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 _),ii2)} as _v)) ->
 102       raise (Semantic ("two or more data types", 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 ("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 ("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 /*(* appears mostly after some fix_xxx in parsing_hack *)*/
 404 %token <string * Ast_c.info> TypedefIdent
 405
 406
 407 /*
 408 (* Some tokens like TOPar and TCPar are used as synchronisation stuff,
 409  * in parsing_hack.ml. So if define special tokens like TOParDefine and
 410  * TCParEOL, then take care to also modify in Token_helpers.
 411  *)
 412 */
 413
 414 %token <Ast_c.info> TOPar TCPar TOBrace TCBrace TOCro TCCro
 415 %token <Ast_c.info> TDot TComma TPtrOp
 416 %token <Ast_c.info> TInc TDec
 417 %token <Ast_c.assignOp * Ast_c.info> TAssign
 418 %token <Ast_c.info> TEq
 419 %token <Ast_c.info> TWhy  TTilde TBang
 420 %token <Ast_c.info> TEllipsis
 421 %token <Ast_c.info> TDotDot
 422
 423 %token <Ast_c.info> TPtVirg
 424 %token <Ast_c.info>
 425        TOrLog TAndLog TOr TXor TAnd  TEqEq TNotEq TInf TSup TInfEq TSupEq
 426        TShl TShr
 427        TPlus TMinus TMul TDiv TMod
 428
 429 %token <Ast_c.info>
 430        Tchar Tshort Tint Tdouble Tfloat Tlong Tunsigned Tsigned Tvoid
 431        Tsize_t Tssize_t Tptrdiff_t
 432        Tauto Tregister Textern Tstatic
 433        Ttypedef
 434        Tconst Tvolatile
 435        Tstruct Tunion Tenum
 436        Tbreak Telse Tswitch Tcase Tcontinue Tfor Tdo Tif  Twhile Treturn
 437        Tgoto Tdefault
 438        Tsizeof
 439
 440 /*(* C99 *)*/
 441 %token <Ast_c.info>
 442        Trestrict
 443
 444 /*(*-----------------------------------------*)*/
 445 /*(* gccext: extra tokens *)*/
 446 /*(*-----------------------------------------*)*/
 447 %token <Ast_c.info> Tasm
 448 %token <Ast_c.info> Tattribute
 449 %token <Ast_c.info> TattributeNoarg
 450 %token <Ast_c.info> Tinline
 451 %token <Ast_c.info> Ttypeof
 452
 453 /*(*-----------------------------------------*)*/
 454 /*(* cppext: extra tokens *)*/
 455 /*(*-----------------------------------------*)*/
 456 /*(* coupling with Token_helpers.is_cpp_token *)*/
 457
 458
 459 /*(*---------------*)*/
 460 /*(* define        *)*/
 461 /*(*---------------*)*/
 462
 463 %token <Ast_c.info> TDefine
 464 %token <(string * Ast_c.info)> TDefParamVariadic
 465
 466 /*(* disappear after fix_tokens_define *)*/
 467 %token <Ast_c.info> TCppEscapedNewline
 468
 469 %token <Ast_c.info> TCppConcatOp
 470
 471 /*(* appear    after fix_tokens_define *)*/
 472 %token <Ast_c.info> TOParDefine
 473 %token <Ast_c.info> TOBraceDefineInit
 474
 475 %token <(string * Ast_c.info)> TIdentDefine /*(* same *)*/
 476 %token <Ast_c.info>            TDefEOL      /*(* same *)*/
 477
 478
 479 /*(*---------------*)*/
 480 /*(* include       *)*/
 481 /*(*---------------*)*/
 482
 483
 484 /*(* used only in lexer_c, then transformed in comment or splitted in tokens *)*/
 485 %token <(string * string * bool ref * Ast_c.info)> TInclude
 486
 487 /*(* tokens coming from above, generated in parse_c from TInclude, etc *)*/
 488 %token <(Ast_c.info * bool ref)> TIncludeStart
 489 %token <(string * Ast_c.info)> TIncludeFilename
 490
 491
 492 /*(*---------------*)*/
 493 /*(* ifdef         *)*/
 494 /*(*---------------*)*/
 495
 496 /*(* coupling: Token_helpers.is_cpp_instruction *)*/
 497 %token <((int * int) option ref * Ast_c.info)>
 498   TIfdef TIfdefelse TIfdefelif TEndif
 499 %token <(bool * (int * int) option ref * Ast_c.info)>
 500   TIfdefBool TIfdefMisc TIfdefVersion
 501
 502 /*(*---------------*)*/
 503 /*(* other         *)*/
 504 /*(*---------------*)*/
 505
 506
 507 %token <string * Ast_c.info> TUndef
 508
 509 %token <Ast_c.info> TCppDirectiveOther
 510
 511 /*(*---------------*)*/
 512 /*(* macro use     *)*/
 513 /*(*---------------*)*/
 514
 515 /*(* appear  after fix_tokens_cpp, cf also parsing_hacks#hint *)*/
 516
 517 %token <(string * Ast_c.info)>            TMacroAttr
 518 %token <(string * Ast_c.info)>            TMacroStmt
 519 %token <(string * Ast_c.info)> TMacroIdentBuilder
 520 /*(* no need value for the moment *)*/
 521 %token <(string * Ast_c.info)>            TMacroString
 522 %token <(string * Ast_c.info)> TMacroDecl
 523 %token <Ast_c.info>            TMacroDeclConst
 524
 525 %token <(string * Ast_c.info)> TMacroIterator
 526 /*(*
 527 %token <(string * Ast_c.info)> TMacroTop
 528 %token <(string * Ast_c.info)> TMacroStructDecl
 529 *)*/
 530
 531 %token <(string * Ast_c.info)>            TMacroAttrStorage
 532
 533
 534 /*(*---------------*)*/
 535 /*(* other         *)*/
 536 /*(*---------------*)*/
 537
 538
 539 /*(* should disappear after parsing_hack *)*/
 540 %token <Ast_c.info> TCommentSkipTagStart TCommentSkipTagEnd
 541
 542
 543 /*(* appear after parsing_hack *)*/
 544 %token <Ast_c.info> TCParEOL
 545
 546 %token <Ast_c.info> TAction
 547
 548
 549 /*(* TCommentMisc still useful ? obsolete ? *)*/
 550 %token <Ast_c.info> TCommentMisc
 551 %token <(Token_c.cppcommentkind * Ast_c.info)> TCommentCpp
 552
 553
 554 /*(*-----------------------------------------*)*/
 555 %token <Ast_c.info> EOF
 556
 557 /*(*-----------------------------------------*)*/
 558
 559 /*(* must be at the top so that it has the lowest priority *)*/
 560 %nonassoc SHIFTHERE
 561
 562 %nonassoc Telse
 563
 564 %left TOrLog
 565 %left TAndLog
 566 %left TOr
 567 %left TXor
 568 %left TAnd
 569 %left TEqEq TNotEq
 570 %left TInf TSup TInfEq TSupEq
 571 %left TShl TShr
 572 %left TPlus TMinus
 573 %left TMul TDiv TMod
 574
 575 /*(*************************************************************************)*/
 576 /*(* Rules type declaration *)*/
 577 /*(*************************************************************************)*/
 578
 579 %start main celem statement expr type_name
 580 %type <Ast_c.program> main
 581 %type <Ast_c.toplevel> celem
 582
 583 %type <Ast_c.statement> statement
 584 %type <Ast_c.expression> expr
 585 %type <Ast_c.fullType> type_name
 586
 587 %%
 588 /*(*************************************************************************)*/
 589 /*
 590 (* TOC:
 591  * toplevel (obsolete)
 592  *
 593  * ident
 594  * expression
 595  * statement
 596  * types with
 597  *   - left part (type_spec, qualif),
 598  *   - right part (declarator, abstract declarator)
 599  *   - aux part (parameters)
 600  * declaration, storage, initializers
 601  * struct
 602  * enum
 603  * cpp directives
 604  * celem (=~ main)
 605  *
 606  * generic workarounds (obrace, cbrace for context setting)
 607  * xxx_list, xxx_opt
 608  *)
 609 */
 610 /*(*************************************************************************)*/
 611
 612 /*(*************************************************************************)*/
 613 /*(* toplevel *)*/
 614 /*(*************************************************************************)*/
 615 /*(* no more used; now that use error recovery *)*/
 616
 617 main:  translation_unit EOF     { $1 }
 618
 619 translation_unit:
 620  | external_declaration
 621      { !LP._lexer_hint.context_stack <- [LP.InTopLevel];   [$1] }
 622  | translation_unit external_declaration
 623      { !LP._lexer_hint.context_stack <- [LP.InTopLevel]; $1 ++ [$2] }
 624
 625
 626
 627 /*(*************************************************************************)*/
 628 /*(* ident *)*/
 629 /*(*************************************************************************)*/
 630
 631 /*(* Why this ? Why not s/ident/TIdent ? cos there is multiple namespaces in C,
 632    * so a label can have the same name that a typedef, same for field and tags
 633    * hence sometimes the use of ident  instead of TIdent.
 634    *)*/
 635 ident:
 636  | TIdent       { $1 }
 637  | TypedefIdent { $1 }
 638
 639
 640 identifier:
 641  | TIdent       { $1 }
 642
 643 /*
 644 (* cppext: string concatenation of idents
 645  * also cppext: gccext: ##args for variadic macro
 646  *)
 647 */
 648 identifier_cpp:
 649  | TIdent
 650      { RegularName (mk_string_wrap $1) }
 651  | ident_extra_cpp { $1 }
 652
 653 ident_cpp:
 654  | TIdent
 655      { RegularName (mk_string_wrap $1) }
 656  | TypedefIdent
 657      { RegularName (mk_string_wrap $1) }
 658  | ident_extra_cpp { $1 }
 659
 660 ident_extra_cpp:
 661  | TIdent TCppConcatOp identifier_cpp_list
 662      {
 663        CppConcatenatedName (
 664          match $3 with
 665          | [] -> raise Impossible
 666          | (x,concatnull)::xs ->
 667              assert(null concatnull);
 668              (mk_string_wrap $1, [])::(x,[$2])::xs
 669        )
 670    }
 671  | TCppConcatOp TIdent
 672      { CppVariadicName (fst $2, [$1; snd $2]) }
 673  | TMacroIdentBuilder TOPar param_define_list TCPar
 674      { CppIdentBuilder ((fst $1, [snd $1;$2;$4]), $3) }
 675
 676 identifier_cpp_list:
 677  | TIdent { [mk_string_wrap $1, []] }
 678  | identifier_cpp_list TCppConcatOp TIdent { $1 ++ [mk_string_wrap $3, [$2]] }
 679
 680 /*(*************************************************************************)*/
 681 /*(* expr *)*/
 682 /*(*************************************************************************)*/
 683
 684 expr:
 685  | assign_expr             { $1 }
 686  | expr TComma assign_expr { mk_e (Sequence ($1,$3)) [$2] }
 687
 688 /*(* bugfix: in C grammar they put unary_expr, but in fact it must be
 689    * cast_expr, otherwise (int * ) xxx = &yy; is not allowed
 690    *)*/
 691 assign_expr:
 692  | cond_expr                     { $1 }
 693  | cast_expr TAssign assign_expr { mk_e(Assignment ($1,fst $2,$3)) [snd $2]}
 694  | cast_expr TEq     assign_expr { mk_e(Assignment ($1,SimpleAssign,$3)) [$2]}
 695
 696 /*(* gccext: allow optional then part hence gcc_opt_expr
 697    * bugfix: in C grammar they put TDotDot cond_expr, but in fact it must be
 698    * assign_expr, otherwise   pnp ? x : x = 0x388  is not allowed
 699    *)*/
 700 cond_expr:
 701  | arith_expr
 702      { $1 }
 703  | arith_expr TWhy gcc_opt_expr TDotDot assign_expr
 704      { mk_e (CondExpr ($1,$3,$5)) [$2;$4] }
 705
 706
 707 arith_expr:
 708  | cast_expr                     { $1 }
 709  | arith_expr TMul    arith_expr { mk_e(Binary ($1, Arith Mul,      $3)) [$2] }
 710  | arith_expr TDiv    arith_expr { mk_e(Binary ($1, Arith Div,      $3)) [$2] }
 711  | arith_expr TMod    arith_expr { mk_e(Binary ($1, Arith Mod,      $3)) [$2] }
 712  | arith_expr TPlus   arith_expr { mk_e(Binary ($1, Arith Plus,     $3)) [$2] }
 713  | arith_expr TMinus  arith_expr { mk_e(Binary ($1, Arith Minus,    $3)) [$2] }
 714  | arith_expr TShl    arith_expr { mk_e(Binary ($1, Arith DecLeft,  $3)) [$2] }
 715  | arith_expr TShr    arith_expr { mk_e(Binary ($1, Arith DecRight, $3)) [$2] }
 716  | arith_expr TInf    arith_expr { mk_e(Binary ($1, Logical Inf,    $3)) [$2] }
 717  | arith_expr TSup    arith_expr { mk_e(Binary ($1, Logical Sup,    $3)) [$2] }
 718  | arith_expr TInfEq  arith_expr { mk_e(Binary ($1, Logical InfEq,  $3)) [$2] }
 719  | arith_expr TSupEq  arith_expr { mk_e(Binary ($1, Logical SupEq,  $3)) [$2] }
 720  | arith_expr TEqEq   arith_expr { mk_e(Binary ($1, Logical Eq,     $3)) [$2] }
 721  | arith_expr TNotEq  arith_expr { mk_e(Binary ($1, Logical NotEq,  $3)) [$2] }
 722  | arith_expr TAnd    arith_expr { mk_e(Binary ($1, Arith And,      $3)) [$2] }
 723  | arith_expr TOr     arith_expr { mk_e(Binary ($1, Arith Or,       $3)) [$2] }
 724  | arith_expr TXor    arith_expr { mk_e(Binary ($1, Arith Xor,      $3)) [$2] }
 725  | arith_expr TAndLog arith_expr { mk_e(Binary ($1, Logical AndLog, $3)) [$2] }
 726  | arith_expr TOrLog  arith_expr { mk_e(Binary ($1, Logical OrLog,  $3)) [$2] }
 727
 728 cast_expr:
 729  | unary_expr                        { $1 }
 730  | topar2 type_name tcpar2 cast_expr { mk_e(Cast ($2, $4)) [$1;$3] }
 731
 732 unary_expr:
 733  | postfix_expr                    { $1 }
 734  | TInc unary_expr                 { mk_e(Infix ($2, Inc))    [$1] }
 735  | TDec unary_expr                 { mk_e(Infix ($2, Dec))    [$1] }
 736  | unary_op cast_expr              { mk_e(Unary ($2, fst $1)) [snd $1] }
 737  | Tsizeof unary_expr              { mk_e(SizeOfExpr ($2))    [$1] }
 738  | Tsizeof topar2 type_name tcpar2 { mk_e(SizeOfType ($3))    [$1;$2;$4] }
 739
 740 unary_op:
 741  | TAnd   { GetRef,     $1 }
 742  | TMul   { DeRef,      $1 }
 743  | TPlus  { UnPlus,     $1 }
 744  | TMinus { UnMinus,    $1 }
 745  | TTilde { Tilde,      $1 }
 746  | TBang  { Not,        $1 }
 747  /*(* gccext: have that a lot in old kernel to get address of local label.
 748     * cf gcc manual "local labels as values".
 749     *)*/
 750  | TAndLog { GetRefLabel, $1 }
 751
 752
 753
 754 postfix_expr:
 755  | primary_expr               { $1 }
 756  | postfix_expr TOCro expr TCCro
 757      { mk_e(ArrayAccess ($1, $3)) [$2;$4] }
 758  | postfix_expr TOPar argument_list_ne TCPar
 759      { mk_e(FunCall ($1, $3)) [$2;$4] }
 760  | postfix_expr TOPar  TCPar  { mk_e(FunCall ($1, [])) [$2;$3] }
 761  | postfix_expr TDot   ident_cpp { mk_e(RecordAccess   ($1,$3)) [$2] }
 762  | postfix_expr TPtrOp ident_cpp { mk_e(RecordPtAccess ($1,$3)) [$2] }
 763  | postfix_expr TInc          { mk_e(Postfix ($1, Inc)) [$2] }
 764  | postfix_expr TDec          { mk_e(Postfix ($1, Dec)) [$2] }
 765
 766  /*(* gccext: also called compound literals *)*/
 767  | topar2 type_name tcpar2 TOBrace TCBrace
 768      { mk_e(Constructor ($2, [])) [$1;$3;$4;$5] }
 769  | topar2 type_name tcpar2 TOBrace initialize_list gcc_comma_opt TCBrace
 770      { mk_e(Constructor ($2, List.rev $5)) ([$1;$3;$4;$7] ++ $6) }
 771
 772 primary_expr:
 773  | identifier_cpp  { mk_e(Ident  ($1)) [] }
 774  | TInt
 775     { let (str,(sign,base)) = fst $1 in
 776       mk_e(Constant (Int (str,Si(sign,base)))) [snd $1] }
 777  | TFloat  { mk_e(Constant (Float  (fst $1))) [snd $1] }
 778  | TString { mk_e(Constant (String (fst $1))) [snd $1] }
 779  | TChar   { mk_e(Constant (Char   (fst $1))) [snd $1] }
 780  | TOPar expr TCPar { mk_e(ParenExpr ($2)) [$1;$3] }  /*(* forunparser: *)*/
 781
 782  /*(* gccext: cppext: TODO better ast ? *)*/
 783  | TMacroString { mk_e(Constant (MultiString [fst $1])) [snd $1] }
 784  | string_elem string_list
 785      { mk_e(Constant (MultiString ["TODO: MultiString"])) ($1 ++ $2) }
 786
 787  /*(* gccext: allow statement as expressions via ({ statement }) *)*/
 788  | TOPar compound TCPar  { mk_e(StatementExpr ($2)) [$1;$3] }
 789
 790
 791
 792 /*(*----------------------------*)*/
 793 /*(* cppext: *)*/
 794 /*(*----------------------------*)*/
 795
 796 /*(* cppext: *)*/
 797 /*(* to avoid conflicts have to introduce a _not_empty (ne) version *)*/
 798 argument_ne:
 799  | assign_expr { Left $1 }
 800  | parameter_decl { Right (ArgType $1)  }
 801  | action_higherordermacro_ne { Right (ArgAction $1) }
 802
 803
 804 argument:
 805  | assign_expr { Left $1 }
 806  | parameter_decl { Right (ArgType $1)  }
 807  /*(* had conflicts before, but julia fixed them *)*/
 808  | action_higherordermacro { Right (ArgAction $1) }
 809
 810 action_higherordermacro_ne:
 811  | taction_list_ne
 812      { if null $1
 813        then ActMisc [Ast_c.fakeInfo()]
 814        else ActMisc $1
 815      }
 816
 817
 818 action_higherordermacro:
 819  | taction_list
 820      { if null $1
 821        then ActMisc [Ast_c.fakeInfo()]
 822        else ActMisc $1
 823      }
 824
 825
 826 /*(*----------------------------*)*/
 827 /*(* workarounds *)*/
 828 /*(*----------------------------*)*/
 829
 830 /*(* would like evalInt $1 but require too much info *)*/
 831 const_expr: cond_expr { $1  }
 832
 833
 834 topar2: TOPar { et "topar2" (); $1  }
 835 tcpar2: TCPar { et "tcpar2" (); $1 (*TODO? et ? sure ? c pas dt plutot ? *) }
 836
 837
 838
 839 /*(*************************************************************************)*/
 840 /*(* statement *)*/
 841 /*(*************************************************************************)*/
 842
 843 statement: statement2 { mk_st (fst $1) (snd $1) }
 844
 845 statement2:
 846  | labeled         { Labeled      (fst $1), snd $1 }
 847  | compound        { Compound     (fst $1), snd $1 }
 848  | expr_statement  { ExprStatement(fst $1), snd $1 }
 849  | selection       { Selection    (fst $1), snd $1 ++ [fakeInfo()] }
 850  | iteration       { Iteration    (fst $1), snd $1 ++ [fakeInfo()] }
 851  | jump TPtVirg    { Jump         (fst $1), snd $1 ++ [$2] }
 852
 853  /*(* gccext: *)*/
 854  | Tasm TOPar asmbody TCPar TPtVirg             { Asm $3, [$1;$2;$4;$5] }
 855  | Tasm Tvolatile TOPar asmbody TCPar TPtVirg   { Asm $4, [$1;$2;$3;$5;$6] }
 856
 857  /*(* cppext: *)*/
 858  | TMacroStmt { MacroStmt, [snd $1] }
 859
 860
 861
 862
 863 /*(* note that case 1: case 2: i++;    would be correctly parsed, but with
 864    * a Case  (1, (Case (2, i++)))  :(
 865    *)*/
 866 labeled:
 867  | ident_cpp        TDotDot statement   { Label ($1, $3),  [$2] }
 868  | Tcase const_expr TDotDot statement   { Case ($2, $4),       [$1; $3] }
 869  | Tcase const_expr TEllipsis const_expr TDotDot statement
 870      { CaseRange ($2, $4, $6), [$1;$3;$5] } /*(* gccext: allow range *)*/
 871  | Tdefault         TDotDot statement   { Default $3,             [$1; $2] }
 872
 873 end_labeled:
 874  /*(* gccext:  allow toto: }
 875     * was generating each 30 shift/Reduce conflicts,
 876     * mais ca va, ca fait ce qu'il faut.
 877     * update: julia fixed the problem by introducing end_labeled
 878     * and modifying below stat_or_decl_list
 879     *)*/
 880  | ident_cpp            TDotDot
 881      { Label ($1, (mk_st (ExprStatement None) Ast_c.noii)), [$2] }
 882  | Tcase const_expr TDotDot
 883      { Case ($2, (mk_st (ExprStatement None) Ast_c.noii)), [$1;$3] }
 884  | Tdefault         TDotDot
 885      { Default (mk_st (ExprStatement None) Ast_c.noii),    [$1; $2] }
 886
 887
 888
 889
 890
 891 compound: tobrace compound2 tcbrace { $2, [$1; $3]  }
 892
 893
 894 /*
 895 (* cppext: because of cpp, some stuff looks like declaration but are in
 896  * fact statement but too hard to figure out, and if parse them as
 897  * expression, then we force to have first decls and then exprs, then
 898  * will have a parse error. So easier to let mix decl/statement.
 899  * Moreover it helps to not make such a difference between decl and
 900  * statement for further coccinelle phases to factorize code.
 901 *)*/
 902 compound2:
 903  |                   { ([]) }
 904  | stat_or_decl_list { $1 }
 905
 906
 907 stat_or_decl_list:
 908  | stat_or_decl                   { [$1] }
 909  /*(* gccext: to avoid conflicts, cf end_labeled above *)*/
 910  | end_labeled  { [StmtElem (mk_st (Labeled  (fst $1)) (snd $1))] }
 911  /*(* old: conflicts | stat_or_decl_list stat_or_decl { $1 ++ [$2] } *)*/
 912  | stat_or_decl stat_or_decl_list { $1 :: $2 }
 913
 914 stat_or_decl:
 915  | decl      { StmtElem (mk_st (Decl ($1 Ast_c.LocalDecl)) Ast_c.noii) }
 916  | statement { StmtElem $1 }
 917
 918   /*(* gccext: *)*/
 919  | function_definition { StmtElem (mk_st (NestedFunc $1) Ast_c.noii) }
 920
 921  /* (* cppext: *)*/
 922  | cpp_directive
 923      { CppDirectiveStmt $1 }
 924  | cpp_ifdef_directive/*(* stat_or_decl_list ...*)*/
 925      { IfdefStmt $1 }
 926
 927
 928
 929
 930
 931 expr_statement:
 932  | TPtVirg      { None,    [$1] }
 933  | expr TPtVirg { Some $1, [$2] }
 934
 935 selection:
 936  | Tif TOPar expr TCPar statement              %prec SHIFTHERE
 937      { If ($3, $5, (mk_st (ExprStatement None) Ast_c.noii)),   [$1;$2;$4] }
 938  | Tif TOPar expr TCPar statement Telse statement
 939      { If ($3, $5, $7),  [$1;$2;$4;$6] }
 940  | Tswitch TOPar expr TCPar statement
 941      { Switch ($3,$5),   [$1;$2;$4]  }
 942
 943 iteration:
 944  | Twhile TOPar expr TCPar statement
 945      { While ($3,$5),                [$1;$2;$4] }
 946  | Tdo statement Twhile TOPar expr TCPar TPtVirg
 947      { DoWhile ($2,$5),              [$1;$3;$4;$6;$7] }
 948  | Tfor TOPar expr_statement expr_statement TCPar statement
 949      { For ($3,$4,(None, []),$6),    [$1;$2;$5]}
 950  | Tfor TOPar expr_statement expr_statement expr TCPar statement
 951      { For ($3,$4,(Some $5, []),$7), [$1;$2;$6] }
 952  /*(* c++ext: for(int i = 0; i < n; i++)*)*/
 953  | Tfor TOPar decl expr_statement expr_opt TCPar statement
 954      {
 955        (* pr2 "DECL in for"; *)
 956        MacroIteration ("toto", [], $7),[$1;$2;$6] (* TODOfake ast, TODO need decl2 ? *)
 957      }
 958  /*(* cppext: *)*/
 959  | TMacroIterator TOPar argument_list_ne TCPar statement
 960      { MacroIteration (fst $1, $3, $5), [snd $1;$2;$4] }
 961  | TMacroIterator TOPar TCPar statement
 962      { MacroIteration (fst $1, [], $4), [snd $1;$2;$3] }
 963
 964 /*(* the ';' in the caller grammar rule will be appended to the infos *)*/
 965 jump:
 966  | Tgoto ident_cpp  { Goto ($2),  [$1] }
 967  | Tcontinue    { Continue,       [$1] }
 968  | Tbreak       { Break,          [$1] }
 969  | Treturn      { Return,         [$1] }
 970  | Treturn expr { ReturnExpr $2,  [$1] }
 971  | Tgoto TMul expr { GotoComputed $3, [$1;$2] }
 972
 973
 974
 975 /*(*----------------------------*)*/
 976 /*(* gccext: *)*/
 977 /*(*----------------------------*)*/
 978 string_elem:
 979  | TString { [snd $1] }
 980  /*(* cppext:  ex= printk (KERN_INFO "xxx" UTS_RELEASE)  *)*/
 981  | TMacroString { [snd $1] }
 982
 983
 984 asmbody:
 985  | string_list colon_asm_list  { $1, $2 }
 986  | string_list { $1, [] } /*(* in old kernel *)*/
 987
 988
 989 colon_asm: TDotDot colon_option_list { Colon $2, [$1]   }
 990
 991 colon_option:
 992  | TString                      { ColonMisc, [snd $1] }
 993  | TString TOPar asm_expr TCPar { ColonExpr $3, [snd $1; $2;$4] }
 994  /*(* cppext: certainly a macro *)*/
 995  | TOCro identifier TCCro TString TOPar asm_expr TCPar
 996      { ColonExpr $6, [$1;snd $2;$3;snd $4; $5; $7 ] }
 997  | identifier                       { ColonMisc, [snd $1] }
 998  | /*(* empty *)*/                  { ColonMisc, [] }
 999
1000 asm_expr: assign_expr { $1 }
1001
1002 /*(*************************************************************************)*/
1003 /*(* types *)*/
1004 /*(*************************************************************************)*/
1005
1006
1007 /*(*-----------------------------------------------------------------------*)*/
1008 /*(* Type spec, left part of a type *)*/
1009 /*(*-----------------------------------------------------------------------*)*/
1010 type_spec2:
1011  | Tvoid                { Right3 (BaseType Void),            [$1] }
1012  | Tchar                { Right3 (BaseType (IntType CChar)), [$1]}
1013  | Tint                 { Right3 (BaseType (IntType (Si (Signed,CInt)))), [$1]}
1014  | Tfloat               { Right3 (BaseType (FloatType CFloat)),  [$1]}
1015  | Tdouble              { Right3 (BaseType (FloatType CDouble)), [$1] }
1016  | Tsize_t              { Right3 (BaseType SizeType),            [$1] }
1017  | Tssize_t             { Right3 (BaseType SSizeType),           [$1] }
1018  | Tptrdiff_t           { Right3 (BaseType PtrDiffType),         [$1] }
1019  | Tshort               { Middle3 Short,  [$1]}
1020  | Tlong                { Middle3 Long,   [$1]}
1021  | Tsigned              { Left3 Signed,   [$1]}
1022  | Tunsigned            { Left3 UnSigned, [$1]}
1023  | struct_or_union_spec { Right3 (fst $1), snd $1 }
1024  | enum_spec            { Right3 (fst $1), snd $1 }
1025
1026  /*
1027  (* parse_typedef_fix1: cant put: TIdent {} cos it make the grammar
1028   * ambiguous, generates lots of conflicts => we must
1029   * use some tricks: we make the lexer and parser cooperate, cf lexerParser.ml.
1030   *
1031   * parse_typedef_fix2: this is not enough, and you must use
1032   * parse_typedef_fix2 to fully manage typedef problems in grammar.
1033   *
1034   * parse_typedef_fix3:
1035   *
1036   * parse_typedef_fix4: try also to do now some consistency checking in
1037   * Parse_c
1038   *)*/
1039  | TypedefIdent
1040      { let name = RegularName (mk_string_wrap $1) in
1041        Right3 (TypeName (name, Ast_c.noTypedefDef())),[] }
1042
1043  | Ttypeof TOPar assign_expr TCPar { Right3 (TypeOfExpr ($3)), [$1;$2;$4] }
1044  | Ttypeof TOPar type_name   TCPar { Right3 (TypeOfType ($3)), [$1;$2;$4] }
1045
1046 /*(*----------------------------*)*/
1047 /*(* workarounds *)*/
1048 /*(*----------------------------*)*/
1049
1050 type_spec: type_spec2    { dt "type" (); $1   }
1051
1052 /*(*-----------------------------------------------------------------------*)*/
1053 /*(* Qualifiers *)*/
1054 /*(*-----------------------------------------------------------------------*)*/
1055
1056 type_qualif:
1057  | Tconst    { {const=true  ; volatile=false}, $1 }
1058  | Tvolatile { {const=false ; volatile=true},  $1 }
1059  /*(* C99 *)*/
1060  | Trestrict { (* TODO *) {const=false ; volatile=false},  $1 }
1061
1062
1063 /*(*-----------------------------------------------------------------------*)*/
1064 /*(* gccext: attributes  *)*/
1065 /*(*-----------------------------------------------------------------------*)*/
1066
1067 attribute:
1068  | Tattribute TOPar /*stuff*/ TCPar { raise Todo }
1069  /*(* cppext: *)*/
1070  | TMacroAttr { Attribute (fst $1), [snd $1] }
1071
1072 attribute_storage:
1073  | TMacroAttrStorage { $1 }
1074
1075 type_qualif_attr:
1076  | type_qualif { $1 }
1077 /*(*TODO !!!!! *)*/
1078  | TMacroAttr { {const=true  ; volatile=false}, snd $1   }
1079
1080 /*(*-----------------------------------------------------------------------*)*/
1081 /*(* Declarator, right part of a type + second part of decl (the ident)  *)*/
1082 /*(*-----------------------------------------------------------------------*)*/
1083
1084 /*
1085 (* declarator return a couple:
1086  *  (name, partial type (a function to be applied to return type))
1087  *
1088  * when int* f(int) we must return Func(Pointer int,int) and not
1089  * Pointer (Func(int,int)
1090  *)*/
1091
1092 declarator:
1093  | pointer direct_d { (fst $2, fun x -> x +> $1 +> (snd $2)  ) }
1094  | direct_d         { $1  }
1095
1096 /*(* so must do  int * const p; if the pointer is constant, not the pointee *)*/
1097 pointer:
1098  | TMul                   { fun x -> mk_ty (Pointer x) [$1] }
1099  | TMul pointer           { fun x -> mk_ty (Pointer ($2 x)) [$1] }
1100  | TMul type_qualif_list
1101      { fun x -> ($2.qualifD, mk_tybis (Pointer x) [$1])}
1102  | TMul type_qualif_list pointer
1103      { fun x -> ($2.qualifD, mk_tybis (Pointer ($3 x)) [$1]) }
1104
1105
1106 direct_d:
1107  | identifier_cpp
1108      { ($1, fun x -> x) }
1109  | TOPar declarator TCPar      /*(* forunparser: old: $2 *)*/
1110      { (fst $2, fun x -> mk_ty (ParenType ((snd $2) x)) [$1;$3]) }
1111  | direct_d tocro            tccro
1112      { (fst $1,fun x->(snd $1) (mk_ty (Array (None,x)) [$2;$3])) }
1113  | direct_d tocro const_expr tccro
1114      { (fst $1,fun x->(snd $1) (mk_ty (Array (Some $3,x)) [$2;$4])) }
1115  | direct_d topar            tcpar
1116      { (fst $1,
1117        fun x->(snd $1)
1118          (mk_ty (FunctionType (x,(([],(false, []))))) [$2;$3]))
1119      }
1120  | direct_d topar parameter_type_list tcpar
1121      { (fst $1,fun x->(snd $1)
1122        (mk_ty (FunctionType (x, $3)) [$2;$4]))
1123      }
1124
1125
1126 /*(*----------------------------*)*/
1127 /*(* workarounds *)*/
1128 /*(*----------------------------*)*/
1129
1130 tocro: TOCro { et "tocro" ();$1 }
1131 tccro: TCCro { dt "tccro" ();$1 }
1132
1133 /*(*-----------------------------------------------------------------------*)*/
1134 abstract_declarator:
1135  | pointer                            { $1 }
1136  |         direct_abstract_declarator { $1 }
1137  | pointer direct_abstract_declarator { fun x -> x +> $2 +> $1 }
1138
1139 direct_abstract_declarator:
1140  | TOPar abstract_declarator TCPar /*(* forunparser: old: $2 *)*/
1141      { fun x -> mk_ty (ParenType ($2 x)) [$1;$3] }
1142
1143  | TOCro            TCCro
1144      { fun x -> mk_ty (Array (None, x)) [$1;$2] }
1145  | TOCro const_expr TCCro
1146      { fun x -> mk_ty (Array (Some $2, x)) [$1;$3] }
1147  | direct_abstract_declarator TOCro            TCCro
1148      { fun x -> $1 (mk_ty (Array (None, x))  [$2;$3]) }
1149  | direct_abstract_declarator TOCro const_expr TCCro
1150      { fun x -> $1 (mk_ty (Array (Some $3,x))  [$2;$4]) }
1151  | TOPar TCPar
1152      { fun x -> mk_ty (FunctionType (x, ([], (false,  [])))) [$1;$2] }
1153  | topar parameter_type_list tcpar
1154      { fun x -> mk_ty (FunctionType (x, $2))  [$1;$3] }
1155 /*(* subtle: here must also use topar, not TOPar, otherwise if have for
1156    * instance   (xxx ( * )(xxx)) cast, then the second xxx may still be a Tident
1157    * but we want to reduce topar, to set the InParameter so that
1158    * parsing_hack can get a chance to change the type of xxx into a typedef.
1159    * That's an example where parsing_hack and the lookahead of ocamlyacc does
1160    * not go very well together ... we got the info too late. We got
1161    * a similar pb with xxx xxx; declaration, cf parsing_hack.ml and the
1162    * "disable typedef cos special case ..." message.
1163 *)*/
1164  | direct_abstract_declarator topar tcpar
1165      { fun x -> $1 (mk_ty (FunctionType (x, (([], (false, []))))) [$2;$3]) }
1166  | direct_abstract_declarator topar parameter_type_list tcpar
1167      { fun x -> $1 (mk_ty (FunctionType (x, $3)) [$2;$4]) }
1168
1169 /*(*-----------------------------------------------------------------------*)*/
1170 /*(* Parameters (use decl_spec not type_spec just for 'register') *)*/
1171 /*(*-----------------------------------------------------------------------*)*/
1172 parameter_type_list:
1173  | parameter_list                  { ($1, (false, []))}
1174  | parameter_list TComma TEllipsis { ($1, (true,  [$2;$3])) }
1175
1176
1177 parameter_decl2:
1178  | decl_spec declaratorp
1179      { let ((returnType,hasreg),iihasreg) = fixDeclSpecForParam $1 in
1180        let (name, ftyp) = $2 in
1181        { p_namei = Some (name);
1182          p_type = ftyp returnType;
1183          p_register = (hasreg, iihasreg);
1184        }
1185      }
1186  | decl_spec abstract_declaratorp
1187      { let ((returnType,hasreg), iihasreg) = fixDeclSpecForParam $1 in
1188        { p_namei = None;
1189          p_type = $2 returnType;
1190          p_register = hasreg, iihasreg;
1191        }
1192      }
1193  | decl_spec
1194      { let ((returnType,hasreg), iihasreg) = fixDeclSpecForParam $1 in
1195        { p_namei = None;
1196          p_type = returnType;
1197          p_register = hasreg, iihasreg;
1198        }
1199      }
1200
1201
1202 /*(*----------------------------*)*/
1203 /*(* workarounds *)*/
1204 /*(*----------------------------*)*/
1205
1206 parameter_decl: parameter_decl2 { et "param" ();  $1 }
1207  | attributes parameter_decl2 { et "param" (); $2 }
1208
1209 declaratorp:
1210  | declarator  { LP.add_ident (str_of_name (fst $1)); $1 }
1211  /*(* gccext: *)*/
1212  | attributes declarator   { LP.add_ident (str_of_name (fst $2)); $2 }
1213  | declarator attributes   { LP.add_ident (str_of_name (fst $1)); $1 }
1214
1215 abstract_declaratorp:
1216  | abstract_declarator { $1 }
1217  /*(* gccext: *)*/
1218  | attributes abstract_declarator { $2 }
1219
1220 /*(*-----------------------------------------------------------------------*)*/
1221 /*(* helper type rules *)*/
1222 /*(*-----------------------------------------------------------------------*)*/
1223
1224 /*(* for struct and also typename *)*/
1225 /*(* cant put decl_spec cos no storage is allowed for field struct *)*/
1226 spec_qualif_list2:
1227  | type_spec                    { addTypeD ($1, nullDecl) }
1228  | type_qualif                  { {nullDecl with qualifD = (fst $1,[snd $1])}}
1229  | type_spec   spec_qualif_list { addTypeD ($1,$2)   }
1230  | type_qualif spec_qualif_list { addQualifD ($1,$2) }
1231
1232 spec_qualif_list: spec_qualif_list2            {  dt "spec_qualif" (); $1 }
1233
1234
1235 /*(* for pointers in direct_declarator and abstract_declarator *)*/
1236 type_qualif_list:
1237  | type_qualif_attr                  { {nullDecl with qualifD = (fst $1,[snd $1])} }
1238  | type_qualif_list type_qualif_attr { addQualifD ($2,$1) }
1239
1240
1241
1242
1243
1244
1245 /*(*-----------------------------------------------------------------------*)*/
1246 /*(* xxx_type_id *)*/
1247 /*(*-----------------------------------------------------------------------*)*/
1248
1249 type_name:
1250  | spec_qualif_list
1251      { let (returnType, _) = fixDeclSpecForDecl $1 in  returnType }
1252  | spec_qualif_list abstract_declaratort
1253      { let (returnType, _) = fixDeclSpecForDecl $1 in $2 returnType }
1254
1255
1256
1257 abstract_declaratort:
1258  | abstract_declarator { $1 }
1259  /*(* gccext: *)*/
1260  | attributes abstract_declarator { $2 }
1261
1262
1263 /*(*************************************************************************)*/
1264 /*(* declaration and initializers *)*/
1265 /*(*************************************************************************)*/
1266
1267 decl2:
1268  | decl_spec TPtVirg
1269      { function local ->
1270        let (returnType,storage) = fixDeclSpecForDecl $1 in
1271        let iistart = Ast_c.fakeInfo () in
1272        DeclList ([{v_namei = None; v_type = returnType;
1273                    v_storage = unwrap storage; v_local = local;
1274                    v_attr = Ast_c.noattr;
1275                    v_type_bis = ref None;
1276                 },[]],
1277                 ($2::iistart::snd storage))
1278      }
1279  | decl_spec init_declarator_list TPtVirg
1280      { function local ->
1281        let (returnType,storage) = fixDeclSpecForDecl $1 in
1282        let iistart = Ast_c.fakeInfo () in
1283        DeclList (
1284          ($2 +> List.map (fun ((((name,f),attrs), ini), iivirg) ->
1285            let s = str_of_name name in
1286            let iniopt =
1287              match ini with
1288              | None -> None
1289              | Some (ini, iini) -> Some (iini, ini)
1290            in
1291            if fst (unwrap storage) =*= StoTypedef
1292            then LP.add_typedef s;
1293            {v_namei = Some (name, iniopt);
1294             v_type = f returnType;
1295             v_storage = unwrap storage;
1296             v_local = local;
1297             v_attr = attrs;
1298             v_type_bis = ref None;
1299            },
1300            iivirg
1301          )
1302          ),  ($3::iistart::snd storage))
1303      }
1304  /*(* cppext: *)*/
1305
1306  | TMacroDecl TOPar argument_list TCPar TPtVirg
1307      { function _ ->
1308        MacroDecl ((fst $1, $3), [snd $1;$2;$4;$5;fakeInfo()]) }
1309  | Tstatic TMacroDecl TOPar argument_list TCPar TPtVirg
1310      { function _ ->
1311        MacroDecl ((fst $2, $4), [snd $2;$3;$5;$6;fakeInfo();$1]) }
1312  | Tstatic TMacroDeclConst TMacroDecl TOPar argument_list TCPar TPtVirg
1313      { function _ ->
1314        MacroDecl ((fst $3, $5), [snd $3;$4;$6;$7;fakeInfo();$1;$2])}
1315
1316
1317 /*(*-----------------------------------------------------------------------*)*/
1318 decl_spec2:
1319  | storage_class_spec      { {nullDecl with storageD = (fst $1, [snd $1]) } }
1320  | type_spec               { addTypeD ($1,nullDecl) }
1321  | type_qualif             { {nullDecl with qualifD  = (fst $1, [snd $1]) } }
1322  | Tinline                 { {nullDecl with inlineD = (true, [$1]) } }
1323  | storage_class_spec decl_spec2 { addStorageD ($1, $2) }
1324  | type_spec          decl_spec2 { addTypeD    ($1, $2) }
1325  | type_qualif        decl_spec2 { addQualifD  ($1, $2) }
1326  | Tinline            decl_spec2 { addInlineD ((true, $1), $2) }
1327
1328 /*(* can simplify by putting all in _opt ? must have at least one otherwise
1329    *  decl_list is ambiguous ? (no cos have ';' between decl)
1330    *)*/
1331
1332
1333 storage_class_spec2:
1334  | Tstatic      { Sto Static,  $1 }
1335  | Textern      { Sto Extern,  $1 }
1336  | Tauto        { Sto Auto,    $1 }
1337  | Tregister    { Sto Register,$1 }
1338  | Ttypedef     { StoTypedef,  $1 }
1339
1340 storage_class_spec:
1341  /*(* gccext: *)*/
1342  | storage_class_spec2 { $1 }
1343  | storage_class_spec2 attribute_storage_list { $1 (* TODO *) }
1344
1345
1346
1347 /*(*----------------------------*)*/
1348 /*(* workarounds *)*/
1349 /*(*----------------------------*)*/
1350
1351 decl:      decl2         { et "decl" (); $1 }
1352 decl_spec: decl_spec2    { dt "declspec" (); $1  }
1353
1354 /*(*-----------------------------------------------------------------------*)*/
1355 /*(* declarators (right part of type and variable) *)*/
1356 /*(*-----------------------------------------------------------------------*)*/
1357 init_declarator2:
1358  | declaratori                  { ($1, None) }
1359  | declaratori teq initialize   { ($1, Some ($3, $2)) }
1360
1361
1362
1363 /*(*----------------------------*)*/
1364 /*(* workarounds *)*/
1365 /*(*----------------------------*)*/
1366 teq: TEq  { et "teq" (); $1 }
1367
1368 init_declarator: init_declarator2  { dt "init" (); $1 }
1369
1370
1371 /*(*----------------------------*)*/
1372 /*(* gccext: *)*/
1373 /*(*----------------------------*)*/
1374
1375 declaratori:
1376  | declarator              { LP.add_ident (str_of_name (fst $1)); $1, Ast_c.noattr }
1377  /*(* gccext: *)*/
1378  | declarator gcc_asm_decl { LP.add_ident (str_of_name (fst $1)); $1, Ast_c.noattr }
1379  /*(* gccext: *)*/
1380  | attributes declarator   { LP.add_ident (str_of_name (fst $2)); $2, $1 }
1381  | declarator attributes   { LP.add_ident (str_of_name (fst $1)); $1, Ast_c.noattr (* TODO *) }
1382
1383
1384
1385 gcc_asm_decl:
1386  | Tasm TOPar asmbody TCPar              {  }
1387  | Tasm Tvolatile TOPar asmbody TCPar   {  }
1388
1389
1390 /*(*-----------------------------------------------------------------------*)*/
1391 initialize:
1392  | assign_expr
1393      { InitExpr $1,                [] }
1394  | tobrace_ini initialize_list gcc_comma_opt_struct  tcbrace_ini
1395      { InitList (List.rev $2),     [$1;$4]++$3 }
1396  | tobrace_ini tcbrace_ini
1397      { InitList [],       [$1;$2] } /*(* gccext: *)*/
1398
1399
1400 /*
1401 (* opti: This time we use the weird order of non-terminal which requires in
1402  * the "caller" to do a List.rev cos quite critical. With this weird order it
1403  * allows yacc to use a constant stack space instead of exploding if we would
1404  * do a  'initialize2 Tcomma initialize_list'.
1405  *)
1406 */
1407 initialize_list:
1408  | initialize2                        { [$1,   []] }
1409  | initialize_list TComma initialize2 { ($3,  [$2])::$1 }
1410
1411
1412 /*(* gccext: condexpr and no assign_expr cos can have ambiguity with comma *)*/
1413 initialize2:
1414  | cond_expr
1415      { InitExpr $1,   [] }
1416  | tobrace_ini initialize_list gcc_comma_opt_struct tcbrace_ini
1417      { InitList (List.rev $2),   [$1;$4]++$3 }
1418  | tobrace_ini tcbrace_ini
1419      { InitList [],  [$1;$2]  }
1420
1421  /*(* gccext: labeled elements, a.k.a designators *)*/
1422  | designator_list TEq initialize2
1423      { InitDesignators ($1, $3), [$2] }
1424
1425  /*(* gccext: old format *)*/
1426  | ident TDotDot initialize2
1427      { InitFieldOld (fst $1, $3),     [snd $1; $2] } /*(* in old kernel *)*/
1428 /* conflict
1429  | TOCro const_expr TCCro initialize2
1430      { InitIndexOld ($2, $4),    [$1;$3] }
1431 */
1432
1433
1434
1435 /*(* they can be nested, can have a .x[3].y *)*/
1436 designator:
1437  | TDot ident
1438      { DesignatorField (fst $2), [$1;snd $2] }
1439  | TOCro const_expr TCCro
1440      { DesignatorIndex ($2),  [$1;$3] }
1441  | TOCro const_expr TEllipsis const_expr TCCro
1442      { DesignatorRange ($2, $4),  [$1;$3;$5] }
1443
1444
1445 /*(*----------------------------*)*/
1446 /*(* workarounds *)*/
1447 /*(*----------------------------*)*/
1448
1449 gcc_comma_opt_struct:
1450  | TComma {  [$1] }
1451  | /*(* empty *)*/  {  [Ast_c.fakeInfo() +> Ast_c.rewrap_str ","]  }
1452
1453
1454
1455
1456
1457
1458
1459
1460 /*(*************************************************************************)*/
1461 /*(* struct *)*/
1462 /*(*************************************************************************)*/
1463
1464 s_or_u_spec2:
1465  | struct_or_union ident tobrace_struct struct_decl_list_gcc tcbrace_struct
1466      { StructUnion (fst $1, Some (fst $2), $4),  [snd $1;snd $2;$3;$5]  }
1467  | struct_or_union       tobrace_struct struct_decl_list_gcc tcbrace_struct
1468      { StructUnion (fst $1, None, $3), [snd $1;$2;$4] }
1469  | struct_or_union ident
1470      { StructUnionName (fst $1, fst $2), [snd $1;snd $2] }
1471
1472 struct_or_union2:
1473  | Tstruct   { Struct, $1 }
1474  | Tunion    { Union, $1 }
1475  /*(* gccext: *)*/
1476  | Tstruct attributes   { Struct, $1 (* TODO *) }
1477  | Tunion  attributes   { Union, $1  (* TODO *) }
1478
1479
1480
1481 struct_decl2:
1482  | field_declaration { DeclarationField $1 }
1483  | TPtVirg { EmptyField $1  }
1484
1485  /*(* no conflict ? no need for a TMacroStruct ? apparently not as at struct
1486     * the rule are slightly different.
1487     *)*/
1488  | identifier TOPar argument_list TCPar TPtVirg
1489      { MacroDeclField ((fst $1, $3), [snd $1;$2;$4;$5;fakeInfo()]) }
1490
1491  /*(* cppext: *)*/
1492  | cpp_directive
1493      { CppDirectiveStruct $1 }
1494  | cpp_ifdef_directive/*(* struct_decl_list ... *)*/
1495      { IfdefStruct $1 }
1496
1497
1498 field_declaration:
1499  | spec_qualif_list struct_declarator_list TPtVirg
1500      {
1501        let (returnType,storage) = fixDeclSpecForDecl $1 in
1502        if fst (unwrap storage) <> NoSto
1503        then internal_error "parsing dont allow this";
1504
1505        FieldDeclList ($2 +> (List.map (fun (f, iivirg) ->
1506          f returnType, iivirg))
1507                          ,[$3])
1508          (* dont need to check if typedef or func initialised cos
1509           * grammar dont allow typedef nor initialiser in struct
1510           *)
1511      }
1512
1513  |  spec_qualif_list TPtVirg
1514      {
1515        (* gccext: allow empty elements if it is a structdef or enumdef *)
1516        let (returnType,storage) = fixDeclSpecForDecl $1 in
1517        if fst (unwrap storage) <> NoSto
1518        then internal_error "parsing dont allow this";
1519
1520        FieldDeclList ([(Simple (None, returnType)) , []], [$2])
1521      }
1522
1523
1524
1525
1526
1527 struct_declarator:
1528  | declaratorsd
1529      { (fun x -> Simple   (Some (fst $1), (snd $1) x)) }
1530  | dotdot const_expr2
1531      { (fun x -> BitField (None, x, $1, $2)) }
1532  | declaratorsd dotdot const_expr2
1533      { (fun x -> BitField (Some (fst $1), ((snd $1) x), $2, $3)) }
1534
1535
1536 /*(*----------------------------*)*/
1537 /*(* workarounds *)*/
1538 /*(*----------------------------*)*/
1539 declaratorsd:
1540  | declarator { (*also ? LP.add_ident (fst (fst $1)); *) $1 }
1541  /*(* gccext: *)*/
1542  | attributes declarator   { $2 }
1543  | declarator attributes   { $1 }
1544
1545
1546
1547
1548 struct_or_union_spec: s_or_u_spec2 { dt "su" (); $1 }
1549 struct_or_union: struct_or_union2 { et "su" (); $1 }
1550 struct_decl: struct_decl2  { et "struct" (); $1 }
1551
1552 dotdot: TDotDot  { et "dotdot" (); $1 }
1553 const_expr2: const_expr { dt "const_expr2" (); $1 }
1554
1555 struct_decl_list_gcc:
1556  | struct_decl_list  { $1 }
1557  | /*(* empty *)*/       { [] } /*(* gccext: allow empty struct *)*/
1558
1559
1560 /*(*************************************************************************)*/
1561 /*(* enum *)*/
1562 /*(*************************************************************************)*/
1563 enum_spec:
1564  | Tenum        tobrace_enum enumerator_list gcc_comma_opt_struct tcbrace_enum
1565      { Enum (None,    $3),           [$1;$2;$5] ++ $4 }
1566  | Tenum ident  tobrace_enum enumerator_list gcc_comma_opt_struct tcbrace_enum
1567      { Enum (Some (fst $2), $4),     [$1; snd $2; $3;$6] ++ $5 }
1568  | Tenum ident
1569      { EnumName (fst $2),       [$1; snd $2] }
1570
1571 enumerator:
1572  | idente                 { $1, None     }
1573  | idente  TEq const_expr { $1, Some ($2, $3) }
1574
1575
1576 /*(*----------------------------*)*/
1577 /*(* workarounds *)*/
1578 /*(*----------------------------*)*/
1579
1580 idente: ident_cpp { LP.add_ident (str_of_name $1); $1 }
1581
1582
1583
1584 /*(*************************************************************************)*/
1585 /*(* function *)*/
1586 /*(*************************************************************************)*/
1587 function_definition: function_def    { fixFunc $1 }
1588
1589 decl_list:
1590  | decl           { [$1 Ast_c.LocalDecl]   }
1591  | decl_list decl { $1 ++ [$2 Ast_c.LocalDecl] }
1592
1593 /* hack : to drop when a better solution is found */
1594 cpp_directive_list:
1595  | cpp_directive                    { }
1596  | cpp_directive_list cpp_directive { }
1597
1598 function_def:
1599  | start_fun compound      { LP.del_scope(); ($1, $2, None) }
1600  | start_fun cpp_directive_list compound { LP.del_scope(); ($1, $3, None) }
1601  | start_fun decl_list compound      {
1602      (* TODO: undo the typedef added ? *)
1603      LP.del_scope();
1604      ($1, $3, Some $2)
1605    }
1606
1607 start_fun: start_fun2
1608   { LP.new_scope();
1609     fix_add_params_ident $1;
1610     (* toreput? !LP._lexer_hint.toplevel <- false;  *)
1611     $1
1612   }
1613
1614 start_fun2: decl_spec declaratorfd
1615      { let (returnType,storage) = fixDeclSpecForFuncDef $1 in
1616        let (id, attrs) = $2 in
1617        (fst id, fixOldCDecl ((snd id) returnType) , storage, attrs)
1618      }
1619
1620 /*(*----------------------------*)*/
1621 /*(* workarounds *)*/
1622 /*(*----------------------------*)*/
1623
1624 /* It would be very nice if we could make declarator aware that this is
1625 coming from a function definition.  Then on the ( and ) cases, it could
1626 set the state to something other than InParameter.  Then the case
1627 (TIdent (s, i1)::(TComma _|TCPar _)::_ , (TComma _ |TOPar _)::_ )
1628 in parsing_hacks.ml would not have to consider K&R variable declarations
1629 as typedefs.  Unfortunately, doing something about this problem seems to
1630 introduce conflicts in the parser. */
1631
1632 declaratorfd:
1633  | declarator { et "declaratorfd" (); $1, Ast_c.noattr }
1634  /*(* gccext: *)*/
1635  | attributes declarator   { et "declaratorfd" (); $2, $1 }
1636  | declarator attributes   { et "declaratorfd" (); $1, Ast_c.noattr }
1637
1638
1639
1640 /*(*************************************************************************)*/
1641 /*(* cpp directives *)*/
1642 /*(*************************************************************************)*/
1643
1644 cpp_directive:
1645  | TIncludeStart TIncludeFilename
1646      {
1647        let (i1, in_ifdef) = $1 in
1648        let (s, i2) = $2 in
1649
1650        (* redo some lexing work :( *)
1651        let inc_file =
1652          match () with
1653          | _ when s =~ "^\"\\(.*\\)\"$" ->
1654              Local (Common.split "/" (matched1 s))
1655          | _ when s =~ "^\\<\\(.*\\)\\>$" ->
1656              NonLocal (Common.split "/" (matched1 s))
1657          | _ ->
1658              Weird s
1659        in
1660        Include { i_include = (inc_file, [i1;i2]);
1661                  i_rel_pos = Ast_c.noRelPos();
1662                  i_is_in_ifdef = !in_ifdef;
1663                  i_content = Ast_c.noi_content;
1664        }
1665      }
1666
1667  | TDefine TIdentDefine define_val TDefEOL
1668      { Define ((fst $2, [$1; snd $2;$4]), (DefineVar, $3)) }
1669
1670  /*
1671  (* The TOParDefine is introduced to avoid ambiguity with previous rules.
1672   * A TOParDefine is a TOPar that was just next to the ident.
1673   *)*/
1674  | TDefine TIdentDefine TOParDefine param_define_list TCPar define_val TDefEOL
1675      { Define
1676          ((fst $2, [$1; snd $2;$7]),
1677            (DefineFunc ($4, [$3;$5]), $6))
1678      }
1679
1680  | TUndef             { Undef (fst $1, [snd $1]) }
1681  | TCppDirectiveOther { PragmaAndCo ([$1]) }
1682
1683
1684
1685
1686
1687 /*(* perhaps better to use assign_expr ? but in that case need
1688    * do a assign_expr_of_string in parse_c
1689    *)*/
1690 define_val:
1691  | expr      { DefineExpr $1 }
1692  | statement { DefineStmt $1 }
1693  | decl      { DefineStmt (mk_st (Decl ($1 Ast_c.NotLocalDecl)) Ast_c.noii) }
1694
1695 /*(*old:
1696   * | TypedefIdent { DefineType (nQ,(TypeName(fst $1,noTypedefDef()),[snd $1]))}
1697   * get conflicts:
1698   * | spec_qualif_list TMul
1699   *   { let (returnType, _) = fixDeclSpecForDecl $1 in  DefineType returnType }
1700   *)
1701 */
1702  | decl_spec
1703      { let returnType = fixDeclSpecForMacro $1 in
1704        DefineType returnType
1705      }
1706  | decl_spec abstract_declarator
1707      { let returnType = fixDeclSpecForMacro $1 in
1708        let typ = $2 returnType in
1709        DefineType typ
1710      }
1711
1712 /* can be in conflict with decl_spec, maybe change fixDeclSpecForMacro
1713  * to also allow storage ?
1714  | storage_class_spec { DefineTodo }
1715  | Tinline { DefineTodo }
1716 */
1717
1718  /*(* a few special cases *)*/
1719  | stat_or_decl stat_or_decl_list { DefineTodo }
1720 /*
1721  | statement statement { DefineTodo }
1722  | decl function_definition { DefineTodo }
1723 */
1724
1725
1726
1727
1728  | function_definition { DefineFunction $1 }
1729
1730  | TOBraceDefineInit initialize_list gcc_comma_opt_struct TCBrace comma_opt
1731     { DefineInit (InitList (List.rev $2), [$1;$4]++$3++$5)  }
1732
1733  /*(* note: had a conflict before when were putting TInt instead of expr *)*/
1734  | Tdo statement Twhile TOPar expr TCPar
1735      {
1736        (* TOREPUT
1737        if fst $5 <> "0"
1738        then pr2 "WEIRD: in macro and have not a while(0)";
1739        *)
1740        DefineDoWhileZero (($2,$5),   [$1;$3;$4;$6])
1741      }
1742
1743  | Tasm TOPar asmbody TCPar              { DefineTodo }
1744  | Tasm Tvolatile TOPar asmbody TCPar    { DefineTodo }
1745
1746  /*(* aliases macro *)*/
1747  | TMacroAttr { DefineTodo }
1748
1749  | /*(* empty *)*/ { DefineEmpty }
1750
1751
1752
1753
1754 param_define:
1755  | TIdent               { mk_string_wrap $1 }
1756  | TypedefIdent         { mk_string_wrap $1 }
1757  | TDefParamVariadic    { mk_string_wrap $1 }
1758  | TEllipsis            { "...", [$1] }
1759  /*(* they reuse keywords :(  *)*/
1760  | Tregister            { "register", [$1] }
1761
1762
1763
1764
1765 cpp_ifdef_directive:
1766  | TIfdef
1767      { let (tag,ii) = $1 in
1768        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)),  [ii]) }
1769  | TIfdefelse
1770      { let (tag,ii) = $1 in
1771        IfdefDirective ((IfdefElse, IfdefTag (Common.some !tag)), [ii]) }
1772  | TIfdefelif
1773      { let (tag,ii) = $1 in
1774        IfdefDirective ((IfdefElseif, IfdefTag (Common.some !tag)), [ii]) }
1775  | TEndif
1776      { let (tag,ii) = $1 in
1777        IfdefDirective ((IfdefEndif, IfdefTag (Common.some !tag)), [ii]) }
1778
1779  | TIfdefBool
1780      { let (_b, tag,ii) = $1 in
1781        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)), [ii]) }
1782  | TIfdefMisc
1783      { let (_b, tag,ii) = $1 in
1784        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)), [ii]) }
1785  | TIfdefVersion
1786      { let (_b, tag,ii) = $1 in
1787        IfdefDirective ((Ifdef, IfdefTag (Common.some !tag)), [ii]) }
1788
1789
1790 /*(* cppext: *)*/
1791 cpp_other:
1792  /*(* no conflict ? no need for a TMacroTop ? apparently not as at toplevel
1793     * the rule are slightly different, they cant be statement and so expr
1794     * at the top, only decl or function definition.
1795     *)*/
1796  | identifier TOPar argument_list TCPar TPtVirg
1797      {
1798        Declaration (MacroDecl ((fst $1, $3), [snd $1;$2;$4;$5;fakeInfo()]))
1799        (* old: MacroTop (fst $1, $3,    [snd $1;$2;$4;$5])  *)
1800      }
1801
1802  /*(* TCParEOL to fix the end-of-stream bug of ocamlyacc *)*/
1803  | identifier TOPar argument_list TCParEOL
1804      { MacroTop (fst $1, $3,    [snd $1;$2;$4;fakeInfo()]) }
1805
1806   /*(* ex: EXPORT_NO_SYMBOLS; *)*/
1807  | identifier TPtVirg { EmptyDef [snd $1;$2] }
1808
1809
1810
1811 /*(*************************************************************************)*/
1812 /*(* celem *)*/
1813 /*(*************************************************************************)*/
1814
1815 external_declaration:
1816  | function_definition               { Definition $1 }
1817  | decl                              { Declaration ($1 Ast_c.NotLocalDecl) }
1818
1819
1820 celem:
1821  | external_declaration                         { $1 }
1822
1823  /*(* cppext: *)*/
1824  | cpp_directive
1825      { CppTop $1 }
1826  | cpp_other
1827      { $1 }
1828  | cpp_ifdef_directive /* (*external_declaration_list ...*)*/
1829      { IfdefTop $1 }
1830
1831  /*(* can have asm declaration at toplevel *)*/
1832  | Tasm TOPar asmbody TCPar TPtVirg             { EmptyDef [$1;$2;$4;$5] }
1833
1834  /*
1835  (* in ~/kernels/src/linux-2.5.2/drivers/isdn/hisax/isdnl3.c sometimes
1836   * the function ends with }; instead of just }
1837   * can also remove this rule and report "parse error" pb to morton
1838   *)*/
1839  | TPtVirg    { EmptyDef [$1] }
1840
1841
1842  | EOF        { FinalDef $1 }
1843
1844
1845
1846
1847 /*(*************************************************************************)*/
1848 /*(* some generic workarounds *)*/
1849 /*(*************************************************************************)*/
1850
1851 tobrace: TOBrace  { LP.push_context LP.InFunction; LP.new_scope (); $1 }
1852 tcbrace: TCBrace  { LP.pop_context();              LP.del_scope (); $1 }
1853
1854 tobrace_enum: TOBrace { LP.push_context LP.InEnum; $1 }
1855 tcbrace_enum: TCBrace { LP.pop_context (); $1 }
1856
1857 tobrace_ini: TOBrace { LP.push_context LP.InInitializer; $1 }
1858 tcbrace_ini: TCBrace { LP.pop_context (); $1 }
1859
1860 tobrace_struct: TOBrace { LP.push_context LP.InStruct; $1}
1861 tcbrace_struct: TCBrace { LP.pop_context (); $1 }
1862
1863
1864
1865
1866 topar: TOPar
1867      { LP.new_scope ();et "topar" ();
1868        LP.push_context LP.InParameter;
1869        $1
1870      }
1871 tcpar: TCPar
1872      { LP.del_scope ();dt "tcpar" ();
1873        LP.pop_context ();
1874        $1
1875      }
1876
1877
1878
1879
1880 /*(*************************************************************************)*/
1881 /*(* xxx_list, xxx_opt *)*/
1882 /*(*************************************************************************)*/
1883
1884
1885 /*(* old:
1886 compound2:
1887  |                            { ([],[]) }
1888  |  statement_list            { ([], $1) }
1889  |  decl_list                 { ($1, []) }
1890  |  decl_list statement_list  { ($1,$2) }
1891
1892 statement_list: stat_or_decl_list { $1 }
1893 *)*/
1894
1895
1896 /*(*
1897 decl_list:
1898  | decl           { [$1]   }
1899  | decl_list decl { $1 ++ [$2] }
1900
1901 statement_list:
1902  | statement { [$1] }
1903  | statement_list statement { $1 ++ [$2] }
1904 *)*/
1905
1906
1907
1908
1909
1910 string_list:
1911  | string_elem { $1 }
1912  | string_list string_elem { $1 ++ $2 }
1913
1914 colon_asm_list:
1915  | colon_asm { [$1] }
1916  | colon_asm_list colon_asm  { $1 ++ [$2] }
1917
1918 colon_option_list:
1919  | colon_option { [$1, []] }
1920  | colon_option_list TComma colon_option { $1 ++ [$3, [$2]] }
1921
1922
1923 argument_list_ne:
1924  | argument_ne                           { [$1, []] }
1925  | argument_list_ne TComma argument { $1 ++ [$3,    [$2]] }
1926
1927 argument_list:
1928  | argument                           { [$1, []] }
1929  | argument_list TComma argument { $1 ++ [$3,    [$2]] }
1930
1931 /*(*
1932 expression_list:
1933  | assign_expr { [$1, []] }
1934  | expression_list TComma assign_expr { $1 ++ [$3,   [$2]] }
1935 *)*/
1936
1937
1938 struct_decl_list:
1939  | struct_decl                   { [$1] }
1940  | struct_decl_list struct_decl  { $1 ++ [$2] }
1941
1942
1943 struct_declarator_list:
1944  | struct_declarator                               { [$1,           []] }
1945  | struct_declarator_list TComma struct_declarator { $1 ++ [$3,     [$2]] }
1946
1947
1948 enumerator_list:
1949  | enumerator                        { [$1,          []]   }
1950  | enumerator_list TComma enumerator { $1 ++ [$3,    [$2]] }
1951
1952
1953 init_declarator_list:
1954  | init_declarator                             { [$1,   []] }
1955  | init_declarator_list TComma init_declarator { $1 ++ [$3,     [$2]] }
1956
1957
1958 parameter_list:
1959  | parameter_decl                       { [$1, []] }
1960  | parameter_list TComma parameter_decl { $1 ++ [$3,  [$2]] }
1961
1962 taction_list_ne:
1963  | TAction                 { [$1] }
1964  | TAction taction_list_ne { $1 :: $2 }
1965
1966 taction_list:
1967 /*old: was generating conflict, hence now taction_list_ne
1968     | (* empty *) { [] }
1969  | TAction { [$1] }
1970  | taction_list TAction { $1 ++ [$2] }
1971 */
1972  |                      { [] }
1973  | TAction taction_list { $1 :: $2 }
1974
1975 param_define_list:
1976  | /*(* empty *)*/ { [] }
1977  | param_define                           { [$1, []] }
1978  | param_define_list TComma param_define  { $1 ++ [$3, [$2]] }
1979
1980 designator_list:
1981  | designator { [$1] }
1982  | designator_list designator { $1 ++ [$2] }
1983
1984 attribute_list:
1985  | attribute { [$1] }
1986  | attribute_list attribute { $1 ++ [$2] }
1987
1988 attribute_storage_list:
1989  | attribute_storage { [$1] }
1990  | attribute_storage_list attribute_storage { $1 ++ [$2] }
1991
1992
1993 attributes: attribute_list { $1 }
1994
1995
1996
1997 /*(* gccext:  which allow a trailing ',' in enum, as in perl *)*/
1998 gcc_comma_opt:
1999  | TComma {  [$1] }
2000  | /*(* empty *)*/  {  []  }
2001
2002 comma_opt:
2003  | TComma {  [$1] }
2004  | /*(* empty *)*/  {  []  }
2005
2006 /*(*
2007 gcc_opt_virg:
2008  | TPtVirg { }
2009  |  { }
2010 *)*/
2011
2012 gcc_opt_expr:
2013  | expr        { Some $1 }
2014  | /*(* empty *)*/ { None  }
2015
2016 /*(*
2017 opt_ptvirg:
2018  | TPtVirg { [$1] }
2019  | { [] }
2020 *)*/
2021
2022
2023 expr_opt:
2024  | expr            { Some $1 }
2025  | /*(* empty *)*/ { None }
2026