1 ------------------------------------------------------------------------------
3 -- GNAT COMPILER COMPONENTS --
9 -- Copyright (C) 1992-2010, Free Software Foundation, Inc. --
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. --
21 -- GNAT was originally developed by the GNAT team at New York University. --
22 -- Extensive contributions were provided by Ada Core Technologies Inc. --
24 ------------------------------------------------------------------------------
26 with Csets; use Csets;
27 with Namet.Sp; use Namet.Sp;
28 with Stylesw; use Stylesw;
29 with Uintp; use Uintp;
31 with GNAT.Spelling_Checker; use GNAT.Spelling_Checker;
40 function Bad_Spelling_Of (T : Token_Type) return Boolean is
41 Tname : constant String := Token_Type'Image (T);
42 -- Characters of token name
44 S : String (1 .. Tname'Last - 4);
45 -- Characters of token name folded to lower case, omitting TOK_ at start
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
53 -- Starting subscripts in M1, M2 for keyword name
55 SL : constant Natural := S'Length;
56 -- Length of expected token name excluding TOK_ at start
59 if Token /= Tok_Identifier then
64 S (J) := Fold_Lower (Tname (J + 4));
67 Get_Name_String (Token_Name);
69 -- A special check for case of PROGRAM used for PROCEDURE
73 and then Name_Buffer (1 .. 7) = "program"
75 Error_Msg_SC -- CODEFIX
76 ("PROCEDURE expected");
80 -- A special check for an illegal abbreviation
82 elsif Name_Len < S'Length
83 and then Name_Len >= 4
84 and then Name_Buffer (1 .. Name_Len) = S (1 .. Name_Len)
86 for J in 1 .. S'Last loop
87 M2 (P2 + J - 1) := Fold_Upper (S (J));
90 Error_Msg_SC (M2 (1 .. P2 - 1 + S'Last));
95 -- Now we go into the full circuit to check for a misspelling
97 -- Never consider something a misspelling if either the actual or
98 -- expected string is less than 3 characters (before this check we
99 -- used to consider i to be a misspelled if in some cases!)
101 if SL < 3 or else Name_Len < 3 then
104 -- Special case: prefix matches, i.e. the leading characters of the
105 -- token that we have exactly match the required keyword. If there
106 -- are at least two characters left over, assume that we have a case
107 -- of two keywords joined together which should not be joined.
109 elsif Name_Len > SL + 1
110 and then S = Name_Buffer (1 .. SL)
112 Scan_Ptr := Token_Ptr + S'Length;
113 Error_Msg_S ("|missing space");
118 if Is_Bad_Spelling_Of (Name_Buffer (1 .. Name_Len), S) then
119 for J in 1 .. S'Last loop
120 M1 (P1 + J - 1) := Fold_Upper (S (J));
123 Error_Msg_SC -- CODFIX
124 (M1 (1 .. P1 - 1 + S'Last));
133 ----------------------
134 -- Check_95_Keyword --
135 ----------------------
137 -- On entry, the caller has checked that current token is an identifier
138 -- whose name matches the name of the 95 keyword New_Tok.
140 procedure Check_95_Keyword (Token_95, Next : Token_Type) is
141 Scan_State : Saved_Scan_State;
144 Save_Scan_State (Scan_State); -- at identifier/keyword
145 Scan; -- past identifier/keyword
148 Restore_Scan_State (Scan_State); -- to identifier
149 Error_Msg_Name_1 := Token_Name;
150 Error_Msg_SC ("(Ada 83) keyword* cannot be used!");
153 Restore_Scan_State (Scan_State); -- to identifier
155 end Check_95_Keyword;
157 ----------------------
158 -- Check_Bad_Layout --
159 ----------------------
161 procedure Check_Bad_Layout is
163 if RM_Column_Check and then Token_Is_At_Start_Of_Line
164 and then Start_Column <= Scope.Table (Scope.Last).Ecol
166 Error_Msg_BC -- CODEFIX
167 ("(style) incorrect layout");
169 end Check_Bad_Layout;
171 --------------------------
172 -- Check_Misspelling_Of --
173 --------------------------
175 procedure Check_Misspelling_Of (T : Token_Type) is
177 if Bad_Spelling_Of (T) then
180 end Check_Misspelling_Of;
182 -----------------------------
183 -- Check_Simple_Expression --
184 -----------------------------
186 procedure Check_Simple_Expression (E : Node_Id) is
188 if Expr_Form = EF_Non_Simple then
189 Error_Msg_N ("this expression must be parenthesized", E);
191 end Check_Simple_Expression;
193 ---------------------------------------
194 -- Check_Simple_Expression_In_Ada_83 --
195 ---------------------------------------
197 procedure Check_Simple_Expression_In_Ada_83 (E : Node_Id) is
199 if Expr_Form = EF_Non_Simple then
200 if Ada_Version = Ada_83 then
201 Error_Msg_N ("(Ada 83) this expression must be parenthesized!", E);
204 end Check_Simple_Expression_In_Ada_83;
206 ------------------------
207 -- Check_Subtype_Mark --
208 ------------------------
210 function Check_Subtype_Mark (Mark : Node_Id) return Node_Id is
212 if Nkind (Mark) = N_Identifier
213 or else Nkind (Mark) = N_Selected_Component
214 or else (Nkind (Mark) = N_Attribute_Reference
215 and then Is_Type_Attribute_Name (Attribute_Name (Mark)))
220 Error_Msg ("subtype mark expected", Sloc (Mark));
223 end Check_Subtype_Mark;
229 function Comma_Present return Boolean is
230 Scan_State : Saved_Scan_State;
234 -- First check, if a comma is present, then a comma is present!
236 if Token = Tok_Comma then
240 -- If we have a right paren, then that is taken as ending the list
241 -- i.e. no comma is present.
243 elsif Token = Tok_Right_Paren then
246 -- If pragmas, then get rid of them and make a recursive call
247 -- to process what follows these pragmas.
249 elsif Token = Tok_Pragma then
251 return Comma_Present;
253 -- At this stage we have an error, and the goal is to decide on whether
254 -- or not we should diagnose an error and report a (non-existent)
255 -- comma as being present, or simply to report no comma is present
257 -- If we are a semicolon, then the question is whether we have a missing
258 -- right paren, or whether the semicolon should have been a comma. To
259 -- guess the right answer, we scan ahead keeping track of the paren
260 -- level, looking for a clue that helps us make the right decision.
262 -- This approach is highly accurate in the single error case, and does
263 -- not make bad mistakes in the multiple error case (indeed we can't
264 -- really make a very bad decision at this point in any case).
266 elsif Token = Tok_Semicolon then
267 Save_Scan_State (Scan_State);
268 Scan; -- past semicolon
270 -- Check for being followed by identifier => which almost certainly
271 -- means we are still in a parameter list and the comma should have
272 -- been a semicolon (such a sequence could not follow a semicolon)
274 if Token = Tok_Identifier then
277 if Token = Tok_Arrow then
282 -- If that test didn't work, loop ahead looking for a comma or
283 -- semicolon at the same parenthesis level. Always remember that
284 -- we can't go badly wrong in an error situation like this!
288 -- Here is the look ahead loop, Paren_Count tells us whether the
289 -- token we are looking at is at the same paren level as the
290 -- suspicious semicolon that we are trying to figure out.
294 -- If we hit another semicolon or an end of file, and we have
295 -- not seen a right paren or another comma on the way, then
296 -- probably the semicolon did end the list. Indeed that is
297 -- certainly the only single error correction possible here.
299 if Token = Tok_Semicolon or else Token = Tok_EOF then
300 Restore_Scan_State (Scan_State);
303 -- A comma at the same paren level as the semicolon is a strong
304 -- indicator that the semicolon should have been a comma, indeed
305 -- again this is the only possible single error correction.
307 elsif Token = Tok_Comma then
308 exit when Paren_Count = 0;
310 -- A left paren just bumps the paren count
312 elsif Token = Tok_Left_Paren then
313 Paren_Count := Paren_Count + 1;
315 -- A right paren that is at the same paren level as the semicolon
316 -- also means that the only possible single error correction is
317 -- to assume that the semicolon should have been a comma. If we
318 -- are not at the same paren level, then adjust the paren level.
320 elsif Token = Tok_Right_Paren then
321 exit when Paren_Count = 0;
322 Paren_Count := Paren_Count - 1;
325 -- Keep going, we haven't made a decision yet
330 -- If we fall through the loop, it means that we found a terminating
331 -- right paren or another comma. In either case it is reasonable to
332 -- assume that the semicolon was really intended to be a comma. Also
333 -- come here for the identifier arrow case.
336 Restore_Scan_State (Scan_State);
337 Error_Msg_SC -- CODEFIX
338 ("|"";"" should be "",""");
339 Scan; -- past the semicolon
342 -- If we are not at semicolon or a right paren, then we base the
343 -- decision on whether or not the next token can be part of an
344 -- expression. If not, then decide that no comma is present (the
345 -- caller will eventually generate a missing right parent message)
347 elsif Token in Token_Class_Eterm then
350 -- Otherwise we assume a comma is present, even if none is present,
351 -- since the next token must be part of an expression, so if we were
352 -- at the end of the list, then there is more than one error present.
355 T_Comma; -- to give error
360 -----------------------
361 -- Discard_Junk_List --
362 -----------------------
364 procedure Discard_Junk_List (L : List_Id) is
365 pragma Warnings (Off, L);
368 end Discard_Junk_List;
370 -----------------------
371 -- Discard_Junk_Node --
372 -----------------------
374 procedure Discard_Junk_Node (N : Node_Id) is
375 pragma Warnings (Off, N);
378 end Discard_Junk_Node;
384 procedure Ignore (T : Token_Type) is
387 if T = Tok_Comma then
388 Error_Msg_SC -- CODEFIX
389 ("|extra "","" ignored");
391 elsif T = Tok_Left_Paren then
392 Error_Msg_SC -- CODEFIX
393 ("|extra ""("" ignored");
395 elsif T = Tok_Right_Paren then
396 Error_Msg_SC -- CODEFIX
397 ("|extra "")"" ignored");
399 elsif T = Tok_Semicolon then
400 Error_Msg_SC -- CODEFIX
401 ("|extra "";"" ignored");
403 elsif T = Tok_Colon then
404 Error_Msg_SC -- CODEFIX
405 ("|extra "":"" ignored");
409 Tname : constant String := Token_Type'Image (Token);
411 Error_Msg_SC ("|extra " & Tname (5 .. Tname'Last) & "ignored");
415 Scan; -- Scan past ignored token
419 ----------------------------
420 -- Is_Reserved_Identifier --
421 ----------------------------
423 function Is_Reserved_Identifier (C : Id_Check := None) return Boolean is
425 if not Is_Reserved_Keyword (Token) then
430 Ident_Casing : constant Casing_Type :=
431 Identifier_Casing (Current_Source_File);
432 Key_Casing : constant Casing_Type :=
433 Keyword_Casing (Current_Source_File);
436 -- If the casing of identifiers and keywords is different in
437 -- this source file, and the casing of this token matches the
438 -- keyword casing, then we return False, since it is pretty
439 -- clearly intended to be a keyword.
441 if Ident_Casing = Unknown
442 or else Key_Casing = Unknown
443 or else Ident_Casing = Key_Casing
444 or else Determine_Token_Casing /= Key_Casing
448 -- Here we have a keyword written clearly with keyword casing.
449 -- In default mode, we would not be willing to consider this as
450 -- a reserved identifier, but if C is set, we may still accept it
454 Scan_State : Saved_Scan_State;
455 OK_Next_Tok : Boolean;
458 Save_Scan_State (Scan_State);
461 if Token_Is_At_Start_Of_Line then
469 when C_Comma_Right_Paren =>
471 Token = Tok_Comma or else Token = Tok_Right_Paren;
473 when C_Comma_Colon =>
475 Token = Tok_Comma or else Token = Tok_Colon;
485 when C_Greater_Greater =>
487 Token = Tok_Greater_Greater;
497 when C_Left_Paren_Semicolon =>
499 Token = Tok_Left_Paren or else Token = Tok_Semicolon;
505 when C_Vertical_Bar_Arrow =>
507 Token = Tok_Vertical_Bar or else Token = Tok_Arrow;
510 Restore_Scan_State (Scan_State);
520 -- If we fall through it is not a reserved identifier
523 end Is_Reserved_Identifier;
525 ----------------------
526 -- Merge_Identifier --
527 ----------------------
529 procedure Merge_Identifier (Prev : Node_Id; Nxt : Token_Type) is
531 if Token /= Tok_Identifier then
536 S : Saved_Scan_State;
543 Restore_Scan_State (S);
550 -- Check exactly one space between identifiers
552 if Source (Token_Ptr - 1) /= ' '
553 or else Int (Token_Ptr) /=
554 Int (Prev_Token_Ptr) + Length_Of_Name (Chars (Prev)) + 1
561 Get_Name_String (Chars (Token_Node));
564 Buf : constant String (1 .. Name_Len) :=
565 Name_Buffer (1 .. Name_Len);
568 Get_Name_String (Chars (Prev));
569 Add_Char_To_Name_Buffer ('_');
570 Add_Str_To_Name_Buffer (Buf);
571 Set_Chars (Prev, Name_Find);
574 Error_Msg_Node_1 := Prev;
575 Error_Msg_SC ("unexpected identifier, possibly & was meant here");
577 end Merge_Identifier;
583 function Next_Token_Is (Tok : Token_Type) return Boolean is
584 Scan_State : Saved_Scan_State;
587 Save_Scan_State (Scan_State);
589 Result := (Token = Tok);
590 Restore_Scan_State (Scan_State);
598 procedure No_Constraint is
600 if Token in Token_Class_Consk then
601 Error_Msg_SC ("constraint not allowed here");
602 Discard_Junk_Node (P_Constraint_Opt);
606 ---------------------
607 -- Pop_Scope_Stack --
608 ---------------------
610 procedure Pop_Scope_Stack is
612 pragma Assert (Scope.Last > 0);
613 Scope.Decrement_Last;
616 Error_Msg_Uint_1 := UI_From_Int (Scope.Last);
617 Error_Msg_SC ("decrement scope stack ptr, new value = ^!");
621 ----------------------
622 -- Push_Scope_Stack --
623 ----------------------
625 procedure Push_Scope_Stack is
627 Scope.Increment_Last;
629 if Style_Check_Max_Nesting_Level
630 and then Scope.Last = Style_Max_Nesting_Level + 1
633 ("(style) maximum nesting level exceeded",
634 First_Non_Blank_Location);
637 Scope.Table (Scope.Last).Junk := False;
638 Scope.Table (Scope.Last).Node := Empty;
641 Error_Msg_Uint_1 := UI_From_Int (Scope.Last);
642 Error_Msg_SC ("increment scope stack ptr, new value = ^!");
644 end Push_Scope_Stack;
646 ----------------------
647 -- Separate_Present --
648 ----------------------
650 function Separate_Present return Boolean is
651 Scan_State : Saved_Scan_State;
654 if Token = Tok_Separate then
657 elsif Token /= Tok_Identifier then
661 Save_Scan_State (Scan_State);
662 Scan; -- past identifier
664 if Token = Tok_Semicolon then
665 Restore_Scan_State (Scan_State);
666 return Bad_Spelling_Of (Tok_Separate);
669 Restore_Scan_State (Scan_State);
673 end Separate_Present;
675 --------------------------
676 -- Signal_Bad_Attribute --
677 --------------------------
679 procedure Signal_Bad_Attribute is
681 Error_Msg_N ("unrecognized attribute&", Token_Node);
683 -- Check for possible misspelling
685 Error_Msg_Name_1 := First_Attribute_Name;
686 while Error_Msg_Name_1 <= Last_Attribute_Name loop
687 if Is_Bad_Spelling_Of (Token_Name, Error_Msg_Name_1) then
688 Error_Msg_N -- CODEFIX
689 ("\possible misspelling of %", Token_Node);
693 Error_Msg_Name_1 := Error_Msg_Name_1 + 1;
695 end Signal_Bad_Attribute;
697 -----------------------------
698 -- Token_Is_At_End_Of_Line --
699 -----------------------------
701 function Token_Is_At_End_Of_Line return Boolean is
705 -- Skip past blanks and horizontal tabs
708 while Source (S) = ' ' or else Source (S) = ASCII.HT loop
712 -- We are at end of line if at a control character (CR/LF/VT/FF/EOF)
713 -- or if we are at the start of an end of line comment sequence.
715 return Source (S) < ' '
716 or else (Source (S) = '-' and then Source (S + 1) = '-');
717 end Token_Is_At_End_Of_Line;
719 -------------------------------
720 -- Token_Is_At_Start_Of_Line --
721 -------------------------------
723 function Token_Is_At_Start_Of_Line return Boolean is
725 return (Token_Ptr = First_Non_Blank_Location or else Token = Tok_EOF);
726 end Token_Is_At_Start_Of_Line;