1------------------------------------------------------------------------------
2--                                                                          --
3--                         GNAT COMPILER COMPONENTS                         --
4--                                                                          --
5--                                 S C N G                                  --
6--                                                                          --
7--                                 B o d y                                  --
8--                                                                          --
9--          Copyright (C) 1992-2012, Free Software Foundation, Inc.         --
10--                                                                          --
11-- GNAT is free software;  you can  redistribute it  and/or modify it under --
12-- terms of the  GNU General Public License as published  by the Free Soft- --
13-- ware  Foundation;  either version 3,  or (at your option) any later ver- --
14-- sion.  GNAT is distributed in the hope that it will be useful, but WITH- --
15-- OUT ANY WARRANTY;  without even the  implied warranty of MERCHANTABILITY --
16-- or FITNESS FOR A PARTICULAR PURPOSE.  See the GNU General Public License --
17-- for  more details.  You should have  received  a copy of the GNU General --
18-- Public License  distributed with GNAT; see file COPYING3.  If not, go to --
19-- http://www.gnu.org/licenses for a complete copy of the license.          --
20--                                                                          --
21-- GNAT was originally developed  by the GNAT team at  New York University. --
22-- Extensive contributions were provided by Ada Core Technologies Inc.      --
23--                                                                          --
24------------------------------------------------------------------------------
25
26with Atree;    use Atree;
27with Csets;    use Csets;
28with Hostparm; use Hostparm;
29with Namet;    use Namet;
30with Opt;      use Opt;
31with Restrict; use Restrict;
32with Rident;   use Rident;
33with Scans;    use Scans;
34with Sinput;   use Sinput;
35with Snames;   use Snames;
36with Stringt;  use Stringt;
37with Stylesw;  use Stylesw;
38with Uintp;    use Uintp;
39with Urealp;   use Urealp;
40with Widechar; use Widechar;
41
42pragma Warnings (Off);
43--  This package is used also by gnatcoll
44with System.CRC32;
45with System.UTF_32;  use System.UTF_32;
46with System.WCh_Con; use System.WCh_Con;
47pragma Warnings (On);
48
49package body Scng is
50
51   use ASCII;
52   --  Make control characters visible
53
54   Special_Characters : array (Character) of Boolean := (others => False);
55   --  For characters that are Special token, the value is True
56
57   Comment_Is_Token : Boolean := False;
58   --  True if comments are tokens
59
60   End_Of_Line_Is_Token : Boolean := False;
61   --  True if End_Of_Line is a token
62
63   -----------------------
64   -- Local Subprograms --
65   -----------------------
66
67   procedure Accumulate_Token_Checksum;
68   pragma Inline (Accumulate_Token_Checksum);
69   --  Called after each numeric literal and identifier/keyword. For keywords,
70   --  the token used is Tok_Identifier. This allows to detect additional
71   --  spaces added in sources when using the builder switch -m.
72
73   procedure Accumulate_Token_Checksum_GNAT_6_3;
74   --  Used in place of Accumulate_Token_Checksum for GNAT versions 5.04 to
75   --  6.3, when Tok_Some was not included in Token_Type and the actual
76   --  Token_Type was used for keywords. This procedure is never used in the
77   --  compiler or gnatmake, only in gprbuild.
78
79   procedure Accumulate_Token_Checksum_GNAT_5_03;
80   --  Used in place of Accumulate_Token_Checksum for GNAT version 5.03, when
81   --  Tok_Interface, Tok_Some, Tok_Synchronized and Tok_Overriding were not
82   --  included in Token_Type and the actual Token_Type was used for keywords.
83   --  This procedure is never used in the compiler or gnatmake, only in
84   --  gprbuild.
85
86   procedure Accumulate_Checksum (C : Character);
87   pragma Inline (Accumulate_Checksum);
88   --  This routine accumulates the checksum given character C. During the
89   --  scanning of a source file, this routine is called with every character
90   --  in the source, excluding blanks, and all control characters (except
91   --  that ESC is included in the checksum). Upper case letters not in string
92   --  literals are folded by the caller. See Sinput spec for the documentation
93   --  of the checksum algorithm. Note: checksum values are only used if we
94   --  generate code, so it is not necessary to worry about making the right
95   --  sequence of calls in any error situation.
96
97   procedure Accumulate_Checksum (C : Char_Code);
98   pragma Inline (Accumulate_Checksum);
99   --  This version is identical, except that the argument, C, is a character
100   --  code value instead of a character. This is used when wide characters
101   --  are scanned. We use the character code rather than the ASCII characters
102   --  so that the checksum is independent of wide character encoding method.
103
104   procedure Initialize_Checksum;
105   pragma Inline (Initialize_Checksum);
106   --  Initialize checksum value
107
108   -------------------------
109   -- Accumulate_Checksum --
110   -------------------------
111
112   procedure Accumulate_Checksum (C : Character) is
113   begin
114      System.CRC32.Update (System.CRC32.CRC32 (Checksum), C);
115   end Accumulate_Checksum;
116
117   procedure Accumulate_Checksum (C : Char_Code) is
118   begin
119      if C > 16#FFFF# then
120         Accumulate_Checksum (Character'Val (C / 2 ** 24));
121         Accumulate_Checksum (Character'Val ((C / 2 ** 16) mod 256));
122         Accumulate_Checksum (Character'Val ((C / 256) mod 256));
123      else
124         Accumulate_Checksum (Character'Val (C / 256));
125      end if;
126
127      Accumulate_Checksum (Character'Val (C mod 256));
128   end Accumulate_Checksum;
129
130   -------------------------------
131   -- Accumulate_Token_Checksum --
132   -------------------------------
133
134   procedure Accumulate_Token_Checksum is
135   begin
136      System.CRC32.Update
137        (System.CRC32.CRC32 (Checksum),
138         Character'Val (Token_Type'Pos (Token)));
139   end Accumulate_Token_Checksum;
140
141   ----------------------------------------
142   -- Accumulate_Token_Checksum_GNAT_6_3 --
143   ----------------------------------------
144
145   procedure Accumulate_Token_Checksum_GNAT_6_3 is
146   begin
147      --  Individual values of Token_Type are used, instead of subranges, so
148      --  that additions or suppressions of enumerated values in type
149      --  Token_Type are detected by the compiler.
150
151      case Token is
152         when Tok_Integer_Literal | Tok_Real_Literal | Tok_String_Literal |
153              Tok_Char_Literal | Tok_Operator_Symbol | Tok_Identifier |
154              Tok_Double_Asterisk | Tok_Ampersand | Tok_Minus | Tok_Plus |
155              Tok_Asterisk | Tok_Mod | Tok_Rem | Tok_Slash | Tok_New |
156              Tok_Abs | Tok_Others | Tok_Null | Tok_Dot | Tok_Apostrophe |
157              Tok_Left_Paren | Tok_Delta | Tok_Digits | Tok_Range |
158              Tok_Right_Paren | Tok_Comma | Tok_And | Tok_Or | Tok_Xor |
159              Tok_Less | Tok_Equal | Tok_Greater | Tok_Not_Equal |
160              Tok_Greater_Equal | Tok_Less_Equal | Tok_In | Tok_Not |
161              Tok_Box | Tok_Colon_Equal | Tok_Colon | Tok_Greater_Greater |
162              Tok_Abstract | Tok_Access | Tok_Aliased | Tok_All | Tok_Array |
163              Tok_At | Tok_Body | Tok_Constant | Tok_Do | Tok_Is |
164              Tok_Interface | Tok_Limited | Tok_Of | Tok_Out | Tok_Record |
165              Tok_Renames | Tok_Reverse =>
166
167            System.CRC32.Update
168              (System.CRC32.CRC32 (Checksum),
169               Character'Val (Token_Type'Pos (Token)));
170
171         when Tok_Some =>
172
173            System.CRC32.Update
174              (System.CRC32.CRC32 (Checksum),
175               Character'Val (Token_Type'Pos (Tok_Identifier)));
176
177         when Tok_Tagged | Tok_Then | Tok_Less_Less | Tok_Abort | Tok_Accept |
178              Tok_Case | Tok_Delay | Tok_Else | Tok_Elsif | Tok_End |
179              Tok_Exception | Tok_Exit | Tok_Goto | Tok_If | Tok_Pragma |
180              Tok_Raise | Tok_Requeue | Tok_Return | Tok_Select |
181              Tok_Terminate | Tok_Until | Tok_When | Tok_Begin | Tok_Declare |
182              Tok_For | Tok_Loop | Tok_While | Tok_Entry | Tok_Protected |
183              Tok_Task | Tok_Type | Tok_Subtype | Tok_Overriding |
184              Tok_Synchronized | Tok_Use | Tok_Function | Tok_Generic |
185              Tok_Package | Tok_Procedure | Tok_Private | Tok_With |
186              Tok_Separate | Tok_EOF | Tok_Semicolon | Tok_Arrow |
187              Tok_Vertical_Bar | Tok_Dot_Dot | Tok_Project | Tok_Extends |
188              Tok_External | Tok_External_As_List | Tok_Comment |
189              Tok_End_Of_Line | Tok_Special | Tok_SPARK_Hide | No_Token =>
190
191            System.CRC32.Update
192              (System.CRC32.CRC32 (Checksum),
193               Character'Val (Token_Type'Pos (Token_Type'Pred (Token))));
194      end case;
195   end Accumulate_Token_Checksum_GNAT_6_3;
196
197   -----------------------------------------
198   -- Accumulate_Token_Checksum_GNAT_5_03 --
199   -----------------------------------------
200
201   procedure Accumulate_Token_Checksum_GNAT_5_03 is
202   begin
203      --  Individual values of Token_Type are used, instead of subranges, so
204      --  that additions or suppressions of enumerated values in type
205      --  Token_Type are detected by the compiler.
206
207      case Token is
208         when Tok_Integer_Literal | Tok_Real_Literal | Tok_String_Literal |
209              Tok_Char_Literal | Tok_Operator_Symbol | Tok_Identifier |
210              Tok_Double_Asterisk | Tok_Ampersand | Tok_Minus | Tok_Plus |
211              Tok_Asterisk | Tok_Mod | Tok_Rem | Tok_Slash | Tok_New |
212              Tok_Abs | Tok_Others | Tok_Null | Tok_Dot | Tok_Apostrophe |
213              Tok_Left_Paren | Tok_Delta | Tok_Digits | Tok_Range |
214              Tok_Right_Paren | Tok_Comma | Tok_And | Tok_Or | Tok_Xor |
215              Tok_Less | Tok_Equal | Tok_Greater | Tok_Not_Equal |
216              Tok_Greater_Equal | Tok_Less_Equal | Tok_In | Tok_Not |
217              Tok_Box | Tok_Colon_Equal | Tok_Colon | Tok_Greater_Greater |
218              Tok_Abstract | Tok_Access | Tok_Aliased | Tok_All | Tok_Array |
219              Tok_At | Tok_Body | Tok_Constant | Tok_Do | Tok_Is =>
220
221            System.CRC32.Update
222              (System.CRC32.CRC32 (Checksum),
223               Character'Val (Token_Type'Pos (Token)));
224
225         when Tok_Interface | Tok_Some | Tok_Overriding | Tok_Synchronized =>
226            System.CRC32.Update
227              (System.CRC32.CRC32 (Checksum),
228               Character'Val (Token_Type'Pos (Tok_Identifier)));
229
230         when Tok_Limited | Tok_Of | Tok_Out | Tok_Record |
231              Tok_Renames | Tok_Reverse =>
232
233            System.CRC32.Update
234              (System.CRC32.CRC32 (Checksum),
235               Character'Val (Token_Type'Pos (Token) - 1));
236
237         when Tok_Tagged | Tok_Then | Tok_Less_Less | Tok_Abort | Tok_Accept |
238              Tok_Case | Tok_Delay | Tok_Else | Tok_Elsif | Tok_End |
239              Tok_Exception | Tok_Exit | Tok_Goto | Tok_If | Tok_Pragma |
240              Tok_Raise | Tok_Requeue | Tok_Return | Tok_Select |
241              Tok_Terminate | Tok_Until | Tok_When | Tok_Begin | Tok_Declare |
242              Tok_For | Tok_Loop | Tok_While | Tok_Entry | Tok_Protected |
243              Tok_Task | Tok_Type | Tok_Subtype =>
244
245            System.CRC32.Update
246              (System.CRC32.CRC32 (Checksum),
247               Character'Val (Token_Type'Pos (Token) - 2));
248
249         when Tok_Use | Tok_Function | Tok_Generic |
250              Tok_Package | Tok_Procedure | Tok_Private | Tok_With |
251              Tok_Separate | Tok_EOF | Tok_Semicolon | Tok_Arrow |
252              Tok_Vertical_Bar | Tok_Dot_Dot | Tok_Project | Tok_Extends |
253              Tok_External | Tok_External_As_List | Tok_Comment |
254              Tok_End_Of_Line | Tok_Special | Tok_SPARK_Hide | No_Token =>
255
256            System.CRC32.Update
257              (System.CRC32.CRC32 (Checksum),
258               Character'Val (Token_Type'Pos (Token) - 4));
259      end case;
260   end Accumulate_Token_Checksum_GNAT_5_03;
261
262   ----------------------------
263   -- Determine_Token_Casing --
264   ----------------------------
265
266   function Determine_Token_Casing return Casing_Type is
267   begin
268      return Determine_Casing (Source (Token_Ptr .. Scan_Ptr - 1));
269   end Determine_Token_Casing;
270
271   -------------------------
272   -- Initialize_Checksum --
273   -------------------------
274
275   procedure Initialize_Checksum is
276   begin
277      System.CRC32.Initialize (System.CRC32.CRC32 (Checksum));
278   end Initialize_Checksum;
279
280   ------------------------
281   -- Initialize_Scanner --
282   ------------------------
283
284   procedure Initialize_Scanner (Index : Source_File_Index) is
285   begin
286      --  Establish reserved words
287
288      Scans.Initialize_Ada_Keywords;
289
290      --  Initialize scan control variables
291
292      Current_Source_File       := Index;
293      Source                    := Source_Text (Current_Source_File);
294      Scan_Ptr                  := Source_First (Current_Source_File);
295      Token                     := No_Token;
296      Token_Ptr                 := Scan_Ptr;
297      Current_Line_Start        := Scan_Ptr;
298      Token_Node                := Empty;
299      Token_Name                := No_Name;
300      Start_Column              := Set_Start_Column;
301      First_Non_Blank_Location  := Scan_Ptr;
302
303      Initialize_Checksum;
304      Wide_Char_Byte_Count := 0;
305
306      --  Do not call Scan, otherwise the License stuff does not work in Scn
307
308   end Initialize_Scanner;
309
310   ------------------------------
311   -- Reset_Special_Characters --
312   ------------------------------
313
314   procedure Reset_Special_Characters is
315   begin
316      Special_Characters := (others => False);
317   end Reset_Special_Characters;
318
319   ----------
320   -- Scan --
321   ----------
322
323   procedure Scan is
324
325      Start_Of_Comment : Source_Ptr;
326      --  Record start of comment position
327
328      Underline_Found : Boolean;
329      --  During scanning of an identifier, set to True if last character
330      --  scanned was an underline or other punctuation character. This
331      --  is used to flag the error of two underlines/punctuations in a
332      --  row or ending an identifier with a underline/punctuation. Here
333      --  punctuation means any UTF_32 character in the Unicode category
334      --  Punctuation,Connector.
335
336      Wptr : Source_Ptr;
337      --  Used to remember start of last wide character scanned
338
339      procedure Check_End_Of_Line;
340      --  Called when end of line encountered. Checks that line is not too
341      --  long, and that other style checks for the end of line are met.
342
343      function Double_Char_Token (C : Character) return Boolean;
344      --  This function is used for double character tokens like := or <>. It
345      --  checks if the character following Source (Scan_Ptr) is C, and if so
346      --  bumps Scan_Ptr past the pair of characters and returns True. A space
347      --  between the two characters is also recognized with an appropriate
348      --  error message being issued. If C is not present, False is returned.
349      --  Note that Double_Char_Token can only be used for tokens defined in
350      --  the Ada syntax (it's use for error cases like && is not appropriate
351      --  since we do not want a junk message for a case like &-space-&).
352
353      procedure Error_Illegal_Character;
354      --  Give illegal character error, Scan_Ptr points to character. On
355      --  return, Scan_Ptr is bumped past the illegal character.
356
357      procedure Error_Illegal_Wide_Character;
358      --  Give illegal wide character message. On return, Scan_Ptr is bumped
359      --  past the illegal character, which may still leave us pointing to
360      --  junk, not much we can do if the escape sequence is messed up!
361
362      procedure Error_Long_Line;
363      --  Signal error of excessively long line
364
365      procedure Error_No_Double_Underline;
366      --  Signal error of two underline or punctuation characters in a row.
367      --  Called with Scan_Ptr pointing to second underline/punctuation char.
368
369      procedure Nlit;
370      --  This is the procedure for scanning out numeric literals. On entry,
371      --  Scan_Ptr points to the digit that starts the numeric literal (the
372      --  checksum for this character has not been accumulated yet). On return
373      --  Scan_Ptr points past the last character of the numeric literal, Token
374      --  and Token_Node are set appropriately, and the checksum is updated.
375
376      procedure Slit;
377      --  This is the procedure for scanning out string literals. On entry,
378      --  Scan_Ptr points to the opening string quote (the checksum for this
379      --  character has not been accumulated yet). On return Scan_Ptr points
380      --  past the closing quote of the string literal, Token and Token_Node
381      --  are set appropriately, and the checksum is updated.
382
383      procedure Skip_Other_Format_Characters;
384      --  Skips past any "other format" category characters at the current
385      --  cursor location (does not skip past spaces or any other characters).
386
387      function Start_Of_Wide_Character return Boolean;
388      --  Returns True if the scan pointer is pointing to the start of a wide
389      --  character sequence, does not modify the scan pointer in any case.
390
391      -----------------------
392      -- Check_End_Of_Line --
393      -----------------------
394
395      procedure Check_End_Of_Line is
396         Len : constant Int :=
397                 Int (Scan_Ptr) -
398                 Int (Current_Line_Start) -
399                 Wide_Char_Byte_Count;
400
401      begin
402         if Style_Check then
403            Style.Check_Line_Terminator (Len);
404         end if;
405
406         --  Deal with checking maximum line length
407
408         if Style_Check and Style_Check_Max_Line_Length then
409            Style.Check_Line_Max_Length (Len);
410
411         --  If style checking is inactive, check maximum line length against
412         --  standard value.
413
414         elsif Len > Max_Line_Length then
415            Error_Long_Line;
416         end if;
417
418         --  Now one more checking circuit. Normally we are only enforcing a
419         --  limit of physical characters, with tabs counting as one character.
420         --  But if after tab expansion we would have a total line length that
421         --  exceeded 32766, that would really cause trouble, because column
422         --  positions would exceed the maximum we allow for a column count.
423         --  Note: the limit is 32766 rather than 32767, since we use a value
424         --  of 32767 for special purposes (see Sinput). Now we really do not
425         --  want to go messing with tabs in the normal case, so what we do is
426         --  to check for a line that has more than 4096 physical characters.
427         --  Any shorter line could not be a problem, even if it was all tabs.
428
429         if Len >= 4096 then
430            declare
431               Col : Natural;
432               Ptr : Source_Ptr;
433
434            begin
435               Col := 1;
436               Ptr := Current_Line_Start;
437               loop
438                  exit when Ptr = Scan_Ptr;
439
440                  if Source (Ptr) = ASCII.HT then
441                     Col := (Col - 1 + 8) / 8 * 8 + 1;
442                  else
443                     Col := Col + 1;
444                  end if;
445
446                  if Col > 32766 then
447                     Error_Msg
448                       ("this line is longer than 32766 characters",
449                        Current_Line_Start);
450                     raise Unrecoverable_Error;
451                  end if;
452
453                  Ptr := Ptr + 1;
454               end loop;
455            end;
456         end if;
457
458         --  Reset wide character byte count for next line
459
460         Wide_Char_Byte_Count := 0;
461      end Check_End_Of_Line;
462
463      -----------------------
464      -- Double_Char_Token --
465      -----------------------
466
467      function Double_Char_Token (C : Character) return Boolean is
468      begin
469         if Source (Scan_Ptr + 1) = C then
470            Accumulate_Checksum (C);
471            Scan_Ptr := Scan_Ptr + 2;
472            return True;
473
474         elsif Source (Scan_Ptr + 1) = ' '
475           and then Source (Scan_Ptr + 2) = C
476         then
477            Scan_Ptr := Scan_Ptr + 1;
478            Error_Msg_S -- CODEFIX
479              ("no space allowed here");
480            Scan_Ptr := Scan_Ptr + 2;
481            return True;
482
483         else
484            return False;
485         end if;
486      end Double_Char_Token;
487
488      -----------------------------
489      -- Error_Illegal_Character --
490      -----------------------------
491
492      procedure Error_Illegal_Character is
493      begin
494         Error_Msg_S ("illegal character");
495         Scan_Ptr := Scan_Ptr + 1;
496      end Error_Illegal_Character;
497
498      ----------------------------------
499      -- Error_Illegal_Wide_Character --
500      ----------------------------------
501
502      procedure Error_Illegal_Wide_Character is
503      begin
504         Scan_Ptr := Scan_Ptr + 1;
505         Error_Msg ("illegal wide character", Wptr);
506      end Error_Illegal_Wide_Character;
507
508      ---------------------
509      -- Error_Long_Line --
510      ---------------------
511
512      procedure Error_Long_Line is
513      begin
514         Error_Msg
515           ("this line is too long",
516            Current_Line_Start + Source_Ptr (Max_Line_Length));
517      end Error_Long_Line;
518
519      -------------------------------
520      -- Error_No_Double_Underline --
521      -------------------------------
522
523      procedure Error_No_Double_Underline is
524      begin
525         Underline_Found := False;
526
527         --  There are four cases, and we special case the messages
528
529         if Source (Scan_Ptr) = '_' then
530            if Source (Scan_Ptr - 1) = '_' then
531               Error_Msg_S -- CODEFIX
532                 ("two consecutive underlines not permitted");
533            else
534               Error_Msg_S ("underline cannot follow punctuation character");
535            end if;
536
537         else
538            if Source (Scan_Ptr - 1) = '_' then
539               Error_Msg_S ("punctuation character cannot follow underline");
540            else
541               Error_Msg_S
542                 ("two consecutive punctuation characters not permitted");
543            end if;
544         end if;
545      end Error_No_Double_Underline;
546
547      ----------
548      -- Nlit --
549      ----------
550
551      procedure Nlit is
552
553         C : Character;
554         --  Current source program character
555
556         Base_Char : Character;
557         --  Either # or : (character at start of based number)
558
559         Base : Int;
560         --  Value of base
561
562         UI_Base : Uint;
563         --  Value of base in Uint format
564
565         UI_Int_Value : Uint;
566         --  Value of integer scanned by Scan_Integer in Uint format
567
568         UI_Num_Value : Uint;
569         --  Value of integer in numeric value being scanned
570
571         Scale : Int;
572         --  Scale value for real literal
573
574         UI_Scale : Uint;
575         --  Scale in Uint format
576
577         Exponent_Is_Negative : Boolean;
578         --  Set true for negative exponent
579
580         Extended_Digit_Value : Int;
581         --  Extended digit value
582
583         Point_Scanned : Boolean;
584         --  Flag for decimal point scanned in numeric literal
585
586         -----------------------
587         -- Local Subprograms --
588         -----------------------
589
590         procedure Error_Digit_Expected;
591         --  Signal error of bad digit, Scan_Ptr points to the location at
592         --  which the digit was expected on input, and is unchanged on return.
593
594         procedure Scan_Integer;
595         --  Procedure to scan integer literal. On entry, Scan_Ptr points to a
596         --  digit, on exit Scan_Ptr points past the last character of the
597         --  integer.
598         --
599         --  For each digit encountered, UI_Int_Value is multiplied by 10, and
600         --  the value of the digit added to the result. In addition, the
601         --  value in Scale is decremented by one for each actual digit
602         --  scanned.
603
604         --------------------------
605         -- Error_Digit_Expected --
606         --------------------------
607
608         procedure Error_Digit_Expected is
609         begin
610            Error_Msg_S ("digit expected");
611         end Error_Digit_Expected;
612
613         ------------------
614         -- Scan_Integer --
615         ------------------
616
617         procedure Scan_Integer is
618            C : Character;
619            --  Next character scanned
620
621         begin
622            C := Source (Scan_Ptr);
623
624            --  Loop through digits (allowing underlines)
625
626            loop
627               Accumulate_Checksum (C);
628               UI_Int_Value :=
629                 UI_Int_Value * 10 + (Character'Pos (C) - Character'Pos ('0'));
630               Scan_Ptr := Scan_Ptr + 1;
631               Scale := Scale - 1;
632               C := Source (Scan_Ptr);
633
634               --  Case of underline encountered
635
636               if C = '_' then
637
638                  --  We do not accumulate the '_' in the checksum, so that
639                  --  1_234 is equivalent to 1234, and does not trigger
640                  --  compilation for "minimal recompilation" (gnatmake -m).
641
642                  loop
643                     Scan_Ptr := Scan_Ptr + 1;
644                     C := Source (Scan_Ptr);
645                     exit when C /= '_';
646                     Error_No_Double_Underline;
647                  end loop;
648
649                  if C not in '0' .. '9' then
650                     Error_Digit_Expected;
651                     exit;
652                  end if;
653
654               else
655                  exit when C not in '0' .. '9';
656               end if;
657            end loop;
658         end Scan_Integer;
659
660      --  Start of Processing for Nlit
661
662      begin
663         Base := 10;
664         UI_Base := Uint_10;
665         UI_Int_Value := Uint_0;
666         Based_Literal_Uses_Colon := False;
667         Scale := 0;
668         Scan_Integer;
669         Point_Scanned := False;
670         UI_Num_Value := UI_Int_Value;
671
672         --  Various possibilities now for continuing the literal are period,
673         --  E/e (for exponent), or :/# (for based literal).
674
675         Scale := 0;
676         C := Source (Scan_Ptr);
677
678         if C = '.' then
679
680            --  Scan out point, but do not scan past .. which is a range
681            --  sequence, and must not be eaten up scanning a numeric literal.
682
683            while C = '.' and then Source (Scan_Ptr + 1) /= '.' loop
684               Accumulate_Checksum ('.');
685
686               if Point_Scanned then
687                  Error_Msg_S ("duplicate point ignored");
688               end if;
689
690               Point_Scanned := True;
691               Scan_Ptr := Scan_Ptr + 1;
692               C := Source (Scan_Ptr);
693
694               if C not in '0' .. '9' then
695                  Error_Msg
696                    ("real literal cannot end with point", Scan_Ptr - 1);
697               else
698                  Scan_Integer;
699                  UI_Num_Value := UI_Int_Value;
700               end if;
701            end loop;
702
703         --  Based literal case. The base is the value we already scanned.
704         --  In the case of colon, we insist that the following character
705         --  is indeed an extended digit or a period. This catches a number
706         --  of common errors, as well as catching the well known tricky
707         --  bug otherwise arising from "x : integer range 1 .. 10:= 6;"
708
709         elsif C = '#'
710           or else (C = ':' and then
711                      (Source (Scan_Ptr + 1) = '.'
712                         or else
713                       Source (Scan_Ptr + 1) in '0' .. '9'
714                         or else
715                       Source (Scan_Ptr + 1) in 'A' .. 'Z'
716                         or else
717                       Source (Scan_Ptr + 1) in 'a' .. 'z'))
718         then
719            Accumulate_Checksum (C);
720            Base_Char := C;
721            UI_Base := UI_Int_Value;
722
723            if Base_Char = ':' then
724               Based_Literal_Uses_Colon := True;
725            end if;
726
727            if UI_Base < 2 or else UI_Base > 16 then
728               Error_Msg_SC ("base not 2-16");
729               UI_Base := Uint_16;
730            end if;
731
732            Base := UI_To_Int (UI_Base);
733            Scan_Ptr := Scan_Ptr + 1;
734
735            --  Scan out extended integer [. integer]
736
737            C := Source (Scan_Ptr);
738            UI_Int_Value := Uint_0;
739            Scale := 0;
740
741            loop
742               if C in '0' .. '9' then
743                  Accumulate_Checksum (C);
744                  Extended_Digit_Value :=
745                    Int'(Character'Pos (C)) - Int'(Character'Pos ('0'));
746
747               elsif C in 'A' .. 'F' then
748                  Accumulate_Checksum (Character'Val (Character'Pos (C) + 32));
749                  Extended_Digit_Value :=
750                    Int'(Character'Pos (C)) - Int'(Character'Pos ('A')) + 10;
751
752               elsif C in 'a' .. 'f' then
753                  Accumulate_Checksum (C);
754                  Extended_Digit_Value :=
755                    Int'(Character'Pos (C)) - Int'(Character'Pos ('a')) + 10;
756
757               else
758                  Error_Msg_S ("extended digit expected");
759                  exit;
760               end if;
761
762               if Extended_Digit_Value >= Base then
763                  Error_Msg_S ("digit '>= base");
764               end if;
765
766               UI_Int_Value := UI_Int_Value * UI_Base + Extended_Digit_Value;
767               Scale := Scale - 1;
768               Scan_Ptr := Scan_Ptr + 1;
769               C := Source (Scan_Ptr);
770
771               if C = '_' then
772                  loop
773                     Accumulate_Checksum ('_');
774                     Scan_Ptr := Scan_Ptr + 1;
775                     C := Source (Scan_Ptr);
776                     exit when C /= '_';
777                     Error_No_Double_Underline;
778                  end loop;
779
780               elsif C = '.' then
781                  Accumulate_Checksum ('.');
782
783                  if Point_Scanned then
784                     Error_Msg_S ("duplicate point ignored");
785                  end if;
786
787                  Scan_Ptr := Scan_Ptr + 1;
788                  C := Source (Scan_Ptr);
789                  Point_Scanned := True;
790                  Scale := 0;
791
792               elsif C = Base_Char then
793                  Accumulate_Checksum (C);
794                  Scan_Ptr := Scan_Ptr + 1;
795                  exit;
796
797               elsif C = '#' or else C = ':' then
798                  Error_Msg_S ("based number delimiters must match");
799                  Scan_Ptr := Scan_Ptr + 1;
800                  exit;
801
802               elsif not Identifier_Char (C) then
803                  if Base_Char = '#' then
804                     Error_Msg_S -- CODEFIX
805                       ("missing '#");
806                  else
807                     Error_Msg_S -- CODEFIX
808                       ("missing ':");
809                  end if;
810
811                  exit;
812               end if;
813
814            end loop;
815
816            UI_Num_Value := UI_Int_Value;
817         end if;
818
819         --  Scan out exponent
820
821         if not Point_Scanned then
822            Scale := 0;
823            UI_Scale := Uint_0;
824         else
825            UI_Scale := UI_From_Int (Scale);
826         end if;
827
828         if Source (Scan_Ptr) = 'e' or else Source (Scan_Ptr) = 'E' then
829            Accumulate_Checksum ('e');
830            Scan_Ptr := Scan_Ptr + 1;
831            Exponent_Is_Negative := False;
832
833            if Source (Scan_Ptr) = '+' then
834               Accumulate_Checksum ('+');
835               Scan_Ptr := Scan_Ptr + 1;
836
837            elsif Source (Scan_Ptr) = '-' then
838               Accumulate_Checksum ('-');
839
840               if not Point_Scanned then
841                  Error_Msg_S
842                    ("negative exponent not allowed for integer literal");
843               else
844                  Exponent_Is_Negative := True;
845               end if;
846
847               Scan_Ptr := Scan_Ptr + 1;
848            end if;
849
850            UI_Int_Value := Uint_0;
851
852            if Source (Scan_Ptr) in '0' .. '9' then
853               Scan_Integer;
854            else
855               Error_Digit_Expected;
856            end if;
857
858            if Exponent_Is_Negative then
859               UI_Scale := UI_Scale - UI_Int_Value;
860            else
861               UI_Scale := UI_Scale + UI_Int_Value;
862            end if;
863         end if;
864
865         --  Case of real literal to be returned
866
867         if Point_Scanned then
868            Token := Tok_Real_Literal;
869            Real_Literal_Value :=
870              UR_From_Components (
871                                  Num   => UI_Num_Value,
872                                  Den   => -UI_Scale,
873                                  Rbase => Base);
874
875         --  Case of integer literal to be returned
876
877         else
878            Token := Tok_Integer_Literal;
879
880            if UI_Scale = 0 then
881               Int_Literal_Value := UI_Num_Value;
882
883            --  Avoid doing possibly expensive calculations in cases like
884            --  parsing 163E800_000# when semantics will not be done anyway.
885            --  This is especially useful when parsing garbled input.
886
887            elsif Operating_Mode /= Check_Syntax
888              and then (Serious_Errors_Detected = 0 or else Try_Semantics)
889            then
890               Int_Literal_Value := UI_Num_Value * UI_Base ** UI_Scale;
891
892            else
893               Int_Literal_Value := No_Uint;
894            end if;
895         end if;
896
897         if Checksum_Accumulate_Token_Checksum then
898            Accumulate_Token_Checksum;
899         end if;
900
901         return;
902      end Nlit;
903
904      ----------
905      -- Slit --
906      ----------
907
908      procedure Slit is
909
910         Delimiter : Character;
911         --  Delimiter (first character of string)
912
913         C : Character;
914         --  Current source program character
915
916         Code : Char_Code;
917         --  Current character code value
918
919         Err : Boolean;
920         --  Error flag for Scan_Wide call
921
922         String_Start : Source_Ptr;
923         --  Point to first character of string
924
925         procedure Error_Bad_String_Char;
926         --  Signal bad character in string/character literal. On entry
927         --  Scan_Ptr points to the improper character encountered during the
928         --  scan. Scan_Ptr is not modified, so it still points to the bad
929         --  character on return.
930
931         procedure Error_Unterminated_String;
932         --  Procedure called if a line terminator character is encountered
933         --  during scanning a string, meaning that the string is not properly
934         --  terminated.
935
936         procedure Set_String;
937         --  Procedure used to distinguish between string and operator symbol.
938         --  On entry the string has been scanned out, and its characters start
939         --  at Token_Ptr and end one character before Scan_Ptr. On exit Token
940         --  is set to Tok_String_Literal/Tok_Operator_Symbol as appropriate,
941         --  and Token_Node is appropriately initialized. In addition, in the
942         --  operator symbol case, Token_Name is appropriately set, and the
943         --  flags [Wide_]Wide_Character_Found are set appropriately.
944
945         ---------------------------
946         -- Error_Bad_String_Char --
947         ---------------------------
948
949         procedure Error_Bad_String_Char is
950            C : constant Character := Source (Scan_Ptr);
951
952         begin
953            if C = HT then
954               Error_Msg_S ("horizontal tab not allowed in string");
955
956            elsif C = VT or else C = FF then
957               Error_Msg_S ("format effector not allowed in string");
958
959            elsif C in Upper_Half_Character then
960               Error_Msg_S ("(Ada 83) upper half character not allowed");
961
962            else
963               Error_Msg_S ("control character not allowed in string");
964            end if;
965         end Error_Bad_String_Char;
966
967         -------------------------------
968         -- Error_Unterminated_String --
969         -------------------------------
970
971         procedure Error_Unterminated_String is
972            S : Source_Ptr;
973
974         begin
975            --  An interesting little refinement. Consider the following
976            --  examples:
977
978            --     A := "this is an unterminated string;
979            --     A := "this is an unterminated string &
980            --     P(A, "this is a parameter that didn't get terminated);
981            --     P("this is a parameter that didn't get terminated, A);
982
983            --  We fiddle a little to do slightly better placement in these
984            --  cases also if there is white space at the end of the line we
985            --  place the flag at the start of this white space, not at the
986            --  end. Note that we only have to test for blanks, since tabs
987            --  aren't allowed in strings in the first place and would have
988            --  caused an error message.
989
990            --  Two more cases that we treat specially are:
991
992            --     A := "this string uses the wrong terminator'
993            --     A := "this string uses the wrong terminator' &
994
995            --  In these cases we give a different error message as well
996
997            --  We actually reposition the scan pointer to the point where we
998            --  place the flag in these cases, since it seems a better bet on
999            --  the original intention.
1000
1001            while Source (Scan_Ptr - 1) = ' '
1002              or else Source (Scan_Ptr - 1) = '&'
1003            loop
1004               Scan_Ptr := Scan_Ptr - 1;
1005               Unstore_String_Char;
1006            end loop;
1007
1008            --  Check for case of incorrect string terminator, but single quote
1009            --  is not considered incorrect if the opening terminator misused
1010            --  a single quote (error message already given).
1011
1012            if Delimiter /= '''
1013              and then Source (Scan_Ptr - 1) = '''
1014            then
1015               Unstore_String_Char;
1016               Error_Msg
1017                 ("incorrect string terminator character", Scan_Ptr - 1);
1018               return;
1019            end if;
1020
1021            --  Backup over semicolon or right-paren/semicolon sequence
1022
1023            if Source (Scan_Ptr - 1) = ';' then
1024               Scan_Ptr := Scan_Ptr - 1;
1025               Unstore_String_Char;
1026
1027               if Source (Scan_Ptr - 1) = ')' then
1028                  Scan_Ptr := Scan_Ptr - 1;
1029                  Unstore_String_Char;
1030               end if;
1031            end if;
1032
1033            --  See if there is a comma in the string, if so, guess that
1034            --  the first comma terminates the string.
1035
1036            S := String_Start;
1037            while S < Scan_Ptr loop
1038               if Source (S) = ',' then
1039                  while Scan_Ptr > S loop
1040                     Scan_Ptr := Scan_Ptr - 1;
1041                     Unstore_String_Char;
1042                  end loop;
1043
1044                  exit;
1045               end if;
1046
1047               S := S + 1;
1048            end loop;
1049
1050            --  Now we have adjusted the scan pointer, give message
1051
1052            Error_Msg_S -- CODEFIX
1053              ("missing string quote");
1054         end Error_Unterminated_String;
1055
1056         ----------------
1057         -- Set_String --
1058         ----------------
1059
1060         procedure Set_String is
1061            Slen : constant Int := Int (Scan_Ptr - Token_Ptr - 2);
1062            C1   : Character;
1063            C2   : Character;
1064            C3   : Character;
1065
1066         begin
1067            --  Token_Name is currently set to Error_Name. The following
1068            --  section of code resets Token_Name to the proper Name_Op_xx
1069            --  value if the string is a valid operator symbol, otherwise it is
1070            --  left set to Error_Name.
1071
1072            if Slen = 1 then
1073               C1 := Source (Token_Ptr + 1);
1074
1075               case C1 is
1076                  when '=' =>
1077                     Token_Name := Name_Op_Eq;
1078
1079                  when '>' =>
1080                     Token_Name := Name_Op_Gt;
1081
1082                  when '<' =>
1083                     Token_Name := Name_Op_Lt;
1084
1085                  when '+' =>
1086                     Token_Name := Name_Op_Add;
1087
1088                  when '-' =>
1089                     Token_Name := Name_Op_Subtract;
1090
1091                  when '&' =>
1092                     Token_Name := Name_Op_Concat;
1093
1094                  when '*' =>
1095                     Token_Name := Name_Op_Multiply;
1096
1097                  when '/' =>
1098                     Token_Name := Name_Op_Divide;
1099
1100                  when others =>
1101                     null;
1102               end case;
1103
1104            elsif Slen = 2 then
1105               C1 := Source (Token_Ptr + 1);
1106               C2 := Source (Token_Ptr + 2);
1107
1108               if C1 = '*' and then C2 = '*' then
1109                  Token_Name := Name_Op_Expon;
1110
1111               elsif C2 = '=' then
1112
1113                  if C1 = '/' then
1114                     Token_Name := Name_Op_Ne;
1115                  elsif C1 = '<' then
1116                     Token_Name := Name_Op_Le;
1117                  elsif C1 = '>' then
1118                     Token_Name := Name_Op_Ge;
1119                  end if;
1120
1121               elsif (C1 = 'O' or else C1 = 'o') and then    -- OR
1122                 (C2 = 'R' or else C2 = 'r')
1123               then
1124                  Token_Name := Name_Op_Or;
1125               end if;
1126
1127            elsif Slen = 3 then
1128               C1 := Source (Token_Ptr + 1);
1129               C2 := Source (Token_Ptr + 2);
1130               C3 := Source (Token_Ptr + 3);
1131
1132               if (C1 = 'A' or else C1 = 'a') and then       -- AND
1133                 (C2 = 'N' or else C2 = 'n') and then
1134                 (C3 = 'D' or else C3 = 'd')
1135               then
1136                  Token_Name := Name_Op_And;
1137
1138               elsif (C1 = 'A' or else C1 = 'a') and then    -- ABS
1139                 (C2 = 'B' or else C2 = 'b') and then
1140                 (C3 = 'S' or else C3 = 's')
1141               then
1142                  Token_Name := Name_Op_Abs;
1143
1144               elsif (C1 = 'M' or else C1 = 'm') and then    -- MOD
1145                 (C2 = 'O' or else C2 = 'o') and then
1146                 (C3 = 'D' or else C3 = 'd')
1147               then
1148                  Token_Name := Name_Op_Mod;
1149
1150               elsif (C1 = 'N' or else C1 = 'n') and then    -- NOT
1151                 (C2 = 'O' or else C2 = 'o') and then
1152                 (C3 = 'T' or else C3 = 't')
1153               then
1154                  Token_Name := Name_Op_Not;
1155
1156               elsif (C1 = 'R' or else C1 = 'r') and then    -- REM
1157                 (C2 = 'E' or else C2 = 'e') and then
1158                 (C3 = 'M' or else C3 = 'm')
1159               then
1160                  Token_Name := Name_Op_Rem;
1161
1162               elsif (C1 = 'X' or else C1 = 'x') and then    -- XOR
1163                 (C2 = 'O' or else C2 = 'o') and then
1164                 (C3 = 'R' or else C3 = 'r')
1165               then
1166                  Token_Name := Name_Op_Xor;
1167               end if;
1168
1169            end if;
1170
1171            --  If it is an operator symbol, then Token_Name is set. If it is
1172            --  some other string value, then Token_Name still contains
1173            --  Error_Name.
1174
1175            if Token_Name = Error_Name then
1176               Token := Tok_String_Literal;
1177
1178            else
1179               Token := Tok_Operator_Symbol;
1180            end if;
1181         end Set_String;
1182
1183      --  Start of processing for Slit
1184
1185      begin
1186         --  On entry, Scan_Ptr points to the opening character of the string
1187         --  which is either a percent, double quote, or apostrophe (single
1188         --  quote). The latter case is an error detected by the character
1189         --  literal circuit.
1190
1191         String_Start := Scan_Ptr;
1192
1193         Delimiter := Source (Scan_Ptr);
1194         Accumulate_Checksum (Delimiter);
1195
1196         Start_String;
1197         Wide_Character_Found      := False;
1198         Wide_Wide_Character_Found := False;
1199         Scan_Ptr := Scan_Ptr + 1;
1200
1201         --  Loop to scan out characters of string literal
1202
1203         loop
1204            C := Source (Scan_Ptr);
1205
1206            if C = Delimiter then
1207               Accumulate_Checksum (C);
1208               Scan_Ptr := Scan_Ptr + 1;
1209               exit when Source (Scan_Ptr) /= Delimiter;
1210               Code := Get_Char_Code (C);
1211               Accumulate_Checksum (C);
1212               Scan_Ptr := Scan_Ptr + 1;
1213
1214            else
1215               if C = '"' and then Delimiter = '%' then
1216                  Error_Msg_S
1217                    ("quote not allowed in percent delimited string");
1218                  Code := Get_Char_Code (C);
1219                  Scan_Ptr := Scan_Ptr + 1;
1220
1221               elsif Start_Of_Wide_Character then
1222                  Wptr := Scan_Ptr;
1223                  Scan_Wide (Source, Scan_Ptr, Code, Err);
1224
1225                  if Err then
1226                     Error_Illegal_Wide_Character;
1227                     Code := Get_Char_Code (' ');
1228                  end if;
1229
1230                  Accumulate_Checksum (Code);
1231
1232                  --  In Ada 95 mode we allow any wide characters in a string
1233                  --  but in Ada 2005, the set of characters allowed has been
1234                  --  restricted to graphic characters.
1235
1236                  if Ada_Version >= Ada_2005
1237                    and then Is_UTF_32_Non_Graphic (UTF_32 (Code))
1238                  then
1239                     Error_Msg
1240                       ("(Ada 2005) non-graphic character not permitted " &
1241                        "in string literal", Wptr);
1242                  end if;
1243
1244               else
1245                  Accumulate_Checksum (C);
1246
1247                  if C not in Graphic_Character then
1248                     if C in Line_Terminator then
1249                        Error_Unterminated_String;
1250                        exit;
1251
1252                     elsif C in Upper_Half_Character then
1253                        if Ada_Version = Ada_83 then
1254                           Error_Bad_String_Char;
1255                        end if;
1256
1257                     else
1258                        Error_Bad_String_Char;
1259                     end if;
1260                  end if;
1261
1262                  Code := Get_Char_Code (C);
1263                  Scan_Ptr := Scan_Ptr + 1;
1264               end if;
1265            end if;
1266
1267            Store_String_Char (Code);
1268
1269            if not In_Character_Range (Code) then
1270               if In_Wide_Character_Range (Code) then
1271                  Wide_Character_Found := True;
1272               else
1273                  Wide_Wide_Character_Found := True;
1274               end if;
1275            end if;
1276         end loop;
1277
1278         String_Literal_Id := End_String;
1279         Set_String;
1280         return;
1281      end Slit;
1282
1283      ----------------------------------
1284      -- Skip_Other_Format_Characters --
1285      ----------------------------------
1286
1287      procedure Skip_Other_Format_Characters is
1288         P    : Source_Ptr;
1289         Code : Char_Code;
1290         Err  : Boolean;
1291
1292      begin
1293         while Start_Of_Wide_Character loop
1294            P := Scan_Ptr;
1295            Scan_Wide (Source, Scan_Ptr, Code, Err);
1296
1297            if not Is_UTF_32_Other (UTF_32 (Code)) then
1298               Scan_Ptr := P;
1299               return;
1300            end if;
1301         end loop;
1302      end Skip_Other_Format_Characters;
1303
1304      -----------------------------
1305      -- Start_Of_Wide_Character --
1306      -----------------------------
1307
1308      function Start_Of_Wide_Character return Boolean is
1309         C : constant Character := Source (Scan_Ptr);
1310
1311      begin
1312         --  ESC encoding method with ESC present
1313
1314         if C = ESC
1315           and then Wide_Character_Encoding_Method in WC_ESC_Encoding_Method
1316         then
1317            return True;
1318
1319         --  Upper half character with upper half encoding
1320
1321         elsif C in Upper_Half_Character and then Upper_Half_Encoding then
1322            return True;
1323
1324         --  Brackets encoding
1325
1326         elsif C = '['
1327           and then Source (Scan_Ptr + 1) = '"'
1328           and then Identifier_Char (Source (Scan_Ptr + 2))
1329         then
1330            return True;
1331
1332         --  Not the start of a wide character
1333
1334         else
1335            return False;
1336         end if;
1337      end Start_Of_Wide_Character;
1338
1339   --  Start of processing for Scan
1340
1341   begin
1342      Prev_Token := Token;
1343      Prev_Token_Ptr := Token_Ptr;
1344      Token_Name := Error_Name;
1345
1346      --  The following loop runs more than once only if a format effector
1347      --  (tab, vertical tab, form  feed, line feed, carriage return) is
1348      --  encountered and skipped, or some error situation, such as an
1349      --  illegal character, is encountered.
1350
1351      <<Scan_Next_Character>>
1352
1353      loop
1354         --  Skip past blanks, loop is opened up for speed
1355
1356         while Source (Scan_Ptr) = ' ' loop
1357            if Source (Scan_Ptr + 1) /= ' ' then
1358               Scan_Ptr := Scan_Ptr + 1;
1359               exit;
1360            end if;
1361
1362            if Source (Scan_Ptr + 2) /= ' ' then
1363               Scan_Ptr := Scan_Ptr + 2;
1364               exit;
1365            end if;
1366
1367            if Source (Scan_Ptr + 3) /= ' ' then
1368               Scan_Ptr := Scan_Ptr + 3;
1369               exit;
1370            end if;
1371
1372            if Source (Scan_Ptr + 4) /= ' ' then
1373               Scan_Ptr := Scan_Ptr + 4;
1374               exit;
1375            end if;
1376
1377            if Source (Scan_Ptr + 5) /= ' ' then
1378               Scan_Ptr := Scan_Ptr + 5;
1379               exit;
1380            end if;
1381
1382            if Source (Scan_Ptr + 6) /= ' ' then
1383               Scan_Ptr := Scan_Ptr + 6;
1384               exit;
1385            end if;
1386
1387            if Source (Scan_Ptr + 7) /= ' ' then
1388               Scan_Ptr := Scan_Ptr + 7;
1389               exit;
1390            end if;
1391
1392            Scan_Ptr := Scan_Ptr + 8;
1393         end loop;
1394
1395         --  We are now at a non-blank character, which is the first character
1396         --  of the token we will scan, and hence the value of Token_Ptr.
1397
1398         Token_Ptr := Scan_Ptr;
1399
1400         --  Here begins the main case statement which transfers control on the
1401         --  basis of the non-blank character we have encountered.
1402
1403         case Source (Scan_Ptr) is
1404
1405         --  Line terminator characters
1406
1407         when CR | LF | FF | VT =>
1408            goto Scan_Line_Terminator;
1409
1410         --  Horizontal tab, just skip past it
1411
1412         when HT =>
1413            if Style_Check then
1414               Style.Check_HT;
1415            end if;
1416
1417            Scan_Ptr := Scan_Ptr + 1;
1418
1419         --  End of file character, treated as an end of file only if it is
1420         --  the last character in the buffer, otherwise it is ignored.
1421
1422         when EOF =>
1423            if Scan_Ptr = Source_Last (Current_Source_File) then
1424               Check_End_Of_Line;
1425
1426               if Style_Check then
1427                  Style.Check_EOF;
1428               end if;
1429
1430               Token := Tok_EOF;
1431               return;
1432            else
1433               Scan_Ptr := Scan_Ptr + 1;
1434            end if;
1435
1436         --  Ampersand
1437
1438         when '&' =>
1439            Accumulate_Checksum ('&');
1440
1441            if Source (Scan_Ptr + 1) = '&' then
1442               Error_Msg_S -- CODEFIX
1443                 ("'&'& should be `AND THEN`");
1444               Scan_Ptr := Scan_Ptr + 2;
1445               Token := Tok_And;
1446               return;
1447
1448            else
1449               Scan_Ptr := Scan_Ptr + 1;
1450               Token := Tok_Ampersand;
1451               return;
1452            end if;
1453
1454         --  Asterisk (can be multiplication operator or double asterisk which
1455         --  is the exponentiation compound delimiter).
1456
1457         when '*' =>
1458            Accumulate_Checksum ('*');
1459
1460            if Source (Scan_Ptr + 1) = '*' then
1461               Accumulate_Checksum ('*');
1462               Scan_Ptr := Scan_Ptr + 2;
1463               Token := Tok_Double_Asterisk;
1464               return;
1465
1466            else
1467               Scan_Ptr := Scan_Ptr + 1;
1468               Token := Tok_Asterisk;
1469               return;
1470            end if;
1471
1472         --  Colon, which can either be an isolated colon, or part of an
1473         --  assignment compound delimiter.
1474
1475         when ':' =>
1476            Accumulate_Checksum (':');
1477
1478            if Double_Char_Token ('=') then
1479               Token := Tok_Colon_Equal;
1480
1481               if Style_Check then
1482                  Style.Check_Colon_Equal;
1483               end if;
1484
1485               return;
1486
1487            elsif Source (Scan_Ptr + 1) = '-'
1488              and then Source (Scan_Ptr + 2) /= '-'
1489            then
1490               Token := Tok_Colon_Equal;
1491               Error_Msg -- CODEFIX
1492                 (":- should be :=", Scan_Ptr);
1493               Scan_Ptr := Scan_Ptr + 2;
1494               return;
1495
1496            else
1497               Scan_Ptr := Scan_Ptr + 1;
1498               Token := Tok_Colon;
1499
1500               if Style_Check then
1501                  Style.Check_Colon;
1502               end if;
1503
1504               return;
1505            end if;
1506
1507         --  Left parenthesis
1508
1509         when '(' =>
1510            Accumulate_Checksum ('(');
1511            Scan_Ptr := Scan_Ptr + 1;
1512            Token := Tok_Left_Paren;
1513
1514            if Style_Check then
1515               Style.Check_Left_Paren;
1516            end if;
1517
1518            return;
1519
1520         --  Left bracket
1521
1522         when '[' =>
1523            if Source (Scan_Ptr + 1) = '"' then
1524               goto Scan_Wide_Character;
1525
1526            else
1527               Error_Msg_S ("illegal character, replaced by ""(""");
1528               Scan_Ptr := Scan_Ptr + 1;
1529               Token := Tok_Left_Paren;
1530               return;
1531            end if;
1532
1533         --  Left brace
1534
1535         when '{' =>
1536            Error_Msg_S ("illegal character, replaced by ""(""");
1537            Scan_Ptr := Scan_Ptr + 1;
1538            Token := Tok_Left_Paren;
1539            return;
1540
1541         --  Comma
1542
1543         when ',' =>
1544            Accumulate_Checksum (',');
1545            Scan_Ptr := Scan_Ptr + 1;
1546            Token := Tok_Comma;
1547
1548            if Style_Check then
1549               Style.Check_Comma;
1550            end if;
1551
1552            return;
1553
1554         --  Dot, which is either an isolated period, or part of a double dot
1555         --  compound delimiter sequence. We also check for the case of a
1556         --  digit following the period, to give a better error message.
1557
1558         when '.' =>
1559            Accumulate_Checksum ('.');
1560
1561            if Double_Char_Token ('.') then
1562               Token := Tok_Dot_Dot;
1563
1564               if Style_Check then
1565                  Style.Check_Dot_Dot;
1566               end if;
1567
1568               return;
1569
1570            elsif Source (Scan_Ptr + 1) in '0' .. '9' then
1571               Error_Msg_S ("numeric literal cannot start with point");
1572               Scan_Ptr := Scan_Ptr + 1;
1573
1574            else
1575               Scan_Ptr := Scan_Ptr + 1;
1576               Token := Tok_Dot;
1577               return;
1578            end if;
1579
1580         --  Equal, which can either be an equality operator, or part of the
1581         --  arrow (=>) compound delimiter.
1582
1583         when '=' =>
1584            Accumulate_Checksum ('=');
1585
1586            if Double_Char_Token ('>') then
1587               Token := Tok_Arrow;
1588
1589               if Style_Check then
1590                  Style.Check_Arrow;
1591               end if;
1592
1593               return;
1594
1595            elsif Source (Scan_Ptr + 1) = '=' then
1596               Error_Msg_S -- CODEFIX
1597                 ("== should be =");
1598               Scan_Ptr := Scan_Ptr + 1;
1599            end if;
1600
1601            Scan_Ptr := Scan_Ptr + 1;
1602            Token := Tok_Equal;
1603            return;
1604
1605         --  Greater than, which can be a greater than operator, greater than
1606         --  or equal operator, or first character of a right label bracket.
1607
1608         when '>' =>
1609            Accumulate_Checksum ('>');
1610
1611            if Double_Char_Token ('=') then
1612               Token := Tok_Greater_Equal;
1613               return;
1614
1615            elsif Double_Char_Token ('>') then
1616               Token := Tok_Greater_Greater;
1617               return;
1618
1619            else
1620               Scan_Ptr := Scan_Ptr + 1;
1621               Token := Tok_Greater;
1622               return;
1623            end if;
1624
1625         --  Less than, which can be a less than operator, less than or equal
1626         --  operator, or the first character of a left label bracket, or the
1627         --  first character of a box (<>) compound delimiter.
1628
1629         when '<' =>
1630            Accumulate_Checksum ('<');
1631
1632            if Double_Char_Token ('=') then
1633               Token := Tok_Less_Equal;
1634               return;
1635
1636            elsif Double_Char_Token ('>') then
1637               Token := Tok_Box;
1638
1639               if Style_Check then
1640                  Style.Check_Box;
1641               end if;
1642
1643               return;
1644
1645            elsif Double_Char_Token ('<') then
1646               Token := Tok_Less_Less;
1647               return;
1648
1649            else
1650               Scan_Ptr := Scan_Ptr + 1;
1651               Token := Tok_Less;
1652               return;
1653            end if;
1654
1655         --  Minus, which is either a subtraction operator, or the first
1656         --  character of double minus starting a comment
1657
1658         when '-' => Minus_Case : begin
1659            if Source (Scan_Ptr + 1) = '>' then
1660               Error_Msg_S ("invalid token");
1661               Scan_Ptr := Scan_Ptr + 2;
1662               Token := Tok_Arrow;
1663               return;
1664
1665            elsif Source (Scan_Ptr + 1) /= '-' then
1666               Accumulate_Checksum ('-');
1667               Scan_Ptr := Scan_Ptr + 1;
1668               Token := Tok_Minus;
1669               return;
1670
1671            --  Comment
1672
1673            else -- Source (Scan_Ptr + 1) = '-' then
1674               if Style_Check then
1675                  Style.Check_Comment;
1676               end if;
1677
1678               Scan_Ptr := Scan_Ptr + 2;
1679
1680               --  If we are in preprocessor mode with Replace_In_Comments set,
1681               --  then we return the "--" as a token on its own.
1682
1683               if Replace_In_Comments then
1684                  Token := Tok_Comment;
1685                  return;
1686               end if;
1687
1688               --  Otherwise scan out the comment
1689
1690               Start_Of_Comment := Scan_Ptr;
1691
1692               --  Loop to scan comment (this loop runs more than once only if
1693               --  a horizontal tab or other non-graphic character is scanned)
1694
1695               loop
1696                  --  Scan to non graphic character (opened up for speed)
1697
1698                  --  Note that we just eat left brackets, which means that
1699                  --  bracket notation cannot be used for end of line
1700                  --  characters in comments. This seems a reasonable choice,
1701                  --  since no one would ever use brackets notation in a real
1702                  --  program in this situation, and if we allow brackets
1703                  --  notation, we forbid some valid comments which contain a
1704                  --  brackets sequence that happens to match an end of line
1705                  --  character.
1706
1707                  loop
1708                     exit when Source (Scan_Ptr) not in Graphic_Character;
1709                     Scan_Ptr := Scan_Ptr + 1;
1710                     exit when Source (Scan_Ptr) not in Graphic_Character;
1711                     Scan_Ptr := Scan_Ptr + 1;
1712                     exit when Source (Scan_Ptr) not in Graphic_Character;
1713                     Scan_Ptr := Scan_Ptr + 1;
1714                     exit when Source (Scan_Ptr) not in Graphic_Character;
1715                     Scan_Ptr := Scan_Ptr + 1;
1716                     exit when Source (Scan_Ptr) not in Graphic_Character;
1717                     Scan_Ptr := Scan_Ptr + 1;
1718                  end loop;
1719
1720                  --  Keep going if horizontal tab
1721
1722                  if Source (Scan_Ptr) = HT then
1723                     if Style_Check then
1724                        Style.Check_HT;
1725                     end if;
1726
1727                     Scan_Ptr := Scan_Ptr + 1;
1728
1729                  --  Terminate scan of comment if line terminator
1730
1731                  elsif Source (Scan_Ptr) in Line_Terminator then
1732                     exit;
1733
1734                  --  Terminate scan of comment if end of file encountered
1735                  --  (embedded EOF character or real last character in file)
1736
1737                  elsif Source (Scan_Ptr) = EOF then
1738                     exit;
1739
1740                  --  If we have a wide character, we have to scan it out,
1741                  --  because it might be a legitimate line terminator
1742
1743                  elsif Start_Of_Wide_Character then
1744                     declare
1745                        Wptr : constant Source_Ptr := Scan_Ptr;
1746                        Code : Char_Code;
1747                        Err  : Boolean;
1748
1749                     begin
1750                        Scan_Wide (Source, Scan_Ptr, Code, Err);
1751
1752                        --  If not well formed wide character, then just skip
1753                        --  past it and ignore it.
1754
1755                        if Err then
1756                           Scan_Ptr := Wptr + 1;
1757
1758                        --  If UTF_32 terminator, terminate comment scan
1759
1760                        elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
1761                           Scan_Ptr := Wptr;
1762                           exit;
1763                        end if;
1764                     end;
1765
1766                  --  Keep going if character in 80-FF range, or is ESC. These
1767                  --  characters are allowed in comments by RM-2.1(1), 2.7(2).
1768                  --  They are allowed even in Ada 83 mode according to the
1769                  --  approved AI. ESC was added to the AI in June 93.
1770
1771                  elsif Source (Scan_Ptr) in Upper_Half_Character
1772                     or else Source (Scan_Ptr) = ESC
1773                  then
1774                     Scan_Ptr := Scan_Ptr + 1;
1775
1776                  --  Otherwise we have an illegal comment character
1777
1778                  else
1779                     Error_Illegal_Character;
1780                  end if;
1781               end loop;
1782
1783               --  Note that, except when comments are tokens, we do NOT
1784               --  execute a return here, instead we fall through to reexecute
1785               --  the scan loop to look for a token.
1786
1787               if Comment_Is_Token then
1788                  Name_Len := Integer (Scan_Ptr - Start_Of_Comment);
1789                  Name_Buffer (1 .. Name_Len) :=
1790                    String (Source (Start_Of_Comment .. Scan_Ptr - 1));
1791                  Comment_Id := Name_Find;
1792                  Token := Tok_Comment;
1793                  return;
1794               end if;
1795
1796               --  If the SPARK restriction is set for this unit, then generate
1797               --  a token Tok_SPARK_Hide for a SPARK HIDE directive.
1798
1799               if Restriction_Check_Required (SPARK)
1800                 and then Source (Start_Of_Comment) = '#'
1801               then
1802                  declare
1803                     Scan_SPARK_Ptr : Source_Ptr;
1804
1805                  begin
1806                     Scan_SPARK_Ptr := Start_Of_Comment + 1;
1807
1808                     --  Scan out blanks
1809
1810                     while Source (Scan_SPARK_Ptr) = ' '
1811                       or else Source (Scan_SPARK_Ptr) = HT
1812                     loop
1813                        Scan_SPARK_Ptr := Scan_SPARK_Ptr + 1;
1814                     end loop;
1815
1816                     --  Recognize HIDE directive. SPARK input cannot be
1817                     --  encoded as wide characters, so only deal with
1818                     --  lower/upper case.
1819
1820                     if (Source (Scan_SPARK_Ptr) = 'h'
1821                          or else Source (Scan_SPARK_Ptr) = 'H')
1822                       and then (Source (Scan_SPARK_Ptr + 1) = 'i'
1823                                  or else Source (Scan_SPARK_Ptr + 1) = 'I')
1824                       and then (Source (Scan_SPARK_Ptr + 2) = 'd'
1825                                  or else Source (Scan_SPARK_Ptr + 2) = 'D')
1826                       and then (Source (Scan_SPARK_Ptr + 3) = 'e'
1827                                  or else Source (Scan_SPARK_Ptr + 3) = 'E')
1828                       and then (Source (Scan_SPARK_Ptr + 4) = ' '
1829                                  or else Source (Scan_SPARK_Ptr + 4) = HT)
1830                     then
1831                        Token := Tok_SPARK_Hide;
1832                        return;
1833                     end if;
1834                  end;
1835               end if;
1836            end if;
1837         end Minus_Case;
1838
1839         --  Double quote or percent starting a string literal
1840
1841         when '"' | '%' =>
1842            Slit;
1843            Post_Scan;
1844            return;
1845
1846         --  Apostrophe. This can either be the start of a character literal,
1847         --  or an isolated apostrophe used in a qualified expression or an
1848         --  attribute. We treat it as a character literal if it does not
1849         --  follow a right parenthesis, identifier, the keyword ALL or
1850         --  a literal. This means that we correctly treat constructs like:
1851
1852         --    A := CHARACTER'('A');
1853
1854         --  Note that RM-2.2(7) does not require a separator between
1855         --  "CHARACTER" and "'" in the above.
1856
1857         when ''' => Char_Literal_Case : declare
1858            Code : Char_Code;
1859            Err  : Boolean;
1860
1861         begin
1862            Accumulate_Checksum (''');
1863            Scan_Ptr := Scan_Ptr + 1;
1864
1865            --  Here is where we make the test to distinguish the cases. Treat
1866            --  as apostrophe if previous token is an identifier, right paren
1867            --  or the reserved word "all" (latter case as in A.all'Address)
1868            --  (or the reserved word "project" in project files). Also treat
1869            --  it as apostrophe after a literal (this catches some legitimate
1870            --  cases, like A."abs"'Address, and also gives better error
1871            --  behavior for impossible cases like 123'xxx).
1872
1873            if Prev_Token = Tok_Identifier
1874               or else Prev_Token = Tok_Right_Paren
1875               or else Prev_Token = Tok_All
1876               or else Prev_Token = Tok_Project
1877               or else Prev_Token in Token_Class_Literal
1878            then
1879               Token := Tok_Apostrophe;
1880
1881               if Style_Check then
1882                  Style.Check_Apostrophe;
1883               end if;
1884
1885               return;
1886
1887            --  Otherwise the apostrophe starts a character literal
1888
1889            else
1890               --  Case of wide character literal
1891
1892               if Start_Of_Wide_Character then
1893                  Wptr := Scan_Ptr;
1894                  Scan_Wide (Source, Scan_Ptr, Code, Err);
1895                  Accumulate_Checksum (Code);
1896
1897                  if Err then
1898                     Error_Illegal_Wide_Character;
1899                     Code := Character'Pos (' ');
1900
1901                  --  In Ada 95 mode we allow any wide character in a character
1902                  --  literal, but in Ada 2005, the set of characters allowed
1903                  --  is restricted to graphic characters.
1904
1905                  elsif Ada_Version >= Ada_2005
1906                    and then Is_UTF_32_Non_Graphic (UTF_32 (Code))
1907                  then
1908                     Error_Msg -- CODEFIX????
1909                       ("(Ada 2005) non-graphic character not permitted " &
1910                        "in character literal", Wptr);
1911                  end if;
1912
1913                  if Source (Scan_Ptr) /= ''' then
1914                        Error_Msg_S ("missing apostrophe");
1915                  else
1916                     Scan_Ptr := Scan_Ptr + 1;
1917                  end if;
1918
1919               --  If we do not find a closing quote in the expected place then
1920               --  assume that we have a misguided attempt at a string literal.
1921
1922               --  However, if previous token is RANGE, then we return an
1923               --  apostrophe instead since this gives better error recovery
1924
1925               elsif Source (Scan_Ptr + 1) /= ''' then
1926                  if Prev_Token = Tok_Range then
1927                     Token := Tok_Apostrophe;
1928                     return;
1929
1930                  else
1931                     Scan_Ptr := Scan_Ptr - 1;
1932                     Error_Msg_S
1933                       ("strings are delimited by double quote character");
1934                     Slit;
1935                     Post_Scan;
1936                     return;
1937                  end if;
1938
1939               --  Otherwise we have a (non-wide) character literal
1940
1941               else
1942                  Accumulate_Checksum (Source (Scan_Ptr));
1943
1944                  if Source (Scan_Ptr) not in Graphic_Character then
1945                     if Source (Scan_Ptr) in Upper_Half_Character then
1946                        if Ada_Version = Ada_83 then
1947                           Error_Illegal_Character;
1948                        end if;
1949
1950                     else
1951                        Error_Illegal_Character;
1952                     end if;
1953                  end if;
1954
1955                  Code := Get_Char_Code (Source (Scan_Ptr));
1956                  Scan_Ptr := Scan_Ptr + 2;
1957               end if;
1958
1959               --  Fall through here with Scan_Ptr updated past the closing
1960               --  quote, and Code set to the Char_Code value for the literal
1961
1962               Accumulate_Checksum (''');
1963               Token := Tok_Char_Literal;
1964               Set_Character_Literal_Name (Code);
1965               Token_Name := Name_Find;
1966               Character_Code := Code;
1967               Post_Scan;
1968               return;
1969            end if;
1970         end Char_Literal_Case;
1971
1972         --  Right parenthesis
1973
1974         when ')' =>
1975            Accumulate_Checksum (')');
1976            Scan_Ptr := Scan_Ptr + 1;
1977            Token := Tok_Right_Paren;
1978
1979            if Style_Check then
1980               Style.Check_Right_Paren;
1981            end if;
1982
1983            return;
1984
1985         --  Right bracket or right brace, treated as right paren
1986
1987         when ']' | '}' =>
1988            Error_Msg_S ("illegal character, replaced by "")""");
1989            Scan_Ptr := Scan_Ptr + 1;
1990            Token := Tok_Right_Paren;
1991            return;
1992
1993         --  Slash (can be division operator or first character of not equal)
1994
1995         when '/' =>
1996            Accumulate_Checksum ('/');
1997
1998            if Double_Char_Token ('=') then
1999               Token := Tok_Not_Equal;
2000               return;
2001            else
2002               Scan_Ptr := Scan_Ptr + 1;
2003               Token := Tok_Slash;
2004               return;
2005            end if;
2006
2007         --  Semicolon
2008
2009         when ';' =>
2010            Accumulate_Checksum (';');
2011            Scan_Ptr := Scan_Ptr + 1;
2012            Token := Tok_Semicolon;
2013
2014            if Style_Check then
2015               Style.Check_Semicolon;
2016            end if;
2017
2018            return;
2019
2020         --  Vertical bar
2021
2022         when '|' => Vertical_Bar_Case : begin
2023            Accumulate_Checksum ('|');
2024
2025            --  Special check for || to give nice message
2026
2027            if Source (Scan_Ptr + 1) = '|' then
2028               Error_Msg_S -- CODEFIX
2029                 ("""'|'|"" should be `OR ELSE`");
2030               Scan_Ptr := Scan_Ptr + 2;
2031               Token := Tok_Or;
2032               return;
2033
2034            else
2035               Scan_Ptr := Scan_Ptr + 1;
2036               Token := Tok_Vertical_Bar;
2037
2038               if Style_Check then
2039                  Style.Check_Vertical_Bar;
2040               end if;
2041
2042               Post_Scan;
2043               return;
2044            end if;
2045         end Vertical_Bar_Case;
2046
2047         --  Exclamation, replacement character for vertical bar
2048
2049         when '!' => Exclamation_Case : begin
2050            Accumulate_Checksum ('!');
2051
2052            if Source (Scan_Ptr + 1) = '=' then
2053               Error_Msg_S -- CODEFIX
2054                 ("'!= should be /=");
2055               Scan_Ptr := Scan_Ptr + 2;
2056               Token := Tok_Not_Equal;
2057               return;
2058
2059            else
2060               Scan_Ptr := Scan_Ptr + 1;
2061               Token := Tok_Vertical_Bar;
2062               Post_Scan;
2063               return;
2064            end if;
2065         end Exclamation_Case;
2066
2067         --  Plus
2068
2069         when '+' => Plus_Case : begin
2070            Accumulate_Checksum ('+');
2071            Scan_Ptr := Scan_Ptr + 1;
2072            Token := Tok_Plus;
2073            return;
2074         end Plus_Case;
2075
2076         --  Digits starting a numeric literal
2077
2078         when '0' .. '9' =>
2079
2080            --  First a bit of a scan ahead to see if we have a case of an
2081            --  identifier starting with a digit (remembering exponent case).
2082
2083            declare
2084               C : constant Character := Source (Scan_Ptr + 1);
2085
2086            begin
2087               --  OK literal if digit followed by digit or underscore
2088
2089               if C in '0' .. '9' or else C = '_' then
2090                  null;
2091
2092               --  OK literal if digit not followed by identifier char
2093
2094               elsif not Identifier_Char (C) then
2095                  null;
2096
2097               --  OK literal if digit followed by e/E followed by digit/sign.
2098               --  We also allow underscore after the E, which is an error, but
2099               --  better handled by Nlit than deciding this is an identifier.
2100
2101               elsif (C = 'e' or else C = 'E')
2102                 and then (Source (Scan_Ptr + 2) in '0' .. '9'
2103                             or else Source (Scan_Ptr + 2) = '+'
2104                             or else Source (Scan_Ptr + 2) = '-'
2105                             or else Source (Scan_Ptr + 2) = '_')
2106               then
2107                  null;
2108
2109               --  Here we have what really looks like an identifier that
2110               --  starts with a digit, so give error msg.
2111
2112               else
2113                  Error_Msg_S ("identifier may not start with digit");
2114                  Name_Len := 1;
2115                  Underline_Found := False;
2116                  Name_Buffer (1) := Source (Scan_Ptr);
2117                  Accumulate_Checksum (Name_Buffer (1));
2118                  Scan_Ptr := Scan_Ptr + 1;
2119                  goto Scan_Identifier;
2120               end if;
2121            end;
2122
2123            --  Here we have an OK integer literal
2124
2125            Nlit;
2126
2127            --  Check for proper delimiter, ignoring other format characters
2128
2129            Skip_Other_Format_Characters;
2130
2131            if Identifier_Char (Source (Scan_Ptr)) then
2132               Error_Msg_S
2133                 ("delimiter required between literal and identifier");
2134            end if;
2135
2136            Post_Scan;
2137            return;
2138
2139         --  Lower case letters
2140
2141         when 'a' .. 'z' =>
2142            Name_Len := 1;
2143            Underline_Found := False;
2144            Name_Buffer (1) := Source (Scan_Ptr);
2145            Accumulate_Checksum (Name_Buffer (1));
2146            Scan_Ptr := Scan_Ptr + 1;
2147            goto Scan_Identifier;
2148
2149         --  Upper case letters
2150
2151         when 'A' .. 'Z' =>
2152            Name_Len := 1;
2153            Underline_Found := False;
2154            Name_Buffer (1) :=
2155              Character'Val (Character'Pos (Source (Scan_Ptr)) + 32);
2156            Accumulate_Checksum (Name_Buffer (1));
2157            Scan_Ptr := Scan_Ptr + 1;
2158            goto Scan_Identifier;
2159
2160         --  Underline character
2161
2162         when '_' =>
2163            if Special_Characters ('_') then
2164               Token_Ptr := Scan_Ptr;
2165               Scan_Ptr := Scan_Ptr + 1;
2166               Token := Tok_Special;
2167               Special_Character := '_';
2168               return;
2169            end if;
2170
2171            Error_Msg_S ("identifier cannot start with underline");
2172            Name_Len := 1;
2173            Name_Buffer (1) := '_';
2174            Scan_Ptr := Scan_Ptr + 1;
2175            Underline_Found := False;
2176            goto Scan_Identifier;
2177
2178         --  Space (not possible, because we scanned past blanks)
2179
2180         when ' ' =>
2181            raise Program_Error;
2182
2183         --  Characters in top half of ASCII 8-bit chart
2184
2185         when Upper_Half_Character =>
2186
2187            --  Wide character case
2188
2189            if Upper_Half_Encoding then
2190               goto Scan_Wide_Character;
2191
2192            --  Otherwise we have OK Latin-1 character
2193
2194            else
2195               --  Upper half characters may possibly be identifier letters
2196               --  but can never be digits, so Identifier_Char can be used to
2197               --  test for a valid start of identifier character.
2198
2199               if Identifier_Char (Source (Scan_Ptr)) then
2200                  Name_Len := 0;
2201                  Underline_Found := False;
2202                  goto Scan_Identifier;
2203               else
2204                  Error_Illegal_Character;
2205               end if;
2206            end if;
2207
2208         when ESC =>
2209
2210            --  ESC character, possible start of identifier if wide characters
2211            --  using ESC encoding are allowed in identifiers, which we can
2212            --  tell by looking at the Identifier_Char flag for ESC, which is
2213            --  only true if these conditions are met. In Ada 2005 mode, may
2214            --  also be valid UTF_32 space or line terminator character.
2215
2216            if Identifier_Char (ESC) then
2217               Name_Len := 0;
2218               goto Scan_Wide_Character;
2219            else
2220               Error_Illegal_Character;
2221            end if;
2222
2223         --  Invalid control characters
2224
2225         when NUL | SOH | STX | ETX | EOT | ENQ | ACK | BEL | BS  | ASCII.SO |
2226              SI  | DLE | DC1 | DC2 | DC3 | DC4 | NAK | SYN | ETB | CAN |
2227              EM  | FS  | GS  | RS  | US  | DEL
2228         =>
2229            Error_Illegal_Character;
2230
2231         --  Invalid graphic characters
2232
2233         when '#' | '$' | '?' | '@' | '`' | '\' | '^' | '~' =>
2234
2235            --  If Set_Special_Character has been called for this character,
2236            --  set Scans.Special_Character and return a Special token.
2237
2238            if Special_Characters (Source (Scan_Ptr)) then
2239               Token_Ptr := Scan_Ptr;
2240               Token := Tok_Special;
2241               Special_Character := Source (Scan_Ptr);
2242               Scan_Ptr := Scan_Ptr + 1;
2243               return;
2244
2245            --  Check for something looking like a preprocessor directive
2246
2247            elsif Source (Scan_Ptr) = '#'
2248              and then (Source (Scan_Ptr + 1 .. Scan_Ptr + 2) = "if"
2249                          or else
2250                        Source (Scan_Ptr + 1 .. Scan_Ptr + 5) = "elsif"
2251                          or else
2252                        Source (Scan_Ptr + 1 .. Scan_Ptr + 4) = "else"
2253                          or else
2254                        Source (Scan_Ptr + 1 .. Scan_Ptr + 3) = "end")
2255            then
2256               Error_Msg_S
2257                 ("preprocessor directive ignored, preprocessor not active");
2258
2259               --  Skip to end of line
2260
2261               loop
2262                  if Source (Scan_Ptr) in Graphic_Character
2263                       or else
2264                     Source (Scan_Ptr) = HT
2265                  then
2266                     Scan_Ptr := Scan_Ptr + 1;
2267
2268                  --  Done if line terminator or EOF
2269
2270                  elsif Source (Scan_Ptr) in Line_Terminator
2271                          or else
2272                        Source (Scan_Ptr) = EOF
2273                  then
2274                     exit;
2275
2276                  --  If we have a wide character, we have to scan it out,
2277                  --  because it might be a legitimate line terminator
2278
2279                  elsif Start_Of_Wide_Character then
2280                     declare
2281                        Wptr : constant Source_Ptr := Scan_Ptr;
2282                        Code : Char_Code;
2283                        Err  : Boolean;
2284
2285                     begin
2286                        Scan_Wide (Source, Scan_Ptr, Code, Err);
2287
2288                        --  If not well formed wide character, then just skip
2289                        --  past it and ignore it.
2290
2291                        if Err then
2292                           Scan_Ptr := Wptr + 1;
2293
2294                        --  If UTF_32 terminator, terminate comment scan
2295
2296                        elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
2297                           Scan_Ptr := Wptr;
2298                           exit;
2299                        end if;
2300                     end;
2301
2302                  --  Else keep going (don't worry about bad comment chars
2303                  --  in this context, we just want to find the end of line.
2304
2305                  else
2306                     Scan_Ptr := Scan_Ptr + 1;
2307                  end if;
2308               end loop;
2309
2310            --  Otherwise, this is an illegal character
2311
2312            else
2313               Error_Illegal_Character;
2314            end if;
2315
2316         --  End switch on non-blank character
2317
2318         end case;
2319
2320      --  End loop past format effectors. The exit from this loop is by
2321      --  executing a return statement following completion of token scan
2322      --  (control never falls out of this loop to the code which follows)
2323
2324      end loop;
2325
2326      --  Wide_Character scanning routine. On entry we have encountered the
2327      --  initial character of a wide character sequence.
2328
2329      <<Scan_Wide_Character>>
2330
2331         declare
2332            Code : Char_Code;
2333            Cat  : Category;
2334            Err  : Boolean;
2335
2336         begin
2337            Wptr := Scan_Ptr;
2338            Scan_Wide (Source, Scan_Ptr, Code, Err);
2339
2340            --  If bad wide character, signal error and continue scan
2341
2342            if Err then
2343               Error_Illegal_Wide_Character;
2344               goto Scan_Next_Character;
2345            end if;
2346
2347            Cat := Get_Category (UTF_32 (Code));
2348
2349            --  If OK letter, reset scan ptr and go scan identifier
2350
2351            if Is_UTF_32_Letter (Cat) then
2352               Scan_Ptr := Wptr;
2353               Name_Len := 0;
2354               Underline_Found := False;
2355               goto Scan_Identifier;
2356
2357            --  If OK wide space, ignore and keep scanning (we do not include
2358            --  any ignored spaces in checksum)
2359
2360            elsif Is_UTF_32_Space (Cat) then
2361               goto Scan_Next_Character;
2362
2363            --  If other format character, ignore and keep scanning (again we
2364            --  do not include in the checksum) (this is for AI-0079).
2365
2366            elsif Is_UTF_32_Other (Cat) then
2367               goto Scan_Next_Character;
2368
2369            --  If OK wide line terminator, terminate current line
2370
2371            elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then
2372               Scan_Ptr := Wptr;
2373               goto Scan_Line_Terminator;
2374
2375            --  Punctuation is an error (at start of identifier)
2376
2377            elsif Is_UTF_32_Punctuation (Cat) then
2378               Error_Msg ("identifier cannot start with punctuation", Wptr);
2379               Scan_Ptr := Wptr;
2380               Name_Len := 0;
2381               Underline_Found := False;
2382               goto Scan_Identifier;
2383
2384            --  Mark character is an error (at start of identifier)
2385
2386            elsif Is_UTF_32_Mark (Cat) then
2387               Error_Msg ("identifier cannot start with mark character", Wptr);
2388               Scan_Ptr := Wptr;
2389               Name_Len := 0;
2390               Underline_Found := False;
2391               goto Scan_Identifier;
2392
2393            --  Extended digit character is an error. Could be bad start of
2394            --  identifier or bad literal. Not worth doing too much to try to
2395            --  distinguish these cases, but we will do a little bit.
2396
2397            elsif Is_UTF_32_Digit (Cat) then
2398               Error_Msg
2399                 ("identifier cannot start with digit character", Wptr);
2400               Scan_Ptr := Wptr;
2401               Name_Len := 0;
2402               Underline_Found := False;
2403               goto Scan_Identifier;
2404
2405            --  All other wide characters are illegal here
2406
2407            else
2408               Error_Illegal_Wide_Character;
2409               goto Scan_Next_Character;
2410            end if;
2411         end;
2412
2413      --  Routine to scan line terminator. On entry Scan_Ptr points to a
2414      --  character which is one of FF,LR,CR,VT, or one of the wide characters
2415      --  that is treated as a line terminator.
2416
2417      <<Scan_Line_Terminator>>
2418
2419         --  Check line too long
2420
2421         Check_End_Of_Line;
2422
2423         --  Set Token_Ptr, if End_Of_Line is a token, for the case when it is
2424         --  a physical line.
2425
2426         if End_Of_Line_Is_Token then
2427            Token_Ptr := Scan_Ptr;
2428         end if;
2429
2430         declare
2431            Physical : Boolean;
2432
2433         begin
2434            Skip_Line_Terminators (Scan_Ptr, Physical);
2435
2436            --  If we are at start of physical line, update scan pointers to
2437            --  reflect the start of the new line.
2438
2439            if Physical then
2440               Current_Line_Start       := Scan_Ptr;
2441               Start_Column             := Set_Start_Column;
2442               First_Non_Blank_Location := Scan_Ptr;
2443
2444               --  If End_Of_Line is a token, we return it as it is a
2445               --  physical line.
2446
2447               if End_Of_Line_Is_Token then
2448                  Token := Tok_End_Of_Line;
2449                  return;
2450               end if;
2451            end if;
2452         end;
2453
2454         goto Scan_Next_Character;
2455
2456      --  Identifier scanning routine. On entry, some initial characters of
2457      --  the identifier may have already been stored in Name_Buffer. If so,
2458      --  Name_Len has the number of characters stored, otherwise Name_Len is
2459      --  set to zero on entry. Underline_Found is also set False on entry.
2460
2461      <<Scan_Identifier>>
2462
2463         --  This loop scans as fast as possible past lower half letters and
2464         --  digits, which we expect to be the most common characters.
2465
2466         loop
2467            if Source (Scan_Ptr) in 'a' .. 'z'
2468              or else Source (Scan_Ptr) in '0' .. '9'
2469            then
2470               Name_Buffer (Name_Len + 1) := Source (Scan_Ptr);
2471               Accumulate_Checksum (Source (Scan_Ptr));
2472
2473            elsif Source (Scan_Ptr) in 'A' .. 'Z' then
2474               Name_Buffer (Name_Len + 1) :=
2475                 Character'Val (Character'Pos (Source (Scan_Ptr)) + 32);
2476               Accumulate_Checksum (Name_Buffer (Name_Len + 1));
2477
2478            else
2479               exit;
2480            end if;
2481
2482            Underline_Found := False;
2483            Scan_Ptr := Scan_Ptr + 1;
2484            Name_Len := Name_Len + 1;
2485         end loop;
2486
2487         --  If we fall through, then we have encountered either an underline
2488         --  character, or an extended identifier character (i.e. one from the
2489         --  upper half), or a wide character, or an identifier terminator. The
2490         --  initial test speeds us up in the most common case where we have
2491         --  an identifier terminator. Note that ESC is an identifier character
2492         --  only if a wide character encoding method that uses ESC encoding
2493         --  is active, so if we find an ESC character we know that we have a
2494         --  wide character.
2495
2496         if Identifier_Char (Source (Scan_Ptr))
2497           or else (Source (Scan_Ptr) in Upper_Half_Character
2498                     and then Upper_Half_Encoding)
2499         then
2500            --  Case of underline
2501
2502            if Source (Scan_Ptr) = '_' then
2503               Accumulate_Checksum ('_');
2504
2505               if Underline_Found then
2506                  Error_No_Double_Underline;
2507               else
2508                  Underline_Found := True;
2509                  Name_Len := Name_Len + 1;
2510                  Name_Buffer (Name_Len) := '_';
2511               end if;
2512
2513               Scan_Ptr := Scan_Ptr + 1;
2514               goto Scan_Identifier;
2515
2516            --  Upper half character
2517
2518            elsif Source (Scan_Ptr) in Upper_Half_Character
2519              and then not Upper_Half_Encoding
2520            then
2521               Accumulate_Checksum (Source (Scan_Ptr));
2522               Store_Encoded_Character
2523                 (Get_Char_Code (Fold_Lower (Source (Scan_Ptr))));
2524               Scan_Ptr := Scan_Ptr + 1;
2525               Underline_Found := False;
2526               goto Scan_Identifier;
2527
2528            --  Left bracket not followed by a quote terminates an identifier.
2529            --  This is an error, but we don't want to give a junk error msg
2530            --  about wide characters in this case!
2531
2532            elsif Source (Scan_Ptr) = '['
2533              and then Source (Scan_Ptr + 1) /= '"'
2534            then
2535               null;
2536
2537            --  We know we have a wide character encoding here (the current
2538            --  character is either ESC, left bracket, or an upper half
2539            --  character depending on the encoding method).
2540
2541            else
2542               --  Scan out the wide character and insert the appropriate
2543               --  encoding into the name table entry for the identifier.
2544
2545               declare
2546                  Code : Char_Code;
2547                  Err  : Boolean;
2548                  Chr  : Character;
2549                  Cat  : Category;
2550
2551               begin
2552                  Wptr := Scan_Ptr;
2553                  Scan_Wide (Source, Scan_Ptr, Code, Err);
2554
2555                  --  If error, signal error
2556
2557                  if Err then
2558                     Error_Illegal_Wide_Character;
2559
2560                  --  If the character scanned is a normal identifier
2561                  --  character, then we treat it that way.
2562
2563                  elsif In_Character_Range (Code)
2564                    and then Identifier_Char (Get_Character (Code))
2565                  then
2566                     Chr := Get_Character (Code);
2567                     Accumulate_Checksum (Chr);
2568                     Store_Encoded_Character
2569                       (Get_Char_Code (Fold_Lower (Chr)));
2570                     Underline_Found := False;
2571
2572                  --  Here if not a normal identifier character
2573
2574                  else
2575                     Cat := Get_Category (UTF_32 (Code));
2576
2577                     --  Wide character in Unicode category "Other, Format"
2578                     --  is not accepted in an identifier. This is because it
2579                     --  it is considered a security risk (AI-0091).
2580
2581                     --  However, it is OK for such a character to appear at
2582                     --  the end of an identifier.
2583
2584                     if Is_UTF_32_Other (Cat) then
2585                        if not Identifier_Char (Source (Scan_Ptr)) then
2586                           goto Scan_Identifier_Complete;
2587                        else
2588                           Error_Msg
2589                             ("identifier cannot contain other_format "
2590                              & "character", Wptr);
2591                           goto Scan_Identifier;
2592                        end if;
2593
2594                     --  Wide character in category Separator,Space terminates
2595
2596                     elsif Is_UTF_32_Space (Cat) then
2597                        goto Scan_Identifier_Complete;
2598                     end if;
2599
2600                     --  Here if wide character is part of the identifier
2601
2602                     --  Make sure we are allowing wide characters in
2603                     --  identifiers. Note that we allow wide character
2604                     --  notation for an OK identifier character. This in
2605                     --  particular allows bracket or other notation to be
2606                     --  used for upper half letters.
2607
2608                     --  Wide characters are always allowed in Ada 2005
2609
2610                     if Identifier_Character_Set /= 'w'
2611                       and then Ada_Version < Ada_2005
2612                     then
2613                        Error_Msg
2614                          ("wide character not allowed in identifier", Wptr);
2615                     end if;
2616
2617                     --  If OK letter, store it folding to upper case. Note
2618                     --  that we include the folded letter in the checksum.
2619
2620                     if Is_UTF_32_Letter (Cat) then
2621                        Code :=
2622                          Char_Code (UTF_32_To_Upper_Case (UTF_32 (Code)));
2623                        Accumulate_Checksum (Code);
2624                        Store_Encoded_Character (Code);
2625                        Underline_Found := False;
2626
2627                     --  If OK extended digit or mark, then store it
2628
2629                     elsif Is_UTF_32_Digit (Cat)
2630                       or else Is_UTF_32_Mark (Cat)
2631                     then
2632                        Accumulate_Checksum (Code);
2633                        Store_Encoded_Character (Code);
2634                        Underline_Found := False;
2635
2636                     --  Wide punctuation is also stored, but counts as an
2637                     --  underline character for error checking purposes.
2638
2639                     elsif Is_UTF_32_Punctuation (Cat) then
2640                        Accumulate_Checksum (Code);
2641
2642                        if Underline_Found then
2643                           declare
2644                              Cend : constant Source_Ptr := Scan_Ptr;
2645                           begin
2646                              Scan_Ptr := Wptr;
2647                              Error_No_Double_Underline;
2648                              Scan_Ptr := Cend;
2649                           end;
2650
2651                        else
2652                           Store_Encoded_Character (Code);
2653                           Underline_Found := True;
2654                        end if;
2655
2656                     --  Any other wide character is not acceptable
2657
2658                     else
2659                        Error_Msg
2660                          ("invalid wide character in identifier", Wptr);
2661                     end if;
2662                  end if;
2663
2664                  goto Scan_Identifier;
2665               end;
2666            end if;
2667         end if;
2668
2669      --  Scan of identifier is complete. The identifier is stored in
2670      --  Name_Buffer, and Scan_Ptr points past the last character.
2671
2672      <<Scan_Identifier_Complete>>
2673         Token_Name := Name_Find;
2674
2675         --  Check for identifier ending with underline or punctuation char
2676
2677         if Underline_Found then
2678            Underline_Found := False;
2679
2680            if Source (Scan_Ptr - 1) = '_' then
2681               Error_Msg
2682                 ("identifier cannot end with underline", Scan_Ptr - 1);
2683            else
2684               Error_Msg
2685                 ("identifier cannot end with punctuation character", Wptr);
2686            end if;
2687         end if;
2688
2689         --  We will assume it is an identifier, not a keyword, so that the
2690         --  checksum is independent of the Ada version.
2691
2692         Token := Tok_Identifier;
2693
2694         --  Here is where we check if it was a keyword
2695
2696         if Is_Keyword_Name (Token_Name) then
2697            if Opt.Checksum_GNAT_6_3 then
2698               Token := Token_Type'Val (Get_Name_Table_Byte (Token_Name));
2699
2700               if Checksum_Accumulate_Token_Checksum then
2701                  if Checksum_GNAT_5_03 then
2702                     Accumulate_Token_Checksum_GNAT_5_03;
2703                  else
2704                     Accumulate_Token_Checksum_GNAT_6_3;
2705                  end if;
2706               end if;
2707
2708            else
2709               Accumulate_Token_Checksum;
2710               Token := Token_Type'Val (Get_Name_Table_Byte (Token_Name));
2711            end if;
2712
2713            --  Keyword style checks
2714
2715            if Style_Check then
2716
2717               --  Deal with possible style check for non-lower case keyword,
2718               --  but we don't treat ACCESS, DELTA, DIGITS, RANGE as keywords
2719               --  for this purpose if they appear as attribute designators.
2720               --  Actually we only check the first character for speed.
2721
2722               --  Ada 2005 (AI-284): Do not apply the style check in case of
2723               --  "pragma Interface"
2724
2725               --  Ada 2005 (AI-340): Do not apply the style check in case of
2726               --  MOD attribute.
2727
2728               if Source (Token_Ptr) <= 'Z'
2729                 and then (Prev_Token /= Tok_Apostrophe
2730                           or else
2731                             (Token /= Tok_Access and then
2732                              Token /= Tok_Delta  and then
2733                              Token /= Tok_Digits and then
2734                              Token /= Tok_Mod    and then
2735                              Token /= Tok_Range))
2736                       and then (Token /= Tok_Interface
2737                                  or else
2738                                    (Token = Tok_Interface
2739                                      and then Prev_Token /= Tok_Pragma))
2740               then
2741                  Style.Non_Lower_Case_Keyword;
2742               end if;
2743
2744               --  Check THEN/ELSE style rules. These do not apply to AND THEN
2745               --  or OR ELSE, and do not apply in if expressions.
2746
2747               if (Token = Tok_Then and then Prev_Token /= Tok_And)
2748                    or else
2749                  (Token = Tok_Else and then Prev_Token /= Tok_Or)
2750               then
2751                  if Inside_If_Expression = 0 then
2752                     Style.Check_Separate_Stmt_Lines;
2753                  end if;
2754               end if;
2755            end if;
2756
2757            --  We must reset Token_Name since this is not an identifier and
2758            --  if we leave Token_Name set, the parser gets confused because
2759            --  it thinks it is dealing with an identifier instead of the
2760            --  corresponding keyword.
2761
2762            Token_Name := No_Name;
2763            return;
2764
2765         --  It is an identifier after all
2766
2767         else
2768            if Checksum_Accumulate_Token_Checksum then
2769               Accumulate_Token_Checksum;
2770            end if;
2771
2772            Post_Scan;
2773            return;
2774         end if;
2775   end Scan;
2776
2777   --------------------------
2778   -- Set_Comment_As_Token --
2779   --------------------------
2780
2781   procedure Set_Comment_As_Token (Value : Boolean) is
2782   begin
2783      Comment_Is_Token := Value;
2784   end Set_Comment_As_Token;
2785
2786   ------------------------------
2787   -- Set_End_Of_Line_As_Token --
2788   ------------------------------
2789
2790   procedure Set_End_Of_Line_As_Token (Value : Boolean) is
2791   begin
2792      End_Of_Line_Is_Token := Value;
2793   end Set_End_Of_Line_As_Token;
2794
2795   ---------------------------
2796   -- Set_Special_Character --
2797   ---------------------------
2798
2799   procedure Set_Special_Character (C : Character) is
2800   begin
2801      case C is
2802         when '#' | '$' | '_' | '?' | '@' | '`' | '\' | '^' | '~' =>
2803            Special_Characters (C) := True;
2804
2805         when others =>
2806            null;
2807      end case;
2808   end Set_Special_Character;
2809
2810   ----------------------
2811   -- Set_Start_Column --
2812   ----------------------
2813
2814   --  Note: it seems at first glance a little expensive to compute this value
2815   --  for every source line (since it is certainly not used for all source
2816   --  lines). On the other hand, it doesn't take much more work to skip past
2817   --  the initial white space on the line counting the columns than it would
2818   --  to scan past the white space using the standard scanning circuits.
2819
2820   function Set_Start_Column return Column_Number is
2821      Start_Column : Column_Number := 0;
2822
2823   begin
2824      --  Outer loop scans past horizontal tab characters
2825
2826      Tabs_Loop : loop
2827
2828         --  Inner loop scans past blanks as fast as possible, bumping Scan_Ptr
2829         --  past the blanks and adjusting Start_Column to account for them.
2830
2831         Blanks_Loop : loop
2832            if Source (Scan_Ptr) = ' ' then
2833               if Source (Scan_Ptr + 1) = ' ' then
2834                  if Source (Scan_Ptr + 2) = ' ' then
2835                     if Source (Scan_Ptr + 3) = ' ' then
2836                        if Source (Scan_Ptr + 4) = ' ' then
2837                           if Source (Scan_Ptr + 5) = ' ' then
2838                              if Source (Scan_Ptr + 6) = ' ' then
2839                                 Scan_Ptr := Scan_Ptr + 7;
2840                                 Start_Column := Start_Column + 7;
2841                              else
2842                                 Scan_Ptr := Scan_Ptr + 6;
2843                                 Start_Column := Start_Column + 6;
2844                                 exit Blanks_Loop;
2845                              end if;
2846                           else
2847                              Scan_Ptr := Scan_Ptr + 5;
2848                              Start_Column := Start_Column + 5;
2849                              exit Blanks_Loop;
2850                           end if;
2851                        else
2852                           Scan_Ptr := Scan_Ptr + 4;
2853                           Start_Column := Start_Column + 4;
2854                           exit Blanks_Loop;
2855                        end if;
2856                     else
2857                        Scan_Ptr := Scan_Ptr + 3;
2858                        Start_Column := Start_Column + 3;
2859                        exit Blanks_Loop;
2860                     end if;
2861                  else
2862                     Scan_Ptr := Scan_Ptr + 2;
2863                     Start_Column := Start_Column + 2;
2864                     exit Blanks_Loop;
2865                  end if;
2866               else
2867                  Scan_Ptr := Scan_Ptr + 1;
2868                  Start_Column := Start_Column + 1;
2869                  exit Blanks_Loop;
2870               end if;
2871            else
2872               exit Blanks_Loop;
2873            end if;
2874         end loop Blanks_Loop;
2875
2876         --  Outer loop keeps going only if a horizontal tab follows
2877
2878         if Source (Scan_Ptr) = HT then
2879            if Style_Check then
2880               Style.Check_HT;
2881            end if;
2882
2883            Scan_Ptr := Scan_Ptr + 1;
2884            Start_Column := (Start_Column / 8) * 8 + 8;
2885         else
2886            exit Tabs_Loop;
2887         end if;
2888      end loop Tabs_Loop;
2889
2890      return Start_Column;
2891
2892   --  A constraint error can happen only if we have a compiler with checks on
2893   --  and a line with a ludicrous number of tabs or spaces at the start. In
2894   --  such a case, we really don't care if Start_Column is right or not.
2895
2896   exception
2897      when Constraint_Error =>
2898         return Start_Column;
2899   end Set_Start_Column;
2900
2901end Scng;
2902