00001
00002
00003
00004
00005
00006
00007
00008
00009
00010
00011
00012
00013
00014
00015
00016
00017
00018
00019
00020
00021
00022
00023
00024
00025
00026
00027
00028
00029
00030
00031
00032
00033
00034 #include <lestes/common.hh>
00035 #include <lestes/lang/cplus/lex/preprocessor.hh>
00036 #include <lestes/lang/cplus/lex/preprocessor.m.hh>
00037 #include <lestes/lang/cplus/lex/cpp_token.hh>
00038 #include <lestes/lang/cplus/lex/evaluator.hh>
00039 #include <lestes/lang/cplus/lex/string_joiner.hh>
00040 #include <lestes/lang/cplus/lex/space_remover.hh>
00041 #include <lestes/lang/cplus/lex/lex_literal.g.hh>
00042 #include <lestes/lang/cplus/lex/file_system.hh>
00043 #include <lestes/lang/cplus/lex/string_translator.hh>
00044 #include <lestes/lang/cplus/lex/token_value.hh>
00045 #include <lestes/lang/cplus/lex/lex_loggers.hh>
00046 #include <lestes/msg/logger.hh>
00047 #include <lestes/msg/logger_util.hh>
00048 #include <lestes/std/source_location.hh>
00049
00050 package(lestes);
00051 package(lang);
00052 package(cplus);
00053 package(lex);
00054
00055
00056
00057
00058
00059
00060
00061 preprocessor::preprocessor(const ptr<file_system> &a_fs, const lstring &file_name):
00062 order(1),
00063 fs(checked(a_fs)),
00064 eva(evaluator::create(a_fs,file_name)),
00065 spr(space_remover::create(eva)),
00066 str(string_translator::create(spr)),
00067 stj(string_joiner::create(str))
00068 {
00069 }
00070
00071
00072
00073
00074
00075 bool preprocessor::pragma_flag_get(void) const
00076 {
00077 return eva->pragma_flag_get();
00078 }
00079
00080
00081
00082
00083
00084
00085
00086 ptr<cpp_token> preprocessor::read(void)
00087 {
00088 preprocessor_logger << "preprocessor::read()\n" << msg::eolog;
00089
00090 ptr<pp_token> t;
00091 ptr<source_location> loc;
00092 ptr<token_value> val;
00093 bool wide;
00094 ptr<lex_literal> lit;
00095 ptr<cpp_token> ct;
00096
00097 while (!ct) {
00098 t = stj->read();
00099 loc = t->location_get();
00100
00101 loc = loc->clone_order(order++);
00102 val = t->value_get();
00103 wide = false;
00104
00105 switch (t->type_get()) {
00106 case pp_token::TOK_FILE_END:
00107 ct = cpp_token::create(loc,cpp_token::TOK_EOF);
00108 break;
00109 case pp_token::TOK_WSTRING_LIT:
00110 wide = true;
00111
00112 case pp_token::TOK_STRING_LIT:
00113 ct = cpp_token::create_literal(loc,lex_string_literal::create(wide),val);
00114 break;
00115 case pp_token::TOK_WCHAR_LIT:
00116 wide = true;
00117
00118 case pp_token::TOK_CHAR_LIT:
00119 ct = cpp_token::create_literal(loc,lex_character_literal::create(wide,val->content_get().length() > 1),val);
00120 break;
00121 case pp_token::TOK_NUMBER_LIT:
00122 preprocessor_logger << "number literal\n" << msg::eolog;
00123 lit = classify_number(loc,val);
00124
00125 if (!lit) {
00126 preprocessor_logger << "the literal is illformed\n" << msg::eolog;
00127
00128 lit = lex_integral_literal::create(lex_integral_literal::OCTAL,lex_integral_literal::NONE,0,1);
00129 val = token_value::create("0");
00130 }
00131
00132 ct = cpp_token::create_literal(loc,lit,val);
00133
00134 break;
00135 case pp_token::TOK_BOOL_LIT:
00136 ct = cpp_token::create_literal(loc,lex_boolean_literal::create(),val);
00137 break;
00138
00139 case pp_token::TOK_LEFT_BRACKET:
00140 ct = cpp_token::create(loc,cpp_token::TOK_LEFT_BRACKET);
00141 break;
00142 case pp_token::TOK_RIGHT_BRACKET:
00143 ct = cpp_token::create(loc,cpp_token::TOK_RIGHT_BRACKET);
00144 break;
00145 case pp_token::TOK_LEFT_BRACE:
00146 ct = cpp_token::create(loc,cpp_token::TOK_LEFT_BRACE);
00147 break;
00148 case pp_token::TOK_RIGHT_BRACE:
00149 ct = cpp_token::create(loc,cpp_token::TOK_RIGHT_BRACE);
00150 break;
00151 case pp_token::TOK_LEFT_PAR:
00152 ct = cpp_token::create(loc,cpp_token::TOK_LEFT_PAR);
00153 break;
00154 case pp_token::TOK_RIGHT_PAR:
00155 ct = cpp_token::create(loc,cpp_token::TOK_RIGHT_PAR);
00156 break;
00157 case pp_token::TOK_SEMICOLON:
00158 ct = cpp_token::create(loc,cpp_token::TOK_SEMICOLON);
00159 break;
00160 case pp_token::TOK_COLON:
00161 ct = cpp_token::create(loc,cpp_token::TOK_COLON);
00162 break;
00163 case pp_token::TOK_QMARK:
00164 ct = cpp_token::create(loc,cpp_token::TOK_QMARK);
00165 break;
00166 case pp_token::TOK_DOT:
00167 ct = cpp_token::create(loc,cpp_token::TOK_DOT);
00168 break;
00169 case pp_token::TOK_PLUS:
00170 ct = cpp_token::create(loc,cpp_token::TOK_PLUS);
00171 break;
00172 case pp_token::TOK_STAR:
00173 ct = cpp_token::create(loc,cpp_token::TOK_STAR);
00174 break;
00175 case pp_token::TOK_PERCENT:
00176 ct = cpp_token::create(loc,cpp_token::TOK_PERCENT);
00177 break;
00178 case pp_token::TOK_SLASH:
00179 ct = cpp_token::create(loc,cpp_token::TOK_SLASH);
00180 break;
00181 case pp_token::TOK_HAT:
00182 ct = cpp_token::create(loc,cpp_token::TOK_HAT);
00183 break;
00184 case pp_token::TOK_AMP:
00185 ct = cpp_token::create(loc,cpp_token::TOK_AMP);
00186 break;
00187 case pp_token::TOK_VBAR:
00188 ct = cpp_token::create(loc,cpp_token::TOK_VBAR);
00189 break;
00190 case pp_token::TOK_TILDE:
00191 ct = cpp_token::create(loc,cpp_token::TOK_TILDE);
00192 break;
00193 case pp_token::TOK_EMARK:
00194 ct = cpp_token::create(loc,cpp_token::TOK_EXCLAMATION);
00195 break;
00196 case pp_token::TOK_EQ:
00197 ct = cpp_token::create(loc,cpp_token::TOK_EQ);
00198 break;
00199 case pp_token::TOK_LT:
00200 ct = cpp_token::create(loc,cpp_token::TOK_LT);
00201 break;
00202 case pp_token::TOK_GT:
00203 ct = cpp_token::create(loc,cpp_token::TOK_GT);
00204 break;
00205 case pp_token::TOK_COMMA:
00206 ct = cpp_token::create(loc,cpp_token::TOK_COMMA);
00207 break;
00208 case pp_token::TOK_MINUS:
00209 ct = cpp_token::create(loc,cpp_token::TOK_MINUS);
00210 break;
00211 case pp_token::TOK_DOT_DOT_DOT:
00212 ct = cpp_token::create(loc,cpp_token::TOK_DOT_DOT_DOT);
00213 break;
00214 case pp_token::TOK_COLON_COLON:
00215 ct = cpp_token::create(loc,cpp_token::TOK_COLON_COLON);
00216 break;
00217 case pp_token::TOK_DOT_STAR:
00218 ct = cpp_token::create(loc,cpp_token::TOK_DOT_STAR);
00219 break;
00220 case pp_token::TOK_PLUS_EQ:
00221 ct = cpp_token::create(loc,cpp_token::TOK_PLUS_EQ);
00222 break;
00223 case pp_token::TOK_MINUS_EQ:
00224 ct = cpp_token::create(loc,cpp_token::TOK_MINUS_EQ);
00225 break;
00226 case pp_token::TOK_STAR_EQ:
00227 ct = cpp_token::create(loc,cpp_token::TOK_STAR_EQ);
00228 break;
00229 case pp_token::TOK_SLASH_EQ:
00230 ct = cpp_token::create(loc,cpp_token::TOK_SLASH_EQ);
00231 break;
00232 case pp_token::TOK_PERCENT_EQ:
00233 ct = cpp_token::create(loc,cpp_token::TOK_PERCENT_EQ);
00234 break;
00235 case pp_token::TOK_HAT_EQ:
00236 ct = cpp_token::create(loc,cpp_token::TOK_HAT_EQ);
00237 break;
00238 case pp_token::TOK_AMP_EQ:
00239 ct = cpp_token::create(loc,cpp_token::TOK_AMP_EQ);
00240 break;
00241 case pp_token::TOK_VBAR_EQ:
00242 ct = cpp_token::create(loc,cpp_token::TOK_VBAR_EQ);
00243 break;
00244 case pp_token::TOK_LT_LT:
00245 ct = cpp_token::create(loc,cpp_token::TOK_LT_LT);
00246 break;
00247 case pp_token::TOK_GT_GT:
00248 ct = cpp_token::create(loc,cpp_token::TOK_GT_GT);
00249 break;
00250 case pp_token::TOK_LT_LT_EQ:
00251 ct = cpp_token::create(loc,cpp_token::TOK_LT_LT_EQ);
00252 break;
00253 case pp_token::TOK_GT_GT_EQ:
00254 ct = cpp_token::create(loc,cpp_token::TOK_GT_GT_EQ);
00255 break;
00256 case pp_token::TOK_EQ_EQ:
00257 ct = cpp_token::create(loc,cpp_token::TOK_EQ_EQ);
00258 break;
00259 case pp_token::TOK_EMARK_EQ:
00260 ct = cpp_token::create(loc,cpp_token::TOK_EXCLAMATION_EQ);
00261 break;
00262 case pp_token::TOK_LT_EQ:
00263 ct = cpp_token::create(loc,cpp_token::TOK_LT_EQ);
00264 break;
00265 case pp_token::TOK_GT_EQ:
00266 ct = cpp_token::create(loc,cpp_token::TOK_GT_EQ);
00267 break;
00268 case pp_token::TOK_AMP_AMP:
00269 ct = cpp_token::create(loc,cpp_token::TOK_AMP_AMP);
00270 break;
00271 case pp_token::TOK_VBAR_VBAR:
00272 ct = cpp_token::create(loc,cpp_token::TOK_VBAR_VBAR);
00273 break;
00274 case pp_token::TOK_PLUS_PLUS:
00275 ct = cpp_token::create(loc,cpp_token::TOK_PLUS_PLUS);
00276 break;
00277 case pp_token::TOK_MINUS_MINUS:
00278 ct = cpp_token::create(loc,cpp_token::TOK_MINUS_MINUS);
00279 break;
00280 case pp_token::TOK_MINUS_GT_STAR:
00281 ct = cpp_token::create(loc,cpp_token::TOK_MINUS_GT_STAR);
00282 break;
00283 case pp_token::TOK_MINUS_GT:
00284 ct = cpp_token::create(loc,cpp_token::TOK_MINUS_GT);
00285 break;
00286 case pp_token::TOK_ASM:
00287 ct = cpp_token::create(loc,cpp_token::TOK_ASM);
00288 break;
00289 case pp_token::TOK_AUTO:
00290 ct = cpp_token::create(loc,cpp_token::TOK_AUTO);
00291 break;
00292 case pp_token::TOK_BOOL:
00293 ct = cpp_token::create(loc,cpp_token::TOK_BOOL);
00294 break;
00295 case pp_token::TOK_BREAK:
00296 ct = cpp_token::create(loc,cpp_token::TOK_BREAK);
00297 break;
00298 case pp_token::TOK_CASE:
00299 ct = cpp_token::create(loc,cpp_token::TOK_CASE);
00300 break;
00301 case pp_token::TOK_CATCH:
00302 ct = cpp_token::create(loc,cpp_token::TOK_CATCH);
00303 break;
00304 case pp_token::TOK_CHAR:
00305 ct = cpp_token::create(loc,cpp_token::TOK_CHAR);
00306 break;
00307 case pp_token::TOK_CLASS:
00308 ct = cpp_token::create(loc,cpp_token::TOK_CLASS);
00309 break;
00310 case pp_token::TOK_CONST:
00311 ct = cpp_token::create(loc,cpp_token::TOK_CONST);
00312 break;
00313 case pp_token::TOK_CONST_CAST:
00314 ct = cpp_token::create(loc,cpp_token::TOK_CONST_CAST);
00315 break;
00316 case pp_token::TOK_CONTINUE:
00317 ct = cpp_token::create(loc,cpp_token::TOK_CONTINUE);
00318 break;
00319 case pp_token::TOK_DEFAULT:
00320 ct = cpp_token::create(loc,cpp_token::TOK_DEFAULT);
00321 break;
00322 case pp_token::TOK_DELETE:
00323 ct = cpp_token::create(loc,cpp_token::TOK_DELETE);
00324 break;
00325 case pp_token::TOK_DO:
00326 ct = cpp_token::create(loc,cpp_token::TOK_DO);
00327 break;
00328 case pp_token::TOK_DOUBLE:
00329 ct = cpp_token::create(loc,cpp_token::TOK_DOUBLE);
00330 break;
00331 case pp_token::TOK_DYNAMIC_CAST:
00332 ct = cpp_token::create(loc,cpp_token::TOK_DYNAMIC_CAST);
00333 break;
00334 case pp_token::TOK_ELSE:
00335 ct = cpp_token::create(loc,cpp_token::TOK_ELSE);
00336 break;
00337 case pp_token::TOK_ENUM:
00338 ct = cpp_token::create(loc,cpp_token::TOK_ENUM);
00339 break;
00340 case pp_token::TOK_EXPLICIT:
00341 ct = cpp_token::create(loc,cpp_token::TOK_EXPLICIT);
00342 break;
00343 case pp_token::TOK_EXPORT:
00344 ct = cpp_token::create(loc,cpp_token::TOK_EXPORT);
00345 break;
00346 case pp_token::TOK_EXTERN:
00347 ct = cpp_token::create(loc,cpp_token::TOK_EXTERN);
00348 break;
00349 case pp_token::TOK_FLOAT:
00350 ct = cpp_token::create(loc,cpp_token::TOK_FLOAT);
00351 break;
00352 case pp_token::TOK_FOR:
00353 ct = cpp_token::create(loc,cpp_token::TOK_FOR);
00354 break;
00355 case pp_token::TOK_FRIEND:
00356 ct = cpp_token::create(loc,cpp_token::TOK_FRIEND);
00357 break;
00358 case pp_token::TOK_GOTO:
00359 ct = cpp_token::create(loc,cpp_token::TOK_GOTO);
00360 break;
00361 case pp_token::TOK_IF:
00362 ct = cpp_token::create(loc,cpp_token::TOK_IF);
00363 break;
00364 case pp_token::TOK_INLINE:
00365 ct = cpp_token::create(loc,cpp_token::TOK_INLINE);
00366 break;
00367 case pp_token::TOK_INT:
00368 ct = cpp_token::create(loc,cpp_token::TOK_INT);
00369 break;
00370 case pp_token::TOK_LONG:
00371 ct = cpp_token::create(loc,cpp_token::TOK_LONG);
00372 break;
00373 case pp_token::TOK_MUTABLE:
00374 ct = cpp_token::create(loc,cpp_token::TOK_MUTABLE);
00375 break;
00376 case pp_token::TOK_NAMESPACE:
00377 ct = cpp_token::create(loc,cpp_token::TOK_NAMESPACE);
00378 break;
00379 case pp_token::TOK_NEW:
00380 ct = cpp_token::create(loc,cpp_token::TOK_NEW);
00381 break;
00382 case pp_token::TOK_OPERATOR:
00383 ct = cpp_token::create(loc,cpp_token::TOK_OPERATOR);
00384 break;
00385 case pp_token::TOK_PRIVATE:
00386 ct = cpp_token::create(loc,cpp_token::TOK_PRIVATE);
00387 break;
00388 case pp_token::TOK_PROTECTED:
00389 ct = cpp_token::create(loc,cpp_token::TOK_PROTECTED);
00390 break;
00391 case pp_token::TOK_PUBLIC:
00392 ct = cpp_token::create(loc,cpp_token::TOK_PUBLIC);
00393 break;
00394 case pp_token::TOK_REGISTER:
00395 ct = cpp_token::create(loc,cpp_token::TOK_REGISTER);
00396 break;
00397 case pp_token::TOK_REINTERPRET_CAST:
00398 ct = cpp_token::create(loc,cpp_token::TOK_REINTERPRET_CAST);
00399 break;
00400 case pp_token::TOK_RETURN:
00401 ct = cpp_token::create(loc,cpp_token::TOK_RETURN);
00402 break;
00403 case pp_token::TOK_SHORT:
00404 ct = cpp_token::create(loc,cpp_token::TOK_SHORT);
00405 break;
00406 case pp_token::TOK_SIGNED:
00407 ct = cpp_token::create(loc,cpp_token::TOK_SIGNED);
00408 break;
00409 case pp_token::TOK_SIZEOF:
00410 ct = cpp_token::create(loc,cpp_token::TOK_SIZEOF);
00411 break;
00412 case pp_token::TOK_STATIC:
00413 ct = cpp_token::create(loc,cpp_token::TOK_STATIC);
00414 break;
00415 case pp_token::TOK_STATIC_CAST:
00416 ct = cpp_token::create(loc,cpp_token::TOK_STATIC_CAST);
00417 break;
00418 case pp_token::TOK_STRUCT:
00419 ct = cpp_token::create(loc,cpp_token::TOK_STRUCT);
00420 break;
00421 case pp_token::TOK_SWITCH:
00422 ct = cpp_token::create(loc,cpp_token::TOK_SWITCH);
00423 break;
00424 case pp_token::TOK_TEMPLATE:
00425 ct = cpp_token::create(loc,cpp_token::TOK_TEMPLATE);
00426 break;
00427 case pp_token::TOK_THIS:
00428 ct = cpp_token::create(loc,cpp_token::TOK_THIS);
00429 break;
00430 case pp_token::TOK_THROW:
00431 ct = cpp_token::create(loc,cpp_token::TOK_THROW);
00432 break;
00433 case pp_token::TOK_TRY:
00434 ct = cpp_token::create(loc,cpp_token::TOK_TRY);
00435 break;
00436 case pp_token::TOK_TYPEDEF:
00437 ct = cpp_token::create(loc,cpp_token::TOK_TYPEDEF);
00438 break;
00439 case pp_token::TOK_TYPEID:
00440 ct = cpp_token::create(loc,cpp_token::TOK_TYPEID);
00441 break;
00442 case pp_token::TOK_TYPENAME:
00443 ct = cpp_token::create(loc,cpp_token::TOK_TYPENAME);
00444 break;
00445 case pp_token::TOK_UNION:
00446 ct = cpp_token::create(loc,cpp_token::TOK_UNION);
00447 break;
00448 case pp_token::TOK_UNSIGNED:
00449 ct = cpp_token::create(loc,cpp_token::TOK_UNSIGNED);
00450 break;
00451 case pp_token::TOK_USING:
00452 ct = cpp_token::create(loc,cpp_token::TOK_USING);
00453 break;
00454 case pp_token::TOK_VIRTUAL:
00455 ct = cpp_token::create(loc,cpp_token::TOK_VIRTUAL);
00456 break;
00457 case pp_token::TOK_VOID:
00458 ct = cpp_token::create(loc,cpp_token::TOK_VOID);
00459 break;
00460 case pp_token::TOK_VOLATILE:
00461 ct = cpp_token::create(loc,cpp_token::TOK_VOLATILE);
00462 break;
00463 case pp_token::TOK_WCHAR:
00464 ct = cpp_token::create(loc,cpp_token::TOK_WCHAR_T);
00465 break;
00466 case pp_token::TOK_WHILE:
00467 ct = cpp_token::create(loc,cpp_token::TOK_WHILE);
00468 break;
00469 case pp_token::TOK_IDENT:
00470 ct = cpp_token::create(loc,cpp_token::TOK_IDENT,val);
00471 break;
00472 case pp_token::TOK_HASH:
00473 case pp_token::TOK_HASH_HASH:
00474
00475 report << invalid_cpp_token << t->spelling_get() << loc;
00476 break;
00477 case pp_token::TOK_OTHER:
00478
00479 report << stray_character << t->spelling_get() << loc;
00480 break;
00481 default:
00482 lassert2(false,"Unexpected token in conversion");
00483 break;
00484 }
00485 }
00486
00487 preprocessor_logger << "returning " << static_cast<ulint>(ct->type_get()) << "\n" << msg::eolog;
00488 preprocessor_logger << "preprocessor::read() end\n" << msg::eolog;
00489 return ct;
00490 }
00491
00492
00493
00494
00495
00496
00497
00498
00499
00500
00501 ptr<lex_literal> preprocessor::classify_number(const ptr<source_location> &loc, const ptr<token_value> &num)
00502 {
00503 preprocessor_logger << "preprocessor::classify_number()\n" << msg::eolog;
00504
00505 lassert(loc);
00506 lassert(num);
00507
00508
00509 enum {
00510 BEGIN,
00511 ZERO,
00512 OCTAL,
00513 DECIMAL,
00514 HEXADECIMAL_START,
00515 HEXADECIMAL,
00516 FLOAT_INTEGRAL,
00517 FLOAT_FRACTION,
00518 FLOAT_EXPONENT_SIGN,
00519 FLOAT_EXPONENT_START,
00520 FLOAT_EXPONENT,
00521 FLOAT_SUFFIX,
00522 FLOAT_END,
00523 INTEGRAL_SUFFIX,
00524 } fstate = BEGIN;
00525
00526 ucn_string str(num->content_get());
00527 ucn u;
00528 lex_integral_literal::base_type base = lex_integral_literal::OCTAL;
00529 bool long_suffix = false;
00530 bool unsigned_suffix = false;
00531 bool float_suffix = false;
00532 ptr<lex_literal> literal;
00533
00534 for (ucn_string::iterator sit = str.begin(), send = str.end();
00535 sit != send; ) {
00536 u = *sit;
00537
00538 preprocessor_logger << "fstate = " << fstate["bzodhHI.+eEfFu"] << ' ' << (ulint)u << '\n' << msg::eolog;
00539
00540 if (!character::is_basic(u) || u == character::ascii_underscore) {
00541
00542 report << invalid_character_in_number << loc;
00543 goto err;
00544 }
00545 switch (fstate) {
00546 case BEGIN:
00547 ++sit;
00548 switch (u) {
00549 case character::ascii_digit_0:
00550 fstate = ZERO;
00551 break;
00552 case character::ascii_digit_1:
00553 case character::ascii_digit_2:
00554 case character::ascii_digit_3:
00555 case character::ascii_digit_4:
00556 case character::ascii_digit_5:
00557 case character::ascii_digit_6:
00558 case character::ascii_digit_7:
00559 case character::ascii_digit_8:
00560 case character::ascii_digit_9:
00561 fstate = DECIMAL;
00562 break;
00563 case character::ascii_dot:
00564 fstate = FLOAT_FRACTION;
00565 break;
00566 default:
00567 lassert2(false,"You should never get here");
00568 }
00569 break;
00570 case ZERO:
00571 if (u == character::ascii_lower_x ||
00572 u == character::ascii_upper_x) {
00573 fstate = HEXADECIMAL_START;
00574 ++sit;
00575 } else if (character::is_odigit(u)) {
00576 fstate = OCTAL;
00577 } else
00578 fstate = INTEGRAL_SUFFIX;
00579 break;
00580 case OCTAL:
00581 if (character::is_odigit(u))
00582 ++sit;
00583 else if (character::is_digit(u))
00584 fstate = FLOAT_INTEGRAL;
00585 else if (u == character::ascii_dot) {
00586 fstate = FLOAT_FRACTION;
00587 ++sit;
00588 } else
00589 fstate = INTEGRAL_SUFFIX;
00590 break;
00591 case DECIMAL:
00592 if (character::is_digit(u))
00593 ++sit;
00594 else if (u == character::ascii_dot) {
00595 ++sit;
00596 fstate = FLOAT_FRACTION;
00597 } else if (u == character::ascii_lower_e ||
00598 u == character::ascii_upper_e) {
00599 ++sit;
00600 fstate = FLOAT_EXPONENT_SIGN;
00601 } else
00602 fstate = INTEGRAL_SUFFIX;
00603 break;
00604 case HEXADECIMAL_START:
00605 if (character::is_xdigit(u)) {
00606 ++sit;
00607 fstate = HEXADECIMAL;
00608 base = lex_integral_literal::HEXADECIMAL;
00609 } else
00610 fstate = INTEGRAL_SUFFIX;
00611 break;
00612 case HEXADECIMAL:
00613 if (character::is_xdigit(u))
00614 ++sit;
00615 else
00616 fstate = INTEGRAL_SUFFIX;
00617 break;
00618 case FLOAT_INTEGRAL:
00619 if (character::is_digit(u))
00620 ++sit;
00621 else if (u == character::ascii_dot) {
00622 ++sit;
00623 fstate = FLOAT_FRACTION;
00624 } else if (u == character::ascii_lower_e ||
00625 u == character::ascii_upper_e) {
00626 ++sit;
00627 fstate = FLOAT_EXPONENT_SIGN;
00628 } else
00629 fstate = FLOAT_SUFFIX;
00630 break;
00631 case FLOAT_FRACTION:
00632 if (character::is_digit(u))
00633 ++sit;
00634 else if (u == character::ascii_lower_e ||
00635 u == character::ascii_upper_e) {
00636 ++sit;
00637 fstate = FLOAT_EXPONENT_SIGN;
00638 } else
00639 fstate = FLOAT_SUFFIX;
00640 break;
00641 case FLOAT_EXPONENT_SIGN:
00642 if (character::is_digit(u)) {
00643 ++sit;
00644 fstate = FLOAT_EXPONENT;
00645 } else if (u == character::ascii_plus) {
00646 ++sit;
00647 fstate = FLOAT_EXPONENT_START;
00648 } else if (u == character::ascii_minus) {
00649 ++sit;
00650 fstate = FLOAT_EXPONENT_START;
00651 } else {
00652
00653 report << floating_exponent_empty << loc;
00654 goto err;
00655 }
00656 break;
00657 case FLOAT_EXPONENT_START:
00658 if (character::is_digit(u)) {
00659 ++sit;
00660 fstate = FLOAT_EXPONENT;
00661 } else {
00662
00663 report << floating_exponent_empty << loc;
00664 goto err;
00665 }
00666 break;
00667 case FLOAT_EXPONENT:
00668 if (character::is_digit(u))
00669 ++sit;
00670 else {
00671 fstate = FLOAT_SUFFIX;
00672 }
00673 break;
00674 case FLOAT_SUFFIX:
00675 ++sit;
00676 if (u == character::ascii_lower_f ||
00677 u == character::ascii_upper_f) {
00678 float_suffix = true;
00679 fstate = FLOAT_END;
00680 break;
00681 } else if (u == character::ascii_lower_l ||
00682 u == character::ascii_upper_l) {
00683 long_suffix = true;
00684 fstate = FLOAT_END;
00685 break;
00686 }
00687
00688 case FLOAT_END:
00689
00690 report << floating_suffix_invalid << loc;
00691 goto err;
00692 case INTEGRAL_SUFFIX:
00693 ++sit;
00694 if (u == character::ascii_lower_u ||
00695 u == character::ascii_upper_u) {
00696 if (unsigned_suffix) {
00697
00698 report << integral_suffix_invalid << loc;
00699 goto err;
00700 }
00701 unsigned_suffix = true;
00702 } else if (u == character::ascii_lower_l ||
00703 u == character::ascii_upper_l) {
00704 if (long_suffix) {
00705
00706 report << integral_suffix_invalid << loc;
00707 goto err;
00708 }
00709 long_suffix = true;
00710 } else {
00711
00712 report << integral_suffix_invalid << loc;
00713 goto err;
00714 }
00715 break;
00716 default:
00717 lassert2(false,"You should never get here");
00718 break;
00719 }
00720 }
00721
00722 switch (fstate) {
00723 case ZERO:
00724 case OCTAL:
00725 case DECIMAL:
00726 case HEXADECIMAL:
00727 case INTEGRAL_SUFFIX:
00728 {
00729 ucn_string::size_type start = (base == lex_integral_literal::HEXADECIMAL) ? 2 : 0;
00730 ucn_string::size_type end = str.length();
00731 lex_integral_literal::suffix_type suffix = lex_integral_literal::NONE;
00732
00733 if (unsigned_suffix) {
00734 if (long_suffix) {
00735 suffix = lex_integral_literal::UNSIGNED_LONG;
00736 end -= 2;
00737 } else {
00738 suffix = lex_integral_literal::UNSIGNED;
00739 end--;
00740 }
00741 } else if (long_suffix) {
00742 suffix = lex_integral_literal::LONG;
00743 end--;
00744 }
00745
00746 preprocessor_logger << "integral su = " << (ulint)suffix <<
00747 " s = " << start << " e = " << end << "\n" << msg::eolog;
00748
00749 literal = lex_integral_literal::create(base,suffix,start,end);
00750 break;
00751 }
00752 case FLOAT_FRACTION:
00753 case FLOAT_EXPONENT:
00754 case FLOAT_SUFFIX:
00755 case FLOAT_END:
00756 {
00757 lex_floating_literal::suffix_type suffix = lex_floating_literal::NONE;
00758 if (long_suffix)
00759 suffix = lex_floating_literal::LONG;
00760 else if (float_suffix)
00761 suffix = lex_floating_literal::FLOAT;
00762
00763 ucn_string::iterator sit = str.begin(), send = str.end();
00764
00765 ucn_string::size_type is = 0, ie = is;
00766
00767 while (sit != send && character::is_digit(*sit)) {
00768 ++ie;
00769 ++sit;
00770 }
00771
00772 ucn_string::size_type fs = ie, fe = fs;
00773 if (sit != send && *sit == character::ascii_dot) {
00774 ++sit;
00775 ++fs;
00776 ++fe;
00777
00778 while (sit != send && character::is_digit(*sit)) {
00779 ++fe;
00780 ++sit;
00781 }
00782 }
00783
00784 bool neg = false;
00785 ucn_string::size_type es = fe, ee = es;
00786 if (sit != send && (*sit == character::ascii_lower_e || *sit == character::ascii_upper_e)) {
00787 ++sit;
00788 ++es;
00789 ++ee;
00790
00791 if (sit != send && !character::is_digit(*sit)) {
00792 neg = *sit == character::ascii_minus;
00793 ++sit;
00794 ++es;
00795 ++ee;
00796 }
00797
00798 while (sit != send && character::is_digit(*sit)) {
00799 ++ee;
00800 ++sit;
00801 }
00802 }
00803
00804 preprocessor_logger << "floating su = " << (ulint)suffix << " n = " << neg <<
00805 " is = " << is << " ie = " << ie << " fs = " << fs << " fe = " << fe <<
00806 " es = " << es << " ee = " << ee << "\n" << msg::eolog;
00807
00808 literal = lex_floating_literal::create(suffix,neg,is,ie,fs,fe,es,ee);
00809 }
00810 break;
00811 case FLOAT_EXPONENT_SIGN:
00812 case FLOAT_EXPONENT_START:
00813
00814 report << floating_exponent_empty << loc;
00815 break;
00816 case HEXADECIMAL_START:
00817
00818 report << integral_empty << loc;
00819 break;
00820 default:
00821 lassert2(false,"You should never get here");
00822 }
00823
00824 err:
00825 preprocessor_logger << "preprocessor::classify_number() end\n" << msg::eolog;
00826
00827 return literal;
00828 }
00829
00830
00831
00832
00833 void preprocessor::gc_mark(void)
00834 {
00835 fs.gc_mark();
00836 eva.gc_mark();
00837 spr.gc_mark();
00838 str.gc_mark();
00839 stj.gc_mark();
00840 ::lestes::std::object::gc_mark();
00841 }
00842
00843
00844
00845
00846
00847
00848 ptr<preprocessor> preprocessor::create(const ptr<file_system> &a_fs, const lstring &file_name)
00849 {
00850 return new preprocessor(a_fs,file_name);
00851 }
00852
00853 end_package(lex);
00854 end_package(cplus);
00855 end_package(lang);
00856 end_package(lestes);
00857
00858
