gcc/
[official-gcc.git] / gcc / ada / par-ch4.adb
blob7926bd11766e16c410c15ee2ffbe46f0b02f69f5
1 ------------------------------------------------------------------------------
2 -- --
3 -- GNAT COMPILER COMPONENTS --
4 -- --
5 -- P A R . C H 4 --
6 -- --
7 -- B o d y --
8 -- --
9 -- Copyright (C) 1992-2014, 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_Loop_Entry => True,
44 Attribute_Old => True,
45 Attribute_Stub_Type => True,
46 Attribute_Version => True,
47 Attribute_Type_Key => True,
48 others => False);
49 -- This map contains True for parameterless attributes that return a
50 -- string or a type. For those attributes, a left parenthesis after
51 -- the attribute should not be analyzed as the beginning of a parameters
52 -- list because it may denote a slice operation (X'Img (1 .. 2)) or
53 -- a type conversion (X'Class (Y)). The Ada2012 attribute 'Old is in
54 -- this category.
56 -- Note: Loop_Entry is in this list because, although it can take an
57 -- optional argument (the loop name), we can't distinguish that at parse
58 -- time from the case where no loop name is given and a legitimate index
59 -- expression is present. So we parse the argument as an indexed component
60 -- and the semantic analysis sorts out this syntactic ambiguity based on
61 -- the type and form of the expression.
63 -- Note that this map designates the minimum set of attributes where a
64 -- construct in parentheses that is not an argument can appear right
65 -- after the attribute. For attributes like 'Size, we do not put them
66 -- in the map. If someone writes X'Size (3), that's illegal in any case,
67 -- but we get a better error message by parsing the (3) as an illegal
68 -- argument to the attribute, rather than some meaningless junk that
69 -- follows the attribute.
71 -----------------------
72 -- Local Subprograms --
73 -----------------------
75 function P_Aggregate_Or_Paren_Expr return Node_Id;
76 function P_Allocator return Node_Id;
77 function P_Case_Expression_Alternative return Node_Id;
78 function P_Record_Or_Array_Component_Association return Node_Id;
79 function P_Factor return Node_Id;
80 function P_Primary return Node_Id;
81 function P_Relation return Node_Id;
82 function P_Term return Node_Id;
84 function P_Binary_Adding_Operator return Node_Kind;
85 function P_Logical_Operator return Node_Kind;
86 function P_Multiplying_Operator return Node_Kind;
87 function P_Relational_Operator return Node_Kind;
88 function P_Unary_Adding_Operator return Node_Kind;
90 procedure Bad_Range_Attribute (Loc : Source_Ptr);
91 -- Called to place complaint about bad range attribute at the given
92 -- source location. Terminates by raising Error_Resync.
94 procedure Check_Bad_Exp;
95 -- Called after scanning a**b, posts error if ** detected
97 procedure P_Membership_Test (N : Node_Id);
98 -- N is the node for a N_In or N_Not_In node whose right operand has not
99 -- yet been processed. It is called just after scanning out the IN keyword.
100 -- On return, either Right_Opnd or Alternatives is set, as appropriate.
102 function P_Range_Attribute_Reference (Prefix_Node : Node_Id) return Node_Id;
103 -- Scan a range attribute reference. The caller has scanned out the
104 -- prefix. The current token is known to be an apostrophe and the
105 -- following token is known to be RANGE.
107 function P_Unparen_Cond_Case_Quant_Expression return Node_Id;
108 -- This function is called with Token pointing to IF, CASE, or FOR, in a
109 -- context that allows a case, conditional, or quantified expression if
110 -- it is surrounded by parentheses. If not surrounded by parentheses, the
111 -- expression is still returned, but an error message is issued.
113 -------------------------
114 -- Bad_Range_Attribute --
115 -------------------------
117 procedure Bad_Range_Attribute (Loc : Source_Ptr) is
118 begin
119 Error_Msg ("range attribute cannot be used in expression!", Loc);
120 Resync_Expression;
121 end Bad_Range_Attribute;
123 -------------------
124 -- Check_Bad_Exp --
125 -------------------
127 procedure Check_Bad_Exp is
128 begin
129 if Token = Tok_Double_Asterisk then
130 Error_Msg_SC ("parenthesization required for '*'*");
131 Scan; -- past **
132 Discard_Junk_Node (P_Primary);
133 Check_Bad_Exp;
134 end if;
135 end Check_Bad_Exp;
137 --------------------------
138 -- 4.1 Name (also 6.4) --
139 --------------------------
141 -- NAME ::=
142 -- DIRECT_NAME | EXPLICIT_DEREFERENCE
143 -- | INDEXED_COMPONENT | SLICE
144 -- | SELECTED_COMPONENT | ATTRIBUTE
145 -- | TYPE_CONVERSION | FUNCTION_CALL
146 -- | CHARACTER_LITERAL
148 -- DIRECT_NAME ::= IDENTIFIER | OPERATOR_SYMBOL
150 -- PREFIX ::= NAME | IMPLICIT_DEREFERENCE
152 -- EXPLICIT_DEREFERENCE ::= NAME . all
154 -- IMPLICIT_DEREFERENCE ::= NAME
156 -- INDEXED_COMPONENT ::= PREFIX (EXPRESSION {, EXPRESSION})
158 -- SLICE ::= PREFIX (DISCRETE_RANGE)
160 -- SELECTED_COMPONENT ::= PREFIX . SELECTOR_NAME
162 -- SELECTOR_NAME ::= IDENTIFIER | CHARACTER_LITERAL | OPERATOR_SYMBOL
164 -- ATTRIBUTE_REFERENCE ::= PREFIX ' ATTRIBUTE_DESIGNATOR
166 -- ATTRIBUTE_DESIGNATOR ::=
167 -- IDENTIFIER [(static_EXPRESSION)]
168 -- | access | delta | digits
170 -- FUNCTION_CALL ::=
171 -- function_NAME
172 -- | function_PREFIX ACTUAL_PARAMETER_PART
174 -- ACTUAL_PARAMETER_PART ::=
175 -- (PARAMETER_ASSOCIATION {,PARAMETER_ASSOCIATION})
177 -- PARAMETER_ASSOCIATION ::=
178 -- [formal_parameter_SELECTOR_NAME =>] EXPLICIT_ACTUAL_PARAMETER
180 -- EXPLICIT_ACTUAL_PARAMETER ::= EXPRESSION | variable_NAME
182 -- Note: syntactically a procedure call looks just like a function call,
183 -- so this routine is in practice used to scan out procedure calls as well.
185 -- On return, Expr_Form is set to either EF_Name or EF_Simple_Name
187 -- Error recovery: can raise Error_Resync
189 -- Note: if on return Token = Tok_Apostrophe, then the apostrophe must be
190 -- followed by either a left paren (qualified expression case), or by
191 -- range (range attribute case). All other uses of apostrophe (i.e. all
192 -- other attributes) are handled in this routine.
194 -- Error recovery: can raise Error_Resync
196 function P_Name return Node_Id is
197 Scan_State : Saved_Scan_State;
198 Name_Node : Node_Id;
199 Prefix_Node : Node_Id;
200 Ident_Node : Node_Id;
201 Expr_Node : Node_Id;
202 Range_Node : Node_Id;
203 Arg_Node : Node_Id;
205 Arg_List : List_Id := No_List; -- kill junk warning
206 Attr_Name : Name_Id := No_Name; -- kill junk warning
208 begin
209 -- Case of not a name
211 if Token not in Token_Class_Name then
213 -- If it looks like start of expression, complain and scan expression
215 if Token in Token_Class_Literal
216 or else Token = Tok_Left_Paren
217 then
218 Error_Msg_SC ("name expected");
219 return P_Expression;
221 -- Otherwise some other junk, not much we can do
223 else
224 Error_Msg_AP ("name expected");
225 raise Error_Resync;
226 end if;
227 end if;
229 -- Loop through designators in qualified name
231 Name_Node := Token_Node;
233 loop
234 Scan; -- past designator
235 exit when Token /= Tok_Dot;
236 Save_Scan_State (Scan_State); -- at dot
237 Scan; -- past dot
239 -- If we do not have another designator after the dot, then join
240 -- the normal circuit to handle a dot extension (may be .all or
241 -- character literal case). Otherwise loop back to scan the next
242 -- designator.
244 if Token not in Token_Class_Desig then
245 goto Scan_Name_Extension_Dot;
246 else
247 Prefix_Node := Name_Node;
248 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
249 Set_Prefix (Name_Node, Prefix_Node);
250 Set_Selector_Name (Name_Node, Token_Node);
251 end if;
252 end loop;
254 -- We have now scanned out a qualified designator. If the last token is
255 -- an operator symbol, then we certainly do not have the Snam case, so
256 -- we can just use the normal name extension check circuit
258 if Prev_Token = Tok_Operator_Symbol then
259 goto Scan_Name_Extension;
260 end if;
262 -- We have scanned out a qualified simple name, check for name extension
263 -- Note that we know there is no dot here at this stage, so the only
264 -- possible cases of name extension are apostrophe and left paren.
266 if Token = Tok_Apostrophe then
267 Save_Scan_State (Scan_State); -- at apostrophe
268 Scan; -- past apostrophe
270 -- Qualified expression in Ada 2012 mode (treated as a name)
272 if Ada_Version >= Ada_2012 and then Token = Tok_Left_Paren then
273 goto Scan_Name_Extension_Apostrophe;
275 -- If left paren not in Ada 2012, then it is not part of the name,
276 -- since qualified expressions are not names in prior versions of
277 -- Ada, so return with Token backed up to point to the apostrophe.
278 -- The treatment for the range attribute is similar (we do not
279 -- consider x'range to be a name in this grammar).
281 elsif Token = Tok_Left_Paren or else Token = Tok_Range then
282 Restore_Scan_State (Scan_State); -- to apostrophe
283 Expr_Form := EF_Simple_Name;
284 return Name_Node;
286 -- Otherwise we have the case of a name extended by an attribute
288 else
289 goto Scan_Name_Extension_Apostrophe;
290 end if;
292 -- Check case of qualified simple name extended by a left parenthesis
294 elsif Token = Tok_Left_Paren then
295 Scan; -- past left paren
296 goto Scan_Name_Extension_Left_Paren;
298 -- Otherwise the qualified simple name is not extended, so return
300 else
301 Expr_Form := EF_Simple_Name;
302 return Name_Node;
303 end if;
305 -- Loop scanning past name extensions. A label is used for control
306 -- transfer for this loop for ease of interfacing with the finite state
307 -- machine in the parenthesis scanning circuit, and also to allow for
308 -- passing in control to the appropriate point from the above code.
310 <<Scan_Name_Extension>>
312 -- Character literal used as name cannot be extended. Also this
313 -- cannot be a call, since the name for a call must be a designator.
314 -- Return in these cases, or if there is no name extension
316 if Token not in Token_Class_Namext
317 or else Prev_Token = Tok_Char_Literal
318 then
319 Expr_Form := EF_Name;
320 return Name_Node;
321 end if;
323 -- Merge here when we know there is a name extension
325 <<Scan_Name_Extension_OK>>
327 if Token = Tok_Left_Paren then
328 Scan; -- past left paren
329 goto Scan_Name_Extension_Left_Paren;
331 elsif Token = Tok_Apostrophe then
332 Save_Scan_State (Scan_State); -- at apostrophe
333 Scan; -- past apostrophe
334 goto Scan_Name_Extension_Apostrophe;
336 else -- Token = Tok_Dot
337 Save_Scan_State (Scan_State); -- at dot
338 Scan; -- past dot
339 goto Scan_Name_Extension_Dot;
340 end if;
342 -- Case of name extended by dot (selection), dot is already skipped
343 -- and the scan state at the point of the dot is saved in Scan_State.
345 <<Scan_Name_Extension_Dot>>
347 -- Explicit dereference case
349 if Token = Tok_All then
350 Prefix_Node := Name_Node;
351 Name_Node := New_Node (N_Explicit_Dereference, Token_Ptr);
352 Set_Prefix (Name_Node, Prefix_Node);
353 Scan; -- past ALL
354 goto Scan_Name_Extension;
356 -- Selected component case
358 elsif Token in Token_Class_Name then
359 Prefix_Node := Name_Node;
360 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
361 Set_Prefix (Name_Node, Prefix_Node);
362 Set_Selector_Name (Name_Node, Token_Node);
363 Scan; -- past selector
364 goto Scan_Name_Extension;
366 -- Reserved identifier as selector
368 elsif Is_Reserved_Identifier then
369 Scan_Reserved_Identifier (Force_Msg => False);
370 Prefix_Node := Name_Node;
371 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
372 Set_Prefix (Name_Node, Prefix_Node);
373 Set_Selector_Name (Name_Node, Token_Node);
374 Scan; -- past identifier used as selector
375 goto Scan_Name_Extension;
377 -- If dot is at end of line and followed by nothing legal,
378 -- then assume end of name and quit (dot will be taken as
379 -- an incorrect form of some other punctuation by our caller).
381 elsif Token_Is_At_Start_Of_Line then
382 Restore_Scan_State (Scan_State);
383 return Name_Node;
385 -- Here if nothing legal after the dot
387 else
388 Error_Msg_AP ("selector expected");
389 raise Error_Resync;
390 end if;
392 -- Here for an apostrophe as name extension. The scan position at the
393 -- apostrophe has already been saved, and the apostrophe scanned out.
395 <<Scan_Name_Extension_Apostrophe>>
397 Scan_Apostrophe : declare
398 function Apostrophe_Should_Be_Semicolon return Boolean;
399 -- Checks for case where apostrophe should probably be
400 -- a semicolon, and if so, gives appropriate message,
401 -- resets the scan pointer to the apostrophe, changes
402 -- the current token to Tok_Semicolon, and returns True.
403 -- Otherwise returns False.
405 ------------------------------------
406 -- Apostrophe_Should_Be_Semicolon --
407 ------------------------------------
409 function Apostrophe_Should_Be_Semicolon return Boolean is
410 begin
411 if Token_Is_At_Start_Of_Line then
412 Restore_Scan_State (Scan_State); -- to apostrophe
413 Error_Msg_SC ("|""''"" should be "";""");
414 Token := Tok_Semicolon;
415 return True;
416 else
417 return False;
418 end if;
419 end Apostrophe_Should_Be_Semicolon;
421 -- Start of processing for Scan_Apostrophe
423 begin
424 -- Check for qualified expression case in Ada 2012 mode
426 if Ada_Version >= Ada_2012 and then Token = Tok_Left_Paren then
427 Name_Node := P_Qualified_Expression (Name_Node);
428 goto Scan_Name_Extension;
430 -- If range attribute after apostrophe, then return with Token
431 -- pointing to the apostrophe. Note that in this case the prefix
432 -- need not be a simple name (cases like A.all'range). Similarly
433 -- if there is a left paren after the apostrophe, then we also
434 -- return with Token pointing to the apostrophe (this is the
435 -- aggregate case, or some error case).
437 elsif Token = Tok_Range or else Token = Tok_Left_Paren then
438 Restore_Scan_State (Scan_State); -- to apostrophe
439 Expr_Form := EF_Name;
440 return Name_Node;
442 -- Here for cases where attribute designator is an identifier
444 elsif Token = Tok_Identifier then
445 Attr_Name := Token_Name;
447 if not Is_Attribute_Name (Attr_Name) then
448 if Apostrophe_Should_Be_Semicolon then
449 Expr_Form := EF_Name;
450 return Name_Node;
452 -- Here for a bad attribute name
454 else
455 Signal_Bad_Attribute;
456 Scan; -- past bad identifier
458 if Token = Tok_Left_Paren then
459 Scan; -- past left paren
461 loop
462 Discard_Junk_Node (P_Expression_If_OK);
463 exit when not Comma_Present;
464 end loop;
466 T_Right_Paren;
467 end if;
469 return Error;
470 end if;
471 end if;
473 if Style_Check then
474 Style.Check_Attribute_Name (False);
475 end if;
477 -- Here for case of attribute designator is not an identifier
479 else
480 if Token = Tok_Delta then
481 Attr_Name := Name_Delta;
483 elsif Token = Tok_Digits then
484 Attr_Name := Name_Digits;
486 elsif Token = Tok_Access then
487 Attr_Name := Name_Access;
489 elsif Token = Tok_Mod and then Ada_Version >= Ada_95 then
490 Attr_Name := Name_Mod;
492 elsif Apostrophe_Should_Be_Semicolon then
493 Expr_Form := EF_Name;
494 return Name_Node;
496 else
497 Error_Msg_AP ("attribute designator expected");
498 raise Error_Resync;
499 end if;
501 if Style_Check then
502 Style.Check_Attribute_Name (True);
503 end if;
504 end if;
506 -- We come here with an OK attribute scanned, and corresponding
507 -- Attribute identifier node stored in Ident_Node.
509 Prefix_Node := Name_Node;
510 Name_Node := New_Node (N_Attribute_Reference, Prev_Token_Ptr);
511 Scan; -- past attribute designator
512 Set_Prefix (Name_Node, Prefix_Node);
513 Set_Attribute_Name (Name_Node, Attr_Name);
515 -- Scan attribute arguments/designator. We skip this if we know
516 -- that the attribute cannot have an argument (see documentation
517 -- of Is_Parameterless_Attribute for further details).
519 if Token = Tok_Left_Paren
520 and then not
521 Is_Parameterless_Attribute (Get_Attribute_Id (Attr_Name))
522 then
523 -- Attribute Update contains an array or record association
524 -- list which provides new values for various components or
525 -- elements. The list is parsed as an aggregate, and we get
526 -- better error handling by knowing that in the parser.
528 if Attr_Name = Name_Update then
529 Set_Expressions (Name_Node, New_List);
530 Append (P_Aggregate, Expressions (Name_Node));
532 -- All other cases of parsing attribute arguments
534 else
535 Set_Expressions (Name_Node, New_List);
536 Scan; -- past left paren
538 loop
539 declare
540 Expr : constant Node_Id := P_Expression_If_OK;
541 Rnam : Node_Id;
543 begin
544 -- Case of => for named notation
546 if Token = Tok_Arrow then
548 -- Named notation allowed only for the special
549 -- case of System'Restriction_Set (No_Dependence =>
550 -- unit_NAME), in which case construct a parameter
551 -- assocation node and append to the arguments.
553 if Attr_Name = Name_Restriction_Set
554 and then Nkind (Expr) = N_Identifier
555 and then Chars (Expr) = Name_No_Dependence
556 then
557 Scan; -- past arrow
558 Rnam := P_Name;
559 Append_To (Expressions (Name_Node),
560 Make_Parameter_Association (Sloc (Rnam),
561 Selector_Name => Expr,
562 Explicit_Actual_Parameter => Rnam));
563 exit;
565 -- For all other cases named notation is illegal
567 else
568 Error_Msg_SC
569 ("named parameters not permitted "
570 & "for attributes");
571 Scan; -- past junk arrow
572 end if;
574 -- Here for normal case (not => for named parameter)
576 else
577 Append (Expr, Expressions (Name_Node));
578 exit when not Comma_Present;
579 end if;
580 end;
581 end loop;
583 T_Right_Paren;
584 end if;
585 end if;
587 goto Scan_Name_Extension;
588 end Scan_Apostrophe;
590 -- Here for left parenthesis extending name (left paren skipped)
592 <<Scan_Name_Extension_Left_Paren>>
594 -- We now have to scan through a list of items, terminated by a
595 -- right parenthesis. The scan is handled by a finite state
596 -- machine. The possibilities are:
598 -- (discrete_range)
600 -- This is a slice. This case is handled in LP_State_Init
602 -- (expression, expression, ..)
604 -- This is interpreted as an indexed component, i.e. as a
605 -- case of a name which can be extended in the normal manner.
606 -- This case is handled by LP_State_Name or LP_State_Expr.
608 -- Note: if and case expressions (without an extra level of
609 -- parentheses) are permitted in this context).
611 -- (..., identifier => expression , ...)
613 -- If there is at least one occurrence of identifier => (but
614 -- none of the other cases apply), then we have a call.
616 -- Test for Id => case
618 if Token = Tok_Identifier then
619 Save_Scan_State (Scan_State); -- at Id
620 Scan; -- past Id
622 -- Test for => (allow := as an error substitute)
624 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
625 Restore_Scan_State (Scan_State); -- to Id
626 Arg_List := New_List;
627 goto LP_State_Call;
629 else
630 Restore_Scan_State (Scan_State); -- to Id
631 end if;
632 end if;
634 -- Here we have an expression after all
636 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
638 -- Check cases of discrete range for a slice
640 -- First possibility: Range_Attribute_Reference
642 if Expr_Form = EF_Range_Attr then
643 Range_Node := Expr_Node;
645 -- Second possibility: Simple_expression .. Simple_expression
647 elsif Token = Tok_Dot_Dot then
648 Check_Simple_Expression (Expr_Node);
649 Range_Node := New_Node (N_Range, Token_Ptr);
650 Set_Low_Bound (Range_Node, Expr_Node);
651 Scan; -- past ..
652 Expr_Node := P_Expression;
653 Check_Simple_Expression (Expr_Node);
654 Set_High_Bound (Range_Node, Expr_Node);
656 -- Third possibility: Type_name range Range
658 elsif Token = Tok_Range then
659 if Expr_Form /= EF_Simple_Name then
660 Error_Msg_SC ("subtype mark must precede RANGE");
661 raise Error_Resync;
662 end if;
664 Range_Node := P_Subtype_Indication (Expr_Node);
666 -- Otherwise we just have an expression. It is true that we might
667 -- have a subtype mark without a range constraint but this case
668 -- is syntactically indistinguishable from the expression case.
670 else
671 Arg_List := New_List;
672 goto LP_State_Expr;
673 end if;
675 -- Fall through here with unmistakable Discrete range scanned,
676 -- which means that we definitely have the case of a slice. The
677 -- Discrete range is in Range_Node.
679 if Token = Tok_Comma then
680 Error_Msg_SC ("slice cannot have more than one dimension");
681 raise Error_Resync;
683 elsif Token /= Tok_Right_Paren then
684 if Token = Tok_Arrow then
686 -- This may be an aggregate that is missing a qualification
688 Error_Msg_SC
689 ("context of aggregate must be a qualified expression");
690 raise Error_Resync;
692 else
693 T_Right_Paren;
694 raise Error_Resync;
695 end if;
697 else
698 Scan; -- past right paren
699 Prefix_Node := Name_Node;
700 Name_Node := New_Node (N_Slice, Sloc (Prefix_Node));
701 Set_Prefix (Name_Node, Prefix_Node);
702 Set_Discrete_Range (Name_Node, Range_Node);
704 -- An operator node is legal as a prefix to other names,
705 -- but not for a slice.
707 if Nkind (Prefix_Node) = N_Operator_Symbol then
708 Error_Msg_N ("illegal prefix for slice", Prefix_Node);
709 end if;
711 -- If we have a name extension, go scan it
713 if Token in Token_Class_Namext then
714 goto Scan_Name_Extension_OK;
716 -- Otherwise return (a slice is a name, but is not a call)
718 else
719 Expr_Form := EF_Name;
720 return Name_Node;
721 end if;
722 end if;
724 -- In LP_State_Expr, we have scanned one or more expressions, and
725 -- so we have a call or an indexed component which is a name. On
726 -- entry we have the expression just scanned in Expr_Node and
727 -- Arg_List contains the list of expressions encountered so far
729 <<LP_State_Expr>>
730 Append (Expr_Node, Arg_List);
732 if Token = Tok_Arrow then
733 Error_Msg
734 ("expect identifier in parameter association", Sloc (Expr_Node));
735 Scan; -- past arrow
737 elsif not Comma_Present then
738 T_Right_Paren;
740 Prefix_Node := Name_Node;
741 Name_Node := New_Node (N_Indexed_Component, Sloc (Prefix_Node));
742 Set_Prefix (Name_Node, Prefix_Node);
743 Set_Expressions (Name_Node, Arg_List);
745 goto Scan_Name_Extension;
746 end if;
748 -- Comma present (and scanned out), test for identifier => case
749 -- Test for identifier => case
751 if Token = Tok_Identifier then
752 Save_Scan_State (Scan_State); -- at Id
753 Scan; -- past Id
755 -- Test for => (allow := as error substitute)
757 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
758 Restore_Scan_State (Scan_State); -- to Id
759 goto LP_State_Call;
761 -- Otherwise it's just an expression after all, so backup
763 else
764 Restore_Scan_State (Scan_State); -- to Id
765 end if;
766 end if;
768 -- Here we have an expression after all, so stay in this state
770 Expr_Node := P_Expression_If_OK;
771 goto LP_State_Expr;
773 -- LP_State_Call corresponds to the situation in which at least one
774 -- instance of Id => Expression has been encountered, so we know that
775 -- we do not have a name, but rather a call. We enter it with the
776 -- scan pointer pointing to the next argument to scan, and Arg_List
777 -- containing the list of arguments scanned so far.
779 <<LP_State_Call>>
781 -- Test for case of Id => Expression (named parameter)
783 if Token = Tok_Identifier then
784 Save_Scan_State (Scan_State); -- at Id
785 Ident_Node := Token_Node;
786 Scan; -- past Id
788 -- Deal with => (allow := as incorrect substitute)
790 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
791 Arg_Node := New_Node (N_Parameter_Association, Prev_Token_Ptr);
792 Set_Selector_Name (Arg_Node, Ident_Node);
793 T_Arrow;
794 Set_Explicit_Actual_Parameter (Arg_Node, P_Expression);
795 Append (Arg_Node, Arg_List);
797 -- If a comma follows, go back and scan next entry
799 if Comma_Present then
800 goto LP_State_Call;
802 -- Otherwise we have the end of a call
804 else
805 Prefix_Node := Name_Node;
806 Name_Node := New_Node (N_Function_Call, Sloc (Prefix_Node));
807 Set_Name (Name_Node, Prefix_Node);
808 Set_Parameter_Associations (Name_Node, Arg_List);
809 T_Right_Paren;
811 if Token in Token_Class_Namext then
812 goto Scan_Name_Extension_OK;
814 -- This is a case of a call which cannot be a name
816 else
817 Expr_Form := EF_Name;
818 return Name_Node;
819 end if;
820 end if;
822 -- Not named parameter: Id started an expression after all
824 else
825 Restore_Scan_State (Scan_State); -- to Id
826 end if;
827 end if;
829 -- Here if entry did not start with Id => which means that it
830 -- is a positional parameter, which is not allowed, since we
831 -- have seen at least one named parameter already.
833 Error_Msg_SC
834 ("positional parameter association " &
835 "not allowed after named one");
837 Expr_Node := P_Expression_If_OK;
839 -- Leaving the '>' in an association is not unusual, so suggest
840 -- a possible fix.
842 if Nkind (Expr_Node) = N_Op_Eq then
843 Error_Msg_N ("\maybe `='>` was intended", Expr_Node);
844 end if;
846 -- We go back to scanning out expressions, so that we do not get
847 -- multiple error messages when several positional parameters
848 -- follow a named parameter.
850 goto LP_State_Expr;
852 -- End of treatment for name extensions starting with left paren
854 -- End of loop through name extensions
856 end P_Name;
858 -- This function parses a restricted form of Names which are either
859 -- designators, or designators preceded by a sequence of prefixes
860 -- that are direct names.
862 -- Error recovery: cannot raise Error_Resync
864 function P_Function_Name return Node_Id is
865 Designator_Node : Node_Id;
866 Prefix_Node : Node_Id;
867 Selector_Node : Node_Id;
868 Dot_Sloc : Source_Ptr := No_Location;
870 begin
871 -- Prefix_Node is set to the gathered prefix so far, Empty means that
872 -- no prefix has been scanned. This allows us to build up the result
873 -- in the required right recursive manner.
875 Prefix_Node := Empty;
877 -- Loop through prefixes
879 loop
880 Designator_Node := Token_Node;
882 if Token not in Token_Class_Desig then
883 return P_Identifier; -- let P_Identifier issue the error message
885 else -- Token in Token_Class_Desig
886 Scan; -- past designator
887 exit when Token /= Tok_Dot;
888 end if;
890 -- Here at a dot, with token just before it in Designator_Node
892 if No (Prefix_Node) then
893 Prefix_Node := Designator_Node;
894 else
895 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
896 Set_Prefix (Selector_Node, Prefix_Node);
897 Set_Selector_Name (Selector_Node, Designator_Node);
898 Prefix_Node := Selector_Node;
899 end if;
901 Dot_Sloc := Token_Ptr;
902 Scan; -- past dot
903 end loop;
905 -- Fall out of the loop having just scanned a designator
907 if No (Prefix_Node) then
908 return Designator_Node;
909 else
910 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
911 Set_Prefix (Selector_Node, Prefix_Node);
912 Set_Selector_Name (Selector_Node, Designator_Node);
913 return Selector_Node;
914 end if;
916 exception
917 when Error_Resync =>
918 return Error;
919 end P_Function_Name;
921 -- This function parses a restricted form of Names which are either
922 -- identifiers, or identifiers preceded by a sequence of prefixes
923 -- that are direct names.
925 -- Error recovery: cannot raise Error_Resync
927 function P_Qualified_Simple_Name return Node_Id is
928 Designator_Node : Node_Id;
929 Prefix_Node : Node_Id;
930 Selector_Node : Node_Id;
931 Dot_Sloc : Source_Ptr := No_Location;
933 begin
934 -- Prefix node is set to the gathered prefix so far, Empty means that
935 -- no prefix has been scanned. This allows us to build up the result
936 -- in the required right recursive manner.
938 Prefix_Node := Empty;
940 -- Loop through prefixes
942 loop
943 Designator_Node := Token_Node;
945 if Token = Tok_Identifier then
946 Scan; -- past identifier
947 exit when Token /= Tok_Dot;
949 elsif Token not in Token_Class_Desig then
950 return P_Identifier; -- let P_Identifier issue the error message
952 else
953 Scan; -- past designator
955 if Token /= Tok_Dot then
956 Error_Msg_SP ("identifier expected");
957 return Error;
958 end if;
959 end if;
961 -- Here at a dot, with token just before it in Designator_Node
963 if No (Prefix_Node) then
964 Prefix_Node := Designator_Node;
965 else
966 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
967 Set_Prefix (Selector_Node, Prefix_Node);
968 Set_Selector_Name (Selector_Node, Designator_Node);
969 Prefix_Node := Selector_Node;
970 end if;
972 Dot_Sloc := Token_Ptr;
973 Scan; -- past dot
974 end loop;
976 -- Fall out of the loop having just scanned an identifier
978 if No (Prefix_Node) then
979 return Designator_Node;
980 else
981 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
982 Set_Prefix (Selector_Node, Prefix_Node);
983 Set_Selector_Name (Selector_Node, Designator_Node);
984 return Selector_Node;
985 end if;
987 exception
988 when Error_Resync =>
989 return Error;
990 end P_Qualified_Simple_Name;
992 -- This procedure differs from P_Qualified_Simple_Name only in that it
993 -- raises Error_Resync if any error is encountered. It only returns after
994 -- scanning a valid qualified simple name.
996 -- Error recovery: can raise Error_Resync
998 function P_Qualified_Simple_Name_Resync return Node_Id is
999 Designator_Node : Node_Id;
1000 Prefix_Node : Node_Id;
1001 Selector_Node : Node_Id;
1002 Dot_Sloc : Source_Ptr := No_Location;
1004 begin
1005 Prefix_Node := Empty;
1007 -- Loop through prefixes
1009 loop
1010 Designator_Node := Token_Node;
1012 if Token = Tok_Identifier then
1013 Scan; -- past identifier
1014 exit when Token /= Tok_Dot;
1016 elsif Token not in Token_Class_Desig then
1017 Discard_Junk_Node (P_Identifier); -- to issue the error message
1018 raise Error_Resync;
1020 else
1021 Scan; -- past designator
1023 if Token /= Tok_Dot then
1024 Error_Msg_SP ("identifier expected");
1025 raise Error_Resync;
1026 end if;
1027 end if;
1029 -- Here at a dot, with token just before it in Designator_Node
1031 if No (Prefix_Node) then
1032 Prefix_Node := Designator_Node;
1033 else
1034 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
1035 Set_Prefix (Selector_Node, Prefix_Node);
1036 Set_Selector_Name (Selector_Node, Designator_Node);
1037 Prefix_Node := Selector_Node;
1038 end if;
1040 Dot_Sloc := Token_Ptr;
1041 Scan; -- past period
1042 end loop;
1044 -- Fall out of the loop having just scanned an identifier
1046 if No (Prefix_Node) then
1047 return Designator_Node;
1048 else
1049 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
1050 Set_Prefix (Selector_Node, Prefix_Node);
1051 Set_Selector_Name (Selector_Node, Designator_Node);
1052 return Selector_Node;
1053 end if;
1054 end P_Qualified_Simple_Name_Resync;
1056 ----------------------
1057 -- 4.1 Direct_Name --
1058 ----------------------
1060 -- Parsed by P_Name and other functions in section 4.1
1062 -----------------
1063 -- 4.1 Prefix --
1064 -----------------
1066 -- Parsed by P_Name (4.1)
1068 -------------------------------
1069 -- 4.1 Explicit Dereference --
1070 -------------------------------
1072 -- Parsed by P_Name (4.1)
1074 -------------------------------
1075 -- 4.1 Implicit_Dereference --
1076 -------------------------------
1078 -- Parsed by P_Name (4.1)
1080 ----------------------------
1081 -- 4.1 Indexed Component --
1082 ----------------------------
1084 -- Parsed by P_Name (4.1)
1086 ----------------
1087 -- 4.1 Slice --
1088 ----------------
1090 -- Parsed by P_Name (4.1)
1092 -----------------------------
1093 -- 4.1 Selected_Component --
1094 -----------------------------
1096 -- Parsed by P_Name (4.1)
1098 ------------------------
1099 -- 4.1 Selector Name --
1100 ------------------------
1102 -- Parsed by P_Name (4.1)
1104 ------------------------------
1105 -- 4.1 Attribute Reference --
1106 ------------------------------
1108 -- Parsed by P_Name (4.1)
1110 -------------------------------
1111 -- 4.1 Attribute Designator --
1112 -------------------------------
1114 -- Parsed by P_Name (4.1)
1116 --------------------------------------
1117 -- 4.1.4 Range Attribute Reference --
1118 --------------------------------------
1120 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
1122 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
1124 -- In the grammar, a RANGE attribute is simply a name, but its use is
1125 -- highly restricted, so in the parser, we do not regard it as a name.
1126 -- Instead, P_Name returns without scanning the 'RANGE part of the
1127 -- attribute, and the caller uses the following function to construct
1128 -- a range attribute in places where it is appropriate.
1130 -- Note that RANGE here is treated essentially as an identifier,
1131 -- rather than a reserved word.
1133 -- The caller has parsed the prefix, i.e. a name, and Token points to
1134 -- the apostrophe. The token after the apostrophe is known to be RANGE
1135 -- at this point. The prefix node becomes the prefix of the attribute.
1137 -- Error_Recovery: Cannot raise Error_Resync
1139 function P_Range_Attribute_Reference
1140 (Prefix_Node : Node_Id)
1141 return Node_Id
1143 Attr_Node : Node_Id;
1145 begin
1146 Attr_Node := New_Node (N_Attribute_Reference, Token_Ptr);
1147 Set_Prefix (Attr_Node, Prefix_Node);
1148 Scan; -- past apostrophe
1150 if Style_Check then
1151 Style.Check_Attribute_Name (True);
1152 end if;
1154 Set_Attribute_Name (Attr_Node, Name_Range);
1155 Scan; -- past RANGE
1157 if Token = Tok_Left_Paren then
1158 Scan; -- past left paren
1159 Set_Expressions (Attr_Node, New_List (P_Expression_If_OK));
1160 T_Right_Paren;
1161 end if;
1163 return Attr_Node;
1164 end P_Range_Attribute_Reference;
1166 ---------------------------------------
1167 -- 4.1.4 Range Attribute Designator --
1168 ---------------------------------------
1170 -- Parsed by P_Range_Attribute_Reference (4.4)
1172 --------------------
1173 -- 4.3 Aggregate --
1174 --------------------
1176 -- AGGREGATE ::= RECORD_AGGREGATE | EXTENSION_AGGREGATE | ARRAY_AGGREGATE
1178 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3), except in the case where
1179 -- an aggregate is known to be required (code statement, extension
1180 -- aggregate), in which cases this routine performs the necessary check
1181 -- that we have an aggregate rather than a parenthesized expression
1183 -- Error recovery: can raise Error_Resync
1185 function P_Aggregate return Node_Id is
1186 Aggr_Sloc : constant Source_Ptr := Token_Ptr;
1187 Aggr_Node : constant Node_Id := P_Aggregate_Or_Paren_Expr;
1189 begin
1190 if Nkind (Aggr_Node) /= N_Aggregate
1191 and then
1192 Nkind (Aggr_Node) /= N_Extension_Aggregate
1193 then
1194 Error_Msg
1195 ("aggregate may not have single positional component", Aggr_Sloc);
1196 return Error;
1197 else
1198 return Aggr_Node;
1199 end if;
1200 end P_Aggregate;
1202 ------------------------------------------------
1203 -- 4.3 Aggregate or Parenthesized Expression --
1204 ------------------------------------------------
1206 -- This procedure parses out either an aggregate or a parenthesized
1207 -- expression (these two constructs are closely related, since a
1208 -- parenthesized expression looks like an aggregate with a single
1209 -- positional component).
1211 -- AGGREGATE ::=
1212 -- RECORD_AGGREGATE | EXTENSION_AGGREGATE | ARRAY_AGGREGATE
1214 -- RECORD_AGGREGATE ::= (RECORD_COMPONENT_ASSOCIATION_LIST)
1216 -- RECORD_COMPONENT_ASSOCIATION_LIST ::=
1217 -- RECORD_COMPONENT_ASSOCIATION {, RECORD_COMPONENT_ASSOCIATION}
1218 -- | null record
1220 -- RECORD_COMPONENT_ASSOCIATION ::=
1221 -- [COMPONENT_CHOICE_LIST =>] EXPRESSION
1223 -- COMPONENT_CHOICE_LIST ::=
1224 -- component_SELECTOR_NAME {| component_SELECTOR_NAME}
1225 -- | others
1227 -- EXTENSION_AGGREGATE ::=
1228 -- (ANCESTOR_PART with RECORD_COMPONENT_ASSOCIATION_LIST)
1230 -- ANCESTOR_PART ::= EXPRESSION | SUBTYPE_MARK
1232 -- ARRAY_AGGREGATE ::=
1233 -- POSITIONAL_ARRAY_AGGREGATE | NAMED_ARRAY_AGGREGATE
1235 -- POSITIONAL_ARRAY_AGGREGATE ::=
1236 -- (EXPRESSION, EXPRESSION {, EXPRESSION})
1237 -- | (EXPRESSION {, EXPRESSION}, others => EXPRESSION)
1238 -- | (EXPRESSION {, EXPRESSION}, others => <>)
1240 -- NAMED_ARRAY_AGGREGATE ::=
1241 -- (ARRAY_COMPONENT_ASSOCIATION {, ARRAY_COMPONENT_ASSOCIATION})
1243 -- PRIMARY ::= (EXPRESSION);
1245 -- Error recovery: can raise Error_Resync
1247 -- Note: POSITIONAL_ARRAY_AGGREGATE rule has been extended to give support
1248 -- to Ada 2005 limited aggregates (AI-287)
1250 function P_Aggregate_Or_Paren_Expr return Node_Id is
1251 Aggregate_Node : Node_Id;
1252 Expr_List : List_Id;
1253 Assoc_List : List_Id;
1254 Expr_Node : Node_Id;
1255 Lparen_Sloc : Source_Ptr;
1256 Scan_State : Saved_Scan_State;
1258 procedure Box_Error;
1259 -- Called if <> is encountered as positional aggregate element. Issues
1260 -- error message and sets Expr_Node to Error.
1262 ---------------
1263 -- Box_Error --
1264 ---------------
1266 procedure Box_Error is
1267 begin
1268 if Ada_Version < Ada_2005 then
1269 Error_Msg_SC ("box in aggregate is an Ada 2005 extension");
1270 end if;
1272 -- Ada 2005 (AI-287): The box notation is allowed only with named
1273 -- notation because positional notation might be error prone. For
1274 -- example, in "(X, <>, Y, <>)", there is no type associated with
1275 -- the boxes, so you might not be leaving out the components you
1276 -- thought you were leaving out.
1278 Error_Msg_SC ("(Ada 2005) box only allowed with named notation");
1279 Scan; -- past box
1280 Expr_Node := Error;
1281 end Box_Error;
1283 -- Start of processing for P_Aggregate_Or_Paren_Expr
1285 begin
1286 Lparen_Sloc := Token_Ptr;
1287 T_Left_Paren;
1289 -- Note on parentheses count. For cases like an if expression, the
1290 -- parens here really count as real parentheses for the paren count,
1291 -- so we adjust the paren count accordingly after scanning the expr.
1293 -- If expression
1295 if Token = Tok_If then
1296 Expr_Node := P_If_Expression;
1297 T_Right_Paren;
1298 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1299 return Expr_Node;
1301 -- Case expression
1303 elsif Token = Tok_Case then
1304 Expr_Node := P_Case_Expression;
1305 T_Right_Paren;
1306 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1307 return Expr_Node;
1309 -- Quantified expression
1311 elsif Token = Tok_For then
1312 Expr_Node := P_Quantified_Expression;
1313 T_Right_Paren;
1314 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1315 return Expr_Node;
1317 -- Note: the mechanism used here of rescanning the initial expression
1318 -- is distinctly unpleasant, but it saves a lot of fiddling in scanning
1319 -- out the discrete choice list.
1321 -- Deal with expression and extension aggregates first
1323 elsif Token /= Tok_Others then
1324 Save_Scan_State (Scan_State); -- at start of expression
1326 -- Deal with (NULL RECORD)
1328 if Token = Tok_Null then
1329 Scan; -- past NULL
1331 if Token = Tok_Record then
1332 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1333 Set_Null_Record_Present (Aggregate_Node, True);
1334 Scan; -- past RECORD
1335 T_Right_Paren;
1336 return Aggregate_Node;
1337 else
1338 Restore_Scan_State (Scan_State); -- to NULL that must be expr
1339 end if;
1340 end if;
1342 -- Scan expression, handling box appearing as positional argument
1344 if Token = Tok_Box then
1345 Box_Error;
1346 else
1347 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
1348 end if;
1350 -- Extension aggregate
1352 if Token = Tok_With then
1353 if Nkind (Expr_Node) = N_Attribute_Reference
1354 and then Attribute_Name (Expr_Node) = Name_Range
1355 then
1356 Bad_Range_Attribute (Sloc (Expr_Node));
1357 return Error;
1358 end if;
1360 if Ada_Version = Ada_83 then
1361 Error_Msg_SC ("(Ada 83) extension aggregate not allowed");
1362 end if;
1364 Aggregate_Node := New_Node (N_Extension_Aggregate, Lparen_Sloc);
1365 Set_Ancestor_Part (Aggregate_Node, Expr_Node);
1366 Scan; -- past WITH
1368 -- Deal with WITH NULL RECORD case
1370 if Token = Tok_Null then
1371 Save_Scan_State (Scan_State); -- at NULL
1372 Scan; -- past NULL
1374 if Token = Tok_Record then
1375 Scan; -- past RECORD
1376 Set_Null_Record_Present (Aggregate_Node, True);
1377 T_Right_Paren;
1378 return Aggregate_Node;
1380 else
1381 Restore_Scan_State (Scan_State); -- to NULL that must be expr
1382 end if;
1383 end if;
1385 if Token /= Tok_Others then
1386 Save_Scan_State (Scan_State);
1387 Expr_Node := P_Expression;
1388 else
1389 Expr_Node := Empty;
1390 end if;
1392 -- Expression
1394 elsif Token = Tok_Right_Paren or else Token in Token_Class_Eterm then
1395 if Nkind (Expr_Node) = N_Attribute_Reference
1396 and then Attribute_Name (Expr_Node) = Name_Range
1397 then
1398 Error_Msg
1399 ("|parentheses not allowed for range attribute", Lparen_Sloc);
1400 Scan; -- past right paren
1401 return Expr_Node;
1402 end if;
1404 -- Bump paren count of expression
1406 if Expr_Node /= Error then
1407 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1408 end if;
1410 T_Right_Paren; -- past right paren (error message if none)
1411 return Expr_Node;
1413 -- Normal aggregate
1415 else
1416 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1417 end if;
1419 -- Others
1421 else
1422 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1423 Expr_Node := Empty;
1424 end if;
1426 -- Prepare to scan list of component associations
1428 Expr_List := No_List; -- don't set yet, maybe all named entries
1429 Assoc_List := No_List; -- don't set yet, maybe all positional entries
1431 -- This loop scans through component associations. On entry to the
1432 -- loop, an expression has been scanned at the start of the current
1433 -- association unless initial token was OTHERS, in which case
1434 -- Expr_Node is set to Empty.
1436 loop
1437 -- Deal with others association first. This is a named association
1439 if No (Expr_Node) then
1440 if No (Assoc_List) then
1441 Assoc_List := New_List;
1442 end if;
1444 Append (P_Record_Or_Array_Component_Association, Assoc_List);
1446 -- Improper use of WITH
1448 elsif Token = Tok_With then
1449 Error_Msg_SC ("WITH must be preceded by single expression in " &
1450 "extension aggregate");
1451 raise Error_Resync;
1453 -- Range attribute can only appear as part of a discrete choice list
1455 elsif Nkind (Expr_Node) = N_Attribute_Reference
1456 and then Attribute_Name (Expr_Node) = Name_Range
1457 and then Token /= Tok_Arrow
1458 and then Token /= Tok_Vertical_Bar
1459 then
1460 Bad_Range_Attribute (Sloc (Expr_Node));
1461 return Error;
1463 -- Assume positional case if comma, right paren, or literal or
1464 -- identifier or OTHERS follows (the latter cases are missing
1465 -- comma cases). Also assume positional if a semicolon follows,
1466 -- which can happen if there are missing parens
1468 elsif Token = Tok_Comma
1469 or else Token = Tok_Right_Paren
1470 or else Token = Tok_Others
1471 or else Token in Token_Class_Lit_Or_Name
1472 or else Token = Tok_Semicolon
1473 then
1474 if Present (Assoc_List) then
1475 Error_Msg_BC -- CODEFIX
1476 ("""='>"" expected (positional association cannot follow " &
1477 "named association)");
1478 end if;
1480 if No (Expr_List) then
1481 Expr_List := New_List;
1482 end if;
1484 Append (Expr_Node, Expr_List);
1486 -- Check for aggregate followed by left parent, maybe missing comma
1488 elsif Nkind (Expr_Node) = N_Aggregate
1489 and then Token = Tok_Left_Paren
1490 then
1491 T_Comma;
1493 if No (Expr_List) then
1494 Expr_List := New_List;
1495 end if;
1497 Append (Expr_Node, Expr_List);
1499 -- Anything else is assumed to be a named association
1501 else
1502 Restore_Scan_State (Scan_State); -- to start of expression
1504 if No (Assoc_List) then
1505 Assoc_List := New_List;
1506 end if;
1508 Append (P_Record_Or_Array_Component_Association, Assoc_List);
1509 end if;
1511 exit when not Comma_Present;
1513 -- If we are at an expression terminator, something is seriously
1514 -- wrong, so let's get out now, before we start eating up stuff
1515 -- that doesn't belong to us.
1517 if Token in Token_Class_Eterm then
1518 Error_Msg_AP
1519 ("expecting expression or component association");
1520 exit;
1521 end if;
1523 -- Deal with misused box
1525 if Token = Tok_Box then
1526 Box_Error;
1528 -- Otherwise initiate for reentry to top of loop by scanning an
1529 -- initial expression, unless the first token is OTHERS.
1531 elsif Token = Tok_Others then
1532 Expr_Node := Empty;
1534 else
1535 Save_Scan_State (Scan_State); -- at start of expression
1536 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
1538 end if;
1539 end loop;
1541 -- All component associations (positional and named) have been scanned
1543 T_Right_Paren;
1544 Set_Expressions (Aggregate_Node, Expr_List);
1545 Set_Component_Associations (Aggregate_Node, Assoc_List);
1546 return Aggregate_Node;
1547 end P_Aggregate_Or_Paren_Expr;
1549 ------------------------------------------------
1550 -- 4.3 Record or Array Component Association --
1551 ------------------------------------------------
1553 -- RECORD_COMPONENT_ASSOCIATION ::=
1554 -- [COMPONENT_CHOICE_LIST =>] EXPRESSION
1555 -- | COMPONENT_CHOICE_LIST => <>
1557 -- COMPONENT_CHOICE_LIST =>
1558 -- component_SELECTOR_NAME {| component_SELECTOR_NAME}
1559 -- | others
1561 -- ARRAY_COMPONENT_ASSOCIATION ::=
1562 -- DISCRETE_CHOICE_LIST => EXPRESSION
1563 -- | DISCRETE_CHOICE_LIST => <>
1565 -- Note: this routine only handles the named cases, including others.
1566 -- Cases where the component choice list is not present have already
1567 -- been handled directly.
1569 -- Error recovery: can raise Error_Resync
1571 -- Note: RECORD_COMPONENT_ASSOCIATION and ARRAY_COMPONENT_ASSOCIATION
1572 -- rules have been extended to give support to Ada 2005 limited
1573 -- aggregates (AI-287)
1575 function P_Record_Or_Array_Component_Association return Node_Id is
1576 Assoc_Node : Node_Id;
1578 begin
1579 Assoc_Node := New_Node (N_Component_Association, Token_Ptr);
1580 Set_Choices (Assoc_Node, P_Discrete_Choice_List);
1581 Set_Sloc (Assoc_Node, Token_Ptr);
1582 TF_Arrow;
1584 if Token = Tok_Box then
1586 -- Ada 2005(AI-287): The box notation is used to indicate the
1587 -- default initialization of aggregate components
1589 if Ada_Version < Ada_2005 then
1590 Error_Msg_SP
1591 ("component association with '<'> is an Ada 2005 extension");
1592 Error_Msg_SP ("\unit must be compiled with -gnat05 switch");
1593 end if;
1595 Set_Box_Present (Assoc_Node);
1596 Scan; -- Past box
1597 else
1598 Set_Expression (Assoc_Node, P_Expression);
1599 end if;
1601 return Assoc_Node;
1602 end P_Record_Or_Array_Component_Association;
1604 -----------------------------
1605 -- 4.3.1 Record Aggregate --
1606 -----------------------------
1608 -- Case of enumeration aggregate is parsed by P_Aggregate (4.3)
1609 -- All other cases are parsed by P_Aggregate_Or_Paren_Expr (4.3)
1611 ----------------------------------------------
1612 -- 4.3.1 Record Component Association List --
1613 ----------------------------------------------
1615 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1617 ----------------------------------
1618 -- 4.3.1 Component Choice List --
1619 ----------------------------------
1621 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1623 --------------------------------
1624 -- 4.3.1 Extension Aggregate --
1625 --------------------------------
1627 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1629 --------------------------
1630 -- 4.3.1 Ancestor Part --
1631 --------------------------
1633 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1635 ----------------------------
1636 -- 4.3.1 Array Aggregate --
1637 ----------------------------
1639 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1641 ---------------------------------------
1642 -- 4.3.1 Positional Array Aggregate --
1643 ---------------------------------------
1645 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1647 ----------------------------------
1648 -- 4.3.1 Named Array Aggregate --
1649 ----------------------------------
1651 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1653 ----------------------------------------
1654 -- 4.3.1 Array Component Association --
1655 ----------------------------------------
1657 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1659 ---------------------
1660 -- 4.4 Expression --
1661 ---------------------
1663 -- This procedure parses EXPRESSION or CHOICE_EXPRESSION
1665 -- EXPRESSION ::=
1666 -- RELATION {LOGICAL_OPERATOR RELATION}
1668 -- CHOICE_EXPRESSION ::=
1669 -- CHOICE_RELATION {LOGICAL_OPERATOR CHOICE_RELATION}
1671 -- LOGICAL_OPERATOR ::= and | and then | or | or else | xor
1673 -- On return, Expr_Form indicates the categorization of the expression
1674 -- EF_Range_Attr is not a possible value (if a range attribute is found,
1675 -- an error message is given, and Error is returned).
1677 -- Error recovery: cannot raise Error_Resync
1679 function P_Expression return Node_Id is
1680 Logical_Op : Node_Kind;
1681 Prev_Logical_Op : Node_Kind;
1682 Op_Location : Source_Ptr;
1683 Node1 : Node_Id;
1684 Node2 : Node_Id;
1686 begin
1687 Node1 := P_Relation;
1689 if Token in Token_Class_Logop then
1690 Prev_Logical_Op := N_Empty;
1692 loop
1693 Op_Location := Token_Ptr;
1694 Logical_Op := P_Logical_Operator;
1696 if Prev_Logical_Op /= N_Empty and then
1697 Logical_Op /= Prev_Logical_Op
1698 then
1699 Error_Msg
1700 ("mixed logical operators in expression", Op_Location);
1701 Prev_Logical_Op := N_Empty;
1702 else
1703 Prev_Logical_Op := Logical_Op;
1704 end if;
1706 Node2 := Node1;
1707 Node1 := New_Op_Node (Logical_Op, Op_Location);
1708 Set_Left_Opnd (Node1, Node2);
1709 Set_Right_Opnd (Node1, P_Relation);
1710 exit when Token not in Token_Class_Logop;
1711 end loop;
1713 Expr_Form := EF_Non_Simple;
1714 end if;
1716 if Token = Tok_Apostrophe then
1717 Bad_Range_Attribute (Token_Ptr);
1718 return Error;
1719 else
1720 return Node1;
1721 end if;
1722 end P_Expression;
1724 -- This function is identical to the normal P_Expression, except that it
1725 -- also permits the appearance of a case, conditional, or quantified
1726 -- expression if the call immediately follows a left paren, and followed
1727 -- by a right parenthesis. These forms are allowed if these conditions
1728 -- are not met, but an error message will be issued.
1730 function P_Expression_If_OK return Node_Id is
1731 begin
1732 -- Case of conditional, case or quantified expression
1734 if Token = Tok_Case or else Token = Tok_If or else Token = Tok_For then
1735 return P_Unparen_Cond_Case_Quant_Expression;
1737 -- Normal case, not case/conditional/quantified expression
1739 else
1740 return P_Expression;
1741 end if;
1742 end P_Expression_If_OK;
1744 -- This function is identical to the normal P_Expression, except that it
1745 -- checks that the expression scan did not stop on a right paren. It is
1746 -- called in all contexts where a right parenthesis cannot legitimately
1747 -- follow an expression.
1749 -- Error recovery: can not raise Error_Resync
1751 function P_Expression_No_Right_Paren return Node_Id is
1752 Expr : constant Node_Id := P_Expression;
1753 begin
1754 Ignore (Tok_Right_Paren);
1755 return Expr;
1756 end P_Expression_No_Right_Paren;
1758 ----------------------------------------
1759 -- 4.4 Expression_Or_Range_Attribute --
1760 ----------------------------------------
1762 -- EXPRESSION ::=
1763 -- RELATION {and RELATION} | RELATION {and then RELATION}
1764 -- | RELATION {or RELATION} | RELATION {or else RELATION}
1765 -- | RELATION {xor RELATION}
1767 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
1769 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
1771 -- On return, Expr_Form indicates the categorization of the expression
1772 -- and EF_Range_Attr is one of the possibilities.
1774 -- Error recovery: cannot raise Error_Resync
1776 -- In the grammar, a RANGE attribute is simply a name, but its use is
1777 -- highly restricted, so in the parser, we do not regard it as a name.
1778 -- Instead, P_Name returns without scanning the 'RANGE part of the
1779 -- attribute, and P_Expression_Or_Range_Attribute handles the range
1780 -- attribute reference. In the normal case where a range attribute is
1781 -- not allowed, an error message is issued by P_Expression.
1783 function P_Expression_Or_Range_Attribute return Node_Id is
1784 Logical_Op : Node_Kind;
1785 Prev_Logical_Op : Node_Kind;
1786 Op_Location : Source_Ptr;
1787 Node1 : Node_Id;
1788 Node2 : Node_Id;
1789 Attr_Node : Node_Id;
1791 begin
1792 Node1 := P_Relation;
1794 if Token = Tok_Apostrophe then
1795 Attr_Node := P_Range_Attribute_Reference (Node1);
1796 Expr_Form := EF_Range_Attr;
1797 return Attr_Node;
1799 elsif Token in Token_Class_Logop then
1800 Prev_Logical_Op := N_Empty;
1802 loop
1803 Op_Location := Token_Ptr;
1804 Logical_Op := P_Logical_Operator;
1806 if Prev_Logical_Op /= N_Empty and then
1807 Logical_Op /= Prev_Logical_Op
1808 then
1809 Error_Msg
1810 ("mixed logical operators in expression", Op_Location);
1811 Prev_Logical_Op := N_Empty;
1812 else
1813 Prev_Logical_Op := Logical_Op;
1814 end if;
1816 Node2 := Node1;
1817 Node1 := New_Op_Node (Logical_Op, Op_Location);
1818 Set_Left_Opnd (Node1, Node2);
1819 Set_Right_Opnd (Node1, P_Relation);
1820 exit when Token not in Token_Class_Logop;
1821 end loop;
1823 Expr_Form := EF_Non_Simple;
1824 end if;
1826 if Token = Tok_Apostrophe then
1827 Bad_Range_Attribute (Token_Ptr);
1828 return Error;
1829 else
1830 return Node1;
1831 end if;
1832 end P_Expression_Or_Range_Attribute;
1834 -- Version that allows a non-parenthesized case, conditional, or quantified
1835 -- expression if the call immediately follows a left paren, and followed
1836 -- by a right parenthesis. These forms are allowed if these conditions
1837 -- are not met, but an error message will be issued.
1839 function P_Expression_Or_Range_Attribute_If_OK return Node_Id is
1840 begin
1841 -- Case of conditional, case or quantified expression
1843 if Token = Tok_Case or else Token = Tok_If or else Token = Tok_For then
1844 return P_Unparen_Cond_Case_Quant_Expression;
1846 -- Normal case, not one of the above expression types
1848 else
1849 return P_Expression_Or_Range_Attribute;
1850 end if;
1851 end P_Expression_Or_Range_Attribute_If_OK;
1853 -------------------
1854 -- 4.4 Relation --
1855 -------------------
1857 -- This procedure scans both relations and choice relations
1859 -- CHOICE_RELATION ::=
1860 -- SIMPLE_EXPRESSION [RELATIONAL_OPERATOR SIMPLE_EXPRESSION]
1862 -- RELATION ::=
1863 -- SIMPLE_EXPRESSION [not] in MEMBERSHIP_CHOICE_LIST
1864 -- | RAISE_EXPRESSION
1866 -- MEMBERSHIP_CHOICE_LIST ::=
1867 -- MEMBERSHIP_CHOICE {'|' MEMBERSHIP CHOICE}
1869 -- MEMBERSHIP_CHOICE ::=
1870 -- CHOICE_EXPRESSION | RANGE | SUBTYPE_MARK
1872 -- RAISE_EXPRESSION ::= raise exception_NAME [with string_EXPRESSION]
1874 -- On return, Expr_Form indicates the categorization of the expression
1876 -- Note: if Token = Tok_Apostrophe on return, then Expr_Form is set to
1877 -- EF_Simple_Name and the following token is RANGE (range attribute case).
1879 -- Error recovery: cannot raise Error_Resync. If an error occurs within an
1880 -- expression, then tokens are scanned until either a non-expression token,
1881 -- a right paren (not matched by a left paren) or a comma, is encountered.
1883 function P_Relation return Node_Id is
1884 Node1, Node2 : Node_Id;
1885 Optok : Source_Ptr;
1887 begin
1888 -- First check for raise expression
1890 if Token = Tok_Raise then
1891 Expr_Form := EF_Non_Simple;
1892 return P_Raise_Expression;
1893 end if;
1895 -- All other cases
1897 Node1 := P_Simple_Expression;
1899 if Token not in Token_Class_Relop then
1900 return Node1;
1902 else
1903 -- Here we have a relational operator following. If so then scan it
1904 -- out. Note that the assignment symbol := is treated as a relational
1905 -- operator to improve the error recovery when it is misused for =.
1906 -- P_Relational_Operator also parses the IN and NOT IN operations.
1908 Optok := Token_Ptr;
1909 Node2 := New_Op_Node (P_Relational_Operator, Optok);
1910 Set_Left_Opnd (Node2, Node1);
1912 -- Case of IN or NOT IN
1914 if Prev_Token = Tok_In then
1915 P_Membership_Test (Node2);
1917 -- Case of relational operator (= /= < <= > >=)
1919 else
1920 Set_Right_Opnd (Node2, P_Simple_Expression);
1921 end if;
1923 Expr_Form := EF_Non_Simple;
1925 if Token in Token_Class_Relop then
1926 Error_Msg_SC ("unexpected relational operator");
1927 raise Error_Resync;
1928 end if;
1930 return Node2;
1931 end if;
1933 -- If any error occurs, then scan to the next expression terminator symbol
1934 -- or comma or right paren at the outer (i.e. current) parentheses level.
1935 -- The flags are set to indicate a normal simple expression.
1937 exception
1938 when Error_Resync =>
1939 Resync_Expression;
1940 Expr_Form := EF_Simple;
1941 return Error;
1942 end P_Relation;
1944 ----------------------------
1945 -- 4.4 Simple Expression --
1946 ----------------------------
1948 -- SIMPLE_EXPRESSION ::=
1949 -- [UNARY_ADDING_OPERATOR] TERM {BINARY_ADDING_OPERATOR TERM}
1951 -- On return, Expr_Form indicates the categorization of the expression
1953 -- Note: if Token = Tok_Apostrophe on return, then Expr_Form is set to
1954 -- EF_Simple_Name and the following token is RANGE (range attribute case).
1956 -- Error recovery: cannot raise Error_Resync. If an error occurs within an
1957 -- expression, then tokens are scanned until either a non-expression token,
1958 -- a right paren (not matched by a left paren) or a comma, is encountered.
1960 -- Note: P_Simple_Expression is called only internally by higher level
1961 -- expression routines. In cases in the grammar where a simple expression
1962 -- is required, the approach is to scan an expression, and then post an
1963 -- appropriate error message if the expression obtained is not simple. This
1964 -- gives better error recovery and treatment.
1966 function P_Simple_Expression return Node_Id is
1967 Scan_State : Saved_Scan_State;
1968 Node1 : Node_Id;
1969 Node2 : Node_Id;
1970 Tokptr : Source_Ptr;
1972 begin
1973 -- Check for cases starting with a name. There are two reasons for
1974 -- special casing. First speed things up by catching a common case
1975 -- without going through several routine layers. Second the caller must
1976 -- be informed via Expr_Form when the simple expression is a name.
1978 if Token in Token_Class_Name then
1979 Node1 := P_Name;
1981 -- Deal with apostrophe cases
1983 if Token = Tok_Apostrophe then
1984 Save_Scan_State (Scan_State); -- at apostrophe
1985 Scan; -- past apostrophe
1987 -- If qualified expression, scan it out and fall through
1989 if Token = Tok_Left_Paren then
1990 Node1 := P_Qualified_Expression (Node1);
1991 Expr_Form := EF_Simple;
1993 -- If range attribute, then we return with Token pointing to the
1994 -- apostrophe. Note: avoid the normal error check on exit. We
1995 -- know that the expression really is complete in this case.
1997 else -- Token = Tok_Range then
1998 Restore_Scan_State (Scan_State); -- to apostrophe
1999 Expr_Form := EF_Simple_Name;
2000 return Node1;
2001 end if;
2002 end if;
2004 -- If an expression terminator follows, the previous processing
2005 -- completely scanned out the expression (a common case), and
2006 -- left Expr_Form set appropriately for returning to our caller.
2008 if Token in Token_Class_Sterm then
2009 null;
2011 -- If we do not have an expression terminator, then complete the
2012 -- scan of a simple expression. This code duplicates the code
2013 -- found in P_Term and P_Factor.
2015 else
2016 if Token = Tok_Double_Asterisk then
2017 if Style_Check then
2018 Style.Check_Exponentiation_Operator;
2019 end if;
2021 Node2 := New_Op_Node (N_Op_Expon, Token_Ptr);
2022 Scan; -- past **
2023 Set_Left_Opnd (Node2, Node1);
2024 Set_Right_Opnd (Node2, P_Primary);
2025 Check_Bad_Exp;
2026 Node1 := Node2;
2027 end if;
2029 loop
2030 exit when Token not in Token_Class_Mulop;
2031 Tokptr := Token_Ptr;
2032 Node2 := New_Op_Node (P_Multiplying_Operator, Tokptr);
2034 if Style_Check then
2035 Style.Check_Binary_Operator;
2036 end if;
2038 Scan; -- past operator
2039 Set_Left_Opnd (Node2, Node1);
2040 Set_Right_Opnd (Node2, P_Factor);
2041 Node1 := Node2;
2042 end loop;
2044 loop
2045 exit when Token not in Token_Class_Binary_Addop;
2046 Tokptr := Token_Ptr;
2047 Node2 := New_Op_Node (P_Binary_Adding_Operator, Tokptr);
2049 if Style_Check then
2050 Style.Check_Binary_Operator;
2051 end if;
2053 Scan; -- past operator
2054 Set_Left_Opnd (Node2, Node1);
2055 Set_Right_Opnd (Node2, P_Term);
2056 Node1 := Node2;
2057 end loop;
2059 Expr_Form := EF_Simple;
2060 end if;
2062 -- Cases where simple expression does not start with a name
2064 else
2065 -- Scan initial sign and initial Term
2067 if Token in Token_Class_Unary_Addop then
2068 Tokptr := Token_Ptr;
2069 Node1 := New_Op_Node (P_Unary_Adding_Operator, Tokptr);
2071 if Style_Check then
2072 Style.Check_Unary_Plus_Or_Minus;
2073 end if;
2075 Scan; -- past operator
2076 Set_Right_Opnd (Node1, P_Term);
2077 else
2078 Node1 := P_Term;
2079 end if;
2081 -- In the following, we special-case a sequence of concatenations of
2082 -- string literals, such as "aaa" & "bbb" & ... & "ccc", with nothing
2083 -- else mixed in. For such a sequence, we return a tree representing
2084 -- "" & "aaabbb...ccc" (a single concatenation). This is done only if
2085 -- the number of concatenations is large. If semantic analysis
2086 -- resolves the "&" to a predefined one, then this folding gives the
2087 -- right answer. Otherwise, semantic analysis will complain about a
2088 -- capacity-exceeded error. The purpose of this trick is to avoid
2089 -- creating a deeply nested tree, which would cause deep recursion
2090 -- during semantics, causing stack overflow. This way, we can handle
2091 -- enormous concatenations in the normal case of predefined "&". We
2092 -- first build up the normal tree, and then rewrite it if
2093 -- appropriate.
2095 declare
2096 Num_Concats_Threshold : constant Positive := 1000;
2097 -- Arbitrary threshold value to enable optimization
2099 First_Node : constant Node_Id := Node1;
2100 Is_Strlit_Concat : Boolean;
2101 -- True iff we've parsed a sequence of concatenations of string
2102 -- literals, with nothing else mixed in.
2104 Num_Concats : Natural;
2105 -- Number of "&" operators if Is_Strlit_Concat is True
2107 begin
2108 Is_Strlit_Concat :=
2109 Nkind (Node1) = N_String_Literal
2110 and then Token = Tok_Ampersand;
2111 Num_Concats := 0;
2113 -- Scan out sequence of terms separated by binary adding operators
2115 loop
2116 exit when Token not in Token_Class_Binary_Addop;
2117 Tokptr := Token_Ptr;
2118 Node2 := New_Op_Node (P_Binary_Adding_Operator, Tokptr);
2119 Scan; -- past operator
2120 Set_Left_Opnd (Node2, Node1);
2121 Node1 := P_Term;
2122 Set_Right_Opnd (Node2, Node1);
2124 -- Check if we're still concatenating string literals
2126 Is_Strlit_Concat :=
2127 Is_Strlit_Concat
2128 and then Nkind (Node2) = N_Op_Concat
2129 and then Nkind (Node1) = N_String_Literal;
2131 if Is_Strlit_Concat then
2132 Num_Concats := Num_Concats + 1;
2133 end if;
2135 Node1 := Node2;
2136 end loop;
2138 -- If we have an enormous series of concatenations of string
2139 -- literals, rewrite as explained above. The Is_Folded_In_Parser
2140 -- flag tells semantic analysis that if the "&" is not predefined,
2141 -- the folded value is wrong.
2143 if Is_Strlit_Concat
2144 and then Num_Concats >= Num_Concats_Threshold
2145 then
2146 declare
2147 Empty_String_Val : String_Id;
2148 -- String_Id for ""
2150 Strlit_Concat_Val : String_Id;
2151 -- Contains the folded value (which will be correct if the
2152 -- "&" operators are the predefined ones).
2154 Cur_Node : Node_Id;
2155 -- For walking up the tree
2157 New_Node : Node_Id;
2158 -- Folded node to replace Node1
2160 Loc : constant Source_Ptr := Sloc (First_Node);
2162 begin
2163 -- Walk up the tree starting at the leftmost string literal
2164 -- (First_Node), building up the Strlit_Concat_Val as we
2165 -- go. Note that we do not use recursion here -- the whole
2166 -- point is to avoid recursively walking that enormous tree.
2168 Start_String;
2169 Store_String_Chars (Strval (First_Node));
2171 Cur_Node := Parent (First_Node);
2172 while Present (Cur_Node) loop
2173 pragma Assert (Nkind (Cur_Node) = N_Op_Concat and then
2174 Nkind (Right_Opnd (Cur_Node)) = N_String_Literal);
2176 Store_String_Chars (Strval (Right_Opnd (Cur_Node)));
2177 Cur_Node := Parent (Cur_Node);
2178 end loop;
2180 Strlit_Concat_Val := End_String;
2182 -- Create new folded node, and rewrite result with a concat-
2183 -- enation of an empty string literal and the folded node.
2185 Start_String;
2186 Empty_String_Val := End_String;
2187 New_Node :=
2188 Make_Op_Concat (Loc,
2189 Make_String_Literal (Loc, Empty_String_Val),
2190 Make_String_Literal (Loc, Strlit_Concat_Val,
2191 Is_Folded_In_Parser => True));
2192 Rewrite (Node1, New_Node);
2193 end;
2194 end if;
2195 end;
2197 -- All done, we clearly do not have name or numeric literal so this
2198 -- is a case of a simple expression which is some other possibility.
2200 Expr_Form := EF_Simple;
2201 end if;
2203 -- Come here at end of simple expression, where we do a couple of
2204 -- special checks to improve error recovery.
2206 -- Special test to improve error recovery. If the current token
2207 -- is a period, then someone is trying to do selection on something
2208 -- that is not a name, e.g. a qualified expression.
2210 if Token = Tok_Dot then
2211 Error_Msg_SC ("prefix for selection is not a name");
2213 -- If qualified expression, comment and continue, otherwise something
2214 -- is pretty nasty so do an Error_Resync call.
2216 if Ada_Version < Ada_2012
2217 and then Nkind (Node1) = N_Qualified_Expression
2218 then
2219 Error_Msg_SC ("\would be legal in Ada 2012 mode");
2220 else
2221 raise Error_Resync;
2222 end if;
2223 end if;
2225 -- Special test to improve error recovery: If the current token is
2226 -- not the first token on a line (as determined by checking the
2227 -- previous token position with the start of the current line),
2228 -- then we insist that we have an appropriate terminating token.
2229 -- Consider the following two examples:
2231 -- 1) if A nad B then ...
2233 -- 2) A := B
2234 -- C := D
2236 -- In the first example, we would like to issue a binary operator
2237 -- expected message and resynchronize to the then. In the second
2238 -- example, we do not want to issue a binary operator message, so
2239 -- that instead we will get the missing semicolon message. This
2240 -- distinction is of course a heuristic which does not always work,
2241 -- but in practice it is quite effective.
2243 -- Note: the one case in which we do not go through this circuit is
2244 -- when we have scanned a range attribute and want to return with
2245 -- Token pointing to the apostrophe. The apostrophe is not normally
2246 -- an expression terminator, and is not in Token_Class_Sterm, but
2247 -- in this special case we know that the expression is complete.
2249 if not Token_Is_At_Start_Of_Line
2250 and then Token not in Token_Class_Sterm
2251 then
2252 -- Normally the right error message is indeed that we expected a
2253 -- binary operator, but in the case of being between a right and left
2254 -- paren, e.g. in an aggregate, a more likely error is missing comma.
2256 if Prev_Token = Tok_Right_Paren and then Token = Tok_Left_Paren then
2257 T_Comma;
2258 else
2259 Error_Msg_AP ("binary operator expected");
2260 end if;
2262 raise Error_Resync;
2264 else
2265 return Node1;
2266 end if;
2268 -- If any error occurs, then scan to next expression terminator symbol
2269 -- or comma, right paren or vertical bar at the outer (i.e. current) paren
2270 -- level. Expr_Form is set to indicate a normal simple expression.
2272 exception
2273 when Error_Resync =>
2274 Resync_Expression;
2275 Expr_Form := EF_Simple;
2276 return Error;
2277 end P_Simple_Expression;
2279 -----------------------------------------------
2280 -- 4.4 Simple Expression or Range Attribute --
2281 -----------------------------------------------
2283 -- SIMPLE_EXPRESSION ::=
2284 -- [UNARY_ADDING_OPERATOR] TERM {BINARY_ADDING_OPERATOR TERM}
2286 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
2288 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
2290 -- Error recovery: cannot raise Error_Resync
2292 function P_Simple_Expression_Or_Range_Attribute return Node_Id is
2293 Sexpr : Node_Id;
2294 Attr_Node : Node_Id;
2296 begin
2297 -- We don't just want to roar ahead and call P_Simple_Expression
2298 -- here, since we want to handle the case of a parenthesized range
2299 -- attribute cleanly.
2301 if Token = Tok_Left_Paren then
2302 declare
2303 Lptr : constant Source_Ptr := Token_Ptr;
2304 Scan_State : Saved_Scan_State;
2306 begin
2307 Save_Scan_State (Scan_State);
2308 Scan; -- past left paren
2309 Sexpr := P_Simple_Expression;
2311 if Token = Tok_Apostrophe then
2312 Attr_Node := P_Range_Attribute_Reference (Sexpr);
2313 Expr_Form := EF_Range_Attr;
2315 if Token = Tok_Right_Paren then
2316 Scan; -- scan past right paren if present
2317 end if;
2319 Error_Msg ("parentheses not allowed for range attribute", Lptr);
2321 return Attr_Node;
2322 end if;
2324 Restore_Scan_State (Scan_State);
2325 end;
2326 end if;
2328 -- Here after dealing with parenthesized range attribute
2330 Sexpr := P_Simple_Expression;
2332 if Token = Tok_Apostrophe then
2333 Attr_Node := P_Range_Attribute_Reference (Sexpr);
2334 Expr_Form := EF_Range_Attr;
2335 return Attr_Node;
2337 else
2338 return Sexpr;
2339 end if;
2340 end P_Simple_Expression_Or_Range_Attribute;
2342 ---------------
2343 -- 4.4 Term --
2344 ---------------
2346 -- TERM ::= FACTOR {MULTIPLYING_OPERATOR FACTOR}
2348 -- Error recovery: can raise Error_Resync
2350 function P_Term return Node_Id is
2351 Node1, Node2 : Node_Id;
2352 Tokptr : Source_Ptr;
2354 begin
2355 Node1 := P_Factor;
2357 loop
2358 exit when Token not in Token_Class_Mulop;
2359 Tokptr := Token_Ptr;
2360 Node2 := New_Op_Node (P_Multiplying_Operator, Tokptr);
2361 Scan; -- past operator
2362 Set_Left_Opnd (Node2, Node1);
2363 Set_Right_Opnd (Node2, P_Factor);
2364 Node1 := Node2;
2365 end loop;
2367 return Node1;
2368 end P_Term;
2370 -----------------
2371 -- 4.4 Factor --
2372 -----------------
2374 -- FACTOR ::= PRIMARY [** PRIMARY] | abs PRIMARY | not PRIMARY
2376 -- Error recovery: can raise Error_Resync
2378 function P_Factor return Node_Id is
2379 Node1 : Node_Id;
2380 Node2 : Node_Id;
2382 begin
2383 if Token = Tok_Abs then
2384 Node1 := New_Op_Node (N_Op_Abs, Token_Ptr);
2386 if Style_Check then
2387 Style.Check_Abs_Not;
2388 end if;
2390 Scan; -- past ABS
2391 Set_Right_Opnd (Node1, P_Primary);
2392 return Node1;
2394 elsif Token = Tok_Not then
2395 Node1 := New_Op_Node (N_Op_Not, Token_Ptr);
2397 if Style_Check then
2398 Style.Check_Abs_Not;
2399 end if;
2401 Scan; -- past NOT
2402 Set_Right_Opnd (Node1, P_Primary);
2403 return Node1;
2405 else
2406 Node1 := P_Primary;
2408 if Token = Tok_Double_Asterisk then
2409 Node2 := New_Op_Node (N_Op_Expon, Token_Ptr);
2410 Scan; -- past **
2411 Set_Left_Opnd (Node2, Node1);
2412 Set_Right_Opnd (Node2, P_Primary);
2413 Check_Bad_Exp;
2414 return Node2;
2415 else
2416 return Node1;
2417 end if;
2418 end if;
2419 end P_Factor;
2421 ------------------
2422 -- 4.4 Primary --
2423 ------------------
2425 -- PRIMARY ::=
2426 -- NUMERIC_LITERAL | null
2427 -- | STRING_LITERAL | AGGREGATE
2428 -- | NAME | QUALIFIED_EXPRESSION
2429 -- | ALLOCATOR | (EXPRESSION) | QUANTIFIED_EXPRESSION
2431 -- Error recovery: can raise Error_Resync
2433 function P_Primary return Node_Id is
2434 Scan_State : Saved_Scan_State;
2435 Node1 : Node_Id;
2437 Lparen : constant Boolean := Prev_Token = Tok_Left_Paren;
2438 -- Remember if previous token is a left parenthesis. This is used to
2439 -- deal with checking whether IF/CASE/FOR expressions appearing as
2440 -- primaries require extra parenthesization.
2442 begin
2443 -- The loop runs more than once only if misplaced pragmas are found
2444 -- or if a misplaced unary minus is skipped.
2446 loop
2447 case Token is
2449 -- Name token can start a name, call or qualified expression, all
2450 -- of which are acceptable possibilities for primary. Note also
2451 -- that string literal is included in name (as operator symbol)
2452 -- and type conversion is included in name (as indexed component).
2454 when Tok_Char_Literal | Tok_Operator_Symbol | Tok_Identifier =>
2455 Node1 := P_Name;
2457 -- All done unless apostrophe follows
2459 if Token /= Tok_Apostrophe then
2460 return Node1;
2462 -- Apostrophe following means that we have either just parsed
2463 -- the subtype mark of a qualified expression, or the prefix
2464 -- or a range attribute.
2466 else -- Token = Tok_Apostrophe
2467 Save_Scan_State (Scan_State); -- at apostrophe
2468 Scan; -- past apostrophe
2470 -- If range attribute, then this is always an error, since
2471 -- the only legitimate case (where the scanned expression is
2472 -- a qualified simple name) is handled at the level of the
2473 -- Simple_Expression processing. This case corresponds to a
2474 -- usage such as 3 + A'Range, which is always illegal.
2476 if Token = Tok_Range then
2477 Restore_Scan_State (Scan_State); -- to apostrophe
2478 Bad_Range_Attribute (Token_Ptr);
2479 return Error;
2481 -- If left paren, then we have a qualified expression.
2482 -- Note that P_Name guarantees that in this case, where
2483 -- Token = Tok_Apostrophe on return, the only two possible
2484 -- tokens following the apostrophe are left paren and
2485 -- RANGE, so we know we have a left paren here.
2487 else -- Token = Tok_Left_Paren
2488 return P_Qualified_Expression (Node1);
2490 end if;
2491 end if;
2493 -- Numeric or string literal
2495 when Tok_Integer_Literal |
2496 Tok_Real_Literal |
2497 Tok_String_Literal =>
2499 Node1 := Token_Node;
2500 Scan; -- past number
2501 return Node1;
2503 -- Left paren, starts aggregate or parenthesized expression
2505 when Tok_Left_Paren =>
2506 declare
2507 Expr : constant Node_Id := P_Aggregate_Or_Paren_Expr;
2509 begin
2510 if Nkind (Expr) = N_Attribute_Reference
2511 and then Attribute_Name (Expr) = Name_Range
2512 then
2513 Bad_Range_Attribute (Sloc (Expr));
2514 end if;
2516 return Expr;
2517 end;
2519 -- Allocator
2521 when Tok_New =>
2522 return P_Allocator;
2524 -- Null
2526 when Tok_Null =>
2527 Scan; -- past NULL
2528 return New_Node (N_Null, Prev_Token_Ptr);
2530 -- Pragma, not allowed here, so just skip past it
2532 when Tok_Pragma =>
2533 P_Pragmas_Misplaced;
2535 -- Deal with IF (possible unparenthesized if expression)
2537 when Tok_If =>
2539 -- If this looks like a real if, defined as an IF appearing at
2540 -- the start of a new line, then we consider we have a missing
2541 -- operand. If in Ada 2012 and the IF is not properly indented
2542 -- for a statement, we prefer to issue a message about an ill-
2543 -- parenthesized if expression.
2545 if Token_Is_At_Start_Of_Line
2546 and then not
2547 (Ada_Version >= Ada_2012
2548 and then Style_Check_Indentation /= 0
2549 and then Start_Column rem Style_Check_Indentation /= 0)
2550 then
2551 Error_Msg_AP ("missing operand");
2552 return Error;
2554 -- If this looks like an if expression, then treat it that way
2555 -- with an error message if not explicitly surrounded by
2556 -- parentheses.
2558 elsif Ada_Version >= Ada_2012 then
2559 Node1 := P_If_Expression;
2561 if not (Lparen and then Token = Tok_Right_Paren) then
2562 Error_Msg
2563 ("if expression must be parenthesized", Sloc (Node1));
2564 end if;
2566 return Node1;
2568 -- Otherwise treat as misused identifier
2570 else
2571 return P_Identifier;
2572 end if;
2574 -- Deal with CASE (possible unparenthesized case expression)
2576 when Tok_Case =>
2578 -- If this looks like a real case, defined as a CASE appearing
2579 -- the start of a new line, then we consider we have a missing
2580 -- operand. If in Ada 2012 and the CASE is not properly
2581 -- indented for a statement, we prefer to issue a message about
2582 -- an ill-parenthesized case expression.
2584 if Token_Is_At_Start_Of_Line
2585 and then not
2586 (Ada_Version >= Ada_2012
2587 and then Style_Check_Indentation /= 0
2588 and then Start_Column rem Style_Check_Indentation /= 0)
2589 then
2590 Error_Msg_AP ("missing operand");
2591 return Error;
2593 -- If this looks like a case expression, then treat it that way
2594 -- with an error message if not within parentheses.
2596 elsif Ada_Version >= Ada_2012 then
2597 Node1 := P_Case_Expression;
2599 if not (Lparen and then Token = Tok_Right_Paren) then
2600 Error_Msg
2601 ("case expression must be parenthesized", Sloc (Node1));
2602 end if;
2604 return Node1;
2606 -- Otherwise treat as misused identifier
2608 else
2609 return P_Identifier;
2610 end if;
2612 -- For [all | some] indicates a quantified expression
2614 when Tok_For =>
2615 if Token_Is_At_Start_Of_Line then
2616 Error_Msg_AP ("misplaced loop");
2617 return Error;
2619 elsif Ada_Version >= Ada_2012 then
2620 Node1 := P_Quantified_Expression;
2622 if not (Lparen and then Token = Tok_Right_Paren) then
2623 Error_Msg
2624 ("quantified expression must be parenthesized",
2625 Sloc (Node1));
2626 end if;
2628 return Node1;
2630 -- Otherwise treat as misused identifier
2632 else
2633 return P_Identifier;
2634 end if;
2636 -- Minus may well be an improper attempt at a unary minus. Give
2637 -- a message, skip the minus and keep going.
2639 when Tok_Minus =>
2640 Error_Msg_SC ("parentheses required for unary minus");
2641 Scan; -- past minus
2643 -- Anything else is illegal as the first token of a primary, but
2644 -- we test for some common errors, to improve error messages.
2646 when others =>
2647 if Is_Reserved_Identifier then
2648 return P_Identifier;
2650 elsif Prev_Token = Tok_Comma then
2651 Error_Msg_SP -- CODEFIX
2652 ("|extra "","" ignored");
2653 raise Error_Resync;
2655 else
2656 Error_Msg_AP ("missing operand");
2657 raise Error_Resync;
2658 end if;
2660 end case;
2661 end loop;
2662 end P_Primary;
2664 -------------------------------
2665 -- 4.4 Quantified_Expression --
2666 -------------------------------
2668 -- QUANTIFIED_EXPRESSION ::=
2669 -- for QUANTIFIER LOOP_PARAMETER_SPECIFICATION => PREDICATE |
2670 -- for QUANTIFIER ITERATOR_SPECIFICATION => PREDICATE
2672 function P_Quantified_Expression return Node_Id is
2673 I_Spec : Node_Id;
2674 Node1 : Node_Id;
2676 begin
2677 Error_Msg_Ada_2012_Feature ("quantified expression", Token_Ptr);
2678 Scan; -- past FOR
2679 Node1 := New_Node (N_Quantified_Expression, Prev_Token_Ptr);
2681 if Token = Tok_All then
2682 Set_All_Present (Node1);
2683 elsif Token /= Tok_Some then
2684 Error_Msg_AP ("missing quantifier");
2685 raise Error_Resync;
2686 end if;
2688 Scan; -- past SOME
2689 I_Spec := P_Loop_Parameter_Specification;
2691 if Nkind (I_Spec) = N_Loop_Parameter_Specification then
2692 Set_Loop_Parameter_Specification (Node1, I_Spec);
2693 else
2694 Set_Iterator_Specification (Node1, I_Spec);
2695 end if;
2697 if Token = Tok_Arrow then
2698 Scan;
2699 Set_Condition (Node1, P_Expression);
2700 return Node1;
2701 else
2702 Error_Msg_AP ("missing arrow");
2703 raise Error_Resync;
2704 end if;
2705 end P_Quantified_Expression;
2707 ---------------------------
2708 -- 4.5 Logical Operator --
2709 ---------------------------
2711 -- LOGICAL_OPERATOR ::= and | or | xor
2713 -- Note: AND THEN and OR ELSE are also treated as logical operators
2714 -- by the parser (even though they are not operators semantically)
2716 -- The value returned is the appropriate Node_Kind code for the operator
2717 -- On return, Token points to the token following the scanned operator.
2719 -- The caller has checked that the first token is a legitimate logical
2720 -- operator token (i.e. is either XOR, AND, OR).
2722 -- Error recovery: cannot raise Error_Resync
2724 function P_Logical_Operator return Node_Kind is
2725 begin
2726 if Token = Tok_And then
2727 if Style_Check then
2728 Style.Check_Binary_Operator;
2729 end if;
2731 Scan; -- past AND
2733 if Token = Tok_Then then
2734 Scan; -- past THEN
2735 return N_And_Then;
2736 else
2737 return N_Op_And;
2738 end if;
2740 elsif Token = Tok_Or then
2741 if Style_Check then
2742 Style.Check_Binary_Operator;
2743 end if;
2745 Scan; -- past OR
2747 if Token = Tok_Else then
2748 Scan; -- past ELSE
2749 return N_Or_Else;
2750 else
2751 return N_Op_Or;
2752 end if;
2754 else -- Token = Tok_Xor
2755 if Style_Check then
2756 Style.Check_Binary_Operator;
2757 end if;
2759 Scan; -- past XOR
2760 return N_Op_Xor;
2761 end if;
2762 end P_Logical_Operator;
2764 ------------------------------
2765 -- 4.5 Relational Operator --
2766 ------------------------------
2768 -- RELATIONAL_OPERATOR ::= = | /= | < | <= | > | >=
2770 -- The value returned is the appropriate Node_Kind code for the operator.
2771 -- On return, Token points to the operator token, NOT past it.
2773 -- The caller has checked that the first token is a legitimate relational
2774 -- operator token (i.e. is one of the operator tokens listed above).
2776 -- Error recovery: cannot raise Error_Resync
2778 function P_Relational_Operator return Node_Kind is
2779 Op_Kind : Node_Kind;
2780 Relop_Node : constant array (Token_Class_Relop) of Node_Kind :=
2781 (Tok_Less => N_Op_Lt,
2782 Tok_Equal => N_Op_Eq,
2783 Tok_Greater => N_Op_Gt,
2784 Tok_Not_Equal => N_Op_Ne,
2785 Tok_Greater_Equal => N_Op_Ge,
2786 Tok_Less_Equal => N_Op_Le,
2787 Tok_In => N_In,
2788 Tok_Not => N_Not_In,
2789 Tok_Box => N_Op_Ne);
2791 begin
2792 if Token = Tok_Box then
2793 Error_Msg_SC -- CODEFIX
2794 ("|""'<'>"" should be ""/=""");
2795 end if;
2797 Op_Kind := Relop_Node (Token);
2799 if Style_Check then
2800 Style.Check_Binary_Operator;
2801 end if;
2803 Scan; -- past operator token
2805 -- Deal with NOT IN, if previous token was NOT, we must have IN now
2807 if Prev_Token = Tok_Not then
2809 -- Style check, for NOT IN, we require one space between NOT and IN
2811 if Style_Check and then Token = Tok_In then
2812 Style.Check_Not_In;
2813 end if;
2815 T_In;
2816 end if;
2818 return Op_Kind;
2819 end P_Relational_Operator;
2821 ---------------------------------
2822 -- 4.5 Binary Adding Operator --
2823 ---------------------------------
2825 -- BINARY_ADDING_OPERATOR ::= + | - | &
2827 -- The value returned is the appropriate Node_Kind code for the operator.
2828 -- On return, Token points to the operator token (NOT past it).
2830 -- The caller has checked that the first token is a legitimate adding
2831 -- operator token (i.e. is one of the operator tokens listed above).
2833 -- Error recovery: cannot raise Error_Resync
2835 function P_Binary_Adding_Operator return Node_Kind is
2836 Addop_Node : constant array (Token_Class_Binary_Addop) of Node_Kind :=
2837 (Tok_Ampersand => N_Op_Concat,
2838 Tok_Minus => N_Op_Subtract,
2839 Tok_Plus => N_Op_Add);
2840 begin
2841 return Addop_Node (Token);
2842 end P_Binary_Adding_Operator;
2844 --------------------------------
2845 -- 4.5 Unary Adding Operator --
2846 --------------------------------
2848 -- UNARY_ADDING_OPERATOR ::= + | -
2850 -- The value returned is the appropriate Node_Kind code for the operator.
2851 -- On return, Token points to the operator token (NOT past it).
2853 -- The caller has checked that the first token is a legitimate adding
2854 -- operator token (i.e. is one of the operator tokens listed above).
2856 -- Error recovery: cannot raise Error_Resync
2858 function P_Unary_Adding_Operator return Node_Kind is
2859 Addop_Node : constant array (Token_Class_Unary_Addop) of Node_Kind :=
2860 (Tok_Minus => N_Op_Minus,
2861 Tok_Plus => N_Op_Plus);
2862 begin
2863 return Addop_Node (Token);
2864 end P_Unary_Adding_Operator;
2866 -------------------------------
2867 -- 4.5 Multiplying Operator --
2868 -------------------------------
2870 -- MULTIPLYING_OPERATOR ::= * | / | mod | rem
2872 -- The value returned is the appropriate Node_Kind code for the operator.
2873 -- On return, Token points to the operator token (NOT past it).
2875 -- The caller has checked that the first token is a legitimate multiplying
2876 -- operator token (i.e. is one of the operator tokens listed above).
2878 -- Error recovery: cannot raise Error_Resync
2880 function P_Multiplying_Operator return Node_Kind is
2881 Mulop_Node : constant array (Token_Class_Mulop) of Node_Kind :=
2882 (Tok_Asterisk => N_Op_Multiply,
2883 Tok_Mod => N_Op_Mod,
2884 Tok_Rem => N_Op_Rem,
2885 Tok_Slash => N_Op_Divide);
2886 begin
2887 return Mulop_Node (Token);
2888 end P_Multiplying_Operator;
2890 --------------------------------------
2891 -- 4.5 Highest Precedence Operator --
2892 --------------------------------------
2894 -- Parsed by P_Factor (4.4)
2896 -- Note: this rule is not in fact used by the grammar at any point
2898 --------------------------
2899 -- 4.6 Type Conversion --
2900 --------------------------
2902 -- Parsed by P_Primary as a Name (4.1)
2904 -------------------------------
2905 -- 4.7 Qualified Expression --
2906 -------------------------------
2908 -- QUALIFIED_EXPRESSION ::=
2909 -- SUBTYPE_MARK ' (EXPRESSION) | SUBTYPE_MARK ' AGGREGATE
2911 -- The caller has scanned the name which is the Subtype_Mark parameter
2912 -- and scanned past the single quote following the subtype mark. The
2913 -- caller has not checked that this name is in fact appropriate for
2914 -- a subtype mark name (i.e. it is a selected component or identifier).
2916 -- Error_Recovery: cannot raise Error_Resync
2918 function P_Qualified_Expression (Subtype_Mark : Node_Id) return Node_Id is
2919 Qual_Node : Node_Id;
2920 begin
2921 Qual_Node := New_Node (N_Qualified_Expression, Prev_Token_Ptr);
2922 Set_Subtype_Mark (Qual_Node, Check_Subtype_Mark (Subtype_Mark));
2923 Set_Expression (Qual_Node, P_Aggregate_Or_Paren_Expr);
2924 return Qual_Node;
2925 end P_Qualified_Expression;
2927 --------------------
2928 -- 4.8 Allocator --
2929 --------------------
2931 -- ALLOCATOR ::=
2932 -- new [SUBPOOL_SPECIFICATION] SUBTYPE_INDICATION
2933 -- | new [SUBPOOL_SPECIFICATION] QUALIFIED_EXPRESSION
2935 -- SUBPOOL_SPECIFICATION ::= (subpool_handle_NAME)
2937 -- The caller has checked that the initial token is NEW
2939 -- Error recovery: can raise Error_Resync
2941 function P_Allocator return Node_Id is
2942 Alloc_Node : Node_Id;
2943 Type_Node : Node_Id;
2944 Null_Exclusion_Present : Boolean;
2946 begin
2947 Alloc_Node := New_Node (N_Allocator, Token_Ptr);
2948 T_New;
2950 -- Scan subpool_specification if present (Ada 2012 (AI05-0111-3))
2952 -- Scan Null_Exclusion if present (Ada 2005 (AI-231))
2954 if Token = Tok_Left_Paren then
2955 Scan; -- past (
2956 Set_Subpool_Handle_Name (Alloc_Node, P_Name);
2957 T_Right_Paren;
2959 Error_Msg_Ada_2012_Feature
2960 ("|subpool specification",
2961 Sloc (Subpool_Handle_Name (Alloc_Node)));
2962 end if;
2964 Null_Exclusion_Present := P_Null_Exclusion;
2965 Set_Null_Exclusion_Present (Alloc_Node, Null_Exclusion_Present);
2966 Type_Node := P_Subtype_Mark_Resync;
2968 if Token = Tok_Apostrophe then
2969 Scan; -- past apostrophe
2970 Set_Expression (Alloc_Node, P_Qualified_Expression (Type_Node));
2971 else
2972 Set_Expression
2973 (Alloc_Node,
2974 P_Subtype_Indication (Type_Node, Null_Exclusion_Present));
2976 -- AI05-0104: An explicit null exclusion is not allowed for an
2977 -- allocator without initialization. In previous versions of the
2978 -- language it just raises constraint error.
2980 if Ada_Version >= Ada_2012 and then Null_Exclusion_Present then
2981 Error_Msg_N
2982 ("an allocator with a subtype indication "
2983 & "cannot have a null exclusion", Alloc_Node);
2984 end if;
2985 end if;
2987 return Alloc_Node;
2988 end P_Allocator;
2990 -----------------------
2991 -- P_Case_Expression --
2992 -----------------------
2994 function P_Case_Expression return Node_Id is
2995 Loc : constant Source_Ptr := Token_Ptr;
2996 Case_Node : Node_Id;
2997 Save_State : Saved_Scan_State;
2999 begin
3000 Error_Msg_Ada_2012_Feature ("|case expression", Token_Ptr);
3001 Scan; -- past CASE
3002 Case_Node :=
3003 Make_Case_Expression (Loc,
3004 Expression => P_Expression_No_Right_Paren,
3005 Alternatives => New_List);
3006 T_Is;
3008 -- We now have scanned out CASE expression IS, scan alternatives
3010 loop
3011 T_When;
3012 Append_To (Alternatives (Case_Node), P_Case_Expression_Alternative);
3014 -- Missing comma if WHEN (more alternatives present)
3016 if Token = Tok_When then
3017 T_Comma;
3019 -- If comma/WHEN, skip comma and we have another alternative
3021 elsif Token = Tok_Comma then
3022 Save_Scan_State (Save_State);
3023 Scan; -- past comma
3025 if Token /= Tok_When then
3026 Restore_Scan_State (Save_State);
3027 exit;
3028 end if;
3030 -- If no comma or WHEN, definitely done
3032 else
3033 exit;
3034 end if;
3035 end loop;
3037 -- If we have an END CASE, diagnose as not needed
3039 if Token = Tok_End then
3040 Error_Msg_SC ("`END CASE` not allowed at end of case expression");
3041 Scan; -- past END
3043 if Token = Tok_Case then
3044 Scan; -- past CASE;
3045 end if;
3046 end if;
3048 -- Return the Case_Expression node
3050 return Case_Node;
3051 end P_Case_Expression;
3053 -----------------------------------
3054 -- P_Case_Expression_Alternative --
3055 -----------------------------------
3057 -- CASE_STATEMENT_ALTERNATIVE ::=
3058 -- when DISCRETE_CHOICE_LIST =>
3059 -- EXPRESSION
3061 -- The caller has checked that and scanned past the initial WHEN token
3062 -- Error recovery: can raise Error_Resync
3064 function P_Case_Expression_Alternative return Node_Id is
3065 Case_Alt_Node : Node_Id;
3066 begin
3067 Case_Alt_Node := New_Node (N_Case_Expression_Alternative, Token_Ptr);
3068 Set_Discrete_Choices (Case_Alt_Node, P_Discrete_Choice_List);
3069 TF_Arrow;
3070 Set_Expression (Case_Alt_Node, P_Expression);
3071 return Case_Alt_Node;
3072 end P_Case_Expression_Alternative;
3074 ---------------------
3075 -- P_If_Expression --
3076 ---------------------
3078 function P_If_Expression return Node_Id is
3080 function P_If_Expression_Internal
3081 (Loc : Source_Ptr;
3082 Cond : Node_Id) return Node_Id;
3083 -- This is the internal recursive routine that does all the work, it is
3084 -- recursive since it is used to process ELSIF parts, which internally
3085 -- are N_If_Expression nodes with the Is_Elsif flag set. The calling
3086 -- sequence is like the outer function except that the caller passes
3087 -- the conditional expression (scanned using P_Expression), and the
3088 -- scan pointer points just past this expression. Loc points to the
3089 -- IF or ELSIF token.
3091 ------------------------------
3092 -- P_If_Expression_Internal --
3093 ------------------------------
3095 function P_If_Expression_Internal
3096 (Loc : Source_Ptr;
3097 Cond : Node_Id) return Node_Id
3099 Exprs : constant List_Id := New_List;
3100 Expr : Node_Id;
3101 State : Saved_Scan_State;
3102 Eptr : Source_Ptr;
3104 begin
3105 -- All cases except where we are at right paren
3107 if Token /= Tok_Right_Paren then
3108 TF_Then;
3109 Append_To (Exprs, P_Condition (Cond));
3110 Append_To (Exprs, P_Expression);
3112 -- Case of right paren (missing THEN phrase). Note that we know this
3113 -- is the IF case, since the caller dealt with this possibility in
3114 -- the ELSIF case.
3116 else
3117 Error_Msg_BC ("missing THEN phrase");
3118 Append_To (Exprs, P_Condition (Cond));
3119 end if;
3121 -- We now have scanned out IF expr THEN expr
3123 -- Check for common error of semicolon before the ELSE
3125 if Token = Tok_Semicolon then
3126 Save_Scan_State (State);
3127 Scan; -- past semicolon
3129 if Token = Tok_Else or else Token = Tok_Elsif then
3130 Error_Msg_SP -- CODEFIX
3131 ("|extra "";"" ignored");
3133 else
3134 Restore_Scan_State (State);
3135 end if;
3136 end if;
3138 -- Scan out ELSIF sequence if present
3140 if Token = Tok_Elsif then
3141 Eptr := Token_Ptr;
3142 Scan; -- past ELSIF
3143 Expr := P_Expression;
3145 -- If we are at a right paren, we assume the ELSIF should be ELSE
3147 if Token = Tok_Right_Paren then
3148 Error_Msg ("ELSIF should be ELSE", Eptr);
3149 Append_To (Exprs, Expr);
3151 -- Otherwise we have an OK ELSIF
3153 else
3154 Expr := P_If_Expression_Internal (Eptr, Expr);
3155 Set_Is_Elsif (Expr);
3156 Append_To (Exprs, Expr);
3157 end if;
3159 -- Scan out ELSE phrase if present
3161 elsif Token = Tok_Else then
3163 -- Scan out ELSE expression
3165 Scan; -- Past ELSE
3166 Append_To (Exprs, P_Expression);
3168 -- Skip redundant ELSE parts
3170 while Token = Tok_Else loop
3171 Error_Msg_SC ("only one ELSE part is allowed");
3172 Scan; -- past ELSE
3173 Discard_Junk_Node (P_Expression);
3174 end loop;
3176 -- Two expression case (implied True, filled in during semantics)
3178 else
3179 null;
3180 end if;
3182 -- If we have an END IF, diagnose as not needed
3184 if Token = Tok_End then
3185 Error_Msg_SC ("`END IF` not allowed at end of if expression");
3186 Scan; -- past END
3188 if Token = Tok_If then
3189 Scan; -- past IF;
3190 end if;
3191 end if;
3193 -- Return the If_Expression node
3195 return Make_If_Expression (Loc, Expressions => Exprs);
3196 end P_If_Expression_Internal;
3198 -- Local variables
3200 Loc : constant Source_Ptr := Token_Ptr;
3201 If_Expr : Node_Id;
3203 -- Start of processing for P_If_Expression
3205 begin
3206 Error_Msg_Ada_2012_Feature ("|if expression", Token_Ptr);
3207 Scan; -- past IF
3208 Inside_If_Expression := Inside_If_Expression + 1;
3209 If_Expr := P_If_Expression_Internal (Loc, P_Expression);
3210 Inside_If_Expression := Inside_If_Expression - 1;
3211 return If_Expr;
3212 end P_If_Expression;
3214 -----------------------
3215 -- P_Membership_Test --
3216 -----------------------
3218 -- MEMBERSHIP_CHOICE_LIST ::= MEMBERHIP_CHOICE {'|' MEMBERSHIP_CHOICE}
3219 -- MEMBERSHIP_CHOICE ::= CHOICE_EXPRESSION | range | subtype_mark
3221 procedure P_Membership_Test (N : Node_Id) is
3222 Alt : constant Node_Id :=
3223 P_Range_Or_Subtype_Mark
3224 (Allow_Simple_Expression => (Ada_Version >= Ada_2012));
3226 begin
3227 -- Set case
3229 if Token = Tok_Vertical_Bar then
3230 Error_Msg_Ada_2012_Feature ("set notation", Token_Ptr);
3231 Set_Alternatives (N, New_List (Alt));
3232 Set_Right_Opnd (N, Empty);
3234 -- Loop to accumulate alternatives
3236 while Token = Tok_Vertical_Bar loop
3237 Scan; -- past vertical bar
3238 Append_To
3239 (Alternatives (N),
3240 P_Range_Or_Subtype_Mark (Allow_Simple_Expression => True));
3241 end loop;
3243 -- Not set case
3245 else
3246 Set_Right_Opnd (N, Alt);
3247 Set_Alternatives (N, No_List);
3248 end if;
3249 end P_Membership_Test;
3251 ------------------------------------------
3252 -- P_Unparen_Cond_Case_Quant_Expression --
3253 ------------------------------------------
3255 function P_Unparen_Cond_Case_Quant_Expression return Node_Id is
3256 Lparen : constant Boolean := Prev_Token = Tok_Left_Paren;
3257 Result : Node_Id;
3259 begin
3260 -- Case expression
3262 if Token = Tok_Case then
3263 Result := P_Case_Expression;
3265 if not (Lparen and then Token = Tok_Right_Paren) then
3266 Error_Msg_N ("case expression must be parenthesized!", Result);
3267 end if;
3269 -- If expression
3271 elsif Token = Tok_If then
3272 Result := P_If_Expression;
3274 if not (Lparen and then Token = Tok_Right_Paren) then
3275 Error_Msg_N ("if expression must be parenthesized!", Result);
3276 end if;
3278 -- Quantified expression
3280 elsif Token = Tok_For then
3281 Result := P_Quantified_Expression;
3283 if not (Lparen and then Token = Tok_Right_Paren) then
3284 Error_Msg_N
3285 ("quantified expression must be parenthesized!", Result);
3286 end if;
3288 -- No other possibility should exist (caller was supposed to check)
3290 else
3291 raise Program_Error;
3292 end if;
3294 -- Return expression (possibly after having given message)
3296 return Result;
3297 end P_Unparen_Cond_Case_Quant_Expression;
3299 end Ch4;