* g++.dg/cpp0x/constexpr-53094-2.C: Ignore non-standard ABI
[official-gcc.git] / gcc / ada / par-ch4.adb
blob185a07d97c2c5541f49513ccee8fb0c1a500d8d6
1 ------------------------------------------------------------------------------
2 -- --
3 -- GNAT COMPILER COMPONENTS --
4 -- --
5 -- P A R . C H 4 --
6 -- --
7 -- B o d y --
8 -- --
9 -- Copyright (C) 1992-2012, 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 ------------------------------------------------------------------------------
26 pragma Style_Checks (All_Checks);
27 -- Turn off subprogram body ordering check. Subprograms are in order
28 -- by RM section rather than alphabetical
30 with Stringt; use Stringt;
32 separate (Par)
33 package body Ch4 is
35 -- Attributes that cannot have arguments
37 Is_Parameterless_Attribute : constant Attribute_Class_Array :=
38 (Attribute_Base => True,
39 Attribute_Body_Version => True,
40 Attribute_Class => True,
41 Attribute_External_Tag => True,
42 Attribute_Img => True,
43 Attribute_Stub_Type => True,
44 Attribute_Version => True,
45 Attribute_Type_Key => True,
46 others => False);
47 -- This map contains True for parameterless attributes that return a
48 -- string or a type. For those attributes, a left parenthesis after
49 -- the attribute should not be analyzed as the beginning of a parameters
50 -- list because it may denote a slice operation (X'Img (1 .. 2)) or
51 -- a type conversion (X'Class (Y)).
53 -- Note that this map designates the minimum set of attributes where a
54 -- construct in parentheses that is not an argument can appear right
55 -- after the attribute. For attributes like 'Size, we do not put them
56 -- in the map. If someone writes X'Size (3), that's illegal in any case,
57 -- but we get a better error message by parsing the (3) as an illegal
58 -- argument to the attribute, rather than some meaningless junk that
59 -- follows the attribute.
61 -----------------------
62 -- Local Subprograms --
63 -----------------------
65 function P_Aggregate_Or_Paren_Expr return Node_Id;
66 function P_Allocator return Node_Id;
67 function P_Case_Expression_Alternative return Node_Id;
68 function P_Record_Or_Array_Component_Association return Node_Id;
69 function P_Factor return Node_Id;
70 function P_Primary return Node_Id;
71 function P_Relation return Node_Id;
72 function P_Term return Node_Id;
74 function P_Binary_Adding_Operator return Node_Kind;
75 function P_Logical_Operator return Node_Kind;
76 function P_Multiplying_Operator return Node_Kind;
77 function P_Relational_Operator return Node_Kind;
78 function P_Unary_Adding_Operator return Node_Kind;
80 procedure Bad_Range_Attribute (Loc : Source_Ptr);
81 -- Called to place complaint about bad range attribute at the given
82 -- source location. Terminates by raising Error_Resync.
84 procedure Check_Bad_Exp;
85 -- Called after scanning a**b, posts error if ** detected
87 procedure P_Membership_Test (N : Node_Id);
88 -- N is the node for a N_In or N_Not_In node whose right operand has not
89 -- yet been processed. It is called just after scanning out the IN keyword.
90 -- On return, either Right_Opnd or Alternatives is set, as appropriate.
92 function P_Range_Attribute_Reference (Prefix_Node : Node_Id) return Node_Id;
93 -- Scan a range attribute reference. The caller has scanned out the
94 -- prefix. The current token is known to be an apostrophe and the
95 -- following token is known to be RANGE.
97 function P_Unparen_Cond_Case_Quant_Expression return Node_Id;
98 -- This function is called with Token pointing to IF, CASE, or FOR, in a
99 -- context that allows a case, conditional, or quantified expression if
100 -- it is surrounded by parentheses. If not surrounded by parentheses, the
101 -- expression is still returned, but an error message is issued.
103 -------------------------
104 -- Bad_Range_Attribute --
105 -------------------------
107 procedure Bad_Range_Attribute (Loc : Source_Ptr) is
108 begin
109 Error_Msg ("range attribute cannot be used in expression!", Loc);
110 Resync_Expression;
111 end Bad_Range_Attribute;
113 -------------------
114 -- Check_Bad_Exp --
115 -------------------
117 procedure Check_Bad_Exp is
118 begin
119 if Token = Tok_Double_Asterisk then
120 Error_Msg_SC ("parenthesization required for '*'*");
121 Scan; -- past **
122 Discard_Junk_Node (P_Primary);
123 Check_Bad_Exp;
124 end if;
125 end Check_Bad_Exp;
127 --------------------------
128 -- 4.1 Name (also 6.4) --
129 --------------------------
131 -- NAME ::=
132 -- DIRECT_NAME | EXPLICIT_DEREFERENCE
133 -- | INDEXED_COMPONENT | SLICE
134 -- | SELECTED_COMPONENT | ATTRIBUTE
135 -- | TYPE_CONVERSION | FUNCTION_CALL
136 -- | CHARACTER_LITERAL
138 -- DIRECT_NAME ::= IDENTIFIER | OPERATOR_SYMBOL
140 -- PREFIX ::= NAME | IMPLICIT_DEREFERENCE
142 -- EXPLICIT_DEREFERENCE ::= NAME . all
144 -- IMPLICIT_DEREFERENCE ::= NAME
146 -- INDEXED_COMPONENT ::= PREFIX (EXPRESSION {, EXPRESSION})
148 -- SLICE ::= PREFIX (DISCRETE_RANGE)
150 -- SELECTED_COMPONENT ::= PREFIX . SELECTOR_NAME
152 -- SELECTOR_NAME ::= IDENTIFIER | CHARACTER_LITERAL | OPERATOR_SYMBOL
154 -- ATTRIBUTE_REFERENCE ::= PREFIX ' ATTRIBUTE_DESIGNATOR
156 -- ATTRIBUTE_DESIGNATOR ::=
157 -- IDENTIFIER [(static_EXPRESSION)]
158 -- | access | delta | digits
160 -- FUNCTION_CALL ::=
161 -- function_NAME
162 -- | function_PREFIX ACTUAL_PARAMETER_PART
164 -- ACTUAL_PARAMETER_PART ::=
165 -- (PARAMETER_ASSOCIATION {,PARAMETER_ASSOCIATION})
167 -- PARAMETER_ASSOCIATION ::=
168 -- [formal_parameter_SELECTOR_NAME =>] EXPLICIT_ACTUAL_PARAMETER
170 -- EXPLICIT_ACTUAL_PARAMETER ::= EXPRESSION | variable_NAME
172 -- Note: syntactically a procedure call looks just like a function call,
173 -- so this routine is in practice used to scan out procedure calls as well.
175 -- On return, Expr_Form is set to either EF_Name or EF_Simple_Name
177 -- Error recovery: can raise Error_Resync
179 -- Note: if on return Token = Tok_Apostrophe, then the apostrophe must be
180 -- followed by either a left paren (qualified expression case), or by
181 -- range (range attribute case). All other uses of apostrophe (i.e. all
182 -- other attributes) are handled in this routine.
184 -- Error recovery: can raise Error_Resync
186 function P_Name return Node_Id is
187 Scan_State : Saved_Scan_State;
188 Name_Node : Node_Id;
189 Prefix_Node : Node_Id;
190 Ident_Node : Node_Id;
191 Expr_Node : Node_Id;
192 Range_Node : Node_Id;
193 Arg_Node : Node_Id;
195 Arg_List : List_Id := No_List; -- kill junk warning
196 Attr_Name : Name_Id := No_Name; -- kill junk warning
198 begin
199 -- Case of not a name
201 if Token not in Token_Class_Name then
203 -- If it looks like start of expression, complain and scan expression
205 if Token in Token_Class_Literal
206 or else Token = Tok_Left_Paren
207 then
208 Error_Msg_SC ("name expected");
209 return P_Expression;
211 -- Otherwise some other junk, not much we can do
213 else
214 Error_Msg_AP ("name expected");
215 raise Error_Resync;
216 end if;
217 end if;
219 -- Loop through designators in qualified name
221 Name_Node := Token_Node;
223 loop
224 Scan; -- past designator
225 exit when Token /= Tok_Dot;
226 Save_Scan_State (Scan_State); -- at dot
227 Scan; -- past dot
229 -- If we do not have another designator after the dot, then join
230 -- the normal circuit to handle a dot extension (may be .all or
231 -- character literal case). Otherwise loop back to scan the next
232 -- designator.
234 if Token not in Token_Class_Desig then
235 goto Scan_Name_Extension_Dot;
236 else
237 Prefix_Node := Name_Node;
238 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
239 Set_Prefix (Name_Node, Prefix_Node);
240 Set_Selector_Name (Name_Node, Token_Node);
241 end if;
242 end loop;
244 -- We have now scanned out a qualified designator. If the last token is
245 -- an operator symbol, then we certainly do not have the Snam case, so
246 -- we can just use the normal name extension check circuit
248 if Prev_Token = Tok_Operator_Symbol then
249 goto Scan_Name_Extension;
250 end if;
252 -- We have scanned out a qualified simple name, check for name extension
253 -- Note that we know there is no dot here at this stage, so the only
254 -- possible cases of name extension are apostrophe and left paren.
256 if Token = Tok_Apostrophe then
257 Save_Scan_State (Scan_State); -- at apostrophe
258 Scan; -- past apostrophe
260 -- Qualified expression in Ada 2012 mode (treated as a name)
262 if Ada_Version >= Ada_2012 and then Token = Tok_Left_Paren then
263 goto Scan_Name_Extension_Apostrophe;
265 -- If left paren not in Ada 2012, then it is not part of the name,
266 -- since qualified expressions are not names in prior versions of
267 -- Ada, so return with Token backed up to point to the apostrophe.
268 -- The treatment for the range attribute is similar (we do not
269 -- consider x'range to be a name in this grammar).
271 elsif Token = Tok_Left_Paren or else Token = Tok_Range then
272 Restore_Scan_State (Scan_State); -- to apostrophe
273 Expr_Form := EF_Simple_Name;
274 return Name_Node;
276 -- Otherwise we have the case of a name extended by an attribute
278 else
279 goto Scan_Name_Extension_Apostrophe;
280 end if;
282 -- Check case of qualified simple name extended by a left parenthesis
284 elsif Token = Tok_Left_Paren then
285 Scan; -- past left paren
286 goto Scan_Name_Extension_Left_Paren;
288 -- Otherwise the qualified simple name is not extended, so return
290 else
291 Expr_Form := EF_Simple_Name;
292 return Name_Node;
293 end if;
295 -- Loop scanning past name extensions. A label is used for control
296 -- transfer for this loop for ease of interfacing with the finite state
297 -- machine in the parenthesis scanning circuit, and also to allow for
298 -- passing in control to the appropriate point from the above code.
300 <<Scan_Name_Extension>>
302 -- Character literal used as name cannot be extended. Also this
303 -- cannot be a call, since the name for a call must be a designator.
304 -- Return in these cases, or if there is no name extension
306 if Token not in Token_Class_Namext
307 or else Prev_Token = Tok_Char_Literal
308 then
309 Expr_Form := EF_Name;
310 return Name_Node;
311 end if;
313 -- Merge here when we know there is a name extension
315 <<Scan_Name_Extension_OK>>
317 if Token = Tok_Left_Paren then
318 Scan; -- past left paren
319 goto Scan_Name_Extension_Left_Paren;
321 elsif Token = Tok_Apostrophe then
322 Save_Scan_State (Scan_State); -- at apostrophe
323 Scan; -- past apostrophe
324 goto Scan_Name_Extension_Apostrophe;
326 else -- Token = Tok_Dot
327 Save_Scan_State (Scan_State); -- at dot
328 Scan; -- past dot
329 goto Scan_Name_Extension_Dot;
330 end if;
332 -- Case of name extended by dot (selection), dot is already skipped
333 -- and the scan state at the point of the dot is saved in Scan_State.
335 <<Scan_Name_Extension_Dot>>
337 -- Explicit dereference case
339 if Token = Tok_All then
340 Prefix_Node := Name_Node;
341 Name_Node := New_Node (N_Explicit_Dereference, Token_Ptr);
342 Set_Prefix (Name_Node, Prefix_Node);
343 Scan; -- past ALL
344 goto Scan_Name_Extension;
346 -- Selected component case
348 elsif Token in Token_Class_Name then
349 Prefix_Node := Name_Node;
350 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
351 Set_Prefix (Name_Node, Prefix_Node);
352 Set_Selector_Name (Name_Node, Token_Node);
353 Scan; -- past selector
354 goto Scan_Name_Extension;
356 -- Reserved identifier as selector
358 elsif Is_Reserved_Identifier then
359 Scan_Reserved_Identifier (Force_Msg => False);
360 Prefix_Node := Name_Node;
361 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
362 Set_Prefix (Name_Node, Prefix_Node);
363 Set_Selector_Name (Name_Node, Token_Node);
364 Scan; -- past identifier used as selector
365 goto Scan_Name_Extension;
367 -- If dot is at end of line and followed by nothing legal,
368 -- then assume end of name and quit (dot will be taken as
369 -- an erroneous form of some other punctuation by our caller).
371 elsif Token_Is_At_Start_Of_Line then
372 Restore_Scan_State (Scan_State);
373 return Name_Node;
375 -- Here if nothing legal after the dot
377 else
378 Error_Msg_AP ("selector expected");
379 raise Error_Resync;
380 end if;
382 -- Here for an apostrophe as name extension. The scan position at the
383 -- apostrophe has already been saved, and the apostrophe scanned out.
385 <<Scan_Name_Extension_Apostrophe>>
387 Scan_Apostrophe : declare
388 function Apostrophe_Should_Be_Semicolon return Boolean;
389 -- Checks for case where apostrophe should probably be
390 -- a semicolon, and if so, gives appropriate message,
391 -- resets the scan pointer to the apostrophe, changes
392 -- the current token to Tok_Semicolon, and returns True.
393 -- Otherwise returns False.
395 ------------------------------------
396 -- Apostrophe_Should_Be_Semicolon --
397 ------------------------------------
399 function Apostrophe_Should_Be_Semicolon return Boolean is
400 begin
401 if Token_Is_At_Start_Of_Line then
402 Restore_Scan_State (Scan_State); -- to apostrophe
403 Error_Msg_SC ("|""''"" should be "";""");
404 Token := Tok_Semicolon;
405 return True;
406 else
407 return False;
408 end if;
409 end Apostrophe_Should_Be_Semicolon;
411 -- Start of processing for Scan_Apostrophe
413 begin
414 -- Check for qualified expression case in Ada 2012 mode
416 if Ada_Version >= Ada_2012 and then Token = Tok_Left_Paren then
417 Name_Node := P_Qualified_Expression (Name_Node);
418 goto Scan_Name_Extension;
420 -- If range attribute after apostrophe, then return with Token
421 -- pointing to the apostrophe. Note that in this case the prefix
422 -- need not be a simple name (cases like A.all'range). Similarly
423 -- if there is a left paren after the apostrophe, then we also
424 -- return with Token pointing to the apostrophe (this is the
425 -- aggregate case, or some error case).
427 elsif Token = Tok_Range or else Token = Tok_Left_Paren then
428 Restore_Scan_State (Scan_State); -- to apostrophe
429 Expr_Form := EF_Name;
430 return Name_Node;
432 -- Here for cases where attribute designator is an identifier
434 elsif Token = Tok_Identifier then
435 Attr_Name := Token_Name;
437 if not Is_Attribute_Name (Attr_Name) then
438 if Apostrophe_Should_Be_Semicolon then
439 Expr_Form := EF_Name;
440 return Name_Node;
442 -- Here for a bad attribute name
444 else
445 Signal_Bad_Attribute;
446 Scan; -- past bad identifier
448 if Token = Tok_Left_Paren then
449 Scan; -- past left paren
451 loop
452 Discard_Junk_Node (P_Expression_If_OK);
453 exit when not Comma_Present;
454 end loop;
456 T_Right_Paren;
457 end if;
459 return Error;
460 end if;
461 end if;
463 if Style_Check then
464 Style.Check_Attribute_Name (False);
465 end if;
467 -- Here for case of attribute designator is not an identifier
469 else
470 if Token = Tok_Delta then
471 Attr_Name := Name_Delta;
473 elsif Token = Tok_Digits then
474 Attr_Name := Name_Digits;
476 elsif Token = Tok_Access then
477 Attr_Name := Name_Access;
479 elsif Token = Tok_Mod and then Ada_Version >= Ada_95 then
480 Attr_Name := Name_Mod;
482 elsif Apostrophe_Should_Be_Semicolon then
483 Expr_Form := EF_Name;
484 return Name_Node;
486 else
487 Error_Msg_AP ("attribute designator expected");
488 raise Error_Resync;
489 end if;
491 if Style_Check then
492 Style.Check_Attribute_Name (True);
493 end if;
494 end if;
496 -- We come here with an OK attribute scanned, and corresponding
497 -- Attribute identifier node stored in Ident_Node.
499 Prefix_Node := Name_Node;
500 Name_Node := New_Node (N_Attribute_Reference, Prev_Token_Ptr);
501 Scan; -- past attribute designator
502 Set_Prefix (Name_Node, Prefix_Node);
503 Set_Attribute_Name (Name_Node, Attr_Name);
505 -- Scan attribute arguments/designator. We skip this if we know
506 -- that the attribute cannot have an argument.
508 if Token = Tok_Left_Paren
509 and then not
510 Is_Parameterless_Attribute (Get_Attribute_Id (Attr_Name))
511 then
512 Set_Expressions (Name_Node, New_List);
514 -- Attribute Update contains an array or record association
515 -- list which provides new values for various components or
516 -- elements. The list is parsed as an aggregate.
518 if Attr_Name = Name_Update then
519 Append (P_Aggregate, Expressions (Name_Node));
521 else
522 Scan; -- past left paren
524 loop
525 declare
526 Expr : constant Node_Id := P_Expression_If_OK;
528 begin
529 if Token = Tok_Arrow then
530 Error_Msg_SC
531 ("named parameters not permitted for attributes");
532 Scan; -- past junk arrow
534 else
535 Append (Expr, Expressions (Name_Node));
536 exit when not Comma_Present;
537 end if;
538 end;
539 end loop;
541 T_Right_Paren;
542 end if;
543 end if;
545 goto Scan_Name_Extension;
546 end Scan_Apostrophe;
548 -- Here for left parenthesis extending name (left paren skipped)
550 <<Scan_Name_Extension_Left_Paren>>
552 -- We now have to scan through a list of items, terminated by a
553 -- right parenthesis. The scan is handled by a finite state
554 -- machine. The possibilities are:
556 -- (discrete_range)
558 -- This is a slice. This case is handled in LP_State_Init
560 -- (expression, expression, ..)
562 -- This is interpreted as an indexed component, i.e. as a
563 -- case of a name which can be extended in the normal manner.
564 -- This case is handled by LP_State_Name or LP_State_Expr.
566 -- Note: if and case expressions (without an extra level of
567 -- parentheses) are permitted in this context).
569 -- (..., identifier => expression , ...)
571 -- If there is at least one occurrence of identifier => (but
572 -- none of the other cases apply), then we have a call.
574 -- Test for Id => case
576 if Token = Tok_Identifier then
577 Save_Scan_State (Scan_State); -- at Id
578 Scan; -- past Id
580 -- Test for => (allow := as an error substitute)
582 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
583 Restore_Scan_State (Scan_State); -- to Id
584 Arg_List := New_List;
585 goto LP_State_Call;
587 else
588 Restore_Scan_State (Scan_State); -- to Id
589 end if;
590 end if;
592 -- Here we have an expression after all
594 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
596 -- Check cases of discrete range for a slice
598 -- First possibility: Range_Attribute_Reference
600 if Expr_Form = EF_Range_Attr then
601 Range_Node := Expr_Node;
603 -- Second possibility: Simple_expression .. Simple_expression
605 elsif Token = Tok_Dot_Dot then
606 Check_Simple_Expression (Expr_Node);
607 Range_Node := New_Node (N_Range, Token_Ptr);
608 Set_Low_Bound (Range_Node, Expr_Node);
609 Scan; -- past ..
610 Expr_Node := P_Expression;
611 Check_Simple_Expression (Expr_Node);
612 Set_High_Bound (Range_Node, Expr_Node);
614 -- Third possibility: Type_name range Range
616 elsif Token = Tok_Range then
617 if Expr_Form /= EF_Simple_Name then
618 Error_Msg_SC ("subtype mark must precede RANGE");
619 raise Error_Resync;
620 end if;
622 Range_Node := P_Subtype_Indication (Expr_Node);
624 -- Otherwise we just have an expression. It is true that we might
625 -- have a subtype mark without a range constraint but this case
626 -- is syntactically indistinguishable from the expression case.
628 else
629 Arg_List := New_List;
630 goto LP_State_Expr;
631 end if;
633 -- Fall through here with unmistakable Discrete range scanned,
634 -- which means that we definitely have the case of a slice. The
635 -- Discrete range is in Range_Node.
637 if Token = Tok_Comma then
638 Error_Msg_SC ("slice cannot have more than one dimension");
639 raise Error_Resync;
641 elsif Token /= Tok_Right_Paren then
642 if Token = Tok_Arrow then
644 -- This may be an aggregate that is missing a qualification
646 Error_Msg_SC
647 ("context of aggregate must be a qualified expression");
648 raise Error_Resync;
650 else
651 T_Right_Paren;
652 raise Error_Resync;
653 end if;
655 else
656 Scan; -- past right paren
657 Prefix_Node := Name_Node;
658 Name_Node := New_Node (N_Slice, Sloc (Prefix_Node));
659 Set_Prefix (Name_Node, Prefix_Node);
660 Set_Discrete_Range (Name_Node, Range_Node);
662 -- An operator node is legal as a prefix to other names,
663 -- but not for a slice.
665 if Nkind (Prefix_Node) = N_Operator_Symbol then
666 Error_Msg_N ("illegal prefix for slice", Prefix_Node);
667 end if;
669 -- If we have a name extension, go scan it
671 if Token in Token_Class_Namext then
672 goto Scan_Name_Extension_OK;
674 -- Otherwise return (a slice is a name, but is not a call)
676 else
677 Expr_Form := EF_Name;
678 return Name_Node;
679 end if;
680 end if;
682 -- In LP_State_Expr, we have scanned one or more expressions, and
683 -- so we have a call or an indexed component which is a name. On
684 -- entry we have the expression just scanned in Expr_Node and
685 -- Arg_List contains the list of expressions encountered so far
687 <<LP_State_Expr>>
688 Append (Expr_Node, Arg_List);
690 if Token = Tok_Arrow then
691 Error_Msg
692 ("expect identifier in parameter association",
693 Sloc (Expr_Node));
694 Scan; -- past arrow
696 elsif not Comma_Present then
697 T_Right_Paren;
698 Prefix_Node := Name_Node;
699 Name_Node := New_Node (N_Indexed_Component, Sloc (Prefix_Node));
700 Set_Prefix (Name_Node, Prefix_Node);
701 Set_Expressions (Name_Node, Arg_List);
702 goto Scan_Name_Extension;
703 end if;
705 -- Comma present (and scanned out), test for identifier => case
706 -- Test for identifier => case
708 if Token = Tok_Identifier then
709 Save_Scan_State (Scan_State); -- at Id
710 Scan; -- past Id
712 -- Test for => (allow := as error substitute)
714 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
715 Restore_Scan_State (Scan_State); -- to Id
716 goto LP_State_Call;
718 -- Otherwise it's just an expression after all, so backup
720 else
721 Restore_Scan_State (Scan_State); -- to Id
722 end if;
723 end if;
725 -- Here we have an expression after all, so stay in this state
727 Expr_Node := P_Expression_If_OK;
728 goto LP_State_Expr;
730 -- LP_State_Call corresponds to the situation in which at least
731 -- one instance of Id => Expression has been encountered, so we
732 -- know that we do not have a name, but rather a call. We enter
733 -- it with the scan pointer pointing to the next argument to scan,
734 -- and Arg_List containing the list of arguments scanned so far.
736 <<LP_State_Call>>
738 -- Test for case of Id => Expression (named parameter)
740 if Token = Tok_Identifier then
741 Save_Scan_State (Scan_State); -- at Id
742 Ident_Node := Token_Node;
743 Scan; -- past Id
745 -- Deal with => (allow := as erroneous substitute)
747 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
748 Arg_Node := New_Node (N_Parameter_Association, Prev_Token_Ptr);
749 Set_Selector_Name (Arg_Node, Ident_Node);
750 T_Arrow;
751 Set_Explicit_Actual_Parameter (Arg_Node, P_Expression);
752 Append (Arg_Node, Arg_List);
754 -- If a comma follows, go back and scan next entry
756 if Comma_Present then
757 goto LP_State_Call;
759 -- Otherwise we have the end of a call
761 else
762 Prefix_Node := Name_Node;
763 Name_Node := New_Node (N_Function_Call, Sloc (Prefix_Node));
764 Set_Name (Name_Node, Prefix_Node);
765 Set_Parameter_Associations (Name_Node, Arg_List);
766 T_Right_Paren;
768 if Token in Token_Class_Namext then
769 goto Scan_Name_Extension_OK;
771 -- This is a case of a call which cannot be a name
773 else
774 Expr_Form := EF_Name;
775 return Name_Node;
776 end if;
777 end if;
779 -- Not named parameter: Id started an expression after all
781 else
782 Restore_Scan_State (Scan_State); -- to Id
783 end if;
784 end if;
786 -- Here if entry did not start with Id => which means that it
787 -- is a positional parameter, which is not allowed, since we
788 -- have seen at least one named parameter already.
790 Error_Msg_SC
791 ("positional parameter association " &
792 "not allowed after named one");
794 Expr_Node := P_Expression_If_OK;
796 -- Leaving the '>' in an association is not unusual, so suggest
797 -- a possible fix.
799 if Nkind (Expr_Node) = N_Op_Eq then
800 Error_Msg_N ("\maybe `='>` was intended", Expr_Node);
801 end if;
803 -- We go back to scanning out expressions, so that we do not get
804 -- multiple error messages when several positional parameters
805 -- follow a named parameter.
807 goto LP_State_Expr;
809 -- End of treatment for name extensions starting with left paren
811 -- End of loop through name extensions
813 end P_Name;
815 -- This function parses a restricted form of Names which are either
816 -- designators, or designators preceded by a sequence of prefixes
817 -- that are direct names.
819 -- Error recovery: cannot raise Error_Resync
821 function P_Function_Name return Node_Id is
822 Designator_Node : Node_Id;
823 Prefix_Node : Node_Id;
824 Selector_Node : Node_Id;
825 Dot_Sloc : Source_Ptr := No_Location;
827 begin
828 -- Prefix_Node is set to the gathered prefix so far, Empty means that
829 -- no prefix has been scanned. This allows us to build up the result
830 -- in the required right recursive manner.
832 Prefix_Node := Empty;
834 -- Loop through prefixes
836 loop
837 Designator_Node := Token_Node;
839 if Token not in Token_Class_Desig then
840 return P_Identifier; -- let P_Identifier issue the error message
842 else -- Token in Token_Class_Desig
843 Scan; -- past designator
844 exit when Token /= Tok_Dot;
845 end if;
847 -- Here at a dot, with token just before it in Designator_Node
849 if No (Prefix_Node) then
850 Prefix_Node := Designator_Node;
851 else
852 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
853 Set_Prefix (Selector_Node, Prefix_Node);
854 Set_Selector_Name (Selector_Node, Designator_Node);
855 Prefix_Node := Selector_Node;
856 end if;
858 Dot_Sloc := Token_Ptr;
859 Scan; -- past dot
860 end loop;
862 -- Fall out of the loop having just scanned a designator
864 if No (Prefix_Node) then
865 return Designator_Node;
866 else
867 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
868 Set_Prefix (Selector_Node, Prefix_Node);
869 Set_Selector_Name (Selector_Node, Designator_Node);
870 return Selector_Node;
871 end if;
873 exception
874 when Error_Resync =>
875 return Error;
876 end P_Function_Name;
878 -- This function parses a restricted form of Names which are either
879 -- identifiers, or identifiers preceded by a sequence of prefixes
880 -- that are direct names.
882 -- Error recovery: cannot raise Error_Resync
884 function P_Qualified_Simple_Name return Node_Id is
885 Designator_Node : Node_Id;
886 Prefix_Node : Node_Id;
887 Selector_Node : Node_Id;
888 Dot_Sloc : Source_Ptr := No_Location;
890 begin
891 -- Prefix node is set to the gathered prefix so far, Empty means that
892 -- no prefix has been scanned. This allows us to build up the result
893 -- in the required right recursive manner.
895 Prefix_Node := Empty;
897 -- Loop through prefixes
899 loop
900 Designator_Node := Token_Node;
902 if Token = Tok_Identifier then
903 Scan; -- past identifier
904 exit when Token /= Tok_Dot;
906 elsif Token not in Token_Class_Desig then
907 return P_Identifier; -- let P_Identifier issue the error message
909 else
910 Scan; -- past designator
912 if Token /= Tok_Dot then
913 Error_Msg_SP ("identifier expected");
914 return Error;
915 end if;
916 end if;
918 -- Here at a dot, with token just before it in Designator_Node
920 if No (Prefix_Node) then
921 Prefix_Node := Designator_Node;
922 else
923 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
924 Set_Prefix (Selector_Node, Prefix_Node);
925 Set_Selector_Name (Selector_Node, Designator_Node);
926 Prefix_Node := Selector_Node;
927 end if;
929 Dot_Sloc := Token_Ptr;
930 Scan; -- past dot
931 end loop;
933 -- Fall out of the loop having just scanned an identifier
935 if No (Prefix_Node) then
936 return Designator_Node;
937 else
938 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
939 Set_Prefix (Selector_Node, Prefix_Node);
940 Set_Selector_Name (Selector_Node, Designator_Node);
941 return Selector_Node;
942 end if;
944 exception
945 when Error_Resync =>
946 return Error;
947 end P_Qualified_Simple_Name;
949 -- This procedure differs from P_Qualified_Simple_Name only in that it
950 -- raises Error_Resync if any error is encountered. It only returns after
951 -- scanning a valid qualified simple name.
953 -- Error recovery: can raise Error_Resync
955 function P_Qualified_Simple_Name_Resync return Node_Id is
956 Designator_Node : Node_Id;
957 Prefix_Node : Node_Id;
958 Selector_Node : Node_Id;
959 Dot_Sloc : Source_Ptr := No_Location;
961 begin
962 Prefix_Node := Empty;
964 -- Loop through prefixes
966 loop
967 Designator_Node := Token_Node;
969 if Token = Tok_Identifier then
970 Scan; -- past identifier
971 exit when Token /= Tok_Dot;
973 elsif Token not in Token_Class_Desig then
974 Discard_Junk_Node (P_Identifier); -- to issue the error message
975 raise Error_Resync;
977 else
978 Scan; -- past designator
980 if Token /= Tok_Dot then
981 Error_Msg_SP ("identifier expected");
982 raise Error_Resync;
983 end if;
984 end if;
986 -- Here at a dot, with token just before it in Designator_Node
988 if No (Prefix_Node) then
989 Prefix_Node := Designator_Node;
990 else
991 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
992 Set_Prefix (Selector_Node, Prefix_Node);
993 Set_Selector_Name (Selector_Node, Designator_Node);
994 Prefix_Node := Selector_Node;
995 end if;
997 Dot_Sloc := Token_Ptr;
998 Scan; -- past period
999 end loop;
1001 -- Fall out of the loop having just scanned an identifier
1003 if No (Prefix_Node) then
1004 return Designator_Node;
1005 else
1006 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
1007 Set_Prefix (Selector_Node, Prefix_Node);
1008 Set_Selector_Name (Selector_Node, Designator_Node);
1009 return Selector_Node;
1010 end if;
1011 end P_Qualified_Simple_Name_Resync;
1013 ----------------------
1014 -- 4.1 Direct_Name --
1015 ----------------------
1017 -- Parsed by P_Name and other functions in section 4.1
1019 -----------------
1020 -- 4.1 Prefix --
1021 -----------------
1023 -- Parsed by P_Name (4.1)
1025 -------------------------------
1026 -- 4.1 Explicit Dereference --
1027 -------------------------------
1029 -- Parsed by P_Name (4.1)
1031 -------------------------------
1032 -- 4.1 Implicit_Dereference --
1033 -------------------------------
1035 -- Parsed by P_Name (4.1)
1037 ----------------------------
1038 -- 4.1 Indexed Component --
1039 ----------------------------
1041 -- Parsed by P_Name (4.1)
1043 ----------------
1044 -- 4.1 Slice --
1045 ----------------
1047 -- Parsed by P_Name (4.1)
1049 -----------------------------
1050 -- 4.1 Selected_Component --
1051 -----------------------------
1053 -- Parsed by P_Name (4.1)
1055 ------------------------
1056 -- 4.1 Selector Name --
1057 ------------------------
1059 -- Parsed by P_Name (4.1)
1061 ------------------------------
1062 -- 4.1 Attribute Reference --
1063 ------------------------------
1065 -- Parsed by P_Name (4.1)
1067 -------------------------------
1068 -- 4.1 Attribute Designator --
1069 -------------------------------
1071 -- Parsed by P_Name (4.1)
1073 --------------------------------------
1074 -- 4.1.4 Range Attribute Reference --
1075 --------------------------------------
1077 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
1079 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
1081 -- In the grammar, a RANGE attribute is simply a name, but its use is
1082 -- highly restricted, so in the parser, we do not regard it as a name.
1083 -- Instead, P_Name returns without scanning the 'RANGE part of the
1084 -- attribute, and the caller uses the following function to construct
1085 -- a range attribute in places where it is appropriate.
1087 -- Note that RANGE here is treated essentially as an identifier,
1088 -- rather than a reserved word.
1090 -- The caller has parsed the prefix, i.e. a name, and Token points to
1091 -- the apostrophe. The token after the apostrophe is known to be RANGE
1092 -- at this point. The prefix node becomes the prefix of the attribute.
1094 -- Error_Recovery: Cannot raise Error_Resync
1096 function P_Range_Attribute_Reference
1097 (Prefix_Node : Node_Id)
1098 return Node_Id
1100 Attr_Node : Node_Id;
1102 begin
1103 Attr_Node := New_Node (N_Attribute_Reference, Token_Ptr);
1104 Set_Prefix (Attr_Node, Prefix_Node);
1105 Scan; -- past apostrophe
1107 if Style_Check then
1108 Style.Check_Attribute_Name (True);
1109 end if;
1111 Set_Attribute_Name (Attr_Node, Name_Range);
1112 Scan; -- past RANGE
1114 if Token = Tok_Left_Paren then
1115 Scan; -- past left paren
1116 Set_Expressions (Attr_Node, New_List (P_Expression_If_OK));
1117 T_Right_Paren;
1118 end if;
1120 return Attr_Node;
1121 end P_Range_Attribute_Reference;
1123 ---------------------------------------
1124 -- 4.1.4 Range Attribute Designator --
1125 ---------------------------------------
1127 -- Parsed by P_Range_Attribute_Reference (4.4)
1129 --------------------
1130 -- 4.3 Aggregate --
1131 --------------------
1133 -- AGGREGATE ::= RECORD_AGGREGATE | EXTENSION_AGGREGATE | ARRAY_AGGREGATE
1135 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3), except in the case where
1136 -- an aggregate is known to be required (code statement, extension
1137 -- aggregate), in which cases this routine performs the necessary check
1138 -- that we have an aggregate rather than a parenthesized expression
1140 -- Error recovery: can raise Error_Resync
1142 function P_Aggregate return Node_Id is
1143 Aggr_Sloc : constant Source_Ptr := Token_Ptr;
1144 Aggr_Node : constant Node_Id := P_Aggregate_Or_Paren_Expr;
1146 begin
1147 if Nkind (Aggr_Node) /= N_Aggregate
1148 and then
1149 Nkind (Aggr_Node) /= N_Extension_Aggregate
1150 then
1151 Error_Msg
1152 ("aggregate may not have single positional component", Aggr_Sloc);
1153 return Error;
1154 else
1155 return Aggr_Node;
1156 end if;
1157 end P_Aggregate;
1159 ------------------------------------------------
1160 -- 4.3 Aggregate or Parenthesized Expression --
1161 ------------------------------------------------
1163 -- This procedure parses out either an aggregate or a parenthesized
1164 -- expression (these two constructs are closely related, since a
1165 -- parenthesized expression looks like an aggregate with a single
1166 -- positional component).
1168 -- AGGREGATE ::=
1169 -- RECORD_AGGREGATE | EXTENSION_AGGREGATE | ARRAY_AGGREGATE
1171 -- RECORD_AGGREGATE ::= (RECORD_COMPONENT_ASSOCIATION_LIST)
1173 -- RECORD_COMPONENT_ASSOCIATION_LIST ::=
1174 -- RECORD_COMPONENT_ASSOCIATION {, RECORD_COMPONENT_ASSOCIATION}
1175 -- | null record
1177 -- RECORD_COMPONENT_ASSOCIATION ::=
1178 -- [COMPONENT_CHOICE_LIST =>] EXPRESSION
1180 -- COMPONENT_CHOICE_LIST ::=
1181 -- component_SELECTOR_NAME {| component_SELECTOR_NAME}
1182 -- | others
1184 -- EXTENSION_AGGREGATE ::=
1185 -- (ANCESTOR_PART with RECORD_COMPONENT_ASSOCIATION_LIST)
1187 -- ANCESTOR_PART ::= EXPRESSION | SUBTYPE_MARK
1189 -- ARRAY_AGGREGATE ::=
1190 -- POSITIONAL_ARRAY_AGGREGATE | NAMED_ARRAY_AGGREGATE
1192 -- POSITIONAL_ARRAY_AGGREGATE ::=
1193 -- (EXPRESSION, EXPRESSION {, EXPRESSION})
1194 -- | (EXPRESSION {, EXPRESSION}, others => EXPRESSION)
1195 -- | (EXPRESSION {, EXPRESSION}, others => <>)
1197 -- NAMED_ARRAY_AGGREGATE ::=
1198 -- (ARRAY_COMPONENT_ASSOCIATION {, ARRAY_COMPONENT_ASSOCIATION})
1200 -- PRIMARY ::= (EXPRESSION);
1202 -- Error recovery: can raise Error_Resync
1204 -- Note: POSITIONAL_ARRAY_AGGREGATE rule has been extended to give support
1205 -- to Ada 2005 limited aggregates (AI-287)
1207 function P_Aggregate_Or_Paren_Expr return Node_Id is
1208 Aggregate_Node : Node_Id;
1209 Expr_List : List_Id;
1210 Assoc_List : List_Id;
1211 Expr_Node : Node_Id;
1212 Lparen_Sloc : Source_Ptr;
1213 Scan_State : Saved_Scan_State;
1215 procedure Box_Error;
1216 -- Called if <> is encountered as positional aggregate element. Issues
1217 -- error message and sets Expr_Node to Error.
1219 ---------------
1220 -- Box_Error --
1221 ---------------
1223 procedure Box_Error is
1224 begin
1225 if Ada_Version < Ada_2005 then
1226 Error_Msg_SC ("box in aggregate is an Ada 2005 extension");
1227 end if;
1229 -- Ada 2005 (AI-287): The box notation is allowed only with named
1230 -- notation because positional notation might be error prone. For
1231 -- example, in "(X, <>, Y, <>)", there is no type associated with
1232 -- the boxes, so you might not be leaving out the components you
1233 -- thought you were leaving out.
1235 Error_Msg_SC ("(Ada 2005) box only allowed with named notation");
1236 Scan; -- past box
1237 Expr_Node := Error;
1238 end Box_Error;
1240 -- Start of processing for P_Aggregate_Or_Paren_Expr
1242 begin
1243 Lparen_Sloc := Token_Ptr;
1244 T_Left_Paren;
1246 -- Note on parentheses count. For cases like an if expression, the
1247 -- parens here really count as real parentheses for the paren count,
1248 -- so we adjust the paren count accordingly after scanning the expr.
1250 -- If expression
1252 if Token = Tok_If then
1253 Expr_Node := P_If_Expression;
1254 T_Right_Paren;
1255 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1256 return Expr_Node;
1258 -- Case expression
1260 elsif Token = Tok_Case then
1261 Expr_Node := P_Case_Expression;
1262 T_Right_Paren;
1263 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1264 return Expr_Node;
1266 -- Quantified expression
1268 elsif Token = Tok_For then
1269 Expr_Node := P_Quantified_Expression;
1270 T_Right_Paren;
1271 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1272 return Expr_Node;
1274 -- Note: the mechanism used here of rescanning the initial expression
1275 -- is distinctly unpleasant, but it saves a lot of fiddling in scanning
1276 -- out the discrete choice list.
1278 -- Deal with expression and extension aggregates first
1280 elsif Token /= Tok_Others then
1281 Save_Scan_State (Scan_State); -- at start of expression
1283 -- Deal with (NULL RECORD)
1285 if Token = Tok_Null then
1286 Scan; -- past NULL
1288 if Token = Tok_Record then
1289 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1290 Set_Null_Record_Present (Aggregate_Node, True);
1291 Scan; -- past RECORD
1292 T_Right_Paren;
1293 return Aggregate_Node;
1294 else
1295 Restore_Scan_State (Scan_State); -- to NULL that must be expr
1296 end if;
1297 end if;
1299 -- Scan expression, handling box appearing as positional argument
1301 if Token = Tok_Box then
1302 Box_Error;
1303 else
1304 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
1305 end if;
1307 -- Extension aggregate
1309 if Token = Tok_With then
1310 if Nkind (Expr_Node) = N_Attribute_Reference
1311 and then Attribute_Name (Expr_Node) = Name_Range
1312 then
1313 Bad_Range_Attribute (Sloc (Expr_Node));
1314 return Error;
1315 end if;
1317 if Ada_Version = Ada_83 then
1318 Error_Msg_SC ("(Ada 83) extension aggregate not allowed");
1319 end if;
1321 Aggregate_Node := New_Node (N_Extension_Aggregate, Lparen_Sloc);
1322 Set_Ancestor_Part (Aggregate_Node, Expr_Node);
1323 Scan; -- past WITH
1325 -- Deal with WITH NULL RECORD case
1327 if Token = Tok_Null then
1328 Save_Scan_State (Scan_State); -- at NULL
1329 Scan; -- past NULL
1331 if Token = Tok_Record then
1332 Scan; -- past RECORD
1333 Set_Null_Record_Present (Aggregate_Node, True);
1334 T_Right_Paren;
1335 return Aggregate_Node;
1337 else
1338 Restore_Scan_State (Scan_State); -- to NULL that must be expr
1339 end if;
1340 end if;
1342 if Token /= Tok_Others then
1343 Save_Scan_State (Scan_State);
1344 Expr_Node := P_Expression;
1345 else
1346 Expr_Node := Empty;
1347 end if;
1349 -- Expression
1351 elsif Token = Tok_Right_Paren or else Token in Token_Class_Eterm then
1352 if Nkind (Expr_Node) = N_Attribute_Reference
1353 and then Attribute_Name (Expr_Node) = Name_Range
1354 then
1355 Error_Msg
1356 ("|parentheses not allowed for range attribute", Lparen_Sloc);
1357 Scan; -- past right paren
1358 return Expr_Node;
1359 end if;
1361 -- Bump paren count of expression
1363 if Expr_Node /= Error then
1364 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1365 end if;
1367 T_Right_Paren; -- past right paren (error message if none)
1368 return Expr_Node;
1370 -- Normal aggregate
1372 else
1373 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1374 end if;
1376 -- Others
1378 else
1379 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1380 Expr_Node := Empty;
1381 end if;
1383 -- Prepare to scan list of component associations
1385 Expr_List := No_List; -- don't set yet, maybe all named entries
1386 Assoc_List := No_List; -- don't set yet, maybe all positional entries
1388 -- This loop scans through component associations. On entry to the
1389 -- loop, an expression has been scanned at the start of the current
1390 -- association unless initial token was OTHERS, in which case
1391 -- Expr_Node is set to Empty.
1393 loop
1394 -- Deal with others association first. This is a named association
1396 if No (Expr_Node) then
1397 if No (Assoc_List) then
1398 Assoc_List := New_List;
1399 end if;
1401 Append (P_Record_Or_Array_Component_Association, Assoc_List);
1403 -- Improper use of WITH
1405 elsif Token = Tok_With then
1406 Error_Msg_SC ("WITH must be preceded by single expression in " &
1407 "extension aggregate");
1408 raise Error_Resync;
1410 -- Range attribute can only appear as part of a discrete choice list
1412 elsif Nkind (Expr_Node) = N_Attribute_Reference
1413 and then Attribute_Name (Expr_Node) = Name_Range
1414 and then Token /= Tok_Arrow
1415 and then Token /= Tok_Vertical_Bar
1416 then
1417 Bad_Range_Attribute (Sloc (Expr_Node));
1418 return Error;
1420 -- Assume positional case if comma, right paren, or literal or
1421 -- identifier or OTHERS follows (the latter cases are missing
1422 -- comma cases). Also assume positional if a semicolon follows,
1423 -- which can happen if there are missing parens
1425 elsif Token = Tok_Comma
1426 or else Token = Tok_Right_Paren
1427 or else Token = Tok_Others
1428 or else Token in Token_Class_Lit_Or_Name
1429 or else Token = Tok_Semicolon
1430 then
1431 if Present (Assoc_List) then
1432 Error_Msg_BC -- CODEFIX
1433 ("""='>"" expected (positional association cannot follow " &
1434 "named association)");
1435 end if;
1437 if No (Expr_List) then
1438 Expr_List := New_List;
1439 end if;
1441 Append (Expr_Node, Expr_List);
1443 -- Check for aggregate followed by left parent, maybe missing comma
1445 elsif Nkind (Expr_Node) = N_Aggregate
1446 and then Token = Tok_Left_Paren
1447 then
1448 T_Comma;
1450 if No (Expr_List) then
1451 Expr_List := New_List;
1452 end if;
1454 Append (Expr_Node, Expr_List);
1456 -- Anything else is assumed to be a named association
1458 else
1459 Restore_Scan_State (Scan_State); -- to start of expression
1461 if No (Assoc_List) then
1462 Assoc_List := New_List;
1463 end if;
1465 Append (P_Record_Or_Array_Component_Association, Assoc_List);
1466 end if;
1468 exit when not Comma_Present;
1470 -- If we are at an expression terminator, something is seriously
1471 -- wrong, so let's get out now, before we start eating up stuff
1472 -- that doesn't belong to us!
1474 if Token in Token_Class_Eterm then
1475 Error_Msg_AP
1476 ("expecting expression or component association");
1477 exit;
1478 end if;
1480 -- Deal with misused box
1482 if Token = Tok_Box then
1483 Box_Error;
1485 -- Otherwise initiate for reentry to top of loop by scanning an
1486 -- initial expression, unless the first token is OTHERS.
1488 elsif Token = Tok_Others then
1489 Expr_Node := Empty;
1491 else
1492 Save_Scan_State (Scan_State); -- at start of expression
1493 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
1495 end if;
1496 end loop;
1498 -- All component associations (positional and named) have been scanned
1500 T_Right_Paren;
1501 Set_Expressions (Aggregate_Node, Expr_List);
1502 Set_Component_Associations (Aggregate_Node, Assoc_List);
1503 return Aggregate_Node;
1504 end P_Aggregate_Or_Paren_Expr;
1506 ------------------------------------------------
1507 -- 4.3 Record or Array Component Association --
1508 ------------------------------------------------
1510 -- RECORD_COMPONENT_ASSOCIATION ::=
1511 -- [COMPONENT_CHOICE_LIST =>] EXPRESSION
1512 -- | COMPONENT_CHOICE_LIST => <>
1514 -- COMPONENT_CHOICE_LIST =>
1515 -- component_SELECTOR_NAME {| component_SELECTOR_NAME}
1516 -- | others
1518 -- ARRAY_COMPONENT_ASSOCIATION ::=
1519 -- DISCRETE_CHOICE_LIST => EXPRESSION
1520 -- | DISCRETE_CHOICE_LIST => <>
1522 -- Note: this routine only handles the named cases, including others.
1523 -- Cases where the component choice list is not present have already
1524 -- been handled directly.
1526 -- Error recovery: can raise Error_Resync
1528 -- Note: RECORD_COMPONENT_ASSOCIATION and ARRAY_COMPONENT_ASSOCIATION
1529 -- rules have been extended to give support to Ada 2005 limited
1530 -- aggregates (AI-287)
1532 function P_Record_Or_Array_Component_Association return Node_Id is
1533 Assoc_Node : Node_Id;
1535 begin
1536 Assoc_Node := New_Node (N_Component_Association, Token_Ptr);
1537 Set_Choices (Assoc_Node, P_Discrete_Choice_List);
1538 Set_Sloc (Assoc_Node, Token_Ptr);
1539 TF_Arrow;
1541 if Token = Tok_Box then
1543 -- Ada 2005(AI-287): The box notation is used to indicate the
1544 -- default initialization of aggregate components
1546 if Ada_Version < Ada_2005 then
1547 Error_Msg_SP
1548 ("component association with '<'> is an Ada 2005 extension");
1549 Error_Msg_SP ("\unit must be compiled with -gnat05 switch");
1550 end if;
1552 Set_Box_Present (Assoc_Node);
1553 Scan; -- Past box
1554 else
1555 Set_Expression (Assoc_Node, P_Expression);
1556 end if;
1558 return Assoc_Node;
1559 end P_Record_Or_Array_Component_Association;
1561 -----------------------------
1562 -- 4.3.1 Record Aggregate --
1563 -----------------------------
1565 -- Case of enumeration aggregate is parsed by P_Aggregate (4.3)
1566 -- All other cases are parsed by P_Aggregate_Or_Paren_Expr (4.3)
1568 ----------------------------------------------
1569 -- 4.3.1 Record Component Association List --
1570 ----------------------------------------------
1572 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1574 ----------------------------------
1575 -- 4.3.1 Component Choice List --
1576 ----------------------------------
1578 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1580 --------------------------------
1581 -- 4.3.1 Extension Aggregate --
1582 --------------------------------
1584 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1586 --------------------------
1587 -- 4.3.1 Ancestor Part --
1588 --------------------------
1590 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1592 ----------------------------
1593 -- 4.3.1 Array Aggregate --
1594 ----------------------------
1596 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1598 ---------------------------------------
1599 -- 4.3.1 Positional Array Aggregate --
1600 ---------------------------------------
1602 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1604 ----------------------------------
1605 -- 4.3.1 Named Array Aggregate --
1606 ----------------------------------
1608 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1610 ----------------------------------------
1611 -- 4.3.1 Array Component Association --
1612 ----------------------------------------
1614 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1616 ---------------------
1617 -- 4.4 Expression --
1618 ---------------------
1620 -- This procedure parses EXPRESSION or CHOICE_EXPRESSION
1622 -- EXPRESSION ::=
1623 -- RELATION {LOGICAL_OPERATOR RELATION}
1625 -- CHOICE_EXPRESSION ::=
1626 -- CHOICE_RELATION {LOGICAL_OPERATOR CHOICE_RELATION}
1628 -- LOGICAL_OPERATOR ::= and | and then | or | or else | xor
1630 -- On return, Expr_Form indicates the categorization of the expression
1631 -- EF_Range_Attr is not a possible value (if a range attribute is found,
1632 -- an error message is given, and Error is returned).
1634 -- Error recovery: cannot raise Error_Resync
1636 function P_Expression return Node_Id is
1637 Logical_Op : Node_Kind;
1638 Prev_Logical_Op : Node_Kind;
1639 Op_Location : Source_Ptr;
1640 Node1 : Node_Id;
1641 Node2 : Node_Id;
1643 begin
1644 Node1 := P_Relation;
1646 if Token in Token_Class_Logop then
1647 Prev_Logical_Op := N_Empty;
1649 loop
1650 Op_Location := Token_Ptr;
1651 Logical_Op := P_Logical_Operator;
1653 if Prev_Logical_Op /= N_Empty and then
1654 Logical_Op /= Prev_Logical_Op
1655 then
1656 Error_Msg
1657 ("mixed logical operators in expression", Op_Location);
1658 Prev_Logical_Op := N_Empty;
1659 else
1660 Prev_Logical_Op := Logical_Op;
1661 end if;
1663 Node2 := Node1;
1664 Node1 := New_Op_Node (Logical_Op, Op_Location);
1665 Set_Left_Opnd (Node1, Node2);
1666 Set_Right_Opnd (Node1, P_Relation);
1667 exit when Token not in Token_Class_Logop;
1668 end loop;
1670 Expr_Form := EF_Non_Simple;
1671 end if;
1673 if Token = Tok_Apostrophe then
1674 Bad_Range_Attribute (Token_Ptr);
1675 return Error;
1676 else
1677 return Node1;
1678 end if;
1679 end P_Expression;
1681 -- This function is identical to the normal P_Expression, except that it
1682 -- also permits the appearance of a case, conditional, or quantified
1683 -- expression if the call immediately follows a left paren, and followed
1684 -- by a right parenthesis. These forms are allowed if these conditions
1685 -- are not met, but an error message will be issued.
1687 function P_Expression_If_OK return Node_Id is
1688 begin
1689 -- Case of conditional, case or quantified expression
1691 if Token = Tok_Case or else Token = Tok_If or else Token = Tok_For then
1692 return P_Unparen_Cond_Case_Quant_Expression;
1694 -- Normal case, not case/conditional/quantified expression
1696 else
1697 return P_Expression;
1698 end if;
1699 end P_Expression_If_OK;
1701 -- This function is identical to the normal P_Expression, except that it
1702 -- checks that the expression scan did not stop on a right paren. It is
1703 -- called in all contexts where a right parenthesis cannot legitimately
1704 -- follow an expression.
1706 -- Error recovery: can not raise Error_Resync
1708 function P_Expression_No_Right_Paren return Node_Id is
1709 Expr : constant Node_Id := P_Expression;
1710 begin
1711 Ignore (Tok_Right_Paren);
1712 return Expr;
1713 end P_Expression_No_Right_Paren;
1715 ----------------------------------------
1716 -- 4.4 Expression_Or_Range_Attribute --
1717 ----------------------------------------
1719 -- EXPRESSION ::=
1720 -- RELATION {and RELATION} | RELATION {and then RELATION}
1721 -- | RELATION {or RELATION} | RELATION {or else RELATION}
1722 -- | RELATION {xor RELATION}
1724 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
1726 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
1728 -- On return, Expr_Form indicates the categorization of the expression
1729 -- and EF_Range_Attr is one of the possibilities.
1731 -- Error recovery: cannot raise Error_Resync
1733 -- In the grammar, a RANGE attribute is simply a name, but its use is
1734 -- highly restricted, so in the parser, we do not regard it as a name.
1735 -- Instead, P_Name returns without scanning the 'RANGE part of the
1736 -- attribute, and P_Expression_Or_Range_Attribute handles the range
1737 -- attribute reference. In the normal case where a range attribute is
1738 -- not allowed, an error message is issued by P_Expression.
1740 function P_Expression_Or_Range_Attribute return Node_Id is
1741 Logical_Op : Node_Kind;
1742 Prev_Logical_Op : Node_Kind;
1743 Op_Location : Source_Ptr;
1744 Node1 : Node_Id;
1745 Node2 : Node_Id;
1746 Attr_Node : Node_Id;
1748 begin
1749 Node1 := P_Relation;
1751 if Token = Tok_Apostrophe then
1752 Attr_Node := P_Range_Attribute_Reference (Node1);
1753 Expr_Form := EF_Range_Attr;
1754 return Attr_Node;
1756 elsif Token in Token_Class_Logop then
1757 Prev_Logical_Op := N_Empty;
1759 loop
1760 Op_Location := Token_Ptr;
1761 Logical_Op := P_Logical_Operator;
1763 if Prev_Logical_Op /= N_Empty and then
1764 Logical_Op /= Prev_Logical_Op
1765 then
1766 Error_Msg
1767 ("mixed logical operators in expression", Op_Location);
1768 Prev_Logical_Op := N_Empty;
1769 else
1770 Prev_Logical_Op := Logical_Op;
1771 end if;
1773 Node2 := Node1;
1774 Node1 := New_Op_Node (Logical_Op, Op_Location);
1775 Set_Left_Opnd (Node1, Node2);
1776 Set_Right_Opnd (Node1, P_Relation);
1777 exit when Token not in Token_Class_Logop;
1778 end loop;
1780 Expr_Form := EF_Non_Simple;
1781 end if;
1783 if Token = Tok_Apostrophe then
1784 Bad_Range_Attribute (Token_Ptr);
1785 return Error;
1786 else
1787 return Node1;
1788 end if;
1789 end P_Expression_Or_Range_Attribute;
1791 -- Version that allows a non-parenthesized case, conditional, or quantified
1792 -- expression if the call immediately follows a left paren, and followed
1793 -- by a right parenthesis. These forms are allowed if these conditions
1794 -- are not met, but an error message will be issued.
1796 function P_Expression_Or_Range_Attribute_If_OK return Node_Id is
1797 begin
1798 -- Case of conditional, case or quantified expression
1800 if Token = Tok_Case or else Token = Tok_If or else Token = Tok_For then
1801 return P_Unparen_Cond_Case_Quant_Expression;
1803 -- Normal case, not one of the above expression types
1805 else
1806 return P_Expression_Or_Range_Attribute;
1807 end if;
1808 end P_Expression_Or_Range_Attribute_If_OK;
1810 -------------------
1811 -- 4.4 Relation --
1812 -------------------
1814 -- This procedure scans both relations and choice relations
1816 -- CHOICE_RELATION ::=
1817 -- SIMPLE_EXPRESSION [RELATIONAL_OPERATOR SIMPLE_EXPRESSION]
1819 -- RELATION ::=
1820 -- SIMPLE_EXPRESSION [not] in MEMBERSHIP_CHOICE_LIST
1822 -- MEMBERSHIP_CHOICE_LIST ::=
1823 -- MEMBERSHIP_CHOICE {'|' MEMBERSHIP CHOICE}
1825 -- MEMBERSHIP_CHOICE ::=
1826 -- CHOICE_EXPRESSION | RANGE | SUBTYPE_MARK
1828 -- On return, Expr_Form indicates the categorization of the expression
1830 -- Note: if Token = Tok_Apostrophe on return, then Expr_Form is set to
1831 -- EF_Simple_Name and the following token is RANGE (range attribute case).
1833 -- Error recovery: cannot raise Error_Resync. If an error occurs within an
1834 -- expression, then tokens are scanned until either a non-expression token,
1835 -- a right paren (not matched by a left paren) or a comma, is encountered.
1837 function P_Relation return Node_Id is
1838 Node1, Node2 : Node_Id;
1839 Optok : Source_Ptr;
1841 begin
1842 Node1 := P_Simple_Expression;
1844 if Token not in Token_Class_Relop then
1845 return Node1;
1847 else
1848 -- Here we have a relational operator following. If so then scan it
1849 -- out. Note that the assignment symbol := is treated as a relational
1850 -- operator to improve the error recovery when it is misused for =.
1851 -- P_Relational_Operator also parses the IN and NOT IN operations.
1853 Optok := Token_Ptr;
1854 Node2 := New_Op_Node (P_Relational_Operator, Optok);
1855 Set_Left_Opnd (Node2, Node1);
1857 -- Case of IN or NOT IN
1859 if Prev_Token = Tok_In then
1860 P_Membership_Test (Node2);
1862 -- Case of relational operator (= /= < <= > >=)
1864 else
1865 Set_Right_Opnd (Node2, P_Simple_Expression);
1866 end if;
1868 Expr_Form := EF_Non_Simple;
1870 if Token in Token_Class_Relop then
1871 Error_Msg_SC ("unexpected relational operator");
1872 raise Error_Resync;
1873 end if;
1875 return Node2;
1876 end if;
1878 -- If any error occurs, then scan to the next expression terminator symbol
1879 -- or comma or right paren at the outer (i.e. current) parentheses level.
1880 -- The flags are set to indicate a normal simple expression.
1882 exception
1883 when Error_Resync =>
1884 Resync_Expression;
1885 Expr_Form := EF_Simple;
1886 return Error;
1887 end P_Relation;
1889 ----------------------------
1890 -- 4.4 Simple Expression --
1891 ----------------------------
1893 -- SIMPLE_EXPRESSION ::=
1894 -- [UNARY_ADDING_OPERATOR] TERM {BINARY_ADDING_OPERATOR TERM}
1896 -- On return, Expr_Form indicates the categorization of the expression
1898 -- Note: if Token = Tok_Apostrophe on return, then Expr_Form is set to
1899 -- EF_Simple_Name and the following token is RANGE (range attribute case).
1901 -- Error recovery: cannot raise Error_Resync. If an error occurs within an
1902 -- expression, then tokens are scanned until either a non-expression token,
1903 -- a right paren (not matched by a left paren) or a comma, is encountered.
1905 -- Note: P_Simple_Expression is called only internally by higher level
1906 -- expression routines. In cases in the grammar where a simple expression
1907 -- is required, the approach is to scan an expression, and then post an
1908 -- appropriate error message if the expression obtained is not simple. This
1909 -- gives better error recovery and treatment.
1911 function P_Simple_Expression return Node_Id is
1912 Scan_State : Saved_Scan_State;
1913 Node1 : Node_Id;
1914 Node2 : Node_Id;
1915 Tokptr : Source_Ptr;
1917 begin
1918 -- Check for cases starting with a name. There are two reasons for
1919 -- special casing. First speed things up by catching a common case
1920 -- without going through several routine layers. Second the caller must
1921 -- be informed via Expr_Form when the simple expression is a name.
1923 if Token in Token_Class_Name then
1924 Node1 := P_Name;
1926 -- Deal with apostrophe cases
1928 if Token = Tok_Apostrophe then
1929 Save_Scan_State (Scan_State); -- at apostrophe
1930 Scan; -- past apostrophe
1932 -- If qualified expression, scan it out and fall through
1934 if Token = Tok_Left_Paren then
1935 Node1 := P_Qualified_Expression (Node1);
1936 Expr_Form := EF_Simple;
1938 -- If range attribute, then we return with Token pointing to the
1939 -- apostrophe. Note: avoid the normal error check on exit. We
1940 -- know that the expression really is complete in this case!
1942 else -- Token = Tok_Range then
1943 Restore_Scan_State (Scan_State); -- to apostrophe
1944 Expr_Form := EF_Simple_Name;
1945 return Node1;
1946 end if;
1947 end if;
1949 -- If an expression terminator follows, the previous processing
1950 -- completely scanned out the expression (a common case), and
1951 -- left Expr_Form set appropriately for returning to our caller.
1953 if Token in Token_Class_Sterm then
1954 null;
1956 -- If we do not have an expression terminator, then complete the
1957 -- scan of a simple expression. This code duplicates the code
1958 -- found in P_Term and P_Factor.
1960 else
1961 if Token = Tok_Double_Asterisk then
1962 if Style_Check then
1963 Style.Check_Exponentiation_Operator;
1964 end if;
1966 Node2 := New_Op_Node (N_Op_Expon, Token_Ptr);
1967 Scan; -- past **
1968 Set_Left_Opnd (Node2, Node1);
1969 Set_Right_Opnd (Node2, P_Primary);
1970 Check_Bad_Exp;
1971 Node1 := Node2;
1972 end if;
1974 loop
1975 exit when Token not in Token_Class_Mulop;
1976 Tokptr := Token_Ptr;
1977 Node2 := New_Op_Node (P_Multiplying_Operator, Tokptr);
1979 if Style_Check then
1980 Style.Check_Binary_Operator;
1981 end if;
1983 Scan; -- past operator
1984 Set_Left_Opnd (Node2, Node1);
1985 Set_Right_Opnd (Node2, P_Factor);
1986 Node1 := Node2;
1987 end loop;
1989 loop
1990 exit when Token not in Token_Class_Binary_Addop;
1991 Tokptr := Token_Ptr;
1992 Node2 := New_Op_Node (P_Binary_Adding_Operator, Tokptr);
1994 if Style_Check then
1995 Style.Check_Binary_Operator;
1996 end if;
1998 Scan; -- past operator
1999 Set_Left_Opnd (Node2, Node1);
2000 Set_Right_Opnd (Node2, P_Term);
2001 Node1 := Node2;
2002 end loop;
2004 Expr_Form := EF_Simple;
2005 end if;
2007 -- Cases where simple expression does not start with a name
2009 else
2010 -- Scan initial sign and initial Term
2012 if Token in Token_Class_Unary_Addop then
2013 Tokptr := Token_Ptr;
2014 Node1 := New_Op_Node (P_Unary_Adding_Operator, Tokptr);
2016 if Style_Check then
2017 Style.Check_Unary_Plus_Or_Minus;
2018 end if;
2020 Scan; -- past operator
2021 Set_Right_Opnd (Node1, P_Term);
2022 else
2023 Node1 := P_Term;
2024 end if;
2026 -- In the following, we special-case a sequence of concatenations of
2027 -- string literals, such as "aaa" & "bbb" & ... & "ccc", with nothing
2028 -- else mixed in. For such a sequence, we return a tree representing
2029 -- "" & "aaabbb...ccc" (a single concatenation). This is done only if
2030 -- the number of concatenations is large. If semantic analysis
2031 -- resolves the "&" to a predefined one, then this folding gives the
2032 -- right answer. Otherwise, semantic analysis will complain about a
2033 -- capacity-exceeded error. The purpose of this trick is to avoid
2034 -- creating a deeply nested tree, which would cause deep recursion
2035 -- during semantics, causing stack overflow. This way, we can handle
2036 -- enormous concatenations in the normal case of predefined "&". We
2037 -- first build up the normal tree, and then rewrite it if
2038 -- appropriate.
2040 declare
2041 Num_Concats_Threshold : constant Positive := 1000;
2042 -- Arbitrary threshold value to enable optimization
2044 First_Node : constant Node_Id := Node1;
2045 Is_Strlit_Concat : Boolean;
2046 -- True iff we've parsed a sequence of concatenations of string
2047 -- literals, with nothing else mixed in.
2049 Num_Concats : Natural;
2050 -- Number of "&" operators if Is_Strlit_Concat is True
2052 begin
2053 Is_Strlit_Concat :=
2054 Nkind (Node1) = N_String_Literal
2055 and then Token = Tok_Ampersand;
2056 Num_Concats := 0;
2058 -- Scan out sequence of terms separated by binary adding operators
2060 loop
2061 exit when Token not in Token_Class_Binary_Addop;
2062 Tokptr := Token_Ptr;
2063 Node2 := New_Op_Node (P_Binary_Adding_Operator, Tokptr);
2064 Scan; -- past operator
2065 Set_Left_Opnd (Node2, Node1);
2066 Node1 := P_Term;
2067 Set_Right_Opnd (Node2, Node1);
2069 -- Check if we're still concatenating string literals
2071 Is_Strlit_Concat :=
2072 Is_Strlit_Concat
2073 and then Nkind (Node2) = N_Op_Concat
2074 and then Nkind (Node1) = N_String_Literal;
2076 if Is_Strlit_Concat then
2077 Num_Concats := Num_Concats + 1;
2078 end if;
2080 Node1 := Node2;
2081 end loop;
2083 -- If we have an enormous series of concatenations of string
2084 -- literals, rewrite as explained above. The Is_Folded_In_Parser
2085 -- flag tells semantic analysis that if the "&" is not predefined,
2086 -- the folded value is wrong.
2088 if Is_Strlit_Concat
2089 and then Num_Concats >= Num_Concats_Threshold
2090 then
2091 declare
2092 Empty_String_Val : String_Id;
2093 -- String_Id for ""
2095 Strlit_Concat_Val : String_Id;
2096 -- Contains the folded value (which will be correct if the
2097 -- "&" operators are the predefined ones).
2099 Cur_Node : Node_Id;
2100 -- For walking up the tree
2102 New_Node : Node_Id;
2103 -- Folded node to replace Node1
2105 Loc : constant Source_Ptr := Sloc (First_Node);
2107 begin
2108 -- Walk up the tree starting at the leftmost string literal
2109 -- (First_Node), building up the Strlit_Concat_Val as we
2110 -- go. Note that we do not use recursion here -- the whole
2111 -- point is to avoid recursively walking that enormous tree.
2113 Start_String;
2114 Store_String_Chars (Strval (First_Node));
2116 Cur_Node := Parent (First_Node);
2117 while Present (Cur_Node) loop
2118 pragma Assert (Nkind (Cur_Node) = N_Op_Concat and then
2119 Nkind (Right_Opnd (Cur_Node)) = N_String_Literal);
2121 Store_String_Chars (Strval (Right_Opnd (Cur_Node)));
2122 Cur_Node := Parent (Cur_Node);
2123 end loop;
2125 Strlit_Concat_Val := End_String;
2127 -- Create new folded node, and rewrite result with a concat-
2128 -- enation of an empty string literal and the folded node.
2130 Start_String;
2131 Empty_String_Val := End_String;
2132 New_Node :=
2133 Make_Op_Concat (Loc,
2134 Make_String_Literal (Loc, Empty_String_Val),
2135 Make_String_Literal (Loc, Strlit_Concat_Val,
2136 Is_Folded_In_Parser => True));
2137 Rewrite (Node1, New_Node);
2138 end;
2139 end if;
2140 end;
2142 -- All done, we clearly do not have name or numeric literal so this
2143 -- is a case of a simple expression which is some other possibility.
2145 Expr_Form := EF_Simple;
2146 end if;
2148 -- Come here at end of simple expression, where we do a couple of
2149 -- special checks to improve error recovery.
2151 -- Special test to improve error recovery. If the current token
2152 -- is a period, then someone is trying to do selection on something
2153 -- that is not a name, e.g. a qualified expression.
2155 if Token = Tok_Dot then
2156 Error_Msg_SC ("prefix for selection is not a name");
2158 -- If qualified expression, comment and continue, otherwise something
2159 -- is pretty nasty so do an Error_Resync call.
2161 if Ada_Version < Ada_2012
2162 and then Nkind (Node1) = N_Qualified_Expression
2163 then
2164 Error_Msg_SC ("\would be legal in Ada 2012 mode");
2165 else
2166 raise Error_Resync;
2167 end if;
2168 end if;
2170 -- Special test to improve error recovery: If the current token is
2171 -- not the first token on a line (as determined by checking the
2172 -- previous token position with the start of the current line),
2173 -- then we insist that we have an appropriate terminating token.
2174 -- Consider the following two examples:
2176 -- 1) if A nad B then ...
2178 -- 2) A := B
2179 -- C := D
2181 -- In the first example, we would like to issue a binary operator
2182 -- expected message and resynchronize to the then. In the second
2183 -- example, we do not want to issue a binary operator message, so
2184 -- that instead we will get the missing semicolon message. This
2185 -- distinction is of course a heuristic which does not always work,
2186 -- but in practice it is quite effective.
2188 -- Note: the one case in which we do not go through this circuit is
2189 -- when we have scanned a range attribute and want to return with
2190 -- Token pointing to the apostrophe. The apostrophe is not normally
2191 -- an expression terminator, and is not in Token_Class_Sterm, but
2192 -- in this special case we know that the expression is complete.
2194 if not Token_Is_At_Start_Of_Line
2195 and then Token not in Token_Class_Sterm
2196 then
2197 -- Normally the right error message is indeed that we expected a
2198 -- binary operator, but in the case of being between a right and left
2199 -- paren, e.g. in an aggregate, a more likely error is missing comma.
2201 if Prev_Token = Tok_Right_Paren and then Token = Tok_Left_Paren then
2202 T_Comma;
2203 else
2204 Error_Msg_AP ("binary operator expected");
2205 end if;
2207 raise Error_Resync;
2209 else
2210 return Node1;
2211 end if;
2213 -- If any error occurs, then scan to next expression terminator symbol
2214 -- or comma, right paren or vertical bar at the outer (i.e. current) paren
2215 -- level. Expr_Form is set to indicate a normal simple expression.
2217 exception
2218 when Error_Resync =>
2219 Resync_Expression;
2220 Expr_Form := EF_Simple;
2221 return Error;
2222 end P_Simple_Expression;
2224 -----------------------------------------------
2225 -- 4.4 Simple Expression or Range Attribute --
2226 -----------------------------------------------
2228 -- SIMPLE_EXPRESSION ::=
2229 -- [UNARY_ADDING_OPERATOR] TERM {BINARY_ADDING_OPERATOR TERM}
2231 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
2233 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
2235 -- Error recovery: cannot raise Error_Resync
2237 function P_Simple_Expression_Or_Range_Attribute return Node_Id is
2238 Sexpr : Node_Id;
2239 Attr_Node : Node_Id;
2241 begin
2242 -- We don't just want to roar ahead and call P_Simple_Expression
2243 -- here, since we want to handle the case of a parenthesized range
2244 -- attribute cleanly.
2246 if Token = Tok_Left_Paren then
2247 declare
2248 Lptr : constant Source_Ptr := Token_Ptr;
2249 Scan_State : Saved_Scan_State;
2251 begin
2252 Save_Scan_State (Scan_State);
2253 Scan; -- past left paren
2254 Sexpr := P_Simple_Expression;
2256 if Token = Tok_Apostrophe then
2257 Attr_Node := P_Range_Attribute_Reference (Sexpr);
2258 Expr_Form := EF_Range_Attr;
2260 if Token = Tok_Right_Paren then
2261 Scan; -- scan past right paren if present
2262 end if;
2264 Error_Msg ("parentheses not allowed for range attribute", Lptr);
2266 return Attr_Node;
2267 end if;
2269 Restore_Scan_State (Scan_State);
2270 end;
2271 end if;
2273 -- Here after dealing with parenthesized range attribute
2275 Sexpr := P_Simple_Expression;
2277 if Token = Tok_Apostrophe then
2278 Attr_Node := P_Range_Attribute_Reference (Sexpr);
2279 Expr_Form := EF_Range_Attr;
2280 return Attr_Node;
2282 else
2283 return Sexpr;
2284 end if;
2285 end P_Simple_Expression_Or_Range_Attribute;
2287 ---------------
2288 -- 4.4 Term --
2289 ---------------
2291 -- TERM ::= FACTOR {MULTIPLYING_OPERATOR FACTOR}
2293 -- Error recovery: can raise Error_Resync
2295 function P_Term return Node_Id is
2296 Node1, Node2 : Node_Id;
2297 Tokptr : Source_Ptr;
2299 begin
2300 Node1 := P_Factor;
2302 loop
2303 exit when Token not in Token_Class_Mulop;
2304 Tokptr := Token_Ptr;
2305 Node2 := New_Op_Node (P_Multiplying_Operator, Tokptr);
2306 Scan; -- past operator
2307 Set_Left_Opnd (Node2, Node1);
2308 Set_Right_Opnd (Node2, P_Factor);
2309 Node1 := Node2;
2310 end loop;
2312 return Node1;
2313 end P_Term;
2315 -----------------
2316 -- 4.4 Factor --
2317 -----------------
2319 -- FACTOR ::= PRIMARY [** PRIMARY] | abs PRIMARY | not PRIMARY
2321 -- Error recovery: can raise Error_Resync
2323 function P_Factor return Node_Id is
2324 Node1 : Node_Id;
2325 Node2 : Node_Id;
2327 begin
2328 if Token = Tok_Abs then
2329 Node1 := New_Op_Node (N_Op_Abs, Token_Ptr);
2331 if Style_Check then
2332 Style.Check_Abs_Not;
2333 end if;
2335 Scan; -- past ABS
2336 Set_Right_Opnd (Node1, P_Primary);
2337 return Node1;
2339 elsif Token = Tok_Not then
2340 Node1 := New_Op_Node (N_Op_Not, Token_Ptr);
2342 if Style_Check then
2343 Style.Check_Abs_Not;
2344 end if;
2346 Scan; -- past NOT
2347 Set_Right_Opnd (Node1, P_Primary);
2348 return Node1;
2350 else
2351 Node1 := P_Primary;
2353 if Token = Tok_Double_Asterisk then
2354 Node2 := New_Op_Node (N_Op_Expon, Token_Ptr);
2355 Scan; -- past **
2356 Set_Left_Opnd (Node2, Node1);
2357 Set_Right_Opnd (Node2, P_Primary);
2358 Check_Bad_Exp;
2359 return Node2;
2360 else
2361 return Node1;
2362 end if;
2363 end if;
2364 end P_Factor;
2366 ------------------
2367 -- 4.4 Primary --
2368 ------------------
2370 -- PRIMARY ::=
2371 -- NUMERIC_LITERAL | null
2372 -- | STRING_LITERAL | AGGREGATE
2373 -- | NAME | QUALIFIED_EXPRESSION
2374 -- | ALLOCATOR | (EXPRESSION) | QUANTIFIED_EXPRESSION
2376 -- Error recovery: can raise Error_Resync
2378 function P_Primary return Node_Id is
2379 Scan_State : Saved_Scan_State;
2380 Node1 : Node_Id;
2382 Lparen : constant Boolean := Prev_Token = Tok_Left_Paren;
2383 -- Remember if previous token is a left parenthesis. This is used to
2384 -- deal with checking whether IF/CASE/FOR expressions appearing as
2385 -- primaries require extra parenthesization.
2387 begin
2388 -- The loop runs more than once only if misplaced pragmas are found
2389 -- or if a misplaced unary minus is skipped.
2391 loop
2392 case Token is
2394 -- Name token can start a name, call or qualified expression, all
2395 -- of which are acceptable possibilities for primary. Note also
2396 -- that string literal is included in name (as operator symbol)
2397 -- and type conversion is included in name (as indexed component).
2399 when Tok_Char_Literal | Tok_Operator_Symbol | Tok_Identifier =>
2400 Node1 := P_Name;
2402 -- All done unless apostrophe follows
2404 if Token /= Tok_Apostrophe then
2405 return Node1;
2407 -- Apostrophe following means that we have either just parsed
2408 -- the subtype mark of a qualified expression, or the prefix
2409 -- or a range attribute.
2411 else -- Token = Tok_Apostrophe
2412 Save_Scan_State (Scan_State); -- at apostrophe
2413 Scan; -- past apostrophe
2415 -- If range attribute, then this is always an error, since
2416 -- the only legitimate case (where the scanned expression is
2417 -- a qualified simple name) is handled at the level of the
2418 -- Simple_Expression processing. This case corresponds to a
2419 -- usage such as 3 + A'Range, which is always illegal.
2421 if Token = Tok_Range then
2422 Restore_Scan_State (Scan_State); -- to apostrophe
2423 Bad_Range_Attribute (Token_Ptr);
2424 return Error;
2426 -- If left paren, then we have a qualified expression.
2427 -- Note that P_Name guarantees that in this case, where
2428 -- Token = Tok_Apostrophe on return, the only two possible
2429 -- tokens following the apostrophe are left paren and
2430 -- RANGE, so we know we have a left paren here.
2432 else -- Token = Tok_Left_Paren
2433 return P_Qualified_Expression (Node1);
2435 end if;
2436 end if;
2438 -- Numeric or string literal
2440 when Tok_Integer_Literal |
2441 Tok_Real_Literal |
2442 Tok_String_Literal =>
2444 Node1 := Token_Node;
2445 Scan; -- past number
2446 return Node1;
2448 -- Left paren, starts aggregate or parenthesized expression
2450 when Tok_Left_Paren =>
2451 declare
2452 Expr : constant Node_Id := P_Aggregate_Or_Paren_Expr;
2454 begin
2455 if Nkind (Expr) = N_Attribute_Reference
2456 and then Attribute_Name (Expr) = Name_Range
2457 then
2458 Bad_Range_Attribute (Sloc (Expr));
2459 end if;
2461 return Expr;
2462 end;
2464 -- Allocator
2466 when Tok_New =>
2467 return P_Allocator;
2469 -- Null
2471 when Tok_Null =>
2472 Scan; -- past NULL
2473 return New_Node (N_Null, Prev_Token_Ptr);
2475 -- Pragma, not allowed here, so just skip past it
2477 when Tok_Pragma =>
2478 P_Pragmas_Misplaced;
2480 -- Deal with IF (possible unparenthesized if expression)
2482 when Tok_If =>
2484 -- If this looks like a real if, defined as an IF appearing at
2485 -- the start of a new line, then we consider we have a missing
2486 -- operand. If in Ada 2012 and the IF is not properly indented
2487 -- for a statement, we prefer to issue a message about an ill-
2488 -- parenthesized if expression.
2490 if Token_Is_At_Start_Of_Line
2491 and then not
2492 (Ada_Version >= Ada_2012
2493 and then Style_Check_Indentation /= 0
2494 and then Start_Column rem Style_Check_Indentation /= 0)
2495 then
2496 Error_Msg_AP ("missing operand");
2497 return Error;
2499 -- If this looks like an if expression, then treat it that way
2500 -- with an error message if not explicitly surrounded by
2501 -- parentheses.
2503 elsif Ada_Version >= Ada_2012 then
2504 Node1 := P_If_Expression;
2506 if not (Lparen and then Token = Tok_Right_Paren) then
2507 Error_Msg
2508 ("if expression must be parenthesized", Sloc (Node1));
2509 end if;
2511 return Node1;
2513 -- Otherwise treat as misused identifier
2515 else
2516 return P_Identifier;
2517 end if;
2519 -- Deal with CASE (possible unparenthesized case expression)
2521 when Tok_Case =>
2523 -- If this looks like a real case, defined as a CASE appearing
2524 -- the start of a new line, then we consider we have a missing
2525 -- operand. If in Ada 2012 and the CASE is not properly
2526 -- indented for a statement, we prefer to issue a message about
2527 -- an ill-parenthesized case expression.
2529 if Token_Is_At_Start_Of_Line
2530 and then not
2531 (Ada_Version >= Ada_2012
2532 and then Style_Check_Indentation /= 0
2533 and then Start_Column rem Style_Check_Indentation /= 0)
2534 then
2535 Error_Msg_AP ("missing operand");
2536 return Error;
2538 -- If this looks like a case expression, then treat it that way
2539 -- with an error message if not within parentheses.
2541 elsif Ada_Version >= Ada_2012 then
2542 Node1 := P_Case_Expression;
2544 if not (Lparen and then Token = Tok_Right_Paren) then
2545 Error_Msg
2546 ("case expression must be parenthesized", Sloc (Node1));
2547 end if;
2549 return Node1;
2551 -- Otherwise treat as misused identifier
2553 else
2554 return P_Identifier;
2555 end if;
2557 -- For [all | some] indicates a quantified expression
2559 when Tok_For =>
2560 if Token_Is_At_Start_Of_Line then
2561 Error_Msg_AP ("misplaced loop");
2562 return Error;
2564 elsif Ada_Version >= Ada_2012 then
2565 Node1 := P_Quantified_Expression;
2567 if not (Lparen and then Token = Tok_Right_Paren) then
2568 Error_Msg
2569 ("quantified expression must be parenthesized",
2570 Sloc (Node1));
2571 end if;
2573 return Node1;
2575 -- Otherwise treat as misused identifier
2577 else
2578 return P_Identifier;
2579 end if;
2581 -- Minus may well be an improper attempt at a unary minus. Give
2582 -- a message, skip the minus and keep going!
2584 when Tok_Minus =>
2585 Error_Msg_SC ("parentheses required for unary minus");
2586 Scan; -- past minus
2588 -- Anything else is illegal as the first token of a primary, but
2589 -- we test for some common errors, to improve error messages.
2591 when others =>
2592 if Is_Reserved_Identifier then
2593 return P_Identifier;
2595 elsif Prev_Token = Tok_Comma then
2596 Error_Msg_SP -- CODEFIX
2597 ("|extra "","" ignored");
2598 raise Error_Resync;
2600 else
2601 Error_Msg_AP ("missing operand");
2602 raise Error_Resync;
2603 end if;
2605 end case;
2606 end loop;
2607 end P_Primary;
2609 -------------------------------
2610 -- 4.4 Quantified_Expression --
2611 -------------------------------
2613 -- QUANTIFIED_EXPRESSION ::=
2614 -- for QUANTIFIER LOOP_PARAMETER_SPECIFICATION => PREDICATE |
2615 -- for QUANTIFIER ITERATOR_SPECIFICATION => PREDICATE
2617 function P_Quantified_Expression return Node_Id is
2618 I_Spec : Node_Id;
2619 Node1 : Node_Id;
2621 begin
2622 if Ada_Version < Ada_2012 then
2623 Error_Msg_SC ("quantified expression is an Ada 2012 feature");
2624 Error_Msg_SC ("\|unit must be compiled with -gnat2012 switch");
2625 end if;
2627 Scan; -- past FOR
2629 Node1 := New_Node (N_Quantified_Expression, Prev_Token_Ptr);
2631 if Token = Tok_All then
2632 Set_All_Present (Node1);
2634 elsif Token /= Tok_Some then
2635 Error_Msg_AP ("missing quantifier");
2636 raise Error_Resync;
2637 end if;
2639 Scan; -- past SOME
2640 I_Spec := P_Loop_Parameter_Specification;
2642 if Nkind (I_Spec) = N_Loop_Parameter_Specification then
2643 Set_Loop_Parameter_Specification (Node1, I_Spec);
2644 else
2645 Set_Iterator_Specification (Node1, I_Spec);
2646 end if;
2648 if Token = Tok_Arrow then
2649 Scan;
2650 Set_Condition (Node1, P_Expression);
2651 return Node1;
2652 else
2653 Error_Msg_AP ("missing arrow");
2654 raise Error_Resync;
2655 end if;
2656 end P_Quantified_Expression;
2658 ---------------------------
2659 -- 4.5 Logical Operator --
2660 ---------------------------
2662 -- LOGICAL_OPERATOR ::= and | or | xor
2664 -- Note: AND THEN and OR ELSE are also treated as logical operators
2665 -- by the parser (even though they are not operators semantically)
2667 -- The value returned is the appropriate Node_Kind code for the operator
2668 -- On return, Token points to the token following the scanned operator.
2670 -- The caller has checked that the first token is a legitimate logical
2671 -- operator token (i.e. is either XOR, AND, OR).
2673 -- Error recovery: cannot raise Error_Resync
2675 function P_Logical_Operator return Node_Kind is
2676 begin
2677 if Token = Tok_And then
2678 if Style_Check then
2679 Style.Check_Binary_Operator;
2680 end if;
2682 Scan; -- past AND
2684 if Token = Tok_Then then
2685 Scan; -- past THEN
2686 return N_And_Then;
2687 else
2688 return N_Op_And;
2689 end if;
2691 elsif Token = Tok_Or then
2692 if Style_Check then
2693 Style.Check_Binary_Operator;
2694 end if;
2696 Scan; -- past OR
2698 if Token = Tok_Else then
2699 Scan; -- past ELSE
2700 return N_Or_Else;
2701 else
2702 return N_Op_Or;
2703 end if;
2705 else -- Token = Tok_Xor
2706 if Style_Check then
2707 Style.Check_Binary_Operator;
2708 end if;
2710 Scan; -- past XOR
2711 return N_Op_Xor;
2712 end if;
2713 end P_Logical_Operator;
2715 ------------------------------
2716 -- 4.5 Relational Operator --
2717 ------------------------------
2719 -- RELATIONAL_OPERATOR ::= = | /= | < | <= | > | >=
2721 -- The value returned is the appropriate Node_Kind code for the operator.
2722 -- On return, Token points to the operator token, NOT past it.
2724 -- The caller has checked that the first token is a legitimate relational
2725 -- operator token (i.e. is one of the operator tokens listed above).
2727 -- Error recovery: cannot raise Error_Resync
2729 function P_Relational_Operator return Node_Kind is
2730 Op_Kind : Node_Kind;
2731 Relop_Node : constant array (Token_Class_Relop) of Node_Kind :=
2732 (Tok_Less => N_Op_Lt,
2733 Tok_Equal => N_Op_Eq,
2734 Tok_Greater => N_Op_Gt,
2735 Tok_Not_Equal => N_Op_Ne,
2736 Tok_Greater_Equal => N_Op_Ge,
2737 Tok_Less_Equal => N_Op_Le,
2738 Tok_In => N_In,
2739 Tok_Not => N_Not_In,
2740 Tok_Box => N_Op_Ne);
2742 begin
2743 if Token = Tok_Box then
2744 Error_Msg_SC -- CODEFIX
2745 ("|""'<'>"" should be ""/=""");
2746 end if;
2748 Op_Kind := Relop_Node (Token);
2750 if Style_Check then
2751 Style.Check_Binary_Operator;
2752 end if;
2754 Scan; -- past operator token
2756 -- Deal with NOT IN, if previous token was NOT, we must have IN now
2758 if Prev_Token = Tok_Not then
2760 -- Style check, for NOT IN, we require one space between NOT and IN
2762 if Style_Check and then Token = Tok_In then
2763 Style.Check_Not_In;
2764 end if;
2766 T_In;
2767 end if;
2769 return Op_Kind;
2770 end P_Relational_Operator;
2772 ---------------------------------
2773 -- 4.5 Binary Adding Operator --
2774 ---------------------------------
2776 -- BINARY_ADDING_OPERATOR ::= + | - | &
2778 -- The value returned is the appropriate Node_Kind code for the operator.
2779 -- On return, Token points to the operator token (NOT past it).
2781 -- The caller has checked that the first token is a legitimate adding
2782 -- operator token (i.e. is one of the operator tokens listed above).
2784 -- Error recovery: cannot raise Error_Resync
2786 function P_Binary_Adding_Operator return Node_Kind is
2787 Addop_Node : constant array (Token_Class_Binary_Addop) of Node_Kind :=
2788 (Tok_Ampersand => N_Op_Concat,
2789 Tok_Minus => N_Op_Subtract,
2790 Tok_Plus => N_Op_Add);
2791 begin
2792 return Addop_Node (Token);
2793 end P_Binary_Adding_Operator;
2795 --------------------------------
2796 -- 4.5 Unary Adding Operator --
2797 --------------------------------
2799 -- UNARY_ADDING_OPERATOR ::= + | -
2801 -- The value returned is the appropriate Node_Kind code for the operator.
2802 -- On return, Token points to the operator token (NOT past it).
2804 -- The caller has checked that the first token is a legitimate adding
2805 -- operator token (i.e. is one of the operator tokens listed above).
2807 -- Error recovery: cannot raise Error_Resync
2809 function P_Unary_Adding_Operator return Node_Kind is
2810 Addop_Node : constant array (Token_Class_Unary_Addop) of Node_Kind :=
2811 (Tok_Minus => N_Op_Minus,
2812 Tok_Plus => N_Op_Plus);
2813 begin
2814 return Addop_Node (Token);
2815 end P_Unary_Adding_Operator;
2817 -------------------------------
2818 -- 4.5 Multiplying Operator --
2819 -------------------------------
2821 -- MULTIPLYING_OPERATOR ::= * | / | mod | rem
2823 -- The value returned is the appropriate Node_Kind code for the operator.
2824 -- On return, Token points to the operator token (NOT past it).
2826 -- The caller has checked that the first token is a legitimate multiplying
2827 -- operator token (i.e. is one of the operator tokens listed above).
2829 -- Error recovery: cannot raise Error_Resync
2831 function P_Multiplying_Operator return Node_Kind is
2832 Mulop_Node : constant array (Token_Class_Mulop) of Node_Kind :=
2833 (Tok_Asterisk => N_Op_Multiply,
2834 Tok_Mod => N_Op_Mod,
2835 Tok_Rem => N_Op_Rem,
2836 Tok_Slash => N_Op_Divide);
2837 begin
2838 return Mulop_Node (Token);
2839 end P_Multiplying_Operator;
2841 --------------------------------------
2842 -- 4.5 Highest Precedence Operator --
2843 --------------------------------------
2845 -- Parsed by P_Factor (4.4)
2847 -- Note: this rule is not in fact used by the grammar at any point!
2849 --------------------------
2850 -- 4.6 Type Conversion --
2851 --------------------------
2853 -- Parsed by P_Primary as a Name (4.1)
2855 -------------------------------
2856 -- 4.7 Qualified Expression --
2857 -------------------------------
2859 -- QUALIFIED_EXPRESSION ::=
2860 -- SUBTYPE_MARK ' (EXPRESSION) | SUBTYPE_MARK ' AGGREGATE
2862 -- The caller has scanned the name which is the Subtype_Mark parameter
2863 -- and scanned past the single quote following the subtype mark. The
2864 -- caller has not checked that this name is in fact appropriate for
2865 -- a subtype mark name (i.e. it is a selected component or identifier).
2867 -- Error_Recovery: cannot raise Error_Resync
2869 function P_Qualified_Expression (Subtype_Mark : Node_Id) return Node_Id is
2870 Qual_Node : Node_Id;
2871 begin
2872 Qual_Node := New_Node (N_Qualified_Expression, Prev_Token_Ptr);
2873 Set_Subtype_Mark (Qual_Node, Check_Subtype_Mark (Subtype_Mark));
2874 Set_Expression (Qual_Node, P_Aggregate_Or_Paren_Expr);
2875 return Qual_Node;
2876 end P_Qualified_Expression;
2878 --------------------
2879 -- 4.8 Allocator --
2880 --------------------
2882 -- ALLOCATOR ::=
2883 -- new [SUBPOOL_SPECIFICATION] SUBTYPE_INDICATION
2884 -- | new [SUBPOOL_SPECIFICATION] QUALIFIED_EXPRESSION
2886 -- SUBPOOL_SPECIFICATION ::= (subpool_handle_NAME)
2888 -- The caller has checked that the initial token is NEW
2890 -- Error recovery: can raise Error_Resync
2892 function P_Allocator return Node_Id is
2893 Alloc_Node : Node_Id;
2894 Type_Node : Node_Id;
2895 Null_Exclusion_Present : Boolean;
2897 begin
2898 Alloc_Node := New_Node (N_Allocator, Token_Ptr);
2899 T_New;
2901 -- Scan subpool_specification if present (Ada 2012 (AI05-0111-3))
2903 -- Scan Null_Exclusion if present (Ada 2005 (AI-231))
2905 if Token = Tok_Left_Paren then
2906 Scan; -- past (
2907 Set_Subpool_Handle_Name (Alloc_Node, P_Name);
2908 T_Right_Paren;
2910 if Ada_Version < Ada_2012 then
2911 Error_Msg_N
2912 ("|subpool specification is an Ada 2012 feature",
2913 Subpool_Handle_Name (Alloc_Node));
2914 Error_Msg_N
2915 ("\|unit must be compiled with -gnat2012 switch",
2916 Subpool_Handle_Name (Alloc_Node));
2917 end if;
2918 end if;
2920 Null_Exclusion_Present := P_Null_Exclusion;
2921 Set_Null_Exclusion_Present (Alloc_Node, Null_Exclusion_Present);
2922 Type_Node := P_Subtype_Mark_Resync;
2924 if Token = Tok_Apostrophe then
2925 Scan; -- past apostrophe
2926 Set_Expression (Alloc_Node, P_Qualified_Expression (Type_Node));
2927 else
2928 Set_Expression
2929 (Alloc_Node,
2930 P_Subtype_Indication (Type_Node, Null_Exclusion_Present));
2932 -- AI05-0104: An explicit null exclusion is not allowed for an
2933 -- allocator without initialization. In previous versions of the
2934 -- language it just raises constraint error.
2936 if Ada_Version >= Ada_2012 and then Null_Exclusion_Present then
2937 Error_Msg_N
2938 ("an allocator with a subtype indication "
2939 & "cannot have a null exclusion", Alloc_Node);
2940 end if;
2941 end if;
2943 return Alloc_Node;
2944 end P_Allocator;
2946 -----------------------
2947 -- P_Case_Expression --
2948 -----------------------
2950 function P_Case_Expression return Node_Id is
2951 Loc : constant Source_Ptr := Token_Ptr;
2952 Case_Node : Node_Id;
2953 Save_State : Saved_Scan_State;
2955 begin
2956 if Ada_Version < Ada_2012 then
2957 Error_Msg_SC ("|case expression is an Ada 2012 feature");
2958 Error_Msg_SC ("\|unit must be compiled with -gnat2012 switch");
2959 end if;
2961 Scan; -- past CASE
2962 Case_Node :=
2963 Make_Case_Expression (Loc,
2964 Expression => P_Expression_No_Right_Paren,
2965 Alternatives => New_List);
2966 T_Is;
2968 -- We now have scanned out CASE expression IS, scan alternatives
2970 loop
2971 T_When;
2972 Append_To (Alternatives (Case_Node), P_Case_Expression_Alternative);
2974 -- Missing comma if WHEN (more alternatives present)
2976 if Token = Tok_When then
2977 T_Comma;
2979 -- If comma/WHEN, skip comma and we have another alternative
2981 elsif Token = Tok_Comma then
2982 Save_Scan_State (Save_State);
2983 Scan; -- past comma
2985 if Token /= Tok_When then
2986 Restore_Scan_State (Save_State);
2987 exit;
2988 end if;
2990 -- If no comma or WHEN, definitely done
2992 else
2993 exit;
2994 end if;
2995 end loop;
2997 -- If we have an END CASE, diagnose as not needed
2999 if Token = Tok_End then
3000 Error_Msg_SC ("`END CASE` not allowed at end of case expression");
3001 Scan; -- past END
3003 if Token = Tok_Case then
3004 Scan; -- past CASE;
3005 end if;
3006 end if;
3008 -- Return the Case_Expression node
3010 return Case_Node;
3011 end P_Case_Expression;
3013 -----------------------------------
3014 -- P_Case_Expression_Alternative --
3015 -----------------------------------
3017 -- CASE_STATEMENT_ALTERNATIVE ::=
3018 -- when DISCRETE_CHOICE_LIST =>
3019 -- EXPRESSION
3021 -- The caller has checked that and scanned past the initial WHEN token
3022 -- Error recovery: can raise Error_Resync
3024 function P_Case_Expression_Alternative return Node_Id is
3025 Case_Alt_Node : Node_Id;
3026 begin
3027 Case_Alt_Node := New_Node (N_Case_Expression_Alternative, Token_Ptr);
3028 Set_Discrete_Choices (Case_Alt_Node, P_Discrete_Choice_List);
3029 TF_Arrow;
3030 Set_Expression (Case_Alt_Node, P_Expression);
3031 return Case_Alt_Node;
3032 end P_Case_Expression_Alternative;
3034 ---------------------
3035 -- P_If_Expression --
3036 ---------------------
3038 function P_If_Expression return Node_Id is
3039 Exprs : constant List_Id := New_List;
3040 Loc : constant Source_Ptr := Token_Ptr;
3041 Expr : Node_Id;
3042 State : Saved_Scan_State;
3044 begin
3045 Inside_If_Expression := Inside_If_Expression + 1;
3047 if Token = Tok_If and then Ada_Version < Ada_2012 then
3048 Error_Msg_SC ("|if expression is an Ada 2012 feature");
3049 Error_Msg_SC ("\|unit must be compiled with -gnat2012 switch");
3050 end if;
3052 Scan; -- past IF or ELSIF
3053 Append_To (Exprs, P_Condition);
3054 TF_Then;
3055 Append_To (Exprs, P_Expression);
3057 -- We now have scanned out IF expr THEN expr
3059 -- Check for common error of semicolon before the ELSE
3061 if Token = Tok_Semicolon then
3062 Save_Scan_State (State);
3063 Scan; -- past semicolon
3065 if Token = Tok_Else or else Token = Tok_Elsif then
3066 Error_Msg_SP -- CODEFIX
3067 ("|extra "";"" ignored");
3069 else
3070 Restore_Scan_State (State);
3071 end if;
3072 end if;
3074 -- Scan out ELSIF sequence if present
3076 if Token = Tok_Elsif then
3077 Expr := P_If_Expression;
3078 Set_Is_Elsif (Expr);
3079 Append_To (Exprs, Expr);
3081 -- Scan out ELSE phrase if present
3083 elsif Token = Tok_Else then
3085 -- Scan out ELSE expression
3087 Scan; -- Past ELSE
3088 Append_To (Exprs, P_Expression);
3090 -- Two expression case (implied True, filled in during semantics)
3092 else
3093 null;
3094 end if;
3096 -- If we have an END IF, diagnose as not needed
3098 if Token = Tok_End then
3099 Error_Msg_SC ("`END IF` not allowed at end of if expression");
3100 Scan; -- past END
3102 if Token = Tok_If then
3103 Scan; -- past IF;
3104 end if;
3105 end if;
3107 Inside_If_Expression := Inside_If_Expression - 1;
3109 -- Return the If_Expression node
3111 return
3112 Make_If_Expression (Loc,
3113 Expressions => Exprs);
3114 end P_If_Expression;
3116 -----------------------
3117 -- P_Membership_Test --
3118 -----------------------
3120 -- MEMBERSHIP_CHOICE_LIST ::= MEMBERHIP_CHOICE {'|' MEMBERSHIP_CHOICE}
3121 -- MEMBERSHIP_CHOICE ::= CHOICE_EXPRESSION | range | subtype_mark
3123 procedure P_Membership_Test (N : Node_Id) is
3124 Alt : constant Node_Id :=
3125 P_Range_Or_Subtype_Mark
3126 (Allow_Simple_Expression => (Ada_Version >= Ada_2012));
3128 begin
3129 -- Set case
3131 if Token = Tok_Vertical_Bar then
3132 if Ada_Version < Ada_2012 then
3133 Error_Msg_SC ("set notation is an Ada 2012 feature");
3134 Error_Msg_SC ("\|unit must be compiled with -gnat2012 switch");
3135 end if;
3137 Set_Alternatives (N, New_List (Alt));
3138 Set_Right_Opnd (N, Empty);
3140 -- Loop to accumulate alternatives
3142 while Token = Tok_Vertical_Bar loop
3143 Scan; -- past vertical bar
3144 Append_To
3145 (Alternatives (N),
3146 P_Range_Or_Subtype_Mark (Allow_Simple_Expression => True));
3147 end loop;
3149 -- Not set case
3151 else
3152 Set_Right_Opnd (N, Alt);
3153 Set_Alternatives (N, No_List);
3154 end if;
3155 end P_Membership_Test;
3157 ------------------------------------------
3158 -- P_Unparen_Cond_Case_Quant_Expression --
3159 ------------------------------------------
3161 function P_Unparen_Cond_Case_Quant_Expression return Node_Id is
3162 Lparen : constant Boolean := Prev_Token = Tok_Left_Paren;
3163 Result : Node_Id;
3165 begin
3166 -- Case expression
3168 if Token = Tok_Case then
3169 Result := P_Case_Expression;
3171 if not (Lparen and then Token = Tok_Right_Paren) then
3172 Error_Msg_N ("case expression must be parenthesized!", Result);
3173 end if;
3175 -- If expression
3177 elsif Token = Tok_If then
3178 Result := P_If_Expression;
3180 if not (Lparen and then Token = Tok_Right_Paren) then
3181 Error_Msg_N ("if expression must be parenthesized!", Result);
3182 end if;
3184 -- Quantified expression
3186 elsif Token = Tok_For then
3187 Result := P_Quantified_Expression;
3189 if not (Lparen and then Token = Tok_Right_Paren) then
3190 Error_Msg_N
3191 ("quantified expression must be parenthesized!", Result);
3192 end if;
3194 -- No other possibility should exist (caller was supposed to check)
3196 else
3197 raise Program_Error;
3198 end if;
3200 -- Return expression (possibly after having given message)
3202 return Result;
3203 end P_Unparen_Cond_Case_Quant_Expression;
3205 end Ch4;