parser.ml

open Location
open Ast

  let gram    = Grammar.create (Plexer.make ())
  let prog    = Grammar.Entry.create gram "prog"
  let expr    = Grammar.Entry.create gram "expression"
  let pat     = Grammar.Entry.create gram "type/pattern expression"
  let regexp  = Grammar.Entry.create gram "type/pattern regexp"
  let const   = Grammar.Entry.create gram "scalar constant"
 
  let rec multi_prod loc = function
    | [ x ] -> x
    | x :: l -> mk loc (Prod (x, multi_prod loc l))
    | [] -> assert false

  let rec tuple loc = function
    | [ x ] -> x
    | x :: l -> mk loc (Pair (x, tuple loc l))
    | [] -> assert false
	
  let char = mk noloc (Internal (Types.char Chars.any))
  let string = Star (Elem char)

  let cst_nil =  mk noloc (Cst (Types.Atom Sequence.nil_atom))

  let seq_of_string s =
    let rec aux accu i = if (i = 0) then accu else aux (s.[i-1]::accu) (i-1) in
    aux [] (String.length s)


  EXTEND
  GLOBAL: prog expr pat regexp const;

  prog: [
    [ l = LIST0 [ p = phrase; ";;" -> mk loc p ]; EOI -> l ]
  ];

  phrase: [
    [ e = expr -> EvalStatement e
    | "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t) ]
  ];

  expr: [
    "top" RIGHTA
    [ "match"; e = SELF; "with"; b = branches -> mk loc (Match (e,b))
    | "map"; e = SELF; "with"; b = branches -> mk loc (Map (e,b))
    | "transform"; e = SELF; "with"; b = branches -> 
	mk noloc (Op ("flatten", [mk loc (Map (e,b))]))
    | "fun"; f = OPT LIDENT; "("; a = LIST1 arrow SEP ";"; ")";
      b = branches -> 
	mk loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
    | (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        mk loc (Match (e1,[p,e2]))
    ]

    |
    [ LIDENT "flatten"; e = expr -> mk loc (Op ("flatten",[e]))
    | e1 = expr; e2 = expr -> mk loc (Apply (e1,e2))
    ]
    
    | 
    [ e1 = expr; "+"; e2 = expr -> mk loc (Op ("+",[e1;e2]))  
    | e1 = expr; "@"; e2 = expr -> mk loc (Op ("@",[e1;e2])) ]
    |
    [ e1 = expr; "*"; e2 = expr -> mk loc (Op ("*",[e1;e2]))  ]

    | "no_appl" 
    [ c = const -> mk loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
	let e = match e with Some e -> e | None -> cst_nil in
	let l = List.flatten l in
        tuple loc (l @ [e])
    | "<"; t = expr_tag_spec; a = expr_attrib_spec; ">"; c = expr ->
	tuple loc [t;a;c]
    | "{"; r = [ expr_record_spec | -> mk loc (RecordLitt []) ]; "}" -> r
    | "!"; t = pat -> mk loc (DebugTyper t)
    | a = LIDENT -> mk loc (Var a)
    ]

  ];

  seq_elem: [
    [ x = STRING -> 
	let s = seq_of_string (Token.eval_string x) in
	List.map 
	  (fun c -> mk loc (Cst (Types.Char (Chars.Unichar.from_char c))))
	  s 
    | e = expr LEVEL "no_appl" -> [e]
    ]
  ];
	  
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
    | "let"; "fun"; f = LIDENT; "("; a = LIST0 arrow SEP ";"; ")";
      b = branches -> 
        let p = mk loc (Capture f) in
	let abst = { fun_name = Some f; fun_iface = a; fun_body = b } in
        let e = mk loc (Abstraction abst) in
        (p,e)
    ] 
  ];

  arrow: [
    [ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
  ];

  branches: [
    [ OPT "|"; l = LIST1 branch SEP "|" ; OPT "end" -> l ]
  ];

  branch: [
    [ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
  ];

	  
  regexp: [ 
    [ x = regexp; "|"; y = regexp -> Alt (x,y) ]
  | [ x = regexp; y = regexp -> Seq (x,y) ]
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (a,x) ]
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
    | x = regexp; "?" -> Alt (x, Epsilon)
    | x = regexp; "??" -> Alt (Epsilon, x) ]
  | [ "("; x = regexp; ")" -> x
    | UIDENT "String" -> string
    | s = STRING ->
	let s = seq_of_string (Token.eval_string s) in
	List.fold_right
	  (fun c accu -> 
	     let c = Chars.Unichar.from_char c in
	     let c = Chars.atom c in
	     Seq (Elem (mk loc (Internal (Types.char c))), accu))
	  s
	  Epsilon
    | e = pat LEVEL "simple" -> Elem e
    ]
  ];

  pat: [ 
      [ x = pat; "where"; 
        b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)] SEP "and"
            -> mk loc (Recurs (x,b)) ]
    | RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
    | "no_arrow" [ x = pat; "|"; y = pat -> mk loc (Or (x,y)) ] 
    | "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y)) 
      | x = pat; "-"; y = pat -> mk loc (Diff (x,y)) ]
    | 
      [ "{"; r = record_spec; "}" -> r
      | LIDENT "_" -> mk loc (Internal Types.any)
      | a = LIDENT -> mk loc (Capture a)
      | "("; a = LIDENT; ":="; c = const; ")" -> mk loc (Constant (a,c))
      | a = UIDENT -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
          let i = Big_int.big_int_of_string i 
	  and j = Big_int.big_int_of_string j in
          mk loc (Internal (Types.interval i j))
      | i = char ; "--"; j = char ->
          mk loc (Internal (Types.char (Chars.char_class i j)))
      | c = const -> mk loc (Internal (Types.constant c))
      | "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
      | "["; r = [ r = regexp -> r | -> Epsilon ];
             q = [ ";"; q = pat -> q 
                 | -> mk noloc (Internal (Sequence.nil_type)) ]; 
             "]" -> mk loc (Regexp (r,q))
      | "<"; t = tag_spec; a = attrib_spec; ">"; c = pat ->
          multi_prod loc [t;a;c]
      ]
    
  ];

  record_spec:
    [ [ r = LIST0 [ l = [LIDENT | UIDENT]; 
                  o = ["=?" -> true | "=" -> false]; 
                  x = pat ->
                    mk loc (Record (Types.label l,o,x))
                ] SEP ";" ->
        match r with
          | [] -> mk loc (Internal Types.Record.any)
          | h::t -> List.fold_left (fun t1 t2 -> mk loc (And (t1,t2))) h t
      ] ];
  
  char:
    [ 
      [ c = CHAR -> Chars.Unichar.from_char (Token.eval_char c)
      | "!"; i = INT -> Chars.Unichar.from_int (int_of_string i) ]
    ];
     

  const:
    [ 
      [ i = INT -> Types.Integer (Big_int.big_int_of_string i)
      | "`"; a = [LIDENT | UIDENT] -> Types.Atom (Types.mk_atom a)
      | c = char -> Types.Char c ]
    ];

  tag_spec:
    [
      [ LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any))) ]
    | [ a = [LIDENT | UIDENT] -> 
	  mk loc (Internal (Types.atom (Atoms.atom (Types.mk_atom a)))) ]
    | [ t = pat -> t ]
    ];

  attrib_spec:
    [ [ r = record_spec -> r | "("; t = pat; ")" -> t ] ];

  expr_record_spec:
    [ [ r = LIST1
	      [ l = [LIDENT | UIDENT]; "="; x = expr -> (Types.label l,x) ] 
	      SEP ";" ->
	  mk loc (RecordLitt r)
      ] ];
  
  expr_tag_spec:
    [
      [ a = [LIDENT | UIDENT] -> 
	  mk loc (Cst (Types.Atom (Types.mk_atom a))) ]
    | [ e = expr LEVEL "no_appl" -> e ]
    ];

  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
      | -> mk loc (RecordLitt []) 
      ] 
    ];
END

let pat = Grammar.Entry.parse pat
let expr = Grammar.Entry.parse expr
let prog = Grammar.Entry.parse prog

module From_string = struct
  let pat s = pat (Stream.of_string s)
  let expr s = expr (Stream.of_string s)
end