gcc/ada/par-util.adb

   1 ------------------------------------------------------------------------------
   2 --                                                                          --
   3 --                         GNAT COMPILER COMPONENTS                         --
   4 --                                                                          --
   5 --                             P A R . U T I L                              --
   6 --                                                                          --
   7 --                                 B o d y                                  --
   8 --                                                                          --
   9 --          Copyright (C) 1992-2008, Free Software Foundation, Inc.         --
  10 --                                                                          --
  11 -- GNAT is free software;  you can  redistribute it  and/or modify it under --
  12 -- terms of the  GNU General Public License as published  by the Free Soft- --
  13 -- ware  Foundation;  either version 3,  or (at your option) any later ver- --
  14 -- sion.  GNAT is distributed in the hope that it will be useful, but WITH- --
  15 -- OUT ANY WARRANTY;  without even the  implied warranty of MERCHANTABILITY --
  16 -- or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License --
  17 -- for  more details.  You should have  received  a copy of the GNU General --
  18 -- Public License  distributed with GNAT; see file COPYING3.  If not, go to --
  19 -- http://www.gnu.org/licenses for a complete copy of the license.          --
  20 --                                                                          --
  21 -- GNAT was originally developed  by the GNAT team at  New York University. --
  22 -- Extensive contributions were provided by Ada Core Technologies Inc.      --
  23 --                                                                          --
  24 ------------------------------------------------------------------------------
  25
  26 with Csets;    use Csets;
  27 with Namet.Sp; use Namet.Sp;
  28 with Stylesw;  use Stylesw;
  29 with Uintp;    use Uintp;
  30
  31 with GNAT.Spelling_Checker; use GNAT.Spelling_Checker;
  32
  33 separate (Par)
  34 package body Util is
  35
  36    ---------------------
  37    -- Bad_Spelling_Of --
  38    ---------------------
  39
  40    function Bad_Spelling_Of (T : Token_Type) return Boolean is
  41       Tname : constant String := Token_Type'Image (T);
  42       --  Characters of token name
  43
  44       S : String (1 .. Tname'Last - 4);
  45       --  Characters of token name folded to lower case, omitting TOK_ at start
  46
  47       M1 : String (1 .. 42) := "incorrect spelling of keyword ************";
  48       M2 : String (1 .. 44) := "illegal abbreviation of keyword ************";
  49       --  Buffers used to construct error message
  50
  51       P1 : constant := 30;
  52       P2 : constant := 32;
  53       --  Starting subscripts in M1, M2 for keyword name
  54
  55       SL : constant Natural := S'Length;
  56       --  Length of expected token name excluding TOK_ at start
  57
  58    begin
  59       if Token /= Tok_Identifier then
  60          return False;
  61       end if;
  62
  63       for J in S'Range loop
  64          S (J) := Fold_Lower (Tname (J + 4));
  65       end loop;
  66
  67       Get_Name_String (Token_Name);
  68
  69       --  A special check for case of PROGRAM used for PROCEDURE
  70
  71       if T = Tok_Procedure
  72         and then Name_Len = 7
  73         and then Name_Buffer (1 .. 7) = "program"
  74       then
  75          Error_Msg_SC ("PROCEDURE expected");
  76          Token := T;
  77          return True;
  78
  79       --  A special check for an illegal abbreviation
  80
  81       elsif Name_Len < S'Length
  82         and then Name_Len >= 4
  83         and then Name_Buffer (1 .. Name_Len) = S (1 .. Name_Len)
  84       then
  85          for J in 1 .. S'Last loop
  86             M2 (P2 + J - 1) := Fold_Upper (S (J));
  87          end loop;
  88
  89          Error_Msg_SC (M2 (1 .. P2 - 1 + S'Last));
  90          Token := T;
  91          return True;
  92       end if;
  93
  94       --  Now we go into the full circuit to check for a misspelling
  95
  96       --  Never consider something a misspelling if either the actual or
  97       --  expected string is less than 3 characters (before this check we
  98       --  used to consider i to be a misspelled if in some cases!)
  99
 100       if SL < 3 or else Name_Len < 3 then
 101          return False;
 102
 103       --  Special case: prefix matches, i.e. the leading characters of the
 104       --  token that we have exactly match the required keyword. If there
 105       --  are at least two characters left over, assume that we have a case
 106       --  of two keywords joined together which should not be joined.
 107
 108       elsif Name_Len > SL + 1
 109         and then S = Name_Buffer (1 .. SL)
 110       then
 111          Scan_Ptr := Token_Ptr + S'Length;
 112          Error_Msg_S ("|missing space");
 113          Token := T;
 114          return True;
 115       end if;
 116
 117       if Is_Bad_Spelling_Of (Name_Buffer (1 .. Name_Len), S) then
 118          for J in 1 .. S'Last loop
 119             M1 (P1 + J - 1) := Fold_Upper (S (J));
 120          end loop;
 121
 122          Error_Msg_SC (M1 (1 .. P1 - 1 + S'Last));
 123          Token := T;
 124          return True;
 125
 126       else
 127          return False;
 128       end if;
 129    end Bad_Spelling_Of;
 130
 131    ----------------------
 132    -- Check_95_Keyword --
 133    ----------------------
 134
 135    --  On entry, the caller has checked that current token is an identifier
 136    --  whose name matches the name of the 95 keyword New_Tok.
 137
 138    procedure Check_95_Keyword (Token_95, Next : Token_Type) is
 139       Scan_State : Saved_Scan_State;
 140
 141    begin
 142       Save_Scan_State (Scan_State); -- at identifier/keyword
 143       Scan; -- past identifier/keyword
 144
 145       if Token = Next then
 146          Restore_Scan_State (Scan_State); -- to identifier
 147          Error_Msg_Name_1 := Token_Name;
 148          Error_Msg_SC ("(Ada 83) keyword* cannot be used!");
 149          Token := Token_95;
 150       else
 151          Restore_Scan_State (Scan_State); -- to identifier
 152       end if;
 153    end Check_95_Keyword;
 154
 155    ----------------------
 156    -- Check_Bad_Layout --
 157    ----------------------
 158
 159    procedure Check_Bad_Layout is
 160    begin
 161       if Style.RM_Column_Check and then Token_Is_At_Start_Of_Line
 162         and then Start_Column <= Scope.Table (Scope.Last).Ecol
 163       then
 164          Error_Msg_BC ("(style) incorrect layout");
 165       end if;
 166    end Check_Bad_Layout;
 167
 168    --------------------------
 169    -- Check_Misspelling_Of --
 170    --------------------------
 171
 172    procedure Check_Misspelling_Of (T : Token_Type) is
 173    begin
 174       if Bad_Spelling_Of (T) then
 175          null;
 176       end if;
 177    end Check_Misspelling_Of;
 178
 179    -----------------------------
 180    -- Check_Simple_Expression --
 181    -----------------------------
 182
 183    procedure Check_Simple_Expression (E : Node_Id) is
 184    begin
 185       if Expr_Form = EF_Non_Simple then
 186          Error_Msg_N ("this expression must be parenthesized", E);
 187       end if;
 188    end Check_Simple_Expression;
 189
 190    ---------------------------------------
 191    -- Check_Simple_Expression_In_Ada_83 --
 192    ---------------------------------------
 193
 194    procedure Check_Simple_Expression_In_Ada_83 (E : Node_Id) is
 195    begin
 196       if Expr_Form = EF_Non_Simple then
 197          if Ada_Version = Ada_83 then
 198             Error_Msg_N ("(Ada 83) this expression must be parenthesized!", E);
 199          end if;
 200       end if;
 201    end Check_Simple_Expression_In_Ada_83;
 202
 203    ------------------------
 204    -- Check_Subtype_Mark --
 205    ------------------------
 206
 207    function Check_Subtype_Mark (Mark : Node_Id) return Node_Id is
 208    begin
 209       if Nkind (Mark) = N_Identifier
 210         or else Nkind (Mark) = N_Selected_Component
 211         or else (Nkind (Mark) = N_Attribute_Reference
 212                   and then Is_Type_Attribute_Name (Attribute_Name (Mark)))
 213         or else Mark = Error
 214       then
 215          return Mark;
 216       else
 217          Error_Msg ("subtype mark expected", Sloc (Mark));
 218          return Error;
 219       end if;
 220    end Check_Subtype_Mark;
 221
 222    -------------------
 223    -- Comma_Present --
 224    -------------------
 225
 226    function Comma_Present return Boolean is
 227       Scan_State  : Saved_Scan_State;
 228       Paren_Count : Nat;
 229
 230    begin
 231       --  First check, if a comma is present, then a comma is present!
 232
 233       if Token = Tok_Comma then
 234          T_Comma;
 235          return True;
 236
 237       --  If we have a right paren, then that is taken as ending the list
 238       --  i.e. no comma is present.
 239
 240       elsif Token = Tok_Right_Paren then
 241          return False;
 242
 243       --  If pragmas, then get rid of them and make a recursive call
 244       --  to process what follows these pragmas.
 245
 246       elsif Token = Tok_Pragma then
 247          P_Pragmas_Misplaced;
 248          return Comma_Present;
 249
 250       --  At this stage we have an error, and the goal is to decide on whether
 251       --  or not we should diagnose an error and report a (non-existent)
 252       --  comma as being present, or simply to report no comma is present
 253
 254       --  If we are a semicolon, then the question is whether we have a missing
 255       --  right paren, or whether the semicolon should have been a comma. To
 256       --  guess the right answer, we scan ahead keeping track of the paren
 257       --  level, looking for a clue that helps us make the right decision.
 258
 259       --  This approach is highly accurate in the single error case, and does
 260       --  not make bad mistakes in the multiple error case (indeed we can't
 261       --  really make a very bad decision at this point in any case).
 262
 263       elsif Token = Tok_Semicolon then
 264          Save_Scan_State (Scan_State);
 265          Scan; -- past semicolon
 266
 267          --  Check for being followed by identifier => which almost certainly
 268          --  means we are still in a parameter list and the comma should have
 269          --  been a semicolon (such a sequence could not follow a semicolon)
 270
 271          if Token = Tok_Identifier then
 272             Scan;
 273
 274             if Token = Tok_Arrow then
 275                goto Assume_Comma;
 276             end if;
 277          end if;
 278
 279          --  If that test didn't work, loop ahead looking for a comma or
 280          --  semicolon at the same parenthesis level. Always remember that
 281          --  we can't go badly wrong in an error situation like this!
 282
 283          Paren_Count := 0;
 284
 285          --  Here is the look ahead loop, Paren_Count tells us whether the
 286          --  token we are looking at is at the same paren level as the
 287          --  suspicious semicolon that we are trying to figure out.
 288
 289          loop
 290
 291             --  If we hit another semicolon or an end of file, and we have
 292             --  not seen a right paren or another comma on the way, then
 293             --  probably the semicolon did end the list. Indeed that is
 294             --  certainly the only single error correction possible here.
 295
 296             if Token = Tok_Semicolon or else Token = Tok_EOF then
 297                Restore_Scan_State (Scan_State);
 298                return False;
 299
 300             --  A comma at the same paren level as the semicolon is a strong
 301             --  indicator that the semicolon should have been a comma, indeed
 302             --  again this is the only possible single error correction.
 303
 304             elsif Token = Tok_Comma then
 305                exit when Paren_Count = 0;
 306
 307             --  A left paren just bumps the paren count
 308
 309             elsif Token = Tok_Left_Paren then
 310                Paren_Count := Paren_Count + 1;
 311
 312             --  A right paren that is at the same paren level as the semicolon
 313             --  also means that the only possible single error correction is
 314             --  to assume that the semicolon should have been a comma. If we
 315             --  are not at the same paren level, then adjust the paren level.
 316
 317             elsif Token = Tok_Right_Paren then
 318                exit when Paren_Count = 0;
 319                Paren_Count := Paren_Count - 1;
 320             end if;
 321
 322             --  Keep going, we haven't made a decision yet
 323
 324             Scan;
 325          end loop;
 326
 327          --  If we fall through the loop, it means that we found a terminating
 328          --  right paren or another comma. In either case it is reasonable to
 329          --  assume that the semicolon was really intended to be a comma. Also
 330          --  come here for the identifier arrow case.
 331
 332          <<Assume_Comma>>
 333             Restore_Scan_State (Scan_State);
 334             Error_Msg_SC ("|"";"" should be "",""");
 335             Scan; -- past the semicolon
 336             return True;
 337
 338       --  If we are not at semicolon or a right paren, then we base the
 339       --  decision on whether or not the next token can be part of an
 340       --  expression. If not, then decide that no comma is present (the
 341       --  caller will eventually generate a missing right parent message)
 342
 343       elsif Token in Token_Class_Eterm then
 344          return False;
 345
 346       --  Otherwise we assume a comma is present, even if none is present,
 347       --  since the next token must be part of an expression, so if we were
 348       --  at the end of the list, then there is more than one error present.
 349
 350       else
 351          T_Comma; -- to give error
 352          return True;
 353       end if;
 354    end Comma_Present;
 355
 356    -----------------------
 357    -- Discard_Junk_List --
 358    -----------------------
 359
 360    procedure Discard_Junk_List (L : List_Id) is
 361       pragma Warnings (Off, L);
 362    begin
 363       null;
 364    end Discard_Junk_List;
 365
 366    -----------------------
 367    -- Discard_Junk_Node --
 368    -----------------------
 369
 370    procedure Discard_Junk_Node (N : Node_Id) is
 371       pragma Warnings (Off, N);
 372    begin
 373       null;
 374    end Discard_Junk_Node;
 375
 376    ------------
 377    -- Ignore --
 378    ------------
 379
 380    procedure Ignore (T : Token_Type) is
 381    begin
 382       while Token = T loop
 383          if T = Tok_Comma then
 384             Error_Msg_SC ("|extra "","" ignored");
 385
 386          elsif T = Tok_Left_Paren then
 387             Error_Msg_SC ("|extra ""("" ignored");
 388
 389          elsif T = Tok_Right_Paren then
 390             Error_Msg_SC ("|extra "")"" ignored");
 391
 392          elsif T = Tok_Semicolon then
 393             Error_Msg_SC ("|extra "";"" ignored");
 394
 395          elsif T = Tok_Colon then
 396             Error_Msg_SC ("|extra "":"" ignored");
 397
 398          else
 399             declare
 400                Tname : constant String := Token_Type'Image (Token);
 401             begin
 402                Error_Msg_SC
 403                  ("|extra " & Tname (5 .. Tname'Last) & "ignored");
 404             end;
 405          end if;
 406
 407          Scan; -- Scan past ignored token
 408       end loop;
 409    end Ignore;
 410
 411    ----------------------------
 412    -- Is_Reserved_Identifier --
 413    ----------------------------
 414
 415    function Is_Reserved_Identifier (C : Id_Check := None) return Boolean is
 416    begin
 417       if not Is_Reserved_Keyword (Token) then
 418          return False;
 419
 420       else
 421          declare
 422             Ident_Casing : constant Casing_Type :=
 423                              Identifier_Casing (Current_Source_File);
 424             Key_Casing   : constant Casing_Type :=
 425                              Keyword_Casing (Current_Source_File);
 426
 427          begin
 428             --  If the casing of identifiers and keywords is different in
 429             --  this source file, and the casing of this token matches the
 430             --  keyword casing, then we return False, since it is pretty
 431             --  clearly intended to be a keyword.
 432
 433             if Ident_Casing = Unknown
 434               or else Key_Casing = Unknown
 435               or else Ident_Casing = Key_Casing
 436               or else Determine_Token_Casing /= Key_Casing
 437             then
 438                return True;
 439
 440             --  Here we have a keyword written clearly with keyword casing.
 441             --  In default mode, we would not be willing to consider this as
 442             --  a reserved identifier, but if C is set, we may still accept it
 443
 444             elsif C /= None then
 445                declare
 446                   Scan_State  : Saved_Scan_State;
 447                   OK_Next_Tok : Boolean;
 448
 449                begin
 450                   Save_Scan_State (Scan_State);
 451                   Scan;
 452
 453                   if Token_Is_At_Start_Of_Line then
 454                      return False;
 455                   end if;
 456
 457                   case C is
 458                      when None =>
 459                         raise Program_Error;
 460
 461                      when C_Comma_Right_Paren =>
 462                         OK_Next_Tok :=
 463                           Token = Tok_Comma or else Token = Tok_Right_Paren;
 464
 465                      when C_Comma_Colon =>
 466                         OK_Next_Tok :=
 467                           Token = Tok_Comma or else Token = Tok_Colon;
 468
 469                      when C_Do =>
 470                         OK_Next_Tok :=
 471                           Token = Tok_Do;
 472
 473                      when C_Dot =>
 474                         OK_Next_Tok :=
 475                           Token = Tok_Dot;
 476
 477                      when C_Greater_Greater =>
 478                         OK_Next_Tok :=
 479                           Token = Tok_Greater_Greater;
 480
 481                      when C_In =>
 482                         OK_Next_Tok :=
 483                           Token = Tok_In;
 484
 485                      when C_Is =>
 486                         OK_Next_Tok :=
 487                           Token = Tok_Is;
 488
 489                      when C_Left_Paren_Semicolon =>
 490                         OK_Next_Tok :=
 491                           Token = Tok_Left_Paren or else Token = Tok_Semicolon;
 492
 493                      when C_Use =>
 494                         OK_Next_Tok :=
 495                           Token = Tok_Use;
 496
 497                      when C_Vertical_Bar_Arrow =>
 498                         OK_Next_Tok :=
 499                           Token = Tok_Vertical_Bar or else Token = Tok_Arrow;
 500                   end case;
 501
 502                   Restore_Scan_State (Scan_State);
 503
 504                   if OK_Next_Tok then
 505                      return True;
 506                   end if;
 507                end;
 508             end if;
 509          end;
 510       end if;
 511
 512       --  If we fall through it is not a reserved identifier
 513
 514       return False;
 515    end Is_Reserved_Identifier;
 516
 517    ----------------------
 518    -- Merge_Identifier --
 519    ----------------------
 520
 521    procedure Merge_Identifier (Prev : Node_Id; Nxt : Token_Type) is
 522    begin
 523       if Token /= Tok_Identifier then
 524          return;
 525       end if;
 526
 527       declare
 528          S : Saved_Scan_State;
 529          T : Token_Type;
 530
 531       begin
 532          Save_Scan_State (S);
 533          Scan;
 534          T := Token;
 535          Restore_Scan_State (S);
 536
 537          if T /= Nxt then
 538             return;
 539          end if;
 540       end;
 541
 542       --  Check exactly one space between identifiers
 543
 544       if Source (Token_Ptr - 1) /= ' '
 545         or else Int (Token_Ptr) /=
 546                   Int (Prev_Token_Ptr) + Length_Of_Name (Chars (Prev)) + 1
 547       then
 548          return;
 549       end if;
 550
 551       --  Do the merge
 552
 553       Get_Name_String (Chars (Token_Node));
 554
 555       declare
 556          Buf : constant String (1 .. Name_Len) :=
 557                  Name_Buffer (1 .. Name_Len);
 558
 559       begin
 560          Get_Name_String (Chars (Prev));
 561          Add_Char_To_Name_Buffer ('_');
 562          Add_Str_To_Name_Buffer (Buf);
 563          Set_Chars (Prev, Name_Find);
 564       end;
 565
 566       Error_Msg_Node_1 := Prev;
 567       Error_Msg_SC
 568         ("unexpected identifier, possibly & was meant here");
 569       Scan;
 570    end Merge_Identifier;
 571
 572    -------------------
 573    -- Next_Token_Is --
 574    -------------------
 575
 576    function Next_Token_Is (Tok : Token_Type) return Boolean is
 577       Scan_State : Saved_Scan_State;
 578       Result     : Boolean;
 579    begin
 580       Save_Scan_State (Scan_State);
 581       Scan;
 582       Result := (Token = Tok);
 583       Restore_Scan_State (Scan_State);
 584       return Result;
 585    end Next_Token_Is;
 586
 587    -------------------
 588    -- No_Constraint --
 589    -------------------
 590
 591    procedure No_Constraint is
 592    begin
 593       if Token in Token_Class_Consk then
 594          Error_Msg_SC ("constraint not allowed here");
 595          Discard_Junk_Node (P_Constraint_Opt);
 596       end if;
 597    end No_Constraint;
 598
 599    ---------------------
 600    -- Pop_Scope_Stack --
 601    ---------------------
 602
 603    procedure Pop_Scope_Stack is
 604    begin
 605       pragma Assert (Scope.Last > 0);
 606       Scope.Decrement_Last;
 607
 608       if Debug_Flag_P then
 609          Error_Msg_Uint_1 := UI_From_Int (Scope.Last);
 610          Error_Msg_SC ("decrement scope stack ptr, new value = ^!");
 611       end if;
 612    end Pop_Scope_Stack;
 613
 614    ----------------------
 615    -- Push_Scope_Stack --
 616    ----------------------
 617
 618    procedure Push_Scope_Stack is
 619    begin
 620       Scope.Increment_Last;
 621
 622       if Style_Check_Max_Nesting_Level
 623         and then Scope.Last = Style_Max_Nesting_Level + 1
 624       then
 625          Error_Msg
 626            ("(style) maximum nesting level exceeded",
 627             First_Non_Blank_Location);
 628       end if;
 629
 630       Scope.Table (Scope.Last).Junk := False;
 631       Scope.Table (Scope.Last).Node := Empty;
 632
 633       if Debug_Flag_P then
 634          Error_Msg_Uint_1 := UI_From_Int (Scope.Last);
 635          Error_Msg_SC ("increment scope stack ptr, new value = ^!");
 636       end if;
 637    end Push_Scope_Stack;
 638
 639    ----------------------
 640    -- Separate_Present --
 641    ----------------------
 642
 643    function Separate_Present return Boolean is
 644       Scan_State : Saved_Scan_State;
 645
 646    begin
 647       if Token = Tok_Separate then
 648          return True;
 649
 650       elsif Token /= Tok_Identifier then
 651          return False;
 652
 653       else
 654          Save_Scan_State (Scan_State);
 655          Scan; -- past identifier
 656
 657          if Token = Tok_Semicolon then
 658             Restore_Scan_State (Scan_State);
 659             return Bad_Spelling_Of (Tok_Separate);
 660
 661          else
 662             Restore_Scan_State (Scan_State);
 663             return False;
 664          end if;
 665       end if;
 666    end Separate_Present;
 667
 668    --------------------------
 669    -- Signal_Bad_Attribute --
 670    --------------------------
 671
 672    procedure Signal_Bad_Attribute is
 673    begin
 674       Error_Msg_N ("unrecognized attribute&", Token_Node);
 675
 676       --  Check for possible misspelling
 677
 678       Error_Msg_Name_1 := First_Attribute_Name;
 679       while Error_Msg_Name_1 <= Last_Attribute_Name loop
 680          if Is_Bad_Spelling_Of (Token_Name, Error_Msg_Name_1) then
 681             Error_Msg_N ("\possible misspelling of %", Token_Node);
 682             exit;
 683          end if;
 684
 685          Error_Msg_Name_1 := Error_Msg_Name_1 + 1;
 686       end loop;
 687    end Signal_Bad_Attribute;
 688
 689    -----------------------------
 690    -- Token_Is_At_End_Of_Line --
 691    -----------------------------
 692
 693    function Token_Is_At_End_Of_Line return Boolean is
 694       S : Source_Ptr;
 695
 696    begin
 697       --  Skip past blanks and horizontal tabs
 698
 699       S := Scan_Ptr;
 700       while Source (S) = ' ' or else Source (S) = ASCII.HT loop
 701          S := S + 1;
 702       end loop;
 703
 704       --  We are at end of line if at a control character (CR/LF/VT/FF/EOF)
 705       --  or if we are at the start of an end of line comment sequence.
 706
 707       return Source (S) < ' '
 708         or else (Source (S) = '-' and then Source (S + 1) = '-');
 709    end Token_Is_At_End_Of_Line;
 710
 711    -------------------------------
 712    -- Token_Is_At_Start_Of_Line --
 713    -------------------------------
 714
 715    function Token_Is_At_Start_Of_Line return Boolean is
 716    begin
 717       return (Token_Ptr = First_Non_Blank_Location or else Token = Tok_EOF);
 718    end Token_Is_At_Start_Of_Line;
 719
 720 end Util;