parsing_c/token_helpers.ml

   1 (* Yoann Padioleau
   2  *
   3  * Copyright (C) 2007, 2008 Ecole des Mines de Nantes
   4  *
   5  * This program is free software; you can redistribute it and/or
   6  * modify it under the terms of the GNU General Public License (GPL)
   7  * version 2 as published by the Free Software Foundation.
   8  *
   9  * This program is distributed in the hope that it will be useful,
  10  * but WITHOUT ANY WARRANTY; without even the implied warranty of
  11  * MERCHANTABILITY or FITNESS FOR A PARTICULAR PURPOSE.  See the
  12  * file license.txt for more details.
  13  *)
  14
  15
  16 open Common
  17
  18 open Parser_c
  19
  20 (*****************************************************************************)
  21 (* Is_xxx, categories *)
  22 (*****************************************************************************)
  23
  24 (* could define a type  token_class = Comment | Ident | Operator | ...
  25  * update: now token_c can maybe do that.
  26  * but still, sometimes tokens belon to multiple classes. Could maybe
  27  * return then a set of classes.
  28  *)
  29
  30 let is_space = function
  31   | TCommentSpace _ -> true
  32   | TCommentNewline _ -> true
  33   | _ -> false
  34
  35 let is_whitespace = is_space
  36
  37 let is_just_comment_or_space = function
  38   | TComment _ -> true
  39   | TCommentSpace _ -> true
  40   | TCommentNewline _ -> true
  41   | _ -> false
  42 let is_real_comment = is_just_comment_or_space
  43
  44 let is_just_comment = function
  45   | TComment _ -> true
  46   | _ -> false
  47
  48
  49
  50
  51 let is_comment = function
  52   | TComment _
  53   | TCommentSpace _ | TCommentNewline _
  54   | TCommentCpp _
  55   | TCommentMisc _ -> true
  56   | _ -> false
  57
  58 (* coupling with comment_annotater_c.ml.
  59  * In fact more tokens than comments are not in the ast, but
  60  * they were usually temporally created by ocamllex and removed
  61  * in parsing_hacks.
  62 *)
  63 let is_not_in_ast = is_comment
  64
  65 let is_fake_comment = function
  66   | TCommentCpp _    | TCommentMisc _
  67       -> true
  68   | _ -> false
  69
  70 let is_not_comment x =
  71   not (is_comment x)
  72
  73
  74 (* ---------------------------------------------------------------------- *)
  75
  76 let is_cpp_instruction = function
  77   | TInclude _
  78   | TDefine _
  79   | TIfdef _   | TIfdefelse _ | TIfdefelif _ | TEndif _
  80   | TIfdefBool _ | TIfdefMisc _ | TIfdefVersion _
  81   | TUndef _
  82   | TCppDirectiveOther _
  83       -> true
  84   | _ -> false
  85
  86
  87 let is_gcc_token = function
  88   | Tasm _
  89   | Tinline _
  90   | Tattribute _
  91   | Ttypeof _
  92       -> true
  93   | _ -> false
  94
  95
  96
  97
  98 (* ---------------------------------------------------------------------- *)
  99 let is_opar = function
 100   | TOPar _ | TOParDefine _ -> true
 101   | _ -> false
 102
 103 let is_cpar = function
 104   | TCPar _ | TCParEOL _ -> true
 105   | _ -> false
 106
 107
 108 let is_obrace = function
 109   | TOBrace _ | TOBraceDefineInit _ -> true
 110   | _ -> false
 111
 112 let is_cbrace = function
 113   | TCBrace _ -> true
 114   | _ -> false
 115
 116
 117
 118
 119 (* ---------------------------------------------------------------------- *)
 120
 121 (* end of file *)
 122 let is_eof = function
 123   | EOF x -> true
 124   | _ -> false
 125
 126
 127 (* end of macro *)
 128 let is_eom = function
 129   | TDefEOL _ -> true
 130   | _ -> false
 131
 132 let is_statement = function
 133   | Tfor _ | Tdo _ | Tif _ | Twhile _ | Treturn _
 134   | Tbreak _ | Telse _ | Tswitch _ | Tcase _ | Tcontinue _
 135   | Tgoto _
 136   | TPtVirg _
 137   | TMacroIterator _
 138       -> true
 139   | _ -> false
 140
 141 (* is_start_of_something is used in parse_c for error recovery, to find
 142  * a synchronisation token.
 143  *
 144  * Would like to put TIdent or TDefine, TIfdef but they can be in the
 145  * middle of a function, for instance with label:.
 146  *
 147  * Could put Typedefident but fired ? it would work in error recovery
 148  * on the already_passed tokens, which has been already gone in the
 149  * Parsing_hacks.lookahead machinery, but it will not work on the
 150  * "next" tokens. But because the namespace for labels is different
 151  * from namespace for ident/typedef, we can use the name for a typedef
 152  * for a label and so dangerous to put Typedefident at true here.
 153  *
 154  * Can look in parser_c.output to know what can be at toplevel
 155  * at the very beginning.
 156  *)
 157
 158 let is_start_of_something = function
 159   | Tchar _  | Tshort _ | Tint _ | Tdouble _ |  Tfloat _ | Tlong _
 160   | Tunsigned _ | Tsigned _ | Tvoid _
 161   | Tauto _ | Tregister _ | Textern _ | Tstatic _
 162   | Tconst _ | Tvolatile _
 163   | Ttypedef _
 164   | Tstruct _ | Tunion _ | Tenum _
 165     -> true
 166   | _ -> false
 167
 168
 169
 170 let is_binary_operator = function
 171   | TOrLog _ | TAndLog _ |  TOr _ |  TXor _ |  TAnd _
 172   | TEqEq _ |  TNotEq _  | TInf _ |  TSup _ |  TInfEq _ |  TSupEq _
 173   | TShl _ | TShr _
 174   | TPlus _ |  TMinus _ |  TMul _ |  TDiv _ |  TMod _
 175         -> true
 176   | _ -> false
 177
 178 let is_stuff_taking_parenthized = function
 179   | Tif _
 180   | Twhile _
 181   | Tswitch _
 182   | Ttypeof _
 183   | TMacroIterator _
 184     -> true
 185   | _ -> false
 186
 187
 188 (* used in the algorithms for "10 most problematic errors" *)
 189 let is_ident_like = function
 190   | TIdent _
 191   | TypedefIdent _
 192   | TIdentDefine  _
 193   | TDefParamVariadic _
 194
 195   | TUnknown _
 196
 197   | TMacroAttr _
 198   | TMacroAttrStorage _
 199   | TMacroStmt _
 200   | TMacroString _
 201   | TMacroDecl _
 202   | TMacroDeclConst _
 203   | TMacroIterator _
 204       -> true
 205
 206   | _ -> false
 207
 208
 209 (*****************************************************************************)
 210 (* Visitors *)
 211 (*****************************************************************************)
 212
 213 (* Because ocamlyacc force us to do it that way. The ocamlyacc token
 214  * cant be a pair of a sum type, it must be directly a sum type.
 215  *)
 216 let info_of_tok = function
 217   | TString ((string, isWchar), i) -> i
 218   | TChar  ((string, isWchar), i) -> i
 219   | TFloat ((string, floatType), i) -> i
 220
 221   | TAssign  (assignOp, i) -> i
 222
 223   | TIdent  (s, i) -> i
 224   | TypedefIdent  (s, i) -> i
 225
 226   | TInt  (s, i) -> i
 227
 228   | TDefine (ii) -> ii
 229   | TInclude (includes, filename, inifdef, i1) ->     i1
 230
 231   | TUndef (s, ii) -> ii
 232   | TCppDirectiveOther (ii) -> ii
 233
 234   | TIncludeStart (i1, inifdef) ->     i1
 235   | TIncludeFilename (s, i1) ->     i1
 236
 237   | TDefEOL (i1) ->     i1
 238   | TOParDefine (i1) ->     i1
 239   | TIdentDefine  (s, i) -> i
 240   | TCppEscapedNewline (ii) -> ii
 241   | TDefParamVariadic (s, i1) ->     i1
 242
 243   | TCppConcatOp (ii) -> ii
 244
 245   | TOBraceDefineInit (i1) ->     i1
 246
 247   | TUnknown             (i) -> i
 248
 249   | TMacroIdentBuilder             (s, i) -> i
 250   | TMacroAttr             (s, i) -> i
 251   | TMacroAttrStorage             (s, i) -> i
 252   | TMacroStmt             (s, i) -> i
 253   | TMacroString             (s, i) -> i
 254   | TMacroDecl             (s, i) -> i
 255   | TMacroDeclConst             (i) -> i
 256   | TMacroIterator             (s,i) -> i
 257 (*  | TMacroTop             (s,i) -> i *)
 258   | TCParEOL (i1) ->     i1
 259
 260   | TAction             (i) -> i
 261
 262   | TComment             (i) -> i
 263   | TCommentSpace        (i) -> i
 264   | TCommentNewline      (i) -> i
 265   | TCommentCpp          (cppkind, i) -> i
 266   | TCommentMisc         (i) -> i
 267
 268   | TCommentSkipTagStart (i) -> i
 269   | TCommentSkipTagEnd (i) -> i
 270
 271   | TIfdef               (_, i) -> i
 272   | TIfdefelse           (_, i) -> i
 273   | TIfdefelif           (_, i) -> i
 274   | TEndif               (_, i) -> i
 275   | TIfdefBool           (b, _, i) -> i
 276   | TIfdefMisc           (b, _, i) -> i
 277   | TIfdefVersion           (b, _, i) -> i
 278
 279   | TOPar                (i) -> i
 280   | TCPar                (i) -> i
 281   | TOBrace              (i) -> i
 282   | TCBrace              (i) -> i
 283   | TOCro                (i) -> i
 284   | TCCro                (i) -> i
 285   | TDot                 (i) -> i
 286   | TComma               (i) -> i
 287   | TPtrOp               (i) -> i
 288   | TInc                 (i) -> i
 289   | TDec                 (i) -> i
 290   | TEq                  (i) -> i
 291   | TWhy                 (i) -> i
 292   | TTilde               (i) -> i
 293   | TBang                (i) -> i
 294   | TEllipsis            (i) -> i
 295   | TDotDot              (i) -> i
 296   | TPtVirg              (i) -> i
 297   | TOrLog               (i) -> i
 298   | TAndLog              (i) -> i
 299   | TOr                  (i) -> i
 300   | TXor                 (i) -> i
 301   | TAnd                 (i) -> i
 302   | TEqEq                (i) -> i
 303   | TNotEq               (i) -> i
 304   | TInf                 (i) -> i
 305   | TSup                 (i) -> i
 306   | TInfEq               (i) -> i
 307   | TSupEq               (i) -> i
 308   | TShl                 (i) -> i
 309   | TShr                 (i) -> i
 310   | TPlus                (i) -> i
 311   | TMinus               (i) -> i
 312   | TMul                 (i) -> i
 313   | TDiv                 (i) -> i
 314   | TMod                 (i) -> i
 315
 316   | Tchar                (i) -> i
 317   | Tshort               (i) -> i
 318   | Tint                 (i) -> i
 319   | Tdouble              (i) -> i
 320   | Tfloat               (i) -> i
 321   | Tlong                (i) -> i
 322   | Tunsigned            (i) -> i
 323   | Tsigned              (i) -> i
 324   | Tvoid                (i) -> i
 325   | Tauto                (i) -> i
 326   | Tregister            (i) -> i
 327   | Textern              (i) -> i
 328   | Tstatic              (i) -> i
 329   | Tconst               (i) -> i
 330   | Tvolatile            (i) -> i
 331
 332   | Trestrict            (i) -> i
 333
 334   | Tstruct              (i) -> i
 335   | Tenum                (i) -> i
 336   | Ttypedef             (i) -> i
 337   | Tunion               (i) -> i
 338   | Tbreak               (i) -> i
 339   | Telse                (i) -> i
 340   | Tswitch              (i) -> i
 341   | Tcase                (i) -> i
 342   | Tcontinue            (i) -> i
 343   | Tfor                 (i) -> i
 344   | Tdo                  (i) -> i
 345   | Tif                  (i) -> i
 346   | Twhile               (i) -> i
 347   | Treturn              (i) -> i
 348   | Tgoto                (i) -> i
 349   | Tdefault             (i) -> i
 350   | Tsizeof              (i) -> i
 351   | Tasm                 (i) -> i
 352   | Tattribute           (i) -> i
 353   | Tinline              (i) -> i
 354   | Ttypeof              (i) -> i
 355
 356   | EOF                  (i) -> i
 357
 358
 359
 360
 361 (* used by tokens to complete the parse_info with filename, line, col infos *)
 362 let visitor_info_of_tok f = function
 363   | TString ((s, isWchar), i)  -> TString ((s, isWchar), f i)
 364   | TChar  ((s, isWchar), i)   -> TChar  ((s, isWchar), f i)
 365   | TFloat ((s, floatType), i) -> TFloat ((s, floatType), f i)
 366   | TAssign  (assignOp, i)     -> TAssign  (assignOp, f i)
 367
 368   | TIdent  (s, i)       -> TIdent  (s, f i)
 369   | TypedefIdent  (s, i) -> TypedefIdent  (s, f i)
 370   | TInt  (s, i)         -> TInt  (s, f i)
 371
 372   | TDefine (i1) -> TDefine(f i1)
 373
 374   | TUndef (s,i1) -> TUndef(s, f i1)
 375   | TCppDirectiveOther (i1) -> TCppDirectiveOther(f i1)
 376
 377   | TInclude (includes, filename, inifdef, i1) ->
 378       TInclude (includes, filename, inifdef, f i1)
 379
 380   | TIncludeStart (i1, inifdef) -> TIncludeStart (f i1, inifdef)
 381   | TIncludeFilename (s, i1) -> TIncludeFilename (s, f i1)
 382
 383   | TCppEscapedNewline (i1) -> TCppEscapedNewline (f i1)
 384   | TDefEOL (i1) -> TDefEOL (f i1)
 385
 386   | TCppConcatOp (ii) -> TCppConcatOp (f ii)
 387
 388   | TOParDefine (i1) -> TOParDefine (f i1)
 389   | TIdentDefine  (s, i) -> TIdentDefine (s, f i)
 390
 391   | TDefParamVariadic (s, i1) -> TDefParamVariadic (s, f i1)
 392
 393   | TOBraceDefineInit (i1) -> TOBraceDefineInit (f i1)
 394
 395
 396   | TUnknown             (i) -> TUnknown                (f i)
 397
 398   | TMacroIdentBuilder             (s, i) -> TMacroIdentBuilder (s, f i)
 399   | TMacroAttr           (s, i)   -> TMacroAttr            (s, f i)
 400   | TMacroAttrStorage           (s, i)   -> TMacroAttrStorage         (s, f i)
 401   | TMacroStmt           (s, i)   -> TMacroStmt            (s, f i)
 402   | TMacroString         (s, i)   -> TMacroString          (s, f i)
 403   | TMacroDecl           (s, i) -> TMacroDecl            (s, f i)
 404   | TMacroDeclConst      (i)   -> TMacroDeclConst       (f i)
 405   | TMacroIterator       (s, i) -> TMacroIterator        (s, f i)
 406 (*  | TMacroTop          (s,i) -> TMacroTop             (s,f i) *)
 407   | TCParEOL (i) ->     TCParEOL (f i)
 408
 409
 410   | TAction               (i) -> TAction             (f i)
 411
 412   | TComment             (i) -> TComment             (f i)
 413   | TCommentSpace        (i) -> TCommentSpace        (f i)
 414   | TCommentNewline      (i) -> TCommentNewline      (f i)
 415   | TCommentCpp          (cppkind, i) -> TCommentCpp (cppkind, f i)
 416   | TCommentMisc         (i) -> TCommentMisc         (f i)
 417
 418   | TCommentSkipTagStart         (i) -> TCommentSkipTagStart         (f i)
 419   | TCommentSkipTagEnd         (i) -> TCommentSkipTagEnd         (f i)
 420
 421   | TIfdef               (t, i) -> TIfdef               (t, f i)
 422   | TIfdefelse           (t, i) -> TIfdefelse           (t, f i)
 423   | TIfdefelif           (t, i) -> TIfdefelif           (t, f i)
 424   | TEndif               (t, i) -> TEndif               (t, f i)
 425   | TIfdefBool           (b, t, i) -> TIfdefBool        (b, t, f i)
 426   | TIfdefMisc           (b, t, i) -> TIfdefMisc        (b, t, f i)
 427   | TIfdefVersion        (b, t, i) -> TIfdefVersion     (b, t, f i)
 428
 429   | TOPar                (i) -> TOPar                (f i)
 430   | TCPar                (i) -> TCPar                (f i)
 431   | TOBrace              (i) -> TOBrace              (f i)
 432   | TCBrace              (i) -> TCBrace              (f i)
 433   | TOCro                (i) -> TOCro                (f i)
 434   | TCCro                (i) -> TCCro                (f i)
 435   | TDot                 (i) -> TDot                 (f i)
 436   | TComma               (i) -> TComma               (f i)
 437   | TPtrOp               (i) -> TPtrOp               (f i)
 438   | TInc                 (i) -> TInc                 (f i)
 439   | TDec                 (i) -> TDec                 (f i)
 440   | TEq                  (i) -> TEq                  (f i)
 441   | TWhy                 (i) -> TWhy                 (f i)
 442   | TTilde               (i) -> TTilde               (f i)
 443   | TBang                (i) -> TBang                (f i)
 444   | TEllipsis            (i) -> TEllipsis            (f i)
 445   | TDotDot              (i) -> TDotDot              (f i)
 446   | TPtVirg              (i) -> TPtVirg              (f i)
 447   | TOrLog               (i) -> TOrLog               (f i)
 448   | TAndLog              (i) -> TAndLog              (f i)
 449   | TOr                  (i) -> TOr                  (f i)
 450   | TXor                 (i) -> TXor                 (f i)
 451   | TAnd                 (i) -> TAnd                 (f i)
 452   | TEqEq                (i) -> TEqEq                (f i)
 453   | TNotEq               (i) -> TNotEq               (f i)
 454   | TInf                 (i) -> TInf                 (f i)
 455   | TSup                 (i) -> TSup                 (f i)
 456   | TInfEq               (i) -> TInfEq               (f i)
 457   | TSupEq               (i) -> TSupEq               (f i)
 458   | TShl                 (i) -> TShl                 (f i)
 459   | TShr                 (i) -> TShr                 (f i)
 460   | TPlus                (i) -> TPlus                (f i)
 461   | TMinus               (i) -> TMinus               (f i)
 462   | TMul                 (i) -> TMul                 (f i)
 463   | TDiv                 (i) -> TDiv                 (f i)
 464   | TMod                 (i) -> TMod                 (f i)
 465   | Tchar                (i) -> Tchar                (f i)
 466   | Tshort               (i) -> Tshort               (f i)
 467   | Tint                 (i) -> Tint                 (f i)
 468   | Tdouble              (i) -> Tdouble              (f i)
 469   | Tfloat               (i) -> Tfloat               (f i)
 470   | Tlong                (i) -> Tlong                (f i)
 471   | Tunsigned            (i) -> Tunsigned            (f i)
 472   | Tsigned              (i) -> Tsigned              (f i)
 473   | Tvoid                (i) -> Tvoid                (f i)
 474   | Tauto                (i) -> Tauto                (f i)
 475   | Tregister            (i) -> Tregister            (f i)
 476   | Textern              (i) -> Textern              (f i)
 477   | Tstatic              (i) -> Tstatic              (f i)
 478   | Tconst               (i) -> Tconst               (f i)
 479   | Tvolatile            (i) -> Tvolatile            (f i)
 480
 481   | Trestrict            (i) -> Trestrict            (f i)
 482
 483   | Tstruct              (i) -> Tstruct              (f i)
 484   | Tenum                (i) -> Tenum                (f i)
 485   | Ttypedef             (i) -> Ttypedef             (f i)
 486   | Tunion               (i) -> Tunion               (f i)
 487   | Tbreak               (i) -> Tbreak               (f i)
 488   | Telse                (i) -> Telse                (f i)
 489   | Tswitch              (i) -> Tswitch              (f i)
 490   | Tcase                (i) -> Tcase                (f i)
 491   | Tcontinue            (i) -> Tcontinue            (f i)
 492   | Tfor                 (i) -> Tfor                 (f i)
 493   | Tdo                  (i) -> Tdo                  (f i)
 494   | Tif                  (i) -> Tif                  (f i)
 495   | Twhile               (i) -> Twhile               (f i)
 496   | Treturn              (i) -> Treturn              (f i)
 497   | Tgoto                (i) -> Tgoto                (f i)
 498   | Tdefault             (i) -> Tdefault             (f i)
 499   | Tsizeof              (i) -> Tsizeof              (f i)
 500   | Tasm                 (i) -> Tasm                 (f i)
 501   | Tattribute           (i) -> Tattribute           (f i)
 502   | Tinline              (i) -> Tinline              (f i)
 503   | Ttypeof              (i) -> Ttypeof              (f i)
 504   | EOF                  (i) -> EOF                  (f i)
 505
 506
 507 (*****************************************************************************)
 508 (* Accessors *)
 509 (*****************************************************************************)
 510
 511 let linecol_of_tok tok =
 512   let info = info_of_tok tok in
 513   Ast_c.line_of_info info, Ast_c.col_of_info info
 514
 515 let col_of_tok x = snd (linecol_of_tok x)
 516 let line_of_tok x = fst (linecol_of_tok x)
 517 let pos_of_tok x =  Ast_c.opos_of_info (info_of_tok x)
 518 let str_of_tok x =  Ast_c.str_of_info (info_of_tok x)
 519 let file_of_tok x = Ast_c.file_of_info (info_of_tok x)
 520 let pinfo_of_tok x = Ast_c.pinfo_of_info (info_of_tok x)
 521
 522 let is_origin x =
 523   match pinfo_of_tok x with Ast_c.OriginTok _ -> true | _ -> false
 524 let is_expanded x =
 525   match pinfo_of_tok x with Ast_c.ExpandedTok _ -> true | _ -> false
 526 let is_fake x =
 527   match pinfo_of_tok x with Ast_c.FakeTok _ -> true | _ -> false
 528 let is_abstract x =
 529   match pinfo_of_tok x with Ast_c.AbstractLineTok _ -> true | _ -> false
 530
 531 (*****************************************************************************)
 532 (* Helpers *)
 533 (*****************************************************************************)
 534 let is_same_line_or_close line tok =
 535   line_of_tok tok =|= line ||
 536   line_of_tok tok =|= line - 1 ||
 537   line_of_tok tok =|= line - 2
 538