1 ------------------------------------------------------------------------------
3 -- GNAT COMPILER COMPONENTS --
9 -- Copyright (C) 1992-2014, Free Software Foundation, Inc. --
11 -- GNAT is free software; you can redistribute it and/or modify it under --
12 -- terms of the GNU General Public License as published by the Free Soft- --
13 -- ware Foundation; either version 3, or (at your option) any later ver- --
14 -- sion. GNAT is distributed in the hope that it will be useful, but WITH- --
15 -- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY --
16 -- or FITNESS FOR A PARTICULAR PURPOSE. --
18 -- As a special exception under Section 7 of GPL version 3, you are granted --
19 -- additional permissions described in the GCC Runtime Library Exception, --
20 -- version 3.1, as published by the Free Software Foundation. --
22 -- You should have received a copy of the GNU General Public License and --
23 -- a copy of the GCC Runtime Library Exception along with this program; --
24 -- see the files COPYING3 and COPYING.RUNTIME respectively. If not, see --
25 -- <http://www.gnu.org/licenses/>. --
27 -- GNAT was originally developed by the GNAT team at New York University. --
28 -- Extensive contributions were provided by Ada Core Technologies Inc. --
30 ------------------------------------------------------------------------------
32 with Namet; use Namet;
33 with Types; use Types;
34 with Uintp; use Uintp;
35 with Urealp; use Urealp;
39 -- The scanner maintains a current state in the global variables defined
40 -- in this package. The call to the Scan routine advances this state to
41 -- the next token. The state is initialized by the call to one of the
42 -- initialization routines in Sinput.
44 -- The following type is used to identify token types returned by Scan.
45 -- The class column in this table indicates the token classes which
46 -- apply to the token, as defined by subsequent subtype declarations.
48 -- Note: Namet.Is_Keyword_Name depends on the fact that the first entry in
49 -- this type declaration is *not* for a reserved word. For details on why
50 -- there is this requirement, see Initialize_Ada_Keywords below.
54 -- Token name Token type Class(es)
56 Tok_Integer_Literal, -- numeric lit Literal, Lit_Or_Name
58 Tok_Real_Literal, -- numeric lit Literal, Lit_Or_Name
60 Tok_String_Literal, -- string lit Literal. Lit_Or_Name
62 Tok_Char_Literal, -- char lit Name, Literal. Lit_Or_Name
64 Tok_Operator_Symbol, -- op symbol Name, Literal, Lit_Or_Name, Desig
66 Tok_Identifier, -- identifier Name, Lit_Or_Name, Desig
68 Tok_Double_Asterisk, -- **
70 Tok_Ampersand, -- & Binary_Addop
71 Tok_Minus, -- - Binary_Addop, Unary_Addop
72 Tok_Plus, -- + Binary_Addop, Unary_Addop
74 Tok_Asterisk, -- * Mulop
85 -- Note: Tok_Raise is in no categories now, it used to be Cterm, Eterm,
86 -- After_SM, but now that Ada 2012 has added raise expressions, the
87 -- raise token can appear anywhere. Note in particular that Tok_Raise
88 -- being in Eterm stopped the parser from recognizing "return raise
89 -- exception-name". This degrades error recovery slightly, and perhaps
90 -- we could do better, but not worth the effort.
95 Tok_Apostrophe, -- ' Namext
97 Tok_Left_Paren, -- ( Namext, Consk
99 Tok_Delta, -- DELTA Atkwd, Sterm, Consk
100 Tok_Digits, -- DIGITS Atkwd, Sterm, Consk
101 Tok_Range, -- RANGE Atkwd, Sterm, Consk
103 Tok_Right_Paren, -- ) Sterm
104 Tok_Comma, -- , Sterm
106 Tok_And, -- AND Logop, Sterm
107 Tok_Or, -- OR Logop, Sterm
108 Tok_Xor, -- XOR Logop, Sterm
110 Tok_Less, -- < Relop, Sterm
111 Tok_Equal, -- = Relop, Sterm
112 Tok_Greater, -- > Relop, Sterm
113 Tok_Not_Equal, -- /= Relop, Sterm
114 Tok_Greater_Equal, -- >= Relop, Sterm
115 Tok_Less_Equal, -- <= Relop, Sterm
117 Tok_In, -- IN Relop, Sterm
118 Tok_Not, -- NOT Relop, Sterm
120 Tok_Box, -- <> Relop, Eterm, Sterm
121 Tok_Colon_Equal, -- := Eterm, Sterm
122 Tok_Colon, -- : Eterm, Sterm
123 Tok_Greater_Greater, -- >> Eterm, Sterm
125 Tok_Abstract, -- ABSTRACT Eterm, Sterm
126 Tok_Access, -- ACCESS Eterm, Sterm
127 Tok_Aliased, -- ALIASED Eterm, Sterm
128 Tok_All, -- ALL Eterm, Sterm
129 Tok_Array, -- ARRAY Eterm, Sterm
130 Tok_At, -- AT Eterm, Sterm
131 Tok_Body, -- BODY Eterm, Sterm
132 Tok_Constant, -- CONSTANT Eterm, Sterm
133 Tok_Do, -- DO Eterm, Sterm
134 Tok_Is, -- IS Eterm, Sterm
135 Tok_Interface, -- INTERFACE Eterm, Sterm
136 Tok_Limited, -- LIMITED Eterm, Sterm
137 Tok_Of, -- OF Eterm, Sterm
138 Tok_Out, -- OUT Eterm, Sterm
139 Tok_Record, -- RECORD Eterm, Sterm
140 Tok_Renames, -- RENAMES Eterm, Sterm
141 Tok_Reverse, -- REVERSE Eterm, Sterm
142 Tok_Some, -- SOME Eterm, Sterm
143 Tok_Tagged, -- TAGGED Eterm, Sterm
144 Tok_Then, -- THEN Eterm, Sterm
146 Tok_Less_Less, -- << Eterm, Sterm, After_SM
148 Tok_Abort, -- ABORT Eterm, Sterm, After_SM
149 Tok_Accept, -- ACCEPT Eterm, Sterm, After_SM
150 Tok_Case, -- CASE Eterm, Sterm, After_SM
151 Tok_Delay, -- DELAY Eterm, Sterm, After_SM
152 Tok_Else, -- ELSE Eterm, Sterm, After_SM
153 Tok_Elsif, -- ELSIF Eterm, Sterm, After_SM
154 Tok_End, -- END Eterm, Sterm, After_SM
155 Tok_Exception, -- EXCEPTION Eterm, Sterm, After_SM
156 Tok_Exit, -- EXIT Eterm, Sterm, After_SM
157 Tok_Goto, -- GOTO Eterm, Sterm, After_SM
158 Tok_If, -- IF Eterm, Sterm, After_SM
159 Tok_Pragma, -- PRAGMA Eterm, Sterm, After_SM
160 Tok_Requeue, -- REQUEUE Eterm, Sterm, After_SM
161 Tok_Return, -- RETURN Eterm, Sterm, After_SM
162 Tok_Select, -- SELECT Eterm, Sterm, After_SM
163 Tok_Terminate, -- TERMINATE Eterm, Sterm, After_SM
164 Tok_Until, -- UNTIL Eterm, Sterm, After_SM
165 Tok_When, -- WHEN Eterm, Sterm, After_SM
167 Tok_Begin, -- BEGIN Eterm, Sterm, After_SM, Labeled_Stmt
168 Tok_Declare, -- DECLARE Eterm, Sterm, After_SM, Labeled_Stmt
169 Tok_For, -- FOR Eterm, Sterm, After_SM, Labeled_Stmt
170 Tok_Loop, -- LOOP Eterm, Sterm, After_SM, Labeled_Stmt
171 Tok_While, -- WHILE Eterm, Sterm, After_SM, Labeled_Stmt
173 Tok_Entry, -- ENTRY Eterm, Sterm, Declk, Deckn, After_SM
174 Tok_Protected, -- PROTECTED Eterm, Sterm, Declk, Deckn, After_SM
175 Tok_Task, -- TASK Eterm, Sterm, Declk, Deckn, After_SM
176 Tok_Type, -- TYPE Eterm, Sterm, Declk, Deckn, After_SM
177 Tok_Subtype, -- SUBTYPE Eterm, Sterm, Declk, Deckn, After_SM
178 Tok_Overriding, -- OVERRIDING Eterm, Sterm, Declk, Declk, After_SM
179 Tok_Synchronized, -- SYNCHRONIZED Eterm, Sterm, Declk, Deckn, After_SM
180 Tok_Use, -- USE Eterm, Sterm, Declk, Deckn, After_SM
182 Tok_Function, -- FUNCTION Eterm, Sterm, Cunit, Declk, After_SM
183 Tok_Generic, -- GENERIC Eterm, Sterm, Cunit, Declk, After_SM
184 Tok_Package, -- PACKAGE Eterm, Sterm, Cunit, Declk, After_SM
185 Tok_Procedure, -- PROCEDURE Eterm, Sterm, Cunit, Declk, After_SM
187 Tok_Private, -- PRIVATE Eterm, Sterm, Cunit, After_SM
188 Tok_With, -- WITH Eterm, Sterm, Cunit, After_SM
189 Tok_Separate, -- SEPARATE Eterm, Sterm, Cunit, After_SM
191 Tok_EOF, -- End of file Eterm, Sterm, Cterm, After_SM
193 Tok_Semicolon, -- ; Eterm, Sterm, Cterm
195 Tok_Arrow, -- => Sterm, Cterm, Chtok
197 Tok_Vertical_Bar, -- | Cterm, Sterm, Chtok
199 Tok_Dot_Dot, -- .. Sterm, Chtok
204 Tok_External_As_List,
205 -- These four entries represent keywords for the project file language
206 -- and can be returned only in the case of scanning project files.
209 -- This entry is used when scanning project files (where it represents
210 -- an entire comment), and in preprocessing with the -C switch set
211 -- (where it represents just the "--" of a comment). For the project
212 -- file case, the text of the comment is stored in Comment_Id.
215 -- Represents an end of line. Not used during normal compilation scans
216 -- where end of line is ignored. Active for preprocessor scanning and
217 -- also when scanning project files (where it is needed because of ???)
220 -- Used only in preprocessor scanning (to represent one of the
221 -- characters '#', '$', '?', '@', '`', '\', '^', '~', or '_'. The
222 -- character value itself is stored in Scans.Special_Character.
225 -- HIDE directive in SPARK
228 -- No_Token is used for initializing Token values to indicate that
229 -- no value has been set yet.
231 -- Note: in the RM, operator symbol is a special case of string literal.
232 -- We distinguish at the lexical level in this compiler, since there are
233 -- many syntactic situations in which only an operator symbol is allowed.
235 -- The following subtype declarations group the token types into classes.
236 -- These are used for class tests in the parser.
238 subtype Token_Class_Numeric_Literal is
239 Token_Type range Tok_Integer_Literal .. Tok_Real_Literal;
242 subtype Token_Class_Literal is
243 Token_Type range Tok_Integer_Literal .. Tok_Operator_Symbol;
246 subtype Token_Class_Lit_Or_Name is
247 Token_Type range Tok_Integer_Literal .. Tok_Identifier;
249 subtype Token_Class_Binary_Addop is
250 Token_Type range Tok_Ampersand .. Tok_Plus;
251 -- Binary adding operator (& + -)
253 subtype Token_Class_Unary_Addop is
254 Token_Type range Tok_Minus .. Tok_Plus;
255 -- Unary adding operator (+ -)
257 subtype Token_Class_Mulop is
258 Token_Type range Tok_Asterisk .. Tok_Slash;
259 -- Multiplying operator
261 subtype Token_Class_Logop is
262 Token_Type range Tok_And .. Tok_Xor;
263 -- Logical operator (and, or, xor)
265 subtype Token_Class_Relop is
266 Token_Type range Tok_Less .. Tok_Box;
267 -- Relational operator (= /= < <= > >= not, in plus <> to catch misuse
268 -- of Pascal style not equal operator).
270 subtype Token_Class_Name is
271 Token_Type range Tok_Char_Literal .. Tok_Identifier;
272 -- First token of name (4.1),
273 -- (identifier, char literal, operator symbol)
275 subtype Token_Class_Desig is
276 Token_Type range Tok_Operator_Symbol .. Tok_Identifier;
277 -- Token which can be a Designator (identifier, operator symbol)
279 subtype Token_Class_Namext is
280 Token_Type range Tok_Dot .. Tok_Left_Paren;
281 -- Name extension tokens. These are tokens which can appear immediately
282 -- after a name to extend it recursively (period, quote, left paren)
284 subtype Token_Class_Consk is
285 Token_Type range Tok_Left_Paren .. Tok_Range;
286 -- Keywords which can start constraint
287 -- (left paren, delta, digits, range)
289 subtype Token_Class_Eterm is
290 Token_Type range Tok_Colon_Equal .. Tok_Semicolon;
291 -- Expression terminators. These tokens can never appear within a simple
292 -- expression. This is used for error recovery purposes (if we encounter
293 -- an error in an expression, we simply scan to the next Eterm token).
295 subtype Token_Class_Sterm is
296 Token_Type range Tok_Delta .. Tok_Dot_Dot;
297 -- Simple_Expression terminators. A Simple_Expression must be followed
298 -- by a token in this class, or an error message is issued complaining
299 -- about a missing binary operator.
301 subtype Token_Class_Atkwd is
302 Token_Type range Tok_Delta .. Tok_Range;
303 -- Attribute keywords. This class includes keywords which can be used
304 -- as an Attribute_Designator, namely DELTA, DIGITS and RANGE
306 subtype Token_Class_Cterm is
307 Token_Type range Tok_EOF .. Tok_Vertical_Bar;
308 -- Choice terminators. These tokens terminate a choice. This is used for
309 -- error recovery purposes (if we encounter an error in a Choice, we
310 -- simply scan to the next Cterm token).
312 subtype Token_Class_Chtok is
313 Token_Type range Tok_Arrow .. Tok_Dot_Dot;
314 -- Choice tokens. These tokens signal a choice when used in an Aggregate
316 subtype Token_Class_Cunit is
317 Token_Type range Tok_Function .. Tok_Separate;
318 -- Tokens which can begin a compilation unit
320 subtype Token_Class_Declk is
321 Token_Type range Tok_Entry .. Tok_Procedure;
322 -- Keywords which start a declaration
324 subtype Token_Class_Deckn is
325 Token_Type range Tok_Entry .. Tok_Use;
326 -- Keywords which start a declaration but can't start a compilation unit
328 subtype Token_Class_After_SM is
329 Token_Type range Tok_Less_Less .. Tok_EOF;
330 -- Tokens which always, or almost always, appear after a semicolon. Used
331 -- in the Resync_Past_Semicolon routine to avoid gobbling up stuff when
332 -- a semicolon is missing. Of significance only for error recovery.
334 subtype Token_Class_Labeled_Stmt is
335 Token_Type range Tok_Begin .. Tok_While;
336 -- Tokens which start labeled statements
338 type Token_Flag_Array is array (Token_Type) of Boolean;
339 Is_Reserved_Keyword : constant Token_Flag_Array :=
341 (Tok_Mod .. Tok_Rem => True,
342 Tok_New .. Tok_Null => True,
343 Tok_Delta .. Tok_Range => True,
344 Tok_And .. Tok_Xor => True,
345 Tok_In .. Tok_Not => True,
346 Tok_Abstract .. Tok_Then => True,
347 Tok_Abort .. Tok_Separate => True,
349 -- Flag array used to test for reserved word
351 procedure Initialize_Ada_Keywords;
352 -- Set up Token_Type values in Names table entries for Ada reserved
353 -- words. This ignores Ada_Version; Ada_Version is taken into account in
354 -- Snames.Is_Keyword_Name.
356 --------------------------
357 -- Scan State Variables --
358 --------------------------
360 -- Note: these variables can only be referenced during the parsing of a
361 -- file. Reference to any of them from Sem or the expander is wrong.
363 -- These variables are initialized as required by Scn.Initialize_Scanner,
364 -- and should not be referenced before such a call. However, there are
365 -- situations in which these variables are saved and restored, and this
366 -- may happen before the first Initialize_Scanner call, resulting in the
367 -- assignment of invalid values. To avoid this, and allow building with
368 -- the -gnatVa switch, we initialize some variables to known valid values.
370 Scan_Ptr : Source_Ptr := No_Location; -- init for -gnatVa
371 -- Current scan pointer location. After a call to Scan, this points
372 -- just past the end of the token just scanned.
374 Token : Token_Type := No_Token; -- init for -gnatVa
375 -- Type of current token
377 Token_Ptr : Source_Ptr := No_Location; -- init for -gnatVa
378 -- Pointer to first character of current token
380 Current_Line_Start : Source_Ptr := No_Location; -- init for -gnatVa
381 -- Pointer to first character of line containing current token
383 Start_Column : Column_Number := No_Column_Number; -- init for -gnatVa
384 -- Starting column number (zero origin) of the first non-blank character
385 -- on the line containing the current token. This is used for error
386 -- recovery circuits which depend on looking at the column line up.
388 Type_Token_Location : Source_Ptr := No_Location; -- init for -gnatVa
389 -- Within a type declaration, gives the location of the TYPE keyword that
390 -- opened the type declaration. Used in checking the end column of a record
391 -- declaration, which can line up either with the TYPE keyword, or with the
392 -- start of the line containing the RECORD keyword.
394 Checksum : Word := 0; -- init for -gnatVa
395 -- Used to accumulate a CRC representing the tokens in the source
396 -- file being compiled. This CRC includes only program tokens, and
397 -- excludes comments.
399 First_Non_Blank_Location : Source_Ptr := No_Location; -- init for -gnatVa
400 -- Location of first non-blank character on the line containing the
401 -- current token (i.e. the location of the character whose column number
402 -- is stored in Start_Column).
404 Token_Node : Node_Id := Empty;
405 -- Node table Id for the current token. This is set only if the current
406 -- token is one for which the scanner constructs a node (i.e. it is an
407 -- identifier, operator symbol, or literal). For other token types,
408 -- Token_Node is undefined.
410 Token_Name : Name_Id := No_Name;
411 -- For identifiers, this is set to the Name_Id of the identifier scanned.
412 -- For all other tokens, Token_Name is set to Error_Name. Note that it
413 -- would be possible for the caller to extract this information from
414 -- Token_Node. We set Token_Name separately for two reasons. First it
415 -- allows a quicker test for a specific identifier. Second, it allows
416 -- a version of the parser to be built that does not build tree nodes,
417 -- usable as a syntax checker.
419 Prev_Token : Token_Type := No_Token;
420 -- Type of previous token
422 Prev_Token_Ptr : Source_Ptr;
423 -- Pointer to first character of previous token
425 Version_To_Be_Found : Boolean;
426 -- This flag is True if the scanner is still looking for an RCS version
427 -- number in a comment. Normally it is initialized to False so that this
428 -- circuit is not activated. If the -dv switch is set, then this flag is
429 -- initialized to True, and then reset when the version number is found.
430 -- We do things this way to minimize the impact on comment scanning.
432 Character_Code : Char_Code;
433 -- Valid only when Token is Tok_Char_Literal. Contains the value of the
436 Real_Literal_Value : Ureal;
437 -- Valid only when Token is Tok_Real_Literal, contains the value of the
440 Int_Literal_Value : Uint;
441 -- Valid only when Token = Tok_Integer_Literal, contains the value of the
444 Based_Literal_Uses_Colon : Boolean;
445 -- Valid only when Token = Tok_Integer_Literal or Tok_Real_Literal. Set
446 -- True only for the case of a based literal using ':' instead of '#'.
448 String_Literal_Id : String_Id;
449 -- Valid only when Token = Tok_String_Literal or Tok_Operator_Symbol.
450 -- Contains the Id for currently scanned string value.
452 Wide_Character_Found : Boolean := False;
453 -- Valid only when Token = Tok_String_Literal. Set True if wide character
454 -- found (i.e. a character that does not fit in Character, but fits in
455 -- Wide_Wide_Character).
457 Wide_Wide_Character_Found : Boolean := False;
458 -- Valid only when Token = Tok_String_Literal. Set True if wide wide
459 -- character found (i.e. a character that does not fit in Character or
462 Special_Character : Character;
463 -- Valid only when Token = Tok_Special. Returns one of the characters
464 -- '#', '$', '?', '@', '`', '\', '^', '~', or '_'.
466 -- Why only this set? What about wide characters???
468 Comment_Id : Name_Id := No_Name;
469 -- Valid only when Token = Tok_Comment. Store the string that follows
470 -- the "--" of a comment when scanning project files.
472 -- Is it really right for this to be a Name rather than a String, what
473 -- about the case of Wide_Wide_Characters???
475 Inside_Depends : Boolean := False;
476 -- Flag set True for parsing the argument of a Depends pragma or aspect
477 -- (used to allow/require non-standard style rules for =>+ with -gnatyt).
479 Inside_If_Expression : Nat := 0;
480 -- This is a counter that is set non-zero while scanning out an if
481 -- expression (incremented on entry, decremented on exit). It is used to
482 -- disconnect format checks that normally apply to keywords THEN, ELSE etc.
484 --------------------------------------------------------
485 -- Procedures for Saving and Restoring the Scan State --
486 --------------------------------------------------------
488 -- The following procedures can be used to save and restore the entire
489 -- scan state. They are used in cases where it is necessary to backup
490 -- the scan during the parse.
492 type Saved_Scan_State is private;
493 -- Used for saving and restoring the scan state
495 procedure Save_Scan_State (Saved_State : out Saved_Scan_State);
496 pragma Inline (Save_Scan_State);
497 -- Saves the current scan state for possible later restoration. Note that
498 -- there is no harm in saving the state and then never restoring it.
500 procedure Restore_Scan_State (Saved_State : Saved_Scan_State);
501 pragma Inline (Restore_Scan_State);
502 -- Restores a scan state saved by a call to Save_Scan_State.
503 -- The saved scan state must refer to the current source file.
506 type Saved_Scan_State is record
507 Save_Scan_Ptr : Source_Ptr;
508 Save_Token : Token_Type;
509 Save_Token_Ptr : Source_Ptr;
510 Save_Current_Line_Start : Source_Ptr;
511 Save_Start_Column : Column_Number;
512 Save_Checksum : Word;
513 Save_First_Non_Blank_Location : Source_Ptr;
514 Save_Token_Node : Node_Id;
515 Save_Token_Name : Name_Id;
516 Save_Prev_Token : Token_Type;
517 Save_Prev_Token_Ptr : Source_Ptr;