taddeus
/
civicaml


			
				
					
						
						
							123456789101112131415161718192021222324252627282930313233343536373839404142434445464748495051525354555657585960616263646566676869707172737475767778798081828384858687888990919293949596979899100101102103104105106107108109110111112113114115116117118119120121122123124125126127128129130131132133134135136137138139140141142143144145146147148149150151152153154155156157158159160161162163164165166167168169170171172173174175176177178179180181182183184185186187188189190191192193194195196197198199200201202203204205206207208209210211212213214215216217218219220221222223224225226227228229230231232233234235236237238239240241242243244245246247248249250251252253254255256257
							(** Type definitions for abstract syntax tree, assembly code, and exceptions. *)

(** {2 Abstract Syntax Tree} *)

(** [(filename, startline, endline, startcol, endcol)]
    Location of node in source code. Used for error reporting. *)
type location = string * int * int * int * int

(** All operators supported by CiviC. *)
type operator =
  | Neg | Not
  | Add | Sub | Mul | Div | Mod
  | Eq | Ne | Lt | Le | Gt | Ge
  | And | Or

(** Wrappers for constant values. This eliminates the need for separate
    [(Bool|Int|Float)Const] nodes, and these values can be passed to {!instr}
    constructors as well. *)
type const =
  | BoolVal of bool
  | IntVal of int
  | FloatVal of float

(** Data types supported by CiviC. [ArrayDims] defines an array type with a set
    of dimensions. {!Dimreduce} replaces this multi-dimensional type with the
    [Array] type, which signifies a one-dimensional array of a certain basic
    type. [Unknown] is used by {!Typecheck} when the result of an operation is
    unclear due to a type error. *)
type ctype =
  | Void | Bool | Int | Float | Array of ctype
  | ArrayDims of ctype * node list
  | Unknown

(** Annotations for {!node}. Each node has a list of zero or more annotations
    that are gradually added through the compiler phases:

    - [Loc] is a location added by the parser.
    - [Depth] is the nesting level of a variable or function, and is added by
      {!Context}.
    - [Type] is a {!ctype} annotation added by {!Typecheck}.
    - [Index] is the stack index of a variable declaration, added by {!Index}.
    - [LabelName] is also added during index analysis, it holds the label name
      of a [FunDef]. *)
and annotation =
  | Loc of location
  | Depth of int
  | Type of ctype
  | Index of int
  | LabelName of string

(** Shorthand for annotation list used in definition of {!node}, should not be
    used elsewhere. *)
and ann = annotation list

(** Abstract Syntax Tree nodes. All attributes except for [ann] (annotations)
    are documented. [Program] to [Arg] are types returned by the parser, the
    rest of the types are annotations or transformations for convenience during
    later traversals. *)
and node =
  (* Global *)

  | Program of node list * ann
    (** list of declarations *)
  | GlobalDec of ctype * string * ann
    (** type, name [extern <type> <name>;] *)
  | GlobalDef of bool * ctype * string * node option * ann
    (** export, type, name, initialisation [<type> <name>[= <initialisation>];] *)
  | FunDec of ctype * string * node list * ann
    (** ret_type, name, params [extern <ret_type> <name>(<params>);] *)
  | FunDef of bool * ctype * string * node list * node * ann
    (** export, ret_type, name, params, body
        [[export ]<ret_type> <name>(<params>) <body>] *)
  | Param of ctype * string * ann
    (** type, name [<type> <name>] *)
  | Dim of string * ann
    (** name [<name>] Used in [ArrayDims] for [Param] of array type. *)

  | VarDecs of node list    (** Holds all [VarDec] nodes in a [FunDef]. *)
  | LocalFuns of node list  (** Holds all local [FunDef] nodes in a [FunDef]. *)

  (* Statements *)

  | VarDec of ctype * string * node option * ann
    (** type, name, initialisation [<type> <name>[= <initialisation>];] *)
  | Assign of string * node list option * node * ann
    (** name, indices, value [<name>[<indices>] = <value>;] *)
  | For of string * node * node * node * node * ann
    (** counter, start, stop, step, body
        [for (int <counter> = <start>, <stop>[, <step>]) <body>] *)
  | Return of node * ann
    (** Return value [return <value>;] *)
  | Expr of node
    (** Expression statement [<expr>;] *)
  | Block of node list
    (** Statement list enclosed in braces: body [{ <body> }] *)
  | If of node * node * ann
    (** condition, body [if (<condition>) <body>] *)
  | IfElse of node * node * node * ann
    (** condition, true_body, false_body [if (<condition>) <true_body> else <false_body>] *)
  | While of node * node * ann
    (** condition, body [while (<condition>) <body>] *)
  | DoWhile of node * node * ann
    (** condition, body [do <body> while (<condition>)] *)

  (* Expressions *)

  | Const of const * ann
    (** Constant value, e.g. [0] or [1.0] or [true]. *)
  | ArrayConst of node list * ann
    (** Array initialisation [[ <expr> , ... ]] *)
  | Var of string * node list option * ann
    (** name, indices [<name> | <name>[<indices>] ] *)
  | Monop of operator * node * ann
    (** operator, operand [<operator><operand>] *)
  | Binop of operator * node * node * ann
    (** operator, left, right [<left> <operator> <right>] *)
  | TypeCast of ctype * node * ann
    (** target_type, value [(target_type)value] *)
  | FunCall of string * node list * ann
    (** name, args [<name>(<args>)] *)
  | Arg of node
    (** Function argument wrapper. *)

  (* Additional types for convenience in traversals
   * Mostly used to annotate existing nodes with information from declarations *)

  | Allocate of node * node list * ann
    (** Array allocation: declaration, dimensions
        [<declaration> = __allocate(<dimensions>);] *)
  | VarUse of node * node list option * ann
    (** Replacement for [Var] with declaration. *)
  | FunUse of node * node list * ann
    (** Replacement for [FunCall] with declaration. *)
  | VarLet of node * node list option * node * ann
    (** Replacement for [Assign] with declaration. *)
  | ArrayInit of node * node list
    (** Wrapper for array initalisation with dimensions, used by {!Desug}. *)
  | Cond of node * node * node * ann
    (** cond, true_expr, false_expr [<cond> ? <true_expr> : <false_expr>]
        Used for short-circuit evaluation. *)
  | DummyNode
    (** Null node, pruned by {!Util.flatten_blocks}, which is called by
        {!Util.traverse_unit} in a traversal. *)

(** {2 Assembly instructions} *)

(** Stack scopes. Correspond to [(load|store|isr)[gln ]] respectively. *)
type stack_scope = Glob | Local | Rel of int | Current

(** Function scopes. *)
type rtn_scope = ExternFun of int | LocalFun of int * string

(** Assembly instructions. *)
type instr =
  | Comment of string                   (** [; <comment>] *)
  | InlineComment of instr * string     (** [<instr>  ; <comment>] *)
  | Label of string                     (** [<label>:] *)

  (* Directives *)
  | Export of string * ctype * ctype list * string
    (** [.export "<name>" <ret_type> [<arg_type_1> ...] <label>] *)
  | Import of string * ctype * ctype list
    (** [.import "<name>" <ret_type> [<arg_type_1> ...]] *)
  | ConstDef of const
    (** [.const <value>] *)
  | Global of ctype
    (** [.global <type>] *)

  | Store of ctype * stack_scope * int  (** [[ifba]store[ gn]] *)

  | Load of ctype * stack_scope * int   (** [[ifb]load[ gn] G] *)
  | LoadConst of ctype * int            (** [[ifb]loadc C] *)
  | LoadImm of const                    (** [[ifb]loadc_[01tf] | iloadc_m1] *)

  (* Operators ] *)
  | Op of operator * ctype
    (** [[ifb](add|sub|mul|div|rem|neg|not|ne|eq|lt|le|gt|ge)] *)
  | Convert of ctype * ctype            (** [i2f | f2i] *)
  | Inc of int * int                    (** [iinc L C] *)
  | Dec of int * int                    (** [idec L C] *)
  | IncOne of int                       (** [iinc_1 L] *)
  | DecOne of int                       (** [idec_1 L] *)

  (* Control flow *)
  | RtnInit of stack_scope              (** [isr[ lg] | isrn N] *)
  | RtnJmp of rtn_scope                 (** [jsr A O | jsre I] *)
  | RtnEnter of int                     (** [esr L] *)
  | Ret of ctype                        (** [[ ifb]return] *)
  | Branch of bool * string             (** [branch_[tf] O] *)
  | Jump of string                      (** [jump O] *)

  (* Stack management *)
  | Pop of ctype                        (** [[ifb]pop] *)

  (* Arrays *)
  | NewArray of ctype * int             (** [[ifb]newa D] *)
  | ArraySize of int                    (** [asize D] *)
  | LoadArray of ctype                  (** [[ifb]loada] *)
  | StoreArray of ctype                 (** [[ifb]storea] *)

  | EmptyLine  (** Empty line, added in between functions for readability. *)

(** {2 Global} *)

(** Intermediate representations, passed from phase to phase. *)
type intermediate =
  | Empty
    (** For phases without input/output (first and last phase). *)
  | FileContent of string * string
    (** Input file content. *)
  | Ast of node
    (** Abstract Syntax Tree. *)
  | Assembly of instr list
    (** List of assembly instructions. *)

(** Container for command-line arguments. *)
type args_record = {
  mutable infile   : string option;
  (** Input filename. *)
  mutable outfile  : string option;
  (** Output filename. *)
  mutable verbose  : int;
  (** Verbosity level. *)
  mutable cpp      : bool;
  (** Run C preprocessor? *)
  mutable optimize : bool;
  (** Run optimization phases? *)
  mutable endphase : string;
  (** Stop at the phase which has the given identifier (see {!Main.phases}). *)
}

(** {2 Exceptions} *)

type error_msg =
  (** General compilation error message (caught by main function). *)
  | Msg of string

  (** Error occurred at a certain location in an input file. Used in combination
      with {!Util.prerr_loc} and {!Util.prerr_loc_msg}. *)
  | LocMsg of location * string

  (** Error occurred at a certain AST node, use {!Util.locof} to get node
      location. *)
  | NodeMsg of node * string

  (** No error message, just quit *)
  | NoMsg

(** Raised error message. Makes the compiler print the message and quit with
    status code 1. *)
exception FatalError of error_msg

(** Catch-all error for traversals that accept a limit set of node types. *)
exception InvalidNode

(** Error raised when a phase receives an unsupported {!intermediate} type. *)
exception InvalidInput