]>
Commit | Line | Data |
---|---|---|
fbf5a39b AC |
1 | ------------------------------------------------------------------------------ |
2 | -- -- | |
3 | -- GNAT COMPILER COMPONENTS -- | |
4 | -- -- | |
5 | -- S C N G -- | |
6 | -- -- | |
7 | -- B o d y -- | |
8 | -- -- | |
4b490c1e | 9 | -- Copyright (C) 1992-2020, Free Software Foundation, Inc. -- |
fbf5a39b AC |
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- -- | |
b5c84c3c | 13 | -- ware Foundation; either version 3, or (at your option) any later ver- -- |
fbf5a39b AC |
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 -- | |
b5c84c3c RD |
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. -- | |
fbf5a39b AC |
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 | ||
ee2ba856 | 26 | with Atree; use Atree; |
fbf5a39b | 27 | with Csets; use Csets; |
23d0d17f | 28 | with Hostparm; use Hostparm; |
fbf5a39b AC |
29 | with Namet; use Namet; |
30 | with Opt; use Opt; | |
f9adb9d4 AC |
31 | with Restrict; use Restrict; |
32 | with Rident; use Rident; | |
fbf5a39b AC |
33 | with Scans; use Scans; |
34 | with Sinput; use Sinput; | |
35 | with Snames; use Snames; | |
36 | with Stringt; use Stringt; | |
37 | with Stylesw; use Stylesw; | |
38 | with Uintp; use Uintp; | |
39 | with Urealp; use Urealp; | |
40 | with Widechar; use Widechar; | |
41 | ||
9e9df9da AC |
42 | pragma Warnings (Off); |
43 | -- This package is used also by gnatcoll | |
fbf5a39b | 44 | with System.CRC32; |
0e701488 | 45 | with System.UTF_32; use System.UTF_32; |
fbf5a39b | 46 | with System.WCh_Con; use System.WCh_Con; |
9e9df9da | 47 | pragma Warnings (On); |
fbf5a39b AC |
48 | |
49 | package 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 | ||
c45b6ae0 AC |
57 | Comment_Is_Token : Boolean := False; |
58 | -- True if comments are tokens | |
59 | ||
fbf5a39b AC |
60 | End_Of_Line_Is_Token : Boolean := False; |
61 | -- True if End_Of_Line is a token | |
62 | ||
63 | ----------------------- | |
64 | -- Local Subprograms -- | |
65 | ----------------------- | |
66 | ||
6b6fcd3e AC |
67 | procedure Accumulate_Token_Checksum; |
68 | pragma Inline (Accumulate_Token_Checksum); | |
1ce9dff3 | 69 | -- Called after each numeric literal and identifier/keyword. For keywords, |
6782b1ef | 70 | -- the token used is Tok_Identifier. This allows detection of additional |
1ce9dff3 VC |
71 | -- spaces added in sources when using the builder switch -m. |
72 | ||
b251750b VC |
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 | |
1ce9dff3 VC |
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. | |
b251750b VC |
83 | -- This procedure is never used in the compiler or gnatmake, only in |
84 | -- gprbuild. | |
6b6fcd3e | 85 | |
fbf5a39b AC |
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 | |
82c80734 | 119 | if C > 16#FFFF# then |
357ac4df RD |
120 | Accumulate_Checksum (Character'Val (C / 2 ** 24)); |
121 | Accumulate_Checksum (Character'Val ((C / 2 ** 16) mod 256)); | |
82c80734 RD |
122 | Accumulate_Checksum (Character'Val ((C / 256) mod 256)); |
123 | else | |
124 | Accumulate_Checksum (Character'Val (C / 256)); | |
125 | end if; | |
126 | ||
fbf5a39b AC |
127 | Accumulate_Checksum (Character'Val (C mod 256)); |
128 | end Accumulate_Checksum; | |
129 | ||
6b6fcd3e AC |
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 | ||
b251750b VC |
141 | ---------------------------------------- |
142 | -- Accumulate_Token_Checksum_GNAT_6_3 -- | |
143 | ---------------------------------------- | |
1ce9dff3 | 144 | |
b251750b | 145 | procedure Accumulate_Token_Checksum_GNAT_6_3 is |
1ce9dff3 VC |
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 | |
d8f43ee6 HK |
152 | when Tok_Abs |
153 | | Tok_Abstract | |
154 | | Tok_Access | |
155 | | Tok_Aliased | |
156 | | Tok_All | |
157 | | Tok_Ampersand | |
158 | | Tok_And | |
159 | | Tok_Apostrophe | |
160 | | Tok_Array | |
161 | | Tok_Asterisk | |
162 | | Tok_At | |
d43584ca | 163 | | Tok_At_Sign |
d8f43ee6 HK |
164 | | Tok_Body |
165 | | Tok_Box | |
166 | | Tok_Char_Literal | |
167 | | Tok_Colon | |
168 | | Tok_Colon_Equal | |
169 | | Tok_Comma | |
170 | | Tok_Constant | |
171 | | Tok_Delta | |
172 | | Tok_Digits | |
173 | | Tok_Do | |
174 | | Tok_Dot | |
175 | | Tok_Double_Asterisk | |
176 | | Tok_Equal | |
177 | | Tok_Greater | |
178 | | Tok_Greater_Equal | |
179 | | Tok_Greater_Greater | |
180 | | Tok_Identifier | |
181 | | Tok_In | |
182 | | Tok_Integer_Literal | |
183 | | Tok_Interface | |
184 | | Tok_Is | |
3c08de34 | 185 | | Tok_Left_Bracket |
d8f43ee6 HK |
186 | | Tok_Left_Paren |
187 | | Tok_Less | |
188 | | Tok_Less_Equal | |
189 | | Tok_Limited | |
190 | | Tok_Minus | |
191 | | Tok_Mod | |
192 | | Tok_New | |
193 | | Tok_Not | |
194 | | Tok_Not_Equal | |
195 | | Tok_Null | |
196 | | Tok_Of | |
197 | | Tok_Operator_Symbol | |
198 | | Tok_Or | |
199 | | Tok_Others | |
200 | | Tok_Out | |
201 | | Tok_Plus | |
202 | | Tok_Range | |
203 | | Tok_Real_Literal | |
204 | | Tok_Record | |
205 | | Tok_Rem | |
206 | | Tok_Renames | |
207 | | Tok_Reverse | |
3c08de34 | 208 | | Tok_Right_Bracket |
d8f43ee6 HK |
209 | | Tok_Right_Paren |
210 | | Tok_Slash | |
211 | | Tok_String_Literal | |
212 | | Tok_Xor | |
213 | => | |
1ce9dff3 VC |
214 | System.CRC32.Update |
215 | (System.CRC32.CRC32 (Checksum), | |
216 | Character'Val (Token_Type'Pos (Token))); | |
217 | ||
218 | when Tok_Some => | |
1ce9dff3 VC |
219 | System.CRC32.Update |
220 | (System.CRC32.CRC32 (Checksum), | |
221 | Character'Val (Token_Type'Pos (Tok_Identifier))); | |
222 | ||
d8f43ee6 HK |
223 | when No_Token |
224 | | Tok_Abort | |
225 | | Tok_Accept | |
226 | | Tok_Arrow | |
227 | | Tok_Begin | |
228 | | Tok_Case | |
229 | | Tok_Comment | |
230 | | Tok_Declare | |
231 | | Tok_Delay | |
232 | | Tok_Dot_Dot | |
233 | | Tok_Else | |
234 | | Tok_Elsif | |
235 | | Tok_End | |
236 | | Tok_End_Of_Line | |
237 | | Tok_Entry | |
238 | | Tok_EOF | |
239 | | Tok_Exception | |
240 | | Tok_Exit | |
241 | | Tok_Extends | |
242 | | Tok_External | |
243 | | Tok_External_As_List | |
244 | | Tok_For | |
245 | | Tok_Function | |
246 | | Tok_Generic | |
247 | | Tok_Goto | |
248 | | Tok_If | |
249 | | Tok_Less_Less | |
250 | | Tok_Loop | |
251 | | Tok_Overriding | |
252 | | Tok_Package | |
253 | | Tok_Pragma | |
254 | | Tok_Private | |
255 | | Tok_Procedure | |
256 | | Tok_Project | |
257 | | Tok_Protected | |
258 | | Tok_Raise | |
259 | | Tok_Requeue | |
260 | | Tok_Return | |
261 | | Tok_Select | |
262 | | Tok_Semicolon | |
263 | | Tok_Separate | |
264 | | Tok_SPARK_Hide | |
265 | | Tok_Special | |
266 | | Tok_Subtype | |
267 | | Tok_Synchronized | |
268 | | Tok_Tagged | |
269 | | Tok_Task | |
270 | | Tok_Terminate | |
271 | | Tok_Then | |
272 | | Tok_Type | |
273 | | Tok_Until | |
274 | | Tok_Use | |
275 | | Tok_Vertical_Bar | |
276 | | Tok_When | |
277 | | Tok_While | |
278 | | Tok_With | |
279 | => | |
1ce9dff3 VC |
280 | System.CRC32.Update |
281 | (System.CRC32.CRC32 (Checksum), | |
282 | Character'Val (Token_Type'Pos (Token_Type'Pred (Token)))); | |
283 | end case; | |
b251750b | 284 | end Accumulate_Token_Checksum_GNAT_6_3; |
1ce9dff3 | 285 | |
b251750b VC |
286 | ----------------------------------------- |
287 | -- Accumulate_Token_Checksum_GNAT_5_03 -- | |
288 | ----------------------------------------- | |
1ce9dff3 | 289 | |
b251750b | 290 | procedure Accumulate_Token_Checksum_GNAT_5_03 is |
1ce9dff3 VC |
291 | begin |
292 | -- Individual values of Token_Type are used, instead of subranges, so | |
293 | -- that additions or suppressions of enumerated values in type | |
294 | -- Token_Type are detected by the compiler. | |
295 | ||
296 | case Token is | |
d8f43ee6 HK |
297 | when Tok_Abs |
298 | | Tok_Abstract | |
299 | | Tok_Access | |
300 | | Tok_Aliased | |
301 | | Tok_All | |
302 | | Tok_Ampersand | |
303 | | Tok_And | |
304 | | Tok_Apostrophe | |
305 | | Tok_Array | |
306 | | Tok_Asterisk | |
307 | | Tok_At | |
ae33543c | 308 | | Tok_At_Sign |
d8f43ee6 HK |
309 | | Tok_Body |
310 | | Tok_Box | |
311 | | Tok_Char_Literal | |
312 | | Tok_Colon | |
313 | | Tok_Colon_Equal | |
314 | | Tok_Comma | |
315 | | Tok_Constant | |
316 | | Tok_Delta | |
317 | | Tok_Digits | |
318 | | Tok_Do | |
319 | | Tok_Dot | |
320 | | Tok_Double_Asterisk | |
321 | | Tok_Equal | |
322 | | Tok_Greater | |
323 | | Tok_Greater_Equal | |
324 | | Tok_Greater_Greater | |
325 | | Tok_Identifier | |
326 | | Tok_In | |
327 | | Tok_Integer_Literal | |
328 | | Tok_Is | |
3c08de34 | 329 | | Tok_Left_Bracket |
d8f43ee6 HK |
330 | | Tok_Left_Paren |
331 | | Tok_Less | |
332 | | Tok_Less_Equal | |
333 | | Tok_Minus | |
334 | | Tok_Mod | |
335 | | Tok_New | |
336 | | Tok_Not | |
337 | | Tok_Not_Equal | |
338 | | Tok_Null | |
339 | | Tok_Operator_Symbol | |
340 | | Tok_Or | |
341 | | Tok_Others | |
342 | | Tok_Plus | |
343 | | Tok_Range | |
344 | | Tok_Real_Literal | |
345 | | Tok_Rem | |
3c08de34 | 346 | | Tok_Right_Bracket |
d8f43ee6 HK |
347 | | Tok_Right_Paren |
348 | | Tok_Slash | |
349 | | Tok_String_Literal | |
350 | | Tok_Xor | |
351 | => | |
1ce9dff3 VC |
352 | System.CRC32.Update |
353 | (System.CRC32.CRC32 (Checksum), | |
354 | Character'Val (Token_Type'Pos (Token))); | |
355 | ||
d8f43ee6 HK |
356 | when Tok_Interface |
357 | | Tok_Overriding | |
358 | | Tok_Some | |
359 | | Tok_Synchronized | |
360 | => | |
1ce9dff3 VC |
361 | System.CRC32.Update |
362 | (System.CRC32.CRC32 (Checksum), | |
363 | Character'Val (Token_Type'Pos (Tok_Identifier))); | |
364 | ||
d8f43ee6 HK |
365 | when Tok_Limited |
366 | | Tok_Of | |
367 | | Tok_Out | |
368 | | Tok_Record | |
369 | | Tok_Renames | |
370 | | Tok_Reverse | |
371 | => | |
1ce9dff3 VC |
372 | System.CRC32.Update |
373 | (System.CRC32.CRC32 (Checksum), | |
374 | Character'Val (Token_Type'Pos (Token) - 1)); | |
375 | ||
d8f43ee6 HK |
376 | when Tok_Abort |
377 | | Tok_Accept | |
378 | | Tok_Begin | |
379 | | Tok_Case | |
380 | | Tok_Declare | |
381 | | Tok_Delay | |
382 | | Tok_Else | |
383 | | Tok_Elsif | |
384 | | Tok_End | |
385 | | Tok_Entry | |
386 | | Tok_Exception | |
387 | | Tok_Exit | |
388 | | Tok_For | |
389 | | Tok_Goto | |
390 | | Tok_If | |
391 | | Tok_Less_Less | |
392 | | Tok_Loop | |
393 | | Tok_Pragma | |
394 | | Tok_Protected | |
395 | | Tok_Raise | |
396 | | Tok_Requeue | |
397 | | Tok_Return | |
398 | | Tok_Select | |
399 | | Tok_Subtype | |
400 | | Tok_Tagged | |
401 | | Tok_Task | |
402 | | Tok_Terminate | |
403 | | Tok_Then | |
404 | | Tok_Type | |
405 | | Tok_Until | |
406 | | Tok_When | |
407 | | Tok_While | |
408 | => | |
1ce9dff3 VC |
409 | System.CRC32.Update |
410 | (System.CRC32.CRC32 (Checksum), | |
411 | Character'Val (Token_Type'Pos (Token) - 2)); | |
412 | ||
d8f43ee6 HK |
413 | when No_Token |
414 | | Tok_Arrow | |
415 | | Tok_Comment | |
416 | | Tok_Dot_Dot | |
417 | | Tok_End_Of_Line | |
418 | | Tok_EOF | |
419 | | Tok_Extends | |
420 | | Tok_External | |
421 | | Tok_External_As_List | |
422 | | Tok_Function | |
423 | | Tok_Generic | |
424 | | Tok_Package | |
425 | | Tok_Private | |
426 | | Tok_Procedure | |
427 | | Tok_Project | |
428 | | Tok_Semicolon | |
429 | | Tok_Separate | |
430 | | Tok_SPARK_Hide | |
431 | | Tok_Special | |
432 | | Tok_Use | |
433 | | Tok_Vertical_Bar | |
434 | | Tok_With | |
435 | => | |
1ce9dff3 VC |
436 | System.CRC32.Update |
437 | (System.CRC32.CRC32 (Checksum), | |
438 | Character'Val (Token_Type'Pos (Token) - 4)); | |
439 | end case; | |
b251750b | 440 | end Accumulate_Token_Checksum_GNAT_5_03; |
1ce9dff3 | 441 | |
29077c18 AC |
442 | ----------------------- |
443 | -- Check_End_Of_Line -- | |
444 | ----------------------- | |
445 | ||
446 | procedure Check_End_Of_Line is | |
447 | Len : constant Int := | |
448 | Int (Scan_Ptr) - | |
449 | Int (Current_Line_Start) - | |
450 | Wide_Char_Byte_Count; | |
451 | ||
452 | -- Start of processing for Check_End_Of_Line | |
453 | ||
454 | begin | |
455 | if Style_Check then | |
456 | Style.Check_Line_Terminator (Len); | |
457 | end if; | |
458 | ||
459 | -- Deal with checking maximum line length | |
460 | ||
461 | if Style_Check and Style_Check_Max_Line_Length then | |
462 | Style.Check_Line_Max_Length (Len); | |
463 | ||
464 | -- If style checking is inactive, check maximum line length against | |
465 | -- standard value. | |
466 | ||
467 | elsif Len > Max_Line_Length then | |
468 | Error_Msg | |
469 | ("this line is too long", | |
470 | Current_Line_Start + Source_Ptr (Max_Line_Length)); | |
471 | end if; | |
472 | ||
473 | -- Now one more checking circuit. Normally we are only enforcing a limit | |
474 | -- of physical characters, with tabs counting as one character. But if | |
475 | -- after tab expansion we would have a total line length that exceeded | |
476 | -- 32766, that would really cause trouble, because column positions | |
477 | -- would exceed the maximum we allow for a column count. Note: the limit | |
478 | -- is 32766 rather than 32767, since we use a value of 32767 for special | |
479 | -- purposes (see Sinput). Now we really do not want to go messing with | |
480 | -- tabs in the normal case, so what we do is to check for a line that | |
481 | -- has more than 4096 physical characters. Any shorter line could not | |
482 | -- be a problem, even if it was all tabs. | |
483 | ||
484 | if Len >= 4096 then | |
485 | declare | |
486 | Col : Natural; | |
487 | Ptr : Source_Ptr; | |
488 | ||
489 | begin | |
490 | Col := 1; | |
491 | Ptr := Current_Line_Start; | |
492 | loop | |
493 | exit when Ptr = Scan_Ptr; | |
494 | ||
495 | if Source (Ptr) = ASCII.HT then | |
496 | Col := (Col - 1 + 8) / 8 * 8 + 1; | |
497 | else | |
498 | Col := Col + 1; | |
499 | end if; | |
500 | ||
501 | if Col > 32766 then | |
502 | Error_Msg | |
503 | ("this line is longer than 32766 characters", | |
504 | Current_Line_Start); | |
505 | raise Unrecoverable_Error; | |
506 | end if; | |
507 | ||
508 | Ptr := Ptr + 1; | |
509 | end loop; | |
510 | end; | |
511 | end if; | |
512 | ||
513 | -- Reset wide character byte count for next line | |
514 | ||
515 | Wide_Char_Byte_Count := 0; | |
516 | end Check_End_Of_Line; | |
517 | ||
fbf5a39b AC |
518 | ---------------------------- |
519 | -- Determine_Token_Casing -- | |
520 | ---------------------------- | |
521 | ||
522 | function Determine_Token_Casing return Casing_Type is | |
523 | begin | |
524 | return Determine_Casing (Source (Token_Ptr .. Scan_Ptr - 1)); | |
525 | end Determine_Token_Casing; | |
526 | ||
527 | ------------------------- | |
528 | -- Initialize_Checksum -- | |
529 | ------------------------- | |
530 | ||
531 | procedure Initialize_Checksum is | |
532 | begin | |
533 | System.CRC32.Initialize (System.CRC32.CRC32 (Checksum)); | |
534 | end Initialize_Checksum; | |
535 | ||
536 | ------------------------ | |
537 | -- Initialize_Scanner -- | |
538 | ------------------------ | |
539 | ||
68e2ea27 | 540 | procedure Initialize_Scanner (Index : Source_File_Index) is |
fbf5a39b | 541 | begin |
82c80734 RD |
542 | -- Establish reserved words |
543 | ||
68e2ea27 | 544 | Scans.Initialize_Ada_Keywords; |
fbf5a39b AC |
545 | |
546 | -- Initialize scan control variables | |
547 | ||
548 | Current_Source_File := Index; | |
549 | Source := Source_Text (Current_Source_File); | |
fbf5a39b AC |
550 | Scan_Ptr := Source_First (Current_Source_File); |
551 | Token := No_Token; | |
552 | Token_Ptr := Scan_Ptr; | |
553 | Current_Line_Start := Scan_Ptr; | |
554 | Token_Node := Empty; | |
555 | Token_Name := No_Name; | |
556 | Start_Column := Set_Start_Column; | |
557 | First_Non_Blank_Location := Scan_Ptr; | |
558 | ||
559 | Initialize_Checksum; | |
d52f1094 | 560 | Wide_Char_Byte_Count := 0; |
fbf5a39b | 561 | |
5f3ab6fb | 562 | -- Do not call Scan, otherwise the License stuff does not work in Scn |
fbf5a39b AC |
563 | |
564 | end Initialize_Scanner; | |
565 | ||
566 | ------------------------------ | |
567 | -- Reset_Special_Characters -- | |
568 | ------------------------------ | |
569 | ||
570 | procedure Reset_Special_Characters is | |
571 | begin | |
572 | Special_Characters := (others => False); | |
573 | end Reset_Special_Characters; | |
574 | ||
575 | ---------- | |
576 | -- Scan -- | |
577 | ---------- | |
578 | ||
579 | procedure Scan is | |
580 | ||
c45b6ae0 | 581 | Start_Of_Comment : Source_Ptr; |
82c80734 RD |
582 | -- Record start of comment position |
583 | ||
584 | Underline_Found : Boolean; | |
585 | -- During scanning of an identifier, set to True if last character | |
586 | -- scanned was an underline or other punctuation character. This | |
587 | -- is used to flag the error of two underlines/punctuations in a | |
588 | -- row or ending an identifier with a underline/punctuation. Here | |
589 | -- punctuation means any UTF_32 character in the Unicode category | |
590 | -- Punctuation,Connector. | |
591 | ||
592 | Wptr : Source_Ptr; | |
593 | -- Used to remember start of last wide character scanned | |
c45b6ae0 | 594 | |
fbf5a39b AC |
595 | function Double_Char_Token (C : Character) return Boolean; |
596 | -- This function is used for double character tokens like := or <>. It | |
597 | -- checks if the character following Source (Scan_Ptr) is C, and if so | |
598 | -- bumps Scan_Ptr past the pair of characters and returns True. A space | |
599 | -- between the two characters is also recognized with an appropriate | |
600 | -- error message being issued. If C is not present, False is returned. | |
601 | -- Note that Double_Char_Token can only be used for tokens defined in | |
602 | -- the Ada syntax (it's use for error cases like && is not appropriate | |
603 | -- since we do not want a junk message for a case like &-space-&). | |
604 | ||
605 | procedure Error_Illegal_Character; | |
82c80734 RD |
606 | -- Give illegal character error, Scan_Ptr points to character. On |
607 | -- return, Scan_Ptr is bumped past the illegal character. | |
fbf5a39b AC |
608 | |
609 | procedure Error_Illegal_Wide_Character; | |
610 | -- Give illegal wide character message. On return, Scan_Ptr is bumped | |
611 | -- past the illegal character, which may still leave us pointing to | |
a90bd866 | 612 | -- junk, not much we can do if the escape sequence is messed up. |
fbf5a39b | 613 | |
fbf5a39b | 614 | procedure Error_No_Double_Underline; |
82c80734 RD |
615 | -- Signal error of two underline or punctuation characters in a row. |
616 | -- Called with Scan_Ptr pointing to second underline/punctuation char. | |
fbf5a39b AC |
617 | |
618 | procedure Nlit; | |
619 | -- This is the procedure for scanning out numeric literals. On entry, | |
620 | -- Scan_Ptr points to the digit that starts the numeric literal (the | |
621 | -- checksum for this character has not been accumulated yet). On return | |
622 | -- Scan_Ptr points past the last character of the numeric literal, Token | |
623 | -- and Token_Node are set appropriately, and the checksum is updated. | |
624 | ||
625 | procedure Slit; | |
626 | -- This is the procedure for scanning out string literals. On entry, | |
627 | -- Scan_Ptr points to the opening string quote (the checksum for this | |
628 | -- character has not been accumulated yet). On return Scan_Ptr points | |
629 | -- past the closing quote of the string literal, Token and Token_Node | |
12a13f01 | 630 | -- are set appropriately, and the checksum is updated. |
fbf5a39b | 631 | |
2385e007 AC |
632 | procedure Skip_Other_Format_Characters; |
633 | -- Skips past any "other format" category characters at the current | |
634 | -- cursor location (does not skip past spaces or any other characters). | |
635 | ||
636 | function Start_Of_Wide_Character return Boolean; | |
637 | -- Returns True if the scan pointer is pointing to the start of a wide | |
638 | -- character sequence, does not modify the scan pointer in any case. | |
639 | ||
fbf5a39b AC |
640 | ----------------------- |
641 | -- Double_Char_Token -- | |
642 | ----------------------- | |
643 | ||
644 | function Double_Char_Token (C : Character) return Boolean is | |
645 | begin | |
646 | if Source (Scan_Ptr + 1) = C then | |
647 | Accumulate_Checksum (C); | |
648 | Scan_Ptr := Scan_Ptr + 2; | |
649 | return True; | |
650 | ||
651 | elsif Source (Scan_Ptr + 1) = ' ' | |
652 | and then Source (Scan_Ptr + 2) = C | |
653 | then | |
654 | Scan_Ptr := Scan_Ptr + 1; | |
ed2233dc AC |
655 | Error_Msg_S -- CODEFIX |
656 | ("no space allowed here"); | |
fbf5a39b AC |
657 | Scan_Ptr := Scan_Ptr + 2; |
658 | return True; | |
659 | ||
660 | else | |
661 | return False; | |
662 | end if; | |
663 | end Double_Char_Token; | |
664 | ||
665 | ----------------------------- | |
666 | -- Error_Illegal_Character -- | |
667 | ----------------------------- | |
668 | ||
669 | procedure Error_Illegal_Character is | |
670 | begin | |
671 | Error_Msg_S ("illegal character"); | |
672 | Scan_Ptr := Scan_Ptr + 1; | |
673 | end Error_Illegal_Character; | |
674 | ||
675 | ---------------------------------- | |
676 | -- Error_Illegal_Wide_Character -- | |
677 | ---------------------------------- | |
678 | ||
679 | procedure Error_Illegal_Wide_Character is | |
680 | begin | |
73f0dc7a | 681 | Scan_Ptr := Scan_Ptr + 1; |
82c80734 | 682 | Error_Msg ("illegal wide character", Wptr); |
fbf5a39b AC |
683 | end Error_Illegal_Wide_Character; |
684 | ||
fbf5a39b AC |
685 | ------------------------------- |
686 | -- Error_No_Double_Underline -- | |
687 | ------------------------------- | |
688 | ||
689 | procedure Error_No_Double_Underline is | |
690 | begin | |
82c80734 RD |
691 | Underline_Found := False; |
692 | ||
693 | -- There are four cases, and we special case the messages | |
694 | ||
695 | if Source (Scan_Ptr) = '_' then | |
696 | if Source (Scan_Ptr - 1) = '_' then | |
483c78cb | 697 | Error_Msg_S -- CODEFIX |
82c80734 RD |
698 | ("two consecutive underlines not permitted"); |
699 | else | |
ed2233dc | 700 | Error_Msg_S ("underline cannot follow punctuation character"); |
82c80734 RD |
701 | end if; |
702 | ||
703 | else | |
704 | if Source (Scan_Ptr - 1) = '_' then | |
ed2233dc | 705 | Error_Msg_S ("punctuation character cannot follow underline"); |
82c80734 | 706 | else |
ed2233dc | 707 | Error_Msg_S |
82c80734 RD |
708 | ("two consecutive punctuation characters not permitted"); |
709 | end if; | |
710 | end if; | |
fbf5a39b AC |
711 | end Error_No_Double_Underline; |
712 | ||
713 | ---------- | |
714 | -- Nlit -- | |
715 | ---------- | |
716 | ||
717 | procedure Nlit is | |
718 | ||
719 | C : Character; | |
720 | -- Current source program character | |
721 | ||
722 | Base_Char : Character; | |
723 | -- Either # or : (character at start of based number) | |
724 | ||
725 | Base : Int; | |
726 | -- Value of base | |
727 | ||
728 | UI_Base : Uint; | |
729 | -- Value of base in Uint format | |
730 | ||
731 | UI_Int_Value : Uint; | |
732 | -- Value of integer scanned by Scan_Integer in Uint format | |
733 | ||
734 | UI_Num_Value : Uint; | |
735 | -- Value of integer in numeric value being scanned | |
736 | ||
737 | Scale : Int; | |
738 | -- Scale value for real literal | |
739 | ||
740 | UI_Scale : Uint; | |
741 | -- Scale in Uint format | |
742 | ||
743 | Exponent_Is_Negative : Boolean; | |
744 | -- Set true for negative exponent | |
745 | ||
746 | Extended_Digit_Value : Int; | |
747 | -- Extended digit value | |
748 | ||
749 | Point_Scanned : Boolean; | |
750 | -- Flag for decimal point scanned in numeric literal | |
751 | ||
752 | ----------------------- | |
753 | -- Local Subprograms -- | |
754 | ----------------------- | |
755 | ||
756 | procedure Error_Digit_Expected; | |
757 | -- Signal error of bad digit, Scan_Ptr points to the location at | |
758 | -- which the digit was expected on input, and is unchanged on return. | |
759 | ||
760 | procedure Scan_Integer; | |
5c211bfd AC |
761 | -- Scan integer literal. On entry, Scan_Ptr points to a digit, on |
762 | -- exit Scan_Ptr points past the last character of the integer. | |
6b6fcd3e | 763 | -- |
82c80734 | 764 | -- For each digit encountered, UI_Int_Value is multiplied by 10, and |
5c211bfd AC |
765 | -- the value of the digit added to the result. In addition, the value |
766 | -- in Scale is decremented by one for each actual digit scanned. | |
fbf5a39b AC |
767 | |
768 | -------------------------- | |
769 | -- Error_Digit_Expected -- | |
770 | -------------------------- | |
771 | ||
772 | procedure Error_Digit_Expected is | |
773 | begin | |
774 | Error_Msg_S ("digit expected"); | |
775 | end Error_Digit_Expected; | |
776 | ||
15ce9ca2 AC |
777 | ------------------ |
778 | -- Scan_Integer -- | |
779 | ------------------ | |
fbf5a39b AC |
780 | |
781 | procedure Scan_Integer is | |
782 | C : Character; | |
783 | -- Next character scanned | |
784 | ||
785 | begin | |
786 | C := Source (Scan_Ptr); | |
787 | ||
788 | -- Loop through digits (allowing underlines) | |
789 | ||
790 | loop | |
791 | Accumulate_Checksum (C); | |
792 | UI_Int_Value := | |
793 | UI_Int_Value * 10 + (Character'Pos (C) - Character'Pos ('0')); | |
794 | Scan_Ptr := Scan_Ptr + 1; | |
795 | Scale := Scale - 1; | |
796 | C := Source (Scan_Ptr); | |
797 | ||
82c80734 RD |
798 | -- Case of underline encountered |
799 | ||
fbf5a39b | 800 | if C = '_' then |
30c20106 AC |
801 | |
802 | -- We do not accumulate the '_' in the checksum, so that | |
803 | -- 1_234 is equivalent to 1234, and does not trigger | |
804 | -- compilation for "minimal recompilation" (gnatmake -m). | |
fbf5a39b AC |
805 | |
806 | loop | |
807 | Scan_Ptr := Scan_Ptr + 1; | |
808 | C := Source (Scan_Ptr); | |
809 | exit when C /= '_'; | |
810 | Error_No_Double_Underline; | |
811 | end loop; | |
812 | ||
813 | if C not in '0' .. '9' then | |
814 | Error_Digit_Expected; | |
815 | exit; | |
816 | end if; | |
817 | ||
818 | else | |
819 | exit when C not in '0' .. '9'; | |
820 | end if; | |
821 | end loop; | |
fbf5a39b AC |
822 | end Scan_Integer; |
823 | ||
704228bd | 824 | -- Start of processing for Nlit |
fbf5a39b AC |
825 | |
826 | begin | |
827 | Base := 10; | |
828 | UI_Base := Uint_10; | |
829 | UI_Int_Value := Uint_0; | |
c90b2058 | 830 | Based_Literal_Uses_Colon := False; |
fbf5a39b AC |
831 | Scale := 0; |
832 | Scan_Integer; | |
fbf5a39b AC |
833 | Point_Scanned := False; |
834 | UI_Num_Value := UI_Int_Value; | |
835 | ||
82c80734 RD |
836 | -- Various possibilities now for continuing the literal are period, |
837 | -- E/e (for exponent), or :/# (for based literal). | |
fbf5a39b AC |
838 | |
839 | Scale := 0; | |
840 | C := Source (Scan_Ptr); | |
841 | ||
842 | if C = '.' then | |
843 | ||
844 | -- Scan out point, but do not scan past .. which is a range | |
845 | -- sequence, and must not be eaten up scanning a numeric literal. | |
846 | ||
847 | while C = '.' and then Source (Scan_Ptr + 1) /= '.' loop | |
848 | Accumulate_Checksum ('.'); | |
849 | ||
850 | if Point_Scanned then | |
851 | Error_Msg_S ("duplicate point ignored"); | |
852 | end if; | |
853 | ||
854 | Point_Scanned := True; | |
855 | Scan_Ptr := Scan_Ptr + 1; | |
856 | C := Source (Scan_Ptr); | |
857 | ||
858 | if C not in '0' .. '9' then | |
859 | Error_Msg | |
860 | ("real literal cannot end with point", Scan_Ptr - 1); | |
861 | else | |
862 | Scan_Integer; | |
863 | UI_Num_Value := UI_Int_Value; | |
864 | end if; | |
865 | end loop; | |
866 | ||
82c80734 RD |
867 | -- Based literal case. The base is the value we already scanned. |
868 | -- In the case of colon, we insist that the following character | |
869 | -- is indeed an extended digit or a period. This catches a number | |
870 | -- of common errors, as well as catching the well known tricky | |
871 | -- bug otherwise arising from "x : integer range 1 .. 10:= 6;" | |
fbf5a39b AC |
872 | |
873 | elsif C = '#' | |
874 | or else (C = ':' and then | |
875 | (Source (Scan_Ptr + 1) = '.' | |
876 | or else | |
877 | Source (Scan_Ptr + 1) in '0' .. '9' | |
878 | or else | |
879 | Source (Scan_Ptr + 1) in 'A' .. 'Z' | |
880 | or else | |
881 | Source (Scan_Ptr + 1) in 'a' .. 'z')) | |
882 | then | |
fbf5a39b AC |
883 | Accumulate_Checksum (C); |
884 | Base_Char := C; | |
885 | UI_Base := UI_Int_Value; | |
886 | ||
c90b2058 AC |
887 | if Base_Char = ':' then |
888 | Based_Literal_Uses_Colon := True; | |
889 | end if; | |
890 | ||
fbf5a39b AC |
891 | if UI_Base < 2 or else UI_Base > 16 then |
892 | Error_Msg_SC ("base not 2-16"); | |
893 | UI_Base := Uint_16; | |
894 | end if; | |
895 | ||
896 | Base := UI_To_Int (UI_Base); | |
897 | Scan_Ptr := Scan_Ptr + 1; | |
898 | ||
899 | -- Scan out extended integer [. integer] | |
900 | ||
901 | C := Source (Scan_Ptr); | |
902 | UI_Int_Value := Uint_0; | |
903 | Scale := 0; | |
904 | ||
905 | loop | |
906 | if C in '0' .. '9' then | |
907 | Accumulate_Checksum (C); | |
908 | Extended_Digit_Value := | |
909 | Int'(Character'Pos (C)) - Int'(Character'Pos ('0')); | |
910 | ||
911 | elsif C in 'A' .. 'F' then | |
912 | Accumulate_Checksum (Character'Val (Character'Pos (C) + 32)); | |
913 | Extended_Digit_Value := | |
914 | Int'(Character'Pos (C)) - Int'(Character'Pos ('A')) + 10; | |
915 | ||
916 | elsif C in 'a' .. 'f' then | |
917 | Accumulate_Checksum (C); | |
918 | Extended_Digit_Value := | |
919 | Int'(Character'Pos (C)) - Int'(Character'Pos ('a')) + 10; | |
920 | ||
921 | else | |
922 | Error_Msg_S ("extended digit expected"); | |
923 | exit; | |
924 | end if; | |
925 | ||
926 | if Extended_Digit_Value >= Base then | |
927 | Error_Msg_S ("digit '>= base"); | |
928 | end if; | |
929 | ||
930 | UI_Int_Value := UI_Int_Value * UI_Base + Extended_Digit_Value; | |
931 | Scale := Scale - 1; | |
932 | Scan_Ptr := Scan_Ptr + 1; | |
933 | C := Source (Scan_Ptr); | |
934 | ||
935 | if C = '_' then | |
936 | loop | |
937 | Accumulate_Checksum ('_'); | |
938 | Scan_Ptr := Scan_Ptr + 1; | |
939 | C := Source (Scan_Ptr); | |
940 | exit when C /= '_'; | |
941 | Error_No_Double_Underline; | |
942 | end loop; | |
943 | ||
944 | elsif C = '.' then | |
945 | Accumulate_Checksum ('.'); | |
946 | ||
947 | if Point_Scanned then | |
948 | Error_Msg_S ("duplicate point ignored"); | |
949 | end if; | |
950 | ||
951 | Scan_Ptr := Scan_Ptr + 1; | |
952 | C := Source (Scan_Ptr); | |
953 | Point_Scanned := True; | |
954 | Scale := 0; | |
955 | ||
956 | elsif C = Base_Char then | |
957 | Accumulate_Checksum (C); | |
958 | Scan_Ptr := Scan_Ptr + 1; | |
959 | exit; | |
960 | ||
961 | elsif C = '#' or else C = ':' then | |
962 | Error_Msg_S ("based number delimiters must match"); | |
963 | Scan_Ptr := Scan_Ptr + 1; | |
964 | exit; | |
965 | ||
966 | elsif not Identifier_Char (C) then | |
967 | if Base_Char = '#' then | |
ed2233dc AC |
968 | Error_Msg_S -- CODEFIX |
969 | ("missing '#"); | |
fbf5a39b | 970 | else |
ed2233dc AC |
971 | Error_Msg_S -- CODEFIX |
972 | ("missing ':"); | |
fbf5a39b AC |
973 | end if; |
974 | ||
975 | exit; | |
976 | end if; | |
977 | ||
978 | end loop; | |
979 | ||
980 | UI_Num_Value := UI_Int_Value; | |
981 | end if; | |
982 | ||
983 | -- Scan out exponent | |
984 | ||
985 | if not Point_Scanned then | |
986 | Scale := 0; | |
987 | UI_Scale := Uint_0; | |
988 | else | |
989 | UI_Scale := UI_From_Int (Scale); | |
990 | end if; | |
991 | ||
992 | if Source (Scan_Ptr) = 'e' or else Source (Scan_Ptr) = 'E' then | |
993 | Accumulate_Checksum ('e'); | |
994 | Scan_Ptr := Scan_Ptr + 1; | |
995 | Exponent_Is_Negative := False; | |
996 | ||
997 | if Source (Scan_Ptr) = '+' then | |
998 | Accumulate_Checksum ('+'); | |
999 | Scan_Ptr := Scan_Ptr + 1; | |
1000 | ||
1001 | elsif Source (Scan_Ptr) = '-' then | |
1002 | Accumulate_Checksum ('-'); | |
1003 | ||
1004 | if not Point_Scanned then | |
1005 | Error_Msg_S | |
1006 | ("negative exponent not allowed for integer literal"); | |
1007 | else | |
1008 | Exponent_Is_Negative := True; | |
1009 | end if; | |
1010 | ||
1011 | Scan_Ptr := Scan_Ptr + 1; | |
1012 | end if; | |
1013 | ||
1014 | UI_Int_Value := Uint_0; | |
1015 | ||
1016 | if Source (Scan_Ptr) in '0' .. '9' then | |
1017 | Scan_Integer; | |
1018 | else | |
1019 | Error_Digit_Expected; | |
1020 | end if; | |
1021 | ||
1022 | if Exponent_Is_Negative then | |
1023 | UI_Scale := UI_Scale - UI_Int_Value; | |
1024 | else | |
1025 | UI_Scale := UI_Scale + UI_Int_Value; | |
1026 | end if; | |
1027 | end if; | |
1028 | ||
1029 | -- Case of real literal to be returned | |
1030 | ||
1031 | if Point_Scanned then | |
1032 | Token := Tok_Real_Literal; | |
1033 | Real_Literal_Value := | |
1034 | UR_From_Components ( | |
1035 | Num => UI_Num_Value, | |
1036 | Den => -UI_Scale, | |
1037 | Rbase => Base); | |
1038 | ||
82c80734 | 1039 | -- Case of integer literal to be returned |
fbf5a39b AC |
1040 | |
1041 | else | |
1042 | Token := Tok_Integer_Literal; | |
1043 | ||
1044 | if UI_Scale = 0 then | |
1045 | Int_Literal_Value := UI_Num_Value; | |
1046 | ||
82c80734 RD |
1047 | -- Avoid doing possibly expensive calculations in cases like |
1048 | -- parsing 163E800_000# when semantics will not be done anyway. | |
1049 | -- This is especially useful when parsing garbled input. | |
fbf5a39b AC |
1050 | |
1051 | elsif Operating_Mode /= Check_Syntax | |
1052 | and then (Serious_Errors_Detected = 0 or else Try_Semantics) | |
1053 | then | |
1054 | Int_Literal_Value := UI_Num_Value * UI_Base ** UI_Scale; | |
1055 | ||
1056 | else | |
1057 | Int_Literal_Value := No_Uint; | |
fbf5a39b | 1058 | end if; |
fbf5a39b AC |
1059 | end if; |
1060 | ||
b251750b VC |
1061 | if Checksum_Accumulate_Token_Checksum then |
1062 | Accumulate_Token_Checksum; | |
1063 | end if; | |
1064 | ||
fbf5a39b | 1065 | return; |
fbf5a39b AC |
1066 | end Nlit; |
1067 | ||
1068 | ---------- | |
1069 | -- Slit -- | |
1070 | ---------- | |
1071 | ||
1072 | procedure Slit is | |
1073 | ||
1074 | Delimiter : Character; | |
1075 | -- Delimiter (first character of string) | |
1076 | ||
1077 | C : Character; | |
1078 | -- Current source program character | |
1079 | ||
1080 | Code : Char_Code; | |
1081 | -- Current character code value | |
1082 | ||
1083 | Err : Boolean; | |
1084 | -- Error flag for Scan_Wide call | |
1085 | ||
a2773bd3 AC |
1086 | String_Start : Source_Ptr; |
1087 | -- Point to first character of string | |
1088 | ||
fbf5a39b AC |
1089 | procedure Error_Bad_String_Char; |
1090 | -- Signal bad character in string/character literal. On entry | |
82c80734 RD |
1091 | -- Scan_Ptr points to the improper character encountered during the |
1092 | -- scan. Scan_Ptr is not modified, so it still points to the bad | |
fbf5a39b AC |
1093 | -- character on return. |
1094 | ||
1095 | procedure Error_Unterminated_String; | |
1096 | -- Procedure called if a line terminator character is encountered | |
1097 | -- during scanning a string, meaning that the string is not properly | |
1098 | -- terminated. | |
1099 | ||
1100 | procedure Set_String; | |
1101 | -- Procedure used to distinguish between string and operator symbol. | |
fd22cd21 RD |
1102 | -- On entry the string has been scanned out, and its characters start |
1103 | -- at Token_Ptr and end one character before Scan_Ptr. On exit Token | |
1104 | -- is set to Tok_String_Literal/Tok_Operator_Symbol as appropriate, | |
1105 | -- and Token_Node is appropriately initialized. In addition, in the | |
1106 | -- operator symbol case, Token_Name is appropriately set, and the | |
1107 | -- flags [Wide_]Wide_Character_Found are set appropriately. | |
fbf5a39b AC |
1108 | |
1109 | --------------------------- | |
1110 | -- Error_Bad_String_Char -- | |
1111 | --------------------------- | |
1112 | ||
1113 | procedure Error_Bad_String_Char is | |
1114 | C : constant Character := Source (Scan_Ptr); | |
1115 | ||
1116 | begin | |
1117 | if C = HT then | |
1118 | Error_Msg_S ("horizontal tab not allowed in string"); | |
1119 | ||
1120 | elsif C = VT or else C = FF then | |
1121 | Error_Msg_S ("format effector not allowed in string"); | |
1122 | ||
1123 | elsif C in Upper_Half_Character then | |
1124 | Error_Msg_S ("(Ada 83) upper half character not allowed"); | |
1125 | ||
1126 | else | |
1127 | Error_Msg_S ("control character not allowed in string"); | |
1128 | end if; | |
1129 | end Error_Bad_String_Char; | |
1130 | ||
1131 | ------------------------------- | |
1132 | -- Error_Unterminated_String -- | |
1133 | ------------------------------- | |
1134 | ||
1135 | procedure Error_Unterminated_String is | |
a2773bd3 AC |
1136 | S : Source_Ptr; |
1137 | ||
fbf5a39b AC |
1138 | begin |
1139 | -- An interesting little refinement. Consider the following | |
1140 | -- examples: | |
1141 | ||
1142 | -- A := "this is an unterminated string; | |
1143 | -- A := "this is an unterminated string & | |
1144 | -- P(A, "this is a parameter that didn't get terminated); | |
a2773bd3 | 1145 | -- P("this is a parameter that didn't get terminated, A); |
fbf5a39b AC |
1146 | |
1147 | -- We fiddle a little to do slightly better placement in these | |
1148 | -- cases also if there is white space at the end of the line we | |
1149 | -- place the flag at the start of this white space, not at the | |
1150 | -- end. Note that we only have to test for blanks, since tabs | |
1151 | -- aren't allowed in strings in the first place and would have | |
1152 | -- caused an error message. | |
1153 | ||
1154 | -- Two more cases that we treat specially are: | |
1155 | ||
1156 | -- A := "this string uses the wrong terminator' | |
1157 | -- A := "this string uses the wrong terminator' & | |
1158 | ||
1159 | -- In these cases we give a different error message as well | |
1160 | ||
1161 | -- We actually reposition the scan pointer to the point where we | |
1162 | -- place the flag in these cases, since it seems a better bet on | |
1163 | -- the original intention. | |
1164 | ||
1165 | while Source (Scan_Ptr - 1) = ' ' | |
1166 | or else Source (Scan_Ptr - 1) = '&' | |
1167 | loop | |
1168 | Scan_Ptr := Scan_Ptr - 1; | |
1169 | Unstore_String_Char; | |
1170 | end loop; | |
1171 | ||
1172 | -- Check for case of incorrect string terminator, but single quote | |
1173 | -- is not considered incorrect if the opening terminator misused | |
1174 | -- a single quote (error message already given). | |
1175 | ||
1176 | if Delimiter /= ''' | |
1177 | and then Source (Scan_Ptr - 1) = ''' | |
1178 | then | |
1179 | Unstore_String_Char; | |
1180 | Error_Msg | |
1181 | ("incorrect string terminator character", Scan_Ptr - 1); | |
1182 | return; | |
1183 | end if; | |
1184 | ||
a2773bd3 AC |
1185 | -- Backup over semicolon or right-paren/semicolon sequence |
1186 | ||
fbf5a39b AC |
1187 | if Source (Scan_Ptr - 1) = ';' then |
1188 | Scan_Ptr := Scan_Ptr - 1; | |
1189 | Unstore_String_Char; | |
1190 | ||
1191 | if Source (Scan_Ptr - 1) = ')' then | |
1192 | Scan_Ptr := Scan_Ptr - 1; | |
1193 | Unstore_String_Char; | |
1194 | end if; | |
1195 | end if; | |
1196 | ||
a2773bd3 AC |
1197 | -- See if there is a comma in the string, if so, guess that |
1198 | -- the first comma terminates the string. | |
1199 | ||
1200 | S := String_Start; | |
1201 | while S < Scan_Ptr loop | |
1202 | if Source (S) = ',' then | |
1203 | while Scan_Ptr > S loop | |
1204 | Scan_Ptr := Scan_Ptr - 1; | |
1205 | Unstore_String_Char; | |
1206 | end loop; | |
1207 | ||
1208 | exit; | |
1209 | end if; | |
1210 | ||
1211 | S := S + 1; | |
1212 | end loop; | |
1213 | ||
1214 | -- Now we have adjusted the scan pointer, give message | |
1215 | ||
ed2233dc | 1216 | Error_Msg_S -- CODEFIX |
2995ebee | 1217 | ("missing string quote"); |
fbf5a39b AC |
1218 | end Error_Unterminated_String; |
1219 | ||
1220 | ---------------- | |
1221 | -- Set_String -- | |
1222 | ---------------- | |
1223 | ||
1224 | procedure Set_String is | |
1225 | Slen : constant Int := Int (Scan_Ptr - Token_Ptr - 2); | |
1226 | C1 : Character; | |
1227 | C2 : Character; | |
1228 | C3 : Character; | |
1229 | ||
1230 | begin | |
1231 | -- Token_Name is currently set to Error_Name. The following | |
1232 | -- section of code resets Token_Name to the proper Name_Op_xx | |
1233 | -- value if the string is a valid operator symbol, otherwise it is | |
1234 | -- left set to Error_Name. | |
1235 | ||
1236 | if Slen = 1 then | |
1237 | C1 := Source (Token_Ptr + 1); | |
1238 | ||
1239 | case C1 is | |
1240 | when '=' => | |
1241 | Token_Name := Name_Op_Eq; | |
1242 | ||
1243 | when '>' => | |
1244 | Token_Name := Name_Op_Gt; | |
1245 | ||
1246 | when '<' => | |
1247 | Token_Name := Name_Op_Lt; | |
1248 | ||
1249 | when '+' => | |
1250 | Token_Name := Name_Op_Add; | |
1251 | ||
1252 | when '-' => | |
1253 | Token_Name := Name_Op_Subtract; | |
1254 | ||
1255 | when '&' => | |
1256 | Token_Name := Name_Op_Concat; | |
1257 | ||
1258 | when '*' => | |
1259 | Token_Name := Name_Op_Multiply; | |
1260 | ||
1261 | when '/' => | |
1262 | Token_Name := Name_Op_Divide; | |
1263 | ||
1264 | when others => | |
1265 | null; | |
1266 | end case; | |
1267 | ||
1268 | elsif Slen = 2 then | |
1269 | C1 := Source (Token_Ptr + 1); | |
1270 | C2 := Source (Token_Ptr + 2); | |
1271 | ||
1272 | if C1 = '*' and then C2 = '*' then | |
1273 | Token_Name := Name_Op_Expon; | |
1274 | ||
1275 | elsif C2 = '=' then | |
1276 | ||
1277 | if C1 = '/' then | |
1278 | Token_Name := Name_Op_Ne; | |
1279 | elsif C1 = '<' then | |
1280 | Token_Name := Name_Op_Le; | |
1281 | elsif C1 = '>' then | |
1282 | Token_Name := Name_Op_Ge; | |
1283 | end if; | |
1284 | ||
1285 | elsif (C1 = 'O' or else C1 = 'o') and then -- OR | |
1286 | (C2 = 'R' or else C2 = 'r') | |
1287 | then | |
1288 | Token_Name := Name_Op_Or; | |
1289 | end if; | |
1290 | ||
1291 | elsif Slen = 3 then | |
1292 | C1 := Source (Token_Ptr + 1); | |
1293 | C2 := Source (Token_Ptr + 2); | |
1294 | C3 := Source (Token_Ptr + 3); | |
1295 | ||
1296 | if (C1 = 'A' or else C1 = 'a') and then -- AND | |
1297 | (C2 = 'N' or else C2 = 'n') and then | |
1298 | (C3 = 'D' or else C3 = 'd') | |
1299 | then | |
1300 | Token_Name := Name_Op_And; | |
1301 | ||
1302 | elsif (C1 = 'A' or else C1 = 'a') and then -- ABS | |
1303 | (C2 = 'B' or else C2 = 'b') and then | |
1304 | (C3 = 'S' or else C3 = 's') | |
1305 | then | |
1306 | Token_Name := Name_Op_Abs; | |
1307 | ||
1308 | elsif (C1 = 'M' or else C1 = 'm') and then -- MOD | |
1309 | (C2 = 'O' or else C2 = 'o') and then | |
1310 | (C3 = 'D' or else C3 = 'd') | |
1311 | then | |
1312 | Token_Name := Name_Op_Mod; | |
1313 | ||
1314 | elsif (C1 = 'N' or else C1 = 'n') and then -- NOT | |
1315 | (C2 = 'O' or else C2 = 'o') and then | |
1316 | (C3 = 'T' or else C3 = 't') | |
1317 | then | |
1318 | Token_Name := Name_Op_Not; | |
1319 | ||
1320 | elsif (C1 = 'R' or else C1 = 'r') and then -- REM | |
1321 | (C2 = 'E' or else C2 = 'e') and then | |
1322 | (C3 = 'M' or else C3 = 'm') | |
1323 | then | |
1324 | Token_Name := Name_Op_Rem; | |
1325 | ||
1326 | elsif (C1 = 'X' or else C1 = 'x') and then -- XOR | |
1327 | (C2 = 'O' or else C2 = 'o') and then | |
1328 | (C3 = 'R' or else C3 = 'r') | |
1329 | then | |
1330 | Token_Name := Name_Op_Xor; | |
1331 | end if; | |
1332 | ||
1333 | end if; | |
1334 | ||
82c80734 RD |
1335 | -- If it is an operator symbol, then Token_Name is set. If it is |
1336 | -- some other string value, then Token_Name still contains | |
1337 | -- Error_Name. | |
fbf5a39b AC |
1338 | |
1339 | if Token_Name = Error_Name then | |
1340 | Token := Tok_String_Literal; | |
1341 | ||
1342 | else | |
1343 | Token := Tok_Operator_Symbol; | |
1344 | end if; | |
fbf5a39b AC |
1345 | end Set_String; |
1346 | ||
82c80734 | 1347 | -- Start of processing for Slit |
fbf5a39b AC |
1348 | |
1349 | begin | |
1350 | -- On entry, Scan_Ptr points to the opening character of the string | |
82c80734 RD |
1351 | -- which is either a percent, double quote, or apostrophe (single |
1352 | -- quote). The latter case is an error detected by the character | |
1353 | -- literal circuit. | |
fbf5a39b | 1354 | |
a2773bd3 AC |
1355 | String_Start := Scan_Ptr; |
1356 | ||
fbf5a39b AC |
1357 | Delimiter := Source (Scan_Ptr); |
1358 | Accumulate_Checksum (Delimiter); | |
fd22cd21 | 1359 | |
fbf5a39b | 1360 | Start_String; |
fd22cd21 RD |
1361 | Wide_Character_Found := False; |
1362 | Wide_Wide_Character_Found := False; | |
fbf5a39b AC |
1363 | Scan_Ptr := Scan_Ptr + 1; |
1364 | ||
1365 | -- Loop to scan out characters of string literal | |
1366 | ||
1367 | loop | |
1368 | C := Source (Scan_Ptr); | |
1369 | ||
1370 | if C = Delimiter then | |
1371 | Accumulate_Checksum (C); | |
1372 | Scan_Ptr := Scan_Ptr + 1; | |
1373 | exit when Source (Scan_Ptr) /= Delimiter; | |
1374 | Code := Get_Char_Code (C); | |
1375 | Accumulate_Checksum (C); | |
1376 | Scan_Ptr := Scan_Ptr + 1; | |
1377 | ||
1378 | else | |
1379 | if C = '"' and then Delimiter = '%' then | |
1380 | Error_Msg_S | |
1381 | ("quote not allowed in percent delimited string"); | |
1382 | Code := Get_Char_Code (C); | |
1383 | Scan_Ptr := Scan_Ptr + 1; | |
1384 | ||
2385e007 | 1385 | elsif Start_Of_Wide_Character then |
82c80734 | 1386 | Wptr := Scan_Ptr; |
fbf5a39b | 1387 | Scan_Wide (Source, Scan_Ptr, Code, Err); |
fbf5a39b AC |
1388 | |
1389 | if Err then | |
1390 | Error_Illegal_Wide_Character; | |
1391 | Code := Get_Char_Code (' '); | |
1392 | end if; | |
1393 | ||
82c80734 RD |
1394 | Accumulate_Checksum (Code); |
1395 | ||
357ac4df RD |
1396 | -- In Ada 95 mode we allow any wide characters in a string |
1397 | -- but in Ada 2005, the set of characters allowed has been | |
1398 | -- restricted to graphic characters. | |
1399 | ||
0791fbe9 | 1400 | if Ada_Version >= Ada_2005 |
c8427bff | 1401 | and then Is_UTF_32_Non_Graphic (UTF_32 (Code)) |
82c80734 RD |
1402 | then |
1403 | Error_Msg | |
1404 | ("(Ada 2005) non-graphic character not permitted " & | |
1405 | "in string literal", Wptr); | |
1406 | end if; | |
1407 | ||
fbf5a39b AC |
1408 | else |
1409 | Accumulate_Checksum (C); | |
1410 | ||
1411 | if C not in Graphic_Character then | |
1412 | if C in Line_Terminator then | |
1413 | Error_Unterminated_String; | |
1414 | exit; | |
1415 | ||
1416 | elsif C in Upper_Half_Character then | |
0ab80019 | 1417 | if Ada_Version = Ada_83 then |
fbf5a39b AC |
1418 | Error_Bad_String_Char; |
1419 | end if; | |
1420 | ||
1421 | else | |
1422 | Error_Bad_String_Char; | |
1423 | end if; | |
1424 | end if; | |
1425 | ||
1426 | Code := Get_Char_Code (C); | |
1427 | Scan_Ptr := Scan_Ptr + 1; | |
1428 | end if; | |
1429 | end if; | |
1430 | ||
1431 | Store_String_Char (Code); | |
1432 | ||
1433 | if not In_Character_Range (Code) then | |
fd22cd21 RD |
1434 | if In_Wide_Character_Range (Code) then |
1435 | Wide_Character_Found := True; | |
1436 | else | |
1437 | Wide_Wide_Character_Found := True; | |
1438 | end if; | |
fbf5a39b AC |
1439 | end if; |
1440 | end loop; | |
1441 | ||
1442 | String_Literal_Id := End_String; | |
1443 | Set_String; | |
1444 | return; | |
fbf5a39b AC |
1445 | end Slit; |
1446 | ||
2385e007 AC |
1447 | ---------------------------------- |
1448 | -- Skip_Other_Format_Characters -- | |
1449 | ---------------------------------- | |
1450 | ||
1451 | procedure Skip_Other_Format_Characters is | |
1452 | P : Source_Ptr; | |
1453 | Code : Char_Code; | |
1454 | Err : Boolean; | |
1455 | ||
1456 | begin | |
1457 | while Start_Of_Wide_Character loop | |
1458 | P := Scan_Ptr; | |
1459 | Scan_Wide (Source, Scan_Ptr, Code, Err); | |
1460 | ||
1461 | if not Is_UTF_32_Other (UTF_32 (Code)) then | |
1462 | Scan_Ptr := P; | |
1463 | return; | |
1464 | end if; | |
1465 | end loop; | |
1466 | end Skip_Other_Format_Characters; | |
1467 | ||
1468 | ----------------------------- | |
1469 | -- Start_Of_Wide_Character -- | |
1470 | ----------------------------- | |
1471 | ||
1472 | function Start_Of_Wide_Character return Boolean is | |
1473 | C : constant Character := Source (Scan_Ptr); | |
1474 | ||
1475 | begin | |
1476 | -- ESC encoding method with ESC present | |
1477 | ||
1478 | if C = ESC | |
1479 | and then Wide_Character_Encoding_Method in WC_ESC_Encoding_Method | |
1480 | then | |
1481 | return True; | |
1482 | ||
1483 | -- Upper half character with upper half encoding | |
1484 | ||
1485 | elsif C in Upper_Half_Character and then Upper_Half_Encoding then | |
1486 | return True; | |
1487 | ||
1488 | -- Brackets encoding | |
1489 | ||
1490 | elsif C = '[' | |
1491 | and then Source (Scan_Ptr + 1) = '"' | |
1492 | and then Identifier_Char (Source (Scan_Ptr + 2)) | |
1493 | then | |
1494 | return True; | |
1495 | ||
1496 | -- Not the start of a wide character | |
1497 | ||
1498 | else | |
1499 | return False; | |
1500 | end if; | |
1501 | end Start_Of_Wide_Character; | |
1502 | ||
82c80734 | 1503 | -- Start of processing for Scan |
fbf5a39b AC |
1504 | |
1505 | begin | |
1506 | Prev_Token := Token; | |
1507 | Prev_Token_Ptr := Token_Ptr; | |
1508 | Token_Name := Error_Name; | |
1509 | ||
1510 | -- The following loop runs more than once only if a format effector | |
1511 | -- (tab, vertical tab, form feed, line feed, carriage return) is | |
1512 | -- encountered and skipped, or some error situation, such as an | |
1513 | -- illegal character, is encountered. | |
1514 | ||
82c80734 RD |
1515 | <<Scan_Next_Character>> |
1516 | ||
fbf5a39b AC |
1517 | loop |
1518 | -- Skip past blanks, loop is opened up for speed | |
1519 | ||
1520 | while Source (Scan_Ptr) = ' ' loop | |
fbf5a39b AC |
1521 | if Source (Scan_Ptr + 1) /= ' ' then |
1522 | Scan_Ptr := Scan_Ptr + 1; | |
1523 | exit; | |
1524 | end if; | |
1525 | ||
1526 | if Source (Scan_Ptr + 2) /= ' ' then | |
1527 | Scan_Ptr := Scan_Ptr + 2; | |
1528 | exit; | |
1529 | end if; | |
1530 | ||
1531 | if Source (Scan_Ptr + 3) /= ' ' then | |
1532 | Scan_Ptr := Scan_Ptr + 3; | |
1533 | exit; | |
1534 | end if; | |
1535 | ||
1536 | if Source (Scan_Ptr + 4) /= ' ' then | |
1537 | Scan_Ptr := Scan_Ptr + 4; | |
1538 | exit; | |
1539 | end if; | |
1540 | ||
1541 | if Source (Scan_Ptr + 5) /= ' ' then | |
1542 | Scan_Ptr := Scan_Ptr + 5; | |
1543 | exit; | |
1544 | end if; | |
1545 | ||
1546 | if Source (Scan_Ptr + 6) /= ' ' then | |
1547 | Scan_Ptr := Scan_Ptr + 6; | |
1548 | exit; | |
1549 | end if; | |
1550 | ||
1551 | if Source (Scan_Ptr + 7) /= ' ' then | |
1552 | Scan_Ptr := Scan_Ptr + 7; | |
1553 | exit; | |
1554 | end if; | |
1555 | ||
1556 | Scan_Ptr := Scan_Ptr + 8; | |
1557 | end loop; | |
1558 | ||
1559 | -- We are now at a non-blank character, which is the first character | |
1560 | -- of the token we will scan, and hence the value of Token_Ptr. | |
1561 | ||
1562 | Token_Ptr := Scan_Ptr; | |
1563 | ||
82c80734 RD |
1564 | -- Here begins the main case statement which transfers control on the |
1565 | -- basis of the non-blank character we have encountered. | |
fbf5a39b AC |
1566 | |
1567 | case Source (Scan_Ptr) is | |
1568 | ||
1569 | -- Line terminator characters | |
1570 | ||
82c80734 RD |
1571 | when CR | LF | FF | VT => |
1572 | goto Scan_Line_Terminator; | |
fbf5a39b AC |
1573 | |
1574 | -- Horizontal tab, just skip past it | |
1575 | ||
1576 | when HT => | |
b11e8d6f RD |
1577 | if Style_Check then |
1578 | Style.Check_HT; | |
1579 | end if; | |
1580 | ||
fbf5a39b AC |
1581 | Scan_Ptr := Scan_Ptr + 1; |
1582 | ||
82c80734 RD |
1583 | -- End of file character, treated as an end of file only if it is |
1584 | -- the last character in the buffer, otherwise it is ignored. | |
fbf5a39b AC |
1585 | |
1586 | when EOF => | |
1587 | if Scan_Ptr = Source_Last (Current_Source_File) then | |
1588 | Check_End_Of_Line; | |
b11e8d6f RD |
1589 | |
1590 | if Style_Check then | |
1591 | Style.Check_EOF; | |
1592 | end if; | |
1593 | ||
fbf5a39b AC |
1594 | Token := Tok_EOF; |
1595 | return; | |
fbf5a39b AC |
1596 | else |
1597 | Scan_Ptr := Scan_Ptr + 1; | |
1598 | end if; | |
1599 | ||
1600 | -- Ampersand | |
1601 | ||
1602 | when '&' => | |
1603 | Accumulate_Checksum ('&'); | |
1604 | ||
1605 | if Source (Scan_Ptr + 1) = '&' then | |
ed2233dc AC |
1606 | Error_Msg_S -- CODEFIX |
1607 | ("'&'& should be `AND THEN`"); | |
fbf5a39b AC |
1608 | Scan_Ptr := Scan_Ptr + 2; |
1609 | Token := Tok_And; | |
1610 | return; | |
1611 | ||
1612 | else | |
1613 | Scan_Ptr := Scan_Ptr + 1; | |
1614 | Token := Tok_Ampersand; | |
1615 | return; | |
1616 | end if; | |
1617 | ||
ae33543c | 1618 | when '@' => |
0f83b044 | 1619 | if Ada_Version < Ada_2020 then |
4b7fd131 | 1620 | Error_Msg ("target_name is an Ada 2020 feature", Scan_Ptr); |
ae33543c ES |
1621 | Scan_Ptr := Scan_Ptr + 1; |
1622 | ||
1623 | else | |
1624 | -- AI12-0125-03 : @ is target_name | |
d43584ca | 1625 | |
ae33543c ES |
1626 | Accumulate_Checksum ('@'); |
1627 | Scan_Ptr := Scan_Ptr + 1; | |
1628 | Token := Tok_At_Sign; | |
1629 | return; | |
1630 | end if; | |
1631 | ||
82c80734 RD |
1632 | -- Asterisk (can be multiplication operator or double asterisk which |
1633 | -- is the exponentiation compound delimiter). | |
fbf5a39b AC |
1634 | |
1635 | when '*' => | |
1636 | Accumulate_Checksum ('*'); | |
1637 | ||
1638 | if Source (Scan_Ptr + 1) = '*' then | |
1639 | Accumulate_Checksum ('*'); | |
1640 | Scan_Ptr := Scan_Ptr + 2; | |
1641 | Token := Tok_Double_Asterisk; | |
1642 | return; | |
1643 | ||
1644 | else | |
1645 | Scan_Ptr := Scan_Ptr + 1; | |
1646 | Token := Tok_Asterisk; | |
1647 | return; | |
1648 | end if; | |
1649 | ||
1650 | -- Colon, which can either be an isolated colon, or part of an | |
1651 | -- assignment compound delimiter. | |
1652 | ||
1653 | when ':' => | |
1654 | Accumulate_Checksum (':'); | |
1655 | ||
1656 | if Double_Char_Token ('=') then | |
1657 | Token := Tok_Colon_Equal; | |
b11e8d6f RD |
1658 | |
1659 | if Style_Check then | |
1660 | Style.Check_Colon_Equal; | |
1661 | end if; | |
1662 | ||
fbf5a39b AC |
1663 | return; |
1664 | ||
1665 | elsif Source (Scan_Ptr + 1) = '-' | |
1666 | and then Source (Scan_Ptr + 2) /= '-' | |
1667 | then | |
1668 | Token := Tok_Colon_Equal; | |
ed2233dc AC |
1669 | Error_Msg -- CODEFIX |
1670 | (":- should be :=", Scan_Ptr); | |
fbf5a39b AC |
1671 | Scan_Ptr := Scan_Ptr + 2; |
1672 | return; | |
1673 | ||
1674 | else | |
1675 | Scan_Ptr := Scan_Ptr + 1; | |
1676 | Token := Tok_Colon; | |
b11e8d6f RD |
1677 | |
1678 | if Style_Check then | |
1679 | Style.Check_Colon; | |
1680 | end if; | |
1681 | ||
fbf5a39b AC |
1682 | return; |
1683 | end if; | |
1684 | ||
1685 | -- Left parenthesis | |
1686 | ||
1687 | when '(' => | |
1688 | Accumulate_Checksum ('('); | |
1689 | Scan_Ptr := Scan_Ptr + 1; | |
1690 | Token := Tok_Left_Paren; | |
b11e8d6f RD |
1691 | |
1692 | if Style_Check then | |
1693 | Style.Check_Left_Paren; | |
1694 | end if; | |
1695 | ||
fbf5a39b AC |
1696 | return; |
1697 | ||
1698 | -- Left bracket | |
1699 | ||
1700 | when '[' => | |
1701 | if Source (Scan_Ptr + 1) = '"' then | |
82c80734 | 1702 | goto Scan_Wide_Character; |
fbf5a39b | 1703 | |
3c08de34 ES |
1704 | elsif Ada_Version = Ada_2020 then |
1705 | Scan_Ptr := Scan_Ptr + 1; | |
1706 | Token := Tok_Left_Bracket; | |
1707 | return; | |
1708 | ||
fbf5a39b AC |
1709 | else |
1710 | Error_Msg_S ("illegal character, replaced by ""("""); | |
1711 | Scan_Ptr := Scan_Ptr + 1; | |
1712 | Token := Tok_Left_Paren; | |
1713 | return; | |
1714 | end if; | |
1715 | ||
1716 | -- Left brace | |
1717 | ||
1718 | when '{' => | |
1719 | Error_Msg_S ("illegal character, replaced by ""("""); | |
1720 | Scan_Ptr := Scan_Ptr + 1; | |
1721 | Token := Tok_Left_Paren; | |
1722 | return; | |
1723 | ||
1724 | -- Comma | |
1725 | ||
1726 | when ',' => | |
1727 | Accumulate_Checksum (','); | |
1728 | Scan_Ptr := Scan_Ptr + 1; | |
1729 | Token := Tok_Comma; | |
b11e8d6f RD |
1730 | |
1731 | if Style_Check then | |
1732 | Style.Check_Comma; | |
1733 | end if; | |
1734 | ||
fbf5a39b AC |
1735 | return; |
1736 | ||
82c80734 RD |
1737 | -- Dot, which is either an isolated period, or part of a double dot |
1738 | -- compound delimiter sequence. We also check for the case of a | |
1739 | -- digit following the period, to give a better error message. | |
fbf5a39b AC |
1740 | |
1741 | when '.' => | |
1742 | Accumulate_Checksum ('.'); | |
1743 | ||
1744 | if Double_Char_Token ('.') then | |
1745 | Token := Tok_Dot_Dot; | |
b11e8d6f RD |
1746 | |
1747 | if Style_Check then | |
1748 | Style.Check_Dot_Dot; | |
1749 | end if; | |
1750 | ||
fbf5a39b AC |
1751 | return; |
1752 | ||
1753 | elsif Source (Scan_Ptr + 1) in '0' .. '9' then | |
1754 | Error_Msg_S ("numeric literal cannot start with point"); | |
1755 | Scan_Ptr := Scan_Ptr + 1; | |
1756 | ||
1757 | else | |
1758 | Scan_Ptr := Scan_Ptr + 1; | |
1759 | Token := Tok_Dot; | |
1760 | return; | |
1761 | end if; | |
1762 | ||
1763 | -- Equal, which can either be an equality operator, or part of the | |
1764 | -- arrow (=>) compound delimiter. | |
1765 | ||
1766 | when '=' => | |
1767 | Accumulate_Checksum ('='); | |
1768 | ||
1769 | if Double_Char_Token ('>') then | |
1770 | Token := Tok_Arrow; | |
b11e8d6f RD |
1771 | |
1772 | if Style_Check then | |
aa3efecd | 1773 | Style.Check_Arrow (Inside_Depends); |
b11e8d6f RD |
1774 | end if; |
1775 | ||
fbf5a39b AC |
1776 | return; |
1777 | ||
1778 | elsif Source (Scan_Ptr + 1) = '=' then | |
ed2233dc AC |
1779 | Error_Msg_S -- CODEFIX |
1780 | ("== should be ="); | |
fbf5a39b AC |
1781 | Scan_Ptr := Scan_Ptr + 1; |
1782 | end if; | |
1783 | ||
1784 | Scan_Ptr := Scan_Ptr + 1; | |
1785 | Token := Tok_Equal; | |
1786 | return; | |
1787 | ||
1788 | -- Greater than, which can be a greater than operator, greater than | |
1789 | -- or equal operator, or first character of a right label bracket. | |
1790 | ||
1791 | when '>' => | |
1792 | Accumulate_Checksum ('>'); | |
1793 | ||
1794 | if Double_Char_Token ('=') then | |
1795 | Token := Tok_Greater_Equal; | |
1796 | return; | |
1797 | ||
1798 | elsif Double_Char_Token ('>') then | |
1799 | Token := Tok_Greater_Greater; | |
1800 | return; | |
1801 | ||
1802 | else | |
1803 | Scan_Ptr := Scan_Ptr + 1; | |
1804 | Token := Tok_Greater; | |
1805 | return; | |
1806 | end if; | |
1807 | ||
1808 | -- Less than, which can be a less than operator, less than or equal | |
1809 | -- operator, or the first character of a left label bracket, or the | |
1810 | -- first character of a box (<>) compound delimiter. | |
1811 | ||
1812 | when '<' => | |
1813 | Accumulate_Checksum ('<'); | |
1814 | ||
1815 | if Double_Char_Token ('=') then | |
1816 | Token := Tok_Less_Equal; | |
1817 | return; | |
1818 | ||
1819 | elsif Double_Char_Token ('>') then | |
1820 | Token := Tok_Box; | |
b11e8d6f RD |
1821 | |
1822 | if Style_Check then | |
1823 | Style.Check_Box; | |
1824 | end if; | |
1825 | ||
fbf5a39b AC |
1826 | return; |
1827 | ||
1828 | elsif Double_Char_Token ('<') then | |
1829 | Token := Tok_Less_Less; | |
1830 | return; | |
1831 | ||
1832 | else | |
1833 | Scan_Ptr := Scan_Ptr + 1; | |
1834 | Token := Tok_Less; | |
1835 | return; | |
1836 | end if; | |
1837 | ||
1838 | -- Minus, which is either a subtraction operator, or the first | |
1839 | -- character of double minus starting a comment | |
1840 | ||
1841 | when '-' => Minus_Case : begin | |
1842 | if Source (Scan_Ptr + 1) = '>' then | |
1843 | Error_Msg_S ("invalid token"); | |
1844 | Scan_Ptr := Scan_Ptr + 2; | |
1845 | Token := Tok_Arrow; | |
1846 | return; | |
1847 | ||
1848 | elsif Source (Scan_Ptr + 1) /= '-' then | |
1849 | Accumulate_Checksum ('-'); | |
1850 | Scan_Ptr := Scan_Ptr + 1; | |
1851 | Token := Tok_Minus; | |
1852 | return; | |
1853 | ||
1854 | -- Comment | |
1855 | ||
1856 | else -- Source (Scan_Ptr + 1) = '-' then | |
b11e8d6f RD |
1857 | if Style_Check then |
1858 | Style.Check_Comment; | |
1859 | end if; | |
1860 | ||
fbf5a39b | 1861 | Scan_Ptr := Scan_Ptr + 2; |
383b2b42 RD |
1862 | |
1863 | -- If we are in preprocessor mode with Replace_In_Comments set, | |
1864 | -- then we return the "--" as a token on its own. | |
1865 | ||
1866 | if Replace_In_Comments then | |
1867 | Token := Tok_Comment; | |
1868 | return; | |
1869 | end if; | |
1870 | ||
1871 | -- Otherwise scan out the comment | |
1872 | ||
c45b6ae0 | 1873 | Start_Of_Comment := Scan_Ptr; |
fbf5a39b AC |
1874 | |
1875 | -- Loop to scan comment (this loop runs more than once only if | |
1876 | -- a horizontal tab or other non-graphic character is scanned) | |
1877 | ||
1878 | loop | |
1879 | -- Scan to non graphic character (opened up for speed) | |
1880 | ||
82c80734 RD |
1881 | -- Note that we just eat left brackets, which means that |
1882 | -- bracket notation cannot be used for end of line | |
1883 | -- characters in comments. This seems a reasonable choice, | |
1884 | -- since no one would ever use brackets notation in a real | |
1885 | -- program in this situation, and if we allow brackets | |
1886 | -- notation, we forbid some valid comments which contain a | |
1887 | -- brackets sequence that happens to match an end of line | |
1888 | -- character. | |
1889 | ||
fbf5a39b AC |
1890 | loop |
1891 | exit when Source (Scan_Ptr) not in Graphic_Character; | |
1892 | Scan_Ptr := Scan_Ptr + 1; | |
1893 | exit when Source (Scan_Ptr) not in Graphic_Character; | |
1894 | Scan_Ptr := Scan_Ptr + 1; | |
1895 | exit when Source (Scan_Ptr) not in Graphic_Character; | |
1896 | Scan_Ptr := Scan_Ptr + 1; | |
1897 | exit when Source (Scan_Ptr) not in Graphic_Character; | |
1898 | Scan_Ptr := Scan_Ptr + 1; | |
1899 | exit when Source (Scan_Ptr) not in Graphic_Character; | |
1900 | Scan_Ptr := Scan_Ptr + 1; | |
1901 | end loop; | |
1902 | ||
1903 | -- Keep going if horizontal tab | |
1904 | ||
1905 | if Source (Scan_Ptr) = HT then | |
b11e8d6f RD |
1906 | if Style_Check then |
1907 | Style.Check_HT; | |
1908 | end if; | |
1909 | ||
fbf5a39b AC |
1910 | Scan_Ptr := Scan_Ptr + 1; |
1911 | ||
1912 | -- Terminate scan of comment if line terminator | |
1913 | ||
1914 | elsif Source (Scan_Ptr) in Line_Terminator then | |
1915 | exit; | |
1916 | ||
1917 | -- Terminate scan of comment if end of file encountered | |
1918 | -- (embedded EOF character or real last character in file) | |
1919 | ||
1920 | elsif Source (Scan_Ptr) = EOF then | |
1921 | exit; | |
1922 | ||
82c80734 RD |
1923 | -- If we have a wide character, we have to scan it out, |
1924 | -- because it might be a legitimate line terminator | |
1925 | ||
2385e007 | 1926 | elsif Start_Of_Wide_Character then |
82c80734 RD |
1927 | declare |
1928 | Wptr : constant Source_Ptr := Scan_Ptr; | |
1929 | Code : Char_Code; | |
1930 | Err : Boolean; | |
1931 | ||
1932 | begin | |
1933 | Scan_Wide (Source, Scan_Ptr, Code, Err); | |
1934 | ||
1935 | -- If not well formed wide character, then just skip | |
1936 | -- past it and ignore it. | |
1937 | ||
1938 | if Err then | |
1939 | Scan_Ptr := Wptr + 1; | |
1940 | ||
1941 | -- If UTF_32 terminator, terminate comment scan | |
1942 | ||
c8427bff | 1943 | elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then |
82c80734 RD |
1944 | Scan_Ptr := Wptr; |
1945 | exit; | |
1946 | end if; | |
1947 | end; | |
1948 | ||
fbf5a39b AC |
1949 | -- Keep going if character in 80-FF range, or is ESC. These |
1950 | -- characters are allowed in comments by RM-2.1(1), 2.7(2). | |
1951 | -- They are allowed even in Ada 83 mode according to the | |
1952 | -- approved AI. ESC was added to the AI in June 93. | |
1953 | ||
1954 | elsif Source (Scan_Ptr) in Upper_Half_Character | |
82c80734 | 1955 | or else Source (Scan_Ptr) = ESC |
fbf5a39b AC |
1956 | then |
1957 | Scan_Ptr := Scan_Ptr + 1; | |
1958 | ||
cd1a470a AC |
1959 | -- Otherwise we have an illegal comment character, ignore |
1960 | -- this error in relaxed semantics mode. | |
fbf5a39b AC |
1961 | |
1962 | else | |
cd1a470a AC |
1963 | if Relaxed_RM_Semantics then |
1964 | Scan_Ptr := Scan_Ptr + 1; | |
1965 | else | |
1966 | Error_Illegal_Character; | |
1967 | end if; | |
fbf5a39b | 1968 | end if; |
fbf5a39b AC |
1969 | end loop; |
1970 | ||
c45b6ae0 AC |
1971 | -- Note that, except when comments are tokens, we do NOT |
1972 | -- execute a return here, instead we fall through to reexecute | |
1973 | -- the scan loop to look for a token. | |
1974 | ||
1975 | if Comment_Is_Token then | |
1976 | Name_Len := Integer (Scan_Ptr - Start_Of_Comment); | |
1977 | Name_Buffer (1 .. Name_Len) := | |
1978 | String (Source (Start_Of_Comment .. Scan_Ptr - 1)); | |
1979 | Comment_Id := Name_Find; | |
1980 | Token := Tok_Comment; | |
1981 | return; | |
1982 | end if; | |
4fbad0ba | 1983 | |
0580d807 AC |
1984 | -- If the SPARK restriction is set for this unit, then generate |
1985 | -- a token Tok_SPARK_Hide for a SPARK HIDE directive. | |
f9adb9d4 | 1986 | |
6480338a | 1987 | if Restriction_Check_Required (SPARK_05) |
f9adb9d4 AC |
1988 | and then Source (Start_Of_Comment) = '#' |
1989 | then | |
4fbad0ba AC |
1990 | declare |
1991 | Scan_SPARK_Ptr : Source_Ptr; | |
1992 | ||
1993 | begin | |
1994 | Scan_SPARK_Ptr := Start_Of_Comment + 1; | |
1995 | ||
1996 | -- Scan out blanks | |
1997 | ||
1998 | while Source (Scan_SPARK_Ptr) = ' ' | |
1999 | or else Source (Scan_SPARK_Ptr) = HT | |
2000 | loop | |
2001 | Scan_SPARK_Ptr := Scan_SPARK_Ptr + 1; | |
2002 | end loop; | |
2003 | ||
2004 | -- Recognize HIDE directive. SPARK input cannot be | |
2005 | -- encoded as wide characters, so only deal with | |
2006 | -- lower/upper case. | |
2007 | ||
2008 | if (Source (Scan_SPARK_Ptr) = 'h' | |
2009 | or else Source (Scan_SPARK_Ptr) = 'H') | |
2010 | and then (Source (Scan_SPARK_Ptr + 1) = 'i' | |
2011 | or else Source (Scan_SPARK_Ptr + 1) = 'I') | |
2012 | and then (Source (Scan_SPARK_Ptr + 2) = 'd' | |
2013 | or else Source (Scan_SPARK_Ptr + 2) = 'D') | |
2014 | and then (Source (Scan_SPARK_Ptr + 3) = 'e' | |
2015 | or else Source (Scan_SPARK_Ptr + 3) = 'E') | |
2016 | and then (Source (Scan_SPARK_Ptr + 4) = ' ' | |
2017 | or else Source (Scan_SPARK_Ptr + 4) = HT) | |
2018 | then | |
2019 | Token := Tok_SPARK_Hide; | |
2020 | return; | |
2021 | end if; | |
2022 | end; | |
2023 | end if; | |
fbf5a39b AC |
2024 | end if; |
2025 | end Minus_Case; | |
2026 | ||
c90b2058 | 2027 | -- Double quote or percent starting a string literal |
fbf5a39b | 2028 | |
c90b2058 | 2029 | when '"' | '%' => |
fbf5a39b AC |
2030 | Slit; |
2031 | Post_Scan; | |
2032 | return; | |
2033 | ||
2034 | -- Apostrophe. This can either be the start of a character literal, | |
2035 | -- or an isolated apostrophe used in a qualified expression or an | |
c79f6efd | 2036 | -- attribute. In the following: |
fbf5a39b AC |
2037 | |
2038 | -- A := CHARACTER'('A'); | |
2039 | ||
c79f6efd BD |
2040 | -- the first apostrophe is treated as an isolated apostrophe, and the |
2041 | -- second one is treated as the start of the character literal 'A'. | |
2042 | -- Note that RM-2.2(7) does not require a separator between "'" and | |
2043 | -- "(" in the above, so we cannot use lookahead to distinguish the | |
2044 | -- cases; we use look-back instead. Analysis of the grammar shows | |
2045 | -- that some tokens can be followed by an apostrophe, and some by a | |
2046 | -- character literal, but none by both. Some cannot be followed by | |
2047 | -- either, so it doesn't matter what we do in those cases, except to | |
2048 | -- get good error behavior. | |
fbf5a39b AC |
2049 | |
2050 | when ''' => Char_Literal_Case : declare | |
2051 | Code : Char_Code; | |
2052 | Err : Boolean; | |
2053 | ||
2054 | begin | |
2055 | Accumulate_Checksum ('''); | |
2056 | Scan_Ptr := Scan_Ptr + 1; | |
2057 | ||
c79f6efd BD |
2058 | -- Distinguish between apostrophe and character literal. It's an |
2059 | -- apostrophe if the previous token is one of the following. | |
2060 | -- Reserved words are included for things like A.all'Address and | |
2061 | -- T'Digits'Img. Strings literals are included for things like | |
2062 | -- "abs"'Address. Other literals are included to give better error | |
2063 | -- behavior for illegal cases like 123'Img. | |
28ad2460 HK |
2064 | -- |
2065 | -- In Ada 2020, a target name (i.e. @) is a valid prefix of an | |
b41c731f | 2066 | -- attribute, and functions like a name. |
fbf5a39b | 2067 | |
2bb988bb AC |
2068 | if Prev_Token = Tok_All |
2069 | or else Prev_Token = Tok_At_Sign | |
2070 | or else Prev_Token = Tok_Delta | |
2071 | or else Prev_Token = Tok_Digits | |
2072 | or else Prev_Token = Tok_Identifier | |
2073 | or else Prev_Token = Tok_Project | |
2074 | or else Prev_Token = Tok_Right_Paren | |
3c08de34 | 2075 | or else Prev_Token = Tok_Right_Bracket |
2bb988bb | 2076 | or else Prev_Token in Token_Class_Literal |
fbf5a39b AC |
2077 | then |
2078 | Token := Tok_Apostrophe; | |
b11e8d6f RD |
2079 | |
2080 | if Style_Check then | |
2081 | Style.Check_Apostrophe; | |
2082 | end if; | |
2083 | ||
fbf5a39b AC |
2084 | return; |
2085 | ||
2086 | -- Otherwise the apostrophe starts a character literal | |
2087 | ||
2088 | else | |
82c80734 | 2089 | -- Case of wide character literal |
fbf5a39b | 2090 | |
2385e007 | 2091 | if Start_Of_Wide_Character then |
82c80734 | 2092 | Wptr := Scan_Ptr; |
fbf5a39b AC |
2093 | Scan_Wide (Source, Scan_Ptr, Code, Err); |
2094 | Accumulate_Checksum (Code); | |
2095 | ||
2096 | if Err then | |
2097 | Error_Illegal_Wide_Character; | |
3fd68ccb | 2098 | Code := Character'Pos (' '); |
357ac4df RD |
2099 | |
2100 | -- In Ada 95 mode we allow any wide character in a character | |
2101 | -- literal, but in Ada 2005, the set of characters allowed | |
2102 | -- is restricted to graphic characters. | |
82c80734 | 2103 | |
0791fbe9 | 2104 | elsif Ada_Version >= Ada_2005 |
c8427bff | 2105 | and then Is_UTF_32_Non_Graphic (UTF_32 (Code)) |
82c80734 | 2106 | then |
ed2233dc | 2107 | Error_Msg -- CODEFIX???? |
82c80734 RD |
2108 | ("(Ada 2005) non-graphic character not permitted " & |
2109 | "in character literal", Wptr); | |
fbf5a39b AC |
2110 | end if; |
2111 | ||
2112 | if Source (Scan_Ptr) /= ''' then | |
ed2233dc | 2113 | Error_Msg_S ("missing apostrophe"); |
fbf5a39b AC |
2114 | else |
2115 | Scan_Ptr := Scan_Ptr + 1; | |
2116 | end if; | |
2117 | ||
2118 | -- If we do not find a closing quote in the expected place then | |
2119 | -- assume that we have a misguided attempt at a string literal. | |
2120 | ||
2121 | -- However, if previous token is RANGE, then we return an | |
2122 | -- apostrophe instead since this gives better error recovery | |
2123 | ||
2124 | elsif Source (Scan_Ptr + 1) /= ''' then | |
fbf5a39b AC |
2125 | if Prev_Token = Tok_Range then |
2126 | Token := Tok_Apostrophe; | |
2127 | return; | |
2128 | ||
2129 | else | |
2130 | Scan_Ptr := Scan_Ptr - 1; | |
2131 | Error_Msg_S | |
2132 | ("strings are delimited by double quote character"); | |
2133 | Slit; | |
2134 | Post_Scan; | |
2135 | return; | |
2136 | end if; | |
2137 | ||
2138 | -- Otherwise we have a (non-wide) character literal | |
2139 | ||
2140 | else | |
2141 | Accumulate_Checksum (Source (Scan_Ptr)); | |
2142 | ||
2143 | if Source (Scan_Ptr) not in Graphic_Character then | |
2144 | if Source (Scan_Ptr) in Upper_Half_Character then | |
0ab80019 | 2145 | if Ada_Version = Ada_83 then |
fbf5a39b AC |
2146 | Error_Illegal_Character; |
2147 | end if; | |
2148 | ||
2149 | else | |
2150 | Error_Illegal_Character; | |
2151 | end if; | |
2152 | end if; | |
2153 | ||
2154 | Code := Get_Char_Code (Source (Scan_Ptr)); | |
2155 | Scan_Ptr := Scan_Ptr + 2; | |
2156 | end if; | |
2157 | ||
2158 | -- Fall through here with Scan_Ptr updated past the closing | |
2159 | -- quote, and Code set to the Char_Code value for the literal | |
2160 | ||
2161 | Accumulate_Checksum ('''); | |
2162 | Token := Tok_Char_Literal; | |
2163 | Set_Character_Literal_Name (Code); | |
2164 | Token_Name := Name_Find; | |
2165 | Character_Code := Code; | |
2166 | Post_Scan; | |
2167 | return; | |
2168 | end if; | |
2169 | end Char_Literal_Case; | |
2170 | ||
2171 | -- Right parenthesis | |
2172 | ||
2173 | when ')' => | |
2174 | Accumulate_Checksum (')'); | |
2175 | Scan_Ptr := Scan_Ptr + 1; | |
2176 | Token := Tok_Right_Paren; | |
b11e8d6f RD |
2177 | |
2178 | if Style_Check then | |
2179 | Style.Check_Right_Paren; | |
2180 | end if; | |
2181 | ||
fbf5a39b AC |
2182 | return; |
2183 | ||
dab8e608 GD |
2184 | -- Right bracket or right brace, treated as right paren but proper |
2185 | -- aggregate delimiter in Ada 2020. | |
fbf5a39b AC |
2186 | |
2187 | when ']' | '}' => | |
3c08de34 ES |
2188 | if Ada_Version >= Ada_2020 then |
2189 | Token := Tok_Right_Bracket; | |
2190 | ||
2191 | else | |
2192 | Error_Msg_S ("illegal character, replaced by "")"""); | |
2193 | Token := Tok_Right_Paren; | |
2194 | end if; | |
2195 | ||
fbf5a39b | 2196 | Scan_Ptr := Scan_Ptr + 1; |
fbf5a39b AC |
2197 | return; |
2198 | ||
2199 | -- Slash (can be division operator or first character of not equal) | |
2200 | ||
2201 | when '/' => | |
2202 | Accumulate_Checksum ('/'); | |
2203 | ||
2204 | if Double_Char_Token ('=') then | |
2205 | Token := Tok_Not_Equal; | |
2206 | return; | |
2207 | else | |
2208 | Scan_Ptr := Scan_Ptr + 1; | |
2209 | Token := Tok_Slash; | |
2210 | return; | |
2211 | end if; | |
2212 | ||
2213 | -- Semicolon | |
2214 | ||
2215 | when ';' => | |
2216 | Accumulate_Checksum (';'); | |
2217 | Scan_Ptr := Scan_Ptr + 1; | |
2218 | Token := Tok_Semicolon; | |
b11e8d6f RD |
2219 | |
2220 | if Style_Check then | |
2221 | Style.Check_Semicolon; | |
2222 | end if; | |
2223 | ||
fbf5a39b AC |
2224 | return; |
2225 | ||
2226 | -- Vertical bar | |
2227 | ||
2228 | when '|' => Vertical_Bar_Case : begin | |
2229 | Accumulate_Checksum ('|'); | |
2230 | ||
2231 | -- Special check for || to give nice message | |
2232 | ||
2233 | if Source (Scan_Ptr + 1) = '|' then | |
ed2233dc AC |
2234 | Error_Msg_S -- CODEFIX |
2235 | ("""'|'|"" should be `OR ELSE`"); | |
fbf5a39b AC |
2236 | Scan_Ptr := Scan_Ptr + 2; |
2237 | Token := Tok_Or; | |
2238 | return; | |
2239 | ||
2240 | else | |
2241 | Scan_Ptr := Scan_Ptr + 1; | |
2242 | Token := Tok_Vertical_Bar; | |
b11e8d6f RD |
2243 | |
2244 | if Style_Check then | |
2245 | Style.Check_Vertical_Bar; | |
2246 | end if; | |
2247 | ||
c90b2058 | 2248 | Post_Scan; |
fbf5a39b AC |
2249 | return; |
2250 | end if; | |
2251 | end Vertical_Bar_Case; | |
2252 | ||
2253 | -- Exclamation, replacement character for vertical bar | |
2254 | ||
2255 | when '!' => Exclamation_Case : begin | |
2256 | Accumulate_Checksum ('!'); | |
fbf5a39b AC |
2257 | |
2258 | if Source (Scan_Ptr + 1) = '=' then | |
ed2233dc AC |
2259 | Error_Msg_S -- CODEFIX |
2260 | ("'!= should be /="); | |
fbf5a39b AC |
2261 | Scan_Ptr := Scan_Ptr + 2; |
2262 | Token := Tok_Not_Equal; | |
2263 | return; | |
2264 | ||
2265 | else | |
2266 | Scan_Ptr := Scan_Ptr + 1; | |
2267 | Token := Tok_Vertical_Bar; | |
c90b2058 | 2268 | Post_Scan; |
fbf5a39b AC |
2269 | return; |
2270 | end if; | |
fbf5a39b AC |
2271 | end Exclamation_Case; |
2272 | ||
2273 | -- Plus | |
2274 | ||
2275 | when '+' => Plus_Case : begin | |
2276 | Accumulate_Checksum ('+'); | |
2277 | Scan_Ptr := Scan_Ptr + 1; | |
2278 | Token := Tok_Plus; | |
2279 | return; | |
2280 | end Plus_Case; | |
2281 | ||
2282 | -- Digits starting a numeric literal | |
2283 | ||
2284 | when '0' .. '9' => | |
daca8389 RD |
2285 | |
2286 | -- First a bit of a scan ahead to see if we have a case of an | |
2287 | -- identifier starting with a digit (remembering exponent case). | |
2288 | ||
2289 | declare | |
2290 | C : constant Character := Source (Scan_Ptr + 1); | |
2291 | ||
2292 | begin | |
2293 | -- OK literal if digit followed by digit or underscore | |
2294 | ||
2295 | if C in '0' .. '9' or else C = '_' then | |
2296 | null; | |
2297 | ||
2298 | -- OK literal if digit not followed by identifier char | |
2299 | ||
2300 | elsif not Identifier_Char (C) then | |
2301 | null; | |
2302 | ||
2303 | -- OK literal if digit followed by e/E followed by digit/sign. | |
2304 | -- We also allow underscore after the E, which is an error, but | |
2305 | -- better handled by Nlit than deciding this is an identifier. | |
2306 | ||
2307 | elsif (C = 'e' or else C = 'E') | |
2308 | and then (Source (Scan_Ptr + 2) in '0' .. '9' | |
2309 | or else Source (Scan_Ptr + 2) = '+' | |
2310 | or else Source (Scan_Ptr + 2) = '-' | |
2311 | or else Source (Scan_Ptr + 2) = '_') | |
2312 | then | |
2313 | null; | |
2314 | ||
2315 | -- Here we have what really looks like an identifier that | |
2316 | -- starts with a digit, so give error msg. | |
2317 | ||
2318 | else | |
2319 | Error_Msg_S ("identifier may not start with digit"); | |
2320 | Name_Len := 1; | |
2321 | Underline_Found := False; | |
2322 | Name_Buffer (1) := Source (Scan_Ptr); | |
2323 | Accumulate_Checksum (Name_Buffer (1)); | |
2324 | Scan_Ptr := Scan_Ptr + 1; | |
2325 | goto Scan_Identifier; | |
2326 | end if; | |
2327 | end; | |
2328 | ||
2329 | -- Here we have an OK integer literal | |
2330 | ||
fbf5a39b AC |
2331 | Nlit; |
2332 | ||
2385e007 AC |
2333 | -- Check for proper delimiter, ignoring other format characters |
2334 | ||
2335 | Skip_Other_Format_Characters; | |
2336 | ||
fbf5a39b AC |
2337 | if Identifier_Char (Source (Scan_Ptr)) then |
2338 | Error_Msg_S | |
2339 | ("delimiter required between literal and identifier"); | |
2340 | end if; | |
daca8389 | 2341 | |
fbf5a39b AC |
2342 | Post_Scan; |
2343 | return; | |
2344 | ||
2345 | -- Lower case letters | |
2346 | ||
2347 | when 'a' .. 'z' => | |
2348 | Name_Len := 1; | |
82c80734 | 2349 | Underline_Found := False; |
fbf5a39b AC |
2350 | Name_Buffer (1) := Source (Scan_Ptr); |
2351 | Accumulate_Checksum (Name_Buffer (1)); | |
2352 | Scan_Ptr := Scan_Ptr + 1; | |
2353 | goto Scan_Identifier; | |
2354 | ||
2355 | -- Upper case letters | |
2356 | ||
2357 | when 'A' .. 'Z' => | |
2358 | Name_Len := 1; | |
82c80734 | 2359 | Underline_Found := False; |
fbf5a39b AC |
2360 | Name_Buffer (1) := |
2361 | Character'Val (Character'Pos (Source (Scan_Ptr)) + 32); | |
2362 | Accumulate_Checksum (Name_Buffer (1)); | |
2363 | Scan_Ptr := Scan_Ptr + 1; | |
2364 | goto Scan_Identifier; | |
2365 | ||
2366 | -- Underline character | |
2367 | ||
2368 | when '_' => | |
2369 | if Special_Characters ('_') then | |
2370 | Token_Ptr := Scan_Ptr; | |
2371 | Scan_Ptr := Scan_Ptr + 1; | |
2372 | Token := Tok_Special; | |
2373 | Special_Character := '_'; | |
2374 | return; | |
2375 | end if; | |
2376 | ||
2377 | Error_Msg_S ("identifier cannot start with underline"); | |
2378 | Name_Len := 1; | |
2379 | Name_Buffer (1) := '_'; | |
2380 | Scan_Ptr := Scan_Ptr + 1; | |
82c80734 | 2381 | Underline_Found := False; |
fbf5a39b AC |
2382 | goto Scan_Identifier; |
2383 | ||
2384 | -- Space (not possible, because we scanned past blanks) | |
2385 | ||
2386 | when ' ' => | |
2387 | raise Program_Error; | |
2388 | ||
2389 | -- Characters in top half of ASCII 8-bit chart | |
2390 | ||
2391 | when Upper_Half_Character => | |
2392 | ||
82c80734 | 2393 | -- Wide character case |
fbf5a39b AC |
2394 | |
2395 | if Upper_Half_Encoding then | |
82c80734 | 2396 | goto Scan_Wide_Character; |
fbf5a39b AC |
2397 | |
2398 | -- Otherwise we have OK Latin-1 character | |
2399 | ||
2400 | else | |
2401 | -- Upper half characters may possibly be identifier letters | |
82c80734 RD |
2402 | -- but can never be digits, so Identifier_Char can be used to |
2403 | -- test for a valid start of identifier character. | |
fbf5a39b AC |
2404 | |
2405 | if Identifier_Char (Source (Scan_Ptr)) then | |
2406 | Name_Len := 0; | |
82c80734 | 2407 | Underline_Found := False; |
fbf5a39b AC |
2408 | goto Scan_Identifier; |
2409 | else | |
2410 | Error_Illegal_Character; | |
2411 | end if; | |
2412 | end if; | |
2413 | ||
2414 | when ESC => | |
2415 | ||
2416 | -- ESC character, possible start of identifier if wide characters | |
2417 | -- using ESC encoding are allowed in identifiers, which we can | |
2418 | -- tell by looking at the Identifier_Char flag for ESC, which is | |
82c80734 RD |
2419 | -- only true if these conditions are met. In Ada 2005 mode, may |
2420 | -- also be valid UTF_32 space or line terminator character. | |
fbf5a39b AC |
2421 | |
2422 | if Identifier_Char (ESC) then | |
2423 | Name_Len := 0; | |
82c80734 | 2424 | goto Scan_Wide_Character; |
fbf5a39b | 2425 | else |
82c80734 | 2426 | Error_Illegal_Character; |
fbf5a39b AC |
2427 | end if; |
2428 | ||
2429 | -- Invalid control characters | |
2430 | ||
d8f43ee6 HK |
2431 | when ACK |
2432 | | ASCII.SO | |
2433 | | BEL | |
2434 | | BS | |
2435 | | CAN | |
2436 | | DC1 | |
2437 | | DC2 | |
2438 | | DC3 | |
2439 | | DC4 | |
2440 | | DEL | |
2441 | | DLE | |
2442 | | EM | |
2443 | | ENQ | |
2444 | | EOT | |
2445 | | ETB | |
2446 | | ETX | |
2447 | | FS | |
2448 | | GS | |
2449 | | NAK | |
2450 | | NUL | |
2451 | | RS | |
2452 | | SI | |
2453 | | SOH | |
2454 | | STX | |
2455 | | SYN | |
2456 | | US | |
fbf5a39b AC |
2457 | => |
2458 | Error_Illegal_Character; | |
2459 | ||
2460 | -- Invalid graphic characters | |
ae33543c ES |
2461 | -- Note that '@' is handled elsewhere, because following AI12-125 |
2462 | -- it denotes the target_name of an assignment. | |
d43584ca | 2463 | |
ae33543c | 2464 | when '#' | '$' | '?' | '`' | '\' | '^' | '~' => |
82c80734 | 2465 | |
fbf5a39b AC |
2466 | -- If Set_Special_Character has been called for this character, |
2467 | -- set Scans.Special_Character and return a Special token. | |
2468 | ||
2469 | if Special_Characters (Source (Scan_Ptr)) then | |
2470 | Token_Ptr := Scan_Ptr; | |
2471 | Token := Tok_Special; | |
2472 | Special_Character := Source (Scan_Ptr); | |
2473 | Scan_Ptr := Scan_Ptr + 1; | |
2474 | return; | |
2475 | ||
2a1f6a1f AC |
2476 | -- Check for something looking like a preprocessor directive |
2477 | ||
2478 | elsif Source (Scan_Ptr) = '#' | |
2479 | and then (Source (Scan_Ptr + 1 .. Scan_Ptr + 2) = "if" | |
2480 | or else | |
2481 | Source (Scan_Ptr + 1 .. Scan_Ptr + 5) = "elsif" | |
2482 | or else | |
2483 | Source (Scan_Ptr + 1 .. Scan_Ptr + 4) = "else" | |
2484 | or else | |
2485 | Source (Scan_Ptr + 1 .. Scan_Ptr + 3) = "end") | |
2486 | then | |
2487 | Error_Msg_S | |
2488 | ("preprocessor directive ignored, preprocessor not active"); | |
2489 | ||
2490 | -- Skip to end of line | |
2491 | ||
2492 | loop | |
2493 | if Source (Scan_Ptr) in Graphic_Character | |
2494 | or else | |
2495 | Source (Scan_Ptr) = HT | |
2496 | then | |
2497 | Scan_Ptr := Scan_Ptr + 1; | |
2498 | ||
2499 | -- Done if line terminator or EOF | |
2500 | ||
2501 | elsif Source (Scan_Ptr) in Line_Terminator | |
2502 | or else | |
2503 | Source (Scan_Ptr) = EOF | |
2504 | then | |
2505 | exit; | |
2506 | ||
2507 | -- If we have a wide character, we have to scan it out, | |
2508 | -- because it might be a legitimate line terminator | |
2509 | ||
2510 | elsif Start_Of_Wide_Character then | |
2511 | declare | |
2512 | Wptr : constant Source_Ptr := Scan_Ptr; | |
2513 | Code : Char_Code; | |
2514 | Err : Boolean; | |
2515 | ||
2516 | begin | |
2517 | Scan_Wide (Source, Scan_Ptr, Code, Err); | |
2518 | ||
2519 | -- If not well formed wide character, then just skip | |
2520 | -- past it and ignore it. | |
2521 | ||
2522 | if Err then | |
2523 | Scan_Ptr := Wptr + 1; | |
2524 | ||
2525 | -- If UTF_32 terminator, terminate comment scan | |
2526 | ||
2527 | elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then | |
2528 | Scan_Ptr := Wptr; | |
2529 | exit; | |
2530 | end if; | |
2531 | end; | |
2532 | ||
2533 | -- Else keep going (don't worry about bad comment chars | |
2534 | -- in this context, we just want to find the end of line. | |
2535 | ||
2536 | else | |
2537 | Scan_Ptr := Scan_Ptr + 1; | |
2538 | end if; | |
2539 | end loop; | |
2540 | ||
82c80734 | 2541 | -- Otherwise, this is an illegal character |
fbf5a39b AC |
2542 | |
2543 | else | |
2544 | Error_Illegal_Character; | |
2545 | end if; | |
2546 | ||
2547 | -- End switch on non-blank character | |
2548 | ||
2549 | end case; | |
2550 | ||
2551 | -- End loop past format effectors. The exit from this loop is by | |
2552 | -- executing a return statement following completion of token scan | |
8f34c90b | 2553 | -- (control never falls out of this loop to the code that follows). |
fbf5a39b AC |
2554 | |
2555 | end loop; | |
2556 | ||
8f34c90b AC |
2557 | pragma Assert (False); |
2558 | ||
82c80734 RD |
2559 | -- Wide_Character scanning routine. On entry we have encountered the |
2560 | -- initial character of a wide character sequence. | |
fbf5a39b | 2561 | |
82c80734 | 2562 | <<Scan_Wide_Character>> |
82c80734 RD |
2563 | declare |
2564 | Code : Char_Code; | |
c8427bff | 2565 | Cat : Category; |
82c80734 | 2566 | Err : Boolean; |
fbf5a39b | 2567 | |
82c80734 RD |
2568 | begin |
2569 | Wptr := Scan_Ptr; | |
2570 | Scan_Wide (Source, Scan_Ptr, Code, Err); | |
fbf5a39b | 2571 | |
82c80734 | 2572 | -- If bad wide character, signal error and continue scan |
fbf5a39b | 2573 | |
82c80734 RD |
2574 | if Err then |
2575 | Error_Illegal_Wide_Character; | |
2576 | goto Scan_Next_Character; | |
c8427bff RD |
2577 | end if; |
2578 | ||
2579 | Cat := Get_Category (UTF_32 (Code)); | |
fbf5a39b | 2580 | |
82c80734 | 2581 | -- If OK letter, reset scan ptr and go scan identifier |
fbf5a39b | 2582 | |
c8427bff | 2583 | if Is_UTF_32_Letter (Cat) then |
82c80734 RD |
2584 | Scan_Ptr := Wptr; |
2585 | Name_Len := 0; | |
2586 | Underline_Found := False; | |
2587 | goto Scan_Identifier; | |
2588 | ||
2589 | -- If OK wide space, ignore and keep scanning (we do not include | |
2590 | -- any ignored spaces in checksum) | |
2591 | ||
c8427bff | 2592 | elsif Is_UTF_32_Space (Cat) then |
82c80734 RD |
2593 | goto Scan_Next_Character; |
2594 | ||
2385e007 AC |
2595 | -- If other format character, ignore and keep scanning (again we |
2596 | -- do not include in the checksum) (this is for AI-0079). | |
2597 | ||
2598 | elsif Is_UTF_32_Other (Cat) then | |
2599 | goto Scan_Next_Character; | |
2600 | ||
82c80734 RD |
2601 | -- If OK wide line terminator, terminate current line |
2602 | ||
c8427bff | 2603 | elsif Is_UTF_32_Line_Terminator (UTF_32 (Code)) then |
82c80734 RD |
2604 | Scan_Ptr := Wptr; |
2605 | goto Scan_Line_Terminator; | |
2606 | ||
2607 | -- Punctuation is an error (at start of identifier) | |
2608 | ||
c8427bff | 2609 | elsif Is_UTF_32_Punctuation (Cat) then |
ed2233dc | 2610 | Error_Msg ("identifier cannot start with punctuation", Wptr); |
82c80734 RD |
2611 | Scan_Ptr := Wptr; |
2612 | Name_Len := 0; | |
2613 | Underline_Found := False; | |
2614 | goto Scan_Identifier; | |
2615 | ||
12a13f01 | 2616 | -- Mark character is an error (at start of identifier) |
82c80734 | 2617 | |
c8427bff | 2618 | elsif Is_UTF_32_Mark (Cat) then |
ed2233dc | 2619 | Error_Msg ("identifier cannot start with mark character", Wptr); |
82c80734 RD |
2620 | Scan_Ptr := Wptr; |
2621 | Name_Len := 0; | |
2622 | Underline_Found := False; | |
2623 | goto Scan_Identifier; | |
2624 | ||
82c80734 RD |
2625 | -- Extended digit character is an error. Could be bad start of |
2626 | -- identifier or bad literal. Not worth doing too much to try to | |
2627 | -- distinguish these cases, but we will do a little bit. | |
2628 | ||
c8427bff | 2629 | elsif Is_UTF_32_Digit (Cat) then |
82c80734 RD |
2630 | Error_Msg |
2631 | ("identifier cannot start with digit character", Wptr); | |
2632 | Scan_Ptr := Wptr; | |
2633 | Name_Len := 0; | |
2634 | Underline_Found := False; | |
2635 | goto Scan_Identifier; | |
2636 | ||
2637 | -- All other wide characters are illegal here | |
fbf5a39b AC |
2638 | |
2639 | else | |
82c80734 RD |
2640 | Error_Illegal_Wide_Character; |
2641 | goto Scan_Next_Character; | |
fbf5a39b | 2642 | end if; |
82c80734 | 2643 | end; |
fbf5a39b | 2644 | |
82c80734 RD |
2645 | -- Routine to scan line terminator. On entry Scan_Ptr points to a |
2646 | -- character which is one of FF,LR,CR,VT, or one of the wide characters | |
12a13f01 | 2647 | -- that is treated as a line terminator. |
fbf5a39b | 2648 | |
82c80734 RD |
2649 | <<Scan_Line_Terminator>> |
2650 | ||
2651 | -- Check line too long | |
2652 | ||
2653 | Check_End_Of_Line; | |
2654 | ||
2655 | -- Set Token_Ptr, if End_Of_Line is a token, for the case when it is | |
2656 | -- a physical line. | |
2657 | ||
2658 | if End_Of_Line_Is_Token then | |
2659 | Token_Ptr := Scan_Ptr; | |
2660 | end if; | |
2661 | ||
2662 | declare | |
2663 | Physical : Boolean; | |
2664 | ||
2665 | begin | |
2666 | Skip_Line_Terminators (Scan_Ptr, Physical); | |
2667 | ||
2668 | -- If we are at start of physical line, update scan pointers to | |
2669 | -- reflect the start of the new line. | |
2670 | ||
2671 | if Physical then | |
2672 | Current_Line_Start := Scan_Ptr; | |
2673 | Start_Column := Set_Start_Column; | |
2674 | First_Non_Blank_Location := Scan_Ptr; | |
2675 | ||
2676 | -- If End_Of_Line is a token, we return it as it is a | |
2677 | -- physical line. | |
2678 | ||
2679 | if End_Of_Line_Is_Token then | |
2680 | Token := Tok_End_Of_Line; | |
2681 | return; | |
2682 | end if; | |
fbf5a39b | 2683 | end if; |
82c80734 RD |
2684 | end; |
2685 | ||
2686 | goto Scan_Next_Character; | |
2687 | ||
2688 | -- Identifier scanning routine. On entry, some initial characters of | |
2689 | -- the identifier may have already been stored in Name_Buffer. If so, | |
12a13f01 | 2690 | -- Name_Len has the number of characters stored, otherwise Name_Len is |
82c80734 RD |
2691 | -- set to zero on entry. Underline_Found is also set False on entry. |
2692 | ||
2693 | <<Scan_Identifier>> | |
2694 | ||
2695 | -- This loop scans as fast as possible past lower half letters and | |
2696 | -- digits, which we expect to be the most common characters. | |
fbf5a39b | 2697 | |
82c80734 RD |
2698 | loop |
2699 | if Source (Scan_Ptr) in 'a' .. 'z' | |
2700 | or else Source (Scan_Ptr) in '0' .. '9' | |
fbf5a39b | 2701 | then |
82c80734 RD |
2702 | Name_Buffer (Name_Len + 1) := Source (Scan_Ptr); |
2703 | Accumulate_Checksum (Source (Scan_Ptr)); | |
fbf5a39b | 2704 | |
82c80734 RD |
2705 | elsif Source (Scan_Ptr) in 'A' .. 'Z' then |
2706 | Name_Buffer (Name_Len + 1) := | |
2707 | Character'Val (Character'Pos (Source (Scan_Ptr)) + 32); | |
2708 | Accumulate_Checksum (Name_Buffer (Name_Len + 1)); | |
fbf5a39b AC |
2709 | |
2710 | else | |
fbf5a39b AC |
2711 | exit; |
2712 | end if; | |
2713 | ||
82c80734 RD |
2714 | Underline_Found := False; |
2715 | Scan_Ptr := Scan_Ptr + 1; | |
2716 | Name_Len := Name_Len + 1; | |
fbf5a39b AC |
2717 | end loop; |
2718 | ||
2719 | -- If we fall through, then we have encountered either an underline | |
2720 | -- character, or an extended identifier character (i.e. one from the | |
82c80734 RD |
2721 | -- upper half), or a wide character, or an identifier terminator. The |
2722 | -- initial test speeds us up in the most common case where we have | |
2723 | -- an identifier terminator. Note that ESC is an identifier character | |
2724 | -- only if a wide character encoding method that uses ESC encoding | |
2725 | -- is active, so if we find an ESC character we know that we have a | |
2726 | -- wide character. | |
fbf5a39b | 2727 | |
7f2401a6 RD |
2728 | if Identifier_Char (Source (Scan_Ptr)) |
2729 | or else (Source (Scan_Ptr) in Upper_Half_Character | |
2730 | and then Upper_Half_Encoding) | |
2731 | then | |
fbf5a39b AC |
2732 | -- Case of underline |
2733 | ||
2734 | if Source (Scan_Ptr) = '_' then | |
2735 | Accumulate_Checksum ('_'); | |
2736 | ||
82c80734 RD |
2737 | if Underline_Found then |
2738 | Error_No_Double_Underline; | |
fbf5a39b | 2739 | else |
82c80734 | 2740 | Underline_Found := True; |
fbf5a39b AC |
2741 | Name_Len := Name_Len + 1; |
2742 | Name_Buffer (Name_Len) := '_'; | |
2743 | end if; | |
2744 | ||
2745 | Scan_Ptr := Scan_Ptr + 1; | |
2746 | goto Scan_Identifier; | |
2747 | ||
2748 | -- Upper half character | |
2749 | ||
2750 | elsif Source (Scan_Ptr) in Upper_Half_Character | |
2751 | and then not Upper_Half_Encoding | |
2752 | then | |
2753 | Accumulate_Checksum (Source (Scan_Ptr)); | |
2754 | Store_Encoded_Character | |
2755 | (Get_Char_Code (Fold_Lower (Source (Scan_Ptr)))); | |
2756 | Scan_Ptr := Scan_Ptr + 1; | |
82c80734 | 2757 | Underline_Found := False; |
fbf5a39b AC |
2758 | goto Scan_Identifier; |
2759 | ||
2760 | -- Left bracket not followed by a quote terminates an identifier. | |
2761 | -- This is an error, but we don't want to give a junk error msg | |
a90bd866 | 2762 | -- about wide characters in this case. |
fbf5a39b AC |
2763 | |
2764 | elsif Source (Scan_Ptr) = '[' | |
2765 | and then Source (Scan_Ptr + 1) /= '"' | |
2766 | then | |
2767 | null; | |
2768 | ||
2769 | -- We know we have a wide character encoding here (the current | |
2770 | -- character is either ESC, left bracket, or an upper half | |
2771 | -- character depending on the encoding method). | |
2772 | ||
2773 | else | |
2774 | -- Scan out the wide character and insert the appropriate | |
2775 | -- encoding into the name table entry for the identifier. | |
2776 | ||
2777 | declare | |
c8427bff RD |
2778 | Code : Char_Code; |
2779 | Err : Boolean; | |
2780 | Chr : Character; | |
2781 | Cat : Category; | |
fbf5a39b AC |
2782 | |
2783 | begin | |
82c80734 | 2784 | Wptr := Scan_Ptr; |
fbf5a39b AC |
2785 | Scan_Wide (Source, Scan_Ptr, Code, Err); |
2786 | ||
2787 | -- If error, signal error | |
2788 | ||
2789 | if Err then | |
2790 | Error_Illegal_Wide_Character; | |
2791 | ||
2792 | -- If the character scanned is a normal identifier | |
2793 | -- character, then we treat it that way. | |
2794 | ||
2795 | elsif In_Character_Range (Code) | |
2796 | and then Identifier_Char (Get_Character (Code)) | |
2797 | then | |
2798 | Chr := Get_Character (Code); | |
2799 | Accumulate_Checksum (Chr); | |
2800 | Store_Encoded_Character | |
2801 | (Get_Char_Code (Fold_Lower (Chr))); | |
82c80734 | 2802 | Underline_Found := False; |
fbf5a39b | 2803 | |
82c80734 | 2804 | -- Here if not a normal identifier character |
fbf5a39b AC |
2805 | |
2806 | else | |
2385e007 AC |
2807 | Cat := Get_Category (UTF_32 (Code)); |
2808 | ||
2809 | -- Wide character in Unicode category "Other, Format" | |
2810 | -- is not accepted in an identifier. This is because it | |
2811 | -- it is considered a security risk (AI-0091). | |
2812 | ||
2813 | -- However, it is OK for such a character to appear at | |
2814 | -- the end of an identifier. | |
2815 | ||
2816 | if Is_UTF_32_Other (Cat) then | |
2817 | if not Identifier_Char (Source (Scan_Ptr)) then | |
2818 | goto Scan_Identifier_Complete; | |
2819 | else | |
2820 | Error_Msg | |
2821 | ("identifier cannot contain other_format " | |
2822 | & "character", Wptr); | |
2823 | goto Scan_Identifier; | |
2824 | end if; | |
2825 | ||
2826 | -- Wide character in category Separator,Space terminates | |
2827 | ||
2828 | elsif Is_UTF_32_Space (Cat) then | |
2829 | goto Scan_Identifier_Complete; | |
2830 | end if; | |
2831 | ||
2832 | -- Here if wide character is part of the identifier | |
2833 | ||
fbf5a39b AC |
2834 | -- Make sure we are allowing wide characters in |
2835 | -- identifiers. Note that we allow wide character | |
82c80734 RD |
2836 | -- notation for an OK identifier character. This in |
2837 | -- particular allows bracket or other notation to be | |
2838 | -- used for upper half letters. | |
fbf5a39b | 2839 | |
5f3ab6fb AC |
2840 | -- Wide characters are always allowed in Ada 2005 |
2841 | ||
2842 | if Identifier_Character_Set /= 'w' | |
0791fbe9 | 2843 | and then Ada_Version < Ada_2005 |
5f3ab6fb | 2844 | then |
fbf5a39b | 2845 | Error_Msg |
2385e007 | 2846 | ("wide character not allowed in identifier", Wptr); |
82c80734 RD |
2847 | end if; |
2848 | ||
2849 | -- If OK letter, store it folding to upper case. Note | |
2850 | -- that we include the folded letter in the checksum. | |
2851 | ||
c8427bff RD |
2852 | if Is_UTF_32_Letter (Cat) then |
2853 | Code := | |
2854 | Char_Code (UTF_32_To_Upper_Case (UTF_32 (Code))); | |
82c80734 RD |
2855 | Accumulate_Checksum (Code); |
2856 | Store_Encoded_Character (Code); | |
2857 | Underline_Found := False; | |
2858 | ||
2859 | -- If OK extended digit or mark, then store it | |
2860 | ||
c8427bff RD |
2861 | elsif Is_UTF_32_Digit (Cat) |
2862 | or else Is_UTF_32_Mark (Cat) | |
82c80734 RD |
2863 | then |
2864 | Accumulate_Checksum (Code); | |
2865 | Store_Encoded_Character (Code); | |
2866 | Underline_Found := False; | |
2867 | ||
2868 | -- Wide punctuation is also stored, but counts as an | |
2869 | -- underline character for error checking purposes. | |
2870 | ||
c8427bff | 2871 | elsif Is_UTF_32_Punctuation (Cat) then |
82c80734 RD |
2872 | Accumulate_Checksum (Code); |
2873 | ||
2874 | if Underline_Found then | |
2875 | declare | |
2876 | Cend : constant Source_Ptr := Scan_Ptr; | |
2877 | begin | |
2878 | Scan_Ptr := Wptr; | |
2879 | Error_No_Double_Underline; | |
2880 | Scan_Ptr := Cend; | |
2881 | end; | |
2882 | ||
2883 | else | |
2884 | Store_Encoded_Character (Code); | |
2885 | Underline_Found := True; | |
2886 | end if; | |
2887 | ||
82c80734 RD |
2888 | -- Any other wide character is not acceptable |
2889 | ||
2890 | else | |
2891 | Error_Msg | |
2892 | ("invalid wide character in identifier", Wptr); | |
fbf5a39b AC |
2893 | end if; |
2894 | end if; | |
fbf5a39b | 2895 | |
82c80734 RD |
2896 | goto Scan_Identifier; |
2897 | end; | |
fbf5a39b AC |
2898 | end if; |
2899 | end if; | |
2900 | ||
82c80734 RD |
2901 | -- Scan of identifier is complete. The identifier is stored in |
2902 | -- Name_Buffer, and Scan_Ptr points past the last character. | |
fbf5a39b | 2903 | |
82c80734 | 2904 | <<Scan_Identifier_Complete>> |
fbf5a39b AC |
2905 | Token_Name := Name_Find; |
2906 | ||
82c80734 RD |
2907 | -- Check for identifier ending with underline or punctuation char |
2908 | ||
2909 | if Underline_Found then | |
2910 | Underline_Found := False; | |
2911 | ||
2912 | if Source (Scan_Ptr - 1) = '_' then | |
2913 | Error_Msg | |
2914 | ("identifier cannot end with underline", Scan_Ptr - 1); | |
2915 | else | |
2916 | Error_Msg | |
2917 | ("identifier cannot end with punctuation character", Wptr); | |
2918 | end if; | |
2919 | end if; | |
2920 | ||
feab3549 AC |
2921 | -- We will assume it is an identifier, not a keyword, so that the |
2922 | -- checksum is independent of the Ada version. | |
2923 | ||
2924 | Token := Tok_Identifier; | |
feab3549 | 2925 | |
fbf5a39b AC |
2926 | -- Here is where we check if it was a keyword |
2927 | ||
68e2ea27 | 2928 | if Is_Keyword_Name (Token_Name) then |
b251750b | 2929 | if Opt.Checksum_GNAT_6_3 then |
1ce9dff3 VC |
2930 | Token := Token_Type'Val (Get_Name_Table_Byte (Token_Name)); |
2931 | ||
b251750b VC |
2932 | if Checksum_Accumulate_Token_Checksum then |
2933 | if Checksum_GNAT_5_03 then | |
2934 | Accumulate_Token_Checksum_GNAT_5_03; | |
2935 | else | |
2936 | Accumulate_Token_Checksum_GNAT_6_3; | |
2937 | end if; | |
1ce9dff3 VC |
2938 | end if; |
2939 | ||
2940 | else | |
2941 | Accumulate_Token_Checksum; | |
2942 | Token := Token_Type'Val (Get_Name_Table_Byte (Token_Name)); | |
2943 | end if; | |
fbf5a39b | 2944 | |
b11e8d6f RD |
2945 | -- Keyword style checks |
2946 | ||
2947 | if Style_Check then | |
2948 | ||
2949 | -- Deal with possible style check for non-lower case keyword, | |
2950 | -- but we don't treat ACCESS, DELTA, DIGITS, RANGE as keywords | |
2951 | -- for this purpose if they appear as attribute designators. | |
2952 | -- Actually we only check the first character for speed. | |
2953 | ||
2954 | -- Ada 2005 (AI-284): Do not apply the style check in case of | |
2955 | -- "pragma Interface" | |
2956 | ||
2957 | -- Ada 2005 (AI-340): Do not apply the style check in case of | |
2958 | -- MOD attribute. | |
2959 | ||
2960 | if Source (Token_Ptr) <= 'Z' | |
2961 | and then (Prev_Token /= Tok_Apostrophe | |
2962 | or else | |
2963 | (Token /= Tok_Access and then | |
2964 | Token /= Tok_Delta and then | |
2965 | Token /= Tok_Digits and then | |
2966 | Token /= Tok_Mod and then | |
2967 | Token /= Tok_Range)) | |
2968 | and then (Token /= Tok_Interface | |
2969 | or else | |
2970 | (Token = Tok_Interface | |
2971 | and then Prev_Token /= Tok_Pragma)) | |
2972 | then | |
2973 | Style.Non_Lower_Case_Keyword; | |
2974 | end if; | |
2975 | ||
b46be8a2 | 2976 | -- Check THEN/ELSE style rules. These do not apply to AND THEN |
9b16cb57 | 2977 | -- or OR ELSE, and do not apply in if expressions. |
b46be8a2 | 2978 | |
b11e8d6f RD |
2979 | if (Token = Tok_Then and then Prev_Token /= Tok_And) |
2980 | or else | |
2981 | (Token = Tok_Else and then Prev_Token /= Tok_Or) | |
2982 | then | |
9b16cb57 | 2983 | if Inside_If_Expression = 0 then |
b46be8a2 RD |
2984 | Style.Check_Separate_Stmt_Lines; |
2985 | end if; | |
b11e8d6f | 2986 | end if; |
fbf5a39b AC |
2987 | end if; |
2988 | ||
82c80734 RD |
2989 | -- We must reset Token_Name since this is not an identifier and |
2990 | -- if we leave Token_Name set, the parser gets confused because | |
2991 | -- it thinks it is dealing with an identifier instead of the | |
2992 | -- corresponding keyword. | |
fbf5a39b AC |
2993 | |
2994 | Token_Name := No_Name; | |
2995 | return; | |
2996 | ||
2997 | -- It is an identifier after all | |
2998 | ||
2999 | else | |
b251750b VC |
3000 | if Checksum_Accumulate_Token_Checksum then |
3001 | Accumulate_Token_Checksum; | |
3002 | end if; | |
3003 | ||
fbf5a39b AC |
3004 | Post_Scan; |
3005 | return; | |
3006 | end if; | |
3007 | end Scan; | |
6b6fcd3e | 3008 | |
c45b6ae0 AC |
3009 | -------------------------- |
3010 | -- Set_Comment_As_Token -- | |
3011 | -------------------------- | |
3012 | ||
3013 | procedure Set_Comment_As_Token (Value : Boolean) is | |
3014 | begin | |
3015 | Comment_Is_Token := Value; | |
3016 | end Set_Comment_As_Token; | |
fbf5a39b AC |
3017 | |
3018 | ------------------------------ | |
3019 | -- Set_End_Of_Line_As_Token -- | |
3020 | ------------------------------ | |
3021 | ||
3022 | procedure Set_End_Of_Line_As_Token (Value : Boolean) is | |
3023 | begin | |
3024 | End_Of_Line_Is_Token := Value; | |
3025 | end Set_End_Of_Line_As_Token; | |
3026 | ||
3027 | --------------------------- | |
3028 | -- Set_Special_Character -- | |
3029 | --------------------------- | |
3030 | ||
3031 | procedure Set_Special_Character (C : Character) is | |
3032 | begin | |
3033 | case C is | |
3034 | when '#' | '$' | '_' | '?' | '@' | '`' | '\' | '^' | '~' => | |
3035 | Special_Characters (C) := True; | |
3036 | ||
3037 | when others => | |
3038 | null; | |
3039 | end case; | |
3040 | end Set_Special_Character; | |
3041 | ||
3042 | ---------------------- | |
3043 | -- Set_Start_Column -- | |
3044 | ---------------------- | |
3045 | ||
3046 | -- Note: it seems at first glance a little expensive to compute this value | |
3047 | -- for every source line (since it is certainly not used for all source | |
3048 | -- lines). On the other hand, it doesn't take much more work to skip past | |
3049 | -- the initial white space on the line counting the columns than it would | |
3050 | -- to scan past the white space using the standard scanning circuits. | |
3051 | ||
3052 | function Set_Start_Column return Column_Number is | |
3053 | Start_Column : Column_Number := 0; | |
3054 | ||
3055 | begin | |
3056 | -- Outer loop scans past horizontal tab characters | |
3057 | ||
3058 | Tabs_Loop : loop | |
3059 | ||
3060 | -- Inner loop scans past blanks as fast as possible, bumping Scan_Ptr | |
3061 | -- past the blanks and adjusting Start_Column to account for them. | |
3062 | ||
3063 | Blanks_Loop : loop | |
3064 | if Source (Scan_Ptr) = ' ' then | |
3065 | if Source (Scan_Ptr + 1) = ' ' then | |
3066 | if Source (Scan_Ptr + 2) = ' ' then | |
3067 | if Source (Scan_Ptr + 3) = ' ' then | |
3068 | if Source (Scan_Ptr + 4) = ' ' then | |
3069 | if Source (Scan_Ptr + 5) = ' ' then | |
3070 | if Source (Scan_Ptr + 6) = ' ' then | |
3071 | Scan_Ptr := Scan_Ptr + 7; | |
3072 | Start_Column := Start_Column + 7; | |
3073 | else | |
3074 | Scan_Ptr := Scan_Ptr + 6; | |
3075 | Start_Column := Start_Column + 6; | |
3076 | exit Blanks_Loop; | |
3077 | end if; | |
3078 | else | |
3079 | Scan_Ptr := Scan_Ptr + 5; | |
3080 | Start_Column := Start_Column + 5; | |
3081 | exit Blanks_Loop; | |
3082 | end if; | |
3083 | else | |
3084 | Scan_Ptr := Scan_Ptr + 4; | |
3085 | Start_Column := Start_Column + 4; | |
3086 | exit Blanks_Loop; | |
3087 | end if; | |
3088 | else | |
3089 | Scan_Ptr := Scan_Ptr + 3; | |
3090 | Start_Column := Start_Column + 3; | |
3091 | exit Blanks_Loop; | |
3092 | end if; | |
3093 | else | |
3094 | Scan_Ptr := Scan_Ptr + 2; | |
3095 | Start_Column := Start_Column + 2; | |
3096 | exit Blanks_Loop; | |
3097 | end if; | |
3098 | else | |
3099 | Scan_Ptr := Scan_Ptr + 1; | |
3100 | Start_Column := Start_Column + 1; | |
3101 | exit Blanks_Loop; | |
3102 | end if; | |
3103 | else | |
3104 | exit Blanks_Loop; | |
3105 | end if; | |
3106 | end loop Blanks_Loop; | |
3107 | ||
3108 | -- Outer loop keeps going only if a horizontal tab follows | |
3109 | ||
3110 | if Source (Scan_Ptr) = HT then | |
b11e8d6f RD |
3111 | if Style_Check then |
3112 | Style.Check_HT; | |
3113 | end if; | |
3114 | ||
fbf5a39b AC |
3115 | Scan_Ptr := Scan_Ptr + 1; |
3116 | Start_Column := (Start_Column / 8) * 8 + 8; | |
3117 | else | |
3118 | exit Tabs_Loop; | |
3119 | end if; | |
fbf5a39b AC |
3120 | end loop Tabs_Loop; |
3121 | ||
3122 | return Start_Column; | |
0e701488 RD |
3123 | |
3124 | -- A constraint error can happen only if we have a compiler with checks on | |
3125 | -- and a line with a ludicrous number of tabs or spaces at the start. In | |
3126 | -- such a case, we really don't care if Start_Column is right or not. | |
3127 | ||
3128 | exception | |
3129 | when Constraint_Error => | |
3130 | return Start_Column; | |
fbf5a39b AC |
3131 | end Set_Start_Column; |
3132 | ||
3133 | end Scng; |