 | Perl_croak(aTHX_ "Unterminated <> operator"); |
| 6601 | |
| 6602 | s++; |
| 6603 | |
| 6604 | /* check for <$fh> |
| 6605 | Remember, only scalar variables are interpreted as filehandles by |
| 6606 | this code. Anything more complex (e.g., <$fh{$num}>) will be |
| 6607 | treated as a glob() call. |
| 6608 | This code makes use of the fact that except for the $ at the front, |
| 6609 | a scalar variable and a filehandle look the same. |
| 6610 | */ |
| 6611 | if (*d == '$' && d[1]) d++; |
| 6612 | |
| 6613 | /* allow <Pkg'VALUE> or <Pkg::VALUE> */ |
| 6614 | while (*d && (isALNUM_lazy_if(d,UTF) || *d == '\'' || *d == ':')) |
| 6615 | d++; |
| 6616 | |
| 6617 | /* If we've tried to read what we allow filehandles to look like, and |
| 6618 | there's still text left, then it must be a glob() and not a getline. |
| 6619 | Use scan_str to pull out the stuff between the <> and treat it |
| 6620 | as nothing more than a string. |
| 6621 | */ |
| 6622 | |
| 6623 | if (d - PL_tokenbuf != len) { |
| 6624 | yylval.ival = OP_GLOB; |
| 6625 | set_csh(); |
| 6626 | s = scan_str(start,FALSE,FALSE); |
| 6627 | if (!s) |
| 6628 | Perl_croak(aTHX_ "Glob not terminated"); |
| 6629 | return s; |
| 6630 | } |
| 6631 | else { |
| 6632 | /* we're in a filehandle read situation */ |
| 6633 | d = PL_tokenbuf; |
| 6634 | |
| 6635 | /* turn <> into <ARGV> */ |
| 6636 | if (!len) |
| 6637 | (void)strcpy(d,"ARGV"); |
| 6638 | |
| 6639 | /* if <$fh>, create the ops to turn the variable into a |
| 6640 | filehandle |
| 6641 | */ |
| 6642 | if (*d == '$') { |
| 6643 | I32 tmp; |
| 6644 | |
| 6645 | /* try to find it in the pad for this block, otherwise find |
| 6646 | add symbol table ops |
| 6647 | */ |
| 6648 | if ((tmp = pad_findmy(d)) != NOT_IN_PAD) { |
| 6649 | OP *o = newOP(OP_PADSV, 0); |
| 6650 | o->op_targ = tmp; |
| 6651 | PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, o); |
| 6652 | } |
| 6653 | else { |
| 6654 | GV *gv = gv_fetchpv(d+1,TRUE, SVt_PV); |
| 6655 | PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, |
| 6656 | newUNOP(OP_RV2SV, 0, |
| 6657 | newGVOP(OP_GV, 0, gv))); |
| 6658 | } |
| 6659 | PL_lex_op->op_flags |= OPf_SPECIAL; |
| 6660 | /* we created the ops in PL_lex_op, so make yylval.ival a null op */ |
| 6661 | yylval.ival = OP_NULL; |
| 6662 | } |
| 6663 | |
| 6664 | /* If it's none of the above, it must be a literal filehandle |
| 6665 | (<Foo::BAR> or <FOO>) so build a simple readline OP */ |
| 6666 | else { |
| 6667 | GV *gv = gv_fetchpv(d,TRUE, SVt_PVIO); |
| 6668 | PL_lex_op = (OP*)newUNOP(OP_READLINE, 0, newGVOP(OP_GV, 0, gv)); |
| 6669 | yylval.ival = OP_NULL; |
| 6670 | } |
| 6671 | } |
| 6672 | |
| 6673 | return s; |
| 6674 | } |
| 6675 | |
| 6676 | |
| 6677 | /* scan_str |
| 6678 | takes: start position in buffer |
| 6679 | keep_quoted preserve \ on the embedded delimiter(s) |
| 6680 | keep_delims preserve the delimiters around the string |
| 6681 | returns: position to continue reading from buffer |
| 6682 | side-effects: multi_start, multi_close, lex_repl or lex_stuff, and |
| 6683 | updates the read buffer. |
| 6684 | |
| 6685 | This subroutine pulls a string out of the input. It is called for: |
| 6686 | q single quotes q(literal text) |
| 6687 | ' single quotes 'literal text' |
| 6688 | qq double quotes qq(interpolate $here please) |
| 6689 | " double quotes "interpolate $here please" |
| 6690 | qx backticks qx(/bin/ls -l) |
| 6691 | ` backticks `/bin/ls -l` |
| 6692 | qw quote words @EXPORT_OK = qw( func() $spam ) |
| 6693 | m// regexp match m/this/ |
| 6694 | s/// regexp substitute s/this/that/ |
| 6695 | tr/// string transliterate tr/this/that/ |
| 6696 | y/// string transliterate y/this/that/ |
| 6697 | ($*@) sub prototypes sub foo ($) |
| 6698 | (stuff) sub attr parameters sub foo : attr(stuff) |
| 6699 | <> readline or globs <FOO>, <>, <$fh>, or <*.c> |
| 6700 | |
| 6701 | In most of these cases (all but <>, patterns and transliterate) |
| 6702 | yylex() calls scan_str(). m// makes yylex() call scan_pat() which |
| 6703 | calls scan_str(). s/// makes yylex() call scan_subst() which calls |
| 6704 | scan_str(). tr/// and y/// make yylex() call scan_trans() which |
| 6705 | calls scan_str(). |
| 6706 | |
| 6707 | It skips whitespace before the string starts, and treats the first |
| 6708 | character as the delimiter. If the delimiter is one of ([{< then |
| 6709 | the corresponding "close" character )]}> is used as the closing |
| 6710 | delimiter. It allows quoting of delimiters, and if the string has |
| 6711 | balanced delimiters ([{<>}]) it allows nesting. |
| 6712 | |
| 6713 | On success, the SV with the resulting string is put into lex_stuff or, |
| 6714 | if that is already non-NULL, into lex_repl. The second case occurs only |
| 6715 | when parsing the RHS of the special constructs s/// and tr/// (y///). |
| 6716 | For convenience, the terminating delimiter character is stuffed into |
| 6717 | SvIVX of the SV. |
| 6718 | */ |
| 6719 | |
| 6720 | STATIC char * |
| 6721 | S_scan_str(pTHX_ char *start, int keep_quoted, int keep_delims) |
| 6722 | { |
| 6723 | SV *sv; /* scalar value: string */ |
| 6724 | char *tmps; /* temp string, used for delimiter matching */ |
| 6725 | register char *s = start; /* current position in the buffer */ |
| 6726 | register char term; /* terminating character */ |
| 6727 | register char *to; /* current position in the sv's data */ |
| 6728 | I32 brackets = 1; /* bracket nesting level */ |
| 6729 | bool has_utf8 = FALSE; /* is there any utf8 content? */ |
| 6730 | |
| 6731 | /* skip space before the delimiter */ |
| 6732 | if (isSPACE(*s)) |
| 6733 | s = skipspace(s); |
| 6734 | |
| 6735 | /* mark where we are, in case we need to report errors */ |
| 6736 | CLINE; |
| 6737 | |
| 6738 | /* after skipping whitespace, the next character is the terminator */ |
| 6739 | term = *s; |
| 6740 | if (!UTF8_IS_INVARIANT((U8)term) && UTF) |
| 6741 | has_utf8 = TRUE; |
| 6742 | |
| 6743 | /* mark where we are */ |
| 6744 | PL_multi_start = CopLINE(PL_curcop); |
| 6745 | PL_multi_open = term; |
| 6746 | |
| 6747 | /* find corresponding closing delimiter */ |
| 6748 | if (term && (tmps = strchr("([{< )]}> )]}>",term))) |
| 6749 | term = tmps[5]; |
| 6750 | PL_multi_close = term; |
| 6751 | |
| 6752 | /* create a new SV to hold the contents. 87 is leak category, I'm |
| 6753 | assuming. 79 is the SV's initial length. What a random number. */ |
| 6754 | sv = NEWSV(87,79); |
| 6755 | sv_upgrade(sv, SVt_PVIV); |
| 6756 | SvIVX(sv) = term; |
| 6757 | (void)SvPOK_only(sv); /* validate pointer */ |
| 6758 | |
| 6759 | /* move past delimiter and try to read a complete string */ |
| 6760 | if (keep_delims) |
| 6761 | sv_catpvn(sv, s, 1); |
| 6762 | s++; |
| 6763 | for (;;) { |
| 6764 | /* extend sv if need be */ |
| 6765 | SvGROW(sv, SvCUR(sv) + (PL_bufend - s) + 1); |
| 6766 | /* set 'to' to the next character in the sv's string */ |
| 6767 | to = SvPVX(sv)+SvCUR(sv); |
| 6768 | |
| 6769 | /* if open delimiter is the close delimiter read unbridle */ |
| 6770 | if (PL_multi_open == PL_multi_close) { |
| 6771 | for (; s < PL_bufend; s++,to++) { |
| 6772 | /* embedded newlines increment the current line number */ |
| 6773 | if (*s == '\n' && !PL_rsfp) |
| 6774 | CopLINE_inc(PL_curcop); |
| 6775 | /* handle quoted delimiters */ |
| 6776 | if (*s == '\\' && s+1 < PL_bufend && term != '\\') { |
| 6777 | if (!keep_quoted && s[1] == term) |
| 6778 | s++; |
| 6779 | /* any other quotes are simply copied straight through */ |
| 6780 | else |
| 6781 | *to++ = *s++; |
| 6782 | } |
| 6783 | /* terminate when run out of buffer (the for() condition), or |
| 6784 | have found the terminator */ |
| 6785 | else if (*s == term) |
| 6786 | break; |
| 6787 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
| 6788 | has_utf8 = TRUE; |
| 6789 | *to = *s; |
| 6790 | } |
| 6791 | } |
| 6792 | |
| 6793 | /* if the terminator isn't the same as the start character (e.g., |
| 6794 | matched brackets), we have to allow more in the quoting, and |
| 6795 | be prepared for nested brackets. |
| 6796 | */ |
| 6797 | else { |
| 6798 | /* read until we run out of string, or we find the terminator */ |
| 6799 | for (; s < PL_bufend; s++,to++) { |
| 6800 | /* embedded newlines increment the line count */ |
| 6801 | if (*s == '\n' && !PL_rsfp) |
| 6802 | CopLINE_inc(PL_curcop); |
| 6803 | /* backslashes can escape the open or closing characters */ |
| 6804 | if (*s == '\\' && s+1 < PL_bufend) { |
| 6805 | if (!keep_quoted && |
| 6806 | ((s[1] == PL_multi_open) || (s[1] == PL_multi_close))) |
| 6807 | s++; |
| 6808 | else |
| 6809 | *to++ = *s++; |
| 6810 | } |
| 6811 | /* allow nested opens and closes */ |
| 6812 | else if (*s == PL_multi_close && --brackets <= 0) |
| 6813 | break; |
| 6814 | else if (*s == PL_multi_open) |
| 6815 | brackets++; |
| 6816 | else if (!has_utf8 && !UTF8_IS_INVARIANT((U8)*s) && UTF) |
| 6817 | has_utf8 = TRUE; |
| 6818 | *to = *s; |
| 6819 | } |
| 6820 | } |
| 6821 | /* terminate the copied string and update the sv's end-of-string */ |
| 6822 | *to = '\0'; |
| 6823 | SvCUR_set(sv, to - SvPVX(sv)); |
| 6824 | |
| 6825 | /* |
| 6826 | * this next chunk reads more into the buffer if we're not done yet |
| 6827 | */ |
| 6828 | |
| 6829 | if (s < PL_bufend) |
| 6830 | break; /* handle case where we are done yet :-) */ |
| 6831 | |
| 6832 | #ifndef PERL_STRICT_CR |
| 6833 | if (to - SvPVX(sv) >= 2) { |
| 6834 | if ((to[-2] == '\r' && to[-1] == '\n') || |
| 6835 | (to[-2] == '\n' && to[-1] == '\r')) |
| 6836 | { |
| 6837 | to[-2] = '\n'; |
| 6838 | to--; |
| 6839 | SvCUR_set(sv, to - SvPVX(sv)); |
| 6840 | } |
| 6841 | else if (to[-1] == '\r') |
| 6842 | to[-1] = '\n'; |
| 6843 | } |
| 6844 | else if (to - SvPVX(sv) == 1 && to[-1] == '\r') |
| 6845 | to[-1] = '\n'; |
| 6846 | #endif |
| 6847 | |
| 6848 | /* if we're out of file, or a read fails, bail and reset the current |
| 6849 | line marker so we can report where the unterminated string began |
| 6850 | */ |
| 6851 | if (!PL_rsfp || |
| 6852 | !(PL_oldoldbufptr = PL_oldbufptr = s = PL_linestart = filter_gets(PL_linestr, PL_rsfp, 0))) { |
| 6853 | sv_free(sv); |
| 6854 | CopLINE_set(PL_curcop, PL_multi_start); |
| 6855 | return Nullch; |
| 6856 | } |
| 6857 | /* we read a line, so increment our line counter */ |
| 6858 | CopLINE_inc(PL_curcop); |
| 6859 | |
| 6860 | /* update debugger info */ |
| 6861 | if (PERLDB_LINE && PL_curstash != PL_debstash) { |
| 6862 | SV *sv = NEWSV(88,0); |
| 6863 | |
| 6864 | sv_upgrade(sv, SVt_PVMG); |
| 6865 | sv_setsv(sv,PL_linestr); |
| 6866 | av_store(CopFILEAV(PL_curcop), (I32)CopLINE(PL_curcop), sv); |
| 6867 | } |
| 6868 | |
| 6869 | /* having changed the buffer, we must update PL_bufend */ |
| 6870 | PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 6871 | PL_last_lop = PL_last_uni = Nullch; |
| 6872 | } |
| 6873 | |
| 6874 | /* at this point, we have successfully read the delimited string */ |
| 6875 | |
| 6876 | if (keep_delims) |
| 6877 | sv_catpvn(sv, s, 1); |
| 6878 | if (has_utf8) |
| 6879 | SvUTF8_on(sv); |
| 6880 | PL_multi_end = CopLINE(PL_curcop); |
| 6881 | s++; |
| 6882 | |
| 6883 | /* if we allocated too much space, give some back */ |
| 6884 | if (SvCUR(sv) + 5 < SvLEN(sv)) { |
| 6885 | SvLEN_set(sv, SvCUR(sv) + 1); |
| 6886 | Renew(SvPVX(sv), SvLEN(sv), char); |
| 6887 | } |
| 6888 | |
| 6889 | /* decide whether this is the first or second quoted string we've read |
| 6890 | for this op |
| 6891 | */ |
| 6892 | |
| 6893 | if (PL_lex_stuff) |
| 6894 | PL_lex_repl = sv; |
| 6895 | else |
| 6896 | PL_lex_stuff = sv; |
| 6897 | return s; |
| 6898 | } |
| 6899 | |
| 6900 | /* |
| 6901 | scan_num |
| 6902 | takes: pointer to position in buffer |
| 6903 | returns: pointer to new position in buffer |
| 6904 | side-effects: builds ops for the constant in yylval.op |
| 6905 | |
| 6906 | Read a number in any of the formats that Perl accepts: |
| 6907 | |
| 6908 | \d(_?\d)*(\.(\d(_?\d)*)?)?[Ee][\+\-]?(\d(_?\d)*) 12 12.34 12. |
| 6909 | \.\d(_?\d)*[Ee][\+\-]?(\d(_?\d)*) .34 |
| 6910 | 0b[01](_?[01])* |
| 6911 | 0[0-7](_?[0-7])* |
| 6912 | 0x[0-9A-Fa-f](_?[0-9A-Fa-f])* |
| 6913 | |
| 6914 | Like most scan_ routines, it uses the PL_tokenbuf buffer to hold the |
| 6915 | thing it reads. |
| 6916 | |
| 6917 | If it reads a number without a decimal point or an exponent, it will |
| 6918 | try converting the number to an integer and see if it can do so |
| 6919 | without loss of precision. |
| 6920 | */ |
| 6921 | |
| 6922 | char * |
| 6923 | Perl_scan_num(pTHX_ char *start, YYSTYPE* lvalp) |
| 6924 | { |
| 6925 | register char *s = start; /* current position in buffer */ |
| 6926 | register char *d; /* destination in temp buffer */ |
| 6927 | register char *e; /* end of temp buffer */ |
| 6928 | NV nv; /* number read, as a double */ |
| 6929 | SV *sv = Nullsv; /* place to put the converted number */ |
| 6930 | bool floatit; /* boolean: int or float? */ |
| 6931 | char *lastub = 0; /* position of last underbar */ |
| 6932 | static char number_too_long[] = "Number too long"; |
| 6933 | |
| 6934 | /* We use the first character to decide what type of number this is */ |
| 6935 | |
| 6936 | switch (*s) { |
| 6937 | default: |
| 6938 | Perl_croak(aTHX_ "panic: scan_num"); |
| 6939 | |
| 6940 | /* if it starts with a 0, it could be an octal number, a decimal in |
| 6941 | 0.13 disguise, or a hexadecimal number, or a binary number. */ |
| 6942 | case '0': |
| 6943 | { |
| 6944 | /* variables: |
| 6945 | u holds the "number so far" |
| 6946 | shift the power of 2 of the base |
| 6947 | (hex == 4, octal == 3, binary == 1) |
| 6948 | overflowed was the number more than we can hold? |
| 6949 | |
| 6950 | Shift is used when we add a digit. It also serves as an "are |
| 6951 | we in octal/hex/binary?" indicator to disallow hex characters |
| 6952 | when in octal mode. |
| 6953 | */ |
| 6954 | NV n = 0.0; |
| 6955 | UV u = 0; |
| 6956 | I32 shift; |
| 6957 | bool overflowed = FALSE; |
| 6958 | static NV nvshift[5] = { 1.0, 2.0, 4.0, 8.0, 16.0 }; |
| 6959 | static char* bases[5] = { "", "binary", "", "octal", |
| 6960 | "hexadecimal" }; |
| 6961 | static char* Bases[5] = { "", "Binary", "", "Octal", |
| 6962 | "Hexadecimal" }; |
| 6963 | static char *maxima[5] = { "", |
| 6964 | "0b11111111111111111111111111111111", |
| 6965 | "", |
| 6966 | "037777777777", |
| 6967 | "0xffffffff" }; |
| 6968 | char *base, *Base, *max; |
| 6969 | |
| 6970 | /* check for hex */ |
| 6971 | if (s[1] == 'x') { |
| 6972 | shift = 4; |
| 6973 | s += 2; |
| 6974 | } else if (s[1] == 'b') { |
| 6975 | shift = 1; |
| 6976 | s += 2; |
| 6977 | } |
| 6978 | /* check for a decimal in disguise */ |
| 6979 | else if (s[1] == '.' || s[1] == 'e' || s[1] == 'E') |
| 6980 | goto decimal; |
| 6981 | /* so it must be octal */ |
| 6982 | else { |
| 6983 | shift = 3; |
| 6984 | s++; |
| 6985 | } |
| 6986 | |
| 6987 | if (*s == '_') { |
| 6988 | if (ckWARN(WARN_SYNTAX)) |
| 6989 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 6990 | "Misplaced _ in number"); |
| 6991 | lastub = s++; |
| 6992 | } |
| 6993 | |
| 6994 | base = bases[shift]; |
| 6995 | Base = Bases[shift]; |
| 6996 | max = maxima[shift]; |
| 6997 | |
| 6998 | /* read the rest of the number */ |
| 6999 | for (;;) { |
| 7000 | /* x is used in the overflow test, |
| 7001 | b is the digit we're adding on. */ |
| 7002 | UV x, b; |
| 7003 | |
| 7004 | switch (*s) { |
| 7005 | |
| 7006 | /* if we don't mention it, we're done */ |
| 7007 | default: |
| 7008 | goto out; |
| 7009 | |
| 7010 | /* _ are ignored -- but warned about if consecutive */ |
| 7011 | case '_': |
| 7012 | if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1) |
| 7013 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7014 | "Misplaced _ in number"); |
| 7015 | lastub = s++; |
| 7016 | break; |
| 7017 | |
| 7018 | /* 8 and 9 are not octal */ |
| 7019 | case '8': case '9': |
| 7020 | if (shift == 3) |
| 7021 | yyerror(Perl_form(aTHX_ "Illegal octal digit '%c'", *s)); |
| 7022 | /* FALL THROUGH */ |
| 7023 | |
| 7024 | /* octal digits */ |
| 7025 | case '2': case '3': case '4': |
| 7026 | case '5': case '6': case '7': |
| 7027 | if (shift == 1) |
| 7028 | yyerror(Perl_form(aTHX_ "Illegal binary digit '%c'", *s)); |
| 7029 | /* FALL THROUGH */ |
| 7030 | |
| 7031 | case '0': case '1': |
| 7032 | b = *s++ & 15; /* ASCII digit -> value of digit */ |
| 7033 | goto digit; |
| 7034 | |
| 7035 | /* hex digits */ |
| 7036 | case 'a': case 'b': case 'c': case 'd': case 'e': case 'f': |
| 7037 | case 'A': case 'B': case 'C': case 'D': case 'E': case 'F': |
| 7038 | /* make sure they said 0x */ |
| 7039 | if (shift != 4) |
| 7040 | goto out; |
| 7041 | b = (*s++ & 7) + 9; |
| 7042 | |
| 7043 | /* Prepare to put the digit we have onto the end |
| 7044 | of the number so far. We check for overflows. |
| 7045 | */ |
| 7046 | |
| 7047 | digit: |
| 7048 | if (!overflowed) { |
| 7049 | x = u << shift; /* make room for the digit */ |
| 7050 | |
| 7051 | if ((x >> shift) != u |
| 7052 | && !(PL_hints & HINT_NEW_BINARY)) { |
| 7053 | overflowed = TRUE; |
| 7054 | n = (NV) u; |
| 7055 | if (ckWARN_d(WARN_OVERFLOW)) |
| 7056 | Perl_warner(aTHX_ WARN_OVERFLOW, |
| 7057 | "Integer overflow in %s number", |
| 7058 | base); |
| 7059 | } else |
| 7060 | u = x | b; /* add the digit to the end */ |
| 7061 | } |
| 7062 | if (overflowed) { |
| 7063 | n *= nvshift[shift]; |
| 7064 | /* If an NV has not enough bits in its |
| 7065 | * mantissa to represent an UV this summing of |
| 7066 | * small low-order numbers is a waste of time |
| 7067 | * (because the NV cannot preserve the |
| 7068 | * low-order bits anyway): we could just |
| 7069 | * remember when did we overflow and in the |
| 7070 | * end just multiply n by the right |
| 7071 | * amount. */ |
| 7072 | n += (NV) b; |
| 7073 | } |
| 7074 | break; |
| 7075 | } |
| 7076 | } |
| 7077 | |
| 7078 | /* if we get here, we had success: make a scalar value from |
| 7079 | the number. |
| 7080 | */ |
| 7081 | out: |
| 7082 | |
| 7083 | /* final misplaced underbar check */ |
| 7084 | if (s[-1] == '_') { |
| 7085 | if (ckWARN(WARN_SYNTAX)) |
| 7086 | Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number"); |
| 7087 | } |
| 7088 | |
| 7089 | sv = NEWSV(92,0); |
| 7090 | if (overflowed) { |
| 7091 | if (ckWARN(WARN_PORTABLE) && n > 4294967295.0) |
| 7092 | Perl_warner(aTHX_ WARN_PORTABLE, |
| 7093 | "%s number > %s non-portable", |
| 7094 | Base, max); |
| 7095 | sv_setnv(sv, n); |
| 7096 | } |
| 7097 | else { |
| 7098 | #if UVSIZE > 4 |
| 7099 | if (ckWARN(WARN_PORTABLE) && u > 0xffffffff) |
| 7100 | Perl_warner(aTHX_ WARN_PORTABLE, |
| 7101 | "%s number > %s non-portable", |
| 7102 | Base, max); |
| 7103 | #endif |
| 7104 | sv_setuv(sv, u); |
| 7105 | } |
| 7106 | if (PL_hints & HINT_NEW_BINARY) |
| 7107 | sv = new_constant(start, s - start, "binary", sv, Nullsv, NULL); |
| 7108 | } |
| 7109 | break; |
| 7110 | |
| 7111 | /* |
| 7112 | handle decimal numbers. |
| 7113 | we're also sent here when we read a 0 as the first digit |
| 7114 | */ |
| 7115 | case '1': case '2': case '3': case '4': case '5': |
| 7116 | case '6': case '7': case '8': case '9': case '.': |
| 7117 | decimal: |
| 7118 | d = PL_tokenbuf; |
| 7119 | e = PL_tokenbuf + sizeof PL_tokenbuf - 6; /* room for various punctuation */ |
| 7120 | floatit = FALSE; |
| 7121 | |
| 7122 | /* read next group of digits and _ and copy into d */ |
| 7123 | while (isDIGIT(*s) || *s == '_') { |
| 7124 | /* skip underscores, checking for misplaced ones |
| 7125 | if -w is on |
| 7126 | */ |
| 7127 | if (*s == '_') { |
| 7128 | if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1) |
| 7129 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7130 | "Misplaced _ in number"); |
| 7131 | lastub = s++; |
| 7132 | } |
| 7133 | else { |
| 7134 | /* check for end of fixed-length buffer */ |
| 7135 | if (d >= e) |
| 7136 | Perl_croak(aTHX_ number_too_long); |
| 7137 | /* if we're ok, copy the character */ |
| 7138 | *d++ = *s++; |
| 7139 | } |
| 7140 | } |
| 7141 | |
| 7142 | /* final misplaced underbar check */ |
| 7143 | if (lastub && s == lastub + 1) { |
| 7144 | if (ckWARN(WARN_SYNTAX)) |
| 7145 | Perl_warner(aTHX_ WARN_SYNTAX, "Misplaced _ in number"); |
| 7146 | } |
| 7147 | |
| 7148 | /* read a decimal portion if there is one. avoid |
| 7149 | 3..5 being interpreted as the number 3. followed |
| 7150 | by .5 |
| 7151 | */ |
| 7152 | if (*s == '.' && s[1] != '.') { |
| 7153 | floatit = TRUE; |
| 7154 | *d++ = *s++; |
| 7155 | |
| 7156 | if (*s == '_') { |
| 7157 | if (ckWARN(WARN_SYNTAX)) |
| 7158 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7159 | "Misplaced _ in number"); |
| 7160 | lastub = s; |
| 7161 | } |
| 7162 | |
| 7163 | /* copy, ignoring underbars, until we run out of digits. |
| 7164 | */ |
| 7165 | for (; isDIGIT(*s) || *s == '_'; s++) { |
| 7166 | /* fixed length buffer check */ |
| 7167 | if (d >= e) |
| 7168 | Perl_croak(aTHX_ number_too_long); |
| 7169 | if (*s == '_') { |
| 7170 | if (ckWARN(WARN_SYNTAX) && lastub && s == lastub + 1) |
| 7171 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7172 | "Misplaced _ in number"); |
| 7173 | lastub = s; |
| 7174 | } |
| 7175 | else |
| 7176 | *d++ = *s; |
| 7177 | } |
| 7178 | /* fractional part ending in underbar? */ |
| 7179 | if (s[-1] == '_') { |
| 7180 | if (ckWARN(WARN_SYNTAX)) |
| 7181 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7182 | "Misplaced _ in number"); |
| 7183 | } |
| 7184 | if (*s == '.' && isDIGIT(s[1])) { |
| 7185 | /* oops, it's really a v-string, but without the "v" */ |
| 7186 | s = start - 1; |
| 7187 | goto vstring; |
| 7188 | } |
| 7189 | } |
| 7190 | |
| 7191 | /* read exponent part, if present */ |
| 7192 | if (*s && strchr("eE",*s) && strchr("+-0123456789_", s[1])) { |
| 7193 | floatit = TRUE; |
| 7194 | s++; |
| 7195 | |
| 7196 | /* regardless of whether user said 3E5 or 3e5, use lower 'e' */ |
| 7197 | *d++ = 'e'; /* At least some Mach atof()s don't grok 'E' */ |
| 7198 | |
| 7199 | /* stray preinitial _ */ |
| 7200 | if (*s == '_') { |
| 7201 | if (ckWARN(WARN_SYNTAX)) |
| 7202 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7203 | "Misplaced _ in number"); |
| 7204 | lastub = s++; |
| 7205 | } |
| 7206 | |
| 7207 | /* allow positive or negative exponent */ |
| 7208 | if (*s == '+' || *s == '-') |
| 7209 | *d++ = *s++; |
| 7210 | |
| 7211 | /* stray initial _ */ |
| 7212 | if (*s == '_') { |
| 7213 | if (ckWARN(WARN_SYNTAX)) |
| 7214 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7215 | "Misplaced _ in number"); |
| 7216 | lastub = s++; |
| 7217 | } |
| 7218 | |
| 7219 | /* read digits of exponent */ |
| 7220 | while (isDIGIT(*s) || *s == '_') { |
| 7221 | if (isDIGIT(*s)) { |
| 7222 | if (d >= e) |
| 7223 | Perl_croak(aTHX_ number_too_long); |
| 7224 | *d++ = *s++; |
| 7225 | } |
| 7226 | else { |
| 7227 | if (ckWARN(WARN_SYNTAX) && |
| 7228 | ((lastub && s == lastub + 1) || |
| 7229 | (!isDIGIT(s[1]) && s[1] != '_'))) |
| 7230 | Perl_warner(aTHX_ WARN_SYNTAX, |
| 7231 | "Misplaced _ in number"); |
| 7232 | lastub = s++; |
| 7233 | } |
| 7234 | } |
| 7235 | } |
| 7236 | |
| 7237 | |
| 7238 | /* make an sv from the string */ |
| 7239 | sv = NEWSV(92,0); |
| 7240 | |
| 7241 | /* |
| 7242 | We try to do an integer conversion first if no characters |
| 7243 | indicating "float" have been found. |
| 7244 | */ |
| 7245 | |
| 7246 | if (!floatit) { |
| 7247 | UV uv; |
| 7248 | int flags = grok_number (PL_tokenbuf, d - PL_tokenbuf, &uv); |
| 7249 | |
| 7250 | if (flags == IS_NUMBER_IN_UV) { |
| 7251 | if (uv <= IV_MAX) |
| 7252 | sv_setiv(sv, uv); /* Prefer IVs over UVs. */ |
| 7253 | else |
| 7254 | sv_setuv(sv, uv); |
| 7255 | } else if (flags == (IS_NUMBER_IN_UV | IS_NUMBER_NEG)) { |
| 7256 | if (uv <= (UV) IV_MIN) |
| 7257 | sv_setiv(sv, -(IV)uv); |
| 7258 | else |
| 7259 | floatit = TRUE; |
| 7260 | } else |
| 7261 | floatit = TRUE; |
| 7262 | } |
| 7263 | if (floatit) { |
| 7264 | /* terminate the string */ |
| 7265 | *d = '\0'; |
| 7266 | nv = Atof(PL_tokenbuf); |
| 7267 | sv_setnv(sv, nv); |
| 7268 | } |
| 7269 | |
| 7270 | if ( floatit ? (PL_hints & HINT_NEW_FLOAT) : |
| 7271 | (PL_hints & HINT_NEW_INTEGER) ) |
| 7272 | sv = new_constant(PL_tokenbuf, d - PL_tokenbuf, |
| 7273 | (floatit ? "float" : "integer"), |
| 7274 | sv, Nullsv, NULL); |
| 7275 | break; |
| 7276 | |
| 7277 | /* if it starts with a v, it could be a v-string */ |
| 7278 | case 'v': |
| 7279 | vstring: |
| 7280 | { |
| 7281 | char *pos = s; |
| 7282 | pos++; |
| 7283 | while (isDIGIT(*pos) || *pos == '_') |
| 7284 | pos++; |
| 7285 | if (!isALPHA(*pos)) { |
| 7286 | UV rev; |
| 7287 | U8 tmpbuf[UTF8_MAXLEN+1]; |
| 7288 | U8 *tmpend; |
| 7289 | s++; /* get past 'v' */ |
| 7290 | |
| 7291 | sv = NEWSV(92,5); |
| 7292 | sv_setpvn(sv, "", 0); |
| 7293 | |
| 7294 | for (;;) { |
| 7295 | if (*s == '0' && isDIGIT(s[1])) |
| 7296 | yyerror("Octal number in vector unsupported"); |
| 7297 | rev = 0; |
| 7298 | { |
| 7299 | /* this is atoi() that tolerates underscores */ |
| 7300 | char *end = pos; |
| 7301 | UV mult = 1; |
| 7302 | while (--end >= s) { |
| 7303 | UV orev; |
| 7304 | if (*end == '_') |
| 7305 | continue; |
| 7306 | orev = rev; |
| 7307 | rev += (*end - '0') * mult; |
| 7308 | mult *= 10; |
| 7309 | if (orev > rev && ckWARN_d(WARN_OVERFLOW)) |
| 7310 | Perl_warner(aTHX_ WARN_OVERFLOW, |
| 7311 | "Integer overflow in decimal number"); |
| 7312 | } |
| 7313 | } |
| 7314 | /* Append native character for the rev point */ |
| 7315 | tmpend = uvchr_to_utf8(tmpbuf, rev); |
| 7316 | sv_catpvn(sv, (const char*)tmpbuf, tmpend - tmpbuf); |
| 7317 | if (!UNI_IS_INVARIANT(NATIVE_TO_UNI(rev))) |
| 7318 | SvUTF8_on(sv); |
| 7319 | if (*pos == '.' && isDIGIT(pos[1])) |
| 7320 | s = ++pos; |
| 7321 | else { |
| 7322 | s = pos; |
| 7323 | break; |
| 7324 | } |
| 7325 | while (isDIGIT(*pos) || *pos == '_') |
| 7326 | pos++; |
| 7327 | } |
| 7328 | SvPOK_on(sv); |
| 7329 | SvREADONLY_on(sv); |
| 7330 | } |
| 7331 | } |
| 7332 | break; |
| 7333 | } |
| 7334 | |
| 7335 | /* make the op for the constant and return */ |
| 7336 | |
| 7337 | if (sv) |
| 7338 | lvalp->opval = newSVOP(OP_CONST, 0, sv); |
| 7339 | else |
| 7340 | lvalp->opval = Nullop; |
| 7341 | |
| 7342 | return s; |
| 7343 | } |
| 7344 | |
| 7345 | STATIC char * |
| 7346 | S_scan_formline(pTHX_ register char *s) |
| 7347 | { |
| 7348 | register char *eol; |
| 7349 | register char *t; |
| 7350 | SV *stuff = newSVpvn("",0); |
| 7351 | bool needargs = FALSE; |
| 7352 | |
| 7353 | while (!needargs) { |
| 7354 | if (*s == '.' || *s == /*{*/'}') { |
| 7355 | /*SUPPRESS 530*/ |
| 7356 | #ifdef PERL_STRICT_CR |
| 7357 | for (t = s+1;SPACE_OR_TAB(*t); t++) ; |
| 7358 | #else |
| 7359 | for (t = s+1;SPACE_OR_TAB(*t) || *t == '\r'; t++) ; |
| 7360 | #endif |
| 7361 | if (*t == '\n' || t == PL_bufend) |
| 7362 | break; |
| 7363 | } |
| 7364 | if (PL_in_eval && !PL_rsfp) { |
| 7365 | eol = strchr(s,'\n'); |
| 7366 | if (!eol++) |
| 7367 | eol = PL_bufend; |
| 7368 | } |
| 7369 | else |
| 7370 | eol = PL_bufend = SvPVX(PL_linestr) + SvCUR(PL_linestr); |
| 7371 | if (*s != '#') { |
| 7372 | for (t = s; t < eol; t++) { |
| 7373 | if (*t == '~' && t[1] == '~' && SvCUR(stuff)) { |
| 7374 | needargs = FALSE; |
| 7375 | goto enough; /* ~~ must be first line in formline */ |
| 7376 | } |
| 7377 | if (*t == '@' || *t == '^') |
| 7378 | needargs = TRUE; |
| 7379 | } |
| 7380 | if (eol > s) { |
| 7381 | sv_catpvn(stuff, s, eol-s); |
| 7382 | #ifndef PERL_STRICT_CR |
| 7383 | if (eol-s > 1 && eol[-2] == '\r' && eol[-1] == '\n') { |
| 7384 | char *end = SvPVX(stuff) + SvCUR(stuff); |
| 7385 | end[-2] = '\n'; |
| 7386 | end[-1] = '\0'; |
| 7387 | SvCUR(stuff)--; |
| 7388 | } |
| 7389 | #endif |
| 7390 | } |
| 7391 | else |
| 7392 | break; |
| 7393 | } |
| 7394 | s = eol; |
| 7395 | if (PL_rsfp) { |
| 7396 | s = filter_gets(PL_linestr, PL_rsfp, 0); |
| 7397 | PL_oldoldbufptr = PL_oldbufptr = PL_bufptr = PL_linestart = SvPVX(PL_linestr); |
| 7398 | PL_bufend = PL_bufptr + SvCUR(PL_linestr); |
| 7399 | PL_last_lop = PL_last_uni = Nullch; |
| 7400 | if (!s) { |
| 7401 | s = PL_bufptr; |
| 7402 | yyerror("Format not terminated"); |
| 7403 | break; |
| 7404 | } |
| 7405 | } |
| 7406 | incline(s); |
| 7407 | } |
| 7408 | enough: |
| 7409 | if (SvCUR(stuff)) { |
| 7410 | PL_expect = XTERM; |
| 7411 | if (needargs) { |
| 7412 | PL_lex_state = LEX_NORMAL; |
| 7413 | PL_nextval[PL_nexttoke].ival = 0; |
| 7414 | force_next(','); |
| 7415 | } |
| 7416 | else |
| 7417 | PL_lex_state = LEX_FORMLINE; |
| 7418 | PL_nextval[PL_nexttoke].opval = (OP*)newSVOP(OP_CONST, 0, stuff); |
| 7419 | force_next(THING); |
| 7420 | PL_nextval[PL_nexttoke].ival = OP_FORMLINE; |
| 7421 | force_next(LSTOP); |
| 7422 | } |
| 7423 | else { |
| 7424 | SvREFCNT_dec(stuff); |
| 7425 | PL_lex_formbrack = 0; |
| 7426 | PL_bufptr = s; |
| 7427 | } |
| 7428 | return s; |
| 7429 | } |
| 7430 | |
| 7431 | STATIC void |
| 7432 | S_set_csh(pTHX) |
| 7433 | { |
| 7434 | #ifdef CSH |
| 7435 | if (!PL_cshlen) |
| 7436 | PL_cshlen = strlen(PL_cshname); |
| 7437 | #endif |
| 7438 | } |
| 7439 | |
| 7440 | I32 |
| 7441 | Perl_start_subparse(pTHX_ I32 is_format, U32 flags) |
| 7442 | { |
| 7443 | I32 oldsavestack_ix = PL_savestack_ix; |
| 7444 | CV* outsidecv = PL_compcv; |
| 7445 | AV* comppadlist; |
| 7446 | |
| 7447 | if (PL_compcv) { |
| 7448 | assert(SvTYPE(PL_compcv) == SVt_PVCV); |
| 7449 | } |
| 7450 | SAVEI32(PL_subline); |
| 7451 | save_item(PL_subname); |
| 7452 | SAVEI32(PL_padix); |
| 7453 | SAVECOMPPAD(); |
| 7454 | SAVESPTR(PL_comppad_name); |
| 7455 | SAVESPTR(PL_compcv); |
| 7456 | SAVEI32(PL_comppad_name_fill); |
| 7457 | SAVEI32(PL_min_intro_pending); |
| 7458 | SAVEI32(PL_max_intro_pending); |
| 7459 | SAVEI32(PL_pad_reset_pending); |
| 7460 | |
| 7461 | PL_compcv = (CV*)NEWSV(1104,0); |
| 7462 | sv_upgrade((SV *)PL_compcv, is_format ? SVt_PVFM : SVt_PVCV); |
| 7463 | CvFLAGS(PL_compcv) |= flags; |
| 7464 | |
| 7465 | PL_comppad = newAV(); |
| 7466 | av_push(PL_comppad, Nullsv); |
| 7467 | PL_curpad = AvARRAY(PL_comppad); |
| 7468 | PL_comppad_name = newAV(); |
| 7469 | PL_comppad_name_fill = 0; |
| 7470 | PL_min_intro_pending = 0; |
| 7471 | PL_padix = 0; |
| 7472 | PL_subline = CopLINE(PL_curcop); |
| 7473 | #ifdef USE_5005THREADS |
| 7474 | av_store(PL_comppad_name, 0, newSVpvn("@_", 2)); |
| 7475 | PL_curpad[0] = (SV*)newAV(); |
| 7476 | SvPADMY_on(PL_curpad[0]); /* XXX Needed? */ |
| 7477 | #endif /* USE_5005THREADS */ |
| 7478 | |
| 7479 | comppadlist = newAV(); |
| 7480 | AvREAL_off(comppadlist); |
| 7481 | av_store(comppadlist, 0, (SV*)PL_comppad_name); |
| 7482 | av_store(comppadlist, 1, (SV*)PL_comppad); |
| 7483 | |
| 7484 | CvPADLIST(PL_compcv) = comppadlist; |
| 7485 | CvOUTSIDE(PL_compcv) = (CV*)SvREFCNT_inc(outsidecv); |
| 7486 | #ifdef USE_5005THREADS |
| 7487 | CvOWNER(PL_compcv) = 0; |
| 7488 | New(666, CvMUTEXP(PL_compcv), 1, perl_mutex); |
| 7489 | MUTEX_INIT(CvMUTEXP(PL_compcv)); |
| 7490 | #endif /* USE_5005THREADS */ |
| 7491 | |
| 7492 | return oldsavestack_ix; |
| 7493 | } |
| 7494 | |
| 7495 | #ifdef __SC__ |
| 7496 | #pragma segment Perl_yylex |
| 7497 | #endif |
| 7498 | int |
| 7499 | Perl_yywarn(pTHX_ char *s) |
| 7500 | { |
| 7501 | PL_in_eval |= EVAL_WARNONLY; |
| 7502 | yyerror(s); |
| 7503 | PL_in_eval &= ~EVAL_WARNONLY; |
| 7504 | return 0; |
| 7505 | } |
| 7506 | |
| 7507 | int |
| 7508 | Perl_yyerror(pTHX_ char *s) |
| 7509 | { |
| 7510 | char *where = NULL; |
| 7511 | char *context = NULL; |
| 7512 | int contlen = -1; |
| 7513 | SV *msg; |
| 7514 | |
| 7515 | if (!yychar || (yychar == ';' && !PL_rsfp)) |
| 7516 | where = "at EOF"; |
| 7517 | else if (PL_bufptr > PL_oldoldbufptr && PL_bufptr - PL_oldoldbufptr < 200 && |
| 7518 | PL_oldoldbufptr != PL_oldbufptr && PL_oldbufptr != PL_bufptr) { |
| 7519 | while (isSPACE(*PL_oldoldbufptr)) |
| 7520 | PL_oldoldbufptr++; |
| 7521 | context = PL_oldoldbufptr; |
| 7522 | contlen = PL_bufptr - PL_oldoldbufptr; |
| 7523 | } |
| 7524 | else if (PL_bufptr > PL_oldbufptr && PL_bufptr - PL_oldbufptr < 200 && |
| 7525 | PL_oldbufptr != PL_bufptr) { |
| 7526 | while (isSPACE(*PL_oldbufptr)) |
| 7527 | PL_oldbufptr++; |
| 7528 | context = PL_oldbufptr; |
| 7529 | contlen = PL_bufptr - PL_oldbufptr; |
| 7530 | } |
| 7531 | else if (yychar > 255) |
| 7532 | where = "next token ???"; |
| 7533 | #ifdef USE_PURE_BISON |
| 7534 | /* GNU Bison sets the value -2 */ |
| 7535 | else if (yychar == -2) { |
| 7536 | #else |
| 7537 | else if ((yychar & 127) == 127) { |
| 7538 | #endif |
| 7539 | if (PL_lex_state == LEX_NORMAL || |
| 7540 | (PL_lex_state == LEX_KNOWNEXT && PL_lex_defer == LEX_NORMAL)) |
| 7541 | where = "at end of line"; |
| 7542 | else if (PL_lex_inpat) |
| 7543 | where = "within pattern"; |
| 7544 | else |
| 7545 | where = "within string"; |
| 7546 | } |
| 7547 | else { |
| 7548 | SV *where_sv = sv_2mortal(newSVpvn("next char ", 10)); |
| 7549 | if (yychar < 32) |
| 7550 | Perl_sv_catpvf(aTHX_ where_sv, "^%c", toCTRL(yychar)); |
| 7551 | else if (isPRINT_LC(yychar)) |
| 7552 | Perl_sv_catpvf(aTHX_ where_sv, "%c", yychar); |
| 7553 | else |
| 7554 | Perl_sv_catpvf(aTHX_ where_sv, "\\%03o", yychar & 255); |
| 7555 | where = SvPVX(where_sv); |
| 7556 | } |
| 7557 | msg = sv_2mortal(newSVpv(s, 0)); |
| 7558 | Perl_sv_catpvf(aTHX_ msg, " at %s line %"IVdf", ", |
| 7559 | CopFILE(PL_curcop), (IV)CopLINE(PL_curcop)); |
| 7560 | if (context) |
| 7561 | Perl_sv_catpvf(aTHX_ msg, "near \"%.*s\"\n", contlen, context); |
| 7562 | else |
| 7563 | Perl_sv_catpvf(aTHX_ msg, "%s\n", where); |
| 7564 | if (PL_multi_start < PL_multi_end && (U32)(CopLINE(PL_curcop) - PL_multi_end) <= 1) { |
| 7565 | Perl_sv_catpvf(aTHX_ msg, |
| 7566 | " (Might be a runaway multi-line %c%c string starting on line %"IVdf")\n", |
| 7567 | (int)PL_multi_open,(int)PL_multi_close,(IV)PL_multi_start); |
| 7568 | PL_multi_end = 0; |
| 7569 | } |
| 7570 | if (PL_in_eval & EVAL_WARNONLY) |
| 7571 | Perl_warn(aTHX_ "%"SVf, msg); |
| 7572 | else |
| 7573 | qerror(msg); |
| 7574 | if (PL_error_count >= 10) { |
| 7575 | if (PL_in_eval && SvCUR(ERRSV)) |
| 7576 | Perl_croak(aTHX_ "%"SVf"%s has too many errors.\n", |
| 7577 | ERRSV, CopFILE(PL_curcop)); |
| 7578 | else |
| 7579 | Perl_croak(aTHX_ "%s has too many errors.\n", |
| 7580 | CopFILE(PL_curcop)); |
| 7581 | } |
| 7582 | PL_in_my = 0; |
| 7583 | PL_in_my_stash = Nullhv; |
| 7584 | return 0; |
| 7585 | } |
| 7586 | #ifdef __SC__ |
| 7587 | #pragma segment Main |
| 7588 | #endif |
| 7589 | |
| 7590 | STATIC char* |
| 7591 | S_swallow_bom(pTHX_ U8 *s) |
| 7592 | { |
| 7593 | STRLEN slen; |
| 7594 | slen = SvCUR(PL_linestr); |
| 7595 | switch (*s) { |
| 7596 | case 0xFF: |
| 7597 | if (s[1] == 0xFE) { |
| 7598 | /* UTF-16 little-endian */ |
| 7599 | if (s[2] == 0 && s[3] == 0) /* UTF-32 little-endian */ |
| 7600 | Perl_croak(aTHX_ "Unsupported script encoding"); |
| 7601 | #ifndef PERL_NO_UTF16_FILTER |
| 7602 | DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-LE script encoding\n")); |
| 7603 | s += 2; |
| 7604 | if (PL_bufend > (char*)s) { |
| 7605 | U8 *news; |
| 7606 | I32 newlen; |
| 7607 | |
| 7608 | filter_add(utf16rev_textfilter, NULL); |
| 7609 | New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8); |
| 7610 | PL_bufend = (char*)utf16_to_utf8_reversed(s, news, |
| 7611 | PL_bufend - (char*)s - 1, |
| 7612 | &newlen); |
| 7613 | Copy(news, s, newlen, U8); |
| 7614 | SvCUR_set(PL_linestr, newlen); |
| 7615 | PL_bufend = SvPVX(PL_linestr) + newlen; |
| 7616 | news[newlen++] = '\0'; |
| 7617 | Safefree(news); |
| 7618 | } |
| 7619 | #else |
| 7620 | Perl_croak(aTHX_ "Unsupported script encoding"); |
| 7621 | #endif |
| 7622 | } |
| 7623 | break; |
| 7624 | case 0xFE: |
| 7625 | if (s[1] == 0xFF) { /* UTF-16 big-endian */ |
| 7626 | #ifndef PERL_NO_UTF16_FILTER |
| 7627 | DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-16BE script encoding\n")); |
| 7628 | s += 2; |
| 7629 | if (PL_bufend > (char *)s) { |
| 7630 | U8 *news; |
| 7631 | I32 newlen; |
| 7632 | |
| 7633 | filter_add(utf16_textfilter, NULL); |
| 7634 | New(898, news, (PL_bufend - (char*)s) * 3 / 2 + 1, U8); |
| 7635 | PL_bufend = (char*)utf16_to_utf8(s, news, |
| 7636 | PL_bufend - (char*)s, |
| 7637 | &newlen); |
| 7638 | Copy(news, s, newlen, U8); |
| 7639 | SvCUR_set(PL_linestr, newlen); |
| 7640 | PL_bufend = SvPVX(PL_linestr) + newlen; |
| 7641 | news[newlen++] = '\0'; |
| 7642 | Safefree(news); |
| 7643 | } |
| 7644 | #else |
| 7645 | Perl_croak(aTHX_ "Unsupported script encoding"); |
| 7646 | #endif |
| 7647 | } |
| 7648 | break; |
| 7649 | case 0xEF: |
| 7650 | if (slen > 2 && s[1] == 0xBB && s[2] == 0xBF) { |
| 7651 | DEBUG_p(PerlIO_printf(Perl_debug_log, "UTF-8 script encoding\n")); |
| 7652 | s += 3; /* UTF-8 */ |
| 7653 | } |
| 7654 | break; |
| 7655 | case 0: |
| 7656 | if (slen > 3 && s[1] == 0 && /* UTF-32 big-endian */ |
| 7657 | s[2] == 0xFE && s[3] == 0xFF) |
| 7658 | { |
| 7659 | Perl_croak(aTHX_ "Unsupported script encoding"); |
| 7660 | } |
| 7661 | } |
| 7662 | return (char*)s; |
| 7663 | } |
| 7664 | |
| 7665 | /* |
| 7666 | * restore_rsfp |
| 7667 | * Restore a source filter. |
| 7668 | */ |
| 7669 | |
| 7670 | static void |
| 7671 | restore_rsfp(pTHX_ void *f) |
| 7672 | { |
| 7673 | PerlIO *fp = (PerlIO*)f; |
| 7674 | |
| 7675 | if (PL_rsfp == PerlIO_stdin()) |
| 7676 | PerlIO_clearerr(PL_rsfp); |
| 7677 | else if (PL_rsfp && (PL_rsfp != fp)) |
| 7678 | PerlIO_close(PL_rsfp); |
| 7679 | PL_rsfp = fp; |
| 7680 | } |
| 7681 | |
| 7682 | #ifndef PERL_NO_UTF16_FILTER |
| 7683 | static I32 |
| 7684 | utf16_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
| 7685 | { |
| 7686 | I32 count = FILTER_READ(idx+1, sv, maxlen); |
| 7687 | if (count) { |
| 7688 | U8* tmps; |
| 7689 | U8* tend; |
| 7690 | I32 newlen; |
| 7691 | New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8); |
| 7692 | if (!*SvPV_nolen(sv)) |
| 7693 | /* Game over, but don't feed an odd-length string to utf16_to_utf8 */ |
| 7694 | return count; |
| 7695 | |
| 7696 | tend = utf16_to_utf8((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen); |
| 7697 | sv_usepvn(sv, (char*)tmps, tend - tmps); |
| 7698 | } |
| 7699 | return count; |
| 7700 | } |
| 7701 | |
| 7702 | static I32 |
| 7703 | utf16rev_textfilter(pTHX_ int idx, SV *sv, int maxlen) |
| 7704 | { |
| 7705 | I32 count = FILTER_READ(idx+1, sv, maxlen); |
| 7706 | if (count) { |
| 7707 | U8* tmps; |
| 7708 | U8* tend; |
| 7709 | I32 newlen; |
| 7710 | if (!*SvPV_nolen(sv)) |
| 7711 | /* Game over, but don't feed an odd-length string to utf16_to_utf8 */ |
| 7712 | return count; |
| 7713 | |
| 7714 | New(898, tmps, SvCUR(sv) * 3 / 2 + 1, U8); |
| 7715 | tend = utf16_to_utf8_reversed((U8*)SvPVX(sv), tmps, SvCUR(sv), &newlen); |
| 7716 | sv_usepvn(sv, (char*)tmps, tend - tmps); |
| 7717 | } |
| 7718 | return count; |
| 7719 | } |
| 7720 | #endif |