| 1 | /* YACC parser for D expressions, for GDB. |
| 2 | |
| 3 | Copyright (C) 2014-2018 Free Software Foundation, Inc. |
| 4 | |
| 5 | This file is part of GDB. |
| 6 | |
| 7 | This program is free software; you can redistribute it and/or modify |
| 8 | it under the terms of the GNU General Public License as published by |
| 9 | the Free Software Foundation; either version 3 of the License, or |
| 10 | (at your option) any later version. |
| 11 | |
| 12 | This program is distributed in the hope that it will be useful, |
| 13 | but WITHOUT ANY WARRANTY; without even the implied warranty of |
| 14 | MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE. See the |
| 15 | GNU General Public License for more details. |
| 16 | |
| 17 | You should have received a copy of the GNU General Public License |
| 18 | along with this program. If not, see <http://www.gnu.org/licenses/>. */ |
| 19 | |
| 20 | /* This file is derived from c-exp.y, jv-exp.y. */ |
| 21 | |
| 22 | /* Parse a D expression from text in a string, |
| 23 | and return the result as a struct expression pointer. |
| 24 | That structure contains arithmetic operations in reverse polish, |
| 25 | with constants represented by operations that are followed by special data. |
| 26 | See expression.h for the details of the format. |
| 27 | What is important here is that it can be built up sequentially |
| 28 | during the process of parsing; the lower levels of the tree always |
| 29 | come first in the result. |
| 30 | |
| 31 | Note that malloc's and realloc's in this file are transformed to |
| 32 | xmalloc and xrealloc respectively by the same sed command in the |
| 33 | makefile that remaps any other malloc/realloc inserted by the parser |
| 34 | generator. Doing this with #defines and trying to control the interaction |
| 35 | with include files (<malloc.h> and <stdlib.h> for example) just became |
| 36 | too messy, particularly when such includes can be inserted at random |
| 37 | times by the parser generator. */ |
| 38 | |
| 39 | %{ |
| 40 | |
| 41 | #include "defs.h" |
| 42 | #include <ctype.h> |
| 43 | #include "expression.h" |
| 44 | #include "value.h" |
| 45 | #include "parser-defs.h" |
| 46 | #include "language.h" |
| 47 | #include "c-lang.h" |
| 48 | #include "d-lang.h" |
| 49 | #include "bfd.h" /* Required by objfiles.h. */ |
| 50 | #include "symfile.h" /* Required by objfiles.h. */ |
| 51 | #include "objfiles.h" /* For have_full_symbols and have_partial_symbols */ |
| 52 | #include "charset.h" |
| 53 | #include "block.h" |
| 54 | |
| 55 | #define parse_type(ps) builtin_type (parse_gdbarch (ps)) |
| 56 | #define parse_d_type(ps) builtin_d_type (parse_gdbarch (ps)) |
| 57 | |
| 58 | /* Remap normal yacc parser interface names (yyparse, yylex, yyerror, |
| 59 | etc). */ |
| 60 | #define GDB_YY_REMAP_PREFIX d_ |
| 61 | #include "yy-remap.h" |
| 62 | |
| 63 | /* The state of the parser, used internally when we are parsing the |
| 64 | expression. */ |
| 65 | |
| 66 | static struct parser_state *pstate = NULL; |
| 67 | |
| 68 | int yyparse (void); |
| 69 | |
| 70 | static int yylex (void); |
| 71 | |
| 72 | static void yyerror (const char *); |
| 73 | |
| 74 | static int type_aggregate_p (struct type *); |
| 75 | |
| 76 | %} |
| 77 | |
| 78 | /* Although the yacc "value" of an expression is not used, |
| 79 | since the result is stored in the structure being created, |
| 80 | other node types do have values. */ |
| 81 | |
| 82 | %union |
| 83 | { |
| 84 | struct { |
| 85 | LONGEST val; |
| 86 | struct type *type; |
| 87 | } typed_val_int; |
| 88 | struct { |
| 89 | gdb_byte val[16]; |
| 90 | struct type *type; |
| 91 | } typed_val_float; |
| 92 | struct symbol *sym; |
| 93 | struct type *tval; |
| 94 | struct typed_stoken tsval; |
| 95 | struct stoken sval; |
| 96 | struct ttype tsym; |
| 97 | struct symtoken ssym; |
| 98 | int ival; |
| 99 | int voidval; |
| 100 | struct block *bval; |
| 101 | enum exp_opcode opcode; |
| 102 | struct stoken_vector svec; |
| 103 | } |
| 104 | |
| 105 | %{ |
| 106 | /* YYSTYPE gets defined by %union */ |
| 107 | static int parse_number (struct parser_state *, const char *, |
| 108 | int, int, YYSTYPE *); |
| 109 | %} |
| 110 | |
| 111 | %token <sval> IDENTIFIER UNKNOWN_NAME |
| 112 | %token <tsym> TYPENAME |
| 113 | %token <voidval> COMPLETE |
| 114 | |
| 115 | /* A NAME_OR_INT is a symbol which is not known in the symbol table, |
| 116 | but which would parse as a valid number in the current input radix. |
| 117 | E.g. "c" when input_radix==16. Depending on the parse, it will be |
| 118 | turned into a name or into a number. */ |
| 119 | |
| 120 | %token <sval> NAME_OR_INT |
| 121 | |
| 122 | %token <typed_val_int> INTEGER_LITERAL |
| 123 | %token <typed_val_float> FLOAT_LITERAL |
| 124 | %token <tsval> CHARACTER_LITERAL |
| 125 | %token <tsval> STRING_LITERAL |
| 126 | |
| 127 | %type <svec> StringExp |
| 128 | %type <tval> BasicType TypeExp |
| 129 | %type <sval> IdentifierExp |
| 130 | %type <ival> ArrayLiteral |
| 131 | |
| 132 | %token ENTRY |
| 133 | %token ERROR |
| 134 | |
| 135 | /* Keywords that have a constant value. */ |
| 136 | %token TRUE_KEYWORD FALSE_KEYWORD NULL_KEYWORD |
| 137 | /* Class 'super' accessor. */ |
| 138 | %token SUPER_KEYWORD |
| 139 | /* Properties. */ |
| 140 | %token CAST_KEYWORD SIZEOF_KEYWORD |
| 141 | %token TYPEOF_KEYWORD TYPEID_KEYWORD |
| 142 | %token INIT_KEYWORD |
| 143 | /* Comparison keywords. */ |
| 144 | /* Type storage classes. */ |
| 145 | %token IMMUTABLE_KEYWORD CONST_KEYWORD SHARED_KEYWORD |
| 146 | /* Non-scalar type keywords. */ |
| 147 | %token STRUCT_KEYWORD UNION_KEYWORD |
| 148 | %token CLASS_KEYWORD INTERFACE_KEYWORD |
| 149 | %token ENUM_KEYWORD TEMPLATE_KEYWORD |
| 150 | %token DELEGATE_KEYWORD FUNCTION_KEYWORD |
| 151 | |
| 152 | %token <sval> DOLLAR_VARIABLE |
| 153 | |
| 154 | %token <opcode> ASSIGN_MODIFY |
| 155 | |
| 156 | %left ',' |
| 157 | %right '=' ASSIGN_MODIFY |
| 158 | %right '?' |
| 159 | %left OROR |
| 160 | %left ANDAND |
| 161 | %left '|' |
| 162 | %left '^' |
| 163 | %left '&' |
| 164 | %left EQUAL NOTEQUAL '<' '>' LEQ GEQ |
| 165 | %right LSH RSH |
| 166 | %left '+' '-' |
| 167 | %left '*' '/' '%' |
| 168 | %right HATHAT |
| 169 | %left IDENTITY NOTIDENTITY |
| 170 | %right INCREMENT DECREMENT |
| 171 | %right '.' '[' '(' |
| 172 | %token DOTDOT |
| 173 | |
| 174 | \f |
| 175 | %% |
| 176 | |
| 177 | start : |
| 178 | Expression |
| 179 | | TypeExp |
| 180 | ; |
| 181 | |
| 182 | /* Expressions, including the comma operator. */ |
| 183 | |
| 184 | Expression: |
| 185 | CommaExpression |
| 186 | ; |
| 187 | |
| 188 | CommaExpression: |
| 189 | AssignExpression |
| 190 | | AssignExpression ',' CommaExpression |
| 191 | { write_exp_elt_opcode (pstate, BINOP_COMMA); } |
| 192 | ; |
| 193 | |
| 194 | AssignExpression: |
| 195 | ConditionalExpression |
| 196 | | ConditionalExpression '=' AssignExpression |
| 197 | { write_exp_elt_opcode (pstate, BINOP_ASSIGN); } |
| 198 | | ConditionalExpression ASSIGN_MODIFY AssignExpression |
| 199 | { write_exp_elt_opcode (pstate, BINOP_ASSIGN_MODIFY); |
| 200 | write_exp_elt_opcode (pstate, $2); |
| 201 | write_exp_elt_opcode (pstate, BINOP_ASSIGN_MODIFY); } |
| 202 | ; |
| 203 | |
| 204 | ConditionalExpression: |
| 205 | OrOrExpression |
| 206 | | OrOrExpression '?' Expression ':' ConditionalExpression |
| 207 | { write_exp_elt_opcode (pstate, TERNOP_COND); } |
| 208 | ; |
| 209 | |
| 210 | OrOrExpression: |
| 211 | AndAndExpression |
| 212 | | OrOrExpression OROR AndAndExpression |
| 213 | { write_exp_elt_opcode (pstate, BINOP_LOGICAL_OR); } |
| 214 | ; |
| 215 | |
| 216 | AndAndExpression: |
| 217 | OrExpression |
| 218 | | AndAndExpression ANDAND OrExpression |
| 219 | { write_exp_elt_opcode (pstate, BINOP_LOGICAL_AND); } |
| 220 | ; |
| 221 | |
| 222 | OrExpression: |
| 223 | XorExpression |
| 224 | | OrExpression '|' XorExpression |
| 225 | { write_exp_elt_opcode (pstate, BINOP_BITWISE_IOR); } |
| 226 | ; |
| 227 | |
| 228 | XorExpression: |
| 229 | AndExpression |
| 230 | | XorExpression '^' AndExpression |
| 231 | { write_exp_elt_opcode (pstate, BINOP_BITWISE_XOR); } |
| 232 | ; |
| 233 | |
| 234 | AndExpression: |
| 235 | CmpExpression |
| 236 | | AndExpression '&' CmpExpression |
| 237 | { write_exp_elt_opcode (pstate, BINOP_BITWISE_AND); } |
| 238 | ; |
| 239 | |
| 240 | CmpExpression: |
| 241 | ShiftExpression |
| 242 | | EqualExpression |
| 243 | | IdentityExpression |
| 244 | | RelExpression |
| 245 | ; |
| 246 | |
| 247 | EqualExpression: |
| 248 | ShiftExpression EQUAL ShiftExpression |
| 249 | { write_exp_elt_opcode (pstate, BINOP_EQUAL); } |
| 250 | | ShiftExpression NOTEQUAL ShiftExpression |
| 251 | { write_exp_elt_opcode (pstate, BINOP_NOTEQUAL); } |
| 252 | ; |
| 253 | |
| 254 | IdentityExpression: |
| 255 | ShiftExpression IDENTITY ShiftExpression |
| 256 | { write_exp_elt_opcode (pstate, BINOP_EQUAL); } |
| 257 | | ShiftExpression NOTIDENTITY ShiftExpression |
| 258 | { write_exp_elt_opcode (pstate, BINOP_NOTEQUAL); } |
| 259 | ; |
| 260 | |
| 261 | RelExpression: |
| 262 | ShiftExpression '<' ShiftExpression |
| 263 | { write_exp_elt_opcode (pstate, BINOP_LESS); } |
| 264 | | ShiftExpression LEQ ShiftExpression |
| 265 | { write_exp_elt_opcode (pstate, BINOP_LEQ); } |
| 266 | | ShiftExpression '>' ShiftExpression |
| 267 | { write_exp_elt_opcode (pstate, BINOP_GTR); } |
| 268 | | ShiftExpression GEQ ShiftExpression |
| 269 | { write_exp_elt_opcode (pstate, BINOP_GEQ); } |
| 270 | ; |
| 271 | |
| 272 | ShiftExpression: |
| 273 | AddExpression |
| 274 | | ShiftExpression LSH AddExpression |
| 275 | { write_exp_elt_opcode (pstate, BINOP_LSH); } |
| 276 | | ShiftExpression RSH AddExpression |
| 277 | { write_exp_elt_opcode (pstate, BINOP_RSH); } |
| 278 | ; |
| 279 | |
| 280 | AddExpression: |
| 281 | MulExpression |
| 282 | | AddExpression '+' MulExpression |
| 283 | { write_exp_elt_opcode (pstate, BINOP_ADD); } |
| 284 | | AddExpression '-' MulExpression |
| 285 | { write_exp_elt_opcode (pstate, BINOP_SUB); } |
| 286 | | AddExpression '~' MulExpression |
| 287 | { write_exp_elt_opcode (pstate, BINOP_CONCAT); } |
| 288 | ; |
| 289 | |
| 290 | MulExpression: |
| 291 | UnaryExpression |
| 292 | | MulExpression '*' UnaryExpression |
| 293 | { write_exp_elt_opcode (pstate, BINOP_MUL); } |
| 294 | | MulExpression '/' UnaryExpression |
| 295 | { write_exp_elt_opcode (pstate, BINOP_DIV); } |
| 296 | | MulExpression '%' UnaryExpression |
| 297 | { write_exp_elt_opcode (pstate, BINOP_REM); } |
| 298 | |
| 299 | UnaryExpression: |
| 300 | '&' UnaryExpression |
| 301 | { write_exp_elt_opcode (pstate, UNOP_ADDR); } |
| 302 | | INCREMENT UnaryExpression |
| 303 | { write_exp_elt_opcode (pstate, UNOP_PREINCREMENT); } |
| 304 | | DECREMENT UnaryExpression |
| 305 | { write_exp_elt_opcode (pstate, UNOP_PREDECREMENT); } |
| 306 | | '*' UnaryExpression |
| 307 | { write_exp_elt_opcode (pstate, UNOP_IND); } |
| 308 | | '-' UnaryExpression |
| 309 | { write_exp_elt_opcode (pstate, UNOP_NEG); } |
| 310 | | '+' UnaryExpression |
| 311 | { write_exp_elt_opcode (pstate, UNOP_PLUS); } |
| 312 | | '!' UnaryExpression |
| 313 | { write_exp_elt_opcode (pstate, UNOP_LOGICAL_NOT); } |
| 314 | | '~' UnaryExpression |
| 315 | { write_exp_elt_opcode (pstate, UNOP_COMPLEMENT); } |
| 316 | | TypeExp '.' SIZEOF_KEYWORD |
| 317 | { write_exp_elt_opcode (pstate, UNOP_SIZEOF); } |
| 318 | | CastExpression |
| 319 | | PowExpression |
| 320 | ; |
| 321 | |
| 322 | CastExpression: |
| 323 | CAST_KEYWORD '(' TypeExp ')' UnaryExpression |
| 324 | { write_exp_elt_opcode (pstate, UNOP_CAST_TYPE); } |
| 325 | /* C style cast is illegal D, but is still recognised in |
| 326 | the grammar, so we keep this around for convenience. */ |
| 327 | | '(' TypeExp ')' UnaryExpression |
| 328 | { write_exp_elt_opcode (pstate, UNOP_CAST_TYPE); } |
| 329 | |
| 330 | ; |
| 331 | |
| 332 | PowExpression: |
| 333 | PostfixExpression |
| 334 | | PostfixExpression HATHAT UnaryExpression |
| 335 | { write_exp_elt_opcode (pstate, BINOP_EXP); } |
| 336 | ; |
| 337 | |
| 338 | PostfixExpression: |
| 339 | PrimaryExpression |
| 340 | | PostfixExpression '.' COMPLETE |
| 341 | { struct stoken s; |
| 342 | mark_struct_expression (pstate); |
| 343 | write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); |
| 344 | s.ptr = ""; |
| 345 | s.length = 0; |
| 346 | write_exp_string (pstate, s); |
| 347 | write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); } |
| 348 | | PostfixExpression '.' IDENTIFIER |
| 349 | { write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); |
| 350 | write_exp_string (pstate, $3); |
| 351 | write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); } |
| 352 | | PostfixExpression '.' IDENTIFIER COMPLETE |
| 353 | { mark_struct_expression (pstate); |
| 354 | write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); |
| 355 | write_exp_string (pstate, $3); |
| 356 | write_exp_elt_opcode (pstate, STRUCTOP_STRUCT); } |
| 357 | | PostfixExpression '.' SIZEOF_KEYWORD |
| 358 | { write_exp_elt_opcode (pstate, UNOP_SIZEOF); } |
| 359 | | PostfixExpression INCREMENT |
| 360 | { write_exp_elt_opcode (pstate, UNOP_POSTINCREMENT); } |
| 361 | | PostfixExpression DECREMENT |
| 362 | { write_exp_elt_opcode (pstate, UNOP_POSTDECREMENT); } |
| 363 | | CallExpression |
| 364 | | IndexExpression |
| 365 | | SliceExpression |
| 366 | ; |
| 367 | |
| 368 | ArgumentList: |
| 369 | AssignExpression |
| 370 | { arglist_len = 1; } |
| 371 | | ArgumentList ',' AssignExpression |
| 372 | { arglist_len++; } |
| 373 | ; |
| 374 | |
| 375 | ArgumentList_opt: |
| 376 | /* EMPTY */ |
| 377 | { arglist_len = 0; } |
| 378 | | ArgumentList |
| 379 | ; |
| 380 | |
| 381 | CallExpression: |
| 382 | PostfixExpression '(' |
| 383 | { start_arglist (); } |
| 384 | ArgumentList_opt ')' |
| 385 | { write_exp_elt_opcode (pstate, OP_FUNCALL); |
| 386 | write_exp_elt_longcst (pstate, (LONGEST) end_arglist ()); |
| 387 | write_exp_elt_opcode (pstate, OP_FUNCALL); } |
| 388 | ; |
| 389 | |
| 390 | IndexExpression: |
| 391 | PostfixExpression '[' ArgumentList ']' |
| 392 | { if (arglist_len > 0) |
| 393 | { |
| 394 | write_exp_elt_opcode (pstate, MULTI_SUBSCRIPT); |
| 395 | write_exp_elt_longcst (pstate, (LONGEST) arglist_len); |
| 396 | write_exp_elt_opcode (pstate, MULTI_SUBSCRIPT); |
| 397 | } |
| 398 | else |
| 399 | write_exp_elt_opcode (pstate, BINOP_SUBSCRIPT); |
| 400 | } |
| 401 | ; |
| 402 | |
| 403 | SliceExpression: |
| 404 | PostfixExpression '[' ']' |
| 405 | { /* Do nothing. */ } |
| 406 | | PostfixExpression '[' AssignExpression DOTDOT AssignExpression ']' |
| 407 | { write_exp_elt_opcode (pstate, TERNOP_SLICE); } |
| 408 | ; |
| 409 | |
| 410 | PrimaryExpression: |
| 411 | '(' Expression ')' |
| 412 | { /* Do nothing. */ } |
| 413 | | IdentifierExp |
| 414 | { struct bound_minimal_symbol msymbol; |
| 415 | char *copy = copy_name ($1); |
| 416 | struct field_of_this_result is_a_field_of_this; |
| 417 | struct block_symbol sym; |
| 418 | |
| 419 | /* Handle VAR, which could be local or global. */ |
| 420 | sym = lookup_symbol (copy, expression_context_block, VAR_DOMAIN, |
| 421 | &is_a_field_of_this); |
| 422 | if (sym.symbol && SYMBOL_CLASS (sym.symbol) != LOC_TYPEDEF) |
| 423 | { |
| 424 | if (symbol_read_needs_frame (sym.symbol)) |
| 425 | innermost_block.update (sym); |
| 426 | write_exp_elt_opcode (pstate, OP_VAR_VALUE); |
| 427 | write_exp_elt_block (pstate, sym.block); |
| 428 | write_exp_elt_sym (pstate, sym.symbol); |
| 429 | write_exp_elt_opcode (pstate, OP_VAR_VALUE); |
| 430 | } |
| 431 | else if (is_a_field_of_this.type != NULL) |
| 432 | { |
| 433 | /* It hangs off of `this'. Must not inadvertently convert from a |
| 434 | method call to data ref. */ |
| 435 | innermost_block.update (sym); |
| 436 | write_exp_elt_opcode (pstate, OP_THIS); |
| 437 | write_exp_elt_opcode (pstate, OP_THIS); |
| 438 | write_exp_elt_opcode (pstate, STRUCTOP_PTR); |
| 439 | write_exp_string (pstate, $1); |
| 440 | write_exp_elt_opcode (pstate, STRUCTOP_PTR); |
| 441 | } |
| 442 | else |
| 443 | { |
| 444 | /* Lookup foreign name in global static symbols. */ |
| 445 | msymbol = lookup_bound_minimal_symbol (copy); |
| 446 | if (msymbol.minsym != NULL) |
| 447 | write_exp_msymbol (pstate, msymbol); |
| 448 | else if (!have_full_symbols () && !have_partial_symbols ()) |
| 449 | error (_("No symbol table is loaded. Use the \"file\" command")); |
| 450 | else |
| 451 | error (_("No symbol \"%s\" in current context."), copy); |
| 452 | } |
| 453 | } |
| 454 | | TypeExp '.' IdentifierExp |
| 455 | { struct type *type = check_typedef ($1); |
| 456 | |
| 457 | /* Check if the qualified name is in the global |
| 458 | context. However if the symbol has not already |
| 459 | been resolved, it's not likely to be found. */ |
| 460 | if (TYPE_CODE (type) == TYPE_CODE_MODULE) |
| 461 | { |
| 462 | struct bound_minimal_symbol msymbol; |
| 463 | struct block_symbol sym; |
| 464 | const char *type_name = TYPE_SAFE_NAME (type); |
| 465 | int type_name_len = strlen (type_name); |
| 466 | std::string name |
| 467 | = string_printf ("%.*s.%.*s", |
| 468 | type_name_len, type_name, |
| 469 | $3.length, $3.ptr); |
| 470 | |
| 471 | sym = |
| 472 | lookup_symbol (name.c_str (), |
| 473 | (const struct block *) NULL, |
| 474 | VAR_DOMAIN, NULL); |
| 475 | if (sym.symbol) |
| 476 | { |
| 477 | write_exp_elt_opcode (pstate, OP_VAR_VALUE); |
| 478 | write_exp_elt_block (pstate, sym.block); |
| 479 | write_exp_elt_sym (pstate, sym.symbol); |
| 480 | write_exp_elt_opcode (pstate, OP_VAR_VALUE); |
| 481 | break; |
| 482 | } |
| 483 | |
| 484 | msymbol = lookup_bound_minimal_symbol (name.c_str ()); |
| 485 | if (msymbol.minsym != NULL) |
| 486 | write_exp_msymbol (pstate, msymbol); |
| 487 | else if (!have_full_symbols () && !have_partial_symbols ()) |
| 488 | error (_("No symbol table is loaded. Use the \"file\" command.")); |
| 489 | else |
| 490 | error (_("No symbol \"%s\" in current context."), |
| 491 | name.c_str ()); |
| 492 | } |
| 493 | |
| 494 | /* Check if the qualified name resolves as a member |
| 495 | of an aggregate or an enum type. */ |
| 496 | if (!type_aggregate_p (type)) |
| 497 | error (_("`%s' is not defined as an aggregate type."), |
| 498 | TYPE_SAFE_NAME (type)); |
| 499 | |
| 500 | write_exp_elt_opcode (pstate, OP_SCOPE); |
| 501 | write_exp_elt_type (pstate, type); |
| 502 | write_exp_string (pstate, $3); |
| 503 | write_exp_elt_opcode (pstate, OP_SCOPE); |
| 504 | } |
| 505 | | DOLLAR_VARIABLE |
| 506 | { write_dollar_variable (pstate, $1); } |
| 507 | | NAME_OR_INT |
| 508 | { YYSTYPE val; |
| 509 | parse_number (pstate, $1.ptr, $1.length, 0, &val); |
| 510 | write_exp_elt_opcode (pstate, OP_LONG); |
| 511 | write_exp_elt_type (pstate, val.typed_val_int.type); |
| 512 | write_exp_elt_longcst (pstate, |
| 513 | (LONGEST) val.typed_val_int.val); |
| 514 | write_exp_elt_opcode (pstate, OP_LONG); } |
| 515 | | NULL_KEYWORD |
| 516 | { struct type *type = parse_d_type (pstate)->builtin_void; |
| 517 | type = lookup_pointer_type (type); |
| 518 | write_exp_elt_opcode (pstate, OP_LONG); |
| 519 | write_exp_elt_type (pstate, type); |
| 520 | write_exp_elt_longcst (pstate, (LONGEST) 0); |
| 521 | write_exp_elt_opcode (pstate, OP_LONG); } |
| 522 | | TRUE_KEYWORD |
| 523 | { write_exp_elt_opcode (pstate, OP_BOOL); |
| 524 | write_exp_elt_longcst (pstate, (LONGEST) 1); |
| 525 | write_exp_elt_opcode (pstate, OP_BOOL); } |
| 526 | | FALSE_KEYWORD |
| 527 | { write_exp_elt_opcode (pstate, OP_BOOL); |
| 528 | write_exp_elt_longcst (pstate, (LONGEST) 0); |
| 529 | write_exp_elt_opcode (pstate, OP_BOOL); } |
| 530 | | INTEGER_LITERAL |
| 531 | { write_exp_elt_opcode (pstate, OP_LONG); |
| 532 | write_exp_elt_type (pstate, $1.type); |
| 533 | write_exp_elt_longcst (pstate, (LONGEST)($1.val)); |
| 534 | write_exp_elt_opcode (pstate, OP_LONG); } |
| 535 | | FLOAT_LITERAL |
| 536 | { write_exp_elt_opcode (pstate, OP_FLOAT); |
| 537 | write_exp_elt_type (pstate, $1.type); |
| 538 | write_exp_elt_floatcst (pstate, $1.val); |
| 539 | write_exp_elt_opcode (pstate, OP_FLOAT); } |
| 540 | | CHARACTER_LITERAL |
| 541 | { struct stoken_vector vec; |
| 542 | vec.len = 1; |
| 543 | vec.tokens = &$1; |
| 544 | write_exp_string_vector (pstate, $1.type, &vec); } |
| 545 | | StringExp |
| 546 | { int i; |
| 547 | write_exp_string_vector (pstate, 0, &$1); |
| 548 | for (i = 0; i < $1.len; ++i) |
| 549 | free ($1.tokens[i].ptr); |
| 550 | free ($1.tokens); } |
| 551 | | ArrayLiteral |
| 552 | { write_exp_elt_opcode (pstate, OP_ARRAY); |
| 553 | write_exp_elt_longcst (pstate, (LONGEST) 0); |
| 554 | write_exp_elt_longcst (pstate, (LONGEST) $1 - 1); |
| 555 | write_exp_elt_opcode (pstate, OP_ARRAY); } |
| 556 | | TYPEOF_KEYWORD '(' Expression ')' |
| 557 | { write_exp_elt_opcode (pstate, OP_TYPEOF); } |
| 558 | ; |
| 559 | |
| 560 | ArrayLiteral: |
| 561 | '[' ArgumentList_opt ']' |
| 562 | { $$ = arglist_len; } |
| 563 | ; |
| 564 | |
| 565 | IdentifierExp: |
| 566 | IDENTIFIER |
| 567 | ; |
| 568 | |
| 569 | StringExp: |
| 570 | STRING_LITERAL |
| 571 | { /* We copy the string here, and not in the |
| 572 | lexer, to guarantee that we do not leak a |
| 573 | string. Note that we follow the |
| 574 | NUL-termination convention of the |
| 575 | lexer. */ |
| 576 | struct typed_stoken *vec = XNEW (struct typed_stoken); |
| 577 | $$.len = 1; |
| 578 | $$.tokens = vec; |
| 579 | |
| 580 | vec->type = $1.type; |
| 581 | vec->length = $1.length; |
| 582 | vec->ptr = (char *) malloc ($1.length + 1); |
| 583 | memcpy (vec->ptr, $1.ptr, $1.length + 1); |
| 584 | } |
| 585 | | StringExp STRING_LITERAL |
| 586 | { /* Note that we NUL-terminate here, but just |
| 587 | for convenience. */ |
| 588 | char *p; |
| 589 | ++$$.len; |
| 590 | $$.tokens |
| 591 | = XRESIZEVEC (struct typed_stoken, $$.tokens, $$.len); |
| 592 | |
| 593 | p = (char *) malloc ($2.length + 1); |
| 594 | memcpy (p, $2.ptr, $2.length + 1); |
| 595 | |
| 596 | $$.tokens[$$.len - 1].type = $2.type; |
| 597 | $$.tokens[$$.len - 1].length = $2.length; |
| 598 | $$.tokens[$$.len - 1].ptr = p; |
| 599 | } |
| 600 | ; |
| 601 | |
| 602 | TypeExp: |
| 603 | '(' TypeExp ')' |
| 604 | { /* Do nothing. */ } |
| 605 | | BasicType |
| 606 | { write_exp_elt_opcode (pstate, OP_TYPE); |
| 607 | write_exp_elt_type (pstate, $1); |
| 608 | write_exp_elt_opcode (pstate, OP_TYPE); } |
| 609 | | BasicType BasicType2 |
| 610 | { $$ = follow_types ($1); |
| 611 | write_exp_elt_opcode (pstate, OP_TYPE); |
| 612 | write_exp_elt_type (pstate, $$); |
| 613 | write_exp_elt_opcode (pstate, OP_TYPE); |
| 614 | } |
| 615 | ; |
| 616 | |
| 617 | BasicType2: |
| 618 | '*' |
| 619 | { push_type (tp_pointer); } |
| 620 | | '*' BasicType2 |
| 621 | { push_type (tp_pointer); } |
| 622 | | '[' INTEGER_LITERAL ']' |
| 623 | { push_type_int ($2.val); |
| 624 | push_type (tp_array); } |
| 625 | | '[' INTEGER_LITERAL ']' BasicType2 |
| 626 | { push_type_int ($2.val); |
| 627 | push_type (tp_array); } |
| 628 | ; |
| 629 | |
| 630 | BasicType: |
| 631 | TYPENAME |
| 632 | { $$ = $1.type; } |
| 633 | ; |
| 634 | |
| 635 | %% |
| 636 | |
| 637 | /* Return true if the type is aggregate-like. */ |
| 638 | |
| 639 | static int |
| 640 | type_aggregate_p (struct type *type) |
| 641 | { |
| 642 | return (TYPE_CODE (type) == TYPE_CODE_STRUCT |
| 643 | || TYPE_CODE (type) == TYPE_CODE_UNION |
| 644 | || TYPE_CODE (type) == TYPE_CODE_MODULE |
| 645 | || (TYPE_CODE (type) == TYPE_CODE_ENUM |
| 646 | && TYPE_DECLARED_CLASS (type))); |
| 647 | } |
| 648 | |
| 649 | /* Take care of parsing a number (anything that starts with a digit). |
| 650 | Set yylval and return the token type; update lexptr. |
| 651 | LEN is the number of characters in it. */ |
| 652 | |
| 653 | /*** Needs some error checking for the float case ***/ |
| 654 | |
| 655 | static int |
| 656 | parse_number (struct parser_state *ps, const char *p, |
| 657 | int len, int parsed_float, YYSTYPE *putithere) |
| 658 | { |
| 659 | ULONGEST n = 0; |
| 660 | ULONGEST prevn = 0; |
| 661 | ULONGEST un; |
| 662 | |
| 663 | int i = 0; |
| 664 | int c; |
| 665 | int base = input_radix; |
| 666 | int unsigned_p = 0; |
| 667 | int long_p = 0; |
| 668 | |
| 669 | /* We have found a "L" or "U" suffix. */ |
| 670 | int found_suffix = 0; |
| 671 | |
| 672 | ULONGEST high_bit; |
| 673 | struct type *signed_type; |
| 674 | struct type *unsigned_type; |
| 675 | |
| 676 | if (parsed_float) |
| 677 | { |
| 678 | char *s, *sp; |
| 679 | |
| 680 | /* Strip out all embedded '_' before passing to parse_float. */ |
| 681 | s = (char *) alloca (len + 1); |
| 682 | sp = s; |
| 683 | while (len-- > 0) |
| 684 | { |
| 685 | if (*p != '_') |
| 686 | *sp++ = *p; |
| 687 | p++; |
| 688 | } |
| 689 | *sp = '\0'; |
| 690 | len = strlen (s); |
| 691 | |
| 692 | /* Check suffix for `i' , `fi' or `li' (idouble, ifloat or ireal). */ |
| 693 | if (len >= 1 && tolower (s[len - 1]) == 'i') |
| 694 | { |
| 695 | if (len >= 2 && tolower (s[len - 2]) == 'f') |
| 696 | { |
| 697 | putithere->typed_val_float.type |
| 698 | = parse_d_type (ps)->builtin_ifloat; |
| 699 | len -= 2; |
| 700 | } |
| 701 | else if (len >= 2 && tolower (s[len - 2]) == 'l') |
| 702 | { |
| 703 | putithere->typed_val_float.type |
| 704 | = parse_d_type (ps)->builtin_ireal; |
| 705 | len -= 2; |
| 706 | } |
| 707 | else |
| 708 | { |
| 709 | putithere->typed_val_float.type |
| 710 | = parse_d_type (ps)->builtin_idouble; |
| 711 | len -= 1; |
| 712 | } |
| 713 | } |
| 714 | /* Check suffix for `f' or `l'' (float or real). */ |
| 715 | else if (len >= 1 && tolower (s[len - 1]) == 'f') |
| 716 | { |
| 717 | putithere->typed_val_float.type |
| 718 | = parse_d_type (ps)->builtin_float; |
| 719 | len -= 1; |
| 720 | } |
| 721 | else if (len >= 1 && tolower (s[len - 1]) == 'l') |
| 722 | { |
| 723 | putithere->typed_val_float.type |
| 724 | = parse_d_type (ps)->builtin_real; |
| 725 | len -= 1; |
| 726 | } |
| 727 | /* Default type if no suffix. */ |
| 728 | else |
| 729 | { |
| 730 | putithere->typed_val_float.type |
| 731 | = parse_d_type (ps)->builtin_double; |
| 732 | } |
| 733 | |
| 734 | if (!parse_float (s, len, |
| 735 | putithere->typed_val_float.type, |
| 736 | putithere->typed_val_float.val)) |
| 737 | return ERROR; |
| 738 | |
| 739 | return FLOAT_LITERAL; |
| 740 | } |
| 741 | |
| 742 | /* Handle base-switching prefixes 0x, 0b, 0 */ |
| 743 | if (p[0] == '0') |
| 744 | switch (p[1]) |
| 745 | { |
| 746 | case 'x': |
| 747 | case 'X': |
| 748 | if (len >= 3) |
| 749 | { |
| 750 | p += 2; |
| 751 | base = 16; |
| 752 | len -= 2; |
| 753 | } |
| 754 | break; |
| 755 | |
| 756 | case 'b': |
| 757 | case 'B': |
| 758 | if (len >= 3) |
| 759 | { |
| 760 | p += 2; |
| 761 | base = 2; |
| 762 | len -= 2; |
| 763 | } |
| 764 | break; |
| 765 | |
| 766 | default: |
| 767 | base = 8; |
| 768 | break; |
| 769 | } |
| 770 | |
| 771 | while (len-- > 0) |
| 772 | { |
| 773 | c = *p++; |
| 774 | if (c == '_') |
| 775 | continue; /* Ignore embedded '_'. */ |
| 776 | if (c >= 'A' && c <= 'Z') |
| 777 | c += 'a' - 'A'; |
| 778 | if (c != 'l' && c != 'u') |
| 779 | n *= base; |
| 780 | if (c >= '0' && c <= '9') |
| 781 | { |
| 782 | if (found_suffix) |
| 783 | return ERROR; |
| 784 | n += i = c - '0'; |
| 785 | } |
| 786 | else |
| 787 | { |
| 788 | if (base > 10 && c >= 'a' && c <= 'f') |
| 789 | { |
| 790 | if (found_suffix) |
| 791 | return ERROR; |
| 792 | n += i = c - 'a' + 10; |
| 793 | } |
| 794 | else if (c == 'l' && long_p == 0) |
| 795 | { |
| 796 | long_p = 1; |
| 797 | found_suffix = 1; |
| 798 | } |
| 799 | else if (c == 'u' && unsigned_p == 0) |
| 800 | { |
| 801 | unsigned_p = 1; |
| 802 | found_suffix = 1; |
| 803 | } |
| 804 | else |
| 805 | return ERROR; /* Char not a digit */ |
| 806 | } |
| 807 | if (i >= base) |
| 808 | return ERROR; /* Invalid digit in this base. */ |
| 809 | /* Portably test for integer overflow. */ |
| 810 | if (c != 'l' && c != 'u') |
| 811 | { |
| 812 | ULONGEST n2 = prevn * base; |
| 813 | if ((n2 / base != prevn) || (n2 + i < prevn)) |
| 814 | error (_("Numeric constant too large.")); |
| 815 | } |
| 816 | prevn = n; |
| 817 | } |
| 818 | |
| 819 | /* An integer constant is an int or a long. An L suffix forces it to |
| 820 | be long, and a U suffix forces it to be unsigned. To figure out |
| 821 | whether it fits, we shift it right and see whether anything remains. |
| 822 | Note that we can't shift sizeof (LONGEST) * HOST_CHAR_BIT bits or |
| 823 | more in one operation, because many compilers will warn about such a |
| 824 | shift (which always produces a zero result). To deal with the case |
| 825 | where it is we just always shift the value more than once, with fewer |
| 826 | bits each time. */ |
| 827 | un = (ULONGEST) n >> 2; |
| 828 | if (long_p == 0 && (un >> 30) == 0) |
| 829 | { |
| 830 | high_bit = ((ULONGEST) 1) << 31; |
| 831 | signed_type = parse_d_type (ps)->builtin_int; |
| 832 | /* For decimal notation, keep the sign of the worked out type. */ |
| 833 | if (base == 10 && !unsigned_p) |
| 834 | unsigned_type = parse_d_type (ps)->builtin_long; |
| 835 | else |
| 836 | unsigned_type = parse_d_type (ps)->builtin_uint; |
| 837 | } |
| 838 | else |
| 839 | { |
| 840 | int shift; |
| 841 | if (sizeof (ULONGEST) * HOST_CHAR_BIT < 64) |
| 842 | /* A long long does not fit in a LONGEST. */ |
| 843 | shift = (sizeof (ULONGEST) * HOST_CHAR_BIT - 1); |
| 844 | else |
| 845 | shift = 63; |
| 846 | high_bit = (ULONGEST) 1 << shift; |
| 847 | signed_type = parse_d_type (ps)->builtin_long; |
| 848 | unsigned_type = parse_d_type (ps)->builtin_ulong; |
| 849 | } |
| 850 | |
| 851 | putithere->typed_val_int.val = n; |
| 852 | |
| 853 | /* If the high bit of the worked out type is set then this number |
| 854 | has to be unsigned_type. */ |
| 855 | if (unsigned_p || (n & high_bit)) |
| 856 | putithere->typed_val_int.type = unsigned_type; |
| 857 | else |
| 858 | putithere->typed_val_int.type = signed_type; |
| 859 | |
| 860 | return INTEGER_LITERAL; |
| 861 | } |
| 862 | |
| 863 | /* Temporary obstack used for holding strings. */ |
| 864 | static struct obstack tempbuf; |
| 865 | static int tempbuf_init; |
| 866 | |
| 867 | /* Parse a string or character literal from TOKPTR. The string or |
| 868 | character may be wide or unicode. *OUTPTR is set to just after the |
| 869 | end of the literal in the input string. The resulting token is |
| 870 | stored in VALUE. This returns a token value, either STRING or |
| 871 | CHAR, depending on what was parsed. *HOST_CHARS is set to the |
| 872 | number of host characters in the literal. */ |
| 873 | |
| 874 | static int |
| 875 | parse_string_or_char (const char *tokptr, const char **outptr, |
| 876 | struct typed_stoken *value, int *host_chars) |
| 877 | { |
| 878 | int quote; |
| 879 | |
| 880 | /* Build the gdb internal form of the input string in tempbuf. Note |
| 881 | that the buffer is null byte terminated *only* for the |
| 882 | convenience of debugging gdb itself and printing the buffer |
| 883 | contents when the buffer contains no embedded nulls. Gdb does |
| 884 | not depend upon the buffer being null byte terminated, it uses |
| 885 | the length string instead. This allows gdb to handle C strings |
| 886 | (as well as strings in other languages) with embedded null |
| 887 | bytes */ |
| 888 | |
| 889 | if (!tempbuf_init) |
| 890 | tempbuf_init = 1; |
| 891 | else |
| 892 | obstack_free (&tempbuf, NULL); |
| 893 | obstack_init (&tempbuf); |
| 894 | |
| 895 | /* Skip the quote. */ |
| 896 | quote = *tokptr; |
| 897 | ++tokptr; |
| 898 | |
| 899 | *host_chars = 0; |
| 900 | |
| 901 | while (*tokptr) |
| 902 | { |
| 903 | char c = *tokptr; |
| 904 | if (c == '\\') |
| 905 | { |
| 906 | ++tokptr; |
| 907 | *host_chars += c_parse_escape (&tokptr, &tempbuf); |
| 908 | } |
| 909 | else if (c == quote) |
| 910 | break; |
| 911 | else |
| 912 | { |
| 913 | obstack_1grow (&tempbuf, c); |
| 914 | ++tokptr; |
| 915 | /* FIXME: this does the wrong thing with multi-byte host |
| 916 | characters. We could use mbrlen here, but that would |
| 917 | make "set host-charset" a bit less useful. */ |
| 918 | ++*host_chars; |
| 919 | } |
| 920 | } |
| 921 | |
| 922 | if (*tokptr != quote) |
| 923 | { |
| 924 | if (quote == '"' || quote == '`') |
| 925 | error (_("Unterminated string in expression.")); |
| 926 | else |
| 927 | error (_("Unmatched single quote.")); |
| 928 | } |
| 929 | ++tokptr; |
| 930 | |
| 931 | /* FIXME: should instead use own language string_type enum |
| 932 | and handle D-specific string suffixes here. */ |
| 933 | if (quote == '\'') |
| 934 | value->type = C_CHAR; |
| 935 | else |
| 936 | value->type = C_STRING; |
| 937 | |
| 938 | value->ptr = (char *) obstack_base (&tempbuf); |
| 939 | value->length = obstack_object_size (&tempbuf); |
| 940 | |
| 941 | *outptr = tokptr; |
| 942 | |
| 943 | return quote == '\'' ? CHARACTER_LITERAL : STRING_LITERAL; |
| 944 | } |
| 945 | |
| 946 | struct token |
| 947 | { |
| 948 | const char *oper; |
| 949 | int token; |
| 950 | enum exp_opcode opcode; |
| 951 | }; |
| 952 | |
| 953 | static const struct token tokentab3[] = |
| 954 | { |
| 955 | {"^^=", ASSIGN_MODIFY, BINOP_EXP}, |
| 956 | {"<<=", ASSIGN_MODIFY, BINOP_LSH}, |
| 957 | {">>=", ASSIGN_MODIFY, BINOP_RSH}, |
| 958 | }; |
| 959 | |
| 960 | static const struct token tokentab2[] = |
| 961 | { |
| 962 | {"+=", ASSIGN_MODIFY, BINOP_ADD}, |
| 963 | {"-=", ASSIGN_MODIFY, BINOP_SUB}, |
| 964 | {"*=", ASSIGN_MODIFY, BINOP_MUL}, |
| 965 | {"/=", ASSIGN_MODIFY, BINOP_DIV}, |
| 966 | {"%=", ASSIGN_MODIFY, BINOP_REM}, |
| 967 | {"|=", ASSIGN_MODIFY, BINOP_BITWISE_IOR}, |
| 968 | {"&=", ASSIGN_MODIFY, BINOP_BITWISE_AND}, |
| 969 | {"^=", ASSIGN_MODIFY, BINOP_BITWISE_XOR}, |
| 970 | {"++", INCREMENT, BINOP_END}, |
| 971 | {"--", DECREMENT, BINOP_END}, |
| 972 | {"&&", ANDAND, BINOP_END}, |
| 973 | {"||", OROR, BINOP_END}, |
| 974 | {"^^", HATHAT, BINOP_END}, |
| 975 | {"<<", LSH, BINOP_END}, |
| 976 | {">>", RSH, BINOP_END}, |
| 977 | {"==", EQUAL, BINOP_END}, |
| 978 | {"!=", NOTEQUAL, BINOP_END}, |
| 979 | {"<=", LEQ, BINOP_END}, |
| 980 | {">=", GEQ, BINOP_END}, |
| 981 | {"..", DOTDOT, BINOP_END}, |
| 982 | }; |
| 983 | |
| 984 | /* Identifier-like tokens. */ |
| 985 | static const struct token ident_tokens[] = |
| 986 | { |
| 987 | {"is", IDENTITY, BINOP_END}, |
| 988 | {"!is", NOTIDENTITY, BINOP_END}, |
| 989 | |
| 990 | {"cast", CAST_KEYWORD, OP_NULL}, |
| 991 | {"const", CONST_KEYWORD, OP_NULL}, |
| 992 | {"immutable", IMMUTABLE_KEYWORD, OP_NULL}, |
| 993 | {"shared", SHARED_KEYWORD, OP_NULL}, |
| 994 | {"super", SUPER_KEYWORD, OP_NULL}, |
| 995 | |
| 996 | {"null", NULL_KEYWORD, OP_NULL}, |
| 997 | {"true", TRUE_KEYWORD, OP_NULL}, |
| 998 | {"false", FALSE_KEYWORD, OP_NULL}, |
| 999 | |
| 1000 | {"init", INIT_KEYWORD, OP_NULL}, |
| 1001 | {"sizeof", SIZEOF_KEYWORD, OP_NULL}, |
| 1002 | {"typeof", TYPEOF_KEYWORD, OP_NULL}, |
| 1003 | {"typeid", TYPEID_KEYWORD, OP_NULL}, |
| 1004 | |
| 1005 | {"delegate", DELEGATE_KEYWORD, OP_NULL}, |
| 1006 | {"function", FUNCTION_KEYWORD, OP_NULL}, |
| 1007 | {"struct", STRUCT_KEYWORD, OP_NULL}, |
| 1008 | {"union", UNION_KEYWORD, OP_NULL}, |
| 1009 | {"class", CLASS_KEYWORD, OP_NULL}, |
| 1010 | {"interface", INTERFACE_KEYWORD, OP_NULL}, |
| 1011 | {"enum", ENUM_KEYWORD, OP_NULL}, |
| 1012 | {"template", TEMPLATE_KEYWORD, OP_NULL}, |
| 1013 | }; |
| 1014 | |
| 1015 | /* This is set if a NAME token appeared at the very end of the input |
| 1016 | string, with no whitespace separating the name from the EOF. This |
| 1017 | is used only when parsing to do field name completion. */ |
| 1018 | static int saw_name_at_eof; |
| 1019 | |
| 1020 | /* This is set if the previously-returned token was a structure operator. |
| 1021 | This is used only when parsing to do field name completion. */ |
| 1022 | static int last_was_structop; |
| 1023 | |
| 1024 | /* Read one token, getting characters through lexptr. */ |
| 1025 | |
| 1026 | static int |
| 1027 | lex_one_token (struct parser_state *par_state) |
| 1028 | { |
| 1029 | int c; |
| 1030 | int namelen; |
| 1031 | unsigned int i; |
| 1032 | const char *tokstart; |
| 1033 | int saw_structop = last_was_structop; |
| 1034 | char *copy; |
| 1035 | |
| 1036 | last_was_structop = 0; |
| 1037 | |
| 1038 | retry: |
| 1039 | |
| 1040 | prev_lexptr = lexptr; |
| 1041 | |
| 1042 | tokstart = lexptr; |
| 1043 | /* See if it is a special token of length 3. */ |
| 1044 | for (i = 0; i < sizeof tokentab3 / sizeof tokentab3[0]; i++) |
| 1045 | if (strncmp (tokstart, tokentab3[i].oper, 3) == 0) |
| 1046 | { |
| 1047 | lexptr += 3; |
| 1048 | yylval.opcode = tokentab3[i].opcode; |
| 1049 | return tokentab3[i].token; |
| 1050 | } |
| 1051 | |
| 1052 | /* See if it is a special token of length 2. */ |
| 1053 | for (i = 0; i < sizeof tokentab2 / sizeof tokentab2[0]; i++) |
| 1054 | if (strncmp (tokstart, tokentab2[i].oper, 2) == 0) |
| 1055 | { |
| 1056 | lexptr += 2; |
| 1057 | yylval.opcode = tokentab2[i].opcode; |
| 1058 | return tokentab2[i].token; |
| 1059 | } |
| 1060 | |
| 1061 | switch (c = *tokstart) |
| 1062 | { |
| 1063 | case 0: |
| 1064 | /* If we're parsing for field name completion, and the previous |
| 1065 | token allows such completion, return a COMPLETE token. |
| 1066 | Otherwise, we were already scanning the original text, and |
| 1067 | we're really done. */ |
| 1068 | if (saw_name_at_eof) |
| 1069 | { |
| 1070 | saw_name_at_eof = 0; |
| 1071 | return COMPLETE; |
| 1072 | } |
| 1073 | else if (saw_structop) |
| 1074 | return COMPLETE; |
| 1075 | else |
| 1076 | return 0; |
| 1077 | |
| 1078 | case ' ': |
| 1079 | case '\t': |
| 1080 | case '\n': |
| 1081 | lexptr++; |
| 1082 | goto retry; |
| 1083 | |
| 1084 | case '[': |
| 1085 | case '(': |
| 1086 | paren_depth++; |
| 1087 | lexptr++; |
| 1088 | return c; |
| 1089 | |
| 1090 | case ']': |
| 1091 | case ')': |
| 1092 | if (paren_depth == 0) |
| 1093 | return 0; |
| 1094 | paren_depth--; |
| 1095 | lexptr++; |
| 1096 | return c; |
| 1097 | |
| 1098 | case ',': |
| 1099 | if (comma_terminates && paren_depth == 0) |
| 1100 | return 0; |
| 1101 | lexptr++; |
| 1102 | return c; |
| 1103 | |
| 1104 | case '.': |
| 1105 | /* Might be a floating point number. */ |
| 1106 | if (lexptr[1] < '0' || lexptr[1] > '9') |
| 1107 | { |
| 1108 | if (parse_completion) |
| 1109 | last_was_structop = 1; |
| 1110 | goto symbol; /* Nope, must be a symbol. */ |
| 1111 | } |
| 1112 | /* FALL THRU. */ |
| 1113 | |
| 1114 | case '0': |
| 1115 | case '1': |
| 1116 | case '2': |
| 1117 | case '3': |
| 1118 | case '4': |
| 1119 | case '5': |
| 1120 | case '6': |
| 1121 | case '7': |
| 1122 | case '8': |
| 1123 | case '9': |
| 1124 | { |
| 1125 | /* It's a number. */ |
| 1126 | int got_dot = 0, got_e = 0, toktype; |
| 1127 | const char *p = tokstart; |
| 1128 | int hex = input_radix > 10; |
| 1129 | |
| 1130 | if (c == '0' && (p[1] == 'x' || p[1] == 'X')) |
| 1131 | { |
| 1132 | p += 2; |
| 1133 | hex = 1; |
| 1134 | } |
| 1135 | |
| 1136 | for (;; ++p) |
| 1137 | { |
| 1138 | /* Hex exponents start with 'p', because 'e' is a valid hex |
| 1139 | digit and thus does not indicate a floating point number |
| 1140 | when the radix is hex. */ |
| 1141 | if ((!hex && !got_e && tolower (p[0]) == 'e') |
| 1142 | || (hex && !got_e && tolower (p[0] == 'p'))) |
| 1143 | got_dot = got_e = 1; |
| 1144 | /* A '.' always indicates a decimal floating point number |
| 1145 | regardless of the radix. If we have a '..' then its the |
| 1146 | end of the number and the beginning of a slice. */ |
| 1147 | else if (!got_dot && (p[0] == '.' && p[1] != '.')) |
| 1148 | got_dot = 1; |
| 1149 | /* This is the sign of the exponent, not the end of the number. */ |
| 1150 | else if (got_e && (tolower (p[-1]) == 'e' || tolower (p[-1]) == 'p') |
| 1151 | && (*p == '-' || *p == '+')) |
| 1152 | continue; |
| 1153 | /* We will take any letters or digits, ignoring any embedded '_'. |
| 1154 | parse_number will complain if past the radix, or if L or U are |
| 1155 | not final. */ |
| 1156 | else if ((*p < '0' || *p > '9') && (*p != '_') |
| 1157 | && ((*p < 'a' || *p > 'z') && (*p < 'A' || *p > 'Z'))) |
| 1158 | break; |
| 1159 | } |
| 1160 | |
| 1161 | toktype = parse_number (par_state, tokstart, p - tokstart, |
| 1162 | got_dot|got_e, &yylval); |
| 1163 | if (toktype == ERROR) |
| 1164 | { |
| 1165 | char *err_copy = (char *) alloca (p - tokstart + 1); |
| 1166 | |
| 1167 | memcpy (err_copy, tokstart, p - tokstart); |
| 1168 | err_copy[p - tokstart] = 0; |
| 1169 | error (_("Invalid number \"%s\"."), err_copy); |
| 1170 | } |
| 1171 | lexptr = p; |
| 1172 | return toktype; |
| 1173 | } |
| 1174 | |
| 1175 | case '@': |
| 1176 | { |
| 1177 | const char *p = &tokstart[1]; |
| 1178 | size_t len = strlen ("entry"); |
| 1179 | |
| 1180 | while (isspace (*p)) |
| 1181 | p++; |
| 1182 | if (strncmp (p, "entry", len) == 0 && !isalnum (p[len]) |
| 1183 | && p[len] != '_') |
| 1184 | { |
| 1185 | lexptr = &p[len]; |
| 1186 | return ENTRY; |
| 1187 | } |
| 1188 | } |
| 1189 | /* FALLTHRU */ |
| 1190 | case '+': |
| 1191 | case '-': |
| 1192 | case '*': |
| 1193 | case '/': |
| 1194 | case '%': |
| 1195 | case '|': |
| 1196 | case '&': |
| 1197 | case '^': |
| 1198 | case '~': |
| 1199 | case '!': |
| 1200 | case '<': |
| 1201 | case '>': |
| 1202 | case '?': |
| 1203 | case ':': |
| 1204 | case '=': |
| 1205 | case '{': |
| 1206 | case '}': |
| 1207 | symbol: |
| 1208 | lexptr++; |
| 1209 | return c; |
| 1210 | |
| 1211 | case '\'': |
| 1212 | case '"': |
| 1213 | case '`': |
| 1214 | { |
| 1215 | int host_len; |
| 1216 | int result = parse_string_or_char (tokstart, &lexptr, &yylval.tsval, |
| 1217 | &host_len); |
| 1218 | if (result == CHARACTER_LITERAL) |
| 1219 | { |
| 1220 | if (host_len == 0) |
| 1221 | error (_("Empty character constant.")); |
| 1222 | else if (host_len > 2 && c == '\'') |
| 1223 | { |
| 1224 | ++tokstart; |
| 1225 | namelen = lexptr - tokstart - 1; |
| 1226 | goto tryname; |
| 1227 | } |
| 1228 | else if (host_len > 1) |
| 1229 | error (_("Invalid character constant.")); |
| 1230 | } |
| 1231 | return result; |
| 1232 | } |
| 1233 | } |
| 1234 | |
| 1235 | if (!(c == '_' || c == '$' |
| 1236 | || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'))) |
| 1237 | /* We must have come across a bad character (e.g. ';'). */ |
| 1238 | error (_("Invalid character '%c' in expression"), c); |
| 1239 | |
| 1240 | /* It's a name. See how long it is. */ |
| 1241 | namelen = 0; |
| 1242 | for (c = tokstart[namelen]; |
| 1243 | (c == '_' || c == '$' || (c >= '0' && c <= '9') |
| 1244 | || (c >= 'a' && c <= 'z') || (c >= 'A' && c <= 'Z'));) |
| 1245 | c = tokstart[++namelen]; |
| 1246 | |
| 1247 | /* The token "if" terminates the expression and is NOT |
| 1248 | removed from the input stream. */ |
| 1249 | if (namelen == 2 && tokstart[0] == 'i' && tokstart[1] == 'f') |
| 1250 | return 0; |
| 1251 | |
| 1252 | /* For the same reason (breakpoint conditions), "thread N" |
| 1253 | terminates the expression. "thread" could be an identifier, but |
| 1254 | an identifier is never followed by a number without intervening |
| 1255 | punctuation. "task" is similar. Handle abbreviations of these, |
| 1256 | similarly to breakpoint.c:find_condition_and_thread. */ |
| 1257 | if (namelen >= 1 |
| 1258 | && (strncmp (tokstart, "thread", namelen) == 0 |
| 1259 | || strncmp (tokstart, "task", namelen) == 0) |
| 1260 | && (tokstart[namelen] == ' ' || tokstart[namelen] == '\t')) |
| 1261 | { |
| 1262 | const char *p = tokstart + namelen + 1; |
| 1263 | |
| 1264 | while (*p == ' ' || *p == '\t') |
| 1265 | p++; |
| 1266 | if (*p >= '0' && *p <= '9') |
| 1267 | return 0; |
| 1268 | } |
| 1269 | |
| 1270 | lexptr += namelen; |
| 1271 | |
| 1272 | tryname: |
| 1273 | |
| 1274 | yylval.sval.ptr = tokstart; |
| 1275 | yylval.sval.length = namelen; |
| 1276 | |
| 1277 | /* Catch specific keywords. */ |
| 1278 | copy = copy_name (yylval.sval); |
| 1279 | for (i = 0; i < sizeof ident_tokens / sizeof ident_tokens[0]; i++) |
| 1280 | if (strcmp (copy, ident_tokens[i].oper) == 0) |
| 1281 | { |
| 1282 | /* It is ok to always set this, even though we don't always |
| 1283 | strictly need to. */ |
| 1284 | yylval.opcode = ident_tokens[i].opcode; |
| 1285 | return ident_tokens[i].token; |
| 1286 | } |
| 1287 | |
| 1288 | if (*tokstart == '$') |
| 1289 | return DOLLAR_VARIABLE; |
| 1290 | |
| 1291 | yylval.tsym.type |
| 1292 | = language_lookup_primitive_type (parse_language (par_state), |
| 1293 | parse_gdbarch (par_state), copy); |
| 1294 | if (yylval.tsym.type != NULL) |
| 1295 | return TYPENAME; |
| 1296 | |
| 1297 | /* Input names that aren't symbols but ARE valid hex numbers, |
| 1298 | when the input radix permits them, can be names or numbers |
| 1299 | depending on the parse. Note we support radixes > 16 here. */ |
| 1300 | if ((tokstart[0] >= 'a' && tokstart[0] < 'a' + input_radix - 10) |
| 1301 | || (tokstart[0] >= 'A' && tokstart[0] < 'A' + input_radix - 10)) |
| 1302 | { |
| 1303 | YYSTYPE newlval; /* Its value is ignored. */ |
| 1304 | int hextype = parse_number (par_state, tokstart, namelen, 0, &newlval); |
| 1305 | if (hextype == INTEGER_LITERAL) |
| 1306 | return NAME_OR_INT; |
| 1307 | } |
| 1308 | |
| 1309 | if (parse_completion && *lexptr == '\0') |
| 1310 | saw_name_at_eof = 1; |
| 1311 | |
| 1312 | return IDENTIFIER; |
| 1313 | } |
| 1314 | |
| 1315 | /* An object of this type is pushed on a FIFO by the "outer" lexer. */ |
| 1316 | struct token_and_value |
| 1317 | { |
| 1318 | int token; |
| 1319 | YYSTYPE value; |
| 1320 | }; |
| 1321 | |
| 1322 | |
| 1323 | /* A FIFO of tokens that have been read but not yet returned to the |
| 1324 | parser. */ |
| 1325 | static std::vector<token_and_value> token_fifo; |
| 1326 | |
| 1327 | /* Non-zero if the lexer should return tokens from the FIFO. */ |
| 1328 | static int popping; |
| 1329 | |
| 1330 | /* Temporary storage for yylex; this holds symbol names as they are |
| 1331 | built up. */ |
| 1332 | static auto_obstack name_obstack; |
| 1333 | |
| 1334 | /* Classify an IDENTIFIER token. The contents of the token are in `yylval'. |
| 1335 | Updates yylval and returns the new token type. BLOCK is the block |
| 1336 | in which lookups start; this can be NULL to mean the global scope. */ |
| 1337 | |
| 1338 | static int |
| 1339 | classify_name (struct parser_state *par_state, const struct block *block) |
| 1340 | { |
| 1341 | struct block_symbol sym; |
| 1342 | char *copy; |
| 1343 | struct field_of_this_result is_a_field_of_this; |
| 1344 | |
| 1345 | copy = copy_name (yylval.sval); |
| 1346 | |
| 1347 | sym = lookup_symbol (copy, block, VAR_DOMAIN, &is_a_field_of_this); |
| 1348 | if (sym.symbol && SYMBOL_CLASS (sym.symbol) == LOC_TYPEDEF) |
| 1349 | { |
| 1350 | yylval.tsym.type = SYMBOL_TYPE (sym.symbol); |
| 1351 | return TYPENAME; |
| 1352 | } |
| 1353 | else if (sym.symbol == NULL) |
| 1354 | { |
| 1355 | /* Look-up first for a module name, then a type. */ |
| 1356 | sym = lookup_symbol (copy, block, MODULE_DOMAIN, NULL); |
| 1357 | if (sym.symbol == NULL) |
| 1358 | sym = lookup_symbol (copy, block, STRUCT_DOMAIN, NULL); |
| 1359 | |
| 1360 | if (sym.symbol != NULL) |
| 1361 | { |
| 1362 | yylval.tsym.type = SYMBOL_TYPE (sym.symbol); |
| 1363 | return TYPENAME; |
| 1364 | } |
| 1365 | |
| 1366 | return UNKNOWN_NAME; |
| 1367 | } |
| 1368 | |
| 1369 | return IDENTIFIER; |
| 1370 | } |
| 1371 | |
| 1372 | /* Like classify_name, but used by the inner loop of the lexer, when a |
| 1373 | name might have already been seen. CONTEXT is the context type, or |
| 1374 | NULL if this is the first component of a name. */ |
| 1375 | |
| 1376 | static int |
| 1377 | classify_inner_name (struct parser_state *par_state, |
| 1378 | const struct block *block, struct type *context) |
| 1379 | { |
| 1380 | struct type *type; |
| 1381 | char *copy; |
| 1382 | |
| 1383 | if (context == NULL) |
| 1384 | return classify_name (par_state, block); |
| 1385 | |
| 1386 | type = check_typedef (context); |
| 1387 | if (!type_aggregate_p (type)) |
| 1388 | return ERROR; |
| 1389 | |
| 1390 | copy = copy_name (yylval.ssym.stoken); |
| 1391 | yylval.ssym.sym = d_lookup_nested_symbol (type, copy, block); |
| 1392 | |
| 1393 | if (yylval.ssym.sym.symbol == NULL) |
| 1394 | return ERROR; |
| 1395 | |
| 1396 | if (SYMBOL_CLASS (yylval.ssym.sym.symbol) == LOC_TYPEDEF) |
| 1397 | { |
| 1398 | yylval.tsym.type = SYMBOL_TYPE (yylval.ssym.sym.symbol); |
| 1399 | return TYPENAME; |
| 1400 | } |
| 1401 | |
| 1402 | return IDENTIFIER; |
| 1403 | } |
| 1404 | |
| 1405 | /* The outer level of a two-level lexer. This calls the inner lexer |
| 1406 | to return tokens. It then either returns these tokens, or |
| 1407 | aggregates them into a larger token. This lets us work around a |
| 1408 | problem in our parsing approach, where the parser could not |
| 1409 | distinguish between qualified names and qualified types at the |
| 1410 | right point. */ |
| 1411 | |
| 1412 | static int |
| 1413 | yylex (void) |
| 1414 | { |
| 1415 | token_and_value current; |
| 1416 | int last_was_dot; |
| 1417 | struct type *context_type = NULL; |
| 1418 | int last_to_examine, next_to_examine, checkpoint; |
| 1419 | const struct block *search_block; |
| 1420 | |
| 1421 | if (popping && !token_fifo.empty ()) |
| 1422 | goto do_pop; |
| 1423 | popping = 0; |
| 1424 | |
| 1425 | /* Read the first token and decide what to do. */ |
| 1426 | current.token = lex_one_token (pstate); |
| 1427 | if (current.token != IDENTIFIER && current.token != '.') |
| 1428 | return current.token; |
| 1429 | |
| 1430 | /* Read any sequence of alternating "." and identifier tokens into |
| 1431 | the token FIFO. */ |
| 1432 | current.value = yylval; |
| 1433 | token_fifo.push_back (current); |
| 1434 | last_was_dot = current.token == '.'; |
| 1435 | |
| 1436 | while (1) |
| 1437 | { |
| 1438 | current.token = lex_one_token (pstate); |
| 1439 | current.value = yylval; |
| 1440 | token_fifo.push_back (current); |
| 1441 | |
| 1442 | if ((last_was_dot && current.token != IDENTIFIER) |
| 1443 | || (!last_was_dot && current.token != '.')) |
| 1444 | break; |
| 1445 | |
| 1446 | last_was_dot = !last_was_dot; |
| 1447 | } |
| 1448 | popping = 1; |
| 1449 | |
| 1450 | /* We always read one extra token, so compute the number of tokens |
| 1451 | to examine accordingly. */ |
| 1452 | last_to_examine = token_fifo.size () - 2; |
| 1453 | next_to_examine = 0; |
| 1454 | |
| 1455 | current = token_fifo[next_to_examine]; |
| 1456 | ++next_to_examine; |
| 1457 | |
| 1458 | /* If we are not dealing with a typename, now is the time to find out. */ |
| 1459 | if (current.token == IDENTIFIER) |
| 1460 | { |
| 1461 | yylval = current.value; |
| 1462 | current.token = classify_name (pstate, expression_context_block); |
| 1463 | current.value = yylval; |
| 1464 | } |
| 1465 | |
| 1466 | /* If the IDENTIFIER is not known, it could be a package symbol, |
| 1467 | first try building up a name until we find the qualified module. */ |
| 1468 | if (current.token == UNKNOWN_NAME) |
| 1469 | { |
| 1470 | name_obstack.clear (); |
| 1471 | obstack_grow (&name_obstack, current.value.sval.ptr, |
| 1472 | current.value.sval.length); |
| 1473 | |
| 1474 | last_was_dot = 0; |
| 1475 | |
| 1476 | while (next_to_examine <= last_to_examine) |
| 1477 | { |
| 1478 | token_and_value next; |
| 1479 | |
| 1480 | next = token_fifo[next_to_examine]; |
| 1481 | ++next_to_examine; |
| 1482 | |
| 1483 | if (next.token == IDENTIFIER && last_was_dot) |
| 1484 | { |
| 1485 | /* Update the partial name we are constructing. */ |
| 1486 | obstack_grow_str (&name_obstack, "."); |
| 1487 | obstack_grow (&name_obstack, next.value.sval.ptr, |
| 1488 | next.value.sval.length); |
| 1489 | |
| 1490 | yylval.sval.ptr = (char *) obstack_base (&name_obstack); |
| 1491 | yylval.sval.length = obstack_object_size (&name_obstack); |
| 1492 | |
| 1493 | current.token = classify_name (pstate, expression_context_block); |
| 1494 | current.value = yylval; |
| 1495 | |
| 1496 | /* We keep going until we find a TYPENAME. */ |
| 1497 | if (current.token == TYPENAME) |
| 1498 | { |
| 1499 | /* Install it as the first token in the FIFO. */ |
| 1500 | token_fifo[0] = current; |
| 1501 | token_fifo.erase (token_fifo.begin () + 1, |
| 1502 | token_fifo.begin () + next_to_examine); |
| 1503 | break; |
| 1504 | } |
| 1505 | } |
| 1506 | else if (next.token == '.' && !last_was_dot) |
| 1507 | last_was_dot = 1; |
| 1508 | else |
| 1509 | { |
| 1510 | /* We've reached the end of the name. */ |
| 1511 | break; |
| 1512 | } |
| 1513 | } |
| 1514 | |
| 1515 | /* Reset our current token back to the start, if we found nothing |
| 1516 | this means that we will just jump to do pop. */ |
| 1517 | current = token_fifo[0]; |
| 1518 | next_to_examine = 1; |
| 1519 | } |
| 1520 | if (current.token != TYPENAME && current.token != '.') |
| 1521 | goto do_pop; |
| 1522 | |
| 1523 | name_obstack.clear (); |
| 1524 | checkpoint = 0; |
| 1525 | if (current.token == '.') |
| 1526 | search_block = NULL; |
| 1527 | else |
| 1528 | { |
| 1529 | gdb_assert (current.token == TYPENAME); |
| 1530 | search_block = expression_context_block; |
| 1531 | obstack_grow (&name_obstack, current.value.sval.ptr, |
| 1532 | current.value.sval.length); |
| 1533 | context_type = current.value.tsym.type; |
| 1534 | checkpoint = 1; |
| 1535 | } |
| 1536 | |
| 1537 | last_was_dot = current.token == '.'; |
| 1538 | |
| 1539 | while (next_to_examine <= last_to_examine) |
| 1540 | { |
| 1541 | token_and_value next; |
| 1542 | |
| 1543 | next = token_fifo[next_to_examine]; |
| 1544 | ++next_to_examine; |
| 1545 | |
| 1546 | if (next.token == IDENTIFIER && last_was_dot) |
| 1547 | { |
| 1548 | int classification; |
| 1549 | |
| 1550 | yylval = next.value; |
| 1551 | classification = classify_inner_name (pstate, search_block, |
| 1552 | context_type); |
| 1553 | /* We keep going until we either run out of names, or until |
| 1554 | we have a qualified name which is not a type. */ |
| 1555 | if (classification != TYPENAME && classification != IDENTIFIER) |
| 1556 | break; |
| 1557 | |
| 1558 | /* Accept up to this token. */ |
| 1559 | checkpoint = next_to_examine; |
| 1560 | |
| 1561 | /* Update the partial name we are constructing. */ |
| 1562 | if (context_type != NULL) |
| 1563 | { |
| 1564 | /* We don't want to put a leading "." into the name. */ |
| 1565 | obstack_grow_str (&name_obstack, "."); |
| 1566 | } |
| 1567 | obstack_grow (&name_obstack, next.value.sval.ptr, |
| 1568 | next.value.sval.length); |
| 1569 | |
| 1570 | yylval.sval.ptr = (char *) obstack_base (&name_obstack); |
| 1571 | yylval.sval.length = obstack_object_size (&name_obstack); |
| 1572 | current.value = yylval; |
| 1573 | current.token = classification; |
| 1574 | |
| 1575 | last_was_dot = 0; |
| 1576 | |
| 1577 | if (classification == IDENTIFIER) |
| 1578 | break; |
| 1579 | |
| 1580 | context_type = yylval.tsym.type; |
| 1581 | } |
| 1582 | else if (next.token == '.' && !last_was_dot) |
| 1583 | last_was_dot = 1; |
| 1584 | else |
| 1585 | { |
| 1586 | /* We've reached the end of the name. */ |
| 1587 | break; |
| 1588 | } |
| 1589 | } |
| 1590 | |
| 1591 | /* If we have a replacement token, install it as the first token in |
| 1592 | the FIFO, and delete the other constituent tokens. */ |
| 1593 | if (checkpoint > 0) |
| 1594 | { |
| 1595 | token_fifo[0] = current; |
| 1596 | if (checkpoint > 1) |
| 1597 | token_fifo.erase (token_fifo.begin () + 1, |
| 1598 | token_fifo.begin () + checkpoint); |
| 1599 | } |
| 1600 | |
| 1601 | do_pop: |
| 1602 | current = token_fifo[0]; |
| 1603 | token_fifo.erase (token_fifo.begin ()); |
| 1604 | yylval = current.value; |
| 1605 | return current.token; |
| 1606 | } |
| 1607 | |
| 1608 | int |
| 1609 | d_parse (struct parser_state *par_state) |
| 1610 | { |
| 1611 | /* Setting up the parser state. */ |
| 1612 | scoped_restore pstate_restore = make_scoped_restore (&pstate); |
| 1613 | gdb_assert (par_state != NULL); |
| 1614 | pstate = par_state; |
| 1615 | |
| 1616 | scoped_restore restore_yydebug = make_scoped_restore (&yydebug, |
| 1617 | parser_debug); |
| 1618 | |
| 1619 | /* Initialize some state used by the lexer. */ |
| 1620 | last_was_structop = 0; |
| 1621 | saw_name_at_eof = 0; |
| 1622 | |
| 1623 | token_fifo.clear (); |
| 1624 | popping = 0; |
| 1625 | name_obstack.clear (); |
| 1626 | |
| 1627 | return yyparse (); |
| 1628 | } |
| 1629 | |
| 1630 | static void |
| 1631 | yyerror (const char *msg) |
| 1632 | { |
| 1633 | if (prev_lexptr) |
| 1634 | lexptr = prev_lexptr; |
| 1635 | |
| 1636 | error (_("A %s in expression, near `%s'."), msg, lexptr); |
| 1637 | } |
| 1638 | |