M2compiler-V3/compiler/src/AST.def

DEFINITION MODULE AST;
(* Abstract syntax tree for m2compiler-V3 (Stage A: node store only).

   This is the foundation for a two-phase frontend: productions build
   nodes here while parsing, and a later walk resolves names and
   lowers to QBE.  Until the grammar is converted, nothing calls it.

   Design notes
   ------------
   - Nodes live in a flat, pre-allocated arena and are referenced by an
     integer index (`Node`), exactly like `SymTab.TypeIndex`.  This
     avoids pointers in the interface and keeps the structure
     self-hosting-friendly (no GC, bounded memory).
   - Every node has a `kind` (NKind) and a `ty` (a SymTab.TypeIndex set
     by the checker), plus up to `MaxChild` children.  This mirrors the
     `armornick/CocoR` AST.java pattern (a Node base with per-construct
     subclasses) but flattened into one record array so Modula-2 can
     store it without variant-class machinery.
   - Text (identifiers, literals, operator spellings) is kept in a
     pooled string table; `TxtIndex` is an index into it.
   - `NoNode` is the null node. *)

CONST
  NoNode = -1;
  NoTxt  = -1;

  (* how many children a node may carry.  Statements/expressions in
     Modula-2 nest shallowly (a body is a list, an expression is a
     spine); 8 covers every construct in the current grammar with room
     to spare. *)
  MaxChild = 8;

  (* ---- node kinds ----
     Grouped by role; the numeric values are stable and may be stored
     in listings. *)

  (* module / unit structure *)
  NkUnit        = 0;   (* [0]=name txt, [1]=decls, [2]=body *)
  NkDefUnit     = 1;
  NkImplUnit    = 2;
  NkDeclSeq     = 3;   (* [0..n)=declarations *)
  NkImport      = 4;   (* [0]=module txt, [1..n)=names (or FROM list) *)

  (* declarations *)
  NkConstDecl   = 10;  (* [0]=name txt, [1]=value expr *)
  NkTypeDecl    = 11;  (* [0]=name txt, [1]=type *)
  NkVarDecl     = 12;  (* [0..n)=name txt, [1]=type  (names then type) *)
  NkProcDecl    = 13;  (* [0]=name, [1]=formals, [2]=result type,
                          [3]=body, [4]=forward flag node *)
  NkClassDecl   = 14;
  NkFieldDecl   = 15;

  (* type constructors *)
  NkTypeIdent   = 20;  (* [0]=name txt (possibly qualified: a.b) *)
  NkSubrange    = 21;  (* [0]=lo expr, [1]=hi expr *)
  NkEnum        = 22;  (* [0..n)=literal name txts *)
  NkArray       = 23;  (* [0..n-2)=index types, [last]=element type *)
  NkRecord      = 24;  (* [0..n)=fields *)
  NkSet         = 25;  (* [0]=base type *)
  NkPointer     = 26;  (* [0]=base type *)
  NkProcType    = 27;  (* [0]=formals, [1]=result type *)

  (* statements *)
  NkAssign      = 40;  (* [0]=designator, [1]=value *)
  NkCall        = 41;  (* [0]=callee, [1..n)=actuals *)
  NkIf          = 42;  (* [0]=cond, [1]=then, [2]=else (or NoNode) *)
  NkWhile       = 43;  (* [0]=cond, [1]=body *)
  NkRepeat      = 44;  (* [0]=body, [1]=until cond *)
  NkLoop        = 45;  (* [0]=body *)
  NkExit        = 46;
  NkFor         = 47;  (* [0]=control var, [1]=lo, [2]=hi, [3]=by, [4]=body *)
  NkCase        = 48;  (* [0]=selector, [1..n)=case arms *)
  NkCaseArm     = 49;  (* [0..n-1)=labels, [n]=body *)
  NkWith        = 50;  (* [0..n-1)=designators, [n]=body *)
  NkReturn      = 51;  (* [0]=expr (or NoNode) *)
  NkBlock       = 52;  (* [0..n)=statements *)
  NkHalt        = 53;
  NkGoto        = 54;  (* [0]=label ident *)
  NkLabel       = 55;  (* [0]=label ident *)

  (* expressions *)
  NkBinExpr     = 70;  (* [0]=left, [1]=right; `op` holds the operator *)
  NkUnary       = 71;  (* [0]=operand; `op` holds the operator *)
  NkDesignator  = 72;  (* [0]=base, [1..n)=selectors *)
  NkSelector    = 73;  (* .field / [idx] / ^ ; `op` distinguishes *)
  NkIdent       = 74;  (* [0]=name txt, [1]=qualifier txt (or NoTxt) *)
  NkIntLit      = 75;  (* [0]=digits txt *)
  NkRealLit     = 76;
  NkStrLit      = 77;
  NkCharLit     = 78;
  NkBraceLit    = 79;  (* [0]=type, [1..n)=elements *)
  NkSetLit      = 80;

  (* ---- relational / arithmetic / selector operator codes ---- *)
  OpAdd = 0;  OpSub = 1;  OpOr  = 2;
  OpMul = 3;  OpDiv = 4;  OpMod = 5;  OpAnd = 6;
  OpEq  = 7;  OpNe  = 8;  OpLt  = 9;  OpLe  = 10;  OpGt = 11;  OpGe = 12;
  OpIn  = 13;  OpNot = 14;

  (* selector codes (NkSelector.op) *)
  SelField = 0;  SelIndex = 1;  SelDeref = 2;

  TYPE
    Node    = INTEGER;   (* index into the arena, or NoNode *)
    TxtIndex = INTEGER;  (* index into the text pool, or NoTxt *)

  (* ---------------- lifecycle ---------------- *)

PROCEDURE Init;
(* Clears the arena and text pool. *)

  (* ---------------- text pool ---------------- *)

PROCEDURE AddTxt (s : ARRAY OF CHAR) : TxtIndex;
(* Interns s, returning its index (deduplicated). *)

PROCEDURE Txt (i : TxtIndex; VAR s : ARRAY OF CHAR);
(* Copies the i-th text into s (empty when out of range). *)


  (* ---------------- node construction ---------------- *)

PROCEDURE MakeNode (kind : INTEGER) : Node;
(* A fresh node of the given kind, no children, ty = SymTab.InvalidType
   (the caller sets it).  NoNode when the arena is full. *)

PROCEDURE MakeLeaf (kind : INTEGER; name : ARRAY OF CHAR) : Node;
(* A fresh node with child[0] set to the interned text of name. *)

PROCEDURE MakeBin (kind, op : INTEGER; l, r : Node) : Node;
(* A fresh node with op and two children (binary expr, assign, …). *)

PROCEDURE MakeUn (kind, op : INTEGER; e : Node) : Node;
(* A fresh node with op and one child (unary expr, selector, …). *)

  (* ---------------- node accessors ---------------- *)

PROCEDURE Kind (n : Node) : INTEGER;

PROCEDURE Op (n : Node) : INTEGER;
PROCEDURE SetOp (n : Node; op : INTEGER);

PROCEDURE Ty (n : Node) : INTEGER;
(* The SymTab.TypeIndex attached by the checker (InvalidType until
   then).  Declared as INTEGER to avoid a circular import with
   SymTab. *)
PROCEDURE SetTy (n : Node; t : INTEGER);

PROCEDURE NChild (n : Node) : CARDINAL;
PROCEDURE Child (n : Node; i : CARDINAL) : Node;
PROCEDURE AddChild (n : Node; c : Node) : BOOLEAN;
PROCEDURE SetChild (n : Node; i : CARDINAL; c : Node);
(* Replaces child i, extending the node when i = NChild. *)

  (* ---------------- listing (debug) ---------------- *)

PROCEDURE Dump (n : Node; depth : CARDINAL);
(* Writes a one-line-per-node indented tree to StdOut (via FileIO).
   Used by the Stage-A self-test and future debugging. *)

END AST.