1------------------------------------------------------------------------------ 2-- -- 3-- GNAT COMPILER COMPONENTS -- 4-- -- 5-- P A R -- 6-- -- 7-- B o d y -- 8-- -- 9-- Copyright (C) 1992-2020, 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 Aspects; use Aspects; 27with Atree; use Atree; 28with Casing; use Casing; 29with Debug; use Debug; 30with Elists; use Elists; 31with Errout; use Errout; 32with Fname; use Fname; 33with Lib; use Lib; 34with Namet; use Namet; 35with Namet.Sp; use Namet.Sp; 36with Nlists; use Nlists; 37with Nmake; use Nmake; 38with Opt; use Opt; 39with Output; use Output; 40with Par_SCO; use Par_SCO; 41with Restrict; use Restrict; 42with Scans; use Scans; 43with Scn; use Scn; 44with Sem_Util; use Sem_Util; 45with Sinput; use Sinput; 46with Sinput.L; use Sinput.L; 47with Sinfo; use Sinfo; 48with Snames; use Snames; 49with Style; 50with Stylesw; use Stylesw; 51with Table; 52with Tbuild; use Tbuild; 53 54--------- 55-- Par -- 56--------- 57 58function Par (Configuration_Pragmas : Boolean) return List_Id is 59 60 Inside_Record_Definition : Boolean := False; 61 -- True within a record definition. Used to control warning for 62 -- redefinition of standard entities (not issued for field names). 63 64 Loop_Block_Count : Nat := 0; 65 -- Counter used for constructing loop/block names (see the routine 66 -- Par.Ch5.Get_Loop_Block_Name). 67 68 Num_Library_Units : Natural := 0; 69 -- Count number of units parsed (relevant only in syntax check only mode, 70 -- since in semantics check mode only a single unit is permitted anyway). 71 72 Save_Config_Attrs : Config_Switches_Type; 73 -- Variable used to save values of config switches while we parse the 74 -- new unit, to be restored on exit for proper recursive behavior. 75 76 -------------------- 77 -- Error Recovery -- 78 -------------------- 79 80 -- When an error is encountered, a call is made to one of the Error_Msg 81 -- routines to record the error. If the syntax scan is not derailed by the 82 -- error (e.g. a complaint that logical operators are inconsistent in an 83 -- EXPRESSION), then control returns from the Error_Msg call, and the 84 -- parse continues unimpeded. 85 86 -- If on the other hand, the Error_Msg represents a situation from which 87 -- the parser cannot recover locally, the exception Error_Resync is raised 88 -- immediately after the call to Error_Msg. Handlers for Error_Resync 89 -- are located at strategic points to resynchronize the parse. For example, 90 -- when an error occurs in a statement, the handler skips to the next 91 -- semicolon and continues the scan from there. 92 93 -- Each parsing procedure contains a note with the heading "Error recovery" 94 -- which shows if it can propagate the Error_Resync exception. In order 95 -- not to propagate the exception, a procedure must either contain its own 96 -- handler for this exception, or it must not call any other routines which 97 -- propagate the exception. 98 99 -- Note: the arrangement of Error_Resync handlers is such that it should 100 -- never be possible to transfer control through a procedure which made 101 -- an entry in the scope stack, invalidating the contents of the stack. 102 103 Error_Resync : exception; 104 -- Exception raised on error that is not handled locally, see above 105 106 Last_Resync_Point : Source_Ptr; 107 -- The resynchronization routines in Par.Sync run a risk of getting 108 -- stuck in an infinite loop if they do not skip a token, and the caller 109 -- keeps repeating the same resync call. On the other hand, if they skip 110 -- a token unconditionally, some recovery opportunities are missed. The 111 -- variable Last_Resync_Point records the token location previously set 112 -- by a Resync call, and if a subsequent Resync call occurs at the same 113 -- location, then the Resync routine does guarantee to skip a token. 114 115 -------------------------------------------- 116 -- Handling Semicolon Used in Place of IS -- 117 -------------------------------------------- 118 119 -- The following global variables are used in handling the error situation 120 -- of using a semicolon in place of IS in a subprogram declaration as in: 121 122 -- procedure X (Y : Integer); 123 -- Q : Integer; 124 -- begin 125 -- ... 126 -- end; 127 128 -- The two contexts in which this can appear are at the outer level, and 129 -- within a declarative region. At the outer level, we know something is 130 -- wrong as soon as we see the Q (or begin, if there are no declarations), 131 -- and we can immediately decide that the semicolon should have been IS. 132 133 -- The situation in a declarative region is more complex. The declaration 134 -- of Q could belong to the outer region, and we do not know that we have 135 -- an error until we hit the begin. It is still not clear at this point 136 -- from a syntactic point of view that something is wrong, because the 137 -- begin could belong to the enclosing subprogram or package. However, we 138 -- can incorporate a bit of semantic knowledge and note that the body of 139 -- X is missing, so we definitely DO have an error. We diagnose this error 140 -- as semicolon in place of IS on the subprogram line. 141 142 -- There are two styles for this diagnostic. If the begin immediately 143 -- follows the semicolon, then we can place a flag (IS expected) right 144 -- on the semicolon. Otherwise we do not detect the error until we hit 145 -- the begin which refers back to the line with the semicolon. 146 147 -- To control the process in the second case, the following global 148 -- variables are set to indicate that we have a subprogram declaration 149 -- whose body is required and has not yet been found. The prefix SIS 150 -- stands for "Subprogram IS" handling. 151 152 SIS_Entry_Active : Boolean := False; 153 -- Set True to indicate that an entry is active (i.e. that a subprogram 154 -- declaration has been encountered, and no body for this subprogram 155 -- has been encountered). The remaining variables other than 156 -- SIS_Aspect_Import_Seen are valid only if this is True. 157 158 SIS_Aspect_Import_Seen : Boolean := False; 159 -- If this is True when a subprogram declaration has been encountered, we 160 -- do not set SIS_Entry_Active, because the Import means there is no body. 161 -- Set False at the start of P_Subprogram, set True when an Import aspect 162 -- specification is seen, and used when P_Subprogram finds a subprogram 163 -- declaration. This is necessary because the aspects are parsed before 164 -- we know we have a subprogram declaration. 165 166 SIS_Labl : Node_Id; 167 -- Subprogram designator 168 169 SIS_Sloc : Source_Ptr; 170 -- Source location of FUNCTION/PROCEDURE keyword 171 172 SIS_Ecol : Column_Number; 173 -- Column number of FUNCTION/PROCEDURE keyword 174 175 SIS_Semicolon_Sloc : Source_Ptr; 176 -- Source location of semicolon at end of subprogram declaration 177 178 SIS_Declaration_Node : Node_Id; 179 -- Pointer to tree node for subprogram declaration 180 181 SIS_Missing_Semicolon_Message : Error_Msg_Id; 182 -- Used to save message ID of missing semicolon message (which will be 183 -- modified to missing IS if necessary). Set to No_Error_Msg in the 184 -- normal (non-error) case. 185 186 -- Five things can happen to an active SIS entry 187 188 -- 1. If a BEGIN is encountered with an SIS entry active, then we have 189 -- exactly the situation in which we know the body of the subprogram is 190 -- missing. After posting an error message, we change the spec to a body, 191 -- rechaining the declarations that intervened between the spec and BEGIN. 192 193 -- 2. Another subprogram declaration or body is encountered. In this 194 -- case the entry gets overwritten with the information for the new 195 -- subprogram declaration. We don't catch some nested cases this way, 196 -- but it doesn't seem worth the effort. 197 198 -- 3. A nested declarative region (e.g. package declaration or package 199 -- body) is encountered. The SIS active indication is reset at the start 200 -- of such a nested region. Again, like case 2, this causes us to miss 201 -- some nested cases, but it doesn't seen worth the effort to stack and 202 -- unstack the SIS information. Maybe we will reconsider this if we ever 203 -- get a complaint about a missed case. 204 205 -- 4. We encounter a valid pragma INTERFACE or IMPORT that effectively 206 -- supplies the missing body. In this case we reset the entry. 207 208 -- 5. We encounter the end of the declarative region without encountering 209 -- a BEGIN first. In this situation we simply reset the entry. We know 210 -- that there is a missing body, but it seems more reasonable to let the 211 -- later semantic checking discover this. 212 213 ---------------------------------------------------- 214 -- Handling of Reserved Words Used as Identifiers -- 215 ---------------------------------------------------- 216 217 -- Note: throughout the parser, the terms reserved word and keyword are 218 -- used interchangeably to refer to the same set of reserved keywords 219 -- (including until, protected, etc). 220 221 -- If a reserved word is used in place of an identifier, the parser where 222 -- possible tries to recover gracefully. In particular, if the keyword is 223 -- clearly spelled using identifier casing, e.g. Until in a source program 224 -- using mixed case identifiers and lower case keywords, then the keyword 225 -- is treated as an identifier if it appears in a place where an identifier 226 -- is required. 227 228 -- The situation is more complex if the keyword is spelled with normal 229 -- keyword casing. In this case, the parser is more reluctant to consider 230 -- it to be intended as an identifier, unless it has some further 231 -- confirmation. 232 233 -- In the case of an identifier appearing in the identifier list of a 234 -- declaration, the appearance of a comma or colon right after the keyword 235 -- on the same line is taken as confirmation. For an enumeration literal, 236 -- a comma or right paren right after the identifier is also treated as 237 -- adequate confirmation. 238 239 -- The following type is used in calls to Is_Reserved_Identifier and 240 -- also to P_Defining_Identifier and P_Identifier. The default for all 241 -- these functions is that reserved words in reserved word case are not 242 -- considered to be reserved identifiers. The Id_Check value indicates 243 -- tokens, which if they appear immediately after the identifier, are 244 -- taken as confirming that the use of an identifier was expected 245 246 type Id_Check is 247 (None, 248 -- Default, no special token test 249 250 C_Comma_Right_Paren, 251 -- Consider as identifier if followed by comma or right paren 252 253 C_Comma_Colon, 254 -- Consider as identifier if followed by comma or colon 255 256 C_Do, 257 -- Consider as identifier if followed by DO 258 259 C_Dot, 260 -- Consider as identifier if followed by period 261 262 C_Greater_Greater, 263 -- Consider as identifier if followed by >> 264 265 C_In, 266 -- Consider as identifier if followed by IN 267 268 C_Is, 269 -- Consider as identifier if followed by IS 270 271 C_Left_Paren_Semicolon, 272 -- Consider as identifier if followed by left paren or semicolon 273 274 C_Use, 275 -- Consider as identifier if followed by USE 276 277 C_Vertical_Bar_Arrow); 278 -- Consider as identifier if followed by | or => 279 280 -------------------------------------------- 281 -- Handling IS Used in Place of Semicolon -- 282 -------------------------------------------- 283 284 -- This is a somewhat trickier situation, and we can't catch it in all 285 -- cases, but we do our best to detect common situations resulting from 286 -- a "cut and paste" operation which forgets to change the IS to semicolon. 287 -- Consider the following example: 288 289 -- package body X is 290 -- procedure A; 291 -- procedure B is 292 -- procedure C; 293 -- ... 294 -- procedure D is 295 -- begin 296 -- ... 297 -- end; 298 -- begin 299 -- ... 300 -- end; 301 302 -- The trouble is that the section of text from PROCEDURE B through END; 303 -- constitutes a valid procedure body, and the danger is that we find out 304 -- far too late that something is wrong (indeed most compilers will behave 305 -- uncomfortably on the above example). 306 307 -- We have two approaches to helping to control this situation. First we 308 -- make every attempt to avoid swallowing the last END; if we can be sure 309 -- that some error will result from doing so. In particular, we won't 310 -- accept the END; unless it is exactly correct (in particular it must not 311 -- have incorrect name tokens), and we won't accept it if it is immediately 312 -- followed by end of file, WITH or SEPARATE (all tokens that unmistakeably 313 -- signal the start of a compilation unit, and which therefore allow us to 314 -- reserve the END; for the outer level.) For more details on this aspect 315 -- of the handling, see package Par.Endh. 316 317 -- If we can avoid eating up the END; then the result in the absence of 318 -- any additional steps would be to post a missing END referring back to 319 -- the subprogram with the bogus IS. Similarly, if the enclosing package 320 -- has no BEGIN, then the result is a missing BEGIN message, which again 321 -- refers back to the subprogram header. 322 323 -- Such an error message is not too bad (it's already a big improvement 324 -- over what many parsers do), but it's not ideal, because the declarations 325 -- following the IS have been absorbed into the wrong scope. In the above 326 -- case, this could result for example in a bogus complaint that the body 327 -- of D was missing from the package. 328 329 -- To catch at least some of these cases, we take the following additional 330 -- steps. First, a subprogram body is marked as having a suspicious IS if 331 -- the declaration line is followed by a line which starts with a symbol 332 -- that can start a declaration in the same column, or to the left of the 333 -- column in which the FUNCTION or PROCEDURE starts (normal style is to 334 -- indent any declarations which really belong a subprogram). If such a 335 -- subprogram encounters a missing BEGIN or missing END, then we decide 336 -- that the IS should have been a semicolon, and the subprogram body node 337 -- is marked (by setting the Bad_Is_Detected flag true. Note that we do 338 -- not do this for library level procedures, only for nested procedures, 339 -- since for library level procedures, we must have a body. 340 341 -- The processing for a declarative part checks to see if the last 342 -- declaration scanned is marked in this way, and if it is, the tree 343 -- is modified to reflect the IS being interpreted as a semicolon. 344 345 --------------------------------------------------- 346 -- Parser Type Definitions and Control Variables -- 347 --------------------------------------------------- 348 349 -- The following variable and associated type declaration are used by the 350 -- expression parsing routines to return more detailed information about 351 -- the categorization of a parsed expression. 352 353 type Expr_Form_Type is ( 354 EF_Simple_Name, -- Simple name, i.e. possibly qualified identifier 355 EF_Name, -- Simple expression which could also be a name 356 EF_Simple, -- Simple expression which is not call or name 357 EF_Range_Attr, -- Range attribute reference 358 EF_Non_Simple); -- Expression that is not a simple expression 359 360 Expr_Form : Expr_Form_Type; 361 362 -- The following type is used for calls to P_Subprogram, P_Package, P_Task, 363 -- P_Protected to indicate which of several possibilities is acceptable. 364 365 type Pf_Rec is record 366 Spcn : Boolean; -- True if specification OK 367 Decl : Boolean; -- True if declaration OK 368 Gins : Boolean; -- True if generic instantiation OK 369 Pbod : Boolean; -- True if proper body OK 370 Rnam : Boolean; -- True if renaming declaration OK 371 Stub : Boolean; -- True if body stub OK 372 Pexp : Boolean; -- True if parameterized expression OK 373 Fil2 : Boolean; -- Filler to fill to 8 bits 374 end record; 375 pragma Pack (Pf_Rec); 376 377 function T return Boolean renames True; 378 function F return Boolean renames False; 379 380 Pf_Decl_Gins_Pbod_Rnam_Stub_Pexp : constant Pf_Rec := 381 Pf_Rec'(F, T, T, T, T, T, T, F); 382 Pf_Decl_Pexp : constant Pf_Rec := 383 Pf_Rec'(F, T, F, F, F, F, T, F); 384 Pf_Decl_Gins_Pbod_Rnam_Pexp : constant Pf_Rec := 385 Pf_Rec'(F, T, T, T, T, F, T, F); 386 Pf_Decl_Pbod_Pexp : constant Pf_Rec := 387 Pf_Rec'(F, T, F, T, F, F, T, F); 388 Pf_Pbod_Pexp : constant Pf_Rec := 389 Pf_Rec'(F, F, F, T, F, F, T, F); 390 Pf_Spcn : constant Pf_Rec := 391 Pf_Rec'(T, F, F, F, F, F, F, F); 392 -- The above are the only allowed values of Pf_Rec arguments 393 394 type SS_Rec is record 395 Eftm : Boolean; -- ELSIF can terminate sequence 396 Eltm : Boolean; -- ELSE can terminate sequence 397 Extm : Boolean; -- EXCEPTION can terminate sequence 398 Ortm : Boolean; -- OR can terminate sequence 399 Sreq : Boolean; -- at least one statement required 400 Tatm : Boolean; -- THEN ABORT can terminate sequence 401 Whtm : Boolean; -- WHEN can terminate sequence 402 Unco : Boolean; -- Unconditional terminate after one statement 403 end record; 404 pragma Pack (SS_Rec); 405 406 SS_Eftm_Eltm_Sreq : constant SS_Rec := SS_Rec'(T, T, F, F, T, F, F, F); 407 SS_Eltm_Ortm_Tatm : constant SS_Rec := SS_Rec'(F, T, F, T, F, T, F, F); 408 SS_Extm_Sreq : constant SS_Rec := SS_Rec'(F, F, T, F, T, F, F, F); 409 SS_None : constant SS_Rec := SS_Rec'(F, F, F, F, F, F, F, F); 410 SS_Ortm_Sreq : constant SS_Rec := SS_Rec'(F, F, F, T, T, F, F, F); 411 SS_Sreq : constant SS_Rec := SS_Rec'(F, F, F, F, T, F, F, F); 412 SS_Sreq_Whtm : constant SS_Rec := SS_Rec'(F, F, F, F, T, F, T, F); 413 SS_Whtm : constant SS_Rec := SS_Rec'(F, F, F, F, F, F, T, F); 414 SS_Unco : constant SS_Rec := SS_Rec'(F, F, F, F, F, F, F, T); 415 416 Goto_List : Elist_Id; 417 -- List of goto nodes appearing in the current compilation. Used to 418 -- recognize natural loops and convert them into bona fide loops for 419 -- optimization purposes. 420 421 Label_List : Elist_Id; 422 -- List of label nodes for labels appearing in the current compilation. 423 -- Used by Par.Labl to construct the corresponding implicit declarations. 424 425 ----------------- 426 -- Scope Table -- 427 ----------------- 428 429 -- The scope table, also referred to as the scope stack, is used to record 430 -- the current scope context. It is organized as a stack, with inner nested 431 -- entries corresponding to higher entries on the stack. An entry is made 432 -- when the parser encounters the opening of a nested construct (such as a 433 -- record, task, package etc.), and then package Par.Endh uses this stack 434 -- to deal with END lines (including properly dealing with END nesting 435 -- errors). 436 437 type SS_End_Type is 438 -- Type of end entry required for this scope. The last two entries are 439 -- used only in the subprogram body case to mark the case of a suspicious 440 -- IS, or a bad IS (i.e. suspicions confirmed by missing BEGIN or END). 441 -- See separate section on dealing with IS used in place of semicolon. 442 -- Note that for many purposes E_Name, E_Suspicious_Is and E_Bad_Is are 443 -- treated the same (E_Suspicious_Is and E_Bad_Is are simply special cases 444 -- of E_Name). They are placed at the end of the enumeration so that a 445 -- test for >= E_Name catches all three cases efficiently. 446 447 (E_Dummy, -- dummy entry at outer level 448 E_Case, -- END CASE; 449 E_If, -- END IF; 450 E_Loop, -- END LOOP; 451 E_Record, -- END RECORD; 452 E_Return, -- END RETURN; 453 E_Select, -- END SELECT; 454 E_Name, -- END [name]; 455 E_Suspicious_Is, -- END [name]; (case of suspicious IS) 456 E_Bad_Is); -- END [name]; (case of bad IS) 457 458 -- The following describes a single entry in the scope table 459 460 type Scope_Table_Entry is record 461 Etyp : SS_End_Type; 462 -- Type of end entry, as per above description 463 464 Lreq : Boolean; 465 -- A flag indicating whether the label, if present, is required to 466 -- appear on the end line. It is referenced only in the case of Etyp is 467 -- equal to E_Name or E_Suspicious_Is where the name may or may not be 468 -- required (yes for labeled block, no in other cases). Note that for 469 -- all cases except begin, the question of whether a label is required 470 -- can be determined from the other fields (for loop, it is required if 471 -- it is present, and for the other constructs it is never required or 472 -- allowed). 473 474 Ecol : Column_Number; 475 -- Contains the absolute column number (with tabs expanded) of the 476 -- expected column of the end assuming normal Ada indentation usage. If 477 -- the RM_Column_Check mode is set, this value is used for generating 478 -- error messages about indentation. Otherwise it is used only to 479 -- control heuristic error recovery actions. This value is zero origin. 480 481 Labl : Node_Id; 482 -- This field is used to provide the name of the construct being parsed 483 -- and indirectly its kind. For loops and blocks, the field contains the 484 -- source name or the generated one. For package specifications, bodies, 485 -- subprogram specifications and bodies the field holds the correponding 486 -- program unit name. For task declarations and bodies, protected types 487 -- and bodies, and accept statements the field hold the name of the type 488 -- or operation. For if-statements, case-statements, return statements, 489 -- and selects, the field is initialized to Error. 490 491 -- Note: this is a bit of an odd (mis)use of Error, since there is no 492 -- Error, but we use this value as a place holder to indicate that it 493 -- is an error to have a label on the end line. 494 495 -- Whenever the field is a name, it is attached to the parent node of 496 -- the construct being parsed. Thus the parent node indicates the kind 497 -- of construct whose parse tree is being built. This is used in error 498 -- recovery. 499 500 Decl : List_Id; 501 -- Points to the list of declarations (i.e. the declarative part) 502 -- associated with this construct. It is set only in the END [name] 503 -- cases, and is set to No_List for all other cases which do not have a 504 -- declarative unit associated with them. This is used for determining 505 -- the proper location for implicit label declarations. 506 507 Node : Node_Id; 508 -- Empty except in the case of entries for IF and CASE statements, in 509 -- which case it contains the N_If_Statement or N_Case_Statement node. 510 -- This is used for setting the End_Span field. 511 512 Sloc : Source_Ptr; 513 -- Source location of the opening token of the construct. This is used 514 -- to refer back to this line in error messages (such as missing or 515 -- incorrect end lines). The Sloc field is not used, and is not set, if 516 -- a label is present (the Labl field provides the text name of the 517 -- label in this case, which is fine for error messages). 518 519 S_Is : Source_Ptr; 520 -- S_Is is relevant only if Etyp is set to E_Suspicious_Is or E_Bad_Is. 521 -- It records the location of the IS that is considered to be 522 -- suspicious. 523 524 Junk : Boolean; 525 -- A boolean flag that is set true if the opening entry is the dubious 526 -- result of some prior error, e.g. a record entry where the record 527 -- keyword was missing. It is used to suppress the issuing of a 528 -- corresponding junk complaint about the end line (we do not want 529 -- to complain about a missing end record when there was no record). 530 end record; 531 532 -- The following declares the scope table itself. The Last field is the 533 -- stack pointer, so that Scope.Table (Scope.Last) is the top entry. The 534 -- oldest entry, at Scope_Stack (0), is a dummy entry with Etyp set to 535 -- E_Dummy, and the other fields undefined. This dummy entry ensures that 536 -- Scope_Stack (Scope_Stack_Ptr).Etyp can always be tested, and that the 537 -- scope stack pointer is always in range. 538 539 package Scope is new Table.Table ( 540 Table_Component_Type => Scope_Table_Entry, 541 Table_Index_Type => Int, 542 Table_Low_Bound => 0, 543 Table_Initial => 50, 544 Table_Increment => 100, 545 Table_Name => "Scope"); 546 547 type Scope_Table_Entry_Ptr is access all Scope_Table_Entry; 548 549 function Scopes (Index : Int) return Scope_Table_Entry_Ptr; 550 -- Return the indicated Scope_Table_Entry. We use a pointer for 551 -- efficiency. Callers should not save the pointer, but should do things 552 -- like Scopes (Scope.Last).Something. Note that there is one place in 553 -- Par.Ch5 that indexes the stack out of bounds, and can't call this. 554 555 function Scopes (Index : Int) return Scope_Table_Entry_Ptr is 556 begin 557 pragma Assert (Index in Scope.First .. Scope.Last); 558 return Scope.Table (Index)'Unrestricted_Access; 559 end Scopes; 560 561 ------------------------------------------ 562 -- Table for Handling Suspicious Labels -- 563 ------------------------------------------ 564 565 -- This is a special data structure which is used to deal very spefifically 566 -- with the following error case 567 568 -- label; 569 -- loop 570 -- ... 571 -- end loop label; 572 573 -- Similar cases apply to FOR, WHILE, DECLARE, or BEGIN 574 575 -- In each case the opening line looks like a procedure call because of 576 -- the semicolon. And the end line looks illegal because of an unexpected 577 -- label. If we did nothing special, we would just diagnose the label on 578 -- the end as unexpected. But that does not help point to the real error 579 -- which is that the semicolon after label should be a colon. 580 581 -- To deal with this, we build an entry in the Suspicious_Labels table 582 -- whenever we encounter an identifier followed by a semicolon, followed 583 -- by one of LOOP, FOR, WHILE, DECLARE, BEGIN. Then this entry is used to 584 -- issue the right message when we hit the END that confirms that this was 585 -- a bad label. 586 587 type Suspicious_Label_Entry is record 588 Proc_Call : Node_Id; 589 -- Node for the procedure call statement built for the label; construct 590 591 Semicolon_Loc : Source_Ptr; 592 -- Location of the possibly wrong semicolon 593 594 Start_Token : Source_Ptr; 595 -- Source location of the LOOP, FOR, WHILE, DECLARE, BEGIN token 596 end record; 597 598 package Suspicious_Labels is new Table.Table ( 599 Table_Component_Type => Suspicious_Label_Entry, 600 Table_Index_Type => Int, 601 Table_Low_Bound => 1, 602 Table_Initial => 50, 603 Table_Increment => 100, 604 Table_Name => "Suspicious_Labels"); 605 606 -- Now when we are about to issue a message complaining about an END label 607 -- that should not be there because it appears to end a construct that has 608 -- no label, we first search the suspicious labels table entry, using the 609 -- source location stored in the scope table as a key. If we find a match, 610 -- then we check that the label on the end matches the name in the call, 611 -- and if so, we issue a message saying the semicolon should be a colon. 612 613 -- Quite a bit of work, but really helpful in the case where it helps, and 614 -- the need for this is based on actual experience with tracking down this 615 -- kind of error (the eye often easily mistakes semicolon for colon). 616 617 -- Note: we actually have enough information to patch up the tree, but 618 -- this may not be worth the effort. Also we could deal with the same 619 -- situation for EXIT with a label, but for now don't bother with that. 620 621 Current_Assign_Node : Node_Id := Empty; 622 -- This is the node of the current assignment statement being compiled. 623 -- It is used to record the presence of target_names on its RHS. This 624 -- context-dependent trick simplifies the analysis of such nodes, where 625 -- the RHS must first be analyzed with expansion disabled. 626 627 --------------------------------- 628 -- Parsing Routines by Chapter -- 629 --------------------------------- 630 631 -- Uncommented declarations in this section simply parse the construct 632 -- corresponding to their name, and return an ID value for the Node or 633 -- List that is created. 634 635 ------------- 636 -- Par.Ch2 -- 637 ------------- 638 639 package Ch2 is 640 function P_Pragma (Skipping : Boolean := False) return Node_Id; 641 -- Scan out a pragma. If Skipping is True, then the caller is skipping 642 -- the pragma in the context of illegal placement (this is used to avoid 643 -- some junk cascaded messages). Some pragmas must be dealt with during 644 -- the parsing phase (e.g. pragma Page, since we can generate a listing 645 -- in syntax only mode). It is possible that the parser uses the rescan 646 -- logic (using Save/Restore_Scan_State) with the effect of calling this 647 -- procedure more than once for the same pragma. All parse-time pragma 648 -- handling must be prepared to handle such multiple calls correctly. 649 650 function P_Identifier (C : Id_Check := None) return Node_Id; 651 -- Scans out an identifier. The parameter C determines the treatment 652 -- of reserved identifiers. See declaration of Id_Check for details. 653 654 function P_Pragmas_Opt return List_Id; 655 -- This function scans for a sequence of pragmas in other than a 656 -- declaration sequence or statement sequence context. All pragmas 657 -- can appear except pragmas Assert and Debug, which are only allowed 658 -- in a declaration or statement sequence context. 659 660 procedure P_Pragmas_Misplaced; 661 -- Skips misplaced pragmas with a complaint 662 663 procedure P_Pragmas_Opt (List : List_Id); 664 -- Parses optional pragmas and appends them to the List 665 end Ch2; 666 667 ------------- 668 -- Par.Ch3 -- 669 ------------- 670 671 package Ch3 is 672 Missing_Begin_Msg : Error_Msg_Id; 673 -- This variable is set by a call to P_Declarative_Part. Normally it 674 -- is set to No_Error_Msg, indicating that no special processing is 675 -- required by the caller. The special case arises when a statement 676 -- is found in the sequence of declarations. In this case the Id of 677 -- the message issued ("declaration expected") is preserved in this 678 -- variable, then the caller can change it to an appropriate missing 679 -- begin message if indeed the BEGIN is missing. 680 681 function P_Array_Type_Definition return Node_Id; 682 function P_Constraint_Opt return Node_Id; 683 function P_Declarative_Part return List_Id; 684 function P_Discrete_Choice_List return List_Id; 685 function P_Discrete_Range return Node_Id; 686 function P_Discrete_Subtype_Definition return Node_Id; 687 function P_Known_Discriminant_Part_Opt return List_Id; 688 function P_Signed_Integer_Type_Definition return Node_Id; 689 function P_Range return Node_Id; 690 function P_Range_Constraint return Node_Id; 691 function P_Record_Definition return Node_Id; 692 function P_Subtype_Mark return Node_Id; 693 function P_Subtype_Mark_Resync return Node_Id; 694 function P_Unknown_Discriminant_Part_Opt return Boolean; 695 696 function P_Basic_Declarative_Items 697 (Declare_Expression : Boolean) return List_Id; 698 -- Used to parse the declarative items in a package visible or 699 -- private part (in which case Declare_Expression is False), and 700 -- the declare_items of a declare_expression (in which case 701 -- Declare_Expression is True). Declare_Expression is used to 702 -- affect the wording of error messages, and to control style 703 -- checking. 704 705 function P_Access_Definition 706 (Null_Exclusion_Present : Boolean) return Node_Id; 707 -- Ada 2005 (AI-231/AI-254): The caller parses the null-exclusion part 708 -- and indicates if it was present 709 710 function P_Access_Type_Definition 711 (Header_Already_Parsed : Boolean := False) return Node_Id; 712 -- Ada 2005 (AI-254): The formal is used to indicate if the caller has 713 -- parsed the null_exclusion part. In this case the caller has also 714 -- removed the ACCESS token 715 716 procedure P_Component_Items (Decls : List_Id); 717 -- Scan out one or more component items and append them to the given 718 -- list. Only scans out more than one declaration in the case where the 719 -- source has a single declaration with multiple defining identifiers. 720 721 function P_Defining_Identifier (C : Id_Check := None) return Node_Id; 722 -- Scan out a defining identifier. The parameter C controls the 723 -- treatment of errors in case a reserved word is scanned. See the 724 -- declaration of this type for details. 725 726 function P_Interface_Type_Definition 727 (Abstract_Present : Boolean) return Node_Id; 728 -- Ada 2005 (AI-251): Parse the interface type definition part. Abstract 729 -- Present indicates if the reserved word "abstract" has been previously 730 -- found. It is used to report an error message because interface types 731 -- are by definition abstract tagged. We generate a record_definition 732 -- node if the list of interfaces is empty; otherwise we generate a 733 -- derived_type_definition node (the first interface in this list is the 734 -- ancestor interface). 735 736 function P_Null_Exclusion 737 (Allow_Anonymous_In_95 : Boolean := False) return Boolean; 738 -- Ada 2005 (AI-231): Parse the null-excluding part. A True result 739 -- indicates that the null-excluding part was present. 740 -- 741 -- Allow_Anonymous_In_95 is True if we are in a context that allows 742 -- anonymous access types in Ada 95, in which case "not null" is legal 743 -- if it precedes "access". 744 745 function P_Subtype_Indication 746 (Not_Null_Present : Boolean := False) return Node_Id; 747 -- Ada 2005 (AI-231): The flag Not_Null_Present indicates that the 748 -- null-excluding part has been scanned out and it was present. 749 750 function P_Range_Or_Subtype_Mark 751 (Allow_Simple_Expression : Boolean := False) return Node_Id; 752 -- Scans out a range or subtype mark, and also permits a general simple 753 -- expression if Allow_Simple_Expression is set to True. 754 755 function Init_Expr_Opt (P : Boolean := False) return Node_Id; 756 -- If an initialization expression is present (:= expression), then 757 -- it is scanned out and returned, otherwise Empty is returned if no 758 -- initialization expression is present. This procedure also handles 759 -- certain common error cases cleanly. The parameter P indicates if 760 -- a right paren can follow the expression (default = no right paren 761 -- allowed). 762 763 procedure Skip_Declaration (S : List_Id); 764 -- Used when scanning statements to skip past a misplaced declaration 765 -- The declaration is scanned out and appended to the given list. 766 -- Token is known to be a declaration token (in Token_Class_Declk) 767 -- on entry, so there definition is a declaration to be scanned. 768 769 function P_Subtype_Indication 770 (Subtype_Mark : Node_Id; 771 Not_Null_Present : Boolean := False) return Node_Id; 772 -- This version of P_Subtype_Indication is called when the caller has 773 -- already scanned out the subtype mark which is passed as a parameter. 774 -- Ada 2005 (AI-231): The flag Not_Null_Present indicates that the 775 -- null-excluding part has been scanned out and it was present. 776 777 function P_Subtype_Mark_Attribute (Type_Node : Node_Id) return Node_Id; 778 -- Parse a subtype mark attribute. The caller has already parsed the 779 -- subtype mark, which is passed in as the argument, and has checked 780 -- that the current token is apostrophe. 781 end Ch3; 782 783 ------------- 784 -- Par.Ch4 -- 785 ------------- 786 787 package Ch4 is 788 function P_Aggregate return Node_Id; 789 function P_Expression return Node_Id; 790 function P_Expression_Or_Range_Attribute return Node_Id; 791 function P_Function_Name return Node_Id; 792 function P_Name return Node_Id; 793 function P_Qualified_Simple_Name return Node_Id; 794 function P_Qualified_Simple_Name_Resync return Node_Id; 795 function P_Simple_Expression return Node_Id; 796 function P_Simple_Expression_Or_Range_Attribute return Node_Id; 797 798 function P_Expression_If_OK return Node_Id; 799 -- Scans out an expression allowing an unparenthesized case expression, 800 -- if expression, or quantified expression to appear without enclosing 801 -- parentheses. However, if such an expression is not preceded by a left 802 -- paren, and followed by a right paren, an error message will be output 803 -- noting that parenthesization is required. 804 805 function P_Expression_No_Right_Paren return Node_Id; 806 -- Scans out an expression in contexts where the expression cannot be 807 -- terminated by a right paren (gives better error recovery if an errant 808 -- right paren is found after the expression). 809 810 function P_Expression_Or_Range_Attribute_If_OK return Node_Id; 811 -- Scans out an expression or range attribute where a conditional 812 -- expression is permitted to appear without surrounding parentheses. 813 -- However, if such an expression is not preceded by a left paren, and 814 -- followed by a right paren, an error message will be output noting 815 -- that parenthesization is required. 816 817 function P_If_Expression return Node_Id; 818 -- Scans out an if expression. Called with Token pointing to the 819 -- IF keyword, and returns pointing to the terminating right paren, 820 -- semicolon or comma, but does not consume this terminating token. 821 822 function P_Qualified_Expression (Subtype_Mark : Node_Id) return Node_Id; 823 -- This routine scans out a qualified expression when the caller has 824 -- already scanned out the name and apostrophe of the construct. 825 826 function P_Quantified_Expression return Node_Id; 827 -- This routine scans out a quantified expression when the caller has 828 -- already scanned out the keyword "for" of the construct. 829 end Ch4; 830 831 ------------- 832 -- Par.Ch5 -- 833 ------------- 834 835 package Ch5 is 836 function P_Condition return Node_Id; 837 -- Scan out and return a condition. Note that an error is given if 838 -- the condition is followed by a right parenthesis. 839 840 function P_Condition (Cond : Node_Id) return Node_Id; 841 -- Similar to the above, but the caller has already scanned out the 842 -- conditional expression and passes it as an argument. This form of 843 -- the call does not check for a following right parenthesis. 844 845 function P_Iterator_Specification (Def_Id : Node_Id) return Node_Id; 846 -- Parse an iterator specification. The defining identifier has already 847 -- been scanned, as it is the common prefix between loop and iterator 848 -- specification. 849 850 function P_Loop_Parameter_Specification return Node_Id; 851 -- Used in loop constructs and quantified expressions. 852 853 function P_Sequence_Of_Statements (SS_Flags : SS_Rec) return List_Id; 854 -- The argument indicates the acceptable termination tokens. 855 -- See body in Par.Ch5 for details of the use of this parameter. 856 857 procedure Parse_Decls_Begin_End (Parent : Node_Id); 858 -- Parses declarations and handled statement sequence, setting 859 -- fields of Parent node appropriately. 860 end Ch5; 861 862 ------------- 863 -- Par.Ch6 -- 864 ------------- 865 866 package Ch6 is 867 function P_Designator return Node_Id; 868 function P_Defining_Program_Unit_Name return Node_Id; 869 function P_Formal_Part return List_Id; 870 function P_Parameter_Profile return List_Id; 871 function P_Return_Statement return Node_Id; 872 function P_Subprogram_Specification return Node_Id; 873 874 procedure P_Mode (Node : Node_Id); 875 -- Sets In_Present and/or Out_Present flags in Node scanning past IN, 876 -- OUT or IN OUT tokens in the source. 877 878 function P_Subprogram (Pf_Flags : Pf_Rec) return Node_Id; 879 -- Scans out any construct starting with either of the keywords 880 -- PROCEDURE or FUNCTION. The parameter indicates which possible 881 -- possible kinds of construct (body, spec, instantiation etc.) 882 -- are permissible in the current context. 883 end Ch6; 884 885 ------------- 886 -- Par.Ch7 -- 887 ------------- 888 889 package Ch7 is 890 function P_Package (Pf_Flags : Pf_Rec) return Node_Id; 891 -- Scans out any construct starting with the keyword PACKAGE. The 892 -- parameter indicates which possible kinds of construct (body, spec, 893 -- instantiation etc.) are permissible in the current context. 894 end Ch7; 895 896 ------------- 897 -- Par.Ch8 -- 898 ------------- 899 900 package Ch8 is 901 procedure P_Use_Clause (Item_List : List_Id); 902 end Ch8; 903 904 ------------- 905 -- Par.Ch9 -- 906 ------------- 907 908 package Ch9 is 909 function P_Abort_Statement return Node_Id; 910 function P_Abortable_Part return Node_Id; 911 function P_Accept_Statement return Node_Id; 912 function P_Delay_Statement return Node_Id; 913 function P_Entry_Body return Node_Id; 914 function P_Protected return Node_Id; 915 function P_Requeue_Statement return Node_Id; 916 function P_Select_Statement return Node_Id; 917 function P_Task return Node_Id; 918 function P_Terminate_Alternative return Node_Id; 919 end Ch9; 920 921 -------------- 922 -- Par.Ch10 -- 923 -------------- 924 925 package Ch10 is 926 function P_Compilation_Unit return Node_Id; 927 -- Note: this function scans a single compilation unit, and checks that 928 -- an end of file follows this unit, diagnosing any unexpected input as 929 -- an error, and then skipping it, so that Token is set to Tok_EOF on 930 -- return. An exception is in syntax-only mode, where multiple 931 -- compilation units are permitted. In this case, P_Compilation_Unit 932 -- does not check for end of file and there may be more compilation 933 -- units to scan. The caller can uniquely detect this situation by the 934 -- fact that Token is not set to Tok_EOF on return. 935 -- 936 -- What about multiple unit/file capability that now exists??? 937 -- 938 -- The Ignore parameter is normally set False. It is set True in the 939 -- multiple unit per file mode if we are skipping past a unit that we 940 -- are not interested in. 941 end Ch10; 942 943 -------------- 944 -- Par.Ch11 -- 945 -------------- 946 947 package Ch11 is 948 function P_Handled_Sequence_Of_Statements return Node_Id; 949 function P_Raise_Expression return Node_Id; 950 function P_Raise_Statement return Node_Id; 951 952 function Parse_Exception_Handlers return List_Id; 953 -- Parses the partial construct EXCEPTION followed by a list of 954 -- exception handlers which appears in a number of productions, and 955 -- returns the list of exception handlers. 956 end Ch11; 957 958 -------------- 959 -- Par.Ch12 -- 960 -------------- 961 962 package Ch12 is 963 function P_Generic return Node_Id; 964 function P_Generic_Actual_Part_Opt return List_Id; 965 end Ch12; 966 967 -------------- 968 -- Par.Ch13 -- 969 -------------- 970 971 package Ch13 is 972 function P_Representation_Clause return Node_Id; 973 974 function Aspect_Specifications_Present 975 (Strict : Boolean := Ada_Version < Ada_2012) return Boolean; 976 -- This function tests whether the next keyword is WITH followed by 977 -- something that looks reasonably like an aspect specification. If so, 978 -- True is returned. Otherwise False is returned. In either case control 979 -- returns with the token pointer unchanged (i.e. pointing to the WITH 980 -- token in the case where True is returned). This function takes care 981 -- of generating appropriate messages if aspect specifications appear 982 -- in versions of Ada prior to Ada 2012. The parameter strict can be 983 -- set to True, to be rather strict about considering something to be 984 -- an aspect specification. If Strict is False, then the circuitry is 985 -- rather more generous in considering something ill-formed to be an 986 -- attempt at an aspect specification. The default is more strict for 987 -- Ada versions before Ada 2012 (where aspect specifications are not 988 -- permitted). Note: this routine never checks the terminator token 989 -- for aspects so it does not matter whether the aspect specifications 990 -- are terminated by semicolon or some other character. 991 -- 992 -- Note: This function also handles the case of WHEN used where WITH 993 -- was intended, and in that case posts an error and returns True. 994 995 procedure P_Aspect_Specifications 996 (Decl : Node_Id; 997 Semicolon : Boolean := True); 998 -- This procedure scans out a series of aspect specifications. If 999 -- argument Semicolon is True, a terminating semicolon is also scanned. 1000 -- If this argument is False, the scan pointer is left pointing past the 1001 -- aspects and the caller must check for a proper terminator. 1002 -- 1003 -- P_Aspect_Specifications is called with the current token pointing 1004 -- to either a WITH keyword starting an aspect specification, or an 1005 -- instance of what shpould be a terminator token. In the former case, 1006 -- the aspect specifications are scanned out including the terminator 1007 -- token if it is a semicolon, and the Has_Aspect_Specifications 1008 -- flag is set in the given declaration node. A list of aspects 1009 -- is built and stored for this declaration node using a call to 1010 -- Set_Aspect_Specifications. If no WITH keyword is present, then this 1011 -- call has no effect other than scanning out the terminator if it is a 1012 -- semicolon (with the exception that it detects WHEN used in place of 1013 -- WITH). 1014 1015 -- If Decl is Error on entry, any scanned aspect specifications are 1016 -- ignored and a message is output saying aspect specifications not 1017 -- permitted here. If Decl is Empty, then scanned aspect specifications 1018 -- are also ignored, but no error message is given (this is used when 1019 -- the caller has already taken care of the error message). 1020 1021 function Get_Aspect_Specifications 1022 (Semicolon : Boolean := True) return List_Id; 1023 -- Parse a list of aspects but do not attach them to a declaration node. 1024 -- Subsidiary to P_Aspect_Specifications procedure. Used when parsing 1025 -- a subprogram specification that may be a declaration or a body. 1026 -- Semicolon has the same meaning as for P_Aspect_Specifications above. 1027 1028 function P_Code_Statement (Subtype_Mark : Node_Id) return Node_Id; 1029 -- Function to parse a code statement. The caller has scanned out 1030 -- the name to be used as the subtype mark (but has not checked that 1031 -- it is suitable for use as a subtype mark, i.e. is either an 1032 -- identifier or a selected component). The current token is an 1033 -- apostrophe and the following token is either a left paren or 1034 -- RANGE (the latter being an error to be caught by P_Code_Statement. 1035 end Ch13; 1036 1037 -- Note: the parsing for annexe J features (i.e. obsolescent features) 1038 -- is found in the logical section where these features would be if 1039 -- they were not obsolescent. In particular: 1040 1041 -- Delta constraint is parsed by P_Delta_Constraint (3.5.9) 1042 -- At clause is parsed by P_At_Clause (13.1) 1043 -- Mod clause is parsed by P_Mod_Clause (13.5.1) 1044 1045 -------------- 1046 -- Par.Endh -- 1047 -------------- 1048 1049 -- Routines for handling end lines, including scope recovery 1050 1051 package Endh is 1052 function Check_End 1053 (Decl : Node_Id := Empty; 1054 Is_Loc : Source_Ptr := No_Location) return Boolean; 1055 -- Called when an end sequence is required. In the absence of an error 1056 -- situation, Token contains Tok_End on entry, but in a missing end 1057 -- case, this may not be the case. Pop_End_Context is used to determine 1058 -- the appropriate action to be taken. The returned result is True if 1059 -- an End sequence was encountered and False if no End sequence was 1060 -- present. This occurs if the END keyword encountered was determined 1061 -- to be improper and deleted (i.e. Pop_End_Context set End_Action to 1062 -- Skip_And_Reject). Note that the END sequence includes a semicolon, 1063 -- except in the case of END RECORD, where a semicolon follows the END 1064 -- RECORD, but is not part of the record type definition itself. 1065 -- 1066 -- If Decl is non-empty, then aspect specifications are permitted 1067 -- following the end, and Decl is the declaration node with which 1068 -- these aspect specifications are to be associated. If Decl is empty, 1069 -- then aspect specifications are not permitted and will generate an 1070 -- error message. 1071 -- 1072 -- Is_Loc is set to other than the default only for the case of a 1073 -- package declaration. It points to the IS keyword of the declaration, 1074 -- and is used to specialize the error messages for misplaced aspect 1075 -- specifications in this case. Note that Decl is always Empty if Is_Loc 1076 -- is set. 1077 1078 procedure End_Skip; 1079 -- Skip past an end sequence. On entry Token contains Tok_End, and we 1080 -- we know that the end sequence is syntactically incorrect, and that 1081 -- an appropriate error message has already been posted. The mission 1082 -- is simply to position the scan pointer to be the best guess of the 1083 -- position after the end sequence. We do not issue any additional 1084 -- error messages while carrying this out. 1085 1086 procedure End_Statements 1087 (Parent : Node_Id := Empty; 1088 Decl : Node_Id := Empty; 1089 Is_Sloc : Source_Ptr := No_Location); 1090 -- Called when an end is required or expected to terminate a sequence 1091 -- of statements. The caller has already made an appropriate entry in 1092 -- the Scope.Table to describe the expected form of the end. This can 1093 -- only be used in cases where the only appropriate terminator is end. 1094 -- If Parent is non-empty, then if a correct END line is encountered, 1095 -- the End_Label field of Parent is set appropriately. 1096 -- 1097 -- If Decl is non-null, then it is a declaration node, and aspect 1098 -- specifications are permitted after the end statement. These aspect 1099 -- specifications, if present, are stored in this declaration node. 1100 -- If Decl is null, then aspect specifications are not permitted after 1101 -- the end statement. 1102 -- 1103 -- In the case where Decl is null, Is_Sloc determines the handling. If 1104 -- it is set to No_Location, then aspect specifications are ignored and 1105 -- an error message is given. Is_Sloc is used in the package declaration 1106 -- case to point to the IS, and is used to specialize the error emssages 1107 -- issued in this case. 1108 end Endh; 1109 1110 -------------- 1111 -- Par.Sync -- 1112 -------------- 1113 1114 -- These procedures are used to resynchronize after errors. Following an 1115 -- error which is not immediately locally recoverable, the exception 1116 -- Error_Resync is raised. The handler for Error_Resync typically calls 1117 -- one of these recovery procedures to resynchronize the source position 1118 -- to a point from which parsing can be restarted. 1119 1120 -- Note: these procedures output an information message that tokens are 1121 -- being skipped, but this message is output only if the option for 1122 -- Multiple_Errors_Per_Line is set in Options. 1123 1124 package Sync is 1125 procedure Resync_Choice; 1126 -- Used if an error occurs scanning a choice. The scan pointer is 1127 -- advanced to the next vertical bar, arrow, or semicolon, whichever 1128 -- comes first. We also quit if we encounter an end of file. 1129 1130 procedure Resync_Cunit; 1131 -- Synchronize to next token which could be the start of a compilation 1132 -- unit, or to the end of file token. 1133 1134 procedure Resync_Expression; 1135 -- Used if an error is detected during the parsing of an expression. 1136 -- It skips past tokens until either a token which cannot be part of 1137 -- an expression is encountered (an expression terminator), or if a 1138 -- comma or right parenthesis or vertical bar is encountered at the 1139 -- current parenthesis level (a parenthesis level counter is maintained 1140 -- to carry out this test). 1141 1142 procedure Resync_Past_Malformed_Aspect; 1143 -- Used when parsing aspect specifications to skip a malformed aspect. 1144 -- The scan pointer is positioned next to a comma, a semicolon or "is" 1145 -- when the aspect applies to a body. 1146 1147 procedure Resync_Past_Semicolon; 1148 -- Used if an error occurs while scanning a sequence of declarations. 1149 -- The scan pointer is positioned past the next semicolon and the scan 1150 -- resumes. The scan is also resumed on encountering a token which 1151 -- starts a declaration (but we make sure to skip at least one token 1152 -- in this case, to avoid getting stuck in a loop). 1153 1154 procedure Resync_Past_Semicolon_Or_To_Loop_Or_Then; 1155 -- Used if an error occurs while scanning a sequence of statements. The 1156 -- scan pointer is positioned past the next semicolon, or to the next 1157 -- occurrence of either then or loop, and the scan resumes. 1158 1159 procedure Resync_Semicolon_List; 1160 -- Used if an error occurs while scanning a parenthesized list of items 1161 -- separated by semicolons. The scan pointer is advanced to the next 1162 -- semicolon or right parenthesis at the outer parenthesis level, or 1163 -- to the next is or RETURN keyword occurrence, whichever comes first. 1164 1165 procedure Resync_To_Semicolon; 1166 -- Similar to Resync_Past_Semicolon, except that the scan pointer is 1167 -- left pointing to the semicolon rather than past it. 1168 1169 procedure Resync_To_When; 1170 -- Used when an error occurs scanning an entry index specification. The 1171 -- scan pointer is positioned to the next WHEN (or to IS or semicolon if 1172 -- either of these appear before WHEN, indicating another error has 1173 -- occurred). 1174 end Sync; 1175 1176 -------------- 1177 -- Par.Tchk -- 1178 -------------- 1179 1180 -- Routines to check for expected tokens 1181 1182 package Tchk is 1183 1184 -- Procedures with names of the form T_xxx, where Tok_xxx is a token 1185 -- name, check that the current token matches the required token, and 1186 -- if so, scan past it. If not, an error is issued indicating that 1187 -- the required token is not present (xxx expected). In most cases, the 1188 -- scan pointer is not moved in the not-found case, but there are some 1189 -- exceptions to this, see for example T_Id, where the scan pointer is 1190 -- moved across a literal appearing where an identifier is expected. 1191 1192 procedure T_Abort; 1193 procedure T_Arrow; 1194 procedure T_At; 1195 procedure T_Body; 1196 procedure T_Box; 1197 procedure T_Colon; 1198 procedure T_Colon_Equal; 1199 procedure T_Comma; 1200 procedure T_Dot_Dot; 1201 procedure T_For; 1202 procedure T_Greater_Greater; 1203 procedure T_Identifier; 1204 procedure T_In; 1205 procedure T_Is; 1206 procedure T_Left_Paren; 1207 procedure T_Loop; 1208 procedure T_Mod; 1209 procedure T_New; 1210 procedure T_Of; 1211 procedure T_Or; 1212 procedure T_Private; 1213 procedure T_Range; 1214 procedure T_Record; 1215 procedure T_Right_Bracket; 1216 procedure T_Right_Paren; 1217 procedure T_Semicolon; 1218 procedure T_Then; 1219 procedure T_Type; 1220 procedure T_Use; 1221 procedure T_When; 1222 procedure T_With; 1223 1224 -- Procedures having names of the form TF_xxx, where Tok_xxx is a token 1225 -- name check that the current token matches the required token, and 1226 -- if so, scan past it. If not, an error message is issued indicating 1227 -- that the required token is not present (xxx expected). 1228 1229 -- If the missing token is at the end of the line, then control returns 1230 -- immediately after posting the message. If there are remaining tokens 1231 -- on the current line, a search is conducted to see if the token 1232 -- appears later on the current line, as follows: 1233 1234 -- A call to Scan_Save is issued and a forward search for the token 1235 -- is carried out. If the token is found on the current line before a 1236 -- semicolon, then it is scanned out and the scan continues from that 1237 -- point. If not the scan is restored to the point where it was missing. 1238 1239 procedure TF_Arrow; 1240 procedure TF_Is; 1241 procedure TF_Loop; 1242 procedure TF_Return; 1243 procedure TF_Semicolon; 1244 procedure TF_Then; 1245 procedure TF_Use; 1246 1247 -- Procedures with names of the form U_xxx, where Tok_xxx is a token 1248 -- name, are just like the corresponding T_xxx procedures except that 1249 -- an error message, if given, is unconditional. 1250 1251 procedure U_Left_Paren; 1252 procedure U_Right_Paren; 1253 end Tchk; 1254 1255 -------------- 1256 -- Par.Util -- 1257 -------------- 1258 1259 package Util is 1260 function Bad_Spelling_Of (T : Token_Type) return Boolean; 1261 -- This function is called in an error situation. It checks if the 1262 -- current token is an identifier whose name is a plausible bad 1263 -- spelling of the given keyword token, and if so, issues an error 1264 -- message, sets Token from T, and returns True. Otherwise Token is 1265 -- unchanged, and False is returned. 1266 1267 procedure Check_Bad_Layout; 1268 -- Check for bad indentation in RM checking mode. Used for statements 1269 -- and declarations. Checks if current token is at start of line and 1270 -- is exdented from the current expected end column, and if so an 1271 -- error message is generated. 1272 1273 procedure Check_Misspelling_Of (T : Token_Type); 1274 pragma Inline (Check_Misspelling_Of); 1275 -- This is similar to the function above, except that it does not 1276 -- return a result. It is typically used in a situation where any 1277 -- identifier is an error, and it makes sense to simply convert it 1278 -- to the given token if it is a plausible misspelling of it. 1279 1280 procedure Check_95_Keyword (Token_95, Next : Token_Type); 1281 -- This routine checks if the token after the current one matches the 1282 -- Next argument. If so, the scan is backed up to the current token 1283 -- and Token_Type is changed to Token_95 after issuing an appropriate 1284 -- error message ("(Ada 83) keyword xx cannot be used"). If not, 1285 -- the scan is backed up with Token_Type unchanged. This routine 1286 -- is used to deal with an attempt to use a 95 keyword in Ada 83 1287 -- mode. The caller has typically checked that the current token, 1288 -- an identifier, matches one of the 95 keywords. 1289 1290 procedure Check_Future_Keyword; 1291 -- Emit a warning if the current token is a valid identifier in the 1292 -- language version in use, but is a reserved word in a later language 1293 -- version (unless the language version in use is Ada 83). 1294 1295 procedure Check_Simple_Expression (E : Node_Id); 1296 -- Given an expression E, that has just been scanned, so that Expr_Form 1297 -- is still set, outputs an error if E is a non-simple expression. E is 1298 -- not modified by this call. 1299 1300 procedure Check_Simple_Expression_In_Ada_83 (E : Node_Id); 1301 -- Like Check_Simple_Expression, except that the error message is only 1302 -- given when operating in Ada 83 mode, and includes "in Ada 83". 1303 1304 function Check_Subtype_Mark (Mark : Node_Id) return Node_Id; 1305 -- Called to check that a node representing a name (or call) is 1306 -- suitable for a subtype mark, i.e, that it is an identifier or 1307 -- a selected component. If so, or if it is already Error, then 1308 -- it is returned unchanged. Otherwise an error message is issued 1309 -- and Error is returned. 1310 1311 function Comma_Present return Boolean; 1312 -- Used in comma delimited lists to determine if a comma is present, or 1313 -- can reasonably be assumed to have been present (an error message is 1314 -- generated in the latter case). If True is returned, the scan has been 1315 -- positioned past the comma. If False is returned, the scan position 1316 -- is unchanged. Note that all comma-delimited lists are terminated by 1317 -- a right paren, so the only legitimate tokens when Comma_Present is 1318 -- called are right paren and comma. If some other token is found, then 1319 -- Comma_Present has the job of deciding whether it is better to pretend 1320 -- a comma was present, post a message for a missing comma and return 1321 -- True, or return False and let the caller diagnose the missing right 1322 -- parenthesis. 1323 1324 procedure Discard_Junk_Node (N : Node_Id); 1325 procedure Discard_Junk_List (L : List_Id); 1326 pragma Inline (Discard_Junk_Node); 1327 pragma Inline (Discard_Junk_List); 1328 -- These procedures do nothing at all, their effect is simply to discard 1329 -- the argument. A typical use is to skip by some junk that is not 1330 -- expected in the current context. 1331 1332 procedure Ignore (T : Token_Type); 1333 -- If current token matches T, then give an error message and skip 1334 -- past it, otherwise the call has no effect at all. T may be any 1335 -- reserved word token, or comma, left or right paren, or semicolon. 1336 1337 function Is_Reserved_Identifier (C : Id_Check := None) return Boolean; 1338 -- Test if current token is a reserved identifier. This test is based 1339 -- on the token being a keyword and being spelled in typical identifier 1340 -- style (i.e. starting with an upper case letter). The parameter C 1341 -- determines the special treatment if a reserved word is encountered 1342 -- that has the normal casing of a reserved word. 1343 1344 procedure Merge_Identifier (Prev : Node_Id; Nxt : Token_Type); 1345 -- Called when the previous token is an identifier (whose Token_Node 1346 -- value is given by Prev) to check if current token is an identifier 1347 -- that can be merged with the previous one adding an underscore. The 1348 -- merge is only attempted if the following token matches Nxt. If all 1349 -- conditions are met, an error message is issued, and the merge is 1350 -- carried out, modifying the Chars field of Prev. 1351 1352 function Next_Token_Is (Tok : Token_Type) return Boolean; 1353 -- Looks at token after current one and returns True if the token type 1354 -- matches Tok. The scan is unconditionally restored on return. 1355 1356 procedure No_Constraint; 1357 -- Called in a place where no constraint is allowed, but one might 1358 -- appear due to a common error (e.g. after the type mark in a procedure 1359 -- parameter. If a constraint is present, an error message is posted, 1360 -- and the constraint is scanned and discarded. 1361 1362 procedure Push_Scope_Stack; 1363 pragma Inline (Push_Scope_Stack); 1364 -- Push a new entry onto the scope stack. Scope.Last (the stack pointer) 1365 -- is incremented. The Junk field is preinitialized to False. The caller 1366 -- is expected to fill in all remaining entries of the new top stack 1367 -- entry at Scopes (Scope.Last). 1368 1369 procedure Pop_Scope_Stack; 1370 -- Pop an entry off the top of the scope stack. Scope_Last (the scope 1371 -- table stack pointer) is decremented by one. It is a fatal error to 1372 -- try to pop off the dummy entry at the bottom of the stack (i.e. 1373 -- Scope.Last must be non-zero at the time of call). 1374 1375 function Separate_Present return Boolean; 1376 -- Determines if the current token is either Tok_Separate, or an 1377 -- identifier that is a possible misspelling of "separate" followed 1378 -- by a semicolon. True is returned if so, otherwise False. 1379 1380 procedure Signal_Bad_Attribute; 1381 -- The current token is an identifier that is supposed to be an 1382 -- attribute identifier but is not. This routine posts appropriate 1383 -- error messages, including a check for a near misspelling. 1384 1385 function Token_Is_At_Start_Of_Line return Boolean; 1386 pragma Inline (Token_Is_At_Start_Of_Line); 1387 -- Determines if the current token is the first token on the line 1388 1389 function Token_Is_At_End_Of_Line return Boolean; 1390 -- Determines if the current token is the last token on the line 1391 1392 procedure Warn_If_Standard_Redefinition (N : Node_Id); 1393 -- Issues a warning if Warn_On_Standard_Redefinition is set True, and 1394 -- the Node N (which is a Defining_Identifier node with the Chars field 1395 -- set) is a renaming of an entity in package Standard. 1396 1397 end Util; 1398 1399 -------------- 1400 -- Par.Prag -- 1401 -------------- 1402 1403 -- The processing for pragmas is split off from chapter 2 1404 1405 function Prag (Pragma_Node : Node_Id; Semi : Source_Ptr) return Node_Id; 1406 -- This function is passed a tree for a pragma that has been scanned out. 1407 -- The pragma is syntactically well formed according to the general syntax 1408 -- for pragmas and the pragma identifier is for one of the recognized 1409 -- pragmas. It performs specific syntactic checks for specific pragmas. 1410 -- The result is the input node if it is OK, or Error otherwise. The 1411 -- reason that this is separated out is to facilitate the addition 1412 -- of implementation defined pragmas. The second parameter records the 1413 -- location of the semicolon following the pragma (this is needed for 1414 -- correct processing of the List and Page pragmas). The returned value 1415 -- is a copy of Pragma_Node, or Error if an error is found. Note that 1416 -- at the point where Prag is called, the right paren ending the pragma 1417 -- has been scanned out, and except in the case of pragma Style_Checks, 1418 -- so has the following semicolon. For Style_Checks, the caller delays 1419 -- the scanning of the semicolon so that it will be scanned using the 1420 -- settings from the Style_Checks pragma preceding it. 1421 1422 -------------- 1423 -- Par.Labl -- 1424 -------------- 1425 1426 procedure Labl; 1427 -- This procedure creates implicit label declarations for all labels that 1428 -- are declared in the current unit. Note that this could conceptually be 1429 -- done at the point where the labels are declared, but it is tricky to do 1430 -- it then, since the tree is not hooked up at the point where the label is 1431 -- declared (e.g. a sequence of statements is not yet attached to its 1432 -- containing scope at the point a label in the sequence is found). 1433 1434 -------------- 1435 -- Par.Load -- 1436 -------------- 1437 1438 procedure Load; 1439 -- This procedure loads all subsidiary units that are required by this 1440 -- unit, including with'ed units, specs for bodies, and parents for child 1441 -- units. It does not load bodies for inlined procedures and generics, 1442 -- since we don't know till semantic analysis is complete what is needed. 1443 1444 ----------- 1445 -- Stubs -- 1446 ----------- 1447 1448 -- The package bodies can see all routines defined in all other subpackages 1449 1450 use Ch2; 1451 use Ch3; 1452 use Ch4; 1453 use Ch5; 1454 use Ch6; 1455 use Ch7; 1456 use Ch8; 1457 use Ch9; 1458 use Ch10; 1459 use Ch11; 1460 use Ch12; 1461 use Ch13; 1462 1463 use Endh; 1464 use Tchk; 1465 use Sync; 1466 use Util; 1467 1468 package body Ch2 is separate; 1469 package body Ch3 is separate; 1470 package body Ch4 is separate; 1471 package body Ch5 is separate; 1472 package body Ch6 is separate; 1473 package body Ch7 is separate; 1474 package body Ch8 is separate; 1475 package body Ch9 is separate; 1476 package body Ch10 is separate; 1477 package body Ch11 is separate; 1478 package body Ch12 is separate; 1479 package body Ch13 is separate; 1480 1481 package body Endh is separate; 1482 package body Tchk is separate; 1483 package body Sync is separate; 1484 package body Util is separate; 1485 1486 function Prag (Pragma_Node : Node_Id; Semi : Source_Ptr) return Node_Id 1487 is separate; 1488 1489 procedure Labl is separate; 1490 procedure Load is separate; 1491 1492 Result : List_Id := Empty_List; 1493 1494-- Start of processing for Par 1495 1496begin 1497 Compiler_State := Parsing; 1498 1499 -- Deal with configuration pragmas case first 1500 1501 if Configuration_Pragmas then 1502 declare 1503 Pragmas : constant List_Id := Empty_List; 1504 P_Node : Node_Id; 1505 1506 begin 1507 loop 1508 if Token = Tok_EOF then 1509 Result := Pragmas; 1510 exit; 1511 1512 elsif Token /= Tok_Pragma then 1513 Error_Msg_SC ("only pragmas allowed in configuration file"); 1514 Result := Error_List; 1515 exit; 1516 1517 else 1518 P_Node := P_Pragma; 1519 1520 if Nkind (P_Node) = N_Pragma then 1521 1522 -- Give error if bad pragma 1523 1524 if not Is_Configuration_Pragma_Name 1525 (Pragma_Name_Unmapped (P_Node)) 1526 and then 1527 Pragma_Name_Unmapped (P_Node) /= Name_Source_Reference 1528 then 1529 if Is_Pragma_Name (Pragma_Name_Unmapped (P_Node)) then 1530 Error_Msg_N 1531 ("only configuration pragmas allowed " & 1532 "in configuration file", P_Node); 1533 else 1534 Error_Msg_N 1535 ("unrecognized pragma in configuration file", 1536 P_Node); 1537 end if; 1538 1539 -- Pragma is OK config pragma, so collect it 1540 1541 else 1542 Append (P_Node, Pragmas); 1543 end if; 1544 end if; 1545 end if; 1546 end loop; 1547 end; 1548 1549 if Config_Files_Store_Basename then 1550 Complete_Source_File_Entry; 1551 end if; 1552 1553 -- Normal case of compilation unit 1554 1555 else 1556 Save_Config_Attrs := Save_Config_Switches; 1557 1558 -- The following loop runs more than once in syntax check mode 1559 -- where we allow multiple compilation units in the same file 1560 -- and in Multiple_Unit_Per_file mode where we skip units till 1561 -- we get to the unit we want. 1562 1563 for Ucount in Pos loop 1564 Set_Config_Switches 1565 (Is_Internal_Unit (Current_Source_Unit), 1566 Main_Unit => Current_Source_Unit = Main_Unit); 1567 1568 -- Initialize scope table and other parser control variables 1569 1570 Compiler_State := Parsing; 1571 Scope.Init; 1572 Scope.Increment_Last; 1573 Scopes (0).Etyp := E_Dummy; 1574 SIS_Entry_Active := False; 1575 Last_Resync_Point := No_Location; 1576 1577 Goto_List := New_Elmt_List; 1578 Label_List := New_Elmt_List; 1579 1580 -- If in multiple unit per file mode, skip past ignored unit 1581 1582 if Ucount < Multiple_Unit_Index then 1583 1584 -- We skip in syntax check only mode, since we don't want to do 1585 -- anything more than skip past the unit and ignore it. This means 1586 -- we skip processing like setting up a unit table entry. 1587 1588 declare 1589 Save_Operating_Mode : constant Operating_Mode_Type := 1590 Operating_Mode; 1591 1592 Save_Style_Check : constant Boolean := Style_Check; 1593 1594 begin 1595 Operating_Mode := Check_Syntax; 1596 Style_Check := False; 1597 Discard_Node (P_Compilation_Unit); 1598 Operating_Mode := Save_Operating_Mode; 1599 Style_Check := Save_Style_Check; 1600 1601 -- If we are at an end of file, and not yet at the right unit, 1602 -- then we have a fatal error. The unit is missing. 1603 1604 if Token = Tok_EOF then 1605 Error_Msg_SC ("file has too few compilation units"); 1606 raise Unrecoverable_Error; 1607 end if; 1608 end; 1609 1610 -- Here if we are not skipping a file in multiple unit per file mode. 1611 -- Parse the unit that we are interested in. Note that in check 1612 -- syntax mode we are interested in all units in the file. 1613 1614 else 1615 declare 1616 Comp_Unit_Node : constant Node_Id := P_Compilation_Unit; 1617 1618 begin 1619 -- If parsing was successful and we are not in check syntax 1620 -- mode, check that language-defined units are compiled in GNAT 1621 -- mode. For this purpose we do NOT consider renamings in annex 1622 -- J as predefined. That allows users to compile their own 1623 -- versions of these files. Another exception is System.RPC 1624 -- and its children. This allows a user to supply their own 1625 -- communication layer. 1626 -- Similarly, we do not generate an error in CodePeer mode, 1627 -- to allow users to analyze third-party compiler packages. 1628 1629 if Comp_Unit_Node /= Error 1630 and then Operating_Mode = Generate_Code 1631 and then Current_Source_Unit = Main_Unit 1632 and then not GNAT_Mode 1633 and then not CodePeer_Mode 1634 then 1635 declare 1636 Uname : constant String := 1637 Get_Name_String 1638 (Unit_Name (Current_Source_Unit)); 1639 Name : String (1 .. Uname'Length - 2); 1640 1641 begin 1642 -- Because Unit_Name includes "%s"/"%b", we need to strip 1643 -- the last two characters to get the real unit name. 1644 1645 Name := Uname (Uname'First .. Uname'Last - 2); 1646 1647 if Name = "ada" or else 1648 Name = "interfaces" or else 1649 Name = "system" 1650 then 1651 Error_Msg 1652 ("language-defined units cannot be recompiled", 1653 Sloc (Unit (Comp_Unit_Node))); 1654 1655 elsif Name'Length > 4 1656 and then 1657 Name (Name'First .. Name'First + 3) = "ada." 1658 then 1659 Error_Msg 1660 ("user-defined descendants of package Ada " & 1661 "are not allowed", 1662 Sloc (Unit (Comp_Unit_Node))); 1663 1664 elsif Name'Length > 11 1665 and then 1666 Name (Name'First .. Name'First + 10) = "interfaces." 1667 then 1668 Error_Msg 1669 ("user-defined descendants of package Interfaces " & 1670 "are not allowed", 1671 Sloc (Unit (Comp_Unit_Node))); 1672 1673 elsif Name'Length > 7 1674 and then Name (Name'First .. Name'First + 6) = "system." 1675 and then Name /= "system.rpc" 1676 and then 1677 (Name'Length < 11 1678 or else Name (Name'First .. Name'First + 10) /= 1679 "system.rpc.") 1680 then 1681 Error_Msg 1682 ("user-defined descendants of package System " & 1683 "are not allowed", 1684 Sloc (Unit (Comp_Unit_Node))); 1685 end if; 1686 end; 1687 end if; 1688 end; 1689 1690 -- All done if at end of file 1691 1692 exit when Token = Tok_EOF; 1693 1694 -- If we are not at an end of file, it means we are in syntax 1695 -- check only mode, and we keep the loop going to parse all 1696 -- remaining units in the file. 1697 1698 end if; 1699 1700 Restore_Config_Switches (Save_Config_Attrs); 1701 end loop; 1702 1703 -- Now that we have completely parsed the source file, we can complete 1704 -- the source file table entry. 1705 1706 Complete_Source_File_Entry; 1707 1708 -- An internal error check, the scope stack should now be empty 1709 1710 pragma Assert (Scope.Last = 0); 1711 1712 -- Here we make the SCO table entries for the main unit 1713 1714 if Generate_SCO then 1715 SCO_Record_Raw (Main_Unit); 1716 end if; 1717 1718 -- Remaining steps are to create implicit label declarations and to load 1719 -- required subsidiary sources. These steps are required only if we are 1720 -- doing semantic checking. 1721 1722 if Operating_Mode /= Check_Syntax or else Debug_Flag_F then 1723 Par.Labl; 1724 Par.Load; 1725 end if; 1726 1727 -- Restore settings of switches saved on entry 1728 1729 Restore_Config_Switches (Save_Config_Attrs); 1730 Set_Comes_From_Source_Default (False); 1731 end if; 1732 1733 Compiler_State := Analyzing; 1734 Current_Source_File := No_Source_File; 1735 return Result; 1736end Par; 1737