COMPILER Umbriel
(* COCO/R Grammar for Umbriel

   Umbriel is a simple Modula-2 / Pascal like language

   This grammar is LL(1), so a Coco/R generated parser will be correct *)

CHARACTERS
  eol      = CHR(13) .
  letter   = "ABCDEFGHIJKLMNOPQRSTUVWXYZabcdefghijklmnopqrstuvwxyz" .
  digit    = "0123456789" .
  noQuote1 = ANY - "'" - eol .
  noQuote2 = ANY - '"' - eol .

IGNORE
  CHR(9) .. CHR(13)
  COMMENTS FROM '(*' TO '*)' NESTED
  (* The language is case sensitive *)

TOKENS
  identifier =  letter { letter | digit } .

  char       =  digit {digit} "C" .

(* The xxC variation uses decimal numbers, not the octal form of Wirth's other
   languages *)

  integer    =    digit { digit }
                | digit { digit }  CONTEXT ("..") .

  real       =  digit { digit } "." { digit }
                [ ( "E" | "e" ) [ "+" | "-" ] digit { digit } ] .

  string     =    "'" { noQuote1 } "'" | '"' { noQuote2 } '"' .

PRODUCTIONS
  Umbriel               = "MODULE" ModuleIdentifier ";"
                             Block ModuleIdentifier "." .

  ModuleIdentifier      = identifier .

(* The same identifier appears in both places *)

  Block                 = { Declaration }
                          [ "BEGIN" StatementSequence ] "END" .
  Declaration           =   "CONST" { ConstantDeclaration ";" }
                          | "TYPE" { TypeDeclaration ";" }
                          | "VAR" { VariableDeclaration ";" }
                          | ProcedureDeclaration ";" .

(* Declarations may be mixed in order, but for all identifiers
   declaration-before-use is required *)

  ConstantDeclaration   = ConstIdentifier "=" ConstExpression .
  ConstIdentifier       = identifier .
  ConstExpression       = Expression .

(* A ConstExpression may only contain constants, literals, and some standard
   function calls *)

  TypeDeclaration       = TypeIdentifier "=" Type .
  Type                  = TypeIdentifier | ArrayType | RecordType .
  TypeIdentifier        = identifier .

  ArrayType             = "ARRAY" IndexType { "," IndexType } "OF" Type .
  IndexType             = "[" ConstExpression ".." ConstExpression "]" .

(* The index type is an anonymous subrange type.  It has to be declared like
   this; implicit subranges like CHAR and BOOLEAN may not be used by quoting
   the type names. *)

  RecordType            = "RECORD" FieldListSequence "END" .
  FieldListSequence     = FieldList { ";" FieldList } .
  FieldList             = [ IdentList ":" Type ] .
  IdentList             = VariableIdentifier { "," VariableIdentifier } .
  VariableIdentifier    = identifier .

(* The syntax allows the "permissive" use of extra ; symbols as in Modula-2,
   and also completely empty records, as in Wirth's other languages *)

  VariableDeclaration   = IdentList ":" TypeIdentifier .

(* Unlike Modula-2 and Pascal, a variable has to be declared in terms of a
   named type.  This is to encourage simple abstraction. *)

  ProcedureDeclaration  = "PROCEDURE" ProcedureIdentifier
                           [ ParameterDeclarations ]
                          ";" Block ProcedureIdentifier .
  ParameterDeclarations = "(" [ FormalParameters ] ")" [ ":" ResultType ] .

  ProcedureIdentifier   = identifier .
  ResultType            = ScalarTypeIdentifier .
  ScalarTypeIdentifier  = TypeIdentifier .

(* Procedures may omit the ( brackets ) if there are no parameters.
   Functions may have no parameters, but require the ( brackets ),
   as in Modula-2.
   Functions may only return INTEGER, REAL, CHAR, COLORS or BOOLEAN values *)

  FormalParameters      = FormalParameter { ";" FormalParameter } .
  FormalParameter       = ValueSpecification | VariableSpecification .
  ValueSpecification    = IdentList ":" TypeIdentifier .
  VariableSpecification = "VAR" IdentList ":" TypeIdentifier .

  StatementSequence     = Statement { ";" Statement } .

  Statement             = [  AssignmentOrCall
                           | IfStatement | CaseStatement
                           | WhileStatement | RepeatStatement
                           | ForStatement | LoopStatement
                           | ExitStatement | ReturnStatement ] .

(* The statements are as in Modula-2, other than WITH *)

  AssignmentOrCall      = VarOrProcIdentifier
                            ( { Selector } ":=" Expression
                              | [ "(" [ ActualParameters ] ")" ] ) .

  VarOrProcIdentifier   = identifier .

  Selector              = "." VariableIdentifier | "[" IndexList "]" .
  IndexList             = OrdinalExpression { "," OrdinalExpression } .
  ActualParameters      = Expression [ FormatSpecifier ]
                          { "," Expression [ FormatSpecifier ] } .

  FormatSpecifier       = ":" IntegerExpression [ ":" IntegerExpression ] .

(* FormatSpecifier is only allowed in calls to the standard procedures
   Write and WriteLn *)

  IfStatement           = "IF" BooleanExpression "THEN" StatementSequence
                          { "ELSIF" BooleanExpression
                             "THEN" StatementSequence }
                          [ "ELSE" StatementSequence ] "END" .

  CaseStatement         = "CASE" Expression "OF" Case { "|" Case }
                          [ "ELSE" StatementSequence ] "END" .
  Case                  = [ CaseLabelList ":" StatementSequence ] .
  CaseLabelList         = CaseLabels { "," CaseLabels } .
  CaseLabels            = ConstExpression [ ".." ConstExpression ] .

(* The syntax allows the "permissive" use of extra | symbols as in Modula-2 *)

  WhileStatement        = "WHILE" BooleanExpression "DO"
                             StatementSequence
                          "END" .

  RepeatStatement       = "REPEAT"
                             StatementSequence
                          "UNTIL" BooleanExpression .

  ForStatement          = "FOR" VariableIdentifier ":=" OrdinalExpression
                          "TO" OrdinalExpression [ "BY" ConstExpression ] "DO"
                             StatementSequence
                          "END" .

(* The BY step requires a constant expression so that the compiler can decide
   at compile time whether it is a "to" or "downto" loop.  The control
   variable must be declared locally, and the implementation tries to detect
   situations where it would be threatened (changed in the body of the loop) *)

  LoopStatement         = "LOOP" StatementSequence "END" .
  ExitStatement         = "EXIT" .

(* An ExitStatement is only allowed within the StatementSequence of a
   LoopStatement.  It transfers control out of the closest enclosing
   LoopStatement. *)

  ReturnStatement       = "RETURN" [ Expression ] .

(* A ReturnStatement within a function must incorporate an Expression; within
   a regular procedure or a program block it must not incorporate an
   Expression. *)

  Expression            = SimpleExpression [ Relation SimpleExpression ] .
  BooleanExpression     = Expression .
  OrdinalExpression     = Expression .
  IntegerExpression     = Expression .

  SimpleExpression      = [ "+" | "-" ] Term { AddOperator Term } .
  Term                  = Factor { MulOperator Factor } .
  Factor                =   ConstantLiteral
                          | VarOrFuncOrConstIdentifier
                            ( { Selector } | "(" [ ActualParameters ] ")" )
                          | NotOperator Factor
                          | "(" Expression ")" .
  ConstantLiteral       = integer | real | char | string .
  VarOrFuncOrConstIdentifier   = identifier .
  NotOperator           = "NOT" | "~" (* synonyms *) .
  MulOperator           = "*" | "/" | "DIV" | "MOD" | AndOperator .
  AndOperator           = "AND" | "&"  (* synonyms *) .
  AddOperator           = "+" | "-" | "OR" .
  Relation              = "=" | "#" | "<>" | "<" | "<=" | ">" | ">=" .

END Umbriel.