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 procedure Skip_Other_Format_Characters;
245 -- Skips past any "other format" category characters at the current
246 -- cursor location (does not skip past spaces or any other characters).
248 function Start_Of_Wide_Character return Boolean;
249 -- Returns True if the scan pointer is pointing to the start of a wide
250 -- character sequence, does not modify the scan pointer in any case.
252 -----------------------
253 -- Check_End_Of_Line --
254 -----------------------
256 procedure Check_End_Of_Line is
257 Len : constant Int :=
259 Int (Current_Line_Start) -
260 Wide_Char_Byte_Count;
264 Style.Check_Line_Terminator (Len);
267 -- Deal with checking maximum line length
269 if Style_Check and Style_Check_Max_Line_Length then
270 Style.Check_Line_Max_Length (Len);
272 -- If style checking is inactive, check maximum line length against
275 elsif Len > Max_Line_Length then
279 -- Now one more checking circuit. Normally we are only enforcing a
280 -- limit of physical characters, with tabs counting as one character.
281 -- But if after tab expansion we would have a total line length that
282 -- exceeded 32766, that would really cause trouble, because column
283 -- positions would exceed the maximum we allow for a column count.
284 -- Note: the limit is 32766 rather than 32767, since we use a value
285 -- of 32767 for special purposes (see Sinput). Now we really do not
286 -- want to go messing with tabs in the normal case, so what we do is
287 -- to check for a line that has more than 4096 physical characters.
288 -- Any shorter line could not be a problem, even if it was all tabs.
297 Ptr := Current_Line_Start;
299 exit when Ptr = Scan_Ptr;
301 if Source (Ptr) = ASCII.HT then
302 Col := (Col - 1 + 8) / 8 * 8 + 1;
309 ("this line is longer than 32766 characters",
311 raise Unrecoverable_Error;
319 -- Reset wide character byte count for next line
321 Wide_Char_Byte_Count := 0;
322 end Check_End_Of_Line;
324 -----------------------
325 -- Double_Char_Token --
326 -----------------------
328 function Double_Char_Token (C : Character) return Boolean is
330 if Source (Scan_Ptr + 1) = C then
331 Accumulate_Checksum (C);
332 Scan_Ptr := Scan_Ptr + 2;
335 elsif Source (Scan_Ptr + 1) = ' '
336 and then Source (Scan_Ptr + 2) = C
338 Scan_Ptr := Scan_Ptr + 1;
339 Error_Msg_S -- CODEFIX
340 ("no space allowed here");
341 Scan_Ptr := Scan_Ptr + 2;
347 end Double_Char_Token;
349 -----------------------------
350 -- Error_Illegal_Character --
351 -----------------------------
353 procedure Error_Illegal_Character is
355 Error_Msg_S ("illegal character");
356 Scan_Ptr := Scan_Ptr + 1;
357 end Error_Illegal_Character;
359 ----------------------------------
360 -- Error_Illegal_Wide_Character --
361 ----------------------------------
363 procedure Error_Illegal_Wide_Character is
365 Scan_Ptr := Scan_Ptr + 1;
366 Error_Msg ("illegal wide character", Wptr);
367 end Error_Illegal_Wide_Character;
369 ---------------------
370 -- Error_Long_Line --
371 ---------------------
373 procedure Error_Long_Line is
376 ("this line is too long",
377 Current_Line_Start + Source_Ptr (Max_Line_Length));
380 -------------------------------
381 -- Error_No_Double_Underline --
382 -------------------------------
384 procedure Error_No_Double_Underline is
386 Underline_Found := False;
388 -- There are four cases, and we special case the messages
390 if Source (Scan_Ptr) = '_' then
391 if Source (Scan_Ptr - 1) = '_' then
392 Error_Msg_S -- CODEFIX
393 ("two consecutive underlines not permitted");
395 Error_Msg_S ("underline cannot follow punctuation character");
399 if Source (Scan_Ptr - 1) = '_' then
400 Error_Msg_S ("punctuation character cannot follow underline");
403 ("two consecutive punctuation characters not permitted");
406 end Error_No_Double_Underline;
415 -- Current source program character
417 Base_Char : Character;
418 -- Either # or : (character at start of based number)
424 -- Value of base in Uint format
427 -- Value of integer scanned by Scan_Integer in Uint format
430 -- Value of integer in numeric value being scanned
433 -- Scale value for real literal
436 -- Scale in Uint format
438 Exponent_Is_Negative : Boolean;
439 -- Set true for negative exponent
441 Extended_Digit_Value : Int;
442 -- Extended digit value
444 Point_Scanned : Boolean;
445 -- Flag for decimal point scanned in numeric literal
447 -----------------------
448 -- Local Subprograms --
449 -----------------------
451 procedure Error_Digit_Expected;
452 -- Signal error of bad digit, Scan_Ptr points to the location at
453 -- which the digit was expected on input, and is unchanged on return.
455 procedure Scan_Integer;
456 -- Procedure to scan integer literal. On entry, Scan_Ptr points to a
457 -- digit, on exit Scan_Ptr points past the last character of the
460 -- For each digit encountered, UI_Int_Value is multiplied by 10, and
461 -- the value of the digit added to the result. In addition, the
462 -- value in Scale is decremented by one for each actual digit
465 --------------------------
466 -- Error_Digit_Expected --
467 --------------------------
469 procedure Error_Digit_Expected is
471 Error_Msg_S ("digit expected");
472 end Error_Digit_Expected;
478 procedure Scan_Integer is
480 -- Next character scanned
483 C := Source (Scan_Ptr);
485 -- Loop through digits (allowing underlines)
488 Accumulate_Checksum (C);
490 UI_Int_Value * 10 + (Character'Pos (C) - Character'Pos ('0'));
491 Scan_Ptr := Scan_Ptr + 1;
493 C := Source (Scan_Ptr);
495 -- Case of underline encountered
499 -- We do not accumulate the '_' in the checksum, so that
500 -- 1_234 is equivalent to 1234, and does not trigger
501 -- compilation for "minimal recompilation" (gnatmake -m).
504 Scan_Ptr := Scan_Ptr + 1;
505 C := Source (Scan_Ptr);
507 Error_No_Double_Underline;
510 if C not in '0' .. '9' then
511 Error_Digit_Expected;
516 exit when C not in '0' .. '9';
521 -- Start of Processing for Nlit
526 UI_Int_Value := Uint_0;
527 Based_Literal_Uses_Colon := False;
530 Point_Scanned := False;
531 UI_Num_Value := UI_Int_Value;
533 -- Various possibilities now for continuing the literal are period,
534 -- E/e (for exponent), or :/# (for based literal).
537 C := Source (Scan_Ptr);
541 -- Scan out point, but do not scan past .. which is a range
542 -- sequence, and must not be eaten up scanning a numeric literal.
544 while C = '.' and then Source (Scan_Ptr + 1) /= '.' loop
545 Accumulate_Checksum ('.');
547 if Point_Scanned then
548 Error_Msg_S ("duplicate point ignored");
551 Point_Scanned := True;
552 Scan_Ptr := Scan_Ptr + 1;
553 C := Source (Scan_Ptr);
555 if C not in '0' .. '9' then
557 ("real literal cannot end with point", Scan_Ptr - 1);
560 UI_Num_Value := UI_Int_Value;
564 -- Based literal case. The base is the value we already scanned.
565 -- In the case of colon, we insist that the following character
566 -- is indeed an extended digit or a period. This catches a number
567 -- of common errors, as well as catching the well known tricky
568 -- bug otherwise arising from "x : integer range 1 .. 10:= 6;"
571 or else (C = ':' and then
572 (Source (Scan_Ptr + 1) = '.'
574 Source (Scan_Ptr + 1) in '0' .. '9'
576 Source (Scan_Ptr + 1) in 'A' .. 'Z'
578 Source (Scan_Ptr + 1) in 'a' .. 'z'))
580 Accumulate_Checksum (C);
582 UI_Base := UI_Int_Value;
584 if Base_Char = ':' then
585 Based_Literal_Uses_Colon := True;
588 if UI_Base < 2 or else UI_Base > 16 then
589 Error_Msg_SC ("base not 2-16");
593 Base := UI_To_Int (UI_Base);
594 Scan_Ptr := Scan_Ptr + 1;
596 -- Scan out extended integer [. integer]
598 C := Source (Scan_Ptr);
599 UI_Int_Value := Uint_0;
603 if C in '0' .. '9' then
604 Accumulate_Checksum (C);
605 Extended_Digit_Value :=
606 Int'(Character'Pos (C)) - Int'(Character'Pos ('0'));
608 elsif C in 'A' .. 'F' then
609 Accumulate_Checksum (Character'Val (Character'Pos (C) + 32));
610 Extended_Digit_Value :=
611 Int'(Character'Pos (C)) - Int'(Character'Pos ('A')) + 10;
613 elsif C in 'a' .. 'f' then
614 Accumulate_Checksum (C);
615 Extended_Digit_Value :=
616 Int'(Character'Pos (C)) - Int'(Character'Pos ('a')) + 10;
619 Error_Msg_S ("extended digit expected");
623 if Extended_Digit_Value >= Base then
624 Error_Msg_S ("digit '>= base");
627 UI_Int_Value := UI_Int_Value * UI_Base + Extended_Digit_Value;
629 Scan_Ptr := Scan_Ptr + 1;
630 C := Source (Scan_Ptr);
634 Accumulate_Checksum ('_');
635 Scan_Ptr := Scan_Ptr + 1;
636 C := Source (Scan_Ptr);
638 Error_No_Double_Underline;
642 Accumulate_Checksum ('.');
644 if Point_Scanned then
645 Error_Msg_S ("duplicate point ignored");
648 Scan_Ptr := Scan_Ptr + 1;
649 C := Source (Scan_Ptr);
650 Point_Scanned := True;
653 elsif C = Base_Char then
654 Accumulate_Checksum (C);
655 Scan_Ptr := Scan_Ptr + 1;
658 elsif C = '#' or else C = ':' then
659 Error_Msg_S ("based number delimiters must match");
660 Scan_Ptr := Scan_Ptr + 1;
663 elsif not Identifier_Char (C) then
664 if Base_Char = '#' then
665 Error_Msg_S -- CODEFIX
668 Error_Msg_S -- CODEFIX
677 UI_Num_Value := UI_Int_Value;
682 if not Point_Scanned then
686 UI_Scale := UI_From_Int (Scale);
689 if Source (Scan_Ptr) = 'e' or else Source (Scan_Ptr) = 'E' then
690 Accumulate_Checksum ('e');
691 Scan_Ptr := Scan_Ptr + 1;
692 Exponent_Is_Negative := False;
694 if Source (Scan_Ptr) = '+' then
695 Accumulate_Checksum ('+');
696 Scan_Ptr := Scan_Ptr + 1;
698 elsif Source (Scan_Ptr) = '-' then
699 Accumulate_Checksum ('-');
701 if not Point_Scanned then
703 ("negative exponent not allowed for integer literal");
705 Exponent_Is_Negative := True;
708 Scan_Ptr := Scan_Ptr + 1;
711 UI_Int_Value := Uint_0;
713 if Source (Scan_Ptr) in '0' .. '9' then
716 Error_Digit_Expected;
719 if Exponent_Is_Negative then
720 UI_Scale := UI_Scale - UI_Int_Value;
722 UI_Scale := UI_Scale + UI_Int_Value;
726 -- Case of real literal to be returned
728 if Point_Scanned then
729 Token := Tok_Real_Literal;
730 Real_Literal_Value :=
736 -- Case of integer literal to be returned
739 Token := Tok_Integer_Literal;
742 Int_Literal_Value := UI_Num_Value;
744 -- Avoid doing possibly expensive calculations in cases like
745 -- parsing 163E800_000# when semantics will not be done anyway.
746 -- This is especially useful when parsing garbled input.
748 elsif Operating_Mode /= Check_Syntax
749 and then (Serious_Errors_Detected = 0 or else Try_Semantics)
751 Int_Literal_Value := UI_Num_Value * UI_Base ** UI_Scale;
754 Int_Literal_Value := No_Uint;
758 Accumulate_Token_Checksum;
768 Delimiter : Character;
769 -- Delimiter (first character of string)
772 -- Current source program character
775 -- Current character code value
778 -- Error flag for Scan_Wide call
780 procedure Error_Bad_String_Char;
781 -- Signal bad character in string/character literal. On entry
782 -- Scan_Ptr points to the improper character encountered during the
783 -- scan. Scan_Ptr is not modified, so it still points to the bad
784 -- character on return.
786 procedure Error_Unterminated_String;
787 -- Procedure called if a line terminator character is encountered
788 -- during scanning a string, meaning that the string is not properly
791 procedure Set_String;
792 -- Procedure used to distinguish between string and operator symbol.
793 -- On entry the string has been scanned out, and its characters start
794 -- at Token_Ptr and end one character before Scan_Ptr. On exit Token
795 -- is set to Tok_String_Literal/Tok_Operator_Symbol as appropriate,
796 -- and Token_Node is appropriately initialized. In addition, in the
797 -- operator symbol case, Token_Name is appropriately set, and the
798 -- flags [Wide_]Wide_Character_Found are set appropriately.
800 ---------------------------
801 -- Error_Bad_String_Char --
802 ---------------------------
804 procedure Error_Bad_String_Char is
805 C : constant Character := Source (Scan_Ptr);
809 Error_Msg_S ("horizontal tab not allowed in string");
811 elsif C = VT or else C = FF then
812 Error_Msg_S ("format effector not allowed in string");
814 elsif C in Upper_Half_Character then
815 Error_Msg_S ("(Ada 83) upper half character not allowed");
818 Error_Msg_S ("control character not allowed in string");
820 end Error_Bad_String_Char;
822 -------------------------------
823 -- Error_Unterminated_String --
824 -------------------------------
826 procedure Error_Unterminated_String is
828 -- An interesting little refinement. Consider the following
831 -- A := "this is an unterminated string;
832 -- A := "this is an unterminated string &
833 -- P(A, "this is a parameter that didn't get terminated);
835 -- We fiddle a little to do slightly better placement in these
836 -- cases also if there is white space at the end of the line we
837 -- place the flag at the start of this white space, not at the
838 -- end. Note that we only have to test for blanks, since tabs
839 -- aren't allowed in strings in the first place and would have
840 -- caused an error message.
842 -- Two more cases that we treat specially are:
844 -- A := "this string uses the wrong terminator'
845 -- A := "this string uses the wrong terminator' &
847 -- In these cases we give a different error message as well
849 -- We actually reposition the scan pointer to the point where we
850 -- place the flag in these cases, since it seems a better bet on
851 -- the original intention.
853 while Source (Scan_Ptr - 1) = ' '
854 or else Source (Scan_Ptr - 1) = '&'
856 Scan_Ptr := Scan_Ptr - 1;
860 -- Check for case of incorrect string terminator, but single quote
861 -- is not considered incorrect if the opening terminator misused
862 -- a single quote (error message already given).
865 and then Source (Scan_Ptr - 1) = '''
869 ("incorrect string terminator character", Scan_Ptr - 1);
873 if Source (Scan_Ptr - 1) = ';' then
874 Scan_Ptr := Scan_Ptr - 1;
877 if Source (Scan_Ptr - 1) = ')' then
878 Scan_Ptr := Scan_Ptr - 1;
883 Error_Msg_S -- CODEFIX
884 ("missing string quote");
885 end Error_Unterminated_String;
891 procedure Set_String is
892 Slen : constant Int := Int (Scan_Ptr - Token_Ptr - 2);
898 -- Token_Name is currently set to Error_Name. The following
899 -- section of code resets Token_Name to the proper Name_Op_xx
900 -- value if the string is a valid operator symbol, otherwise it is
901 -- left set to Error_Name.
904 C1 := Source (Token_Ptr + 1);
908 Token_Name := Name_Op_Eq;
911 Token_Name := Name_Op_Gt;
914 Token_Name := Name_Op_Lt;
917 Token_Name := Name_Op_Add;
920 Token_Name := Name_Op_Subtract;
923 Token_Name := Name_Op_Concat;
926 Token_Name := Name_Op_Multiply;
929 Token_Name := Name_Op_Divide;
936 C1 := Source (Token_Ptr + 1);
937 C2 := Source (Token_Ptr + 2);
939 if C1 = '*' and then C2 = '*' then
940 Token_Name := Name_Op_Expon;
945 Token_Name := Name_Op_Ne;
947 Token_Name := Name_Op_Le;
949 Token_Name := Name_Op_Ge;
952 elsif (C1 = 'O' or else C1 = 'o') and then -- OR
953 (C2 = 'R' or else C2 = 'r')
955 Token_Name := Name_Op_Or;
959 C1 := Source (Token_Ptr + 1);
960 C2 := Source (Token_Ptr + 2);
961 C3 := Source (Token_Ptr + 3);
963 if (C1 = 'A' or else C1 = 'a') and then -- AND
964 (C2 = 'N' or else C2 = 'n') and then
965 (C3 = 'D' or else C3 = 'd')
967 Token_Name := Name_Op_And;
969 elsif (C1 = 'A' or else C1 = 'a') and then -- ABS
970 (C2 = 'B' or else C2 = 'b') and then
971 (C3 = 'S' or else C3 = 's')
973 Token_Name := Name_Op_Abs;
975 elsif (C1 = 'M' or else C1 = 'm') and then -- MOD
976 (C2 = 'O' or else C2 = 'o') and then
977 (C3 = 'D' or else C3 = 'd')
979 Token_Name := Name_Op_Mod;
981 elsif (C1 = 'N' or else C1 = 'n') and then -- NOT
982 (C2 = 'O' or else C2 = 'o') and then
983 (C3 = 'T' or else C3 = 't')
985 Token_Name := Name_Op_Not;
987 elsif (C1 = 'R' or else C1 = 'r') and then -- REM
988 (C2 = 'E' or else C2 = 'e') and then
989 (C3 = 'M' or else C3 = 'm')
991 Token_Name := Name_Op_Rem;
993 elsif (C1 = 'X' or else C1 = 'x') and then -- XOR
994 (C2 = 'O' or else C2 = 'o') and then
995 (C3 = 'R' or else C3 = 'r')
997 Token_Name := Name_Op_Xor;
1002 -- If it is an operator symbol, then Token_Name is set. If it is
1003 -- some other string value, then Token_Name still contains
1006 if Token_Name = Error_Name then
1007 Token := Tok_String_Literal;
1010 Token := Tok_Operator_Symbol;
1014 -- Start of processing for Slit
1017 -- On entry, Scan_Ptr points to the opening character of the string
1018 -- which is either a percent, double quote, or apostrophe (single
1019 -- quote). The latter case is an error detected by the character
1022 Delimiter := Source (Scan_Ptr);
1023 Accumulate_Checksum (Delimiter);
1026 Wide_Character_Found := False;
1027 Wide_Wide_Character_Found := False;
1028 Scan_Ptr := Scan_Ptr + 1;
1030 -- Loop to scan out characters of string literal
1033 C := Source (Scan_Ptr);
1035 if C = Delimiter then
1036 Accumulate_Checksum (C);
1037 Scan_Ptr := Scan_Ptr + 1;
1038 exit when Source (Scan_Ptr) /= Delimiter;
1039 Code := Get_Char_Code (C);
1040 Accumulate_Checksum (C);
1041 Scan_Ptr := Scan_Ptr + 1;
1044 if C = '"' and then Delimiter = '%' then
1046 ("quote not allowed in percent delimited string");
1047 Code := Get_Char_Code (C);
1048 Scan_Ptr := Scan_Ptr + 1;
1050 elsif Start_Of_Wide_Character then
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_2005
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 ----------------------------------
1113 -- Skip_Other_Format_Characters --
1114 ----------------------------------
1116 procedure Skip_Other_Format_Characters is
1122 while Start_Of_Wide_Character loop
1124 Scan_Wide (Source, Scan_Ptr, Code, Err);
1126 if not Is_UTF_32_Other (UTF_32 (Code)) then
1131 end Skip_Other_Format_Characters;
1133 -----------------------------
1134 -- Start_Of_Wide_Character --
1135 -----------------------------
1137 function Start_Of_Wide_Character return Boolean is
1138 C : constant Character := Source (Scan_Ptr);
1141 -- ESC encoding method with ESC present
1144 and then Wide_Character_Encoding_Method in WC_ESC_Encoding_Method
1148 -- Upper half character with upper half encoding
1150 elsif C in Upper_Half_Character and then Upper_Half_Encoding then
1153 -- Brackets encoding
1156 and then Source (Scan_Ptr + 1) = '"'
1157 and then Identifier_Char (Source (Scan_Ptr + 2))
1161 -- Not the start of a wide character
1166 end Start_Of_Wide_Character;
1168 -- Start of processing for Scan
1171 Prev_Token := Token;
1172 Prev_Token_Ptr := Token_Ptr;
1173 Token_Name := Error_Name;
1175 -- The following loop runs more than once only if a format effector
1176 -- (tab, vertical tab, form feed, line feed, carriage return) is
1177 -- encountered and skipped, or some error situation, such as an
1178 -- illegal character, is encountered.
1180 <<Scan_Next_Character>>
1183 -- Skip past blanks, loop is opened up for speed
1185 while Source (Scan_Ptr) = ' ' loop
1186 if Source (Scan_Ptr + 1) /= ' ' then
1187 Scan_Ptr := Scan_Ptr + 1;
1191 if Source (Scan_Ptr + 2) /= ' ' then
1192 Scan_Ptr := Scan_Ptr + 2;
1196 if Source (Scan_Ptr + 3) /= ' ' then
1197 Scan_Ptr := Scan_Ptr + 3;
1201 if Source (Scan_Ptr + 4) /= ' ' then
1202 Scan_Ptr := Scan_Ptr + 4;
1206 if Source (Scan_Ptr + 5) /= ' ' then
1207 Scan_Ptr := Scan_Ptr + 5;
1211 if Source (Scan_Ptr + 6) /= ' ' then
1212 Scan_Ptr := Scan_Ptr + 6;
1216 if Source (Scan_Ptr + 7) /= ' ' then
1217 Scan_Ptr := Scan_Ptr + 7;
1221 Scan_Ptr := Scan_Ptr + 8;
1224 -- We are now at a non-blank character, which is the first character
1225 -- of the token we will scan, and hence the value of Token_Ptr.
1227 Token_Ptr := Scan_Ptr;
1229 -- Here begins the main case statement which transfers control on the
1230 -- basis of the non-blank character we have encountered.
1232 case Source (Scan_Ptr) is
1234 -- Line terminator characters
1236 when CR | LF | FF | VT =>
1237 goto Scan_Line_Terminator;
1239 -- Horizontal tab, just skip past it
1246 Scan_Ptr := Scan_Ptr + 1;
1248 -- End of file character, treated as an end of file only if it is
1249 -- the last character in the buffer, otherwise it is ignored.
1252 if Scan_Ptr = Source_Last (Current_Source_File) then
1262 Scan_Ptr := Scan_Ptr + 1;
1268 Accumulate_Checksum ('&');
1270 if Source (Scan_Ptr + 1) = '&' then
1271 Error_Msg_S -- CODEFIX
1272 ("'&'& should be `AND THEN`");
1273 Scan_Ptr := Scan_Ptr + 2;
1278 Scan_Ptr := Scan_Ptr + 1;
1279 Token := Tok_Ampersand;
1283 -- Asterisk (can be multiplication operator or double asterisk which
1284 -- is the exponentiation compound delimiter).
1287 Accumulate_Checksum ('*');
1289 if Source (Scan_Ptr + 1) = '*' then
1290 Accumulate_Checksum ('*');
1291 Scan_Ptr := Scan_Ptr + 2;
1292 Token := Tok_Double_Asterisk;
1296 Scan_Ptr := Scan_Ptr + 1;
1297 Token := Tok_Asterisk;
1301 -- Colon, which can either be an isolated colon, or part of an
1302 -- assignment compound delimiter.
1305 Accumulate_Checksum (':');
1307 if Double_Char_Token ('=') then
1308 Token := Tok_Colon_Equal;
1311 Style.Check_Colon_Equal;
1316 elsif Source (Scan_Ptr + 1) = '-'
1317 and then Source (Scan_Ptr + 2) /= '-'
1319 Token := Tok_Colon_Equal;
1320 Error_Msg -- CODEFIX
1321 (":- should be :=", Scan_Ptr);
1322 Scan_Ptr := Scan_Ptr + 2;
1326 Scan_Ptr := Scan_Ptr + 1;
1339 Accumulate_Checksum ('(');
1340 Scan_Ptr := Scan_Ptr + 1;
1341 Token := Tok_Left_Paren;
1344 Style.Check_Left_Paren;
1352 if Source (Scan_Ptr + 1) = '"' then
1353 goto Scan_Wide_Character;
1356 Error_Msg_S ("illegal character, replaced by ""(""");
1357 Scan_Ptr := Scan_Ptr + 1;
1358 Token := Tok_Left_Paren;
1365 Error_Msg_S ("illegal character, replaced by ""(""");
1366 Scan_Ptr := Scan_Ptr + 1;
1367 Token := Tok_Left_Paren;
1373 Accumulate_Checksum (',');
1374 Scan_Ptr := Scan_Ptr + 1;
1383 -- Dot, which is either an isolated period, or part of a double dot
1384 -- compound delimiter sequence. We also check for the case of a
1385 -- digit following the period, to give a better error message.
1388 Accumulate_Checksum ('.');
1390 if Double_Char_Token ('.') then
1391 Token := Tok_Dot_Dot;
1394 Style.Check_Dot_Dot;
1399 elsif Source (Scan_Ptr + 1) in '0' .. '9' then
1400 Error_Msg_S ("numeric literal cannot start with point");
1401 Scan_Ptr := Scan_Ptr + 1;
1404 Scan_Ptr := Scan_Ptr + 1;
1409 -- Equal, which can either be an equality operator, or part of the
1410 -- arrow (=>) compound delimiter.
1413 Accumulate_Checksum ('=');
1415 if Double_Char_Token ('>') then
1424 elsif Source (Scan_Ptr + 1) = '=' then
1425 Error_Msg_S -- CODEFIX
1427 Scan_Ptr := Scan_Ptr + 1;
1430 Scan_Ptr := Scan_Ptr + 1;
1434 -- Greater than, which can be a greater than operator, greater than
1435 -- or equal operator, or first character of a right label bracket.
1438 Accumulate_Checksum ('>');
1440 if Double_Char_Token ('=') then
1441 Token := Tok_Greater_Equal;
1444 elsif Double_Char_Token ('>') then
1445 Token := Tok_Greater_Greater;
1449 Scan_Ptr := Scan_Ptr + 1;
1450 Token := Tok_Greater;
1454 -- Less than, which can be a less than operator, less than or equal
1455 -- operator, or the first character of a left label bracket, or the
1456 -- first character of a box (<>) compound delimiter.
1459 Accumulate_Checksum ('<');
1461 if Double_Char_Token ('=') then
1462 Token := Tok_Less_Equal;
1465 elsif Double_Char_Token ('>') then
1474 elsif Double_Char_Token ('<') then
1475 Token := Tok_Less_Less;
1479 Scan_Ptr := Scan_Ptr + 1;
1484 -- Minus, which is either a subtraction operator, or the first
1485 -- character of double minus starting a comment
1487 when '-' => Minus_Case : begin
1488 if Source (Scan_Ptr + 1) = '>' then
1489 Error_Msg_S ("invalid token");
1490 Scan_Ptr := Scan_Ptr + 2;
1494 elsif Source (Scan_Ptr + 1) /= '-' then
1495 Accumulate_Checksum ('-');
1496 Scan_Ptr := Scan_Ptr + 1;
1502 else -- Source (Scan_Ptr + 1) = '-' then
1504 Style.Check_Comment;
1507 Scan_Ptr := Scan_Ptr + 2;
1509 -- If we are in preprocessor mode with Replace_In_Comments set,
1510 -- then we return the "--" as a token on its own.
1512 if Replace_In_Comments then
1513 Token := Tok_Comment;
1517 -- Otherwise scan out the comment
1519 Start_Of_Comment := Scan_Ptr;
1521 -- Loop to scan comment (this loop runs more than once only if
1522 -- a horizontal tab or other non-graphic character is scanned)
1525 -- Scan to non graphic character (opened up for speed)
1527 -- Note that we just eat left brackets, which means that
1528 -- bracket notation cannot be used for end of line
1529 -- characters in comments. This seems a reasonable choice,
1530 -- since no one would ever use brackets notation in a real
1531 -- program in this situation, and if we allow brackets
1532 -- notation, we forbid some valid comments which contain a
1533 -- brackets sequence that happens to match an end of line
1537 exit when Source (Scan_Ptr) not in Graphic_Character;
1538 Scan_Ptr := Scan_Ptr + 1;
1539 exit when Source (Scan_Ptr) not in Graphic_Character;
1540 Scan_Ptr := Scan_Ptr + 1;
1541 exit when Source (Scan_Ptr) not in Graphic_Character;
1542 Scan_Ptr := Scan_Ptr + 1;
1543 exit when Source (Scan_Ptr) not in Graphic_Character;
1544 Scan_Ptr := Scan_Ptr + 1;
1545 exit when Source (Scan_Ptr) not in Graphic_Character;
1546 Scan_Ptr := Scan_Ptr + 1;
1549 -- Keep going if horizontal tab
1551 if Source (Scan_Ptr) = HT then
1556 Scan_Ptr := Scan_Ptr + 1;
1558 -- Terminate scan of comment if line terminator
1560 elsif Source (Scan_Ptr) in Line_Terminator then
1563 -- Terminate scan of comment if end of file encountered
1564 -- (embedded EOF character or real last character in file)
1566 elsif Source (Scan_Ptr) = EOF then
1569 -- If we have a wide character, we have to scan it out,
1570 -- because it might be a legitimate line terminator
1572 elsif Start_Of_Wide_Character then
1574 Wptr : constant Source_Ptr := Scan_Ptr;
1579 Scan_Wide (Source, Scan_Ptr, Code, Err);
1581 -- If not well formed wide character, then just skip
1582 -- past it and ignore it.
1585 Scan_Ptr := Wptr + 1;
1587 -- If UTF_32 terminator, terminate comment scan
1589 elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
1595 -- Keep going if character in 80-FF range, or is ESC. These
1596 -- characters are allowed in comments by RM-2.1(1), 2.7(2).
1597 -- They are allowed even in Ada 83 mode according to the
1598 -- approved AI. ESC was added to the AI in June 93.
1600 elsif Source (Scan_Ptr) in Upper_Half_Character
1601 or else Source (Scan_Ptr) = ESC
1603 Scan_Ptr := Scan_Ptr + 1;
1605 -- Otherwise we have an illegal comment character
1608 Error_Illegal_Character;
1612 -- Note that, except when comments are tokens, we do NOT
1613 -- execute a return here, instead we fall through to reexecute
1614 -- the scan loop to look for a token.
1616 if Comment_Is_Token then
1617 Name_Len := Integer (Scan_Ptr - Start_Of_Comment);
1618 Name_Buffer (1 .. Name_Len) :=
1619 String (Source (Start_Of_Comment .. Scan_Ptr - 1));
1620 Comment_Id := Name_Find;
1621 Token := Tok_Comment;
1627 -- Double quote or percent starting a string literal
1634 -- Apostrophe. This can either be the start of a character literal,
1635 -- or an isolated apostrophe used in a qualified expression or an
1636 -- attribute. We treat it as a character literal if it does not
1637 -- follow a right parenthesis, identifier, the keyword ALL or
1638 -- a literal. This means that we correctly treat constructs like:
1640 -- A := CHARACTER'('A');
1642 -- Note that RM-2.2(7) does not require a separator between
1643 -- "CHARACTER" and "'" in the above.
1645 when ''' => Char_Literal_Case : declare
1650 Accumulate_Checksum (''');
1651 Scan_Ptr := Scan_Ptr + 1;
1653 -- Here is where we make the test to distinguish the cases. Treat
1654 -- as apostrophe if previous token is an identifier, right paren
1655 -- or the reserved word "all" (latter case as in A.all'Address)
1656 -- (or the reserved word "project" in project files). Also treat
1657 -- it as apostrophe after a literal (this catches some legitimate
1658 -- cases, like A."abs"'Address, and also gives better error
1659 -- behavior for impossible cases like 123'xxx).
1661 if Prev_Token = Tok_Identifier
1662 or else Prev_Token = Tok_Right_Paren
1663 or else Prev_Token = Tok_All
1664 or else Prev_Token = Tok_Project
1665 or else Prev_Token in Token_Class_Literal
1667 Token := Tok_Apostrophe;
1670 Style.Check_Apostrophe;
1675 -- Otherwise the apostrophe starts a character literal
1678 -- Case of wide character literal
1680 if Start_Of_Wide_Character then
1682 Scan_Wide (Source, Scan_Ptr, Code, Err);
1683 Accumulate_Checksum (Code);
1686 Error_Illegal_Wide_Character;
1687 Code := Character'Pos (' ');
1689 -- In Ada 95 mode we allow any wide character in a character
1690 -- literal, but in Ada 2005, the set of characters allowed
1691 -- is restricted to graphic characters.
1693 elsif Ada_Version >= Ada_2005
1694 and then Is_UTF_32_Non_Graphic (UTF_32 (Code))
1696 Error_Msg -- CODEFIX????
1697 ("(Ada 2005) non-graphic character not permitted " &
1698 "in character literal", Wptr);
1701 if Source (Scan_Ptr) /= ''' then
1702 Error_Msg_S ("missing apostrophe");
1704 Scan_Ptr := Scan_Ptr + 1;
1707 -- If we do not find a closing quote in the expected place then
1708 -- assume that we have a misguided attempt at a string literal.
1710 -- However, if previous token is RANGE, then we return an
1711 -- apostrophe instead since this gives better error recovery
1713 elsif Source (Scan_Ptr + 1) /= ''' then
1714 if Prev_Token = Tok_Range then
1715 Token := Tok_Apostrophe;
1719 Scan_Ptr := Scan_Ptr - 1;
1721 ("strings are delimited by double quote character");
1727 -- Otherwise we have a (non-wide) character literal
1730 Accumulate_Checksum (Source (Scan_Ptr));
1732 if Source (Scan_Ptr) not in Graphic_Character then
1733 if Source (Scan_Ptr) in Upper_Half_Character then
1734 if Ada_Version = Ada_83 then
1735 Error_Illegal_Character;
1739 Error_Illegal_Character;
1743 Code := Get_Char_Code (Source (Scan_Ptr));
1744 Scan_Ptr := Scan_Ptr + 2;
1747 -- Fall through here with Scan_Ptr updated past the closing
1748 -- quote, and Code set to the Char_Code value for the literal
1750 Accumulate_Checksum (''');
1751 Token := Tok_Char_Literal;
1752 Set_Character_Literal_Name (Code);
1753 Token_Name := Name_Find;
1754 Character_Code := Code;
1758 end Char_Literal_Case;
1760 -- Right parenthesis
1763 Accumulate_Checksum (')');
1764 Scan_Ptr := Scan_Ptr + 1;
1765 Token := Tok_Right_Paren;
1768 Style.Check_Right_Paren;
1773 -- Right bracket or right brace, treated as right paren
1776 Error_Msg_S ("illegal character, replaced by "")""");
1777 Scan_Ptr := Scan_Ptr + 1;
1778 Token := Tok_Right_Paren;
1781 -- Slash (can be division operator or first character of not equal)
1784 Accumulate_Checksum ('/');
1786 if Double_Char_Token ('=') then
1787 Token := Tok_Not_Equal;
1790 Scan_Ptr := Scan_Ptr + 1;
1798 Accumulate_Checksum (';');
1799 Scan_Ptr := Scan_Ptr + 1;
1800 Token := Tok_Semicolon;
1803 Style.Check_Semicolon;
1810 when '|' => Vertical_Bar_Case : begin
1811 Accumulate_Checksum ('|');
1813 -- Special check for || to give nice message
1815 if Source (Scan_Ptr + 1) = '|' then
1816 Error_Msg_S -- CODEFIX
1817 ("""'|'|"" should be `OR ELSE`");
1818 Scan_Ptr := Scan_Ptr + 2;
1823 Scan_Ptr := Scan_Ptr + 1;
1824 Token := Tok_Vertical_Bar;
1827 Style.Check_Vertical_Bar;
1833 end Vertical_Bar_Case;
1835 -- Exclamation, replacement character for vertical bar
1837 when '!' => Exclamation_Case : begin
1838 Accumulate_Checksum ('!');
1840 if Source (Scan_Ptr + 1) = '=' then
1841 Error_Msg_S -- CODEFIX
1842 ("'!= should be /=");
1843 Scan_Ptr := Scan_Ptr + 2;
1844 Token := Tok_Not_Equal;
1848 Scan_Ptr := Scan_Ptr + 1;
1849 Token := Tok_Vertical_Bar;
1853 end Exclamation_Case;
1857 when '+' => Plus_Case : begin
1858 Accumulate_Checksum ('+');
1859 Scan_Ptr := Scan_Ptr + 1;
1864 -- Digits starting a numeric literal
1868 -- First a bit of a scan ahead to see if we have a case of an
1869 -- identifier starting with a digit (remembering exponent case).
1872 C : constant Character := Source (Scan_Ptr + 1);
1875 -- OK literal if digit followed by digit or underscore
1877 if C in '0' .. '9' or else C = '_' then
1880 -- OK literal if digit not followed by identifier char
1882 elsif not Identifier_Char (C) then
1885 -- OK literal if digit followed by e/E followed by digit/sign.
1886 -- We also allow underscore after the E, which is an error, but
1887 -- better handled by Nlit than deciding this is an identifier.
1889 elsif (C = 'e' or else C = 'E')
1890 and then (Source (Scan_Ptr + 2) in '0' .. '9'
1891 or else Source (Scan_Ptr + 2) = '+'
1892 or else Source (Scan_Ptr + 2) = '-'
1893 or else Source (Scan_Ptr + 2) = '_')
1897 -- Here we have what really looks like an identifier that
1898 -- starts with a digit, so give error msg.
1901 Error_Msg_S ("identifier may not start with digit");
1903 Underline_Found := False;
1904 Name_Buffer (1) := Source (Scan_Ptr);
1905 Accumulate_Checksum (Name_Buffer (1));
1906 Scan_Ptr := Scan_Ptr + 1;
1907 goto Scan_Identifier;
1911 -- Here we have an OK integer literal
1915 -- Check for proper delimiter, ignoring other format characters
1917 Skip_Other_Format_Characters;
1919 if Identifier_Char (Source (Scan_Ptr)) then
1921 ("delimiter required between literal and identifier");
1927 -- Lower case letters
1931 Underline_Found := False;
1932 Name_Buffer (1) := Source (Scan_Ptr);
1933 Accumulate_Checksum (Name_Buffer (1));
1934 Scan_Ptr := Scan_Ptr + 1;
1935 goto Scan_Identifier;
1937 -- Upper case letters
1941 Underline_Found := False;
1943 Character'Val (Character'Pos (Source (Scan_Ptr)) + 32);
1944 Accumulate_Checksum (Name_Buffer (1));
1945 Scan_Ptr := Scan_Ptr + 1;
1946 goto Scan_Identifier;
1948 -- Underline character
1951 if Special_Characters ('_') then
1952 Token_Ptr := Scan_Ptr;
1953 Scan_Ptr := Scan_Ptr + 1;
1954 Token := Tok_Special;
1955 Special_Character := '_';
1959 Error_Msg_S ("identifier cannot start with underline");
1961 Name_Buffer (1) := '_';
1962 Scan_Ptr := Scan_Ptr + 1;
1963 Underline_Found := False;
1964 goto Scan_Identifier;
1966 -- Space (not possible, because we scanned past blanks)
1969 raise Program_Error;
1971 -- Characters in top half of ASCII 8-bit chart
1973 when Upper_Half_Character =>
1975 -- Wide character case
1977 if Upper_Half_Encoding then
1978 goto Scan_Wide_Character;
1980 -- Otherwise we have OK Latin-1 character
1983 -- Upper half characters may possibly be identifier letters
1984 -- but can never be digits, so Identifier_Char can be used to
1985 -- test for a valid start of identifier character.
1987 if Identifier_Char (Source (Scan_Ptr)) then
1989 Underline_Found := False;
1990 goto Scan_Identifier;
1992 Error_Illegal_Character;
1998 -- ESC character, possible start of identifier if wide characters
1999 -- using ESC encoding are allowed in identifiers, which we can
2000 -- tell by looking at the Identifier_Char flag for ESC, which is
2001 -- only true if these conditions are met. In Ada 2005 mode, may
2002 -- also be valid UTF_32 space or line terminator character.
2004 if Identifier_Char (ESC) then
2006 goto Scan_Wide_Character;
2008 Error_Illegal_Character;
2011 -- Invalid control characters
2013 when NUL | SOH | STX | ETX | EOT | ENQ | ACK | BEL | BS | ASCII.SO |
2014 SI | DLE | DC1 | DC2 | DC3 | DC4 | NAK | SYN | ETB | CAN |
2015 EM | FS | GS | RS | US | DEL
2017 Error_Illegal_Character;
2019 -- Invalid graphic characters
2021 when '#' | '$' | '?' | '@' | '`' | '\' | '^' | '~' =>
2023 -- If Set_Special_Character has been called for this character,
2024 -- set Scans.Special_Character and return a Special token.
2026 if Special_Characters (Source (Scan_Ptr)) then
2027 Token_Ptr := Scan_Ptr;
2028 Token := Tok_Special;
2029 Special_Character := Source (Scan_Ptr);
2030 Scan_Ptr := Scan_Ptr + 1;
2033 -- Otherwise, this is an illegal character
2036 Error_Illegal_Character;
2039 -- End switch on non-blank character
2043 -- End loop past format effectors. The exit from this loop is by
2044 -- executing a return statement following completion of token scan
2045 -- (control never falls out of this loop to the code which follows)
2049 -- Wide_Character scanning routine. On entry we have encountered the
2050 -- initial character of a wide character sequence.
2052 <<Scan_Wide_Character>>
2061 Scan_Wide (Source, Scan_Ptr, Code, Err);
2063 -- If bad wide character, signal error and continue scan
2066 Error_Illegal_Wide_Character;
2067 goto Scan_Next_Character;
2070 Cat := Get_Category (UTF_32 (Code));
2072 -- If OK letter, reset scan ptr and go scan identifier
2074 if Is_UTF_32_Letter (Cat) then
2077 Underline_Found := False;
2078 goto Scan_Identifier;
2080 -- If OK wide space, ignore and keep scanning (we do not include
2081 -- any ignored spaces in checksum)
2083 elsif Is_UTF_32_Space (Cat) then
2084 goto Scan_Next_Character;
2086 -- If other format character, ignore and keep scanning (again we
2087 -- do not include in the checksum) (this is for AI-0079).
2089 elsif Is_UTF_32_Other (Cat) then
2090 goto Scan_Next_Character;
2092 -- If OK wide line terminator, terminate current line
2094 elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
2096 goto Scan_Line_Terminator;
2098 -- Punctuation is an error (at start of identifier)
2100 elsif Is_UTF_32_Punctuation (Cat) then
2101 Error_Msg ("identifier cannot start with punctuation", Wptr);
2104 Underline_Found := False;
2105 goto Scan_Identifier;
2107 -- Mark character is an error (at start of identifier)
2109 elsif Is_UTF_32_Mark (Cat) then
2110 Error_Msg ("identifier cannot start with mark character", Wptr);
2113 Underline_Found := False;
2114 goto Scan_Identifier;
2116 -- Extended digit character is an error. Could be bad start of
2117 -- identifier or bad literal. Not worth doing too much to try to
2118 -- distinguish these cases, but we will do a little bit.
2120 elsif Is_UTF_32_Digit (Cat) then
2122 ("identifier cannot start with digit character", Wptr);
2125 Underline_Found := False;
2126 goto Scan_Identifier;
2128 -- All other wide characters are illegal here
2131 Error_Illegal_Wide_Character;
2132 goto Scan_Next_Character;
2136 -- Routine to scan line terminator. On entry Scan_Ptr points to a
2137 -- character which is one of FF,LR,CR,VT, or one of the wide characters
2138 -- that is treated as a line terminator.
2140 <<Scan_Line_Terminator>>
2142 -- Check line too long
2146 -- Set Token_Ptr, if End_Of_Line is a token, for the case when it is
2149 if End_Of_Line_Is_Token then
2150 Token_Ptr := Scan_Ptr;
2157 Skip_Line_Terminators (Scan_Ptr, Physical);
2159 -- If we are at start of physical line, update scan pointers to
2160 -- reflect the start of the new line.
2163 Current_Line_Start := Scan_Ptr;
2164 Start_Column := Set_Start_Column;
2165 First_Non_Blank_Location := Scan_Ptr;
2167 -- If End_Of_Line is a token, we return it as it is a
2170 if End_Of_Line_Is_Token then
2171 Token := Tok_End_Of_Line;
2177 goto Scan_Next_Character;
2179 -- Identifier scanning routine. On entry, some initial characters of
2180 -- the identifier may have already been stored in Name_Buffer. If so,
2181 -- Name_Len has the number of characters stored, otherwise Name_Len is
2182 -- set to zero on entry. Underline_Found is also set False on entry.
2186 -- This loop scans as fast as possible past lower half letters and
2187 -- digits, which we expect to be the most common characters.
2190 if Source (Scan_Ptr) in 'a' .. 'z'
2191 or else Source (Scan_Ptr) in '0' .. '9'
2193 Name_Buffer (Name_Len + 1) := Source (Scan_Ptr);
2194 Accumulate_Checksum (Source (Scan_Ptr));
2196 elsif Source (Scan_Ptr) in 'A' .. 'Z' then
2197 Name_Buffer (Name_Len + 1) :=
2198 Character'Val (Character'Pos (Source (Scan_Ptr)) + 32);
2199 Accumulate_Checksum (Name_Buffer (Name_Len + 1));
2205 Underline_Found := False;
2206 Scan_Ptr := Scan_Ptr + 1;
2207 Name_Len := Name_Len + 1;
2210 -- If we fall through, then we have encountered either an underline
2211 -- character, or an extended identifier character (i.e. one from the
2212 -- upper half), or a wide character, or an identifier terminator. The
2213 -- initial test speeds us up in the most common case where we have
2214 -- an identifier terminator. Note that ESC is an identifier character
2215 -- only if a wide character encoding method that uses ESC encoding
2216 -- is active, so if we find an ESC character we know that we have a
2219 if Identifier_Char (Source (Scan_Ptr))
2220 or else (Source (Scan_Ptr) in Upper_Half_Character
2221 and then Upper_Half_Encoding)
2223 -- Case of underline
2225 if Source (Scan_Ptr) = '_' then
2226 Accumulate_Checksum ('_');
2228 if Underline_Found then
2229 Error_No_Double_Underline;
2231 Underline_Found := True;
2232 Name_Len := Name_Len + 1;
2233 Name_Buffer (Name_Len) := '_';
2236 Scan_Ptr := Scan_Ptr + 1;
2237 goto Scan_Identifier;
2239 -- Upper half character
2241 elsif Source (Scan_Ptr) in Upper_Half_Character
2242 and then not Upper_Half_Encoding
2244 Accumulate_Checksum (Source (Scan_Ptr));
2245 Store_Encoded_Character
2246 (Get_Char_Code (Fold_Lower (Source (Scan_Ptr))));
2247 Scan_Ptr := Scan_Ptr + 1;
2248 Underline_Found := False;
2249 goto Scan_Identifier;
2251 -- Left bracket not followed by a quote terminates an identifier.
2252 -- This is an error, but we don't want to give a junk error msg
2253 -- about wide characters in this case!
2255 elsif Source (Scan_Ptr) = '['
2256 and then Source (Scan_Ptr + 1) /= '"'
2260 -- We know we have a wide character encoding here (the current
2261 -- character is either ESC, left bracket, or an upper half
2262 -- character depending on the encoding method).
2265 -- Scan out the wide character and insert the appropriate
2266 -- encoding into the name table entry for the identifier.
2276 Scan_Wide (Source, Scan_Ptr, Code, Err);
2278 -- If error, signal error
2281 Error_Illegal_Wide_Character;
2283 -- If the character scanned is a normal identifier
2284 -- character, then we treat it that way.
2286 elsif In_Character_Range (Code)
2287 and then Identifier_Char (Get_Character (Code))
2289 Chr := Get_Character (Code);
2290 Accumulate_Checksum (Chr);
2291 Store_Encoded_Character
2292 (Get_Char_Code (Fold_Lower (Chr)));
2293 Underline_Found := False;
2295 -- Here if not a normal identifier character
2298 Cat := Get_Category (UTF_32 (Code));
2300 -- Wide character in Unicode category "Other, Format"
2301 -- is not accepted in an identifier. This is because it
2302 -- it is considered a security risk (AI-0091).
2304 -- However, it is OK for such a character to appear at
2305 -- the end of an identifier.
2307 if Is_UTF_32_Other (Cat) then
2308 if not Identifier_Char (Source (Scan_Ptr)) then
2309 goto Scan_Identifier_Complete;
2312 ("identifier cannot contain other_format "
2313 & "character", Wptr);
2314 goto Scan_Identifier;
2317 -- Wide character in category Separator,Space terminates
2319 elsif Is_UTF_32_Space (Cat) then
2320 goto Scan_Identifier_Complete;
2323 -- Here if wide character is part of the identifier
2325 -- Make sure we are allowing wide characters in
2326 -- identifiers. Note that we allow wide character
2327 -- notation for an OK identifier character. This in
2328 -- particular allows bracket or other notation to be
2329 -- used for upper half letters.
2331 -- Wide characters are always allowed in Ada 2005
2333 if Identifier_Character_Set /= 'w'
2334 and then Ada_Version < Ada_2005
2337 ("wide character not allowed in identifier", Wptr);
2340 -- If OK letter, store it folding to upper case. Note
2341 -- that we include the folded letter in the checksum.
2343 if Is_UTF_32_Letter (Cat) then
2345 Char_Code (UTF_32_To_Upper_Case (UTF_32 (Code)));
2346 Accumulate_Checksum (Code);
2347 Store_Encoded_Character (Code);
2348 Underline_Found := False;
2350 -- If OK extended digit or mark, then store it
2352 elsif Is_UTF_32_Digit (Cat)
2353 or else Is_UTF_32_Mark (Cat)
2355 Accumulate_Checksum (Code);
2356 Store_Encoded_Character (Code);
2357 Underline_Found := False;
2359 -- Wide punctuation is also stored, but counts as an
2360 -- underline character for error checking purposes.
2362 elsif Is_UTF_32_Punctuation (Cat) then
2363 Accumulate_Checksum (Code);
2365 if Underline_Found then
2367 Cend : constant Source_Ptr := Scan_Ptr;
2370 Error_No_Double_Underline;
2375 Store_Encoded_Character (Code);
2376 Underline_Found := True;
2379 -- Any other wide character is not acceptable
2383 ("invalid wide character in identifier", Wptr);
2387 goto Scan_Identifier;
2392 -- Scan of identifier is complete. The identifier is stored in
2393 -- Name_Buffer, and Scan_Ptr points past the last character.
2395 <<Scan_Identifier_Complete>>
2396 Token_Name := Name_Find;
2398 -- Check for identifier ending with underline or punctuation char
2400 if Underline_Found then
2401 Underline_Found := False;
2403 if Source (Scan_Ptr - 1) = '_' then
2405 ("identifier cannot end with underline", Scan_Ptr - 1);
2408 ("identifier cannot end with punctuation character", Wptr);
2412 -- We will assume it is an identifier, not a keyword, so that the
2413 -- checksum is independent of the Ada version.
2415 Token := Tok_Identifier;
2416 Accumulate_Token_Checksum;
2418 -- Here is where we check if it was a keyword
2420 if Is_Keyword_Name (Token_Name) then
2421 Token := Token_Type'Val (Get_Name_Table_Byte (Token_Name));
2423 -- Keyword style checks
2427 -- Deal with possible style check for non-lower case keyword,
2428 -- but we don't treat ACCESS, DELTA, DIGITS, RANGE as keywords
2429 -- for this purpose if they appear as attribute designators.
2430 -- Actually we only check the first character for speed.
2432 -- Ada 2005 (AI-284): Do not apply the style check in case of
2433 -- "pragma Interface"
2435 -- Ada 2005 (AI-340): Do not apply the style check in case of
2438 if Source (Token_Ptr) <= 'Z'
2439 and then (Prev_Token /= Tok_Apostrophe
2441 (Token /= Tok_Access and then
2442 Token /= Tok_Delta and then
2443 Token /= Tok_Digits and then
2444 Token /= Tok_Mod and then
2445 Token /= Tok_Range))
2446 and then (Token /= Tok_Interface
2448 (Token = Tok_Interface
2449 and then Prev_Token /= Tok_Pragma))
2451 Style.Non_Lower_Case_Keyword;
2454 -- Check THEN/ELSE style rules. These do not apply to AND THEN
2455 -- or OR ELSE, and do not apply in conditional expressions.
2457 if (Token = Tok_Then and then Prev_Token /= Tok_And)
2459 (Token = Tok_Else and then Prev_Token /= Tok_Or)
2461 if Inside_Conditional_Expression = 0 then
2462 Style.Check_Separate_Stmt_Lines;
2467 -- We must reset Token_Name since this is not an identifier and
2468 -- if we leave Token_Name set, the parser gets confused because
2469 -- it thinks it is dealing with an identifier instead of the
2470 -- corresponding keyword.
2472 Token_Name := No_Name;
2475 -- It is an identifier after all
2483 --------------------------
2484 -- Set_Comment_As_Token --
2485 --------------------------
2487 procedure Set_Comment_As_Token (Value : Boolean) is
2489 Comment_Is_Token := Value;
2490 end Set_Comment_As_Token;
2492 ------------------------------
2493 -- Set_End_Of_Line_As_Token --
2494 ------------------------------
2496 procedure Set_End_Of_Line_As_Token (Value : Boolean) is
2498 End_Of_Line_Is_Token := Value;
2499 end Set_End_Of_Line_As_Token;
2501 ---------------------------
2502 -- Set_Special_Character --
2503 ---------------------------
2505 procedure Set_Special_Character (C : Character) is
2508 when '#' | '$' | '_' | '?' | '@' | '`' | '\' | '^' | '~' =>
2509 Special_Characters (C) := True;
2514 end Set_Special_Character;
2516 ----------------------
2517 -- Set_Start_Column --
2518 ----------------------
2520 -- Note: it seems at first glance a little expensive to compute this value
2521 -- for every source line (since it is certainly not used for all source
2522 -- lines). On the other hand, it doesn't take much more work to skip past
2523 -- the initial white space on the line counting the columns than it would
2524 -- to scan past the white space using the standard scanning circuits.
2526 function Set_Start_Column return Column_Number is
2527 Start_Column : Column_Number := 0;
2530 -- Outer loop scans past horizontal tab characters
2534 -- Inner loop scans past blanks as fast as possible, bumping Scan_Ptr
2535 -- past the blanks and adjusting Start_Column to account for them.
2538 if Source (Scan_Ptr) = ' ' then
2539 if Source (Scan_Ptr + 1) = ' ' then
2540 if Source (Scan_Ptr + 2) = ' ' then
2541 if Source (Scan_Ptr + 3) = ' ' then
2542 if Source (Scan_Ptr + 4) = ' ' then
2543 if Source (Scan_Ptr + 5) = ' ' then
2544 if Source (Scan_Ptr + 6) = ' ' then
2545 Scan_Ptr := Scan_Ptr + 7;
2546 Start_Column := Start_Column + 7;
2548 Scan_Ptr := Scan_Ptr + 6;
2549 Start_Column := Start_Column + 6;
2553 Scan_Ptr := Scan_Ptr + 5;
2554 Start_Column := Start_Column + 5;
2558 Scan_Ptr := Scan_Ptr + 4;
2559 Start_Column := Start_Column + 4;
2563 Scan_Ptr := Scan_Ptr + 3;
2564 Start_Column := Start_Column + 3;
2568 Scan_Ptr := Scan_Ptr + 2;
2569 Start_Column := Start_Column + 2;
2573 Scan_Ptr := Scan_Ptr + 1;
2574 Start_Column := Start_Column + 1;
2580 end loop Blanks_Loop;
2582 -- Outer loop keeps going only if a horizontal tab follows
2584 if Source (Scan_Ptr) = HT then
2589 Scan_Ptr := Scan_Ptr + 1;
2590 Start_Column := (Start_Column / 8) * 8 + 8;
2596 return Start_Column;
2598 -- A constraint error can happen only if we have a compiler with checks on
2599 -- and a line with a ludicrous number of tabs or spaces at the start. In
2600 -- such a case, we really don't care if Start_Column is right or not.
2603 when Constraint_Error =>
2604 return Start_Column;
2605 end Set_Start_Column;