parser.ml

open Location
open Ast
open Ident
open Printf

(*
let ()  = Grammar.error_verbose := true
*)

exception Error of string
let error (i,j) s = Location.raise_loc i j (Error s)

let gram    = Grammar.gcreate (Wlexer.lexer Wlexer.token Wlexer.latin1_engine)

let true_atom = Atoms.mk_ascii "true"
let false_atom = Atoms.mk_ascii "false"
let true_type = Types.atom (Atoms.atom true_atom)
let false_type = Types.atom (Atoms.atom false_atom)

let parse_ident = U.mk_latin1

let id_dummy = ident (U.mk "$$$")

(* TODO: NS *)
let label = parse_ident
let ident s = ident (parse_ident s)

let prog    = Grammar.Entry.create gram "prog"
let top_phrases   = Grammar.Entry.create gram "toplevel phrases"
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 exp pos e = LocatedExpr (loc_of_pos pos,e)

let rec multi_prod loc = function
  | [ x ] -> x
  | x :: l -> mk loc (Prod (x, multi_prod loc l))
  | [] -> assert false
      
let rec tuple = function
  | [ x ] -> x
  | x :: l -> Pair (x, tuple l)
  | [] -> assert false

let tuple_queue = 
  List.fold_right (fun x q -> Pair (x, q))


let char = mknoloc (Internal (Types.char Chars.any))
let string_regexp = Star (Elem char)
	       
let cst_nil =  Cst (Const_internal (Types.Atom Sequence.nil_atom))
		 
let seq_of_string s =
  let s = Encodings.Utf8.mk s in
  let rec aux i j =
    if Encodings.Utf8.equal_index i j then []
    else let (c,i) = Encodings.Utf8.next s i in c :: (aux i j)
  in
  aux (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s)


let parse_char loc s =
  match seq_of_string s with
    | [ c ] -> c
    | _ -> error loc "Character litteral must have length 1"

let include_stack = ref []

let protect_exn f g =
  try let x = f () in g (); x
  with e -> g (); raise e

let is_fun_decl =
   Grammar.Entry.of_parser gram "[is_fun_decl]"
     (fun strm ->
       match Stream.npeek 3 strm with
	 | [ ("", "fun"); ("LIDENT", _); ("", "(") ]
	 | [ ("LIDENT", _) ; ("", "(") ; _ ] -> ()
	 | _ -> raise Stream.Failure
     )

let dot_RE = Pcre.regexp "\\."

EXTEND
  GLOBAL: top_phrases prog expr pat regexp const;

  top_phrases: [
    [ l = LIST0 phrase; ";;" -> List.flatten l  ]
  ];

  prog: [
    [ l = LIST0 [ p = phrase ; OPT ";;" -> p ]; EOI -> List.flatten l ]
  ];

  uident: [ [ x = UIDENT -> parse_ident x ] ];

  phrase: [
    [ (f,p,e) = let_binding -> 
	if f then [ mk loc (FunDecl e) ] else
	  [ mk loc (LetDecl (p,e)) ]
    | (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        [ mk loc (EvalStatement (exp loc (Match (e1,[p,e2])))) ]
    | "type"; x = uident; "="; t = pat -> [ mk loc (TypeDecl (x,t)) ]
    | "type"; x = LIDENT -> error loc "Type identifiers must be capitalized"
    | "schema"; name = UIDENT; "="; uri = STRING2 ->
	protect_op "schema";
        let schema_doc = Schema_xml.pxp_tree_of uri in
        let schema = Schema_parser.parse_schema schema_doc in
        [ mk loc (SchemaDecl (name, schema))]
    | (name,ns) = namespace_binding ->
	[ mk loc (Namespace (name, ns)) ]
    | (name,ns) = namespace_binding; "in"; e2 = expr LEVEL "top" ->
	let e = exp loc (NamespaceIn (name, ns, e2)) in
        [ mk loc (EvalStatement (exp loc e)) ]
    | "debug"; d = debug_directive -> [ mk loc (Debug d) ]
    | DIRECTIVE "#quit" -> [ mk loc (Directive `Quit) ]
    | DIRECTIVE "#env" -> [ mk loc (Directive `Env) ]
    | DIRECTIVE "#reinit_ns" -> [ mk loc (Directive `Reinit_ns) ]
    | "include"; s = STRING2 ->
	let s = 
	  if Filename.is_relative s 
	  then Filename.concat (Location.current_dir ()) s
	  else s in
	protect_op "File inclusion";
	(* avoid looping; should issue an error ? *)
	(* it is possible to have looping with x/../x/../x/.. ....
	   Need to canonicalize filename *)
	if List.mem s !include_stack then [] 
	else (
	  include_stack := s :: !include_stack;
	  Location.push_source (`File s);
	  protect_exn
	    (fun () ->
	       let chan = open_in s in
	       protect_exn
		 (fun () ->
		    let input = Stream.of_channel chan in
		    Grammar.Entry.parse prog input)
		 (fun () -> close_in chan))
	    (fun () ->
	       Location.pop_source ();
	       include_stack := List.tl !include_stack)
	)
    ] | 
    [ e = expr -> [ mk loc (EvalStatement e) ]
    ]
  ];

  debug_directive: [
    [ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
    | LIDENT "accept"; p = pat -> `Accept p
    | LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
    | LIDENT "sample"; t = pat -> `Sample t
    | LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
    ]
  ];

  keyword: [
    [ a = 
	[ "map" | "match" | "with" | "try" | "xtransform"
	| "if" | "then"	| "else"
	| "transform" | "fun" | "in"
	| "let" | "type" | "debug" | "include"
        | "and" | "validate" | "schema" | "namespace"
	]
	-> a
    ]
  ];

  expr: [
    "top" RIGHTA
    [ "match"; e = SELF; "with"; b = branches -> 
	exp loc (Match (e,b))
    | "try"; e = SELF; "with"; b = branches -> 
	exp loc (Try (e,b))
    | "map"; e = SELF; "with"; b = branches -> 
	exp loc (Map (e,b))
    | "xtransform"; e = SELF; "with"; b = branches -> 
	exp loc (Xtrans (e,b))
    | "if"; e = SELF; "then"; e1 = SELF; "else"; e2 = SELF ->
	let p1 = mk loc (Internal true_type)
	and p2 = mk loc (Internal false_type) in
	exp loc (Match (e, [p1,e1; p2,e2]))
    | "transform"; e = SELF; "with"; b = branches -> 
	exp loc (Transform (e,b))
    | "validate"; e = SELF; "with"; schema = UIDENT; "#";
      typ = [ UIDENT | LIDENT | keyword ] ->
        exp loc (Validate (e, schema, typ))
    | "fun"; (f,a,b) = fun_decl ->
	exp loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
    | (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        exp loc (Match (e1,[p,e2]))
    | (name,ns) = namespace_binding; "in"; e2 = expr LEVEL "top" ->
	exp loc (NamespaceIn (name, ns, e2))
    | e = expr; ":"; p = pat ->
	exp loc (Forget (e,p))
    ]

    
    | 
    [ e1 = expr; op = ["=" | "<=" | "<<" | ">>" | ">=" ]; e2 = expr -> 
	let op = match op with
	  | "<<" -> "<"
	  | ">>" -> ">"
	  | s -> s in
	exp loc (Op (op,[e1;e2]))
    ]

    | 
    [ e1 = expr; op = ["+" | "-" | "@" ]; e2 = expr -> 
	exp loc (Op (op,[e1;e2]))
    | e = expr; "\\"; l = [LIDENT | UIDENT | keyword ] -> 
	exp loc (RemoveField (e, label l)) 
    ]
    |
    [ e1 = expr; op = ["*"]; e2 = expr -> exp loc (Op (op,[e1;e2]))  
    | e = expr; op = "/"; p = pat ->

	let tag = mk loc (Internal (Types.atom (Atoms.any))) in
	let att = mk loc (Internal Types.Record.any) in
	let any = mk loc (Internal (Types.any)) in
	let re = Star(Alt(SeqCapture(id_dummy,Elem p), Elem any)) in
	let ct = mk loc (Regexp (re,any)) in
        let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
	let b = (p,  Var id_dummy) in
	exp loc (Transform (e,[b]))
    ]
    |
    [ e = expr;  "."; l = [LIDENT | UIDENT | keyword ] -> 
	exp loc (Dot (e, label l)) 
    ]  
 
    |
    [ op = [ LIDENT "flatten"
	   | LIDENT "load_xml"
	   | LIDENT "load_file" | LIDENT "load_file_utf8"
	   | LIDENT "load_html"
	   | LIDENT "print_xml" | LIDENT "print_xml_utf8"
	   | LIDENT "print"
	   | LIDENT "int_of"
	   | LIDENT "string_of"
	   | LIDENT "atom_of"
	   | LIDENT "raise"
	   ]; 
      e = expr -> exp loc (Op (op,[e]))
    | op = [ LIDENT "dump_to_file" | LIDENT "dump_to_file_utf8" ];
      e1 = expr LEVEL "no_appl"; e2 = expr -> exp loc (Op (op, [e1;e2])) 
    | e1 = SELF; LIDENT "div"; e2 = expr -> exp loc (Op ("/", [e1;e2]))
    | e1 = SELF; LIDENT "mod"; e2 = expr -> exp loc (Op ("mod", [e1;e2]))
    | e1 = SELF; e2 = expr -> exp loc (Apply (e1,e2))
    ]

    | "no_appl" 
    [ c = const -> exp loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> exp loc (tuple 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.fold_right 
		  (fun x q ->
		     match x with
		       | `String (loc,i,j,s) -> exp loc (String (i,j,s,q))
		       | `Elems l -> tuple_queue l q
		       | `Explode x -> Op ("@",[x;q])
		  ) l e
	in
	exp loc l
    | "<"; t = [ "("; e = expr; ")" -> e
	       | a = tag -> exp loc (Cst a)
	       ];
	a = expr_attrib_spec; ">"; c = expr ->
(*	  let t = Pair (cst_nil, t) in *)
	  exp loc (Xml (t, Pair (a,c)))
    | "{"; r = [ expr_record_spec | -> exp loc (RecordLitt []) ]; "}" -> r
    | s = STRING2 ->
	let s = U.mk s in
	exp loc (String (U.start_index s, U.end_index s, s, cst_nil))
    | a = LIDENT -> exp loc (Var (ident a))
    ]

  ];

  tag: [ [ a = [ LIDENT | UIDENT | keyword ] -> 
	     Const_atom (parse_ident a) ] ];

  tag_type: [
    [ LIDENT "_" ->  mk loc (Internal (Types.atom Atoms.any))
    | a = [ LIDENT | UIDENT | keyword ] -> mk loc (AtomT (parse_ident a))
    | t = ANY_IN_NS -> mk loc (NsT (parse_ident t)) 
    ]
  ];

  seq_elem: [
    [ x = STRING1 -> 
	let s = U.mk x in
	`String (loc, U.start_index s, U.end_index s, s)
    | e = expr LEVEL "no_appl" -> `Elems [e]
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
    ]
  ];
	
  namespace_binding: [
    [ "namespace"; 
	name = [ name = [ UIDENT | LIDENT | keyword ]; "=" ->
		   parse_ident name
	       | -> U.mk "" ];
	uri = STRING2 ->
	  let ns = Ns.mk (parse_ident uri) in
	  (name,ns)
    ]
  ];

  
  let_binding: [
    [ "let"; is_fun_decl; OPT "fun"; (f,a,b) = fun_decl ->
	let f = match f with Some x -> x | None -> assert false in
	let p = mk loc (Capture f) in
	let abst = { fun_name = Some f; fun_iface = a; fun_body = b } in
        let e = exp loc (Abstraction abst) in
        (true,p,e)
    | "let"; p = pat; "="; e = expr -> (false,p,e)
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (false,p, Forget (e,t))
    ] 
  ];

 fun_decl_after_lparen: [
(* need an hack to do this, because both productions would
   match   [ OPT LIDENT; "("; pat ] .... *)
   [ p1 = pat LEVEL "no_arrow";
     res = [ "->"; p2 = pat;
	     a = [ ";"; a = LIST0 arrow SEP ";" -> a | -> [] ];
	     ")"; b = branches -> `Classic (p2,a,b)
	   | ":"; targ1 = pat;
	     args = LIST0 [ ","; arg = pat; ":"; targ = pat -> (arg,targ) ]; 
	     ")"; ":"; tres = pat ; 
	     "="; body = expr ->
	       `Compact (targ1,args,tres,body)
	   ] ->
       match res with
	 | `Classic (p2,a,b) -> (p1,p2)::a,b
	 | `Compact (targ1,args,tres,body) ->
	     let args = (p1,targ1) :: args in
	     let targ = multi_prod nopos (List.map snd args) in
	     let arg = multi_prod nopos (List.map fst args) in
	     let b = [arg, body] in
	     let a = [targ,tres] in
	     (a,b) ] ];


 fun_decl: [
   [ f = OPT [ x = LIDENT -> ident x]; "("; (a,b) = fun_decl_after_lparen ->
	    (f,a,b)
   ]
 ];

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

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

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

	  
  regexp: [ 
    [ x = regexp; "|"; y = regexp -> 
	match (x,y) with
	  | Elem x, Elem y -> Elem (mk loc (Or (x,y)))
	  | _ -> Alt (x,y) 
    ]
  | [ x = regexp; y = regexp -> Seq (x,y) ]
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident 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
    | "("; a = LIDENT; ":="; c = const; ")" -> 
	Elem (mk loc (Constant ((ident a,c))))
    | UIDENT "PCDATA" -> string_regexp
    | i = STRING1; "--"; j = STRING1 ->
	let i = Chars.mk_int (parse_char loc i)
	and j = Chars.mk_int (parse_char loc j) in
        Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
    |  s = STRING1 ->
	List.fold_right
	  (fun c accu -> 
	     let c = Chars.mk_int c in
	     let c = Chars.atom c in
	     Seq (Elem (mknoloc (Internal (Types.char c))), accu))
	  (seq_of_string s)
	  Epsilon 
    | e = pat LEVEL "simple" -> Elem e
    ]
  ];

  pat: [ 
      [ x = pat; LIDENT "where"; 
        b = LIST1 [ a = uident; "="; y = pat -> (a,y)
		  | LIDENT -> error loc "Type/pattern identifiers must be capitalized"
		  ] 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; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
      | LIDENT "_" -> mk loc (Internal Types.any)
      | a = LIDENT -> mk loc (Capture (ident a))
      | "("; a = LIDENT; ":="; c = const; ")" -> 
	  mk loc (Constant (ident a,c))
      | schema = UIDENT; "#"; typ = [ UIDENT | LIDENT | keyword ];
        k = OPT [ "as"; k = [ "element" | "type" | "attribute" ] -> k ] ->
          let kind =
            match k with
            | None -> `Any
            | Some "element" -> `Element
            | Some "type" -> `Type
            | Some "attribute" -> `Attribute
            | _ -> assert false
          in
          mk loc (SchemaVar (kind, schema, typ))
      | a = uident -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
          let i =  Intervals.mk i 
	  and j =  Intervals.mk j in
          mk loc (Internal (Types.interval (Intervals.bounded i j)))
      | i = INT -> 
          let i =  Intervals.mk i  in
          mk loc (Internal (Types.interval (Intervals.atom i)))
      | "*"; "--"; j = INT ->
	  let j =  Intervals.mk j in
          mk loc (Internal (Types.interval (Intervals.left j)))
      | i = INT; "--"; "*" ->
	  let i = Intervals.mk i in
          mk loc (Internal (Types.interval (Intervals.right i)))
      | i = char ->
          mk loc (Internal (Types.char (Chars.char_class i i)))
      | i = char ; "--"; j = char ->
          mk loc (Internal (Types.char (Chars.char_class i j)))
      | "`"; c = tag_type -> c
      | c = const -> 
	  (match c with
	     | Const_atom l -> mk loc (AtomT l)
	     | Const_internal c -> mk loc (Internal (Types.constant c))
	  )
      | "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
      | "["; r = [ r = regexp -> r | -> Epsilon ];
             q = [ ";"; q = pat -> q 
                 | -> mknoloc (Internal (Sequence.nil_type)) ]; 
             "]" -> mk loc (Regexp (r,q))
      | "<"; t =
            [ x = tag_type -> x
            | "("; t = pat; ")" -> t ];
	a = attrib_spec; ">"; c = pat ->
(*	  let t = mk loc (Prod (mk loc (Internal Sequence.nil_type), t)) in *)
          mk loc (XmlT (t, multi_prod loc [a;c]))
      | s = STRING2 ->
	  let s = 
	    List.map 
	      (fun c -> 
		 mknoloc (Internal
			     (Types.char
				(Chars.atom
				   (Chars.mk_int c))))) 
	      (seq_of_string s) in
	  let s = s @ [mknoloc (Internal (Sequence.nil_type))] in
	  multi_prod loc s
      ]
    
  ];

  record_spec:
    [ [ r = LIST0 [ l = [LIDENT | UIDENT | keyword ]; "=";
                  o = [ "?" -> true | -> false]; 
                  x = pat -> 
		    let x = if o then mk loc (Optional x) else x in
		    (label l, x)
                ] SEP ";" ->
	  r
      ] ];
  
  char:
    [ 
      [ c = STRING1 -> Chars.mk_int (parse_char loc c) ]
    ];
     

  const:
    [ 
      [ i = INT -> Const_internal (Types.Integer (Intervals.mk i))
      | "`"; a = tag -> a
      | c = char -> Const_internal (Types.Char c) ]
    ];


  attrib_spec:
    [ [ r = record_spec -> mk loc (Record (true,r)) 
      | "("; t = pat; ")" -> t 
      | "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
      ] ];

  expr_record_spec:
    [ [ r = LIST1
	      [ l = [LIDENT | UIDENT | keyword ]; "="; x = expr -> 
		  (label l,x) ] 
	      SEP ";" ->
	  exp loc (RecordLitt r)
      ] ];
  
  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
      | -> exp loc (RecordLitt []) 
      ] 
    ];
END

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

let sync () = 
  match !Wlexer.lexbuf with
    | None -> ()
    | Some lb ->
	let rec aux () =
	  match !Wlexer.last_tok with
	    | ("",";;") | ("EOI","") -> ()
	    | _ -> 
		Wlexer.last_tok := Wlexer.token Wlexer.latin1_engine lb; 
		aux ()
	in
	aux ()