1 ------------------------------------------------------------------------------
3 -- GNAT COMPILER COMPONENTS --
9 -- Copyright (C) 1992-2010, Free Software Foundation, Inc. --
11 -- GNAT is free software; you can redistribute it and/or modify it under --
12 -- terms of the GNU General Public License as published by the Free Soft- --
13 -- ware Foundation; either version 3, or (at your option) any later ver- --
14 -- sion. GNAT is distributed in the hope that it will be useful, but WITH- --
15 -- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY --
16 -- or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License --
17 -- for more details. You should have received a copy of the GNU General --
18 -- Public License distributed with GNAT; see file COPYING3. If not, go to --
19 -- http://www.gnu.org/licenses for a complete copy of the license. --
21 -- GNAT was originally developed by the GNAT team at New York University. --
22 -- Extensive contributions were provided by Ada Core Technologies Inc. --
24 ------------------------------------------------------------------------------
26 with Csets; use Csets;
27 with Err_Vars; use Err_Vars;
28 with Hostparm; use Hostparm;
29 with Namet; use Namet;
31 with Scans; use Scans;
32 with Sinput; use Sinput;
33 with Snames; use Snames;
34 with Stringt; use Stringt;
35 with Stylesw; use Stylesw;
36 with Uintp; use Uintp;
37 with Urealp; use Urealp;
38 with Widechar; use Widechar;
40 pragma Warnings (Off);
41 -- This package is used also by gnatcoll
43 with System.UTF_32; use System.UTF_32;
44 with System.WCh_Con; use System.WCh_Con;
50 -- Make control characters visible
52 Special_Characters : array (Character) of Boolean := (others => False);
53 -- For characters that are Special token, the value is True
55 Comment_Is_Token : Boolean := False;
56 -- True if comments are tokens
58 End_Of_Line_Is_Token : Boolean := False;
59 -- True if End_Of_Line is a token
61 -----------------------
62 -- Local Subprograms --
63 -----------------------
65 procedure Accumulate_Token_Checksum;
66 pragma Inline (Accumulate_Token_Checksum);
68 procedure Accumulate_Checksum (C : Character);
69 pragma Inline (Accumulate_Checksum);
70 -- This routine accumulates the checksum given character C. During the
71 -- scanning of a source file, this routine is called with every character
72 -- in the source, excluding blanks, and all control characters (except
73 -- that ESC is included in the checksum). Upper case letters not in string
74 -- literals are folded by the caller. See Sinput spec for the documentation
75 -- of the checksum algorithm. Note: checksum values are only used if we
76 -- generate code, so it is not necessary to worry about making the right
77 -- sequence of calls in any error situation.
79 procedure Accumulate_Checksum (C : Char_Code);
80 pragma Inline (Accumulate_Checksum);
81 -- This version is identical, except that the argument, C, is a character
82 -- code value instead of a character. This is used when wide characters
83 -- are scanned. We use the character code rather than the ASCII characters
84 -- so that the checksum is independent of wide character encoding method.
86 procedure Initialize_Checksum;
87 pragma Inline (Initialize_Checksum);
88 -- Initialize checksum value
90 -------------------------
91 -- Accumulate_Checksum --
92 -------------------------
94 procedure Accumulate_Checksum (C : Character) is
96 System.CRC32.Update (System.CRC32.CRC32 (Checksum), C);
97 end Accumulate_Checksum;
99 procedure Accumulate_Checksum (C : Char_Code) is
102 Accumulate_Checksum (Character'Val (C / 2 ** 24));
103 Accumulate_Checksum (Character'Val ((C / 2 ** 16) mod 256));
104 Accumulate_Checksum (Character'Val ((C / 256) mod 256));
106 Accumulate_Checksum (Character'Val (C / 256));
109 Accumulate_Checksum (Character'Val (C mod 256));
110 end Accumulate_Checksum;
112 -------------------------------
113 -- Accumulate_Token_Checksum --
114 -------------------------------
116 procedure Accumulate_Token_Checksum is
119 (System.CRC32.CRC32 (Checksum),
120 Character'Val (Token_Type'Pos (Token)));
121 end Accumulate_Token_Checksum;
123 ----------------------------
124 -- Determine_Token_Casing --
125 ----------------------------
127 function Determine_Token_Casing return Casing_Type is
129 return Determine_Casing (Source (Token_Ptr .. Scan_Ptr - 1));
130 end Determine_Token_Casing;
132 -------------------------
133 -- Initialize_Checksum --
134 -------------------------
136 procedure Initialize_Checksum is
138 System.CRC32.Initialize (System.CRC32.CRC32 (Checksum));
139 end Initialize_Checksum;
141 ------------------------
142 -- Initialize_Scanner --
143 ------------------------
145 procedure Initialize_Scanner (Index : Source_File_Index) is
147 -- Establish reserved words
149 Scans.Initialize_Ada_Keywords;
151 -- Initialize scan control variables
153 Current_Source_File := Index;
154 Source := Source_Text (Current_Source_File);
155 Scan_Ptr := Source_First (Current_Source_File);
157 Token_Ptr := Scan_Ptr;
158 Current_Line_Start := Scan_Ptr;
160 Token_Name := No_Name;
161 Start_Column := Set_Start_Column;
162 First_Non_Blank_Location := Scan_Ptr;
165 Wide_Char_Byte_Count := 0;
167 -- Do not call Scan, otherwise the License stuff does not work in Scn
169 end Initialize_Scanner;
171 ------------------------------
172 -- Reset_Special_Characters --
173 ------------------------------
175 procedure Reset_Special_Characters is
177 Special_Characters := (others => False);
178 end Reset_Special_Characters;
186 Start_Of_Comment : Source_Ptr;
187 -- Record start of comment position
189 Underline_Found : Boolean;
190 -- During scanning of an identifier, set to True if last character
191 -- scanned was an underline or other punctuation character. This
192 -- is used to flag the error of two underlines/punctuations in a
193 -- row or ending an identifier with a underline/punctuation. Here
194 -- punctuation means any UTF_32 character in the Unicode category
195 -- Punctuation,Connector.
198 -- Used to remember start of last wide character scanned
200 procedure Check_End_Of_Line;
201 -- Called when end of line encountered. Checks that line is not too
202 -- long, and that other style checks for the end of line are met.
204 function Double_Char_Token (C : Character) return Boolean;
205 -- This function is used for double character tokens like := or <>. It
206 -- checks if the character following Source (Scan_Ptr) is C, and if so
207 -- bumps Scan_Ptr past the pair of characters and returns True. A space
208 -- between the two characters is also recognized with an appropriate
209 -- error message being issued. If C is not present, False is returned.
210 -- Note that Double_Char_Token can only be used for tokens defined in
211 -- the Ada syntax (it's use for error cases like && is not appropriate
212 -- since we do not want a junk message for a case like &-space-&).
214 procedure Error_Illegal_Character;
215 -- Give illegal character error, Scan_Ptr points to character. On
216 -- return, Scan_Ptr is bumped past the illegal character.
218 procedure Error_Illegal_Wide_Character;
219 -- Give illegal wide character message. On return, Scan_Ptr is bumped
220 -- past the illegal character, which may still leave us pointing to
221 -- junk, not much we can do if the escape sequence is messed up!
223 procedure Error_Long_Line;
224 -- Signal error of excessively long line
226 procedure Error_No_Double_Underline;
227 -- Signal error of two underline or punctuation characters in a row.
228 -- Called with Scan_Ptr pointing to second underline/punctuation char.
231 -- This is the procedure for scanning out numeric literals. On entry,
232 -- Scan_Ptr points to the digit that starts the numeric literal (the
233 -- checksum for this character has not been accumulated yet). On return
234 -- Scan_Ptr points past the last character of the numeric literal, Token
235 -- and Token_Node are set appropriately, and the checksum is updated.
238 -- This is the procedure for scanning out string literals. On entry,
239 -- Scan_Ptr points to the opening string quote (the checksum for this
240 -- character has not been accumulated yet). On return Scan_Ptr points
241 -- past the closing quote of the string literal, Token and Token_Node
242 -- are set appropriately, and the checksum is updated.
244 -----------------------
245 -- Check_End_Of_Line --
246 -----------------------
248 procedure Check_End_Of_Line is
249 Len : constant Int :=
251 Int (Current_Line_Start) -
252 Wide_Char_Byte_Count;
256 Style.Check_Line_Terminator (Len);
259 -- Deal with checking maximum line length
261 if Style_Check and Style_Check_Max_Line_Length then
262 Style.Check_Line_Max_Length (Len);
264 -- If style checking is inactive, check maximum line length against
267 elsif Len > Max_Line_Length then
271 -- Now one more checking circuit. Normally we are only enforcing a
272 -- limit of physical characters, with tabs counting as one character.
273 -- But if after tab expansion we would have a total line length that
274 -- exceeded 32766, that would really cause trouble, because column
275 -- positions would exceed the maximum we allow for a column count.
276 -- Note: the limit is 32766 rather than 32767, since we use a value
277 -- of 32767 for special purposes (see Sinput). Now we really do not
278 -- want to go messing with tabs in the normal case, so what we do is
279 -- to check for a line that has more than 4096 physical characters.
280 -- Any shorter line could not be a problem, even if it was all tabs.
289 Ptr := Current_Line_Start;
291 exit when Ptr = Scan_Ptr;
293 if Source (Ptr) = ASCII.HT then
294 Col := (Col - 1 + 8) / 8 * 8 + 1;
301 ("this line is longer than 32766 characters",
303 raise Unrecoverable_Error;
311 -- Reset wide character byte count for next line
313 Wide_Char_Byte_Count := 0;
314 end Check_End_Of_Line;
316 -----------------------
317 -- Double_Char_Token --
318 -----------------------
320 function Double_Char_Token (C : Character) return Boolean is
322 if Source (Scan_Ptr + 1) = C then
323 Accumulate_Checksum (C);
324 Scan_Ptr := Scan_Ptr + 2;
327 elsif Source (Scan_Ptr + 1) = ' '
328 and then Source (Scan_Ptr + 2) = C
330 Scan_Ptr := Scan_Ptr + 1;
331 Error_Msg_S -- CODEFIX
332 ("no space allowed here");
333 Scan_Ptr := Scan_Ptr + 2;
339 end Double_Char_Token;
341 -----------------------------
342 -- Error_Illegal_Character --
343 -----------------------------
345 procedure Error_Illegal_Character is
347 Error_Msg_S ("illegal character");
348 Scan_Ptr := Scan_Ptr + 1;
349 end Error_Illegal_Character;
351 ----------------------------------
352 -- Error_Illegal_Wide_Character --
353 ----------------------------------
355 procedure Error_Illegal_Wide_Character is
357 Scan_Ptr := Scan_Ptr + 1;
358 Error_Msg ("illegal wide character", Wptr);
359 end Error_Illegal_Wide_Character;
361 ---------------------
362 -- Error_Long_Line --
363 ---------------------
365 procedure Error_Long_Line is
368 ("this line is too long",
369 Current_Line_Start + Source_Ptr (Max_Line_Length));
372 -------------------------------
373 -- Error_No_Double_Underline --
374 -------------------------------
376 procedure Error_No_Double_Underline is
378 Underline_Found := False;
380 -- There are four cases, and we special case the messages
382 if Source (Scan_Ptr) = '_' then
383 if Source (Scan_Ptr - 1) = '_' then
384 Error_Msg_S -- CODEFIX
385 ("two consecutive underlines not permitted");
387 Error_Msg_S ("underline cannot follow punctuation character");
391 if Source (Scan_Ptr - 1) = '_' then
392 Error_Msg_S ("punctuation character cannot follow underline");
395 ("two consecutive punctuation characters not permitted");
398 end Error_No_Double_Underline;
407 -- Current source program character
409 Base_Char : Character;
410 -- Either # or : (character at start of based number)
416 -- Value of base in Uint format
419 -- Value of integer scanned by Scan_Integer in Uint format
422 -- Value of integer in numeric value being scanned
425 -- Scale value for real literal
428 -- Scale in Uint format
430 Exponent_Is_Negative : Boolean;
431 -- Set true for negative exponent
433 Extended_Digit_Value : Int;
434 -- Extended digit value
436 Point_Scanned : Boolean;
437 -- Flag for decimal point scanned in numeric literal
439 -----------------------
440 -- Local Subprograms --
441 -----------------------
443 procedure Error_Digit_Expected;
444 -- Signal error of bad digit, Scan_Ptr points to the location at
445 -- which the digit was expected on input, and is unchanged on return.
447 procedure Scan_Integer;
448 -- Procedure to scan integer literal. On entry, Scan_Ptr points to a
449 -- digit, on exit Scan_Ptr points past the last character of the
452 -- For each digit encountered, UI_Int_Value is multiplied by 10, and
453 -- the value of the digit added to the result. In addition, the
454 -- value in Scale is decremented by one for each actual digit
457 --------------------------
458 -- Error_Digit_Expected --
459 --------------------------
461 procedure Error_Digit_Expected is
463 Error_Msg_S ("digit expected");
464 end Error_Digit_Expected;
470 procedure Scan_Integer is
472 -- Next character scanned
475 C := Source (Scan_Ptr);
477 -- Loop through digits (allowing underlines)
480 Accumulate_Checksum (C);
482 UI_Int_Value * 10 + (Character'Pos (C) - Character'Pos ('0'));
483 Scan_Ptr := Scan_Ptr + 1;
485 C := Source (Scan_Ptr);
487 -- Case of underline encountered
491 -- We do not accumulate the '_' in the checksum, so that
492 -- 1_234 is equivalent to 1234, and does not trigger
493 -- compilation for "minimal recompilation" (gnatmake -m).
496 Scan_Ptr := Scan_Ptr + 1;
497 C := Source (Scan_Ptr);
499 Error_No_Double_Underline;
502 if C not in '0' .. '9' then
503 Error_Digit_Expected;
508 exit when C not in '0' .. '9';
513 -- Start of Processing for Nlit
518 UI_Int_Value := Uint_0;
519 Based_Literal_Uses_Colon := False;
522 Point_Scanned := False;
523 UI_Num_Value := UI_Int_Value;
525 -- Various possibilities now for continuing the literal are period,
526 -- E/e (for exponent), or :/# (for based literal).
529 C := Source (Scan_Ptr);
533 -- Scan out point, but do not scan past .. which is a range
534 -- sequence, and must not be eaten up scanning a numeric literal.
536 while C = '.' and then Source (Scan_Ptr + 1) /= '.' loop
537 Accumulate_Checksum ('.');
539 if Point_Scanned then
540 Error_Msg_S ("duplicate point ignored");
543 Point_Scanned := True;
544 Scan_Ptr := Scan_Ptr + 1;
545 C := Source (Scan_Ptr);
547 if C not in '0' .. '9' then
549 ("real literal cannot end with point", Scan_Ptr - 1);
552 UI_Num_Value := UI_Int_Value;
556 -- Based literal case. The base is the value we already scanned.
557 -- In the case of colon, we insist that the following character
558 -- is indeed an extended digit or a period. This catches a number
559 -- of common errors, as well as catching the well known tricky
560 -- bug otherwise arising from "x : integer range 1 .. 10:= 6;"
563 or else (C = ':' and then
564 (Source (Scan_Ptr + 1) = '.'
566 Source (Scan_Ptr + 1) in '0' .. '9'
568 Source (Scan_Ptr + 1) in 'A' .. 'Z'
570 Source (Scan_Ptr + 1) in 'a' .. 'z'))
572 Accumulate_Checksum (C);
574 UI_Base := UI_Int_Value;
576 if Base_Char = ':' then
577 Based_Literal_Uses_Colon := True;
580 if UI_Base < 2 or else UI_Base > 16 then
581 Error_Msg_SC ("base not 2-16");
585 Base := UI_To_Int (UI_Base);
586 Scan_Ptr := Scan_Ptr + 1;
588 -- Scan out extended integer [. integer]
590 C := Source (Scan_Ptr);
591 UI_Int_Value := Uint_0;
595 if C in '0' .. '9' then
596 Accumulate_Checksum (C);
597 Extended_Digit_Value :=
598 Int'(Character'Pos (C)) - Int'(Character'Pos ('0'));
600 elsif C in 'A' .. 'F' then
601 Accumulate_Checksum (Character'Val (Character'Pos (C) + 32));
602 Extended_Digit_Value :=
603 Int'(Character'Pos (C)) - Int'(Character'Pos ('A')) + 10;
605 elsif C in 'a' .. 'f' then
606 Accumulate_Checksum (C);
607 Extended_Digit_Value :=
608 Int'(Character'Pos (C)) - Int'(Character'Pos ('a')) + 10;
611 Error_Msg_S ("extended digit expected");
615 if Extended_Digit_Value >= Base then
616 Error_Msg_S ("digit '>= base");
619 UI_Int_Value := UI_Int_Value * UI_Base + Extended_Digit_Value;
621 Scan_Ptr := Scan_Ptr + 1;
622 C := Source (Scan_Ptr);
626 Accumulate_Checksum ('_');
627 Scan_Ptr := Scan_Ptr + 1;
628 C := Source (Scan_Ptr);
630 Error_No_Double_Underline;
634 Accumulate_Checksum ('.');
636 if Point_Scanned then
637 Error_Msg_S ("duplicate point ignored");
640 Scan_Ptr := Scan_Ptr + 1;
641 C := Source (Scan_Ptr);
642 Point_Scanned := True;
645 elsif C = Base_Char then
646 Accumulate_Checksum (C);
647 Scan_Ptr := Scan_Ptr + 1;
650 elsif C = '#' or else C = ':' then
651 Error_Msg_S ("based number delimiters must match");
652 Scan_Ptr := Scan_Ptr + 1;
655 elsif not Identifier_Char (C) then
656 if Base_Char = '#' then
657 Error_Msg_S -- CODEFIX
660 Error_Msg_S -- CODEFIX
669 UI_Num_Value := UI_Int_Value;
674 if not Point_Scanned then
678 UI_Scale := UI_From_Int (Scale);
681 if Source (Scan_Ptr) = 'e' or else Source (Scan_Ptr) = 'E' then
682 Accumulate_Checksum ('e');
683 Scan_Ptr := Scan_Ptr + 1;
684 Exponent_Is_Negative := False;
686 if Source (Scan_Ptr) = '+' then
687 Accumulate_Checksum ('+');
688 Scan_Ptr := Scan_Ptr + 1;
690 elsif Source (Scan_Ptr) = '-' then
691 Accumulate_Checksum ('-');
693 if not Point_Scanned then
695 ("negative exponent not allowed for integer literal");
697 Exponent_Is_Negative := True;
700 Scan_Ptr := Scan_Ptr + 1;
703 UI_Int_Value := Uint_0;
705 if Source (Scan_Ptr) in '0' .. '9' then
708 Error_Digit_Expected;
711 if Exponent_Is_Negative then
712 UI_Scale := UI_Scale - UI_Int_Value;
714 UI_Scale := UI_Scale + UI_Int_Value;
718 -- Case of real literal to be returned
720 if Point_Scanned then
721 Token := Tok_Real_Literal;
722 Real_Literal_Value :=
728 -- Case of integer literal to be returned
731 Token := Tok_Integer_Literal;
734 Int_Literal_Value := UI_Num_Value;
736 -- Avoid doing possibly expensive calculations in cases like
737 -- parsing 163E800_000# when semantics will not be done anyway.
738 -- This is especially useful when parsing garbled input.
740 elsif Operating_Mode /= Check_Syntax
741 and then (Serious_Errors_Detected = 0 or else Try_Semantics)
743 Int_Literal_Value := UI_Num_Value * UI_Base ** UI_Scale;
746 Int_Literal_Value := No_Uint;
750 Accumulate_Token_Checksum;
760 Delimiter : Character;
761 -- Delimiter (first character of string)
764 -- Current source program character
767 -- Current character code value
770 -- Error flag for Scan_Wide call
772 procedure Error_Bad_String_Char;
773 -- Signal bad character in string/character literal. On entry
774 -- Scan_Ptr points to the improper character encountered during the
775 -- scan. Scan_Ptr is not modified, so it still points to the bad
776 -- character on return.
778 procedure Error_Unterminated_String;
779 -- Procedure called if a line terminator character is encountered
780 -- during scanning a string, meaning that the string is not properly
783 procedure Set_String;
784 -- Procedure used to distinguish between string and operator symbol.
785 -- On entry the string has been scanned out, and its characters start
786 -- at Token_Ptr and end one character before Scan_Ptr. On exit Token
787 -- is set to Tok_String_Literal/Tok_Operator_Symbol as appropriate,
788 -- and Token_Node is appropriately initialized. In addition, in the
789 -- operator symbol case, Token_Name is appropriately set, and the
790 -- flags [Wide_]Wide_Character_Found are set appropriately.
792 ---------------------------
793 -- Error_Bad_String_Char --
794 ---------------------------
796 procedure Error_Bad_String_Char is
797 C : constant Character := Source (Scan_Ptr);
801 Error_Msg_S ("horizontal tab not allowed in string");
803 elsif C = VT or else C = FF then
804 Error_Msg_S ("format effector not allowed in string");
806 elsif C in Upper_Half_Character then
807 Error_Msg_S ("(Ada 83) upper half character not allowed");
810 Error_Msg_S ("control character not allowed in string");
812 end Error_Bad_String_Char;
814 -------------------------------
815 -- Error_Unterminated_String --
816 -------------------------------
818 procedure Error_Unterminated_String is
820 -- An interesting little refinement. Consider the following
823 -- A := "this is an unterminated string;
824 -- A := "this is an unterminated string &
825 -- P(A, "this is a parameter that didn't get terminated);
827 -- We fiddle a little to do slightly better placement in these
828 -- cases also if there is white space at the end of the line we
829 -- place the flag at the start of this white space, not at the
830 -- end. Note that we only have to test for blanks, since tabs
831 -- aren't allowed in strings in the first place and would have
832 -- caused an error message.
834 -- Two more cases that we treat specially are:
836 -- A := "this string uses the wrong terminator'
837 -- A := "this string uses the wrong terminator' &
839 -- In these cases we give a different error message as well
841 -- We actually reposition the scan pointer to the point where we
842 -- place the flag in these cases, since it seems a better bet on
843 -- the original intention.
845 while Source (Scan_Ptr - 1) = ' '
846 or else Source (Scan_Ptr - 1) = '&'
848 Scan_Ptr := Scan_Ptr - 1;
852 -- Check for case of incorrect string terminator, but single quote
853 -- is not considered incorrect if the opening terminator misused
854 -- a single quote (error message already given).
857 and then Source (Scan_Ptr - 1) = '''
861 ("incorrect string terminator character", Scan_Ptr - 1);
865 if Source (Scan_Ptr - 1) = ';' then
866 Scan_Ptr := Scan_Ptr - 1;
869 if Source (Scan_Ptr - 1) = ')' then
870 Scan_Ptr := Scan_Ptr - 1;
875 Error_Msg_S -- CODEFIX
876 ("missing string quote");
877 end Error_Unterminated_String;
883 procedure Set_String is
884 Slen : constant Int := Int (Scan_Ptr - Token_Ptr - 2);
890 -- Token_Name is currently set to Error_Name. The following
891 -- section of code resets Token_Name to the proper Name_Op_xx
892 -- value if the string is a valid operator symbol, otherwise it is
893 -- left set to Error_Name.
896 C1 := Source (Token_Ptr + 1);
900 Token_Name := Name_Op_Eq;
903 Token_Name := Name_Op_Gt;
906 Token_Name := Name_Op_Lt;
909 Token_Name := Name_Op_Add;
912 Token_Name := Name_Op_Subtract;
915 Token_Name := Name_Op_Concat;
918 Token_Name := Name_Op_Multiply;
921 Token_Name := Name_Op_Divide;
928 C1 := Source (Token_Ptr + 1);
929 C2 := Source (Token_Ptr + 2);
931 if C1 = '*' and then C2 = '*' then
932 Token_Name := Name_Op_Expon;
937 Token_Name := Name_Op_Ne;
939 Token_Name := Name_Op_Le;
941 Token_Name := Name_Op_Ge;
944 elsif (C1 = 'O' or else C1 = 'o') and then -- OR
945 (C2 = 'R' or else C2 = 'r')
947 Token_Name := Name_Op_Or;
951 C1 := Source (Token_Ptr + 1);
952 C2 := Source (Token_Ptr + 2);
953 C3 := Source (Token_Ptr + 3);
955 if (C1 = 'A' or else C1 = 'a') and then -- AND
956 (C2 = 'N' or else C2 = 'n') and then
957 (C3 = 'D' or else C3 = 'd')
959 Token_Name := Name_Op_And;
961 elsif (C1 = 'A' or else C1 = 'a') and then -- ABS
962 (C2 = 'B' or else C2 = 'b') and then
963 (C3 = 'S' or else C3 = 's')
965 Token_Name := Name_Op_Abs;
967 elsif (C1 = 'M' or else C1 = 'm') and then -- MOD
968 (C2 = 'O' or else C2 = 'o') and then
969 (C3 = 'D' or else C3 = 'd')
971 Token_Name := Name_Op_Mod;
973 elsif (C1 = 'N' or else C1 = 'n') and then -- NOT
974 (C2 = 'O' or else C2 = 'o') and then
975 (C3 = 'T' or else C3 = 't')
977 Token_Name := Name_Op_Not;
979 elsif (C1 = 'R' or else C1 = 'r') and then -- REM
980 (C2 = 'E' or else C2 = 'e') and then
981 (C3 = 'M' or else C3 = 'm')
983 Token_Name := Name_Op_Rem;
985 elsif (C1 = 'X' or else C1 = 'x') and then -- XOR
986 (C2 = 'O' or else C2 = 'o') and then
987 (C3 = 'R' or else C3 = 'r')
989 Token_Name := Name_Op_Xor;
994 -- If it is an operator symbol, then Token_Name is set. If it is
995 -- some other string value, then Token_Name still contains
998 if Token_Name = Error_Name then
999 Token := Tok_String_Literal;
1002 Token := Tok_Operator_Symbol;
1006 -- Start of processing for Slit
1009 -- On entry, Scan_Ptr points to the opening character of the string
1010 -- which is either a percent, double quote, or apostrophe (single
1011 -- quote). The latter case is an error detected by the character
1014 Delimiter := Source (Scan_Ptr);
1015 Accumulate_Checksum (Delimiter);
1018 Wide_Character_Found := False;
1019 Wide_Wide_Character_Found := False;
1020 Scan_Ptr := Scan_Ptr + 1;
1022 -- Loop to scan out characters of string literal
1025 C := Source (Scan_Ptr);
1027 if C = Delimiter then
1028 Accumulate_Checksum (C);
1029 Scan_Ptr := Scan_Ptr + 1;
1030 exit when Source (Scan_Ptr) /= Delimiter;
1031 Code := Get_Char_Code (C);
1032 Accumulate_Checksum (C);
1033 Scan_Ptr := Scan_Ptr + 1;
1036 if C = '"' and then Delimiter = '%' then
1038 ("quote not allowed in percent delimited string");
1039 Code := Get_Char_Code (C);
1040 Scan_Ptr := Scan_Ptr + 1;
1043 and then Wide_Character_Encoding_Method
1044 in WC_ESC_Encoding_Method)
1045 or else (C in Upper_Half_Character
1046 and then Upper_Half_Encoding)
1048 and then Source (Scan_Ptr + 1) = '"'
1049 and then Identifier_Char (Source (Scan_Ptr + 2)))
1052 Scan_Wide (Source, Scan_Ptr, Code, Err);
1055 Error_Illegal_Wide_Character;
1056 Code := Get_Char_Code (' ');
1059 Accumulate_Checksum (Code);
1061 -- In Ada 95 mode we allow any wide characters in a string
1062 -- but in Ada 2005, the set of characters allowed has been
1063 -- restricted to graphic characters.
1065 if Ada_Version >= Ada_05
1066 and then Is_UTF_32_Non_Graphic (UTF_32 (Code))
1069 ("(Ada 2005) non-graphic character not permitted " &
1070 "in string literal", Wptr);
1074 Accumulate_Checksum (C);
1076 if C not in Graphic_Character then
1077 if C in Line_Terminator then
1078 Error_Unterminated_String;
1081 elsif C in Upper_Half_Character then
1082 if Ada_Version = Ada_83 then
1083 Error_Bad_String_Char;
1087 Error_Bad_String_Char;
1091 Code := Get_Char_Code (C);
1092 Scan_Ptr := Scan_Ptr + 1;
1096 Store_String_Char (Code);
1098 if not In_Character_Range (Code) then
1099 if In_Wide_Character_Range (Code) then
1100 Wide_Character_Found := True;
1102 Wide_Wide_Character_Found := True;
1107 String_Literal_Id := End_String;
1112 -- Start of processing for Scan
1115 Prev_Token := Token;
1116 Prev_Token_Ptr := Token_Ptr;
1117 Token_Name := Error_Name;
1119 -- The following loop runs more than once only if a format effector
1120 -- (tab, vertical tab, form feed, line feed, carriage return) is
1121 -- encountered and skipped, or some error situation, such as an
1122 -- illegal character, is encountered.
1124 <<Scan_Next_Character>>
1127 -- Skip past blanks, loop is opened up for speed
1129 while Source (Scan_Ptr) = ' ' loop
1130 if Source (Scan_Ptr + 1) /= ' ' then
1131 Scan_Ptr := Scan_Ptr + 1;
1135 if Source (Scan_Ptr + 2) /= ' ' then
1136 Scan_Ptr := Scan_Ptr + 2;
1140 if Source (Scan_Ptr + 3) /= ' ' then
1141 Scan_Ptr := Scan_Ptr + 3;
1145 if Source (Scan_Ptr + 4) /= ' ' then
1146 Scan_Ptr := Scan_Ptr + 4;
1150 if Source (Scan_Ptr + 5) /= ' ' then
1151 Scan_Ptr := Scan_Ptr + 5;
1155 if Source (Scan_Ptr + 6) /= ' ' then
1156 Scan_Ptr := Scan_Ptr + 6;
1160 if Source (Scan_Ptr + 7) /= ' ' then
1161 Scan_Ptr := Scan_Ptr + 7;
1165 Scan_Ptr := Scan_Ptr + 8;
1168 -- We are now at a non-blank character, which is the first character
1169 -- of the token we will scan, and hence the value of Token_Ptr.
1171 Token_Ptr := Scan_Ptr;
1173 -- Here begins the main case statement which transfers control on the
1174 -- basis of the non-blank character we have encountered.
1176 case Source (Scan_Ptr) is
1178 -- Line terminator characters
1180 when CR | LF | FF | VT =>
1181 goto Scan_Line_Terminator;
1183 -- Horizontal tab, just skip past it
1190 Scan_Ptr := Scan_Ptr + 1;
1192 -- End of file character, treated as an end of file only if it is
1193 -- the last character in the buffer, otherwise it is ignored.
1196 if Scan_Ptr = Source_Last (Current_Source_File) then
1206 Scan_Ptr := Scan_Ptr + 1;
1212 Accumulate_Checksum ('&');
1214 if Source (Scan_Ptr + 1) = '&' then
1215 Error_Msg_S -- CODEFIX
1216 ("'&'& should be `AND THEN`");
1217 Scan_Ptr := Scan_Ptr + 2;
1222 Scan_Ptr := Scan_Ptr + 1;
1223 Token := Tok_Ampersand;
1227 -- Asterisk (can be multiplication operator or double asterisk which
1228 -- is the exponentiation compound delimiter).
1231 Accumulate_Checksum ('*');
1233 if Source (Scan_Ptr + 1) = '*' then
1234 Accumulate_Checksum ('*');
1235 Scan_Ptr := Scan_Ptr + 2;
1236 Token := Tok_Double_Asterisk;
1240 Scan_Ptr := Scan_Ptr + 1;
1241 Token := Tok_Asterisk;
1245 -- Colon, which can either be an isolated colon, or part of an
1246 -- assignment compound delimiter.
1249 Accumulate_Checksum (':');
1251 if Double_Char_Token ('=') then
1252 Token := Tok_Colon_Equal;
1255 Style.Check_Colon_Equal;
1260 elsif Source (Scan_Ptr + 1) = '-'
1261 and then Source (Scan_Ptr + 2) /= '-'
1263 Token := Tok_Colon_Equal;
1264 Error_Msg -- CODEFIX
1265 (":- should be :=", Scan_Ptr);
1266 Scan_Ptr := Scan_Ptr + 2;
1270 Scan_Ptr := Scan_Ptr + 1;
1283 Accumulate_Checksum ('(');
1284 Scan_Ptr := Scan_Ptr + 1;
1285 Token := Tok_Left_Paren;
1288 Style.Check_Left_Paren;
1296 if Source (Scan_Ptr + 1) = '"' then
1297 goto Scan_Wide_Character;
1300 Error_Msg_S ("illegal character, replaced by ""(""");
1301 Scan_Ptr := Scan_Ptr + 1;
1302 Token := Tok_Left_Paren;
1309 Error_Msg_S ("illegal character, replaced by ""(""");
1310 Scan_Ptr := Scan_Ptr + 1;
1311 Token := Tok_Left_Paren;
1317 Accumulate_Checksum (',');
1318 Scan_Ptr := Scan_Ptr + 1;
1327 -- Dot, which is either an isolated period, or part of a double dot
1328 -- compound delimiter sequence. We also check for the case of a
1329 -- digit following the period, to give a better error message.
1332 Accumulate_Checksum ('.');
1334 if Double_Char_Token ('.') then
1335 Token := Tok_Dot_Dot;
1338 Style.Check_Dot_Dot;
1343 elsif Source (Scan_Ptr + 1) in '0' .. '9' then
1344 Error_Msg_S ("numeric literal cannot start with point");
1345 Scan_Ptr := Scan_Ptr + 1;
1348 Scan_Ptr := Scan_Ptr + 1;
1353 -- Equal, which can either be an equality operator, or part of the
1354 -- arrow (=>) compound delimiter.
1357 Accumulate_Checksum ('=');
1359 if Double_Char_Token ('>') then
1368 elsif Source (Scan_Ptr + 1) = '=' then
1369 Error_Msg_S -- CODEFIX
1371 Scan_Ptr := Scan_Ptr + 1;
1374 Scan_Ptr := Scan_Ptr + 1;
1378 -- Greater than, which can be a greater than operator, greater than
1379 -- or equal operator, or first character of a right label bracket.
1382 Accumulate_Checksum ('>');
1384 if Double_Char_Token ('=') then
1385 Token := Tok_Greater_Equal;
1388 elsif Double_Char_Token ('>') then
1389 Token := Tok_Greater_Greater;
1393 Scan_Ptr := Scan_Ptr + 1;
1394 Token := Tok_Greater;
1398 -- Less than, which can be a less than operator, less than or equal
1399 -- operator, or the first character of a left label bracket, or the
1400 -- first character of a box (<>) compound delimiter.
1403 Accumulate_Checksum ('<');
1405 if Double_Char_Token ('=') then
1406 Token := Tok_Less_Equal;
1409 elsif Double_Char_Token ('>') then
1418 elsif Double_Char_Token ('<') then
1419 Token := Tok_Less_Less;
1423 Scan_Ptr := Scan_Ptr + 1;
1428 -- Minus, which is either a subtraction operator, or the first
1429 -- character of double minus starting a comment
1431 when '-' => Minus_Case : begin
1432 if Source (Scan_Ptr + 1) = '>' then
1433 Error_Msg_S ("invalid token");
1434 Scan_Ptr := Scan_Ptr + 2;
1438 elsif Source (Scan_Ptr + 1) /= '-' then
1439 Accumulate_Checksum ('-');
1440 Scan_Ptr := Scan_Ptr + 1;
1446 else -- Source (Scan_Ptr + 1) = '-' then
1448 Style.Check_Comment;
1451 Scan_Ptr := Scan_Ptr + 2;
1453 -- If we are in preprocessor mode with Replace_In_Comments set,
1454 -- then we return the "--" as a token on its own.
1456 if Replace_In_Comments then
1457 Token := Tok_Comment;
1461 -- Otherwise scan out the comment
1463 Start_Of_Comment := Scan_Ptr;
1465 -- Loop to scan comment (this loop runs more than once only if
1466 -- a horizontal tab or other non-graphic character is scanned)
1469 -- Scan to non graphic character (opened up for speed)
1471 -- Note that we just eat left brackets, which means that
1472 -- bracket notation cannot be used for end of line
1473 -- characters in comments. This seems a reasonable choice,
1474 -- since no one would ever use brackets notation in a real
1475 -- program in this situation, and if we allow brackets
1476 -- notation, we forbid some valid comments which contain a
1477 -- brackets sequence that happens to match an end of line
1481 exit when Source (Scan_Ptr) not in Graphic_Character;
1482 Scan_Ptr := Scan_Ptr + 1;
1483 exit when Source (Scan_Ptr) not in Graphic_Character;
1484 Scan_Ptr := Scan_Ptr + 1;
1485 exit when Source (Scan_Ptr) not in Graphic_Character;
1486 Scan_Ptr := Scan_Ptr + 1;
1487 exit when Source (Scan_Ptr) not in Graphic_Character;
1488 Scan_Ptr := Scan_Ptr + 1;
1489 exit when Source (Scan_Ptr) not in Graphic_Character;
1490 Scan_Ptr := Scan_Ptr + 1;
1493 -- Keep going if horizontal tab
1495 if Source (Scan_Ptr) = HT then
1500 Scan_Ptr := Scan_Ptr + 1;
1502 -- Terminate scan of comment if line terminator
1504 elsif Source (Scan_Ptr) in Line_Terminator then
1507 -- Terminate scan of comment if end of file encountered
1508 -- (embedded EOF character or real last character in file)
1510 elsif Source (Scan_Ptr) = EOF then
1513 -- If we have a wide character, we have to scan it out,
1514 -- because it might be a legitimate line terminator
1516 elsif (Source (Scan_Ptr) = ESC
1517 and then Identifier_Char (ESC))
1519 (Source (Scan_Ptr) in Upper_Half_Character
1520 and then Upper_Half_Encoding)
1523 Wptr : constant Source_Ptr := Scan_Ptr;
1528 Scan_Wide (Source, Scan_Ptr, Code, Err);
1530 -- If not well formed wide character, then just skip
1531 -- past it and ignore it.
1534 Scan_Ptr := Wptr + 1;
1536 -- If UTF_32 terminator, terminate comment scan
1538 elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
1544 -- Keep going if character in 80-FF range, or is ESC. These
1545 -- characters are allowed in comments by RM-2.1(1), 2.7(2).
1546 -- They are allowed even in Ada 83 mode according to the
1547 -- approved AI. ESC was added to the AI in June 93.
1549 elsif Source (Scan_Ptr) in Upper_Half_Character
1550 or else Source (Scan_Ptr) = ESC
1552 Scan_Ptr := Scan_Ptr + 1;
1554 -- Otherwise we have an illegal comment character
1557 Error_Illegal_Character;
1561 -- Note that, except when comments are tokens, we do NOT
1562 -- execute a return here, instead we fall through to reexecute
1563 -- the scan loop to look for a token.
1565 if Comment_Is_Token then
1566 Name_Len := Integer (Scan_Ptr - Start_Of_Comment);
1567 Name_Buffer (1 .. Name_Len) :=
1568 String (Source (Start_Of_Comment .. Scan_Ptr - 1));
1569 Comment_Id := Name_Find;
1570 Token := Tok_Comment;
1576 -- Double quote or percent starting a string literal
1583 -- Apostrophe. This can either be the start of a character literal,
1584 -- or an isolated apostrophe used in a qualified expression or an
1585 -- attribute. We treat it as a character literal if it does not
1586 -- follow a right parenthesis, identifier, the keyword ALL or
1587 -- a literal. This means that we correctly treat constructs like:
1589 -- A := CHARACTER'('A');
1591 -- Note that RM-2.2(7) does not require a separator between
1592 -- "CHARACTER" and "'" in the above.
1594 when ''' => Char_Literal_Case : declare
1599 Accumulate_Checksum (''');
1600 Scan_Ptr := Scan_Ptr + 1;
1602 -- Here is where we make the test to distinguish the cases. Treat
1603 -- as apostrophe if previous token is an identifier, right paren
1604 -- or the reserved word "all" (latter case as in A.all'Address)
1605 -- (or the reserved word "project" in project files). Also treat
1606 -- it as apostrophe after a literal (this catches some legitimate
1607 -- cases, like A."abs"'Address, and also gives better error
1608 -- behavior for impossible cases like 123'xxx).
1610 if Prev_Token = Tok_Identifier
1611 or else Prev_Token = Tok_Right_Paren
1612 or else Prev_Token = Tok_All
1613 or else Prev_Token = Tok_Project
1614 or else Prev_Token in Token_Class_Literal
1616 Token := Tok_Apostrophe;
1619 Style.Check_Apostrophe;
1624 -- Otherwise the apostrophe starts a character literal
1627 -- Case of wide character literal
1629 if (Source (Scan_Ptr) = ESC
1631 Wide_Character_Encoding_Method in WC_ESC_Encoding_Method)
1633 (Source (Scan_Ptr) in Upper_Half_Character
1635 Upper_Half_Encoding)
1637 (Source (Scan_Ptr) = '['
1639 Source (Scan_Ptr + 1) = '"')
1642 Scan_Wide (Source, Scan_Ptr, Code, Err);
1643 Accumulate_Checksum (Code);
1646 Error_Illegal_Wide_Character;
1647 Code := Character'Pos (' ');
1649 -- In Ada 95 mode we allow any wide character in a character
1650 -- literal, but in Ada 2005, the set of characters allowed
1651 -- is restricted to graphic characters.
1653 elsif Ada_Version >= Ada_05
1654 and then Is_UTF_32_Non_Graphic (UTF_32 (Code))
1656 Error_Msg -- CODEFIX????
1657 ("(Ada 2005) non-graphic character not permitted " &
1658 "in character literal", Wptr);
1661 if Source (Scan_Ptr) /= ''' then
1662 Error_Msg_S ("missing apostrophe");
1664 Scan_Ptr := Scan_Ptr + 1;
1667 -- If we do not find a closing quote in the expected place then
1668 -- assume that we have a misguided attempt at a string literal.
1670 -- However, if previous token is RANGE, then we return an
1671 -- apostrophe instead since this gives better error recovery
1673 elsif Source (Scan_Ptr + 1) /= ''' then
1674 if Prev_Token = Tok_Range then
1675 Token := Tok_Apostrophe;
1679 Scan_Ptr := Scan_Ptr - 1;
1681 ("strings are delimited by double quote character");
1687 -- Otherwise we have a (non-wide) character literal
1690 Accumulate_Checksum (Source (Scan_Ptr));
1692 if Source (Scan_Ptr) not in Graphic_Character then
1693 if Source (Scan_Ptr) in Upper_Half_Character then
1694 if Ada_Version = Ada_83 then
1695 Error_Illegal_Character;
1699 Error_Illegal_Character;
1703 Code := Get_Char_Code (Source (Scan_Ptr));
1704 Scan_Ptr := Scan_Ptr + 2;
1707 -- Fall through here with Scan_Ptr updated past the closing
1708 -- quote, and Code set to the Char_Code value for the literal
1710 Accumulate_Checksum (''');
1711 Token := Tok_Char_Literal;
1712 Set_Character_Literal_Name (Code);
1713 Token_Name := Name_Find;
1714 Character_Code := Code;
1718 end Char_Literal_Case;
1720 -- Right parenthesis
1723 Accumulate_Checksum (')');
1724 Scan_Ptr := Scan_Ptr + 1;
1725 Token := Tok_Right_Paren;
1728 Style.Check_Right_Paren;
1733 -- Right bracket or right brace, treated as right paren
1736 Error_Msg_S ("illegal character, replaced by "")""");
1737 Scan_Ptr := Scan_Ptr + 1;
1738 Token := Tok_Right_Paren;
1741 -- Slash (can be division operator or first character of not equal)
1744 Accumulate_Checksum ('/');
1746 if Double_Char_Token ('=') then
1747 Token := Tok_Not_Equal;
1750 Scan_Ptr := Scan_Ptr + 1;
1758 Accumulate_Checksum (';');
1759 Scan_Ptr := Scan_Ptr + 1;
1760 Token := Tok_Semicolon;
1763 Style.Check_Semicolon;
1770 when '|' => Vertical_Bar_Case : begin
1771 Accumulate_Checksum ('|');
1773 -- Special check for || to give nice message
1775 if Source (Scan_Ptr + 1) = '|' then
1776 Error_Msg_S -- CODEFIX
1777 ("""'|'|"" should be `OR ELSE`");
1778 Scan_Ptr := Scan_Ptr + 2;
1783 Scan_Ptr := Scan_Ptr + 1;
1784 Token := Tok_Vertical_Bar;
1787 Style.Check_Vertical_Bar;
1793 end Vertical_Bar_Case;
1795 -- Exclamation, replacement character for vertical bar
1797 when '!' => Exclamation_Case : begin
1798 Accumulate_Checksum ('!');
1800 if Source (Scan_Ptr + 1) = '=' then
1801 Error_Msg_S -- CODEFIX
1802 ("'!= should be /=");
1803 Scan_Ptr := Scan_Ptr + 2;
1804 Token := Tok_Not_Equal;
1808 Scan_Ptr := Scan_Ptr + 1;
1809 Token := Tok_Vertical_Bar;
1813 end Exclamation_Case;
1817 when '+' => Plus_Case : begin
1818 Accumulate_Checksum ('+');
1819 Scan_Ptr := Scan_Ptr + 1;
1824 -- Digits starting a numeric literal
1828 -- First a bit of a scan ahead to see if we have a case of an
1829 -- identifier starting with a digit (remembering exponent case).
1832 C : constant Character := Source (Scan_Ptr + 1);
1835 -- OK literal if digit followed by digit or underscore
1837 if C in '0' .. '9' or else C = '_' then
1840 -- OK literal if digit not followed by identifier char
1842 elsif not Identifier_Char (C) then
1845 -- OK literal if digit followed by e/E followed by digit/sign.
1846 -- We also allow underscore after the E, which is an error, but
1847 -- better handled by Nlit than deciding this is an identifier.
1849 elsif (C = 'e' or else C = 'E')
1850 and then (Source (Scan_Ptr + 2) in '0' .. '9'
1851 or else Source (Scan_Ptr + 2) = '+'
1852 or else Source (Scan_Ptr + 2) = '-'
1853 or else Source (Scan_Ptr + 2) = '_')
1857 -- Here we have what really looks like an identifier that
1858 -- starts with a digit, so give error msg.
1861 Error_Msg_S ("identifier may not start with digit");
1863 Underline_Found := False;
1864 Name_Buffer (1) := Source (Scan_Ptr);
1865 Accumulate_Checksum (Name_Buffer (1));
1866 Scan_Ptr := Scan_Ptr + 1;
1867 goto Scan_Identifier;
1871 -- Here we have an OK integer literal
1875 if Identifier_Char (Source (Scan_Ptr)) then
1877 ("delimiter required between literal and identifier");
1883 -- Lower case letters
1887 Underline_Found := False;
1888 Name_Buffer (1) := Source (Scan_Ptr);
1889 Accumulate_Checksum (Name_Buffer (1));
1890 Scan_Ptr := Scan_Ptr + 1;
1891 goto Scan_Identifier;
1893 -- Upper case letters
1897 Underline_Found := False;
1899 Character'Val (Character'Pos (Source (Scan_Ptr)) + 32);
1900 Accumulate_Checksum (Name_Buffer (1));
1901 Scan_Ptr := Scan_Ptr + 1;
1902 goto Scan_Identifier;
1904 -- Underline character
1907 if Special_Characters ('_') then
1908 Token_Ptr := Scan_Ptr;
1909 Scan_Ptr := Scan_Ptr + 1;
1910 Token := Tok_Special;
1911 Special_Character := '_';
1915 Error_Msg_S ("identifier cannot start with underline");
1917 Name_Buffer (1) := '_';
1918 Scan_Ptr := Scan_Ptr + 1;
1919 Underline_Found := False;
1920 goto Scan_Identifier;
1922 -- Space (not possible, because we scanned past blanks)
1925 raise Program_Error;
1927 -- Characters in top half of ASCII 8-bit chart
1929 when Upper_Half_Character =>
1931 -- Wide character case
1933 if Upper_Half_Encoding then
1934 goto Scan_Wide_Character;
1936 -- Otherwise we have OK Latin-1 character
1939 -- Upper half characters may possibly be identifier letters
1940 -- but can never be digits, so Identifier_Char can be used to
1941 -- test for a valid start of identifier character.
1943 if Identifier_Char (Source (Scan_Ptr)) then
1945 Underline_Found := False;
1946 goto Scan_Identifier;
1948 Error_Illegal_Character;
1954 -- ESC character, possible start of identifier if wide characters
1955 -- using ESC encoding are allowed in identifiers, which we can
1956 -- tell by looking at the Identifier_Char flag for ESC, which is
1957 -- only true if these conditions are met. In Ada 2005 mode, may
1958 -- also be valid UTF_32 space or line terminator character.
1960 if Identifier_Char (ESC) then
1962 goto Scan_Wide_Character;
1964 Error_Illegal_Character;
1967 -- Invalid control characters
1969 when NUL | SOH | STX | ETX | EOT | ENQ | ACK | BEL | BS | ASCII.SO |
1970 SI | DLE | DC1 | DC2 | DC3 | DC4 | NAK | SYN | ETB | CAN |
1971 EM | FS | GS | RS | US | DEL
1973 Error_Illegal_Character;
1975 -- Invalid graphic characters
1977 when '#' | '$' | '?' | '@' | '`' | '\' | '^' | '~' =>
1979 -- If Set_Special_Character has been called for this character,
1980 -- set Scans.Special_Character and return a Special token.
1982 if Special_Characters (Source (Scan_Ptr)) then
1983 Token_Ptr := Scan_Ptr;
1984 Token := Tok_Special;
1985 Special_Character := Source (Scan_Ptr);
1986 Scan_Ptr := Scan_Ptr + 1;
1989 -- Otherwise, this is an illegal character
1992 Error_Illegal_Character;
1995 -- End switch on non-blank character
1999 -- End loop past format effectors. The exit from this loop is by
2000 -- executing a return statement following completion of token scan
2001 -- (control never falls out of this loop to the code which follows)
2005 -- Wide_Character scanning routine. On entry we have encountered the
2006 -- initial character of a wide character sequence.
2008 <<Scan_Wide_Character>>
2017 Scan_Wide (Source, Scan_Ptr, Code, Err);
2019 -- If bad wide character, signal error and continue scan
2022 Error_Illegal_Wide_Character;
2023 goto Scan_Next_Character;
2026 Cat := Get_Category (UTF_32 (Code));
2028 -- If OK letter, reset scan ptr and go scan identifier
2030 if Is_UTF_32_Letter (Cat) then
2033 Underline_Found := False;
2034 goto Scan_Identifier;
2036 -- If OK wide space, ignore and keep scanning (we do not include
2037 -- any ignored spaces in checksum)
2039 elsif Is_UTF_32_Space (Cat) then
2040 goto Scan_Next_Character;
2042 -- If OK wide line terminator, terminate current line
2044 elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
2046 goto Scan_Line_Terminator;
2048 -- Punctuation is an error (at start of identifier)
2050 elsif Is_UTF_32_Punctuation (Cat) then
2051 Error_Msg ("identifier cannot start with punctuation", Wptr);
2054 Underline_Found := False;
2055 goto Scan_Identifier;
2057 -- Mark character is an error (at start of identifier)
2059 elsif Is_UTF_32_Mark (Cat) then
2060 Error_Msg ("identifier cannot start with mark character", Wptr);
2063 Underline_Found := False;
2064 goto Scan_Identifier;
2066 -- Other format character is an error (at start of identifier)
2068 elsif Is_UTF_32_Other (Cat) then
2070 ("identifier cannot start with other format character", Wptr);
2073 Underline_Found := False;
2074 goto Scan_Identifier;
2076 -- Extended digit character is an error. Could be bad start of
2077 -- identifier or bad literal. Not worth doing too much to try to
2078 -- distinguish these cases, but we will do a little bit.
2080 elsif Is_UTF_32_Digit (Cat) then
2082 ("identifier cannot start with digit character", Wptr);
2085 Underline_Found := False;
2086 goto Scan_Identifier;
2088 -- All other wide characters are illegal here
2091 Error_Illegal_Wide_Character;
2092 goto Scan_Next_Character;
2096 -- Routine to scan line terminator. On entry Scan_Ptr points to a
2097 -- character which is one of FF,LR,CR,VT, or one of the wide characters
2098 -- that is treated as a line terminator.
2100 <<Scan_Line_Terminator>>
2102 -- Check line too long
2106 -- Set Token_Ptr, if End_Of_Line is a token, for the case when it is
2109 if End_Of_Line_Is_Token then
2110 Token_Ptr := Scan_Ptr;
2117 Skip_Line_Terminators (Scan_Ptr, Physical);
2119 -- If we are at start of physical line, update scan pointers to
2120 -- reflect the start of the new line.
2123 Current_Line_Start := Scan_Ptr;
2124 Start_Column := Set_Start_Column;
2125 First_Non_Blank_Location := Scan_Ptr;
2127 -- If End_Of_Line is a token, we return it as it is a
2130 if End_Of_Line_Is_Token then
2131 Token := Tok_End_Of_Line;
2137 goto Scan_Next_Character;
2139 -- Identifier scanning routine. On entry, some initial characters of
2140 -- the identifier may have already been stored in Name_Buffer. If so,
2141 -- Name_Len has the number of characters stored, otherwise Name_Len is
2142 -- set to zero on entry. Underline_Found is also set False on entry.
2146 -- This loop scans as fast as possible past lower half letters and
2147 -- digits, which we expect to be the most common characters.
2150 if Source (Scan_Ptr) in 'a' .. 'z'
2151 or else Source (Scan_Ptr) in '0' .. '9'
2153 Name_Buffer (Name_Len + 1) := Source (Scan_Ptr);
2154 Accumulate_Checksum (Source (Scan_Ptr));
2156 elsif Source (Scan_Ptr) in 'A' .. 'Z' then
2157 Name_Buffer (Name_Len + 1) :=
2158 Character'Val (Character'Pos (Source (Scan_Ptr)) + 32);
2159 Accumulate_Checksum (Name_Buffer (Name_Len + 1));
2165 Underline_Found := False;
2166 Scan_Ptr := Scan_Ptr + 1;
2167 Name_Len := Name_Len + 1;
2170 -- If we fall through, then we have encountered either an underline
2171 -- character, or an extended identifier character (i.e. one from the
2172 -- upper half), or a wide character, or an identifier terminator. The
2173 -- initial test speeds us up in the most common case where we have
2174 -- an identifier terminator. Note that ESC is an identifier character
2175 -- only if a wide character encoding method that uses ESC encoding
2176 -- is active, so if we find an ESC character we know that we have a
2179 if Identifier_Char (Source (Scan_Ptr))
2180 or else (Source (Scan_Ptr) in Upper_Half_Character
2181 and then Upper_Half_Encoding)
2183 -- Case of underline
2185 if Source (Scan_Ptr) = '_' then
2186 Accumulate_Checksum ('_');
2188 if Underline_Found then
2189 Error_No_Double_Underline;
2191 Underline_Found := True;
2192 Name_Len := Name_Len + 1;
2193 Name_Buffer (Name_Len) := '_';
2196 Scan_Ptr := Scan_Ptr + 1;
2197 goto Scan_Identifier;
2199 -- Upper half character
2201 elsif Source (Scan_Ptr) in Upper_Half_Character
2202 and then not Upper_Half_Encoding
2204 Accumulate_Checksum (Source (Scan_Ptr));
2205 Store_Encoded_Character
2206 (Get_Char_Code (Fold_Lower (Source (Scan_Ptr))));
2207 Scan_Ptr := Scan_Ptr + 1;
2208 Underline_Found := False;
2209 goto Scan_Identifier;
2211 -- Left bracket not followed by a quote terminates an identifier.
2212 -- This is an error, but we don't want to give a junk error msg
2213 -- about wide characters in this case!
2215 elsif Source (Scan_Ptr) = '['
2216 and then Source (Scan_Ptr + 1) /= '"'
2220 -- We know we have a wide character encoding here (the current
2221 -- character is either ESC, left bracket, or an upper half
2222 -- character depending on the encoding method).
2225 -- Scan out the wide character and insert the appropriate
2226 -- encoding into the name table entry for the identifier.
2236 Scan_Wide (Source, Scan_Ptr, Code, Err);
2238 -- If error, signal error
2241 Error_Illegal_Wide_Character;
2243 -- If the character scanned is a normal identifier
2244 -- character, then we treat it that way.
2246 elsif In_Character_Range (Code)
2247 and then Identifier_Char (Get_Character (Code))
2249 Chr := Get_Character (Code);
2250 Accumulate_Checksum (Chr);
2251 Store_Encoded_Character
2252 (Get_Char_Code (Fold_Lower (Chr)));
2253 Underline_Found := False;
2255 -- Here if not a normal identifier character
2258 -- Make sure we are allowing wide characters in
2259 -- identifiers. Note that we allow wide character
2260 -- notation for an OK identifier character. This in
2261 -- particular allows bracket or other notation to be
2262 -- used for upper half letters.
2264 -- Wide characters are always allowed in Ada 2005
2266 if Identifier_Character_Set /= 'w'
2267 and then Ada_Version < Ada_05
2270 ("wide character not allowed in identifier", Wptr);
2273 Cat := Get_Category (UTF_32 (Code));
2275 -- If OK letter, store it folding to upper case. Note
2276 -- that we include the folded letter in the checksum.
2278 if Is_UTF_32_Letter (Cat) then
2280 Char_Code (UTF_32_To_Upper_Case (UTF_32 (Code)));
2281 Accumulate_Checksum (Code);
2282 Store_Encoded_Character (Code);
2283 Underline_Found := False;
2285 -- If OK extended digit or mark, then store it
2287 elsif Is_UTF_32_Digit (Cat)
2288 or else Is_UTF_32_Mark (Cat)
2290 Accumulate_Checksum (Code);
2291 Store_Encoded_Character (Code);
2292 Underline_Found := False;
2294 -- Wide punctuation is also stored, but counts as an
2295 -- underline character for error checking purposes.
2297 elsif Is_UTF_32_Punctuation (Cat) then
2298 Accumulate_Checksum (Code);
2300 if Underline_Found then
2302 Cend : constant Source_Ptr := Scan_Ptr;
2305 Error_No_Double_Underline;
2310 Store_Encoded_Character (Code);
2311 Underline_Found := True;
2314 -- Wide character in Unicode category "Other, Format"
2315 -- is accepted in an identifier, but is ignored and not
2316 -- stored. It seems reasonable to exclude it from the
2319 -- Note that it is correct (see AI-395) to simply strip
2320 -- other format characters, before testing for double
2321 -- underlines, or for reserved words).
2323 elsif Is_UTF_32_Other (Cat) then
2326 -- Wide character in category Separator,Space terminates
2328 elsif Is_UTF_32_Space (Cat) then
2329 goto Scan_Identifier_Complete;
2331 -- Any other wide character is not acceptable
2335 ("invalid wide character in identifier", Wptr);
2339 goto Scan_Identifier;
2344 -- Scan of identifier is complete. The identifier is stored in
2345 -- Name_Buffer, and Scan_Ptr points past the last character.
2347 <<Scan_Identifier_Complete>>
2348 Token_Name := Name_Find;
2350 -- Check for identifier ending with underline or punctuation char
2352 if Underline_Found then
2353 Underline_Found := False;
2355 if Source (Scan_Ptr - 1) = '_' then
2357 ("identifier cannot end with underline", Scan_Ptr - 1);
2360 ("identifier cannot end with punctuation character", Wptr);
2364 -- Here is where we check if it was a keyword
2366 if Is_Keyword_Name (Token_Name) then
2367 Token := Token_Type'Val (Get_Name_Table_Byte (Token_Name));
2369 -- Keyword style checks
2373 -- Deal with possible style check for non-lower case keyword,
2374 -- but we don't treat ACCESS, DELTA, DIGITS, RANGE as keywords
2375 -- for this purpose if they appear as attribute designators.
2376 -- Actually we only check the first character for speed.
2378 -- Ada 2005 (AI-284): Do not apply the style check in case of
2379 -- "pragma Interface"
2381 -- Ada 2005 (AI-340): Do not apply the style check in case of
2384 if Source (Token_Ptr) <= 'Z'
2385 and then (Prev_Token /= Tok_Apostrophe
2387 (Token /= Tok_Access and then
2388 Token /= Tok_Delta and then
2389 Token /= Tok_Digits and then
2390 Token /= Tok_Mod and then
2391 Token /= Tok_Range))
2392 and then (Token /= Tok_Interface
2394 (Token = Tok_Interface
2395 and then Prev_Token /= Tok_Pragma))
2397 Style.Non_Lower_Case_Keyword;
2400 -- Check THEN/ELSE style rules. These do not apply to AND THEN
2401 -- or OR ELSE, and do not apply in conditional expressions.
2403 if (Token = Tok_Then and then Prev_Token /= Tok_And)
2405 (Token = Tok_Else and then Prev_Token /= Tok_Or)
2407 if Inside_Conditional_Expression = 0 then
2408 Style.Check_Separate_Stmt_Lines;
2413 -- We must reset Token_Name since this is not an identifier and
2414 -- if we leave Token_Name set, the parser gets confused because
2415 -- it thinks it is dealing with an identifier instead of the
2416 -- corresponding keyword.
2418 Token_Name := No_Name;
2419 Accumulate_Token_Checksum;
2422 -- It is an identifier after all
2425 Token := Tok_Identifier;
2426 Accumulate_Token_Checksum;
2432 --------------------------
2433 -- Set_Comment_As_Token --
2434 --------------------------
2436 procedure Set_Comment_As_Token (Value : Boolean) is
2438 Comment_Is_Token := Value;
2439 end Set_Comment_As_Token;
2441 ------------------------------
2442 -- Set_End_Of_Line_As_Token --
2443 ------------------------------
2445 procedure Set_End_Of_Line_As_Token (Value : Boolean) is
2447 End_Of_Line_Is_Token := Value;
2448 end Set_End_Of_Line_As_Token;
2450 ---------------------------
2451 -- Set_Special_Character --
2452 ---------------------------
2454 procedure Set_Special_Character (C : Character) is
2457 when '#' | '$' | '_' | '?' | '@' | '`' | '\' | '^' | '~' =>
2458 Special_Characters (C) := True;
2463 end Set_Special_Character;
2465 ----------------------
2466 -- Set_Start_Column --
2467 ----------------------
2469 -- Note: it seems at first glance a little expensive to compute this value
2470 -- for every source line (since it is certainly not used for all source
2471 -- lines). On the other hand, it doesn't take much more work to skip past
2472 -- the initial white space on the line counting the columns than it would
2473 -- to scan past the white space using the standard scanning circuits.
2475 function Set_Start_Column return Column_Number is
2476 Start_Column : Column_Number := 0;
2479 -- Outer loop scans past horizontal tab characters
2483 -- Inner loop scans past blanks as fast as possible, bumping Scan_Ptr
2484 -- past the blanks and adjusting Start_Column to account for them.
2487 if Source (Scan_Ptr) = ' ' then
2488 if Source (Scan_Ptr + 1) = ' ' then
2489 if Source (Scan_Ptr + 2) = ' ' then
2490 if Source (Scan_Ptr + 3) = ' ' then
2491 if Source (Scan_Ptr + 4) = ' ' then
2492 if Source (Scan_Ptr + 5) = ' ' then
2493 if Source (Scan_Ptr + 6) = ' ' then
2494 Scan_Ptr := Scan_Ptr + 7;
2495 Start_Column := Start_Column + 7;
2497 Scan_Ptr := Scan_Ptr + 6;
2498 Start_Column := Start_Column + 6;
2502 Scan_Ptr := Scan_Ptr + 5;
2503 Start_Column := Start_Column + 5;
2507 Scan_Ptr := Scan_Ptr + 4;
2508 Start_Column := Start_Column + 4;
2512 Scan_Ptr := Scan_Ptr + 3;
2513 Start_Column := Start_Column + 3;
2517 Scan_Ptr := Scan_Ptr + 2;
2518 Start_Column := Start_Column + 2;
2522 Scan_Ptr := Scan_Ptr + 1;
2523 Start_Column := Start_Column + 1;
2529 end loop Blanks_Loop;
2531 -- Outer loop keeps going only if a horizontal tab follows
2533 if Source (Scan_Ptr) = HT then
2538 Scan_Ptr := Scan_Ptr + 1;
2539 Start_Column := (Start_Column / 8) * 8 + 8;
2545 return Start_Column;
2547 -- A constraint error can happen only if we have a compiler with checks on
2548 -- and a line with a ludicrous number of tabs or spaces at the start. In
2549 -- such a case, we really don't care if Start_Column is right or not.
2552 when Constraint_Error =>
2553 return Start_Column;
2554 end Set_Start_Column;