PR tree-optimization/43833
[official-gcc/alias-decl.git] / gcc / ada / par-ch4.adb
blob2bb9d25fcc12add6ed1152387fec4635be0e3a60
1 ------------------------------------------------------------------------------
2 -- --
3 -- GNAT COMPILER COMPONENTS --
4 -- --
5 -- P A R . C H 4 --
6 -- --
7 -- B o d y --
8 -- --
9 -- Copyright (C) 1992-2009, 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_Body_Version => True,
39 Attribute_External_Tag => True,
40 Attribute_Img => True,
41 Attribute_Version => True,
42 Attribute_Base => True,
43 Attribute_Class => True,
44 Attribute_Stub_Type => True,
45 others => False);
46 -- This map contains True for parameterless attributes that return a
47 -- string or a type. For those attributes, a left parenthesis after
48 -- the attribute should not be analyzed as the beginning of a parameters
49 -- list because it may denote a slice operation (X'Img (1 .. 2)) or
50 -- a type conversion (X'Class (Y)).
52 -- Note that this map designates the minimum set of attributes where a
53 -- construct in parentheses that is not an argument can appear right
54 -- after the attribute. For attributes like 'Size, we do not put them
55 -- in the map. If someone writes X'Size (3), that's illegal in any case,
56 -- but we get a better error message by parsing the (3) as an illegal
57 -- argument to the attribute, rather than some meaningless junk that
58 -- follows the attribute.
60 -----------------------
61 -- Local Subprograms --
62 -----------------------
64 function P_Aggregate_Or_Paren_Expr return Node_Id;
65 function P_Allocator return Node_Id;
66 function P_Record_Or_Array_Component_Association return Node_Id;
67 function P_Factor return Node_Id;
68 function P_Primary return Node_Id;
69 function P_Relation return Node_Id;
70 function P_Term return Node_Id;
72 function P_Binary_Adding_Operator return Node_Kind;
73 function P_Logical_Operator return Node_Kind;
74 function P_Multiplying_Operator return Node_Kind;
75 function P_Relational_Operator return Node_Kind;
76 function P_Unary_Adding_Operator return Node_Kind;
78 procedure Bad_Range_Attribute (Loc : Source_Ptr);
79 -- Called to place complaint about bad range attribute at the given
80 -- source location. Terminates by raising Error_Resync.
82 procedure P_Membership_Test (N : Node_Id);
83 -- N is the node for a N_In or N_Not_In node whose right operand has not
84 -- yet been processed. It is called just after scanning out the IN keyword.
85 -- On return, either Right_Opnd or Alternatives is set, as appropriate.
87 function P_Range_Attribute_Reference (Prefix_Node : Node_Id) return Node_Id;
88 -- Scan a range attribute reference. The caller has scanned out the
89 -- prefix. The current token is known to be an apostrophe and the
90 -- following token is known to be RANGE.
92 -------------------------
93 -- Bad_Range_Attribute --
94 -------------------------
96 procedure Bad_Range_Attribute (Loc : Source_Ptr) is
97 begin
98 Error_Msg ("range attribute cannot be used in expression!", Loc);
99 Resync_Expression;
100 end Bad_Range_Attribute;
102 --------------------------
103 -- 4.1 Name (also 6.4) --
104 --------------------------
106 -- NAME ::=
107 -- DIRECT_NAME | EXPLICIT_DEREFERENCE
108 -- | INDEXED_COMPONENT | SLICE
109 -- | SELECTED_COMPONENT | ATTRIBUTE
110 -- | TYPE_CONVERSION | FUNCTION_CALL
111 -- | CHARACTER_LITERAL
113 -- DIRECT_NAME ::= IDENTIFIER | OPERATOR_SYMBOL
115 -- PREFIX ::= NAME | IMPLICIT_DEREFERENCE
117 -- EXPLICIT_DEREFERENCE ::= NAME . all
119 -- IMPLICIT_DEREFERENCE ::= NAME
121 -- INDEXED_COMPONENT ::= PREFIX (EXPRESSION {, EXPRESSION})
123 -- SLICE ::= PREFIX (DISCRETE_RANGE)
125 -- SELECTED_COMPONENT ::= PREFIX . SELECTOR_NAME
127 -- SELECTOR_NAME ::= IDENTIFIER | CHARACTER_LITERAL | OPERATOR_SYMBOL
129 -- ATTRIBUTE_REFERENCE ::= PREFIX ' ATTRIBUTE_DESIGNATOR
131 -- ATTRIBUTE_DESIGNATOR ::=
132 -- IDENTIFIER [(static_EXPRESSION)]
133 -- | access | delta | digits
135 -- FUNCTION_CALL ::=
136 -- function_NAME
137 -- | function_PREFIX ACTUAL_PARAMETER_PART
139 -- ACTUAL_PARAMETER_PART ::=
140 -- (PARAMETER_ASSOCIATION {,PARAMETER_ASSOCIATION})
142 -- PARAMETER_ASSOCIATION ::=
143 -- [formal_parameter_SELECTOR_NAME =>] EXPLICIT_ACTUAL_PARAMETER
145 -- EXPLICIT_ACTUAL_PARAMETER ::= EXPRESSION | variable_NAME
147 -- Note: syntactically a procedure call looks just like a function call,
148 -- so this routine is in practice used to scan out procedure calls as well.
150 -- On return, Expr_Form is set to either EF_Name or EF_Simple_Name
152 -- Error recovery: can raise Error_Resync
154 -- Note: if on return Token = Tok_Apostrophe, then the apostrophe must be
155 -- followed by either a left paren (qualified expression case), or by
156 -- range (range attribute case). All other uses of apostrophe (i.e. all
157 -- other attributes) are handled in this routine.
159 -- Error recovery: can raise Error_Resync
161 function P_Name return Node_Id is
162 Scan_State : Saved_Scan_State;
163 Name_Node : Node_Id;
164 Prefix_Node : Node_Id;
165 Ident_Node : Node_Id;
166 Expr_Node : Node_Id;
167 Range_Node : Node_Id;
168 Arg_Node : Node_Id;
170 Arg_List : List_Id := No_List; -- kill junk warning
171 Attr_Name : Name_Id := No_Name; -- kill junk warning
173 begin
174 -- Case of not a name
176 if Token not in Token_Class_Name then
178 -- If it looks like start of expression, complain and scan expression
180 if Token in Token_Class_Literal
181 or else Token = Tok_Left_Paren
182 then
183 Error_Msg_SC ("name expected");
184 return P_Expression;
186 -- Otherwise some other junk, not much we can do
188 else
189 Error_Msg_AP ("name expected");
190 raise Error_Resync;
191 end if;
192 end if;
194 -- Loop through designators in qualified name
196 Name_Node := Token_Node;
198 loop
199 Scan; -- past designator
200 exit when Token /= Tok_Dot;
201 Save_Scan_State (Scan_State); -- at dot
202 Scan; -- past dot
204 -- If we do not have another designator after the dot, then join
205 -- the normal circuit to handle a dot extension (may be .all or
206 -- character literal case). Otherwise loop back to scan the next
207 -- designator.
209 if Token not in Token_Class_Desig then
210 goto Scan_Name_Extension_Dot;
211 else
212 Prefix_Node := Name_Node;
213 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
214 Set_Prefix (Name_Node, Prefix_Node);
215 Set_Selector_Name (Name_Node, Token_Node);
216 end if;
217 end loop;
219 -- We have now scanned out a qualified designator. If the last token is
220 -- an operator symbol, then we certainly do not have the Snam case, so
221 -- we can just use the normal name extension check circuit
223 if Prev_Token = Tok_Operator_Symbol then
224 goto Scan_Name_Extension;
225 end if;
227 -- We have scanned out a qualified simple name, check for name extension
228 -- Note that we know there is no dot here at this stage, so the only
229 -- possible cases of name extension are apostrophe and left paren.
231 if Token = Tok_Apostrophe then
232 Save_Scan_State (Scan_State); -- at apostrophe
233 Scan; -- past apostrophe
235 -- If left paren, then this might be a qualified expression, but we
236 -- are only in the business of scanning out names, so return with
237 -- Token backed up to point to the apostrophe. The treatment for
238 -- the range attribute is similar (we do not consider x'range to
239 -- be a name in this grammar).
241 if Token = Tok_Left_Paren or else Token = Tok_Range then
242 Restore_Scan_State (Scan_State); -- to apostrophe
243 Expr_Form := EF_Simple_Name;
244 return Name_Node;
246 -- Otherwise we have the case of a name extended by an attribute
248 else
249 goto Scan_Name_Extension_Apostrophe;
250 end if;
252 -- Check case of qualified simple name extended by a left parenthesis
254 elsif Token = Tok_Left_Paren then
255 Scan; -- past left paren
256 goto Scan_Name_Extension_Left_Paren;
258 -- Otherwise the qualified simple name is not extended, so return
260 else
261 Expr_Form := EF_Simple_Name;
262 return Name_Node;
263 end if;
265 -- Loop scanning past name extensions. A label is used for control
266 -- transfer for this loop for ease of interfacing with the finite state
267 -- machine in the parenthesis scanning circuit, and also to allow for
268 -- passing in control to the appropriate point from the above code.
270 <<Scan_Name_Extension>>
272 -- Character literal used as name cannot be extended. Also this
273 -- cannot be a call, since the name for a call must be a designator.
274 -- Return in these cases, or if there is no name extension
276 if Token not in Token_Class_Namext
277 or else Prev_Token = Tok_Char_Literal
278 then
279 Expr_Form := EF_Name;
280 return Name_Node;
281 end if;
283 -- Merge here when we know there is a name extension
285 <<Scan_Name_Extension_OK>>
287 if Token = Tok_Left_Paren then
288 Scan; -- past left paren
289 goto Scan_Name_Extension_Left_Paren;
291 elsif Token = Tok_Apostrophe then
292 Save_Scan_State (Scan_State); -- at apostrophe
293 Scan; -- past apostrophe
294 goto Scan_Name_Extension_Apostrophe;
296 else -- Token = Tok_Dot
297 Save_Scan_State (Scan_State); -- at dot
298 Scan; -- past dot
299 goto Scan_Name_Extension_Dot;
300 end if;
302 -- Case of name extended by dot (selection), dot is already skipped
303 -- and the scan state at the point of the dot is saved in Scan_State.
305 <<Scan_Name_Extension_Dot>>
307 -- Explicit dereference case
309 if Token = Tok_All then
310 Prefix_Node := Name_Node;
311 Name_Node := New_Node (N_Explicit_Dereference, Token_Ptr);
312 Set_Prefix (Name_Node, Prefix_Node);
313 Scan; -- past ALL
314 goto Scan_Name_Extension;
316 -- Selected component case
318 elsif Token in Token_Class_Name then
319 Prefix_Node := Name_Node;
320 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
321 Set_Prefix (Name_Node, Prefix_Node);
322 Set_Selector_Name (Name_Node, Token_Node);
323 Scan; -- past selector
324 goto Scan_Name_Extension;
326 -- Reserved identifier as selector
328 elsif Is_Reserved_Identifier then
329 Scan_Reserved_Identifier (Force_Msg => False);
330 Prefix_Node := Name_Node;
331 Name_Node := New_Node (N_Selected_Component, Prev_Token_Ptr);
332 Set_Prefix (Name_Node, Prefix_Node);
333 Set_Selector_Name (Name_Node, Token_Node);
334 Scan; -- past identifier used as selector
335 goto Scan_Name_Extension;
337 -- If dot is at end of line and followed by nothing legal,
338 -- then assume end of name and quit (dot will be taken as
339 -- an erroneous form of some other punctuation by our caller).
341 elsif Token_Is_At_Start_Of_Line then
342 Restore_Scan_State (Scan_State);
343 return Name_Node;
345 -- Here if nothing legal after the dot
347 else
348 Error_Msg_AP ("selector expected");
349 raise Error_Resync;
350 end if;
352 -- Here for an apostrophe as name extension. The scan position at the
353 -- apostrophe has already been saved, and the apostrophe scanned out.
355 <<Scan_Name_Extension_Apostrophe>>
357 Scan_Apostrophe : declare
358 function Apostrophe_Should_Be_Semicolon return Boolean;
359 -- Checks for case where apostrophe should probably be
360 -- a semicolon, and if so, gives appropriate message,
361 -- resets the scan pointer to the apostrophe, changes
362 -- the current token to Tok_Semicolon, and returns True.
363 -- Otherwise returns False.
365 function Apostrophe_Should_Be_Semicolon return Boolean is
366 begin
367 if Token_Is_At_Start_Of_Line then
368 Restore_Scan_State (Scan_State); -- to apostrophe
369 Error_Msg_SC ("|""''"" should be "";""");
370 Token := Tok_Semicolon;
371 return True;
372 else
373 return False;
374 end if;
375 end Apostrophe_Should_Be_Semicolon;
377 -- Start of processing for Scan_Apostrophe
379 begin
380 -- If range attribute after apostrophe, then return with Token
381 -- pointing to the apostrophe. Note that in this case the prefix
382 -- need not be a simple name (cases like A.all'range). Similarly
383 -- if there is a left paren after the apostrophe, then we also
384 -- return with Token pointing to the apostrophe (this is the
385 -- qualified expression case).
387 if Token = Tok_Range or else Token = Tok_Left_Paren then
388 Restore_Scan_State (Scan_State); -- to apostrophe
389 Expr_Form := EF_Name;
390 return Name_Node;
392 -- Here for cases where attribute designator is an identifier
394 elsif Token = Tok_Identifier then
395 Attr_Name := Token_Name;
397 if not Is_Attribute_Name (Attr_Name) then
398 if Apostrophe_Should_Be_Semicolon then
399 Expr_Form := EF_Name;
400 return Name_Node;
402 -- Here for a bad attribute name
404 else
405 Signal_Bad_Attribute;
406 Scan; -- past bad identifier
408 if Token = Tok_Left_Paren then
409 Scan; -- past left paren
411 loop
412 Discard_Junk_Node (P_Expression_If_OK);
413 exit when not Comma_Present;
414 end loop;
416 T_Right_Paren;
417 end if;
419 return Error;
420 end if;
421 end if;
423 if Style_Check then
424 Style.Check_Attribute_Name (False);
425 end if;
427 -- Here for case of attribute designator is not an identifier
429 else
430 if Token = Tok_Delta then
431 Attr_Name := Name_Delta;
433 elsif Token = Tok_Digits then
434 Attr_Name := Name_Digits;
436 elsif Token = Tok_Access then
437 Attr_Name := Name_Access;
439 elsif Token = Tok_Mod and then Ada_Version = Ada_05 then
440 Attr_Name := Name_Mod;
442 elsif Apostrophe_Should_Be_Semicolon then
443 Expr_Form := EF_Name;
444 return Name_Node;
446 else
447 Error_Msg_AP ("attribute designator expected");
448 raise Error_Resync;
449 end if;
451 if Style_Check then
452 Style.Check_Attribute_Name (True);
453 end if;
454 end if;
456 -- We come here with an OK attribute scanned, and the
457 -- corresponding Attribute identifier node stored in Ident_Node.
459 Prefix_Node := Name_Node;
460 Name_Node := New_Node (N_Attribute_Reference, Prev_Token_Ptr);
461 Scan; -- past attribute designator
462 Set_Prefix (Name_Node, Prefix_Node);
463 Set_Attribute_Name (Name_Node, Attr_Name);
465 -- Scan attribute arguments/designator. We skip this if we know
466 -- that the attribute cannot have an argument.
468 if Token = Tok_Left_Paren
469 and then not
470 Is_Parameterless_Attribute (Get_Attribute_Id (Attr_Name))
471 then
472 Set_Expressions (Name_Node, New_List);
473 Scan; -- past left paren
475 loop
476 declare
477 Expr : constant Node_Id := P_Expression_If_OK;
479 begin
480 if Token = Tok_Arrow then
481 Error_Msg_SC
482 ("named parameters not permitted for attributes");
483 Scan; -- past junk arrow
485 else
486 Append (Expr, Expressions (Name_Node));
487 exit when not Comma_Present;
488 end if;
489 end;
490 end loop;
492 T_Right_Paren;
493 end if;
495 goto Scan_Name_Extension;
496 end Scan_Apostrophe;
498 -- Here for left parenthesis extending name (left paren skipped)
500 <<Scan_Name_Extension_Left_Paren>>
502 -- We now have to scan through a list of items, terminated by a
503 -- right parenthesis. The scan is handled by a finite state
504 -- machine. The possibilities are:
506 -- (discrete_range)
508 -- This is a slice. This case is handled in LP_State_Init
510 -- (expression, expression, ..)
512 -- This is interpreted as an indexed component, i.e. as a
513 -- case of a name which can be extended in the normal manner.
514 -- This case is handled by LP_State_Name or LP_State_Expr.
516 -- Note: conditional expressions (without an extra level of
517 -- parentheses) are permitted in this context).
519 -- (..., identifier => expression , ...)
521 -- If there is at least one occurrence of identifier => (but
522 -- none of the other cases apply), then we have a call.
524 -- Test for Id => case
526 if Token = Tok_Identifier then
527 Save_Scan_State (Scan_State); -- at Id
528 Scan; -- past Id
530 -- Test for => (allow := as an error substitute)
532 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
533 Restore_Scan_State (Scan_State); -- to Id
534 Arg_List := New_List;
535 goto LP_State_Call;
537 else
538 Restore_Scan_State (Scan_State); -- to Id
539 end if;
540 end if;
542 -- Here we have an expression after all
544 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
546 -- Check cases of discrete range for a slice
548 -- First possibility: Range_Attribute_Reference
550 if Expr_Form = EF_Range_Attr then
551 Range_Node := Expr_Node;
553 -- Second possibility: Simple_expression .. Simple_expression
555 elsif Token = Tok_Dot_Dot then
556 Check_Simple_Expression (Expr_Node);
557 Range_Node := New_Node (N_Range, Token_Ptr);
558 Set_Low_Bound (Range_Node, Expr_Node);
559 Scan; -- past ..
560 Expr_Node := P_Expression;
561 Check_Simple_Expression (Expr_Node);
562 Set_High_Bound (Range_Node, Expr_Node);
564 -- Third possibility: Type_name range Range
566 elsif Token = Tok_Range then
567 if Expr_Form /= EF_Simple_Name then
568 Error_Msg_SC -- CODEFIX???
569 ("subtype mark must precede RANGE");
570 raise Error_Resync;
571 end if;
573 Range_Node := P_Subtype_Indication (Expr_Node);
575 -- Otherwise we just have an expression. It is true that we might
576 -- have a subtype mark without a range constraint but this case
577 -- is syntactically indistinguishable from the expression case.
579 else
580 Arg_List := New_List;
581 goto LP_State_Expr;
582 end if;
584 -- Fall through here with unmistakable Discrete range scanned,
585 -- which means that we definitely have the case of a slice. The
586 -- Discrete range is in Range_Node.
588 if Token = Tok_Comma then
589 Error_Msg_SC ("slice cannot have more than one dimension");
590 raise Error_Resync;
592 elsif Token /= Tok_Right_Paren then
593 T_Right_Paren;
594 raise Error_Resync;
596 else
597 Scan; -- past right paren
598 Prefix_Node := Name_Node;
599 Name_Node := New_Node (N_Slice, Sloc (Prefix_Node));
600 Set_Prefix (Name_Node, Prefix_Node);
601 Set_Discrete_Range (Name_Node, Range_Node);
603 -- An operator node is legal as a prefix to other names,
604 -- but not for a slice.
606 if Nkind (Prefix_Node) = N_Operator_Symbol then
607 Error_Msg_N ("illegal prefix for slice", Prefix_Node);
608 end if;
610 -- If we have a name extension, go scan it
612 if Token in Token_Class_Namext then
613 goto Scan_Name_Extension_OK;
615 -- Otherwise return (a slice is a name, but is not a call)
617 else
618 Expr_Form := EF_Name;
619 return Name_Node;
620 end if;
621 end if;
623 -- In LP_State_Expr, we have scanned one or more expressions, and
624 -- so we have a call or an indexed component which is a name. On
625 -- entry we have the expression just scanned in Expr_Node and
626 -- Arg_List contains the list of expressions encountered so far
628 <<LP_State_Expr>>
629 Append (Expr_Node, Arg_List);
631 if Token = Tok_Arrow then
632 Error_Msg
633 ("expect identifier in parameter association",
634 Sloc (Expr_Node));
635 Scan; -- past arrow.
637 elsif not Comma_Present then
638 T_Right_Paren;
639 Prefix_Node := Name_Node;
640 Name_Node := New_Node (N_Indexed_Component, Sloc (Prefix_Node));
641 Set_Prefix (Name_Node, Prefix_Node);
642 Set_Expressions (Name_Node, Arg_List);
643 goto Scan_Name_Extension;
644 end if;
646 -- Comma present (and scanned out), test for identifier => case
647 -- Test for identifier => case
649 if Token = Tok_Identifier then
650 Save_Scan_State (Scan_State); -- at Id
651 Scan; -- past Id
653 -- Test for => (allow := as error substitute)
655 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
656 Restore_Scan_State (Scan_State); -- to Id
657 goto LP_State_Call;
659 -- Otherwise it's just an expression after all, so backup
661 else
662 Restore_Scan_State (Scan_State); -- to Id
663 end if;
664 end if;
666 -- Here we have an expression after all, so stay in this state
668 Expr_Node := P_Expression_If_OK;
669 goto LP_State_Expr;
671 -- LP_State_Call corresponds to the situation in which at least
672 -- one instance of Id => Expression has been encountered, so we
673 -- know that we do not have a name, but rather a call. We enter
674 -- it with the scan pointer pointing to the next argument to scan,
675 -- and Arg_List containing the list of arguments scanned so far.
677 <<LP_State_Call>>
679 -- Test for case of Id => Expression (named parameter)
681 if Token = Tok_Identifier then
682 Save_Scan_State (Scan_State); -- at Id
683 Ident_Node := Token_Node;
684 Scan; -- past Id
686 -- Deal with => (allow := as erroneous substitute)
688 if Token = Tok_Arrow or else Token = Tok_Colon_Equal then
689 Arg_Node := New_Node (N_Parameter_Association, Prev_Token_Ptr);
690 Set_Selector_Name (Arg_Node, Ident_Node);
691 T_Arrow;
692 Set_Explicit_Actual_Parameter (Arg_Node, P_Expression);
693 Append (Arg_Node, Arg_List);
695 -- If a comma follows, go back and scan next entry
697 if Comma_Present then
698 goto LP_State_Call;
700 -- Otherwise we have the end of a call
702 else
703 Prefix_Node := Name_Node;
704 Name_Node := New_Node (N_Function_Call, Sloc (Prefix_Node));
705 Set_Name (Name_Node, Prefix_Node);
706 Set_Parameter_Associations (Name_Node, Arg_List);
707 T_Right_Paren;
709 if Token in Token_Class_Namext then
710 goto Scan_Name_Extension_OK;
712 -- This is a case of a call which cannot be a name
714 else
715 Expr_Form := EF_Name;
716 return Name_Node;
717 end if;
718 end if;
720 -- Not named parameter: Id started an expression after all
722 else
723 Restore_Scan_State (Scan_State); -- to Id
724 end if;
725 end if;
727 -- Here if entry did not start with Id => which means that it
728 -- is a positional parameter, which is not allowed, since we
729 -- have seen at least one named parameter already.
731 Error_Msg_SC
732 ("positional parameter association " &
733 "not allowed after named one");
735 Expr_Node := P_Expression_If_OK;
737 -- Leaving the '>' in an association is not unusual, so suggest
738 -- a possible fix.
740 if Nkind (Expr_Node) = N_Op_Eq then
741 Error_Msg_N ("\maybe `='>` was intended", Expr_Node);
742 end if;
744 -- We go back to scanning out expressions, so that we do not get
745 -- multiple error messages when several positional parameters
746 -- follow a named parameter.
748 goto LP_State_Expr;
750 -- End of treatment for name extensions starting with left paren
752 -- End of loop through name extensions
754 end P_Name;
756 -- This function parses a restricted form of Names which are either
757 -- designators, or designators preceded by a sequence of prefixes
758 -- that are direct names.
760 -- Error recovery: cannot raise Error_Resync
762 function P_Function_Name return Node_Id is
763 Designator_Node : Node_Id;
764 Prefix_Node : Node_Id;
765 Selector_Node : Node_Id;
766 Dot_Sloc : Source_Ptr := No_Location;
768 begin
769 -- Prefix_Node is set to the gathered prefix so far, Empty means that
770 -- no prefix has been scanned. This allows us to build up the result
771 -- in the required right recursive manner.
773 Prefix_Node := Empty;
775 -- Loop through prefixes
777 loop
778 Designator_Node := Token_Node;
780 if Token not in Token_Class_Desig then
781 return P_Identifier; -- let P_Identifier issue the error message
783 else -- Token in Token_Class_Desig
784 Scan; -- past designator
785 exit when Token /= Tok_Dot;
786 end if;
788 -- Here at a dot, with token just before it in Designator_Node
790 if No (Prefix_Node) then
791 Prefix_Node := Designator_Node;
792 else
793 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
794 Set_Prefix (Selector_Node, Prefix_Node);
795 Set_Selector_Name (Selector_Node, Designator_Node);
796 Prefix_Node := Selector_Node;
797 end if;
799 Dot_Sloc := Token_Ptr;
800 Scan; -- past dot
801 end loop;
803 -- Fall out of the loop having just scanned a designator
805 if No (Prefix_Node) then
806 return Designator_Node;
807 else
808 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
809 Set_Prefix (Selector_Node, Prefix_Node);
810 Set_Selector_Name (Selector_Node, Designator_Node);
811 return Selector_Node;
812 end if;
814 exception
815 when Error_Resync =>
816 return Error;
817 end P_Function_Name;
819 -- This function parses a restricted form of Names which are either
820 -- identifiers, or identifiers preceded by a sequence of prefixes
821 -- that are direct names.
823 -- Error recovery: cannot raise Error_Resync
825 function P_Qualified_Simple_Name return Node_Id is
826 Designator_Node : Node_Id;
827 Prefix_Node : Node_Id;
828 Selector_Node : Node_Id;
829 Dot_Sloc : Source_Ptr := No_Location;
831 begin
832 -- Prefix node is set to the gathered prefix so far, Empty means that
833 -- no prefix has been scanned. This allows us to build up the result
834 -- in the required right recursive manner.
836 Prefix_Node := Empty;
838 -- Loop through prefixes
840 loop
841 Designator_Node := Token_Node;
843 if Token = Tok_Identifier then
844 Scan; -- past identifier
845 exit when Token /= Tok_Dot;
847 elsif Token not in Token_Class_Desig then
848 return P_Identifier; -- let P_Identifier issue the error message
850 else
851 Scan; -- past designator
853 if Token /= Tok_Dot then
854 Error_Msg_SP ("identifier expected");
855 return Error;
856 end if;
857 end if;
859 -- Here at a dot, with token just before it in Designator_Node
861 if No (Prefix_Node) then
862 Prefix_Node := Designator_Node;
863 else
864 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
865 Set_Prefix (Selector_Node, Prefix_Node);
866 Set_Selector_Name (Selector_Node, Designator_Node);
867 Prefix_Node := Selector_Node;
868 end if;
870 Dot_Sloc := Token_Ptr;
871 Scan; -- past dot
872 end loop;
874 -- Fall out of the loop having just scanned an identifier
876 if No (Prefix_Node) then
877 return Designator_Node;
878 else
879 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
880 Set_Prefix (Selector_Node, Prefix_Node);
881 Set_Selector_Name (Selector_Node, Designator_Node);
882 return Selector_Node;
883 end if;
885 exception
886 when Error_Resync =>
887 return Error;
888 end P_Qualified_Simple_Name;
890 -- This procedure differs from P_Qualified_Simple_Name only in that it
891 -- raises Error_Resync if any error is encountered. It only returns after
892 -- scanning a valid qualified simple name.
894 -- Error recovery: can raise Error_Resync
896 function P_Qualified_Simple_Name_Resync return Node_Id is
897 Designator_Node : Node_Id;
898 Prefix_Node : Node_Id;
899 Selector_Node : Node_Id;
900 Dot_Sloc : Source_Ptr := No_Location;
902 begin
903 Prefix_Node := Empty;
905 -- Loop through prefixes
907 loop
908 Designator_Node := Token_Node;
910 if Token = Tok_Identifier then
911 Scan; -- past identifier
912 exit when Token /= Tok_Dot;
914 elsif Token not in Token_Class_Desig then
915 Discard_Junk_Node (P_Identifier); -- to issue the error message
916 raise Error_Resync;
918 else
919 Scan; -- past designator
921 if Token /= Tok_Dot then
922 Error_Msg_SP ("identifier expected");
923 raise Error_Resync;
924 end if;
925 end if;
927 -- Here at a dot, with token just before it in Designator_Node
929 if No (Prefix_Node) then
930 Prefix_Node := Designator_Node;
931 else
932 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
933 Set_Prefix (Selector_Node, Prefix_Node);
934 Set_Selector_Name (Selector_Node, Designator_Node);
935 Prefix_Node := Selector_Node;
936 end if;
938 Dot_Sloc := Token_Ptr;
939 Scan; -- past period
940 end loop;
942 -- Fall out of the loop having just scanned an identifier
944 if No (Prefix_Node) then
945 return Designator_Node;
946 else
947 Selector_Node := New_Node (N_Selected_Component, Dot_Sloc);
948 Set_Prefix (Selector_Node, Prefix_Node);
949 Set_Selector_Name (Selector_Node, Designator_Node);
950 return Selector_Node;
951 end if;
952 end P_Qualified_Simple_Name_Resync;
954 ----------------------
955 -- 4.1 Direct_Name --
956 ----------------------
958 -- Parsed by P_Name and other functions in section 4.1
960 -----------------
961 -- 4.1 Prefix --
962 -----------------
964 -- Parsed by P_Name (4.1)
966 -------------------------------
967 -- 4.1 Explicit Dereference --
968 -------------------------------
970 -- Parsed by P_Name (4.1)
972 -------------------------------
973 -- 4.1 Implicit_Dereference --
974 -------------------------------
976 -- Parsed by P_Name (4.1)
978 ----------------------------
979 -- 4.1 Indexed Component --
980 ----------------------------
982 -- Parsed by P_Name (4.1)
984 ----------------
985 -- 4.1 Slice --
986 ----------------
988 -- Parsed by P_Name (4.1)
990 -----------------------------
991 -- 4.1 Selected_Component --
992 -----------------------------
994 -- Parsed by P_Name (4.1)
996 ------------------------
997 -- 4.1 Selector Name --
998 ------------------------
1000 -- Parsed by P_Name (4.1)
1002 ------------------------------
1003 -- 4.1 Attribute Reference --
1004 ------------------------------
1006 -- Parsed by P_Name (4.1)
1008 -------------------------------
1009 -- 4.1 Attribute Designator --
1010 -------------------------------
1012 -- Parsed by P_Name (4.1)
1014 --------------------------------------
1015 -- 4.1.4 Range Attribute Reference --
1016 --------------------------------------
1018 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
1020 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
1022 -- In the grammar, a RANGE attribute is simply a name, but its use is
1023 -- highly restricted, so in the parser, we do not regard it as a name.
1024 -- Instead, P_Name returns without scanning the 'RANGE part of the
1025 -- attribute, and the caller uses the following function to construct
1026 -- a range attribute in places where it is appropriate.
1028 -- Note that RANGE here is treated essentially as an identifier,
1029 -- rather than a reserved word.
1031 -- The caller has parsed the prefix, i.e. a name, and Token points to
1032 -- the apostrophe. The token after the apostrophe is known to be RANGE
1033 -- at this point. The prefix node becomes the prefix of the attribute.
1035 -- Error_Recovery: Cannot raise Error_Resync
1037 function P_Range_Attribute_Reference
1038 (Prefix_Node : Node_Id)
1039 return Node_Id
1041 Attr_Node : Node_Id;
1043 begin
1044 Attr_Node := New_Node (N_Attribute_Reference, Token_Ptr);
1045 Set_Prefix (Attr_Node, Prefix_Node);
1046 Scan; -- past apostrophe
1048 if Style_Check then
1049 Style.Check_Attribute_Name (True);
1050 end if;
1052 Set_Attribute_Name (Attr_Node, Name_Range);
1053 Scan; -- past RANGE
1055 if Token = Tok_Left_Paren then
1056 Scan; -- past left paren
1057 Set_Expressions (Attr_Node, New_List (P_Expression_If_OK));
1058 T_Right_Paren;
1059 end if;
1061 return Attr_Node;
1062 end P_Range_Attribute_Reference;
1064 ---------------------------------------
1065 -- 4.1.4 Range Attribute Designator --
1066 ---------------------------------------
1068 -- Parsed by P_Range_Attribute_Reference (4.4)
1070 --------------------
1071 -- 4.3 Aggregate --
1072 --------------------
1074 -- AGGREGATE ::= RECORD_AGGREGATE | EXTENSION_AGGREGATE | ARRAY_AGGREGATE
1076 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3), except in the case where
1077 -- an aggregate is known to be required (code statement, extension
1078 -- aggregate), in which cases this routine performs the necessary check
1079 -- that we have an aggregate rather than a parenthesized expression
1081 -- Error recovery: can raise Error_Resync
1083 function P_Aggregate return Node_Id is
1084 Aggr_Sloc : constant Source_Ptr := Token_Ptr;
1085 Aggr_Node : constant Node_Id := P_Aggregate_Or_Paren_Expr;
1087 begin
1088 if Nkind (Aggr_Node) /= N_Aggregate
1089 and then
1090 Nkind (Aggr_Node) /= N_Extension_Aggregate
1091 then
1092 Error_Msg
1093 ("aggregate may not have single positional component", Aggr_Sloc);
1094 return Error;
1095 else
1096 return Aggr_Node;
1097 end if;
1098 end P_Aggregate;
1100 ------------------------------------------------
1101 -- 4.3 Aggregate or Parenthesized Expression --
1102 ------------------------------------------------
1104 -- This procedure parses out either an aggregate or a parenthesized
1105 -- expression (these two constructs are closely related, since a
1106 -- parenthesized expression looks like an aggregate with a single
1107 -- positional component).
1109 -- AGGREGATE ::=
1110 -- RECORD_AGGREGATE | EXTENSION_AGGREGATE | ARRAY_AGGREGATE
1112 -- RECORD_AGGREGATE ::= (RECORD_COMPONENT_ASSOCIATION_LIST)
1114 -- RECORD_COMPONENT_ASSOCIATION_LIST ::=
1115 -- RECORD_COMPONENT_ASSOCIATION {, RECORD_COMPONENT_ASSOCIATION}
1116 -- | null record
1118 -- RECORD_COMPONENT_ASSOCIATION ::=
1119 -- [COMPONENT_CHOICE_LIST =>] EXPRESSION
1121 -- COMPONENT_CHOICE_LIST ::=
1122 -- component_SELECTOR_NAME {| component_SELECTOR_NAME}
1123 -- | others
1125 -- EXTENSION_AGGREGATE ::=
1126 -- (ANCESTOR_PART with RECORD_COMPONENT_ASSOCIATION_LIST)
1128 -- ANCESTOR_PART ::= EXPRESSION | SUBTYPE_MARK
1130 -- ARRAY_AGGREGATE ::=
1131 -- POSITIONAL_ARRAY_AGGREGATE | NAMED_ARRAY_AGGREGATE
1133 -- POSITIONAL_ARRAY_AGGREGATE ::=
1134 -- (EXPRESSION, EXPRESSION {, EXPRESSION})
1135 -- | (EXPRESSION {, EXPRESSION}, others => EXPRESSION)
1136 -- | (EXPRESSION {, EXPRESSION}, others => <>)
1138 -- NAMED_ARRAY_AGGREGATE ::=
1139 -- (ARRAY_COMPONENT_ASSOCIATION {, ARRAY_COMPONENT_ASSOCIATION})
1141 -- PRIMARY ::= (EXPRESSION);
1143 -- Error recovery: can raise Error_Resync
1145 -- Note: POSITIONAL_ARRAY_AGGREGATE rule has been extended to give support
1146 -- to Ada 2005 limited aggregates (AI-287)
1148 function P_Aggregate_Or_Paren_Expr return Node_Id is
1149 Aggregate_Node : Node_Id;
1150 Expr_List : List_Id;
1151 Assoc_List : List_Id;
1152 Expr_Node : Node_Id;
1153 Lparen_Sloc : Source_Ptr;
1154 Scan_State : Saved_Scan_State;
1156 begin
1157 Lparen_Sloc := Token_Ptr;
1158 T_Left_Paren;
1160 -- Conditional expression case
1162 if Token = Tok_If then
1163 Expr_Node := P_Conditional_Expression;
1164 T_Right_Paren;
1165 return Expr_Node;
1167 -- Note: the mechanism used here of rescanning the initial expression
1168 -- is distinctly unpleasant, but it saves a lot of fiddling in scanning
1169 -- out the discrete choice list.
1171 -- Deal with expression and extension aggregate cases first
1173 elsif Token /= Tok_Others then
1174 Save_Scan_State (Scan_State); -- at start of expression
1176 -- Deal with (NULL RECORD) case
1178 if Token = Tok_Null then
1179 Scan; -- past NULL
1181 if Token = Tok_Record then
1182 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1183 Set_Null_Record_Present (Aggregate_Node, True);
1184 Scan; -- past RECORD
1185 T_Right_Paren;
1186 return Aggregate_Node;
1187 else
1188 Restore_Scan_State (Scan_State); -- to NULL that must be expr
1189 end if;
1190 end if;
1192 -- Ada 2005 (AI-287): The box notation is allowed only with named
1193 -- notation because positional notation might be error prone. For
1194 -- example, in "(X, <>, Y, <>)", there is no type associated with
1195 -- the boxes, so you might not be leaving out the components you
1196 -- thought you were leaving out.
1198 if Ada_Version >= Ada_05 and then Token = Tok_Box then
1199 Error_Msg_SC ("(Ada 2005) box notation only allowed with "
1200 & "named notation");
1201 Scan; -- past BOX
1202 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1203 return Aggregate_Node;
1204 end if;
1206 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
1208 -- Extension aggregate case
1210 if Token = Tok_With then
1212 if Nkind (Expr_Node) = N_Attribute_Reference
1213 and then Attribute_Name (Expr_Node) = Name_Range
1214 then
1215 Bad_Range_Attribute (Sloc (Expr_Node));
1216 return Error;
1217 end if;
1219 if Ada_Version = Ada_83 then
1220 Error_Msg_SC ("(Ada 83) extension aggregate not allowed");
1221 end if;
1223 Aggregate_Node := New_Node (N_Extension_Aggregate, Lparen_Sloc);
1224 Set_Ancestor_Part (Aggregate_Node, Expr_Node);
1225 Scan; -- past WITH
1227 -- Deal with WITH NULL RECORD case
1229 if Token = Tok_Null then
1230 Save_Scan_State (Scan_State); -- at NULL
1231 Scan; -- past NULL
1233 if Token = Tok_Record then
1234 Scan; -- past RECORD
1235 Set_Null_Record_Present (Aggregate_Node, True);
1236 T_Right_Paren;
1237 return Aggregate_Node;
1239 else
1240 Restore_Scan_State (Scan_State); -- to NULL that must be expr
1241 end if;
1242 end if;
1244 if Token /= Tok_Others then
1245 Save_Scan_State (Scan_State);
1246 Expr_Node := P_Expression;
1247 else
1248 Expr_Node := Empty;
1249 end if;
1251 -- Expression case
1253 elsif Token = Tok_Right_Paren or else Token in Token_Class_Eterm then
1254 if Nkind (Expr_Node) = N_Attribute_Reference
1255 and then Attribute_Name (Expr_Node) = Name_Range
1256 then
1257 Error_Msg
1258 ("|parentheses not allowed for range attribute", Lparen_Sloc);
1259 Scan; -- past right paren
1260 return Expr_Node;
1261 end if;
1263 -- Bump paren count of expression
1265 if Expr_Node /= Error then
1266 Set_Paren_Count (Expr_Node, Paren_Count (Expr_Node) + 1);
1267 end if;
1269 T_Right_Paren; -- past right paren (error message if none)
1270 return Expr_Node;
1272 -- Normal aggregate case
1274 else
1275 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1276 end if;
1278 -- Others case
1280 else
1281 Aggregate_Node := New_Node (N_Aggregate, Lparen_Sloc);
1282 Expr_Node := Empty;
1283 end if;
1285 -- Prepare to scan list of component associations
1287 Expr_List := No_List; -- don't set yet, maybe all named entries
1288 Assoc_List := No_List; -- don't set yet, maybe all positional entries
1290 -- This loop scans through component associations. On entry to the
1291 -- loop, an expression has been scanned at the start of the current
1292 -- association unless initial token was OTHERS, in which case
1293 -- Expr_Node is set to Empty.
1295 loop
1296 -- Deal with others association first. This is a named association
1298 if No (Expr_Node) then
1299 if No (Assoc_List) then
1300 Assoc_List := New_List;
1301 end if;
1303 Append (P_Record_Or_Array_Component_Association, Assoc_List);
1305 -- Improper use of WITH
1307 elsif Token = Tok_With then
1308 Error_Msg_SC ("WITH must be preceded by single expression in " &
1309 "extension aggregate");
1310 raise Error_Resync;
1312 -- A range attribute can only appear as part of a discrete choice
1313 -- list.
1315 elsif Nkind (Expr_Node) = N_Attribute_Reference
1316 and then Attribute_Name (Expr_Node) = Name_Range
1317 and then Token /= Tok_Arrow
1318 and then Token /= Tok_Vertical_Bar
1319 then
1320 Bad_Range_Attribute (Sloc (Expr_Node));
1321 return Error;
1323 -- Assume positional case if comma, right paren, or literal or
1324 -- identifier or OTHERS follows (the latter cases are missing
1325 -- comma cases). Also assume positional if a semicolon follows,
1326 -- which can happen if there are missing parens
1328 elsif Token = Tok_Comma
1329 or else Token = Tok_Right_Paren
1330 or else Token = Tok_Others
1331 or else Token in Token_Class_Lit_Or_Name
1332 or else Token = Tok_Semicolon
1333 then
1334 if Present (Assoc_List) then
1335 Error_Msg_BC
1336 ("""='>"" expected (positional association cannot follow " &
1337 "named association)");
1338 end if;
1340 if No (Expr_List) then
1341 Expr_List := New_List;
1342 end if;
1344 Append (Expr_Node, Expr_List);
1346 -- Check for aggregate followed by left parent, maybe missing comma
1348 elsif Nkind (Expr_Node) = N_Aggregate
1349 and then Token = Tok_Left_Paren
1350 then
1351 T_Comma;
1353 if No (Expr_List) then
1354 Expr_List := New_List;
1355 end if;
1357 Append (Expr_Node, Expr_List);
1359 -- Anything else is assumed to be a named association
1361 else
1362 Restore_Scan_State (Scan_State); -- to start of expression
1364 if No (Assoc_List) then
1365 Assoc_List := New_List;
1366 end if;
1368 Append (P_Record_Or_Array_Component_Association, Assoc_List);
1369 end if;
1371 exit when not Comma_Present;
1373 -- If we are at an expression terminator, something is seriously
1374 -- wrong, so let's get out now, before we start eating up stuff
1375 -- that doesn't belong to us!
1377 if Token in Token_Class_Eterm then
1378 Error_Msg_AP ("expecting expression or component association");
1379 exit;
1380 end if;
1382 -- Otherwise initiate for reentry to top of loop by scanning an
1383 -- initial expression, unless the first token is OTHERS.
1385 if Token = Tok_Others then
1386 Expr_Node := Empty;
1387 else
1388 Save_Scan_State (Scan_State); -- at start of expression
1389 Expr_Node := P_Expression_Or_Range_Attribute_If_OK;
1391 end if;
1392 end loop;
1394 -- All component associations (positional and named) have been scanned
1396 T_Right_Paren;
1397 Set_Expressions (Aggregate_Node, Expr_List);
1398 Set_Component_Associations (Aggregate_Node, Assoc_List);
1399 return Aggregate_Node;
1400 end P_Aggregate_Or_Paren_Expr;
1402 ------------------------------------------------
1403 -- 4.3 Record or Array Component Association --
1404 ------------------------------------------------
1406 -- RECORD_COMPONENT_ASSOCIATION ::=
1407 -- [COMPONENT_CHOICE_LIST =>] EXPRESSION
1408 -- | COMPONENT_CHOICE_LIST => <>
1410 -- COMPONENT_CHOICE_LIST =>
1411 -- component_SELECTOR_NAME {| component_SELECTOR_NAME}
1412 -- | others
1414 -- ARRAY_COMPONENT_ASSOCIATION ::=
1415 -- DISCRETE_CHOICE_LIST => EXPRESSION
1416 -- | DISCRETE_CHOICE_LIST => <>
1418 -- Note: this routine only handles the named cases, including others.
1419 -- Cases where the component choice list is not present have already
1420 -- been handled directly.
1422 -- Error recovery: can raise Error_Resync
1424 -- Note: RECORD_COMPONENT_ASSOCIATION and ARRAY_COMPONENT_ASSOCIATION
1425 -- rules have been extended to give support to Ada 2005 limited
1426 -- aggregates (AI-287)
1428 function P_Record_Or_Array_Component_Association return Node_Id is
1429 Assoc_Node : Node_Id;
1431 begin
1432 Assoc_Node := New_Node (N_Component_Association, Token_Ptr);
1433 Set_Choices (Assoc_Node, P_Discrete_Choice_List);
1434 Set_Sloc (Assoc_Node, Token_Ptr);
1435 TF_Arrow;
1437 if Token = Tok_Box then
1439 -- Ada 2005(AI-287): The box notation is used to indicate the
1440 -- default initialization of aggregate components
1442 if Ada_Version < Ada_05 then
1443 Error_Msg_SP
1444 ("component association with '<'> is an Ada 2005 extension");
1445 Error_Msg_SP ("\unit must be compiled with -gnat05 switch");
1446 end if;
1448 Set_Box_Present (Assoc_Node);
1449 Scan; -- Past box
1450 else
1451 Set_Expression (Assoc_Node, P_Expression);
1452 end if;
1454 return Assoc_Node;
1455 end P_Record_Or_Array_Component_Association;
1457 -----------------------------
1458 -- 4.3.1 Record Aggregate --
1459 -----------------------------
1461 -- Case of enumeration aggregate is parsed by P_Aggregate (4.3)
1462 -- All other cases are parsed by P_Aggregate_Or_Paren_Expr (4.3)
1464 ----------------------------------------------
1465 -- 4.3.1 Record Component Association List --
1466 ----------------------------------------------
1468 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1470 ----------------------------------
1471 -- 4.3.1 Component Choice List --
1472 ----------------------------------
1474 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1476 --------------------------------
1477 -- 4.3.1 Extension Aggregate --
1478 --------------------------------
1480 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1482 --------------------------
1483 -- 4.3.1 Ancestor Part --
1484 --------------------------
1486 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1488 ----------------------------
1489 -- 4.3.1 Array Aggregate --
1490 ----------------------------
1492 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1494 ---------------------------------------
1495 -- 4.3.1 Positional Array Aggregate --
1496 ---------------------------------------
1498 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1500 ----------------------------------
1501 -- 4.3.1 Named Array Aggregate --
1502 ----------------------------------
1504 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1506 ----------------------------------------
1507 -- 4.3.1 Array Component Association --
1508 ----------------------------------------
1510 -- Parsed by P_Aggregate_Or_Paren_Expr (4.3)
1512 ---------------------
1513 -- 4.4 Expression --
1514 ---------------------
1516 -- EXPRESSION ::=
1517 -- RELATION {and RELATION} | RELATION {and then RELATION}
1518 -- | RELATION {or RELATION} | RELATION {or else RELATION}
1519 -- | RELATION {xor RELATION}
1521 -- On return, Expr_Form indicates the categorization of the expression
1522 -- EF_Range_Attr is not a possible value (if a range attribute is found,
1523 -- an error message is given, and Error is returned).
1525 -- Error recovery: cannot raise Error_Resync
1527 function P_Expression return Node_Id is
1528 Logical_Op : Node_Kind;
1529 Prev_Logical_Op : Node_Kind;
1530 Op_Location : Source_Ptr;
1531 Node1 : Node_Id;
1532 Node2 : Node_Id;
1534 begin
1535 Node1 := P_Relation;
1537 if Token in Token_Class_Logop then
1538 Prev_Logical_Op := N_Empty;
1540 loop
1541 Op_Location := Token_Ptr;
1542 Logical_Op := P_Logical_Operator;
1544 if Prev_Logical_Op /= N_Empty and then
1545 Logical_Op /= Prev_Logical_Op
1546 then
1547 Error_Msg
1548 ("mixed logical operators in expression", Op_Location);
1549 Prev_Logical_Op := N_Empty;
1550 else
1551 Prev_Logical_Op := Logical_Op;
1552 end if;
1554 Node2 := Node1;
1555 Node1 := New_Op_Node (Logical_Op, Op_Location);
1556 Set_Left_Opnd (Node1, Node2);
1557 Set_Right_Opnd (Node1, P_Relation);
1558 exit when Token not in Token_Class_Logop;
1559 end loop;
1561 Expr_Form := EF_Non_Simple;
1562 end if;
1564 if Token = Tok_Apostrophe then
1565 Bad_Range_Attribute (Token_Ptr);
1566 return Error;
1567 else
1568 return Node1;
1569 end if;
1570 end P_Expression;
1572 -- This function is identical to the normal P_Expression, except that it
1573 -- also permits the appearence of a conditional expression without the
1574 -- usual surrounding parentheses.
1576 function P_Expression_If_OK return Node_Id is
1577 begin
1578 if Token = Tok_If then
1579 return P_Conditional_Expression;
1580 else
1581 return P_Expression;
1582 end if;
1583 end P_Expression_If_OK;
1585 -- This function is identical to the normal P_Expression, except that it
1586 -- checks that the expression scan did not stop on a right paren. It is
1587 -- called in all contexts where a right parenthesis cannot legitimately
1588 -- follow an expression.
1590 -- Error recovery: can not raise Error_Resync
1592 function P_Expression_No_Right_Paren return Node_Id is
1593 Expr : constant Node_Id := P_Expression;
1594 begin
1595 Ignore (Tok_Right_Paren);
1596 return Expr;
1597 end P_Expression_No_Right_Paren;
1599 ----------------------------------------
1600 -- 4.4 Expression_Or_Range_Attribute --
1601 ----------------------------------------
1603 -- EXPRESSION ::=
1604 -- RELATION {and RELATION} | RELATION {and then RELATION}
1605 -- | RELATION {or RELATION} | RELATION {or else RELATION}
1606 -- | RELATION {xor RELATION}
1608 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
1610 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
1612 -- On return, Expr_Form indicates the categorization of the expression
1613 -- and EF_Range_Attr is one of the possibilities.
1615 -- Error recovery: cannot raise Error_Resync
1617 -- In the grammar, a RANGE attribute is simply a name, but its use is
1618 -- highly restricted, so in the parser, we do not regard it as a name.
1619 -- Instead, P_Name returns without scanning the 'RANGE part of the
1620 -- attribute, and P_Expression_Or_Range_Attribute handles the range
1621 -- attribute reference. In the normal case where a range attribute is
1622 -- not allowed, an error message is issued by P_Expression.
1624 function P_Expression_Or_Range_Attribute return Node_Id is
1625 Logical_Op : Node_Kind;
1626 Prev_Logical_Op : Node_Kind;
1627 Op_Location : Source_Ptr;
1628 Node1 : Node_Id;
1629 Node2 : Node_Id;
1630 Attr_Node : Node_Id;
1632 begin
1633 Node1 := P_Relation;
1635 if Token = Tok_Apostrophe then
1636 Attr_Node := P_Range_Attribute_Reference (Node1);
1637 Expr_Form := EF_Range_Attr;
1638 return Attr_Node;
1640 elsif Token in Token_Class_Logop then
1641 Prev_Logical_Op := N_Empty;
1643 loop
1644 Op_Location := Token_Ptr;
1645 Logical_Op := P_Logical_Operator;
1647 if Prev_Logical_Op /= N_Empty and then
1648 Logical_Op /= Prev_Logical_Op
1649 then
1650 Error_Msg
1651 ("mixed logical operators in expression", Op_Location);
1652 Prev_Logical_Op := N_Empty;
1653 else
1654 Prev_Logical_Op := Logical_Op;
1655 end if;
1657 Node2 := Node1;
1658 Node1 := New_Op_Node (Logical_Op, Op_Location);
1659 Set_Left_Opnd (Node1, Node2);
1660 Set_Right_Opnd (Node1, P_Relation);
1661 exit when Token not in Token_Class_Logop;
1662 end loop;
1664 Expr_Form := EF_Non_Simple;
1665 end if;
1667 if Token = Tok_Apostrophe then
1668 Bad_Range_Attribute (Token_Ptr);
1669 return Error;
1670 else
1671 return Node1;
1672 end if;
1673 end P_Expression_Or_Range_Attribute;
1675 -- Version that allows a non-parenthesized conditional expression
1677 function P_Expression_Or_Range_Attribute_If_OK return Node_Id is
1678 begin
1679 if Token = Tok_If then
1680 return P_Conditional_Expression;
1681 else
1682 return P_Expression_Or_Range_Attribute;
1683 end if;
1684 end P_Expression_Or_Range_Attribute_If_OK;
1686 -------------------
1687 -- 4.4 Relation --
1688 -------------------
1690 -- RELATION ::=
1691 -- SIMPLE_EXPRESSION [RELATIONAL_OPERATOR SIMPLE_EXPRESSION]
1692 -- | SIMPLE_EXPRESSION [not] in RANGE
1693 -- | SIMPLE_EXPRESSION [not] in SUBTYPE_MARK
1695 -- On return, Expr_Form indicates the categorization of the expression
1697 -- Note: if Token = Tok_Apostrophe on return, then Expr_Form is set to
1698 -- EF_Simple_Name and the following token is RANGE (range attribute case).
1700 -- Error recovery: cannot raise Error_Resync. If an error occurs within an
1701 -- expression, then tokens are scanned until either a non-expression token,
1702 -- a right paren (not matched by a left paren) or a comma, is encountered.
1704 function P_Relation return Node_Id is
1705 Node1, Node2 : Node_Id;
1706 Optok : Source_Ptr;
1708 begin
1709 Node1 := P_Simple_Expression;
1711 if Token not in Token_Class_Relop then
1712 return Node1;
1714 else
1715 -- Here we have a relational operator following. If so then scan it
1716 -- out. Note that the assignment symbol := is treated as a relational
1717 -- operator to improve the error recovery when it is misused for =.
1718 -- P_Relational_Operator also parses the IN and NOT IN operations.
1720 Optok := Token_Ptr;
1721 Node2 := New_Op_Node (P_Relational_Operator, Optok);
1722 Set_Left_Opnd (Node2, Node1);
1724 -- Case of IN or NOT IN
1726 if Prev_Token = Tok_In then
1727 P_Membership_Test (Node2);
1729 -- Case of relational operator (= /= < <= > >=)
1731 else
1732 Set_Right_Opnd (Node2, P_Simple_Expression);
1733 end if;
1735 Expr_Form := EF_Non_Simple;
1737 if Token in Token_Class_Relop then
1738 Error_Msg_SC ("unexpected relational operator");
1739 raise Error_Resync;
1740 end if;
1742 return Node2;
1743 end if;
1745 -- If any error occurs, then scan to the next expression terminator symbol
1746 -- or comma or right paren at the outer (i.e. current) parentheses level.
1747 -- The flags are set to indicate a normal simple expression.
1749 exception
1750 when Error_Resync =>
1751 Resync_Expression;
1752 Expr_Form := EF_Simple;
1753 return Error;
1754 end P_Relation;
1756 ----------------------------
1757 -- 4.4 Simple Expression --
1758 ----------------------------
1760 -- SIMPLE_EXPRESSION ::=
1761 -- [UNARY_ADDING_OPERATOR] TERM {BINARY_ADDING_OPERATOR TERM}
1763 -- On return, Expr_Form indicates the categorization of the expression
1765 -- Note: if Token = Tok_Apostrophe on return, then Expr_Form is set to
1766 -- EF_Simple_Name and the following token is RANGE (range attribute case).
1768 -- Error recovery: cannot raise Error_Resync. If an error occurs within an
1769 -- expression, then tokens are scanned until either a non-expression token,
1770 -- a right paren (not matched by a left paren) or a comma, is encountered.
1772 -- Note: P_Simple_Expression is called only internally by higher level
1773 -- expression routines. In cases in the grammar where a simple expression
1774 -- is required, the approach is to scan an expression, and then post an
1775 -- appropriate error message if the expression obtained is not simple. This
1776 -- gives better error recovery and treatment.
1778 function P_Simple_Expression return Node_Id is
1779 Scan_State : Saved_Scan_State;
1780 Node1 : Node_Id;
1781 Node2 : Node_Id;
1782 Tokptr : Source_Ptr;
1784 begin
1785 -- Check for cases starting with a name. There are two reasons for
1786 -- special casing. First speed things up by catching a common case
1787 -- without going through several routine layers. Second the caller must
1788 -- be informed via Expr_Form when the simple expression is a name.
1790 if Token in Token_Class_Name then
1791 Node1 := P_Name;
1793 -- Deal with apostrophe cases
1795 if Token = Tok_Apostrophe then
1796 Save_Scan_State (Scan_State); -- at apostrophe
1797 Scan; -- past apostrophe
1799 -- If qualified expression, scan it out and fall through
1801 if Token = Tok_Left_Paren then
1802 Node1 := P_Qualified_Expression (Node1);
1803 Expr_Form := EF_Simple;
1805 -- If range attribute, then we return with Token pointing to the
1806 -- apostrophe. Note: avoid the normal error check on exit. We
1807 -- know that the expression really is complete in this case!
1809 else -- Token = Tok_Range then
1810 Restore_Scan_State (Scan_State); -- to apostrophe
1811 Expr_Form := EF_Simple_Name;
1812 return Node1;
1813 end if;
1814 end if;
1816 -- If an expression terminator follows, the previous processing
1817 -- completely scanned out the expression (a common case), and
1818 -- left Expr_Form set appropriately for returning to our caller.
1820 if Token in Token_Class_Sterm then
1821 null;
1823 -- If we do not have an expression terminator, then complete the
1824 -- scan of a simple expression. This code duplicates the code
1825 -- found in P_Term and P_Factor.
1827 else
1828 if Token = Tok_Double_Asterisk then
1829 if Style_Check then
1830 Style.Check_Exponentiation_Operator;
1831 end if;
1833 Node2 := New_Op_Node (N_Op_Expon, Token_Ptr);
1834 Scan; -- past **
1835 Set_Left_Opnd (Node2, Node1);
1836 Set_Right_Opnd (Node2, P_Primary);
1837 Node1 := Node2;
1838 end if;
1840 loop
1841 exit when Token not in Token_Class_Mulop;
1842 Tokptr := Token_Ptr;
1843 Node2 := New_Op_Node (P_Multiplying_Operator, Tokptr);
1845 if Style_Check then
1846 Style.Check_Binary_Operator;
1847 end if;
1849 Scan; -- past operator
1850 Set_Left_Opnd (Node2, Node1);
1851 Set_Right_Opnd (Node2, P_Factor);
1852 Node1 := Node2;
1853 end loop;
1855 loop
1856 exit when Token not in Token_Class_Binary_Addop;
1857 Tokptr := Token_Ptr;
1858 Node2 := New_Op_Node (P_Binary_Adding_Operator, Tokptr);
1860 if Style_Check then
1861 Style.Check_Binary_Operator;
1862 end if;
1864 Scan; -- past operator
1865 Set_Left_Opnd (Node2, Node1);
1866 Set_Right_Opnd (Node2, P_Term);
1867 Node1 := Node2;
1868 end loop;
1870 Expr_Form := EF_Simple;
1871 end if;
1873 -- Cases where simple expression does not start with a name
1875 else
1876 -- Scan initial sign and initial Term
1878 if Token in Token_Class_Unary_Addop then
1879 Tokptr := Token_Ptr;
1880 Node1 := New_Op_Node (P_Unary_Adding_Operator, Tokptr);
1882 if Style_Check then
1883 Style.Check_Unary_Plus_Or_Minus;
1884 end if;
1886 Scan; -- past operator
1887 Set_Right_Opnd (Node1, P_Term);
1888 else
1889 Node1 := P_Term;
1890 end if;
1892 -- In the following, we special-case a sequence of concatenations of
1893 -- string literals, such as "aaa" & "bbb" & ... & "ccc", with nothing
1894 -- else mixed in. For such a sequence, we return a tree representing
1895 -- "" & "aaabbb...ccc" (a single concatenation). This is done only if
1896 -- the number of concatenations is large. If semantic analysis
1897 -- resolves the "&" to a predefined one, then this folding gives the
1898 -- right answer. Otherwise, semantic analysis will complain about a
1899 -- capacity-exceeded error. The purpose of this trick is to avoid
1900 -- creating a deeply nested tree, which would cause deep recursion
1901 -- during semantics, causing stack overflow. This way, we can handle
1902 -- enormous concatenations in the normal case of predefined "&". We
1903 -- first build up the normal tree, and then rewrite it if
1904 -- appropriate.
1906 declare
1907 Num_Concats_Threshold : constant Positive := 1000;
1908 -- Arbitrary threshold value to enable optimization
1910 First_Node : constant Node_Id := Node1;
1911 Is_Strlit_Concat : Boolean;
1912 -- True iff we've parsed a sequence of concatenations of string
1913 -- literals, with nothing else mixed in.
1915 Num_Concats : Natural;
1916 -- Number of "&" operators if Is_Strlit_Concat is True
1918 begin
1919 Is_Strlit_Concat :=
1920 Nkind (Node1) = N_String_Literal
1921 and then Token = Tok_Ampersand;
1922 Num_Concats := 0;
1924 -- Scan out sequence of terms separated by binary adding operators
1926 loop
1927 exit when Token not in Token_Class_Binary_Addop;
1928 Tokptr := Token_Ptr;
1929 Node2 := New_Op_Node (P_Binary_Adding_Operator, Tokptr);
1930 Scan; -- past operator
1931 Set_Left_Opnd (Node2, Node1);
1932 Node1 := P_Term;
1933 Set_Right_Opnd (Node2, Node1);
1935 -- Check if we're still concatenating string literals
1937 Is_Strlit_Concat :=
1938 Is_Strlit_Concat
1939 and then Nkind (Node2) = N_Op_Concat
1940 and then Nkind (Node1) = N_String_Literal;
1942 if Is_Strlit_Concat then
1943 Num_Concats := Num_Concats + 1;
1944 end if;
1946 Node1 := Node2;
1947 end loop;
1949 -- If we have an enormous series of concatenations of string
1950 -- literals, rewrite as explained above. The Is_Folded_In_Parser
1951 -- flag tells semantic analysis that if the "&" is not predefined,
1952 -- the folded value is wrong.
1954 if Is_Strlit_Concat
1955 and then Num_Concats >= Num_Concats_Threshold
1956 then
1957 declare
1958 Empty_String_Val : String_Id;
1959 -- String_Id for ""
1961 Strlit_Concat_Val : String_Id;
1962 -- Contains the folded value (which will be correct if the
1963 -- "&" operators are the predefined ones).
1965 Cur_Node : Node_Id;
1966 -- For walking up the tree
1968 New_Node : Node_Id;
1969 -- Folded node to replace Node1
1971 Loc : constant Source_Ptr := Sloc (First_Node);
1973 begin
1974 -- Walk up the tree starting at the leftmost string literal
1975 -- (First_Node), building up the Strlit_Concat_Val as we
1976 -- go. Note that we do not use recursion here -- the whole
1977 -- point is to avoid recursively walking that enormous tree.
1979 Start_String;
1980 Store_String_Chars (Strval (First_Node));
1982 Cur_Node := Parent (First_Node);
1983 while Present (Cur_Node) loop
1984 pragma Assert (Nkind (Cur_Node) = N_Op_Concat and then
1985 Nkind (Right_Opnd (Cur_Node)) = N_String_Literal);
1987 Store_String_Chars (Strval (Right_Opnd (Cur_Node)));
1988 Cur_Node := Parent (Cur_Node);
1989 end loop;
1991 Strlit_Concat_Val := End_String;
1993 -- Create new folded node, and rewrite result with a concat-
1994 -- enation of an empty string literal and the folded node.
1996 Start_String;
1997 Empty_String_Val := End_String;
1998 New_Node :=
1999 Make_Op_Concat (Loc,
2000 Make_String_Literal (Loc, Empty_String_Val),
2001 Make_String_Literal (Loc, Strlit_Concat_Val,
2002 Is_Folded_In_Parser => True));
2003 Rewrite (Node1, New_Node);
2004 end;
2005 end if;
2006 end;
2008 -- All done, we clearly do not have name or numeric literal so this
2009 -- is a case of a simple expression which is some other possibility.
2011 Expr_Form := EF_Simple;
2012 end if;
2014 -- Come here at end of simple expression, where we do a couple of
2015 -- special checks to improve error recovery.
2017 -- Special test to improve error recovery. If the current token
2018 -- is a period, then someone is trying to do selection on something
2019 -- that is not a name, e.g. a qualified expression.
2021 if Token = Tok_Dot then
2022 Error_Msg_SC ("prefix for selection is not a name");
2023 raise Error_Resync;
2024 end if;
2026 -- Special test to improve error recovery: If the current token is
2027 -- not the first token on a line (as determined by checking the
2028 -- previous token position with the start of the current line),
2029 -- then we insist that we have an appropriate terminating token.
2030 -- Consider the following two examples:
2032 -- 1) if A nad B then ...
2034 -- 2) A := B
2035 -- C := D
2037 -- In the first example, we would like to issue a binary operator
2038 -- expected message and resynchronize to the then. In the second
2039 -- example, we do not want to issue a binary operator message, so
2040 -- that instead we will get the missing semicolon message. This
2041 -- distinction is of course a heuristic which does not always work,
2042 -- but in practice it is quite effective.
2044 -- Note: the one case in which we do not go through this circuit is
2045 -- when we have scanned a range attribute and want to return with
2046 -- Token pointing to the apostrophe. The apostrophe is not normally
2047 -- an expression terminator, and is not in Token_Class_Sterm, but
2048 -- in this special case we know that the expression is complete.
2050 if not Token_Is_At_Start_Of_Line
2051 and then Token not in Token_Class_Sterm
2052 then
2053 -- Normally the right error message is indeed that we expected a
2054 -- binary operator, but in the case of being between a right and left
2055 -- paren, e.g. in an aggregate, a more likely error is missing comma.
2057 if Prev_Token = Tok_Right_Paren and then Token = Tok_Left_Paren then
2058 T_Comma;
2059 else
2060 Error_Msg_AP ("binary operator expected");
2061 end if;
2063 raise Error_Resync;
2065 else
2066 return Node1;
2067 end if;
2069 -- If any error occurs, then scan to next expression terminator symbol
2070 -- or comma, right paren or vertical bar at the outer (i.e. current) paren
2071 -- level. Expr_Form is set to indicate a normal simple expression.
2073 exception
2074 when Error_Resync =>
2075 Resync_Expression;
2076 Expr_Form := EF_Simple;
2077 return Error;
2078 end P_Simple_Expression;
2080 -----------------------------------------------
2081 -- 4.4 Simple Expression or Range Attribute --
2082 -----------------------------------------------
2084 -- SIMPLE_EXPRESSION ::=
2085 -- [UNARY_ADDING_OPERATOR] TERM {BINARY_ADDING_OPERATOR TERM}
2087 -- RANGE_ATTRIBUTE_REFERENCE ::= PREFIX ' RANGE_ATTRIBUTE_DESIGNATOR
2089 -- RANGE_ATTRIBUTE_DESIGNATOR ::= range [(static_EXPRESSION)]
2091 -- Error recovery: cannot raise Error_Resync
2093 function P_Simple_Expression_Or_Range_Attribute return Node_Id is
2094 Sexpr : Node_Id;
2095 Attr_Node : Node_Id;
2097 begin
2098 -- We don't just want to roar ahead and call P_Simple_Expression
2099 -- here, since we want to handle the case of a parenthesized range
2100 -- attribute cleanly.
2102 if Token = Tok_Left_Paren then
2103 declare
2104 Lptr : constant Source_Ptr := Token_Ptr;
2105 Scan_State : Saved_Scan_State;
2107 begin
2108 Save_Scan_State (Scan_State);
2109 Scan; -- past left paren
2110 Sexpr := P_Simple_Expression;
2112 if Token = Tok_Apostrophe then
2113 Attr_Node := P_Range_Attribute_Reference (Sexpr);
2114 Expr_Form := EF_Range_Attr;
2116 if Token = Tok_Right_Paren then
2117 Scan; -- scan past right paren if present
2118 end if;
2120 Error_Msg ("parentheses not allowed for range attribute", Lptr);
2122 return Attr_Node;
2123 end if;
2125 Restore_Scan_State (Scan_State);
2126 end;
2127 end if;
2129 -- Here after dealing with parenthesized range attribute
2131 Sexpr := P_Simple_Expression;
2133 if Token = Tok_Apostrophe then
2134 Attr_Node := P_Range_Attribute_Reference (Sexpr);
2135 Expr_Form := EF_Range_Attr;
2136 return Attr_Node;
2138 else
2139 return Sexpr;
2140 end if;
2141 end P_Simple_Expression_Or_Range_Attribute;
2143 ---------------
2144 -- 4.4 Term --
2145 ---------------
2147 -- TERM ::= FACTOR {MULTIPLYING_OPERATOR FACTOR}
2149 -- Error recovery: can raise Error_Resync
2151 function P_Term return Node_Id is
2152 Node1, Node2 : Node_Id;
2153 Tokptr : Source_Ptr;
2155 begin
2156 Node1 := P_Factor;
2158 loop
2159 exit when Token not in Token_Class_Mulop;
2160 Tokptr := Token_Ptr;
2161 Node2 := New_Op_Node (P_Multiplying_Operator, Tokptr);
2162 Scan; -- past operator
2163 Set_Left_Opnd (Node2, Node1);
2164 Set_Right_Opnd (Node2, P_Factor);
2165 Node1 := Node2;
2166 end loop;
2168 return Node1;
2169 end P_Term;
2171 -----------------
2172 -- 4.4 Factor --
2173 -----------------
2175 -- FACTOR ::= PRIMARY [** PRIMARY] | abs PRIMARY | not PRIMARY
2177 -- Error recovery: can raise Error_Resync
2179 function P_Factor return Node_Id is
2180 Node1 : Node_Id;
2181 Node2 : Node_Id;
2183 begin
2184 if Token = Tok_Abs then
2185 Node1 := New_Op_Node (N_Op_Abs, Token_Ptr);
2187 if Style_Check then
2188 Style.Check_Abs_Not;
2189 end if;
2191 Scan; -- past ABS
2192 Set_Right_Opnd (Node1, P_Primary);
2193 return Node1;
2195 elsif Token = Tok_Not then
2196 Node1 := New_Op_Node (N_Op_Not, Token_Ptr);
2198 if Style_Check then
2199 Style.Check_Abs_Not;
2200 end if;
2202 Scan; -- past NOT
2203 Set_Right_Opnd (Node1, P_Primary);
2204 return Node1;
2206 else
2207 Node1 := P_Primary;
2209 if Token = Tok_Double_Asterisk then
2210 Node2 := New_Op_Node (N_Op_Expon, Token_Ptr);
2211 Scan; -- past **
2212 Set_Left_Opnd (Node2, Node1);
2213 Set_Right_Opnd (Node2, P_Primary);
2214 return Node2;
2215 else
2216 return Node1;
2217 end if;
2218 end if;
2219 end P_Factor;
2221 ------------------
2222 -- 4.4 Primary --
2223 ------------------
2225 -- PRIMARY ::=
2226 -- NUMERIC_LITERAL | null
2227 -- | STRING_LITERAL | AGGREGATE
2228 -- | NAME | QUALIFIED_EXPRESSION
2229 -- | ALLOCATOR | (EXPRESSION)
2231 -- Error recovery: can raise Error_Resync
2233 function P_Primary return Node_Id is
2234 Scan_State : Saved_Scan_State;
2235 Node1 : Node_Id;
2237 begin
2238 -- The loop runs more than once only if misplaced pragmas are found
2240 loop
2241 case Token is
2243 -- Name token can start a name, call or qualified expression, all
2244 -- of which are acceptable possibilities for primary. Note also
2245 -- that string literal is included in name (as operator symbol)
2246 -- and type conversion is included in name (as indexed component).
2248 when Tok_Char_Literal | Tok_Operator_Symbol | Tok_Identifier =>
2249 Node1 := P_Name;
2251 -- All done unless apostrophe follows
2253 if Token /= Tok_Apostrophe then
2254 return Node1;
2256 -- Apostrophe following means that we have either just parsed
2257 -- the subtype mark of a qualified expression, or the prefix
2258 -- or a range attribute.
2260 else -- Token = Tok_Apostrophe
2261 Save_Scan_State (Scan_State); -- at apostrophe
2262 Scan; -- past apostrophe
2264 -- If range attribute, then this is always an error, since
2265 -- the only legitimate case (where the scanned expression is
2266 -- a qualified simple name) is handled at the level of the
2267 -- Simple_Expression processing. This case corresponds to a
2268 -- usage such as 3 + A'Range, which is always illegal.
2270 if Token = Tok_Range then
2271 Restore_Scan_State (Scan_State); -- to apostrophe
2272 Bad_Range_Attribute (Token_Ptr);
2273 return Error;
2275 -- If left paren, then we have a qualified expression.
2276 -- Note that P_Name guarantees that in this case, where
2277 -- Token = Tok_Apostrophe on return, the only two possible
2278 -- tokens following the apostrophe are left paren and
2279 -- RANGE, so we know we have a left paren here.
2281 else -- Token = Tok_Left_Paren
2282 return P_Qualified_Expression (Node1);
2284 end if;
2285 end if;
2287 -- Numeric or string literal
2289 when Tok_Integer_Literal |
2290 Tok_Real_Literal |
2291 Tok_String_Literal =>
2293 Node1 := Token_Node;
2294 Scan; -- past number
2295 return Node1;
2297 -- Left paren, starts aggregate or parenthesized expression
2299 when Tok_Left_Paren =>
2300 declare
2301 Expr : constant Node_Id := P_Aggregate_Or_Paren_Expr;
2303 begin
2304 if Nkind (Expr) = N_Attribute_Reference
2305 and then Attribute_Name (Expr) = Name_Range
2306 then
2307 Bad_Range_Attribute (Sloc (Expr));
2308 end if;
2310 return Expr;
2311 end;
2313 -- Allocator
2315 when Tok_New =>
2316 return P_Allocator;
2318 -- Null
2320 when Tok_Null =>
2321 Scan; -- past NULL
2322 return New_Node (N_Null, Prev_Token_Ptr);
2324 -- Pragma, not allowed here, so just skip past it
2326 when Tok_Pragma =>
2327 P_Pragmas_Misplaced;
2329 -- Deal with IF (possible unparenthesized conditional expression)
2331 when Tok_If =>
2333 -- If this looks like a real if, defined as an IF appearing at
2334 -- the start of a new line, then we consider we have a missing
2335 -- operand.
2337 if Token_Is_At_Start_Of_Line then
2338 Error_Msg_AP ("missing operand");
2339 return Error;
2341 -- If this looks like a conditional expression, then treat it
2342 -- that way with an error messasge.
2344 elsif Extensions_Allowed then
2345 Error_Msg_SC
2346 ("conditional expression must be parenthesized");
2347 return P_Conditional_Expression;
2349 -- Otherwise treat as misused identifier
2351 else
2352 return P_Identifier;
2353 end if;
2355 -- Anything else is illegal as the first token of a primary, but
2356 -- we test for a reserved identifier so that it is treated nicely
2358 when others =>
2359 if Is_Reserved_Identifier then
2360 return P_Identifier;
2362 elsif Prev_Token = Tok_Comma then
2363 Error_Msg_SP ("|extra "","" ignored");
2364 raise Error_Resync;
2366 else
2367 Error_Msg_AP ("missing operand");
2368 raise Error_Resync;
2369 end if;
2371 end case;
2372 end loop;
2373 end P_Primary;
2375 ---------------------------
2376 -- 4.5 Logical Operator --
2377 ---------------------------
2379 -- LOGICAL_OPERATOR ::= and | or | xor
2381 -- Note: AND THEN and OR ELSE are also treated as logical operators
2382 -- by the parser (even though they are not operators semantically)
2384 -- The value returned is the appropriate Node_Kind code for the operator
2385 -- On return, Token points to the token following the scanned operator.
2387 -- The caller has checked that the first token is a legitimate logical
2388 -- operator token (i.e. is either XOR, AND, OR).
2390 -- Error recovery: cannot raise Error_Resync
2392 function P_Logical_Operator return Node_Kind is
2393 begin
2394 if Token = Tok_And then
2395 if Style_Check then
2396 Style.Check_Binary_Operator;
2397 end if;
2399 Scan; -- past AND
2401 if Token = Tok_Then then
2402 Scan; -- past THEN
2403 return N_And_Then;
2404 else
2405 return N_Op_And;
2406 end if;
2408 elsif Token = Tok_Or then
2409 if Style_Check then
2410 Style.Check_Binary_Operator;
2411 end if;
2413 Scan; -- past OR
2415 if Token = Tok_Else then
2416 Scan; -- past ELSE
2417 return N_Or_Else;
2418 else
2419 return N_Op_Or;
2420 end if;
2422 else -- Token = Tok_Xor
2423 if Style_Check then
2424 Style.Check_Binary_Operator;
2425 end if;
2427 Scan; -- past XOR
2428 return N_Op_Xor;
2429 end if;
2430 end P_Logical_Operator;
2432 ------------------------------
2433 -- 4.5 Relational Operator --
2434 ------------------------------
2436 -- RELATIONAL_OPERATOR ::= = | /= | < | <= | > | >=
2438 -- The value returned is the appropriate Node_Kind code for the operator.
2439 -- On return, Token points to the operator token, NOT past it.
2441 -- The caller has checked that the first token is a legitimate relational
2442 -- operator token (i.e. is one of the operator tokens listed above).
2444 -- Error recovery: cannot raise Error_Resync
2446 function P_Relational_Operator return Node_Kind is
2447 Op_Kind : Node_Kind;
2448 Relop_Node : constant array (Token_Class_Relop) of Node_Kind :=
2449 (Tok_Less => N_Op_Lt,
2450 Tok_Equal => N_Op_Eq,
2451 Tok_Greater => N_Op_Gt,
2452 Tok_Not_Equal => N_Op_Ne,
2453 Tok_Greater_Equal => N_Op_Ge,
2454 Tok_Less_Equal => N_Op_Le,
2455 Tok_In => N_In,
2456 Tok_Not => N_Not_In,
2457 Tok_Box => N_Op_Ne);
2459 begin
2460 if Token = Tok_Box then
2461 Error_Msg_SC ("|""'<'>"" should be ""/=""");
2462 end if;
2464 Op_Kind := Relop_Node (Token);
2466 if Style_Check then
2467 Style.Check_Binary_Operator;
2468 end if;
2470 Scan; -- past operator token
2472 if Prev_Token = Tok_Not then
2473 T_In;
2474 end if;
2476 return Op_Kind;
2477 end P_Relational_Operator;
2479 ---------------------------------
2480 -- 4.5 Binary Adding Operator --
2481 ---------------------------------
2483 -- BINARY_ADDING_OPERATOR ::= + | - | &
2485 -- The value returned is the appropriate Node_Kind code for the operator.
2486 -- On return, Token points to the operator token (NOT past it).
2488 -- The caller has checked that the first token is a legitimate adding
2489 -- operator token (i.e. is one of the operator tokens listed above).
2491 -- Error recovery: cannot raise Error_Resync
2493 function P_Binary_Adding_Operator return Node_Kind is
2494 Addop_Node : constant array (Token_Class_Binary_Addop) of Node_Kind :=
2495 (Tok_Ampersand => N_Op_Concat,
2496 Tok_Minus => N_Op_Subtract,
2497 Tok_Plus => N_Op_Add);
2498 begin
2499 return Addop_Node (Token);
2500 end P_Binary_Adding_Operator;
2502 --------------------------------
2503 -- 4.5 Unary Adding Operator --
2504 --------------------------------
2506 -- UNARY_ADDING_OPERATOR ::= + | -
2508 -- The value returned is the appropriate Node_Kind code for the operator.
2509 -- On return, Token points to the operator token (NOT past it).
2511 -- The caller has checked that the first token is a legitimate adding
2512 -- operator token (i.e. is one of the operator tokens listed above).
2514 -- Error recovery: cannot raise Error_Resync
2516 function P_Unary_Adding_Operator return Node_Kind is
2517 Addop_Node : constant array (Token_Class_Unary_Addop) of Node_Kind :=
2518 (Tok_Minus => N_Op_Minus,
2519 Tok_Plus => N_Op_Plus);
2520 begin
2521 return Addop_Node (Token);
2522 end P_Unary_Adding_Operator;
2524 -------------------------------
2525 -- 4.5 Multiplying Operator --
2526 -------------------------------
2528 -- MULTIPLYING_OPERATOR ::= * | / | mod | rem
2530 -- The value returned is the appropriate Node_Kind code for the operator.
2531 -- On return, Token points to the operator token (NOT past it).
2533 -- The caller has checked that the first token is a legitimate multiplying
2534 -- operator token (i.e. is one of the operator tokens listed above).
2536 -- Error recovery: cannot raise Error_Resync
2538 function P_Multiplying_Operator return Node_Kind is
2539 Mulop_Node : constant array (Token_Class_Mulop) of Node_Kind :=
2540 (Tok_Asterisk => N_Op_Multiply,
2541 Tok_Mod => N_Op_Mod,
2542 Tok_Rem => N_Op_Rem,
2543 Tok_Slash => N_Op_Divide);
2544 begin
2545 return Mulop_Node (Token);
2546 end P_Multiplying_Operator;
2548 --------------------------------------
2549 -- 4.5 Highest Precedence Operator --
2550 --------------------------------------
2552 -- Parsed by P_Factor (4.4)
2554 -- Note: this rule is not in fact used by the grammar at any point!
2556 --------------------------
2557 -- 4.6 Type Conversion --
2558 --------------------------
2560 -- Parsed by P_Primary as a Name (4.1)
2562 -------------------------------
2563 -- 4.7 Qualified Expression --
2564 -------------------------------
2566 -- QUALIFIED_EXPRESSION ::=
2567 -- SUBTYPE_MARK ' (EXPRESSION) | SUBTYPE_MARK ' AGGREGATE
2569 -- The caller has scanned the name which is the Subtype_Mark parameter
2570 -- and scanned past the single quote following the subtype mark. The
2571 -- caller has not checked that this name is in fact appropriate for
2572 -- a subtype mark name (i.e. it is a selected component or identifier).
2574 -- Error_Recovery: cannot raise Error_Resync
2576 function P_Qualified_Expression (Subtype_Mark : Node_Id) return Node_Id is
2577 Qual_Node : Node_Id;
2578 begin
2579 Qual_Node := New_Node (N_Qualified_Expression, Prev_Token_Ptr);
2580 Set_Subtype_Mark (Qual_Node, Check_Subtype_Mark (Subtype_Mark));
2581 Set_Expression (Qual_Node, P_Aggregate_Or_Paren_Expr);
2582 return Qual_Node;
2583 end P_Qualified_Expression;
2585 --------------------
2586 -- 4.8 Allocator --
2587 --------------------
2589 -- ALLOCATOR ::=
2590 -- new [NULL_EXCLUSION] SUBTYPE_INDICATION | new QUALIFIED_EXPRESSION
2592 -- The caller has checked that the initial token is NEW
2594 -- Error recovery: can raise Error_Resync
2596 function P_Allocator return Node_Id is
2597 Alloc_Node : Node_Id;
2598 Type_Node : Node_Id;
2599 Null_Exclusion_Present : Boolean;
2601 begin
2602 Alloc_Node := New_Node (N_Allocator, Token_Ptr);
2603 T_New;
2605 -- Scan Null_Exclusion if present (Ada 2005 (AI-231))
2607 Null_Exclusion_Present := P_Null_Exclusion;
2608 Set_Null_Exclusion_Present (Alloc_Node, Null_Exclusion_Present);
2609 Type_Node := P_Subtype_Mark_Resync;
2611 if Token = Tok_Apostrophe then
2612 Scan; -- past apostrophe
2613 Set_Expression (Alloc_Node, P_Qualified_Expression (Type_Node));
2614 else
2615 Set_Expression
2616 (Alloc_Node,
2617 P_Subtype_Indication (Type_Node, Null_Exclusion_Present));
2618 end if;
2620 return Alloc_Node;
2621 end P_Allocator;
2623 ------------------------------
2624 -- P_Conditional_Expression --
2625 ------------------------------
2627 function P_Conditional_Expression return Node_Id is
2628 Exprs : constant List_Id := New_List;
2629 Loc : constant Source_Ptr := Token_Ptr;
2630 Expr : Node_Id;
2631 State : Saved_Scan_State;
2633 begin
2634 Inside_Conditional_Expression := Inside_Conditional_Expression + 1;
2636 if Token = Tok_If and then not Extensions_Allowed then
2637 Error_Msg_SC ("|conditional expression is an Ada extension");
2638 Error_Msg_SC ("\|use -gnatX switch to compile this unit");
2639 end if;
2641 Scan; -- past IF or ELSIF
2642 Append_To (Exprs, P_Expression_No_Right_Paren);
2643 TF_Then;
2644 Append_To (Exprs, P_Expression);
2646 -- We now have scanned out IF expr THEN expr
2648 -- Check for common error of semicolon before the ELSE
2650 if Token = Tok_Semicolon then
2651 Save_Scan_State (State);
2652 Scan; -- past semicolon
2654 if Token = Tok_Else or else Token = Tok_Elsif then
2655 Error_Msg_SP ("|extra "";"" ignored");
2657 else
2658 Restore_Scan_State (State);
2659 end if;
2660 end if;
2662 -- Scan out ELSIF sequence if present
2664 if Token = Tok_Elsif then
2665 Expr := P_Conditional_Expression;
2666 Set_Is_Elsif (Expr);
2667 Append_To (Exprs, Expr);
2669 -- Scan out ELSE phrase if present
2671 elsif Token = Tok_Else then
2673 -- Scan out ELSE expression
2675 Scan; -- Past ELSE
2676 Append_To (Exprs, P_Expression);
2678 -- Two expression case (implied True, filled in during semantics)
2680 else
2681 null;
2682 end if;
2684 -- If we have an END IF, diagnose as not needed
2686 if Token = Tok_End then
2687 Error_Msg_SC
2688 ("`END IF` not allowed at end of conditional expression");
2689 Scan; -- past END
2691 if Token = Tok_If then
2692 Scan; -- past IF;
2693 end if;
2694 end if;
2696 Inside_Conditional_Expression := Inside_Conditional_Expression - 1;
2698 -- Return the Conditional_Expression node
2700 return
2701 Make_Conditional_Expression (Loc,
2702 Expressions => Exprs);
2703 end P_Conditional_Expression;
2705 -----------------------
2706 -- P_Membership_Test --
2707 -----------------------
2709 procedure P_Membership_Test (N : Node_Id) is
2710 Alt : constant Node_Id :=
2711 P_Range_Or_Subtype_Mark
2712 (Allow_Simple_Expression => Extensions_Allowed);
2714 begin
2715 -- Set case
2717 if Token = Tok_Vertical_Bar then
2718 if not Extensions_Allowed then
2719 Error_Msg_SC ("set notation is a language extension");
2720 Error_Msg_SC ("\|use -gnatX switch to compile this unit");
2721 end if;
2723 Set_Alternatives (N, New_List (Alt));
2724 Set_Right_Opnd (N, Empty);
2726 -- Loop to accumulate alternatives
2728 while Token = Tok_Vertical_Bar loop
2729 Scan; -- past vertical bar
2730 Append_To
2731 (Alternatives (N),
2732 P_Range_Or_Subtype_Mark (Allow_Simple_Expression => True));
2733 end loop;
2735 -- Not set case
2737 else
2738 Set_Right_Opnd (N, Alt);
2739 Set_Alternatives (N, No_List);
2740 end if;
2741 end P_Membership_Test;
2743 end Ch4;