parsing_c/parser_c.mly

   1 %{
   2 (* Yoann Padioleau
   3  *
   4  * Copyright (C) 2002, 2006, 2007, 2008, 2009 Yoann Padioleau
   5  *
   6  * This program is free software; you can redistribute it and/or
   7  * modify it under the terms of the GNU General Public License (GPL)
   8  * version 2 as published by the Free Software Foundation.
   9  *
  10  * This program is distributed in the hope that it will be useful,
  11  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  12  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  13  * file license.txt for more details.
  14  *)
  15 open Common
  16
  17 open Ast_c
  18
  19 module LP = Lexer_parser
  20 open Lexer_parser (* for the fields *)
  21
  22 open Semantic_c (* Semantic exn *)
  23
  24
  25 (*****************************************************************************)
  26 (* Wrappers *)
  27 (*****************************************************************************)
  28 let warning s v =
  29   if !Flag_parsing_c.verbose_parsing
  30   then Common.warning ("PARSING: " ^ s) v
  31   else v
  32
  33 let pr2, pr2_once = Common.mk_pr2_wrappers Flag_parsing_c.verbose_parsing
  34
  35 (*****************************************************************************)
  36 (* Parse helpers functions *)
  37 (*****************************************************************************)
  38
  39 (*-------------------------------------------------------------------------- *)
  40 (* Type related *)
  41 (*-------------------------------------------------------------------------- *)
  42
  43 type shortLong      = Short  | Long | LongLong
  44
  45 type decl = {
  46   storageD: storagebis wrap;
  47   typeD: ((sign option) * (shortLong option) * (typeCbis option)) wrap;
  48   qualifD: typeQualifierbis wrap;
  49   inlineD: bool             wrap;
  50   (* note: have a full_info: parse_info list; to remember ordering
  51    * between storage, qualifier, type ? well this info is already in
  52    * the Ast_c.info, just have to sort them to get good order *)
  53 }
  54
  55 let nullDecl = {
  56   storageD = NoSto, [];
  57   typeD = (None, None, None), [];
  58   qualifD = nullQualif;
  59   inlineD = false, [];
  60 }
  61 let fake_pi = Common.fake_parse_info
  62
  63 let addStorageD  = function
  64   | ((x,ii), ({storageD = (NoSto,[])} as v)) -> { v with storageD = (x, [ii]) }
  65   | ((x,ii), ({storageD = (y, ii2)} as v)) ->
  66       if x =*= y then warning "duplicate storage classes" v
  67       else raise (Semantic ("multiple storage classes", fake_pi))
  68
  69 let addInlineD  = function
  70   | ((true,ii), ({inlineD = (false,[])} as v)) -> { v with inlineD=(true,[ii])}
  71   | ((true,ii), ({inlineD = (true, ii2)} as v)) -> warning "duplicate inline" v
  72   | _ -> raise Impossible
  73
  74
  75 let addTypeD     = function
  76   | ((Left3 Signed,ii)   ,({typeD = ((Some Signed,  b,c),ii2)} as v)) ->
  77       warning "duplicate 'signed'"   v
  78   | ((Left3 UnSigned,ii) ,({typeD = ((Some UnSigned,b,c),ii2)} as v)) ->
  79       warning "duplicate 'unsigned'" v
  80   | ((Left3 _,ii),        ({typeD = ((Some _,b,c),ii2)} as _v)) ->
  81       raise (Semantic ("both signed and unsigned specified", fake_pi))
  82   | ((Left3 x,ii),        ({typeD = ((None,b,c),ii2)} as v))   ->
  83       {v with typeD = (Some x,b,c),ii ++ ii2}
  84
  85   | ((Middle3 Short,ii),  ({typeD = ((a,Some Short,c),ii2)} as v)) ->
  86       warning "duplicate 'short'" v
  87
  88
  89   (* gccext: long long allowed *)
  90   | ((Middle3 Long,ii),   ({typeD = ((a,Some Long ,c),ii2)} as v)) ->
  91       { v with typeD = (a, Some LongLong, c),ii++ii2 }
  92   | ((Middle3 Long,ii),   ({typeD = ((a,Some LongLong ,c),ii2)} as v)) ->
  93       warning "triplicate 'long'" v
  94
  95
  96   | ((Middle3 _,ii),      ({typeD = ((a,Some _,c),ii2)} as _v)) ->
  97       raise (Semantic ("both long and short specified", fake_pi))
  98   | ((Middle3 x,ii),      ({typeD = ((a,None,c),ii2)} as v))  ->
  99       {v with typeD = (a, Some x,c),ii++ii2}
 100
 101   | ((Right3 t,ii),       ({typeD = ((a,b,Some x),ii2)} as _v)) ->
 102       raise (Semantic ((Printf.sprintf "two or more data types: t %s ii %s\ntypeD %s ii2 %s\n" (Dumper.dump t) (Dumper.dump ii) (Dumper.dump x) (Dumper.dump ii2)), fake_pi))
 103   | ((Right3 t,ii),       ({typeD = ((a,b,None),ii2)} as v))   ->
 104       {v with typeD = (a,b, Some t),ii++ii2}
 105
 106
 107 let addQualif = function
 108   | ({const=true},   ({const=true} as x)) ->   warning "duplicate 'const'" x
 109   | ({volatile=true},({volatile=true} as x))-> warning "duplicate 'volatile'" x
 110   | ({const=true},    v) -> {v with const=true}
 111   | ({volatile=true}, v) -> {v with volatile=true}
 112   | _ ->
 113       internal_error "there is no noconst or novolatile keyword"
 114
 115 let addQualifD ((qu,ii), ({qualifD = (v,ii2)} as x)) =
 116   { x with qualifD = (addQualif (qu, v),ii::ii2) }
 117
 118
 119 (*-------------------------------------------------------------------------- *)
 120 (* Declaration/Function related *)
 121 (*-------------------------------------------------------------------------- *)
 122
 123
 124 (* stdC: type section, basic integer types (and ritchie)
 125  * To understand the code, just look at the result (right part of the PM)
 126  * and go back.
 127  *)
 128 let (fixDeclSpecForDecl: decl -> (fullType * (storage wrap)))  = function
 129  {storageD = (st,iist);
 130   qualifD = (qu,iiq);
 131   typeD = (ty,iit);
 132   inlineD = (inline,iinl);
 133   } ->
 134    let ty',iit' =
 135    (match ty with
 136  | (None,None,None)       ->
 137      (* generate fake_info, otherwise type_annotater can crash in
 138       * offset.
 139       *)
 140      warning "type defaults to 'int'" (defaultInt, [fakeInfo fake_pi])
 141  | (None, None, Some t)   -> (t, iit)
 142
 143  | (Some sign,   None, (None| Some (BaseType (IntType (Si (_,CInt))))))  ->
 144      BaseType(IntType (Si (sign, CInt))), iit
 145  | ((None|Some Signed),Some x,(None|Some(BaseType(IntType (Si (_,CInt)))))) ->
 146      BaseType(IntType (Si (Signed, [Short,CShort; Long, CLong; LongLong, CLongLong] +> List.assoc x))), iit
 147  | (Some UnSigned, Some x, (None| Some (BaseType (IntType (Si (_,CInt))))))->
 148      BaseType(IntType (Si (UnSigned, [Short,CShort; Long, CLong; LongLong, CLongLong] +> List.assoc x))), iit
 149  | (Some sign,   None, (Some (BaseType (IntType CChar))))   ->
 150      BaseType(IntType (Si (sign, CChar2))), iit
 151  | (None, Some Long,(Some(BaseType(FloatType CDouble))))    ->
 152      BaseType (FloatType (CLongDouble)), iit
 153
 154  | (Some _,_, Some _) ->
 155      (*mine*)
 156      raise (Semantic ("signed, unsigned valid only for char and int", fake_pi))
 157  | (_,Some _,(Some(BaseType(FloatType (CFloat|CLongDouble))))) ->
 158      raise (Semantic ("long or short specified with floatint type", fake_pi))
 159  | (_,Some Short,(Some(BaseType(FloatType CDouble)))) ->
 160      raise (Semantic ("the only valid combination is long double", fake_pi))
 161
 162  | (_, Some _, Some _) ->
 163      (* mine *)
 164      raise (Semantic ("long, short valid only for int or float", fake_pi))
 165
 166      (* if do short uint i, then gcc say parse error, strange ? it is
 167       * not a parse error, it is just that we dont allow with typedef
 168       * either short/long or signed/unsigned. In fact, with
 169       * parse_typedef_fix2 (with et() and dt()) now I say too parse
 170       * error so this code is executed only when do short struct
 171       * {....} and never with a typedef cos now we parse short uint i
 172       * as short ident ident => parse error (cos after first short i
 173       * pass in dt() mode) *)
 174
 175    )
 176    in
 177    ((qu, iiq),
 178    (ty', iit'))
 179      ,((st, inline),iist++iinl)
 180
 181
 182 let fixDeclSpecForParam = function ({storageD = (st,iist)} as r) ->
 183   let ((qu,ty) as v,_st) = fixDeclSpecForDecl r in
 184   match st with
 185   | (Sto Register) -> (v, true), iist
 186   | NoSto -> (v, false), iist
 187   | _ ->
 188       raise
 189         (Semantic ("storage class specified for parameter of function",
 190                   fake_pi))
 191
 192 let fixDeclSpecForMacro = function ({storageD = (st,iist)} as r) ->
 193   let ((qu,ty) as v,_st) = fixDeclSpecForDecl r in
 194   match st with
 195   | NoSto -> v
 196   | _ ->
 197       raise
 198         (Semantic ("storage class specified for macro type decl",
 199                   fake_pi))
 200
 201
 202 let fixDeclSpecForFuncDef x =
 203   let (returnType,storage) = fixDeclSpecForDecl x in
 204   (match fst (unwrap storage) with
 205   | StoTypedef ->
 206       raise (Semantic ("function definition declared 'typedef'", fake_pi))
 207   | _ -> (returnType, storage)
 208   )
 209
 210
 211 (* parameter: (this is the context where we give parameter only when
 212  * in func DEFINITION not in funct DECLARATION) We must have a name.
 213  * This function ensure that we give only parameterTypeDecl with well
 214  * formed Classic constructor todo?: do we accept other declaration
 215  * in ? so I must add them to the compound of the deffunc. I dont
 216  * have to handle typedef pb here cos C forbid to do VF f { ... }
 217  * with VF a typedef of func cos here we dont see the name of the
 218  * argument (in the typedef)
 219  *)
 220 let (fixOldCDecl: fullType -> fullType) = fun ty ->
 221   match Ast_c.unwrap_typeC ty with
 222   | FunctionType (fullt, (params, (b, iib))) ->
 223
 224       (* stdC: If the prototype declaration declares a parameter for a
 225        * function that you are defining (it is part of a function
 226        * definition), then you must write a name within the declarator.
 227        * Otherwise, you can omit the name. *)
 228       (match params with
 229       | [{p_namei = None; p_type = ty2},_] ->
 230           (match Ast_c.unwrap_typeC ty2 with
 231           | BaseType Void ->
 232               ty
 233           | _ ->
 234               pr2_once ("SEMANTIC:parameter name omitted, but I continue");
 235               ty
 236           )
 237
 238       | params ->
 239           (params +> List.iter (fun (param,_) ->
 240             match param with
 241             | {p_namei = None} ->
 242               (* if majuscule, then certainly macro-parameter *)
 243                 pr2_once ("SEMANTIC:parameter name omitted, but I continue");
 244             | _ -> ()
 245           ));
 246           ty
 247       )
 248
 249         (* todo? can we declare prototype in the decl or structdef,
 250            ... => length <> but good kan meme *)
 251   | _ ->
 252       (* gcc say parse error but dont see why *)
 253       raise (Semantic ("seems this is not a function", fake_pi))
 254
 255
 256 let fixFunc (typ, compound, old_style_opt) =
 257   let (cp,iicp) = compound in
 258
 259   let (name, ty,   (st,iist),  attrs) = typ in
 260
 261   let (qu, tybis) = ty in
 262
 263   match Ast_c.unwrap_typeC ty with
 264   | FunctionType (fullt, (params,abool)) ->
 265       let iifunc = Ast_c.get_ii_typeC_take_care tybis in
 266
 267       let iistart = Ast_c.fakeInfo () in
 268       assert (qu =*= nullQualif);
 269
 270       (match params with
 271       | [{p_namei= None; p_type = ty2}, _] ->
 272           (match Ast_c.unwrap_typeC ty2 with
 273           | BaseType Void ->  ()
 274           | _ ->
 275                 (* failwith "internal errror: fixOldCDecl not good" *)
 276               ()
 277           )
 278       | params ->
 279           params +> List.iter (function
 280           | ({p_namei = Some s}, _) -> ()
 281           | _ -> ()
 282                 (* failwith "internal errror: fixOldCDecl not good" *)
 283           )
 284       );
 285       (* bugfix: cf tests_c/function_pointer4.c.
 286        * Apparemment en C on peut syntaxiquement ecrire ca:
 287        *
 288        *   void a(int)(int x);
 289        * mais apres gcc gueule au niveau semantique avec:
 290        *   xxx.c:1: error: 'a' declared as function returning a function
 291        * Je ne faisais pas cette verif. Sur du code comme
 292        *   void METH(foo)(int x) { ...} , le parser croit (a tort) que foo
 293        * est un typedef, et donc c'est parsé comme l'exemple precedent,
 294        * ce qui ensuite confuse l'unparser qui n'est pas habitué
 295        * a avoir dans le returnType un FunctionType et qui donc
 296        * pr_elem les ii dans le mauvais sens ce qui genere au final
 297        * une exception. Hence this fix to at least detect the error
 298        * at parsing time (not unparsing time).
 299        *)
 300       (match Ast_c.unwrap_typeC fullt with
 301       | FunctionType _ ->
 302           let s = Ast_c.str_of_name name in
 303           let iis = Ast_c.info_of_name name in
 304           pr2 (spf "WEIRD: %s declared as function returning a function." s);
 305           pr2 (spf "This is probably because of a macro. Extend standard.h");
 306           raise (Semantic (spf "error: %s " s, Ast_c.parse_info_of_info iis))
 307       | _ -> ()
 308       );
 309
 310       (* it must be nullQualif,cos parser construct only this*)
 311       {f_name = name;
 312        f_type = (fullt, (params, abool));
 313        f_storage = st;
 314        f_body = cp;
 315        f_attr = attrs;
 316        f_old_c_style = old_style_opt;
 317       },
 318       (iifunc++iicp++[iistart]++iist)
 319   | _ ->
 320       raise
 321         (Semantic
 322             ("you are trying to do a function definition but you dont give " ^
 323              "any parameter", fake_pi))
 324
 325
 326 (*-------------------------------------------------------------------------- *)
 327 (* parse_typedef_fix2 *)
 328 (*-------------------------------------------------------------------------- *)
 329
 330 let dt s () =
 331   if !Flag_parsing_c.debug_etdt then pr2 ("<" ^ s);
 332   LP.disable_typedef ()
 333
 334 let et s () =
 335   if !Flag_parsing_c.debug_etdt then pr2 (">" ^ s);
 336   LP.enable_typedef ()
 337
 338
 339 let fix_add_params_ident x =
 340   let (s, ty, st, _attrs) = x in
 341   match Ast_c.unwrap_typeC ty with
 342   | FunctionType (fullt, (params, bool)) ->
 343
 344       (match params with
 345       | [{p_namei=None; p_type=ty2}, _] ->
 346           (match Ast_c.unwrap_typeC ty2 with
 347           | BaseType Void -> ()
 348           | _ ->
 349               (* failwith "internal errror: fixOldCDecl not good" *)
 350               ()
 351           )
 352       | params ->
 353           params +> List.iter (function
 354           | ({p_namei= Some name}, _) ->
 355               LP.add_ident (Ast_c.str_of_name s)
 356           | _ ->
 357               ()
 358                 (* failwith "internal errror: fixOldCDecl not good" *)
 359           )
 360       )
 361   | _ -> ()
 362
 363
 364
 365 (*-------------------------------------------------------------------------- *)
 366 (* shortcuts *)
 367 (*-------------------------------------------------------------------------- *)
 368
 369 let mk_e e ii = Ast_c.mk_e e ii
 370
 371 let mk_string_wrap (s,info) = (s, [info])
 372
 373 %}
 374
 375 /*(*****************************************************************************)*/
 376 /*(* Tokens *)*/
 377 /*(*************************************************************************)*/
 378
 379 /*
 380 (*
 381  * Some tokens are not even used in this file because they are filtered
 382  * in some intermediate phase. But they still must be declared because
 383  * ocamllex may generate them, or some intermediate phase may also
 384  * generate them (like some functions in parsing_hacks.ml)
 385  *)
 386 */
 387
 388 %token <Ast_c.info> TUnknown /*(* unrecognized token *)*/
 389
 390 /*(* coupling: Token_helpers.is_real_comment *)*/
 391 %token <Ast_c.info> TCommentSpace TCommentNewline   TComment
 392
 393 /*(*-----------------------------------------*)*/
 394 /*(* the normal tokens *)*/
 395 /*(*-----------------------------------------*)*/
 396
 397 %token <(string * (Ast_c.sign * Ast_c.base)) * Ast_c.info> TInt
 398 %token <(string * Ast_c.floatType) * Ast_c.info> TFloat
 399 %token <(string * Ast_c.isWchar) * Ast_c.info>   TChar
 400 %token <(string * Ast_c.isWchar) * Ast_c.info>   TString
 401
 402 %token <string * Ast_c.info> TIdent
 403 %token <string * Ast_c.info> TKRParam
 404 %token <string * Ast_c.info> Tconstructorname /* parsing_hack for c++ */
 405 /*(* appears mostly after some fix_xxx in parsing_hack *)*/
 406 %token <string * Ast_c.info> TypedefIdent
 407
 408
 409 /*
 410 (* Some tokens like TOPar and TCPar are used as synchronisation stuff,
 411  * in parsing_hack.ml. So if define special tokens like TOParDefine and
 412  * TCParEOL, then take care to also modify in Token_helpers.
 413  *)
 414 */
 415
 416 %token <Ast_c.info> TOPar TCPar TOBrace TCBrace TOCro TCCro
 417 %token <Ast_c.info> TDot TComma TPtrOp
 418 %token <Ast_c.info> TInc TDec
 419 %token <Ast_c.assignOp * Ast_c.info> TAssign
 420 %token <Ast_c.info> TEq
 421 %token <Ast_c.info> TWhy  TTilde TBang
 422 %token <Ast_c.info> TEllipsis
 423 %token <Ast_c.info> TDotDot
 424
 425 %token <Ast_c.info> TPtVirg
 426 %token <Ast_c.info>
 427        TOrLog TAndLog TOr TXor TAnd  TEqEq TNotEq TInf TSup TInfEq TSupEq
 428        TShl TShr
 429        TPlus TMinus TMul TDiv TMod
 430
 431 %token <Ast_c.info>
 432        Tchar Tshort Tint Tdouble Tfloat Tlong Tunsigned Tsigned Tvoid
 433        Tsize_t Tssize_t Tptrdiff_t
 434        Tauto Tregister Textern Tstatic
 435        Ttypedef
 436        Tconst Tvolatile
 437        Tstruct Tunion Tenum
 438        Tbreak Telse Tswitch Tcase Tcontinue Tfor Tdo Tif  Twhile Treturn
 439        Tgoto Tdefault
 440        Tsizeof Tnew Tdelete TOParCplusplusInit
 441
 442 /*(* C99 *)*/
 443 %token <Ast_c.info>
 444        Trestrict
 445
 446 /*(*-----------------------------------------*)*/
 447 /*(* gccext: extra tokens *)*/
 448 /*(*-----------------------------------------*)*/
 449 %token <Ast_c.info> Tasm
 450 %token <Ast_c.info> Tattribute
 451 %token <Ast_c.info> TattributeNoarg
 452 %token <Ast_c.info> Tinline
 453 %token <Ast_c.info> Ttypeof
 454
 455 /*(*-----------------------------------------*)*/
 456 /*(* cppext: extra tokens *)*/
 457 /*(*-----------------------------------------*)*/
 458 /*(* coupling with Token_helpers.is_cpp_token *)*/
 459
 460
 461 /*(*---------------*)*/
 462 /*(* define        *)*/
 463 /*(*---------------*)*/
 464
 465 %token <Ast_c.info> TDefine
 466 %token <(string * Ast_c.info)> TDefParamVariadic
 467
 468 /*(* disappear after fix_tokens_define *)*/
 469 %token <Ast_c.info> TCppEscapedNewline
 470
 471 %token <Ast_c.info> TCppConcatOp
 472
 473 /*(* appear    after fix_tokens_define *)*/
 474 %token <Ast_c.info> TOParDefine
 475 %token <Ast_c.info> TOBraceDefineInit
 476
 477 %token <(string * Ast_c.info)> TIdentDefine /*(* same *)*/
 478 %token <Ast_c.info>            TDefEOL      /*(* same *)*/
 479
 480
 481 /*(*---------------*)*/
 482 /*(* include       *)*/
 483 /*(*---------------*)*/
 484
 485
 486 /*(* used only in lexer_c, then transformed in comment or splitted in tokens *)*/
 487 %token <(string * string * bool ref * Ast_c.info)> TInclude
 488
 489 /*(* tokens coming from above, generated in parse_c from TInclude, etc *)*/
 490 %token <(Ast_c.info * bool ref)> TIncludeStart
 491 %token <(string * Ast_c.info)> TIncludeFilename
 492
 493
 494 /*(*---------------*)*/
 495 /*(* ifdef         *)*/
 496 /*(*---------------*)*/
 497
 498 /*(* coupling: Token_helpers.is_cpp_instruction *)*/
 499 %token <((int * int) option ref * Ast_c.info)>
 500   TIfdef TIfdefelse TIfdefelif TEndif
 501 %token <(bool * (int * int) option ref * Ast_c.info)>
 502   TIfdefBool TIfdefMisc TIfdefVersion
 503
 504 /*(*---------------*)*/
 505 /*(* other         *)*/
 506 /*(*---------------*)*/
 507
 508 %token <Ast_c.info> TUndef
 509
 510 %token <Ast_c.info> TCppDirectiveOther
 511
 512 /*(*---------------*)*/
 513 /*(* macro use     *)*/
 514 /*(*---------------*)*/
 515
 516 /*(* appear  after fix_tokens_cpp, cf also parsing_hacks#hint *)*/
 517
 518 %token <(string * Ast_c.info)>            TMacroAttr
 519 %token <(string * Ast_c.info)>            TMacroStmt
 520 %token <(string * Ast_c.info)> TMacroIdentBuilder
 521 /*(* no need value for the moment *)*/
 522 %token <(string * Ast_c.info)>            TMacroString
 523 %token <(string * Ast_c.info)> TMacroDecl
 524 %token <Ast_c.info>            TMacroDeclConst
 525
 526 %token <(string * Ast_c.info)> TMacroIterator
 527 /*(*
 528 %token <(string * Ast_c.info)> TMacroTop
 529 %token <(string * Ast_c.info)> TMacroStructDecl
 530 *)*/
 531
 532 %token <(string * Ast_c.info)>            TMacroAttrStorage
 533
 534
 535 /*(*---------------*)*/
 536 /*(* other         *)*/
 537 /*(*---------------*)*/
 538
 539
 540 /*(* should disappear after parsing_hack *)*/
 541 %token <Ast_c.info> TCommentSkipTagStart TCommentSkipTagEnd
 542
 543
 544 /*(* appear after parsing_hack *)*/
 545 %token <Ast_c.info> TCParEOL
 546
 547 %token <Ast_c.info> TAction
 548
 549
 550 /*(* TCommentMisc still useful ? obsolete ? *)*/
 551 %token <Ast_c.info> TCommentMisc
 552 %token <(Token_c.cppcommentkind * Ast_c.info)> TCommentCpp
 553
 554
 555 /*(*-----------------------------------------*)*/
 556 %token <Ast_c.info> EOF
 557
 558 /*(*-----------------------------------------*)*/
 559
 560 /*(* must be at the top so that it has the lowest priority *)*/
 561 %nonassoc SHIFTHERE
 562
 563 %nonassoc Telse
 564
 565 %left TOrLog
 566 %left TAndLog
 567 %left TOr
 568 %left TXor
 569 %left TAnd
 570 %left TEqEq TNotEq
 571 %left TInf TSup TInfEq TSupEq
 572 %left TShl TShr
 573 %left TPlus TMinus
 574 %left TMul TDiv TMod
 575
 576 /*(*************************************************************************)*/
 577 /*(* Rules type declaration *)*/
 578 /*(*************************************************************************)*/
 579
 580 %start main celem statement expr type_name
 581 %type <Ast_c.program> main
 582 %type <Ast_c.toplevel> celem
 583
 584 %type <Ast_c.statement> statement
 585 %type <Ast_c.expression> expr
 586 %type <Ast_c.fullType> type_name
 587
 588 %%
 589 /*(*************************************************************************)*/
 590 /*
 591 (* TOC:
 592  * toplevel (obsolete)
 593  *
 594  * ident
 595  * expression
 596  * statement
 597  * types with
 598  *   - left part (type_spec, qualif),
 599  *   - right part (declarator, abstract declarator)
 600  *   - aux part (parameters)
 601  * declaration, storage, initializers
 602  * struct
 603  * enum
 604  * cpp directives
 605  * celem (=~ main)
 606  *
 607  * generic workarounds (obrace, cbrace for context setting)
 608  * xxx_list, xxx_opt
 609  *)
 610 */
 611 /*(*************************************************************************)*/
 612
 613 /*(*************************************************************************)*/
 614 /*(* toplevel *)*/
 615 /*(*************************************************************************)*/
 616 /*(* no more used; now that use error recovery *)*/
 617
 618 main:
 619  translation_unit EOF     { $1 }
 620
 621 translation_unit:
 622  | external_declaration
 623      { !LP._lexer_hint.context_stack <- [LP.InTopLevel];   [$1] }
 624  | translation_unit external_declaration
 625      { !LP._lexer_hint.context_stack <- [LP.InTopLevel]; $1 ++ [$2] }
 626
 627
 628
 629 /*(*************************************************************************)*/
 630 /*(* ident *)*/
 631 /*(*************************************************************************)*/
 632
 633 /*(* Why this ? Why not s/ident/TIdent ? cos there is multiple namespaces in C,
 634    * so a label can have the same name that a typedef, same for field and tags
 635    * hence sometimes the use of ident  instead of TIdent.
 636    *)*/
 637 ident:
 638  | TIdent       { $1 }
 639  | TypedefIdent { $1 }
 640
 641
 642 identifier:
 643  | TIdent       { $1 }
 644
 645 /*
 646 (* cppext: string concatenation of idents
 647  * also cppext: gccext: ##args for variadic macro
 648  *)
 649 */
 650 identifier_cpp:
 651  | TIdent
 652      { RegularName (mk_string_wrap $1) }
 653  | ident_extra_cpp { $1 }
 654
 655 ident_cpp:
 656  | TIdent
 657      { RegularName (mk_string_wrap $1) }
 658  | TypedefIdent
 659      { RegularName (mk_string_wrap $1) }
 660  | ident_extra_cpp { $1 }
 661
 662 ident_extra_cpp:
 663  | TIdent TCppConcatOp identifier_cpp_list
 664      {
 665        CppConcatenatedName (
 666          match $3 with
 667          | [] -> raise Impossible
 668          | (x,concatnull)::xs ->
 669              assert(null concatnull);
 670              (mk_string_wrap $1, [])::(x,[$2])::xs
 671        )
 672    }
 673  | TCppConcatOp TIdent
 674      { CppVariadicName (fst $2, [$1; snd $2]) }
 675  | TMacroIdentBuilder TOPar param_define_list TCPar
 676      { CppIdentBuilder ((fst $1, [snd $1;$2;$4]), $3) }
 677
 678 identifier_cpp_list:
 679  | TIdent { [mk_string_wrap $1, []] }
 680  | identifier_cpp_list TCppConcatOp TIdent { $1 ++ [mk_string_wrap $3, [$2]] }
 681
 682 /*(*************************************************************************)*/
 683 /*(* expr *)*/
 684 /*(*************************************************************************)*/
 685
 686 expr:
 687  | assign_expr             { $1 }
 688  | expr TComma assign_expr { mk_e (Sequence ($1,$3)) [$2] }
 689
 690 /*(* bugfix: in C grammar they put unary_expr, but in fact it must be
 691    * cast_expr, otherwise (int * ) xxx = &yy; is not allowed
 692    *)*/
 693 assign_expr:
 694  | cond_expr                     { $1 }
 695  | cast_expr TAssign assign_expr { mk_e(Assignment ($1,fst $2,$3)) [snd $2]}
 696  | cast_expr TEq     assign_expr { mk_e(Assignment ($1,SimpleAssign,$3)) [$2]}
 697
 698 /*(* gccext: allow optional then part hence gcc_opt_expr
 699    * bugfix: in C grammar they put TDotDot cond_expr, but in fact it must be
 700    * assign_expr, otherwise   pnp ? x : x = 0x388  is not allowed
 701    *)*/
 702 cond_expr:
 703  | arith_expr
 704      { $1 }
 705  | arith_expr TWhy gcc_opt_expr TDotDot assign_expr
 706      { mk_e (CondExpr ($1,$3,$5)) [$2;$4] }
 707
 708
 709 arith_expr:
 710  | cast_expr                     { $1 }
 711  | arith_expr TMul    arith_expr { mk_e(Binary ($1, Arith Mul,      $3)) [$2] }
 712  | arith_expr TDiv    arith_expr { mk_e(Binary ($1, Arith Div,      $3)) [$2] }
 713  | arith_expr TMod    arith_expr { mk_e(Binary ($1, Arith Mod,      $3)) [$2] }
 714  | arith_expr TPlus   arith_expr { mk_e(Binary ($1, Arith Plus,     $3)) [$2] }
 715  | arith_expr TMinus  arith_expr { mk_e(Binary ($1, Arith Minus,    $3)) [$2] }
 716  | arith_expr TShl    arith_expr { mk_e(Binary ($1, Arith DecLeft,  $3)) [$2] }
 717  | arith_expr TShr    arith_expr { mk_e(Binary ($1, Arith DecRight, $3)) [$2] }
 718  | arith_expr TInf    arith_expr { mk_e(Binary ($1, Logical Inf,    $3)) [$2] }
 719  | arith_expr TSup    arith_expr { mk_e(Binary ($1, Logical Sup,    $3)) [$2] }
 720  | arith_expr TInfEq  arith_expr { mk_e(Binary ($1, Logical InfEq,  $3)) [$2] }
 721  | arith_expr TSupEq  arith_expr { mk_e(Binary ($1, Logical SupEq,  $3)) [$2] }
 722  | arith_expr TEqEq   arith_expr { mk_e(Binary ($1, Logical Eq,     $3)) [$2] }
 723  | arith_expr TNotEq  arith_expr { mk_e(Binary ($1, Logical NotEq,  $3)) [$2] }
 724  | arith_expr TAnd    arith_expr { mk_e(Binary ($1, Arith And,      $3)) [$2] }
 725  | arith_expr TOr     arith_expr { mk_e(Binary ($1, Arith Or,       $3)) [$2] }
 726  | arith_expr TXor    arith_expr { mk_e(Binary ($1, Arith Xor,      $3)) [$2] }
 727  | arith_expr TAndLog arith_expr { mk_e(Binary ($1, Logical AndLog, $3)) [$2] }
 728  | arith_expr TOrLog  arith_expr { mk_e(Binary ($1, Logical OrLog,  $3)) [$2] }
 729
 730 cast_expr:
 731  | unary_expr                        { $1 }
 732  | topar2 type_name tcpar2 cast_expr { mk_e(Cast ($2, $4)) [$1;$3] }
 733
 734 unary_expr:
 735  | postfix_expr                    { $1 }
 736  | TInc unary_expr                 { mk_e(Infix ($2, Inc))    [$1] }
 737  | TDec unary_expr                 { mk_e(Infix ($2, Dec))    [$1] }
 738  | unary_op cast_expr              { mk_e(Unary ($2, fst $1)) [snd $1] }
 739  | Tsizeof unary_expr              { mk_e(SizeOfExpr ($2))    [$1] }
 740  | Tsizeof topar2 type_name tcpar2 { mk_e(SizeOfType ($3))    [$1;$2;$4] }
 741  | Tnew new_argument               { mk_e(New $2)             [$1] }
 742  | Tdelete cast_expr               { mk_e(Delete $2)          [$1] }
 743
 744 new_argument:
 745  | TIdent TOPar argument_list_ne TCPar
 746      { let fn = mk_e(Ident (RegularName (mk_string_wrap $1))) [] in
 747        Left (mk_e(FunCall (fn, $3)) [$2;$4]) }
 748  | TIdent TOPar TCPar
 749      { let fn = mk_e(Ident (RegularName (mk_string_wrap $1))) [] in
 750        Left(mk_e(FunCall (fn, [])) [$2;$3]) }
 751  | TypedefIdent TOPar argument_list_ne TCPar
 752      { let fn = mk_e(Ident (RegularName (mk_string_wrap $1))) [] in
 753        Left (mk_e(FunCall (fn, $3)) [$2;$4]) }
 754  | TypedefIdent TOPar TCPar
 755      { let fn = mk_e(Ident (RegularName (mk_string_wrap $1))) [] in
 756        Left (mk_e(FunCall (fn, [])) [$2;$3]) }
 757  | type_spec
 758      { let ty = addTypeD ($1,nullDecl) in
 759        let ((returnType,hasreg), iihasreg) = fixDeclSpecForParam ty in
 760        Right (ArgType { p_namei = None; p_type = returnType;
 761                         p_register = hasreg, iihasreg;
 762                       } )
 763      }
 764  | new_argument TOCro expr TCCro
 765      {
 766        match $1 with
 767          Left(e) -> Left(mk_e(ArrayAccess (e, $3)) [$2;$4])
 768        | Right(ArgType(ty)) -> (* lots of hacks to make the right type *)
 769            let fty = mk_ty (Array (Some $3, ty.Ast_c.p_type)) [$2;$4] in
 770            let pty = { ty with p_type = fty } in
 771            Right(ArgType pty)
 772        | _ -> raise Impossible
 773      }
 774
 775 unary_op:
 776  | TAnd   { GetRef,     $1 }
 777  | TMul   { DeRef,      $1 }
 778  | TPlus  { UnPlus,     $1 }
 779  | TMinus { UnMinus,    $1 }
 780  | TTilde { Tilde,      $1 }
 781  | TBang  { Not,        $1 }
 782  /*(* gccext: have that a lot in old kernel to get address of local label.
 783     * cf gcc manual "local labels as values".
 784     *)*/
 785  | TAndLog { GetRefLabel, $1 }
 786
 787 postfix_expr:
 788  | primary_expr               { $1 }
 789  | postfix_expr TOCro expr TCCro
 790      { mk_e(ArrayAccess ($1, $3)) [$2;$4] }
 791  | postfix_expr TOPar argument_list_ne TCPar
 792      { mk_e(FunCall ($1, $3)) [$2;$4] }
 793  | postfix_expr TOPar  TCPar  { mk_e(FunCall ($1, [])) [$2;$3] }
 794  | postfix_expr TDot   ident_cpp { mk_e(RecordAccess   ($1,$3)) [$2] }
 795  | postfix_expr TPtrOp ident_cpp { mk_e(RecordPtAccess ($1,$3)) [$2] }
 796  | postfix_expr TInc          { mk_e(Postfix ($1, Inc)) [$2] }
 797  | postfix_expr TDec          { mk_e(Postfix ($1, Dec)) [$2] }
 798
 799  /*(* gccext: also called compound literals *)*/
 800  | topar2 type_name tcpar2 TOBrace TCBrace
 801      { mk_e(Constructor ($2, (InitList [], [$4;$5]))) [$1;$3] }
 802  | topar2 type_name tcpar2 TOBrace initialize_list gcc_comma_opt TCBrace
 803      { mk_e(Constructor ($2, (InitList (List.rev $5),[$4;$7]++$6))) [$1;$3] }
 804
 805
 806 primary_expr:
 807  | identifier_cpp  { mk_e(Ident  ($1)) [] }
 808  | TInt
 809     { let (str,(sign,base)) = fst $1 in
 810       mk_e(Constant (Int (str,Si(sign,base)))) [snd $1] }
 811  | TFloat  { mk_e(Constant (Float  (fst $1))) [snd $1] }
 812  | TString { mk_e(Constant (String (fst $1))) [snd $1] }
 813  | TChar   { mk_e(Constant (Char   (fst $1))) [snd $1] }
 814  | TOPar expr TCPar { mk_e(ParenExpr ($2)) [$1;$3] }  /*(* forunparser: *)*/
 815
 816  /*(* gccext: cppext: TODO better ast ? *)*/
 817  | TMacroString { mk_e(Constant (MultiString [fst $1])) [snd $1] }
 818  | string_elem string_list
 819      { mk_e(Constant (MultiString ["TODO: MultiString"])) ($1 ++ $2) }
 820
 821  /*(* gccext: allow statement as expressions via ({ statement }) *)*/
 822  | TOPar compound TCPar  { mk_e(StatementExpr ($2)) [$1;$3] }
 823
 824
 825
 826 /*(*----------------------------*)*/
 827 /*(* cppext: *)*/
 828 /*(*----------------------------*)*/
 829
 830 /*(* cppext: *)*/
 831 /*(* to avoid conflicts have to introduce a _not_empty (ne) version *)*/
 832 argument_ne:
 833  | assign_expr { Left $1 }
 834  | parameter_decl { Right (ArgType $1)  }
 835  | action_higherordermacro_ne { Right (ArgAction $1) }
 836
 837
 838 argument:
 839  | assign_expr { Left $1 }
 840  | parameter_decl { Right (ArgType $1)  }
 841  /*(* had conflicts before, but julia fixed them *)*/
 842  | action_higherordermacro { Right (ArgAction $1) }
 843
 844 action_higherordermacro_ne:
 845  | taction_list_ne
 846      { if null $1
 847        then ActMisc [Ast_c.fakeInfo()]
 848        else ActMisc $1
 849      }
 850
 851
 852 action_higherordermacro:
 853  | taction_list
 854      { if null $1
 855        then ActMisc [Ast_c.fakeInfo()]
 856        else ActMisc $1
 857      }
 858
 859
 860 /*(*----------------------------*)*/
 861 /*(* workarounds *)*/
 862 /*(*----------------------------*)*/
 863
 864 /*(* would like evalInt $1 but require too much info *)*/
 865 const_expr: cond_expr { $1  }
 866
 867
 868 topar2: TOPar { et "topar2" (); $1  }
 869 tcpar2: TCPar { et "tcpar2" (); $1 (*TODO? et ? sure ? c pas dt plutot ? *) }
 870
 871
 872
 873 /*(*************************************************************************)*/
 874 /*(* statement *)*/
 875 /*(*************************************************************************)*/
 876
 877 statement: statement2 { mk_st (fst $1) (snd $1) }
 878
 879 statement2:
 880  | labeled         { Labeled      (fst $1), snd $1 }
 881  | compound        { Compound     (fst $1), snd $1 }
 882  | expr_statement  { ExprStatement(fst $1), snd $1 }
 883  | selection       { Selection    (fst $1), snd $1 ++ [fakeInfo()] }
 884  | iteration       { Iteration    (fst $1), snd $1 ++ [fakeInfo()] }
 885  | jump TPtVirg    { Jump         (fst $1), snd $1 ++ [$2] }
 886
 887  /*(* gccext: *)*/
 888  | Tasm TOPar asmbody TCPar TPtVirg             { Asm $3, [$1;$2;$4;$5] }
 889  | Tasm Tvolatile TOPar asmbody TCPar TPtVirg   { Asm $4, [$1;$2;$3;$5;$6] }
 890
 891  /*(* cppext: *)*/
 892  | TMacroStmt { MacroStmt, [snd $1] }
 893
 894
 895
 896
 897 /*(* note that case 1: case 2: i++;    would be correctly parsed, but with
 898    * a Case  (1, (Case (2, i++)))  :(
 899    *)*/
 900 labeled:
 901  | ident_cpp        TDotDot sw_stat_or_decl   { Label ($1, $3),  [$2] }
 902  | Tcase const_expr TDotDot sw_stat_or_decl   { Case ($2, $4),       [$1; $3] }
 903  | Tcase const_expr TEllipsis const_expr TDotDot sw_stat_or_decl
 904      { CaseRange ($2, $4, $6), [$1;$3;$5] } /*(* gccext: allow range *)*/
 905  | Tdefault    TDotDot sw_stat_or_decl   { Default $3,             [$1; $2] }
 906
 907 sw_stat_or_decl:
 908  | decl      { mk_st (Decl ($1 Ast_c.LocalDecl)) Ast_c.noii }
 909  | statement { $1 }
 910
 911
 912 end_labeled:
 913  /*(* gccext:  allow toto: }
 914     * was generating each 30 shift/Reduce conflicts,
 915     * mais ca va, ca fait ce qu'il faut.
 916     * update: julia fixed the problem by introducing end_labeled
 917     * and modifying below stat_or_decl_list
 918     *)*/
 919  | ident_cpp            TDotDot
 920      { Label ($1, (mk_st (ExprStatement None) Ast_c.noii)), [$2] }
 921  | Tcase const_expr TDotDot
 922      { Case ($2, (mk_st (ExprStatement None) Ast_c.noii)), [$1;$3] }
 923  | Tdefault         TDotDot
 924      { Default (mk_st (ExprStatement None) Ast_c.noii),    [$1; $2] }
 925
 926
 927
 928
 929
 930 compound: tobrace compound2 tcbrace { $2, [$1; $3]  }
 931
 932
 933 /*
 934 (* cppext: because of cpp, some stuff looks like declaration but are in
 935  * fact statement but too hard to figure out, and if parse them as
 936  * expression, then we force to have first decls and then exprs, then
 937  * will have a parse error. So easier to let mix decl/statement.
 938  * Moreover it helps to not make such a difference between decl and
 939  * statement for further coccinelle phases to factorize code.
 940 *)*/
 941 compound2:
 942  |                   { ([]) }
 943  | stat_or_decl_list { $1 }
 944
 945
 946 stat_or_decl_list:
 947  | stat_or_decl                   { [$1] }
 948  /*(* gccext: to avoid conflicts, cf end_labeled above *)*/
 949  | end_labeled  { [StmtElem (mk_st (Labeled  (fst $1)) (snd $1))] }
 950  /*(* old: conflicts | stat_or_decl_list stat_or_decl { $1 ++ [$2] } *)*/
 951  | stat_or_decl stat_or_decl_list { $1 :: $2 }
 952
 953 stat_or_decl:
 954  | decl      { StmtElem (mk_st (Decl ($1 Ast_c.LocalDecl)) Ast_c.noii) }
 955  | statement { StmtElem $1 }
 956
 957   /*(* gccext: *)*/
 958  | function_definition { StmtElem (mk_st (NestedFunc $1) Ast_c.noii) }
 959
 960  /* (* cppext: *)*/
 961  | cpp_directive
 962      { CppDirectiveStmt $1 }
 963  | cpp_ifdef_directive/*(* stat_or_decl_list ...*)*/
 964      { IfdefStmt $1 }
 965
 966
 967 expr_statement:
 968  | TPtVirg      { None,    [$1] }
 969  | expr TPtVirg { Some $1, [$2] }
 970
 971 selection:
 972  | Tif TOPar expr TCPar statement              %prec SHIFTHERE
 973      { If ($3, $5, (mk_st (ExprStatement None) Ast_c.noii)),   [$1;$2;$4] }
 974  | Tif TOPar expr TCPar statement Telse statement
 975      { If ($3, $5, $7),  [$1;$2;$4;$6] }
 976  | Tswitch TOPar expr TCPar statement
 977      { Switch ($3,$5),   [$1;$2;$4]  }
 978
 979 iteration:
 980  | Twhile TOPar expr TCPar statement
 981      { While ($3,$5),                [$1;$2;$4] }
 982  | Tdo statement Twhile TOPar expr TCPar TPtVirg
 983      { DoWhile ($2,$5),              [$1;$3;$4;$6;$7] }
 984  | Tfor TOPar expr_statement expr_statement TCPar statement
 985      { For ($3,$4,(None, []),$6),    [$1;$2;$5]}
 986  | Tfor TOPar expr_statement expr_statement expr TCPar statement
 987      { For ($3,$4,(Some $5, []),$7), [$1;$2;$6] }
 988  /*(* c++ext: for(int i = 0; i < n; i++)*)*/
 989  | Tfor TOPar decl expr_statement expr_opt TCPar statement
 990      {
 991        (* pr2 "DECL in for"; *)
 992        MacroIteration ("toto", [], $7),[$1;$2;$6] (* TODOfake ast, TODO need decl2 ? *)
 993      }
 994  /*(* cppext: *)*/
 995  | TMacroIterator TOPar argument_list_ne TCPar statement
 996      { MacroIteration (fst $1, $3, $5), [snd $1;$2;$4] }
 997  | TMacroIterator TOPar TCPar statement
 998      { MacroIteration (fst $1, [], $4), [snd $1;$2;$3] }
 999
1000 /*(* the ';' in the caller grammar rule will be appended to the infos *)*/
1001 jump:
1002  | Tgoto ident_cpp  { Goto ($2),  [$1] }
1003  | Tcontinue    { Continue,       [$1] }
1004  | Tbreak       { Break,          [$1] }
1005  | Treturn      { Return,         [$1] }
1006  | Treturn expr { ReturnExpr $2,  [$1] }
1007  | Tgoto TMul expr { GotoComputed $3, [$1;$2] }
1008
1009
1010
1011 /*(*----------------------------*)*/
1012 /*(* gccext: *)*/
1013 /*(*----------------------------*)*/
1014 string_elem:
1015  | TString { [snd $1] }
1016  /*(* cppext:  ex= printk (KERN_INFO "xxx" UTS_RELEASE)  *)*/
1017  | TMacroString { [snd $1] }
1018
1019
1020 asmbody:
1021  | string_list colon_asm_list  { $1, $2 }
1022  | string_list { $1, [] } /*(* in old kernel *)*/
1023
1024
1025 colon_asm: TDotDot colon_option_list { Colon $2, [$1]   }
1026
1027 colon_option:
1028  | TString                      { ColonMisc, [snd $1] }
1029  | TString TOPar asm_expr TCPar { ColonExpr $3, [snd $1; $2;$4] }
1030  /*(* cppext: certainly a macro *)*/
1031  | TOCro identifier TCCro TString TOPar asm_expr TCPar
1032      { ColonExpr $6, [$1;snd $2;$3;snd $4; $5; $7 ] }
1033  | identifier                       { ColonMisc, [snd $1] }
1034  | /*(* empty *)*/                  { ColonMisc, [] }
1035
1036 asm_expr: assign_expr { $1 }
1037
1038 /*(*************************************************************************)*/
1039 /*(* types *)*/
1040 /*(*************************************************************************)*/
1041
1042
1043 /*(*-----------------------------------------------------------------------*)*/
1044 /*(* Type spec, left part of a type *)*/
1045 /*(*-----------------------------------------------------------------------*)*/
1046 type_spec2:
1047  | Tvoid                { Right3 (BaseType Void),            [$1] }
1048  | Tchar                { Right3 (BaseType (IntType CChar)), [$1]}
1049  | Tint                 { Right3 (BaseType (IntType (Si (Signed,CInt)))), [$1]}
1050  | Tfloat               { Right3 (BaseType (FloatType CFloat)),  [$1]}
1051  | Tdouble              { Right3 (BaseType (FloatType CDouble)), [$1] }
1052  | Tsize_t              { Right3 (BaseType SizeType),            [$1] }
1053  | Tssize_t             { Right3 (BaseType SSizeType),           [$1] }
1054  | Tptrdiff_t           { Right3 (BaseType PtrDiffType),         [$1] }
1055  | Tshort               { Middle3 Short,  [$1]}
1056  | Tlong                { Middle3 Long,   [$1]}
1057  | Tsigned              { Left3 Signed,   [$1]}
1058  | Tunsigned            { Left3 UnSigned, [$1]}
1059  | struct_or_union_spec { Right3 (fst $1), snd $1 }
1060  | enum_spec            { Right3 (fst $1), snd $1 }
1061
1062  /*
1063  (* parse_typedef_fix1: cant put: TIdent {} cos it make the grammar
1064   * ambiguous, generates lots of conflicts => we must
1065   * use some tricks: we make the lexer and parser cooperate, cf lexerParser.ml.
1066   *
1067   * parse_typedef_fix2: this is not enough, and you must use
1068   * parse_typedef_fix2 to fully manage typedef problems in grammar.
1069   *
1070   * parse_typedef_fix3:
1071   *
1072   * parse_typedef_fix4: try also to do now some consistency checking in
1073   * Parse_c
1074   *)*/
1075  | TypedefIdent
1076      { let name = RegularName (mk_string_wrap $1) in
1077        Right3 (TypeName (name, Ast_c.noTypedefDef())),[] }
1078
1079  | Ttypeof TOPar assign_expr TCPar { Right3 (TypeOfExpr ($3)), [$1;$2;$4] }
1080  | Ttypeof TOPar type_name   TCPar { Right3 (TypeOfType ($3)), [$1;$2;$4] }
1081
1082 /*(*----------------------------*)*/
1083 /*(* workarounds *)*/
1084 /*(*----------------------------*)*/
1085
1086 type_spec: type_spec2    { dt "type" (); $1   }
1087
1088 /*(*-----------------------------------------------------------------------*)*/
1089 /*(* Qualifiers *)*/
1090 /*(*-----------------------------------------------------------------------*)*/
1091
1092 type_qualif:
1093  | Tconst    { {const=true  ; volatile=false}, $1 }
1094  | Tvolatile { {const=false ; volatile=true},  $1 }
1095  /*(* C99 *)*/
1096  | Trestrict { (* TODO *) {const=false ; volatile=false},  $1 }
1097
1098
1099 /*(*-----------------------------------------------------------------------*)*/
1100 /*(* gccext: attributes  *)*/
1101 /*(*-----------------------------------------------------------------------*)*/
1102
1103 attribute:
1104  | Tattribute TOPar /*stuff*/ TCPar { raise Todo }
1105  /*(* cppext: *)*/
1106  | TMacroAttr { Attribute (fst $1), [snd $1] }
1107
1108 attribute_storage:
1109  | TMacroAttrStorage { $1 }
1110
1111 type_qualif_attr:
1112  | type_qualif { $1 }
1113 /*(*TODO !!!!! *)*/
1114  | TMacroAttr { {const=true  ; volatile=false}, snd $1   }
1115
1116 /*(*-----------------------------------------------------------------------*)*/
1117 /*(* Declarator, right part of a type + second part of decl (the ident)  *)*/
1118 /*(*-----------------------------------------------------------------------*)*/
1119
1120 /*
1121 (* declarator return a couple:
1122  *  (name, partial type (a function to be applied to return type))
1123  *
1124  * when int* f(int) we must return Func(Pointer int,int) and not
1125  * Pointer (Func(int,int)
1126  *)*/
1127
1128 declarator:
1129  | pointer direct_d { (fst $2, fun x -> x +> $1 +> (snd $2)  ) }
1130  | direct_d         { $1  }
1131
1132 /*(* so must do  int * const p; if the pointer is constant, not the pointee *)*/
1133 pointer:
1134  | tmul                   { fun x -> mk_ty (Pointer x) [$1] }
1135  | tmul pointer           { fun x -> mk_ty (Pointer ($2 x)) [$1] }
1136  | tmul type_qualif_list
1137      { fun x -> ($2.qualifD, mk_tybis (Pointer x) [$1])}
1138  | tmul type_qualif_list pointer
1139      { fun x -> ($2.qualifD, mk_tybis (Pointer ($3 x)) [$1]) }
1140
1141 tmul:
1142    TMul { $1 }
1143  | TAnd
1144      { if !Flag.c_plus_plus
1145      then $1
1146      else
1147        let i = Ast_c.parse_info_of_info $1 in
1148        raise (Semantic("& not allowed in C types, try -c++ option", i)) }
1149
1150
1151 direct_d:
1152  | identifier_cpp
1153      { ($1, fun x -> x) }
1154  | TOPar declarator TCPar      /*(* forunparser: old: $2 *)*/
1155      { (fst $2, fun x -> mk_ty (ParenType ((snd $2) x)) [$1;$3]) }
1156  | direct_d tocro            tccro
1157      { (fst $1,fun x->(snd $1) (mk_ty (Array (None,x)) [$2;$3])) }
1158  | direct_d tocro const_expr tccro
1159      { (fst $1,fun x->(snd $1) (mk_ty (Array (Some $3,x)) [$2;$4])) }
1160  | direct_d topar            tcpar
1161      { (fst $1,
1162        fun x->(snd $1)
1163          (mk_ty (FunctionType (x,(([],(false, []))))) [$2;$3]))
1164      }
1165  | direct_d topar parameter_type_list tcpar
1166      { (fst $1,fun x->(snd $1)
1167        (mk_ty (FunctionType (x, $3)) [$2;$4]))
1168      }
1169
1170
1171 /*(*----------------------------*)*/
1172 /*(* workarounds *)*/
1173 /*(*----------------------------*)*/
1174
1175 tocro: TOCro { et "tocro" ();$1 }
1176 tccro: TCCro { dt "tccro" ();$1 }
1177
1178 /*(*-----------------------------------------------------------------------*)*/
1179 abstract_declarator:
1180  | pointer                            { $1 }
1181  |         direct_abstract_declarator { $1 }
1182  | pointer direct_abstract_declarator { fun x -> x +> $2 +> $1 }
1183
1184 direct_abstract_declarator:
1185  | TOPar abstract_declarator TCPar /*(* forunparser: old: $2 *)*/
1186      { fun x -> mk_ty (ParenType ($2 x)) [$1;$3] }
1187
1188  | TOCro            TCCro
1189      { fun x -> mk_ty (Array (None, x)) [$1;$2] }
1190  | TOCro const_expr TCCro
1191      { fun x -> mk_ty (Array (Some $2, x)) [$1;$3] }
1192  | direct_abstract_declarator TOCro            TCCro
1193      { fun x -> $1 (mk_ty (Array (None, x))  [$2;$3]) }
1194  | direct_abstract_declarator TOCro const_expr TCCro
1195      { fun x -> $1 (mk_ty (Array (Some $3,x))  [$2;$4]) }
1196  | TOPar TCPar
1197      { fun x -> mk_ty (FunctionType (x, ([], (false,  [])))) [$1;$2] }
1198  | topar parameter_type_list tcpar
1199      { fun x -> mk_ty (FunctionType (x, $2))  [$1;$3] }
1200 /*(* subtle: here must also use topar, not TOPar, otherwise if have for
1201    * instance   (xxx ( * )(xxx)) cast, then the second xxx may still be a Tident
1202    * but we want to reduce topar, to set the InParameter so that
1203    * parsing_hack can get a chance to change the type of xxx into a typedef.
1204    * That's an example where parsing_hack and the lookahead of ocamlyacc does
1205    * not go very well together ... we got the info too late. We got
1206    * a similar pb with xxx xxx; declaration, cf parsing_hack.ml and the
1207    * "disable typedef cos special case ..." message.
1208 *)*/
1209  | direct_abstract_declarator topar tcpar
1210      { fun x -> $1 (mk_ty (FunctionType (x, (([], (false, []))))) [$2;$3]) }
1211  | direct_abstract_declarator topar parameter_type_list tcpar
1212      { fun x -> $1 (mk_ty (FunctionType (x, $3)) [$2;$4]) }
1213
1214 /*(*-----------------------------------------------------------------------*)*/
1215 /*(* Parameters (use decl_spec not type_spec just for 'register') *)*/
1216 /*(*-----------------------------------------------------------------------*)*/
1217 parameter_type_list:
1218  | parameter_list                  { ($1, (false, []))}
1219  | parameter_list TComma TEllipsis { ($1, (true,  [$2;$3])) }
1220
1221
1222 parameter_decl2:
1223    TKRParam {
1224      let name = RegularName (mk_string_wrap $1) in
1225      LP.add_ident (str_of_name name);
1226      { p_namei = Some name;
1227        p_type = mk_ty NoType [];
1228        p_register = (false, []);
1229      }
1230    }
1231  | decl_spec declaratorp
1232      { let ((returnType,hasreg),iihasreg) = fixDeclSpecForParam $1 in
1233        let (name, ftyp) = $2 in
1234        { p_namei = Some (name);
1235          p_type = ftyp returnType;
1236          p_register = (hasreg, iihasreg);
1237        }
1238      }
1239  | decl_spec abstract_declaratorp
1240      { let ((returnType,hasreg), iihasreg) = fixDeclSpecForParam $1 in
1241        { p_namei = None;
1242          p_type = $2 returnType;
1243          p_register = hasreg, iihasreg;
1244        }
1245      }
1246  | decl_spec
1247      { let ((returnType,hasreg), iihasreg) = fixDeclSpecForParam $1 in
1248        { p_namei = None;
1249          p_type = returnType;
1250          p_register = hasreg, iihasreg;
1251        }
1252      }
1253
1254
1255 /*(*----------------------------*)*/
1256 /*(* workarounds *)*/
1257 /*(*----------------------------*)*/
1258
1259 parameter_decl: parameter_decl2 { et "param" ();  $1 }
1260  | attributes parameter_decl2 { et "param" (); $2 }
1261
1262 declaratorp:
1263  | declarator  { LP.add_ident (str_of_name (fst $1)); $1 }
1264  /*(* gccext: *)*/
1265  | attributes declarator   { LP.add_ident (str_of_name (fst $2)); $2 }
1266  | declarator attributes   { LP.add_ident (str_of_name (fst $1)); $1 }
1267
1268 abstract_declaratorp:
1269  | abstract_declarator { $1 }
1270  /*(* gccext: *)*/
1271  | attributes abstract_declarator { $2 }
1272
1273 /*(*-----------------------------------------------------------------------*)*/
1274 /*(* helper type rules *)*/
1275 /*(*-----------------------------------------------------------------------*)*/
1276
1277 /*(* for struct and also typename *)*/
1278 /*(* cant put decl_spec cos no storage is allowed for field struct *)*/
1279 spec_qualif_list2:
1280  | type_spec                    { addTypeD ($1, nullDecl) }
1281  | type_qualif                  { {nullDecl with qualifD = (fst $1,[snd $1])}}
1282  | type_spec   spec_qualif_list { addTypeD ($1,$2)   }
1283  | type_qualif spec_qualif_list { addQualifD ($1,$2) }
1284
1285 spec_qualif_list: spec_qualif_list2            {  dt "spec_qualif" (); $1 }
1286
1287
1288 /*(* for pointers in direct_declarator and abstract_declarator *)*/
1289 type_qualif_list:
1290  | type_qualif_attr                  { {nullDecl with qualifD = (fst $1,[snd $1])} }
1291  | type_qualif_list type_qualif_attr { addQualifD ($2,$1) }
1292
1293
1294
1295
1296
1297
1298 /*(*-----------------------------------------------------------------------*)*/
1299 /*(* xxx_type_id *)*/
1300 /*(*-----------------------------------------------------------------------*)*/
1301
1302 type_name:
1303  | spec_qualif_list
1304      { let (returnType, _) = fixDeclSpecForDecl $1 in  returnType }
1305  | spec_qualif_list abstract_declaratort
1306      { let (returnType, _) = fixDeclSpecForDecl $1 in $2 returnType }
1307
1308
1309
1310 abstract_declaratort:
1311  | abstract_declarator { $1 }
1312  /*(* gccext: *)*/
1313  | attributes abstract_declarator { $2 }
1314
1315
1316 /*(*************************************************************************)*/
1317 /*(* declaration and initializers *)*/
1318 /*(*************************************************************************)*/
1319
1320 decl2:
1321  | decl_spec TPtVirg
1322      { function local ->
1323        let (returnType,storage) = fixDeclSpecForDecl $1 in
1324        let iistart = Ast_c.fakeInfo () in
1325        DeclList ([{v_namei = None; v_type = returnType;
1326                    v_storage = unwrap storage; v_local = local;
1327                    v_attr = Ast_c.noattr;
1328                    v_type_bis = ref None;
1329                 },[]],
1330                 ($2::iistart::snd storage))
1331      }
1332  | decl_spec init_declarator_list TPtVirg
1333      { function local ->
1334        let (returnType,storage) = fixDeclSpecForDecl $1 in
1335        let iistart = Ast_c.fakeInfo () in
1336        DeclList (
1337          ($2 +> List.map (fun ((((name,f),attrs), ini), iivirg) ->
1338            let s = str_of_name name in
1339            if fst (unwrap storage) =*= StoTypedef
1340            then LP.add_typedef s;
1341            {v_namei = Some (name, ini);
1342             v_type = f returnType;
1343             v_storage = unwrap storage;
1344             v_local = local;
1345             v_attr = attrs;
1346             v_type_bis = ref None;
1347            },
1348            iivirg
1349          )
1350          ),  ($3::iistart::snd storage))
1351      }
1352  /*(* cppext: *)*/
1353
1354  | TMacroDecl TOPar argument_list TCPar TPtVirg
1355      { function _ ->
1356        MacroDecl ((fst $1, $3, true), [snd $1;$2;$4;$5;fakeInfo()]) }
1357  | Tstatic TMacroDecl TOPar argument_list TCPar TPtVirg
1358      { function _ ->
1359        MacroDecl ((fst $2, $4, true), [snd $2;$3;$5;$6;fakeInfo();$1]) }
1360  | Tstatic TMacroDeclConst TMacroDecl TOPar argument_list TCPar TPtVirg
1361      { function _ ->
1362        MacroDecl ((fst $3, $5, true), [snd $3;$4;$6;$7;fakeInfo();$1;$2])}
1363
1364
1365  | TMacroDecl TOPar argument_list TCPar teq initialize TPtVirg
1366      { function _ ->
1367        MacroDeclInit ((fst $1, $3, $6), [snd $1;$2;$4;$5;$7;fakeInfo()]) }
1368  | Tstatic TMacroDecl TOPar argument_list TCPar teq initialize TPtVirg
1369      { function _ ->
1370        MacroDeclInit ((fst $2, $4, $7),[snd $2;$3;$5;$6;$8;fakeInfo();$1]) }
1371  | Tstatic TMacroDeclConst TMacroDecl TOPar argument_list TCPar
1372      teq initialize TPtVirg
1373      { function _ ->
1374        MacroDeclInit
1375          ((fst $3, $5, $8), [snd $3;$4;$6;$7;$9;fakeInfo();$1;$2])}
1376
1377
1378 /*(*-----------------------------------------------------------------------*)*/
1379 decl_spec2:
1380  | storage_class_spec      { {nullDecl with storageD = (fst $1, [snd $1]) } }
1381  | type_spec               { addTypeD ($1,nullDecl) }
1382  | type_qualif             { {nullDecl with qualifD  = (fst $1, [snd $1]) } }
1383  | Tinline                 { {nullDecl with inlineD = (true, [$1]) } }
1384  | storage_class_spec decl_spec2 { addStorageD ($1, $2) }
1385  | type_spec          decl_spec2 { addTypeD    ($1, $2) }
1386  | type_qualif        decl_spec2 { addQualifD  ($1, $2) }
1387  | Tinline            decl_spec2 { addInlineD ((true, $1), $2) }
1388
1389 /*(* can simplify by putting all in _opt ? must have at least one otherwise
1390    *  decl_list is ambiguous ? (no cos have ';' between decl)
1391    *)*/
1392
1393
1394 storage_class_spec2:
1395  | Tstatic      { Sto Static,  $1 }
1396  | Textern      { Sto Extern,  $1 }
1397  | Tauto        { Sto Auto,    $1 }
1398  | Tregister    { Sto Register,$1 }
1399  | Ttypedef     { StoTypedef,  $1 }
1400
1401 storage_class_spec:
1402  /*(* gccext: *)*/
1403  | storage_class_spec2 { $1 }
1404  | storage_class_spec2 attribute_storage_list { $1 (* TODO *) }
1405
1406
1407
1408 /*(*----------------------------*)*/
1409 /*(* workarounds *)*/
1410 /*(*----------------------------*)*/
1411
1412 decl:      decl2         { et "decl" (); $1 }
1413 decl_spec: decl_spec2    { dt "declspec" (); $1  }
1414
1415 /*(*-----------------------------------------------------------------------*)*/
1416 /*(* declarators (right part of type and variable) *)*/
1417 /*(*-----------------------------------------------------------------------*)*/
1418 init_declarator2:
1419  | declaratori                  { ($1, NoInit) }
1420  | declaratori teq initialize   { ($1, ValInit($2, $3)) }
1421  /* C++ only */
1422  | declaratori TOParCplusplusInit argument_list TCPar
1423      { ($1, ConstrInit($3,[$2;$4])) }
1424
1425
1426 /*(*----------------------------*)*/
1427 /*(* workarounds *)*/
1428 /*(*----------------------------*)*/
1429 teq: TEq  { et "teq" (); $1 }
1430
1431 init_declarator: init_declarator2  { dt "init" (); $1 }
1432
1433
1434 /*(*----------------------------*)*/
1435 /*(* gccext: *)*/
1436 /*(*----------------------------*)*/
1437
1438 declaratori:
1439  | declarator              { LP.add_ident (str_of_name (fst $1)); $1, Ast_c.noattr }
1440  /*(* gccext: *)*/
1441  | declarator gcc_asm_decl { LP.add_ident (str_of_name (fst $1)); $1, Ast_c.noattr }
1442  /*(* gccext: *)*/
1443  | attributes declarator   { LP.add_ident (str_of_name (fst $2)); $2, $1 }
1444  | declarator attributes   { LP.add_ident (str_of_name (fst $1)); $1, Ast_c.noattr (* TODO *) }
1445
1446
1447
1448 gcc_asm_decl:
1449  | Tasm TOPar asmbody TCPar              {  }
1450  | Tasm Tvolatile TOPar asmbody TCPar   {  }
1451
1452
1453 /*(*-----------------------------------------------------------------------*)*/
1454 initialize:
1455  | assign_expr
1456      { InitExpr $1,                [] }
1457  | tobrace_ini initialize_list gcc_comma_opt_struct  tcbrace_ini
1458      { InitList (List.rev $2),     [$1;$4]++$3 }
1459  | tobrace_ini tcbrace_ini
1460      { InitList [],       [$1;$2] } /*(* gccext: *)*/
1461
1462
1463 /*
1464 (* opti: This time we use the weird order of non-terminal which requires in
1465  * the "caller" to do a List.rev cos quite critical. With this weird order it
1466  * allows yacc to use a constant stack space instead of exploding if we would
1467  * do a  'initialize2 Tcomma initialize_list'.
1468  *)
1469 */
1470 initialize_list:
1471  | initialize2                        { [$1,   []] }
1472  | initialize_list TComma initialize2 { ($3,  [$2])::$1 }
1473
1474
1475 /*(* gccext: condexpr and no assign_expr cos can have ambiguity with comma *)*/
1476 initialize2:
1477  | cond_expr
1478      { InitExpr $1,   [] }
1479  | tobrace_ini initialize_list gcc_comma_opt_struct tcbrace_ini
1480      { InitList (List.rev $2),   [$1;$4]++$3 }
1481  | tobrace_ini tcbrace_ini
1482      { InitList [],  [$1;$2]  }
1483
1484  /*(* gccext: labeled elements, a.k.a designators *)*/
1485  | designator_list TEq initialize2
1486      { InitDesignators ($1, $3), [$2] }
1487
1488  /*(* gccext: old format *)*/
1489  | ident TDotDot initialize2
1490      { InitFieldOld (fst $1, $3),     [snd $1; $2] } /*(* in old kernel *)*/
1491 /* conflict
1492  | TOCro const_expr TCCro initialize2
1493      { InitIndexOld ($2, $4),    [$1;$3] }
1494 */
1495
1496
1497
1498 /*(* they can be nested, can have a .x[3].y *)*/
1499 designator:
1500  | TDot ident
1501      { DesignatorField (fst $2), [$1;snd $2] }
1502  | TOCro const_expr TCCro
1503      { DesignatorIndex ($2),  [$1;$3] }
1504  | TOCro const_expr TEllipsis const_expr TCCro
1505      { DesignatorRange ($2, $4),  [$1;$3;$5] }
1506
1507
1508 /*(*----------------------------*)*/
1509 /*(* workarounds *)*/
1510 /*(*----------------------------*)*/
1511
1512 gcc_comma_opt_struct:
1513  | TComma {  [$1] }
1514  | /*(* empty *)*/  {  [Ast_c.fakeInfo() +> Ast_c.rewrap_str ","]  }
1515
1516
1517
1518
1519
1520
1521
1522
1523 /*(*************************************************************************)*/
1524 /*(* struct *)*/
1525 /*(*************************************************************************)*/
1526
1527 s_or_u_spec2:
1528  | struct_or_union ident tobrace_struct struct_decl_list_gcc tcbrace_struct
1529      { StructUnion (fst $1, Some (fst $2), $4),  [snd $1;snd $2;$3;$5]  }
1530  | struct_or_union       tobrace_struct struct_decl_list_gcc tcbrace_struct
1531      { StructUnion (fst $1, None, $3), [snd $1;$2;$4] }
1532  | struct_or_union ident
1533      { StructUnionName (fst $1, fst $2), [snd $1;snd $2] }
1534
1535 struct_or_union2:
1536  | Tstruct   { Struct, $1 }
1537  | Tunion    { Union, $1 }
1538  /*(* gccext: *)*/
1539  | Tstruct attributes   { Struct, $1 (* TODO *) }
1540  | Tunion  attributes   { Union, $1  (* TODO *) }
1541
1542
1543
1544 struct_decl2:
1545  | field_declaration { DeclarationField $1 }
1546  | TPtVirg { EmptyField $1  }
1547
1548  /*(* no conflict ? no need for a TMacroStruct ? apparently not as at struct
1549     * the rule are slightly different.
1550     *)*/
1551  | identifier TOPar argument_list TCPar TPtVirg
1552      { MacroDeclField ((fst $1, $3), [snd $1;$2;$4;$5;fakeInfo()]) }
1553
1554  /*(* cppext: *)*/
1555  | cpp_directive
1556      { CppDirectiveStruct $1 }
1557  | cpp_ifdef_directive/*(* struct_decl_list ... *)*/
1558      { IfdefStruct $1 }
1559
1560
1561 field_declaration:
1562  | spec_qualif_list struct_declarator_list TPtVirg
1563      {
1564        let (returnType,storage) = fixDeclSpecForDecl $1 in
1565        if fst (unwrap storage) <> NoSto
1566        then internal_error "parsing dont allow this";
1567
1568        FieldDeclList ($2 +> (List.map (fun (f, iivirg) ->
1569          f returnType, iivirg))
1570                          ,[$3])
1571          (* dont need to check if typedef or func initialised cos
1572           * grammar dont allow typedef nor initialiser in struct
1573           *)
1574      }
1575
1576  |  spec_qualif_list TPtVirg
1577      {
1578        (* gccext: allow empty elements if it is a structdef or enumdef *)
1579        let (returnType,storage) = fixDeclSpecForDecl $1 in
1580        if fst (unwrap storage) <> NoSto
1581        then internal_error "parsing dont allow this";
1582
1583        FieldDeclList ([(Simple (None, returnType)) , []], [$2])
1584      }
1585
1586
1587
1588
1589
1590 struct_declarator:
1591  | declaratorsd
1592      { (fun x -> Simple   (Some (fst $1), (snd $1) x)) }
1593  | dotdot const_expr2
1594      { (fun x -> BitField (None, x, $1, $2)) }
1595  | declaratorsd dotdot const_expr2
1596      { (fun x -> BitField (Some (fst $1), ((snd $1) x), $2, $3)) }
1597
1598
1599 /*(*----------------------------*)*/
1600 /*(* workarounds *)*/
1601 /*(*----------------------------*)*/
1602 declaratorsd:
1603  | declarator { (*also ? LP.add_ident (fst (fst $1)); *) $1 }
1604  /*(* gccext: *)*/
1605  | attributes declarator   { $2 }
1606  | declarator attributes   { $1 }
1607
1608
1609
1610
1611 struct_or_union_spec: s_or_u_spec2 { dt "su" (); $1 }
1612 struct_or_union: struct_or_union2 { et "su" (); $1 }
1613 struct_decl: struct_decl2  { et "struct" (); $1 }
1614
1615 dotdot: TDotDot  { et "dotdot" (); $1 }
1616 const_expr2: const_expr { dt "const_expr2" (); $1 }
1617
1618 struct_decl_list_gcc:
1619  | struct_decl_list  { $1 }
1620  | /*(* empty *)*/       { [] } /*(* gccext: allow empty struct *)*/
1621
1622
1623 /*(*************************************************************************)*/
1624 /*(* enum *)*/
1625 /*(*************************************************************************)*/
1626 enum_spec:
1627  | Tenum        tobrace_enum enumerator_list gcc_comma_opt_struct tcbrace_enum
1628      { Enum (None,    $3),           [$1;$2;$5] ++ $4 }
1629  | Tenum ident  tobrace_enum enumerator_list gcc_comma_opt_struct tcbrace_enum
1630      { Enum (Some (fst $2), $4),     [$1; snd $2; $3;$6] ++ $5 }
1631  | Tenum ident
1632      { EnumName (fst $2),       [$1; snd $2] }
1633
1634 enumerator:
1635  | idente                 { $1, None     }
1636  | idente  TEq const_expr { $1, Some ($2, $3) }
1637
1638
1639 /*(*----------------------------*)*/
1640 /*(* workarounds *)*/
1641 /*(*----------------------------*)*/
1642
1643 idente: ident_cpp { LP.add_ident (str_of_name $1); $1 }
1644
1645
1646
1647 /*(*************************************************************************)*/
1648 /*(* function *)*/
1649 /*(*************************************************************************)*/
1650 function_definition: function_def    { fixFunc $1 }
1651
1652 decl_list:
1653  | decl           { [$1 Ast_c.LocalDecl]   }
1654  | decl_list decl { $1 ++ [$2 Ast_c.LocalDecl] }
1655
1656 /* hack : to drop when a better solution is found */
1657 cpp_directive_list:
1658  | cpp_directive                    { }
1659  | cpp_directive_list cpp_directive { }
1660
1661 function_def:
1662  | start_fun compound      { LP.del_scope(); ($1, $2, None) }
1663  | start_fun cpp_directive_list compound { LP.del_scope(); ($1, $3, None) }
1664  | start_fun decl_list compound      {
1665      (* TODO: undo the typedef added ? *)
1666      LP.del_scope();
1667      ($1, $3, Some $2)
1668    }
1669
1670 start_fun: start_fun2
1671   { LP.new_scope();
1672     fix_add_params_ident $1;
1673     (* toreput? !LP._lexer_hint.toplevel <- false;  *)
1674     $1
1675   }
1676
1677 start_fun2: decl_spec declaratorfd
1678      { let (returnType,storage) = fixDeclSpecForFuncDef $1 in
1679        let (id, attrs) = $2 in
1680        (fst id, fixOldCDecl ((snd id) returnType) , storage, attrs)
1681      }
1682    | ctor_dtor { $1 }
1683
1684 ctor_dtor:
1685  | Tconstructorname topar tcpar {
1686      let id = RegularName (mk_string_wrap $1) in
1687      let ret = mk_ty NoType [] in
1688      let ty = mk_ty (FunctionType (ret, (([], (false, []))))) [$2;$3] in
1689      let storage = ((NoSto,false),[]) in
1690      let attrs = [] in
1691      (id, ty, storage, attrs) }
1692  | Tconstructorname topar parameter_type_list tcpar {
1693      let id = RegularName (mk_string_wrap $1) in
1694      let ret = mk_ty NoType [] in
1695      let ty = mk_ty (FunctionType (ret, $3)) [$2;$4] in
1696      let storage = ((NoSto,false),[]) in
1697      let attrs = [] in
1698      (id, ty, storage, attrs) }
1699
1700 /*(*----------------------------*)*/
1701 /*(* workarounds *)*/
1702 /*(*----------------------------*)*/
1703
1704 /* It would be very nice if we could make declarator aware that this is
1705 coming from a function definition.  Then on the ( and ) cases, it could
1706 set the state to something other than InParameter.  Then the case
1707 (TIdent (s, i1)::(TComma _|TCPar _)::_ , (TComma _ |TOPar _)::_ )
1708 in parsing_hacks.ml would not have to consider K&R variable declarations
1709 as typedefs.  Unfortunately, doing something about this problem seems to
1710 introduce conflicts in the parser. */
1711
1712 declaratorfd:
1713  | declarator { et "declaratorfd" (); $1, Ast_c.noattr }
1714  /*(* gccext: *)*/
1715  | attributes declarator   { et "declaratorfd" (); $2, $1 }
1716  | declarator attributes   { et "declaratorfd" (); $1, Ast_c.noattr }
1717
1718
1719
1720 /*(*************************************************************************)*/
1721 /*(* cpp directives *)*/
1722 /*(*************************************************************************)*/
1723
1724 cpp_directive:
1725  | TIncludeStart TIncludeFilename
1726      {
1727        let (i1, in_ifdef) = $1 in
1728        let (s, i2) = $2 in
1729
1730        (* redo some lexing work :( *)
1731        let inc_file =
1732          match () with
1733          | _ when s =~ "^\"\\(.*\\)\"$" ->
1734              Local (Common.split "/" (matched1 s))
1735          | _ when s =~ "^\\<\\(.*\\)\\>$" ->
1736              NonLocal (Common.split "/" (matched1 s))
1737          | _ ->
1738              Weird s
1739        in
1740        Include { i_include = (inc_file, [i1;i2]);
1741                  i_rel_pos = Ast_c.noRelPos();
1742                  i_is_in_ifdef = !in_ifdef;
1743                  i_content = Ast_c.noi_content;
1744        }
1745      }
1746
1747  | TDefine TIdentDefine define_val TDefEOL
1748      { Define ((fst $2, [$1; snd $2;$4]), (DefineVar, $3)) }
1749
1750  /*
1751  (* The TOParDefine is introduced to avoid ambiguity with previous rules.
1752   * A TOParDefine is a TOPar that was just next to the ident.
1753   *)*/
1754  | TDefine TIdentDefine TOParDefine param_define_list TCPar define_val TDefEOL
1755      { Define
1756          ((fst $2, [$1; snd $2; $7]),
1757            (DefineFunc ($4, [$3;$5]), $6))
1758      }
1759
1760  | TUndef TIdentDefine TDefEOL
1761      { Define((fst $2, [$1; snd $2; $3]), (Undef,DefineEmpty)) }
1762  | TCppDirectiveOther { PragmaAndCo ([$1]) }
1763
1764
1765
1766
1767
1768 /*(* perhaps better to use assign_expr ? but in that case need
1769    * do a assign_expr_of_string in parse_c
1770    *)*/
1771 define_val:
1772  | expr      { DefineExpr $1 }
1773  | statement { DefineStmt $1 }
1774  | decl      { DefineStmt (mk_st (Decl ($1 Ast_c.NotLocalDecl)) Ast_c.noii) }
1775
1776 /*(*old:
1777   * | TypedefIdent { DefineType (nQ,(TypeName(fst $1,noTypedefDef()),[snd $1]))}
1778   * get conflicts:
1779   * | spec_qualif_list TMul
1780   *   { let (returnType, _) = fixDeclSpecForDecl $1 in  DefineType returnType }
1781   *)
1782 */
1783  | decl_spec
1784      { let returnType = fixDeclSpecForMacro $1 in
1785        DefineType returnType
1786      }
1787  | decl_spec abstract_declarator
1788      { let returnType = fixDeclSpecForMacro $1 in
1789        let typ = $2 returnType in
1790        DefineType typ
1791      }
1792
1793 /* can be in conflict with decl_spec, maybe change fixDeclSpecForMacro
1794  * to also allow storage ?
1795  | storage_class_spec { DefineTodo }
1796  | Tinline { DefineTodo }
1797 */
1798
1799  /*(* a few special cases *)*/
1800  | stat_or_decl stat_or_decl_list { DefineTodo }
1801 /*
1802  | statement statement { DefineTodo }
1803  | decl function_definition { DefineTodo }
1804 */
1805
1806
1807
1808
1809  | function_definition { DefineFunction $1 }
1810
1811  | TOBraceDefineInit initialize_list gcc_comma_opt_struct TCBrace comma_opt
1812     { DefineInit (InitList (List.rev $2), [$1;$4]++$3++$5)  }
1813
1814  /*(* note: had a conflict before when were putting TInt instead of expr *)*/
1815  | Tdo statement Twhile TOPar expr TCPar
1816      {
1817        (* TOREPUT
1818        if fst $5 <> "0"
1819        then pr2 "WEIRD: in macro and have not a while(0)";
1820        *)
1821        DefineDoWhileZero (($2,$5),   [$1;$3;$4;$6])
1822      }
1823
1824  | Tasm TOPar asmbody TCPar              { DefineTodo }
1825  | Tasm Tvolatile TOPar asmbody TCPar    { DefineTodo }
1826
1827  /*(* aliases macro *)*/
1828  | TMacroAttr { DefineTodo }
1829
1830  | /*(* empty *)*/ { DefineEmpty }
1831
1832
1833
1834
1835 param_define:
1836  | TIdent               { mk_string_wrap $1 }
1837  | TypedefIdent         { mk_string_wrap $1 }
1838  | TDefParamVariadic    { mk_string_wrap $1 }
1839  | TEllipsis            { "...", [$1] }
1840  /*(* they reuse keywords :(  *)*/
1841  | Tregister            { "register", [$1] }
1842
1843
1844
1845
1846 cpp_ifdef_directive:
1847  | TIfdef
1848      { let (tag,ii) = $1 in
1849        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)),  [ii]) }
1850  | TIfdefelse
1851      { let (tag,ii) = $1 in
1852        IfdefDirective ((IfdefElse, IfdefTag (Common.some !tag)), [ii]) }
1853  | TIfdefelif
1854      { let (tag,ii) = $1 in
1855        IfdefDirective ((IfdefElseif, IfdefTag (Common.some !tag)), [ii]) }
1856  | TEndif
1857      { let (tag,ii) = $1 in
1858        IfdefDirective ((IfdefEndif, IfdefTag (Common.some !tag)), [ii]) }
1859
1860  | TIfdefBool
1861      { let (_b, tag,ii) = $1 in
1862        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)), [ii]) }
1863  | TIfdefMisc
1864      { let (_b, tag,ii) = $1 in
1865        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)), [ii]) }
1866  | TIfdefVersion
1867      { let (_b, tag,ii) = $1 in
1868        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)), [ii]) }
1869
1870
1871 /*(* cppext: *)*/
1872 cpp_other:
1873  /*(* no conflict ? no need for a TMacroTop ? apparently not as at toplevel
1874     * the rule are slightly different, they cant be statement and so expr
1875     * at the top, only decl or function definition.
1876     *)*/
1877  | identifier TOPar argument_list TCPar TPtVirg
1878      {
1879        Declaration(MacroDecl((fst $1, $3, true), [snd $1;$2;$4;$5;fakeInfo()]))
1880        (* old: MacroTop (fst $1, $3,    [snd $1;$2;$4;$5])  *)
1881      }
1882
1883  /*(* TCParEOL to fix the end-of-stream bug of ocamlyacc *)*/
1884  | identifier TOPar argument_list TCParEOL
1885      { Declaration (MacroDecl ((fst $1, $3, false), [snd $1;$2;$4;fakeInfo()])) }
1886
1887   /*(* ex: EXPORT_NO_SYMBOLS; *)*/
1888  | identifier TPtVirg { EmptyDef [snd $1;$2] }
1889
1890
1891
1892 /*(*************************************************************************)*/
1893 /*(* celem *)*/
1894 /*(*************************************************************************)*/
1895
1896 external_declaration:
1897  | function_definition               { Definition $1 }
1898  | decl                              { Declaration ($1 Ast_c.NotLocalDecl) }
1899
1900
1901 celem:
1902    | external_declaration                         { $1 }
1903
1904  /*(* cppext: *)*/
1905  | cpp_directive
1906      { CppTop $1 }
1907  | cpp_other
1908      { $1 }
1909  | cpp_ifdef_directive /* (*external_declaration_list ...*)*/
1910      { IfdefTop $1 }
1911
1912  /*(* can have asm declaration at toplevel *)*/
1913  | Tasm TOPar asmbody TCPar TPtVirg             { EmptyDef [$1;$2;$4;$5] }
1914
1915  /*
1916  (* in ~/kernels/src/linux-2.5.2/drivers/isdn/hisax/isdnl3.c sometimes
1917   * the function ends with }; instead of just }
1918   * can also remove this rule and report "parse error" pb to morton
1919   *)*/
1920  | TPtVirg    { EmptyDef [$1] }
1921
1922
1923  | EOF        { FinalDef $1 }
1924
1925
1926
1927
1928 /*(*************************************************************************)*/
1929 /*(* some generic workarounds *)*/
1930 /*(*************************************************************************)*/
1931
1932 tobrace: TOBrace  { LP.push_context LP.InFunction; LP.new_scope (); $1 }
1933 tcbrace: TCBrace  { LP.pop_context();              LP.del_scope (); $1 }
1934
1935 tobrace_enum: TOBrace { LP.push_context LP.InEnum; $1 }
1936 tcbrace_enum: TCBrace { LP.pop_context (); $1 }
1937
1938 tobrace_ini: TOBrace { LP.push_context LP.InInitializer; $1 }
1939 tcbrace_ini: TCBrace { LP.pop_context (); $1 }
1940
1941 tobrace_struct: TOBrace { LP.push_context LP.InStruct; $1}
1942 tcbrace_struct: TCBrace { LP.pop_context (); $1 }
1943
1944
1945
1946
1947 topar: TOPar
1948      { LP.new_scope ();et "topar" ();
1949        LP.push_context LP.InParameter;
1950        $1
1951      }
1952 tcpar: TCPar
1953      { LP.del_scope ();dt "tcpar" ();
1954        LP.pop_context ();
1955        $1
1956      }
1957
1958
1959
1960
1961 /*(*************************************************************************)*/
1962 /*(* xxx_list, xxx_opt *)*/
1963 /*(*************************************************************************)*/
1964
1965
1966 /*(* old:
1967 compound2:
1968  |                            { ([],[]) }
1969  |  statement_list            { ([], $1) }
1970  |  decl_list                 { ($1, []) }
1971  |  decl_list statement_list  { ($1,$2) }
1972
1973 statement_list: stat_or_decl_list { $1 }
1974 *)*/
1975
1976
1977 /*(*
1978 decl_list:
1979  | decl           { [$1]   }
1980  | decl_list decl { $1 ++ [$2] }
1981
1982 statement_list:
1983  | statement { [$1] }
1984  | statement_list statement { $1 ++ [$2] }
1985 *)*/
1986
1987
1988
1989
1990
1991 string_list:
1992  | string_elem { $1 }
1993  | string_list string_elem { $1 ++ $2 }
1994
1995 colon_asm_list:
1996  | colon_asm { [$1] }
1997  | colon_asm_list colon_asm  { $1 ++ [$2] }
1998
1999 colon_option_list:
2000  | colon_option { [$1, []] }
2001  | colon_option_list TComma colon_option { $1 ++ [$3, [$2]] }
2002
2003
2004 argument_list_ne:
2005  | argument_ne                           { [$1, []] }
2006  | argument_list_ne TComma argument { $1 ++ [$3,    [$2]] }
2007
2008 argument_list:
2009  | argument                           { [$1, []] }
2010  | argument_list TComma argument { $1 ++ [$3,    [$2]] }
2011
2012 /*(*
2013 expression_list:
2014  | assign_expr { [$1, []] }
2015  | expression_list TComma assign_expr { $1 ++ [$3,   [$2]] }
2016 *)*/
2017
2018
2019 struct_decl_list:
2020  | struct_decl                   { [$1] }
2021  | struct_decl_list struct_decl  { $1 ++ [$2] }
2022
2023
2024 struct_declarator_list:
2025  | struct_declarator                               { [$1,           []] }
2026  | struct_declarator_list TComma struct_declarator { $1 ++ [$3,     [$2]] }
2027
2028
2029 enumerator_list:
2030  | enumerator                        { [$1,          []]   }
2031  | enumerator_list TComma enumerator { $1 ++ [$3,    [$2]] }
2032
2033
2034 init_declarator_list:
2035  | init_declarator                             { [$1,   []] }
2036  | init_declarator_list TComma init_declarator { $1 ++ [$3,     [$2]] }
2037
2038
2039 parameter_list:
2040  | parameter_decl                       { [$1, []] }
2041  | parameter_list TComma parameter_decl { $1 ++ [$3,  [$2]] }
2042
2043 taction_list_ne:
2044  | TAction                 { [$1] }
2045  | TAction taction_list_ne { $1 :: $2 }
2046
2047 taction_list:
2048 /*old: was generating conflict, hence now taction_list_ne
2049     | (* empty *) { [] }
2050  | TAction { [$1] }
2051  | taction_list TAction { $1 ++ [$2] }
2052 */
2053  |                      { [] }
2054  | TAction taction_list { $1 :: $2 }
2055
2056 param_define_list:
2057  | /*(* empty *)*/ { [] }
2058  | param_define                           { [$1, []] }
2059  | param_define_list TComma param_define  { $1 ++ [$3, [$2]] }
2060
2061 designator_list:
2062  | designator { [$1] }
2063  | designator_list designator { $1 ++ [$2] }
2064
2065 attribute_list:
2066  | attribute { [$1] }
2067  | attribute_list attribute { $1 ++ [$2] }
2068
2069 attribute_storage_list:
2070  | attribute_storage { [$1] }
2071  | attribute_storage_list attribute_storage { $1 ++ [$2] }
2072
2073
2074 attributes: attribute_list { $1 }
2075
2076
2077
2078 /*(* gccext:  which allow a trailing ',' in enum, as in perl *)*/
2079 gcc_comma_opt:
2080  | TComma {  [$1] }
2081  | /*(* empty *)*/  {  []  }
2082
2083 comma_opt:
2084  | TComma {  [$1] }
2085  | /*(* empty *)*/  {  []  }
2086
2087 /*(*
2088 gcc_opt_virg:
2089  | TPtVirg { }
2090  |  { }
2091 *)*/
2092
2093 gcc_opt_expr:
2094  | expr        { Some $1 }
2095  | /*(* empty *)*/ { None  }
2096
2097 /*(*
2098 opt_ptvirg:
2099  | TPtVirg { [$1] }
2100  | { [] }
2101 *)*/
2102
2103
2104 expr_opt:
2105  | expr            { Some $1 }
2106  | /*(* empty *)*/ { None }
2107