1------------------------------------------------------------------------------ 2-- -- 3-- GNAT COMPILER COMPONENTS -- 4-- -- 5-- S C A N S -- 6-- -- 7-- S p e c -- 8-- -- 9-- Copyright (C) 1992-2021, Free Software Foundation, Inc. -- 10-- -- 11-- GNAT is free software; you can redistribute it and/or modify it under -- 12-- terms of the GNU General Public License as published by the Free Soft- -- 13-- ware Foundation; either version 3, or (at your option) any later ver- -- 14-- sion. GNAT is distributed in the hope that it will be useful, but WITH- -- 15-- OUT ANY WARRANTY; without even the implied warranty of MERCHANTABILITY -- 16-- or FITNESS FOR A PARTICULAR PURPOSE. See the GNU General Public License -- 17-- for more details. You should have received a copy of the GNU General -- 18-- Public License distributed with GNAT; see file COPYING3. If not, go to -- 19-- http://www.gnu.org/licenses for a complete copy of the license. -- 20-- -- 21-- GNAT was originally developed by the GNAT team at New York University. -- 22-- Extensive contributions were provided by Ada Core Technologies Inc. -- 23-- -- 24------------------------------------------------------------------------------ 25 26with Namet; use Namet; 27with Types; use Types; 28with Uintp; use Uintp; 29with Urealp; use Urealp; 30 31package Scans is 32 33-- The scanner maintains a current state in the global variables defined 34-- in this package. The call to the Scan routine advances this state to 35-- the next token. The state is initialized by the call to one of the 36-- initialization routines in Sinput. 37 38 -- The following type is used to identify token types returned by Scan. 39 -- The class column in this table indicates the token classes which 40 -- apply to the token, as defined by subsequent subtype declarations. 41 42 type Token_Type is ( 43 44 -- Token name Token type Class(es) 45 46 Tok_Integer_Literal, -- numeric lit Literal, Lit_Or_Name 47 48 Tok_Real_Literal, -- numeric lit Literal, Lit_Or_Name 49 50 Tok_String_Literal, -- string lit Literal. Lit_Or_Name 51 52 Tok_Char_Literal, -- char lit Name, Literal. Lit_Or_Name 53 54 Tok_Operator_Symbol, -- op symbol Name, Literal, Lit_Or_Name, Desig 55 56 Tok_Identifier, -- identifier Name, Lit_Or_Name, Desig 57 58 Tok_At_Sign, -- @ AI12-0125-3 : target name 59 60 Tok_Double_Asterisk, -- ** 61 62 Tok_Ampersand, -- & Binary_Addop 63 Tok_Minus, -- - Binary_Addop, Unary_Addop 64 Tok_Plus, -- + Binary_Addop, Unary_Addop 65 66 Tok_Asterisk, -- * Mulop 67 Tok_Mod, -- MOD Mulop 68 Tok_Rem, -- REM Mulop 69 Tok_Slash, -- / Mulop 70 71 Tok_New, -- NEW 72 73 Tok_Abs, -- ABS 74 Tok_Others, -- OTHERS 75 Tok_Null, -- NULL 76 77 -- Note: Tok_Raise is in no categories now, it used to be Cterm, Eterm, 78 -- After_SM, but now that Ada 2012 has added raise expressions, the 79 -- raise token can appear anywhere. Note in particular that Tok_Raise 80 -- being in Eterm stopped the parser from recognizing "return raise 81 -- exception-name". This degrades error recovery slightly, and perhaps 82 -- we could do better, but not worth the effort. 83 84 -- Ada 2022 introduces square brackets as delimiters for array and 85 -- container aggregates. 86 87 Tok_Raise, -- RAISE 88 89 Tok_Dot, -- . Namext 90 Tok_Apostrophe, -- ' Namext 91 92 Tok_Left_Bracket, -- [ Namest 93 Tok_Left_Paren, -- ( Namext, Consk 94 95 Tok_Delta, -- DELTA Atkwd, Sterm, Consk 96 Tok_Digits, -- DIGITS Atkwd, Sterm, Consk 97 Tok_Range, -- RANGE Atkwd, Sterm, Consk 98 99 Tok_Right_Paren, -- ) Sterm 100 Tok_Right_Bracket, -- ] Sterm 101 Tok_Comma, -- , Sterm 102 103 Tok_And, -- AND Logop, Sterm 104 Tok_Or, -- OR Logop, Sterm 105 Tok_Xor, -- XOR Logop, Sterm 106 107 Tok_Less, -- < Relop, Sterm 108 Tok_Equal, -- = Relop, Sterm 109 Tok_Greater, -- > Relop, Sterm 110 Tok_Not_Equal, -- /= Relop, Sterm 111 Tok_Greater_Equal, -- >= Relop, Sterm 112 Tok_Less_Equal, -- <= Relop, Sterm 113 114 Tok_In, -- IN Relop, Sterm 115 Tok_Not, -- NOT Relop, Sterm 116 117 Tok_Box, -- <> Relop, Eterm, Sterm 118 Tok_Colon_Equal, -- := Eterm, Sterm 119 Tok_Colon, -- : Eterm, Sterm 120 Tok_Greater_Greater, -- >> Eterm, Sterm 121 122 Tok_Abstract, -- ABSTRACT Eterm, Sterm 123 Tok_Access, -- ACCESS Eterm, Sterm 124 Tok_Aliased, -- ALIASED Eterm, Sterm 125 Tok_All, -- ALL Eterm, Sterm 126 Tok_Array, -- ARRAY Eterm, Sterm 127 Tok_At, -- AT Eterm, Sterm 128 Tok_Body, -- BODY Eterm, Sterm 129 Tok_Constant, -- CONSTANT Eterm, Sterm 130 Tok_Do, -- DO Eterm, Sterm 131 Tok_Is, -- IS Eterm, Sterm 132 Tok_Interface, -- INTERFACE Eterm, Sterm 133 Tok_Limited, -- LIMITED Eterm, Sterm 134 Tok_Of, -- OF Eterm, Sterm 135 Tok_Out, -- OUT Eterm, Sterm 136 Tok_Record, -- RECORD Eterm, Sterm 137 Tok_Renames, -- RENAMES Eterm, Sterm 138 Tok_Reverse, -- REVERSE Eterm, Sterm 139 Tok_Some, -- SOME Eterm, Sterm 140 Tok_Tagged, -- TAGGED Eterm, Sterm 141 Tok_Then, -- THEN Eterm, Sterm 142 143 Tok_Less_Less, -- << Eterm, Sterm, After_SM 144 145 Tok_Abort, -- ABORT Eterm, Sterm, After_SM 146 Tok_Accept, -- ACCEPT Eterm, Sterm, After_SM 147 Tok_Case, -- CASE Eterm, Sterm, After_SM 148 Tok_Delay, -- DELAY Eterm, Sterm, After_SM 149 Tok_Else, -- ELSE Eterm, Sterm, After_SM 150 Tok_Elsif, -- ELSIF Eterm, Sterm, After_SM 151 Tok_End, -- END Eterm, Sterm, After_SM 152 Tok_Exception, -- EXCEPTION Eterm, Sterm, After_SM 153 Tok_Exit, -- EXIT Eterm, Sterm, After_SM 154 Tok_Goto, -- GOTO Eterm, Sterm, After_SM 155 Tok_If, -- IF Eterm, Sterm, After_SM 156 Tok_Pragma, -- PRAGMA Eterm, Sterm, After_SM 157 Tok_Requeue, -- REQUEUE Eterm, Sterm, After_SM 158 Tok_Return, -- RETURN Eterm, Sterm, After_SM 159 Tok_Select, -- SELECT Eterm, Sterm, After_SM 160 Tok_Terminate, -- TERMINATE Eterm, Sterm, After_SM 161 Tok_Until, -- UNTIL Eterm, Sterm, After_SM 162 Tok_When, -- WHEN Eterm, Sterm, After_SM 163 164 Tok_Begin, -- BEGIN Eterm, Sterm, After_SM, Labeled_Stmt 165 Tok_Declare, -- DECLARE Eterm, Sterm, After_SM, Labeled_Stmt 166 Tok_For, -- FOR Eterm, Sterm, After_SM, Labeled_Stmt 167 Tok_Loop, -- LOOP Eterm, Sterm, After_SM, Labeled_Stmt 168 Tok_While, -- WHILE Eterm, Sterm, After_SM, Labeled_Stmt 169 170 Tok_Entry, -- ENTRY Eterm, Sterm, Declk, Deckn, After_SM 171 Tok_Protected, -- PROTECTED Eterm, Sterm, Declk, Deckn, After_SM 172 Tok_Task, -- TASK Eterm, Sterm, Declk, Deckn, After_SM 173 Tok_Type, -- TYPE Eterm, Sterm, Declk, Deckn, After_SM 174 Tok_Subtype, -- SUBTYPE Eterm, Sterm, Declk, Deckn, After_SM 175 Tok_Overriding, -- OVERRIDING Eterm, Sterm, Declk, Declk, After_SM 176 Tok_Synchronized, -- SYNCHRONIZED Eterm, Sterm, Declk, Deckn, After_SM 177 Tok_Use, -- USE Eterm, Sterm, Declk, Deckn, After_SM 178 179 Tok_Function, -- FUNCTION Eterm, Sterm, Cunit, Declk, After_SM 180 Tok_Generic, -- GENERIC Eterm, Sterm, Cunit, Declk, After_SM 181 Tok_Package, -- PACKAGE Eterm, Sterm, Cunit, Declk, After_SM 182 Tok_Procedure, -- PROCEDURE Eterm, Sterm, Cunit, Declk, After_SM 183 184 Tok_Private, -- PRIVATE Eterm, Sterm, Cunit, After_SM 185 Tok_With, -- WITH Eterm, Sterm, Cunit, After_SM 186 Tok_Separate, -- SEPARATE Eterm, Sterm, Cunit, After_SM 187 188 Tok_EOF, -- End of file Eterm, Sterm, Cterm, After_SM 189 190 Tok_Semicolon, -- ; Eterm, Sterm, Cterm 191 192 Tok_Arrow, -- => Sterm, Cterm, Chtok 193 194 Tok_Vertical_Bar, -- | Cterm, Sterm, Chtok 195 196 Tok_Dot_Dot, -- .. Sterm, Chtok 197 198 Tok_Project, 199 Tok_Extends, 200 Tok_External, 201 Tok_External_As_List, 202 -- These four entries represent keywords for the project file language 203 -- and can be returned only in the case of scanning project files. 204 205 Tok_Comment, 206 -- This entry is used when scanning project files (where it represents 207 -- an entire comment), and in preprocessing with the -C switch set 208 -- (where it represents just the "--" of a comment). For the project 209 -- file case, the text of the comment is stored in Comment_Id. 210 211 Tok_End_Of_Line, 212 -- Represents an end of line. Not used during normal compilation scans 213 -- where end of line is ignored. Active for preprocessor scanning and 214 -- also when scanning project files (where it is needed because of ???) 215 216 Tok_Special, 217 -- AI12-0125-03 : target name as abbreviation for LHS 218 219 -- Otherwise used only in preprocessor scanning (to represent one of 220 -- the characters '#', '$', '?', '@', '`', '\', '^', '~', or '_'. The 221 -- character value itself is stored in Scans.Special_Character. 222 223 No_Token); 224 -- No_Token is used for initializing Token values to indicate that 225 -- no value has been set yet. 226 227 function Keyword_Name (Token : Token_Type) return Name_Id; 228 -- Given a token that is a reserved word, return the corresponding Name_Id 229 -- in lower case. E.g. Keyword_Name (Tok_Begin) = Name_Find ("begin"). 230 -- It is an error to pass any other kind of token. 231 232 -- Note: in the RM, operator symbol is a special case of string literal. 233 -- We distinguish at the lexical level in this compiler, since there are 234 -- many syntactic situations in which only an operator symbol is allowed. 235 236 -- The following subtype declarations group the token types into classes. 237 -- These are used for class tests in the parser. 238 239 subtype Token_Class_Numeric_Literal is 240 Token_Type range Tok_Integer_Literal .. Tok_Real_Literal; 241 -- Numeric literal 242 243 subtype Token_Class_Literal is 244 Token_Type range Tok_Integer_Literal .. Tok_Operator_Symbol; 245 -- Literal 246 247 subtype Token_Class_Lit_Or_Name is 248 Token_Type range Tok_Integer_Literal .. Tok_Identifier; 249 250 subtype Token_Class_Binary_Addop is 251 Token_Type range Tok_Ampersand .. Tok_Plus; 252 -- Binary adding operator (& + -) 253 254 subtype Token_Class_Unary_Addop is 255 Token_Type range Tok_Minus .. Tok_Plus; 256 -- Unary adding operator (+ -) 257 258 subtype Token_Class_Mulop is 259 Token_Type range Tok_Asterisk .. Tok_Slash; 260 -- Multiplying operator 261 262 subtype Token_Class_Logop is 263 Token_Type range Tok_And .. Tok_Xor; 264 -- Logical operator (and, or, xor) 265 266 subtype Token_Class_Relop is 267 Token_Type range Tok_Less .. Tok_Box; 268 -- Relational operator (= /= < <= > >= not, in plus <> to catch misuse 269 -- of Pascal style not equal operator). 270 271 subtype Token_Class_Name is 272 Token_Type range Tok_Char_Literal .. Tok_At_Sign; 273 -- First token of name (4.1), 274 -- (identifier, char literal, operator symbol) 275 -- Includes '@' after Ada2012 corrigendum. 276 277 subtype Token_Class_Desig is 278 Token_Type range Tok_Operator_Symbol .. Tok_At_Sign; 279 -- Token which can be a Designator (identifier, operator symbol) 280 281 subtype Token_Class_Namext is 282 Token_Type range Tok_Dot .. Tok_Left_Paren; 283 -- Name extension tokens. These are tokens which can appear immediately 284 -- after a name to extend it recursively (period, quote, left paren) 285 286 subtype Token_Class_Consk is 287 Token_Type range Tok_Left_Paren .. Tok_Range; 288 -- Keywords which can start constraint 289 -- (left paren, delta, digits, range) 290 291 subtype Token_Class_Eterm is 292 Token_Type range Tok_Colon_Equal .. Tok_Semicolon; 293 -- Expression terminators. These tokens can never appear within a simple 294 -- expression. This is used for error recovery purposes (if we encounter 295 -- an error in an expression, we simply scan to the next Eterm token). 296 297 subtype Token_Class_Sterm is 298 Token_Type range Tok_Delta .. Tok_Dot_Dot; 299 -- Simple_Expression terminators. A Simple_Expression must be followed 300 -- by a token in this class, or an error message is issued complaining 301 -- about a missing binary operator. 302 303 subtype Token_Class_Atkwd is 304 Token_Type range Tok_Delta .. Tok_Range; 305 -- Attribute keywords. This class includes keywords which can be used 306 -- as an Attribute_Designator, namely DELTA, DIGITS and RANGE 307 308 subtype Token_Class_Cterm is 309 Token_Type range Tok_EOF .. Tok_Vertical_Bar; 310 -- Choice terminators. These tokens terminate a choice. This is used for 311 -- error recovery purposes (if we encounter an error in a Choice, we 312 -- simply scan to the next Cterm token). 313 314 subtype Token_Class_Chtok is 315 Token_Type range Tok_Arrow .. Tok_Dot_Dot; 316 -- Choice tokens. These tokens signal a choice when used in an Aggregate 317 318 subtype Token_Class_Cunit is 319 Token_Type range Tok_Function .. Tok_Separate; 320 -- Tokens which can begin a compilation unit 321 322 subtype Token_Class_Declk is 323 Token_Type range Tok_Entry .. Tok_Procedure; 324 -- Keywords which start a declaration 325 326 subtype Token_Class_Deckn is 327 Token_Type range Tok_Entry .. Tok_Use; 328 -- Keywords which start a declaration but can't start a compilation unit 329 330 subtype Token_Class_After_SM is 331 Token_Type range Tok_Less_Less .. Tok_EOF; 332 -- Tokens which always, or almost always, appear after a semicolon. Used 333 -- in the Resync_Past_Semicolon routine to avoid gobbling up stuff when 334 -- a semicolon is missing. Of significance only for error recovery. 335 336 subtype Token_Class_Labeled_Stmt is 337 Token_Type range Tok_Begin .. Tok_While; 338 -- Tokens which start labeled statements 339 340 type Token_Flag_Array is array (Token_Type) of Boolean; 341 Is_Reserved_Keyword : constant Token_Flag_Array := 342 Token_Flag_Array' 343 (Tok_Mod .. Tok_Rem => True, 344 Tok_New .. Tok_Null => True, 345 Tok_Delta .. Tok_Range => True, 346 Tok_And .. Tok_Xor => True, 347 Tok_In .. Tok_Not => True, 348 Tok_Abstract .. Tok_Then => True, 349 Tok_Abort .. Tok_Separate => True, 350 others => False); 351 -- Flag array used to test for reserved word 352 353 procedure Initialize_Ada_Keywords; 354 -- Set up Token_Type values in Names table entries for Ada reserved 355 -- words. This ignores Ada_Version; Ada_Version is taken into account in 356 -- Snames.Is_Keyword_Name. 357 358 -------------------------- 359 -- Scan State Variables -- 360 -------------------------- 361 362 -- Note: these variables can only be referenced during the parsing of a 363 -- file. Reference to any of them from Sem or the expander is wrong. 364 365 -- These variables are initialized as required by Scn.Initialize_Scanner, 366 -- and should not be referenced before such a call. However, there are 367 -- situations in which these variables are saved and restored, and this 368 -- may happen before the first Initialize_Scanner call, resulting in the 369 -- assignment of invalid values. To avoid this, and allow building with 370 -- the -gnatVa switch, we initialize some variables to known valid values. 371 372 Scan_Ptr : Source_Ptr := No_Location; -- init for -gnatVa 373 -- Current scan pointer location. After a call to Scan, this points 374 -- just past the end of the token just scanned. 375 376 Token : Token_Type := No_Token; -- init for -gnatVa 377 -- Type of current token 378 379 Token_Ptr : Source_Ptr := No_Location; -- init for -gnatVa 380 -- Pointer to first character of current token 381 382 Current_Line_Start : Source_Ptr := No_Location; -- init for -gnatVa 383 -- Pointer to first character of line containing current token 384 385 Start_Column : Column_Number := No_Column_Number; -- init for -gnatVa 386 -- Starting column number (zero origin) of the first non-blank character 387 -- on the line containing the current token. This is used for error 388 -- recovery circuits which depend on looking at the column line up. 389 390 Type_Token_Location : Source_Ptr := No_Location; -- init for -gnatVa 391 -- Within a type declaration, gives the location of the TYPE keyword that 392 -- opened the type declaration. Used in checking the end column of a record 393 -- declaration, which can line up either with the TYPE keyword, or with the 394 -- start of the line containing the RECORD keyword. 395 396 Checksum : Word := 0; -- init for -gnatVa 397 -- Used to accumulate a CRC representing the tokens in the source 398 -- file being compiled. This CRC includes only program tokens, and 399 -- excludes comments. 400 401 Limited_Checksum : Word := 0; 402 -- Used to accumulate a CRC representing significant tokens in the 403 -- limited view of a package, i.e. visible type names and related 404 -- tagged indicators. 405 406 First_Non_Blank_Location : Source_Ptr := No_Location; -- init for -gnatVa 407 -- Location of first non-blank character on the line containing the 408 -- current token (i.e. the location of the character whose column number 409 -- is stored in Start_Column). 410 411 Token_Node : Node_Id := Empty; 412 -- Node table Id for the current token. This is set only if the current 413 -- token is one for which the scanner constructs a node (i.e. it is an 414 -- identifier, operator symbol, or literal). For other token types, 415 -- Token_Node is undefined. 416 417 Token_Name : Name_Id := No_Name; 418 -- For identifiers, this is set to the Name_Id of the identifier scanned. 419 -- For all other tokens, Token_Name is set to Error_Name. Note that it 420 -- would be possible for the caller to extract this information from 421 -- Token_Node. We set Token_Name separately for two reasons. First it 422 -- allows a quicker test for a specific identifier. Second, it allows 423 -- a version of the parser to be built that does not build tree nodes, 424 -- usable as a syntax checker. 425 426 Prev_Token : Token_Type := No_Token; 427 -- Type of previous token 428 429 Prev_Token_Ptr : Source_Ptr; 430 -- Pointer to first character of previous token 431 432 Version_To_Be_Found : Boolean; 433 -- This flag is True if the scanner is still looking for an RCS version 434 -- number in a comment. Normally it is initialized to False so that this 435 -- circuit is not activated. If the -dv switch is set, then this flag is 436 -- initialized to True, and then reset when the version number is found. 437 -- We do things this way to minimize the impact on comment scanning. 438 439 Character_Code : Char_Code; 440 -- Valid only when Token is Tok_Char_Literal. Contains the value of the 441 -- scanned literal. 442 443 Real_Literal_Value : Ureal; 444 -- Valid only when Token is Tok_Real_Literal. Contains the value of the 445 -- scanned literal. 446 447 Int_Literal_Value : Uint; 448 -- Valid only when Token = Tok_Integer_Literal, and we are not in 449 -- syntax-only mode. Contains the value of the scanned literal. 450 451 Based_Literal_Uses_Colon : Boolean; 452 -- Valid only when Token = Tok_Integer_Literal or Tok_Real_Literal. Set 453 -- True only for the case of a based literal using ':' instead of '#'. 454 455 String_Literal_Id : String_Id; 456 -- Valid only when Token = Tok_String_Literal or Tok_Operator_Symbol. 457 -- Contains the Id for currently scanned string value. 458 459 Wide_Character_Found : Boolean := False; 460 -- Valid only when Token = Tok_String_Literal. Set True if wide character 461 -- found (i.e. a character that does not fit in Character, but fits in 462 -- Wide_Wide_Character). 463 464 Wide_Wide_Character_Found : Boolean := False; 465 -- Valid only when Token = Tok_String_Literal. Set True if wide wide 466 -- character found (i.e. a character that does not fit in Character or 467 -- Wide_Character). 468 469 Special_Character : Character; 470 -- AI12-0125-03 : '@' as target name is handled elsewhere. 471 -- Valid only when Token = Tok_Special. Returns one of the characters 472 -- '#', '$', '?', '`', '\', '^', '~', or '_'. 473 -- 474 -- Why only this set? What about wide characters??? 475 476 Comment_Id : Name_Id := No_Name; 477 -- Valid only when Token = Tok_Comment. Store the string that follows 478 -- the "--" of a comment when scanning project files. 479 -- 480 -- Is it really right for this to be a Name rather than a String, what 481 -- about the case of Wide_Wide_Characters??? 482 483 Inside_Depends : Boolean := False; 484 -- True while parsing the argument of a Depends or Refined_Depends pragma 485 -- or aspect. Used to allow/require nonstandard style rules for =>+ with 486 -- -gnatyt. 487 488 Inside_If_Expression : Nat := 0; 489 -- This is a counter that is set non-zero while scanning out an if 490 -- expression (incremented on entry, decremented on exit). It is used to 491 -- disconnect format checks that normally apply to keywords THEN, ELSE etc. 492 493 Inside_Pragma : Boolean := False; 494 -- True within a pragma. Used to avoid complaining about reserved words 495 -- within pragmas (see Scan_Reserved_Identifier). 496 497 -------------------------------------------------------- 498 -- Procedures for Saving and Restoring the Scan State -- 499 -------------------------------------------------------- 500 501 -- The following procedures can be used to save and restore the entire 502 -- scan state. They are used in cases where it is necessary to backup 503 -- the scan during the parse. 504 505 type Saved_Scan_State is private; 506 -- Used for saving and restoring the scan state 507 508 procedure Save_Scan_State (Saved_State : out Saved_Scan_State); 509 pragma Inline (Save_Scan_State); 510 -- Saves the current scan state for possible later restoration. Note that 511 -- there is no harm in saving the state and then never restoring it. 512 513 procedure Restore_Scan_State (Saved_State : Saved_Scan_State); 514 pragma Inline (Restore_Scan_State); 515 -- Restores a scan state saved by a call to Save_Scan_State. 516 -- The saved scan state must refer to the current source file. 517 518private 519 type Saved_Scan_State is record 520 Save_Scan_Ptr : Source_Ptr; 521 Save_Token : Token_Type; 522 Save_Token_Ptr : Source_Ptr; 523 Save_Current_Line_Start : Source_Ptr; 524 Save_Start_Column : Column_Number; 525 Save_Checksum : Word; 526 Save_First_Non_Blank_Location : Source_Ptr; 527 Save_Token_Node : Node_Id; 528 Save_Token_Name : Name_Id; 529 Save_Prev_Token : Token_Type; 530 Save_Prev_Token_Ptr : Source_Ptr; 531 end record; 532 533end Scans; 534