viewcvs.cgi/parser/parser.ml

open Location
open Ast
open Ident

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

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


let parse_ident = Encodings.Utf8.mk_latin1

let label s = LabelPool.mk (parse_ident s)

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 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 (Types.Atom Sequence.nil_atom)
                 
let seq_of_string pos s =
  let s =  Encodings.Utf8.mk s in
  let (pos,_) = pos in
  let rec aux pos i j =
    if Encodings.Utf8.equal_index i j then []
    else
      let (len,i) = Encodings.Utf8.next s i in
      let (c,i) = Encodings.Utf8.next s i in
      ((pos,pos+len),c) :: (aux (pos + len) i j)
  in
  aux (pos+1) (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s)

let get_string s =
  let s = Encodings.Utf8.mk s in
  let b = Buffer.create 32 in
  let rec aux i j =
    if Encodings.Utf8.equal_index i j then ()
    else
      let (len,i) = Encodings.Utf8.next s i in
      let (c,i) = Encodings.Utf8.next s i in
      Encodings.Utf8.store b c;
      aux i j
  in
  aux (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s);
  Buffer.contents b

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

let make_record loc r =
  LabelMap.from_list (fun _ _ -> error loc "Duplicated record field") r

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

let char_list loc s = 
  let s = seq_of_string loc s in
  List.map (fun (loc,c) -> exp loc (Cst (Types.Char (Chars.mk_int c)))) s


let include_stack = ref []

EXTEND
  GLOBAL: prog expr pat regexp const;

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

  phrase: [
    [ (p,e) = let_binding -> [ mk loc (LetDecl (p,e)) ]
    | (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        [ mk loc (EvalStatement (exp loc (Match (e1,[p,e2])))) ]
    | LIDENT "type"; x = UIDENT; "="; t = pat -> [ mk loc (TypeDecl (x,t)) ]
    | LIDENT "type"; x = LIDENT -> [ error loc "Type identifiers must be capitalized" ]
    | LIDENT "debug"; d = debug_directive -> [ mk loc (Debug d) ]
    | LIDENT "include"; s = STRING2 ->
        let s = get_string s in
        protect_op "File inclusion";
        (* avoid looping; should issue an error ? *)
        if List.mem s !include_stack then [] 
        else (
          include_stack := s :: !include_stack;
          let chan = open_in s in
          Location.push_source (`File s);
          let input = Stream.of_channel chan in
          let l = Grammar.Entry.parse prog input in
          close_in chan;
          Location.pop_source ();
          include_stack := List.tl !include_stack;
          l
        )
    ] | 
    [ 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 "normal_record"; t = pat -> `Normal_record t
    | LIDENT "compile2"; t = pat; p = LIST1 pat -> `Compile2 (t,p)
    | LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
    ]
  ];

  keyword: [
    [ a = 
        [ "map" | "match" | "with" | "try" | "xtransform"
        | "if" | "then" | "else"
        | "transform" | "fun" | "in"
        | "let"
        ]
        -> a
    ]
  ];

  expr: [
    "top" RIGHTA
    [ "match"; e = SELF; "with"; b = branches -> exp loc (Match (e,b))
    | "try"; e = SELF; "with"; b = branches -> 
        let id = ident (U.mk "x") in
        let default = 
          mknoloc (Capture id), 
          Op ("raise",[Var id]) in
        exp loc (Try (e,b@[default]))
    | "map"; e = SELF; "with"; b = branches -> exp loc (Map (false,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 (Builtin.true_type))
        and p2 = mk loc (Internal (Builtin.false_type)) in
        exp loc (Match (e, [p1,e1; p2,e2]))
    | "transform"; e = SELF; "with"; b = branches -> 
        exp loc (Op ("flatten", [Map (true,e,b)]))
    | "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]))
    | 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 id = ident (U.mk "x") in
        let re = Star(Alt(SeqCapture(id,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) in
        exp loc (Op ("flatten", [Map (false,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 "raise" 
           | LIDENT "int_of"
           | LIDENT "string_of"
           | LIDENT "atom_of"
           ]; 
      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
                       | `Elems l -> tuple_queue l q
                       | `Explode x -> Op ("@",[x;q])
                  ) l e
        in
        exp loc l
    | "<"; t = [ "("; e = expr; ")" -> e
               | a = [ LIDENT | UIDENT | keyword ] -> 
                   let a = parse_ident a in
                   exp loc (Cst (Types.Atom (Atoms.mk a))) ];
        a = expr_attrib_spec; ">"; c = expr ->
        exp loc (Xml (t, Pair (a,c)))
    | "{"; r = [ expr_record_spec | -> exp loc (RecordLitt LabelMap.empty) ]; "}" -> r
    | s = STRING2 ->
        exp loc (tuple (char_list loc s @ [cst_nil]))
    | a = LIDENT -> exp loc (Var (ident (parse_ident a)))
    ]

  ];

  seq_elem: [
    [ x = STRING1 -> `Elems (char_list loc x)
    | e = expr LEVEL "no_appl" -> `Elems [e]
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
    ]
  ];
          
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, Forget (e,t))
    | "let"; "fun"; (f,a,b) = fun_decl ->
        let p = match f with 
          | Some x -> mk loc (Capture x)
          | _ -> failwith "Function name mandatory in let fun declarations"
        in
        let abst = { fun_name = f; fun_iface = a; fun_body = b } in
        let e = exp loc (Abstraction abst) in
        (p,e);
    ] 
  ];

 fun_decl: [
(* need an hack to do this, because both productions would
   match   [ OPT LIDENT; "("; pat ] .... *)
   [ f = OPT [ x = LIDENT -> ident (parse_ident x)]; "("; 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) -> f,(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
            (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 (parse_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 (parse_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 ->
        let s = seq_of_string loc s in
        List.fold_right
          (fun (loc,c) accu -> 
             let c = Chars.mk_int 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; 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 (parse_ident a)))
      | "("; a = LIDENT; ":="; c = const; ")" -> 
          mk loc (Constant (ident (parse_ident a),c))
      | 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 = const -> 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 = [ LIDENT | UIDENT | keyword ] ->
            let a = if x = "_" then Atoms.any else Atoms.atom (Atoms.mk (parse_ident x)) in
            mk loc (Internal (Types.atom a))
        | "("; t = pat; ")" -> t ];
        a = attrib_spec; ">"; c = pat ->
          mk loc (XmlT (t, multi_prod loc [a;c]))
      | s = STRING2 ->
          let s = seq_of_string loc s in
          let s = List.map 
                    (fun (loc,c) -> 
                       mk loc (Internal
                                 (Types.char
                                    (Chars.atom
                                       (Chars.mk_int c))))) s in
          let s = s @ [mk loc (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 ";" ->
          make_record loc r
      ] ];
  
  char:
    [ 
      [ c = STRING1 -> Chars.mk_int (parse_char loc c) ]
    ];
     

  const:
    [ 
      [ i = INT -> Types.Integer (Intervals.mk i)
      | "`"; a = [LIDENT | UIDENT | keyword ] -> Types.Atom (Atoms.mk (parse_ident a))
      | c = char -> 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 (make_record loc r))
      ] ];
  
  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
      | -> exp loc (RecordLitt (LabelMap.empty)) 
      ] 
    ];
END

let pat'  = Grammar.Entry.create gram "type/pattern expression"
EXTEND GLOBAL: pat pat';
  pat': [  [ p = pat; EOI -> p ] ];
END

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


module From_string = struct
  let pat s = Grammar.Entry.parse pat' (Stream.of_string s)
  let expr s = expr (Stream.of_string s)
end

1	open Location
2	open Ast
3	open Ident
4
5	(*
6	let () = Grammar.error_verbose := true
7	*)
8
9	let gram = Grammar.gcreate (Wlexer.lexer Wlexer.token Wlexer.latin1_engine)
10
11
12	let parse_ident = Encodings.Utf8.mk_latin1
13
14	let label s = LabelPool.mk (parse_ident s)
15
16	let prog = Grammar.Entry.create gram "prog"
17	let expr = Grammar.Entry.create gram "expression"
18	let pat = Grammar.Entry.create gram "type/pattern expression"
19	let regexp = Grammar.Entry.create gram "type/pattern regexp"
20	let const = Grammar.Entry.create gram "scalar constant"
21
22	let exp pos e = LocatedExpr (loc_of_pos pos,e)
23
24	let rec multi_prod loc = function
25	\| [ x ] -> x
26	\| x :: l -> mk loc (Prod (x, multi_prod loc l))
27	\| [] -> assert false
28
29	let rec tuple = function
30	\| [ x ] -> x
31	\| x :: l -> Pair (x, tuple l)
32	\| [] -> assert false
33
34	let tuple_queue =
35	List.fold_right (fun x q -> Pair (x, q))
36
37
38	let char = mknoloc (Internal (Types.char Chars.any))
39	let string_regexp = Star (Elem char)
40
41	let cst_nil = Cst (Types.Atom Sequence.nil_atom)
42
43	let seq_of_string pos s =
44	let s = Encodings.Utf8.mk s in
45	let (pos,_) = pos in
46	let rec aux pos i j =
47	if Encodings.Utf8.equal_index i j then []
48	else
49	let (len,i) = Encodings.Utf8.next s i in
50	let (c,i) = Encodings.Utf8.next s i in
51	((pos,pos+len),c) :: (aux (pos + len) i j)
52	in
53	aux (pos+1) (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s)
54
55	let get_string s =
56	let s = Encodings.Utf8.mk s in
57	let b = Buffer.create 32 in
58	let rec aux i j =
59	if Encodings.Utf8.equal_index i j then ()
60	else
61	let (len,i) = Encodings.Utf8.next s i in
62	let (c,i) = Encodings.Utf8.next s i in
63	Encodings.Utf8.store b c;
64	aux i j
65	in
66	aux (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s);
67	Buffer.contents b
68
69	exception Error of string
70	let error (i,j) s = Location.raise_loc i j (Error s)
71
72	let make_record loc r =
73	LabelMap.from_list (fun _ _ -> error loc "Duplicated record field") r
74
75	let parse_char loc s =
76	let s = seq_of_string loc s in
77	match s with
78	\| [ loc,c ] -> c
79	\| _ -> error loc "Character litteral must have length 1"
80
81	let char_list loc s =
82	let s = seq_of_string loc s in
83	List.map (fun (loc,c) -> exp loc (Cst (Types.Char (Chars.mk_int c)))) s
84
85
86	let include_stack = ref []
87
88	EXTEND
89	GLOBAL: prog expr pat regexp const;
90
91	prog: [
92	[ l = LIST0 [ p = phrase; ";;" -> p ]; EOI -> List.flatten l ]
93	];
94
95	phrase: [
96	[ (p,e) = let_binding -> [ mk loc (LetDecl (p,e)) ]
97	\| (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
98	[ mk loc (EvalStatement (exp loc (Match (e1,[p,e2])))) ]
99	\| LIDENT "type"; x = UIDENT; "="; t = pat -> [ mk loc (TypeDecl (x,t)) ]
100	\| LIDENT "type"; x = LIDENT -> [ error loc "Type identifiers must be capitalized" ]
101	\| LIDENT "debug"; d = debug_directive -> [ mk loc (Debug d) ]
102	\| LIDENT "include"; s = STRING2 ->
103	let s = get_string s in
104	protect_op "File inclusion";
105	(* avoid looping; should issue an error ? *)
106	if List.mem s !include_stack then []
107	else (
108	include_stack := s :: !include_stack;
109	let chan = open_in s in
110	Location.push_source (`File s);
111	let input = Stream.of_channel chan in
112	let l = Grammar.Entry.parse prog input in
113	close_in chan;
114	Location.pop_source ();
115	include_stack := List.tl !include_stack;
116	l
117	)
118	] \|
119	[ e = expr -> [ mk loc (EvalStatement e) ]
120	]
121	];
122
123	debug_directive: [
124	[ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
125	\| LIDENT "accept"; p = pat -> `Accept p
126	\| LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
127	\| LIDENT "normal_record"; t = pat -> `Normal_record t
128	\| LIDENT "compile2"; t = pat; p = LIST1 pat -> `Compile2 (t,p)
129	\| LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
130	]
131	];
132
133	keyword: [
134	[ a =
135	[ "map" \| "match" \| "with" \| "try" \| "xtransform"
136	\| "if" \| "then" \| "else"
137	\| "transform" \| "fun" \| "in"
138	\| "let"
139	]
140	-> a
141	]
142	];
143
144	expr: [
145	"top" RIGHTA
146	[ "match"; e = SELF; "with"; b = branches -> exp loc (Match (e,b))
147	\| "try"; e = SELF; "with"; b = branches ->
148	let id = ident (U.mk "x") in
149	let default =
150	mknoloc (Capture id),
151	Op ("raise",[Var id]) in
152	exp loc (Try (e,b@[default]))
153	\| "map"; e = SELF; "with"; b = branches -> exp loc (Map (false,e,b))
154	\| "xtransform"; e = SELF; "with"; b = branches -> exp loc (Xtrans (e,b))
155	\| "if"; e = SELF; "then"; e1 = SELF; "else"; e2 = SELF ->
156	let p1 = mk loc (Internal (Builtin.true_type))
157	and p2 = mk loc (Internal (Builtin.false_type)) in
158	exp loc (Match (e, [p1,e1; p2,e2]))
159	\| "transform"; e = SELF; "with"; b = branches ->
160	exp loc (Op ("flatten", [Map (true,e,b)]))
161	\| "fun"; (f,a,b) = fun_decl ->
162	exp loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
163	\| (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
164	exp loc (Match (e1,[p,e2]))
165	\| e = expr; ":"; p = pat ->
166	exp loc (Forget (e,p))
167	]
168
169
170	\|
171	[ e1 = expr; op = ["=" \| "<=" \| "<<" \| ">>" \| ">=" ]; e2 = expr ->
172	let op = match op with
173	\| "<<" -> "<"
174	\| ">>" -> ">"
175	\| s -> s in
176	exp loc (Op (op,[e1;e2]))
177	]
178
179	\|
180	[ e1 = expr; op = ["+" \| "-" \| "@" ]; e2 = expr ->
181	exp loc (Op (op,[e1;e2]))
182	\| e = expr; "\\"; l = [LIDENT \| UIDENT \| keyword ] ->
183	exp loc (RemoveField (e, label l))
184	]
185	\|
186	[ e1 = expr; op = ["*"]; e2 = expr -> exp loc (Op (op,[e1;e2]))
187	\| e = expr; op = "/"; p = pat ->
188
189	let tag = mk loc (Internal (Types.atom (Atoms.any))) in
190	let att = mk loc (Internal Types.Record.any) in
191	let any = mk loc (Internal (Types.any)) in
192	let id = ident (U.mk "x") in
193	let re = Star(Alt(SeqCapture(id,Elem p), Elem any)) in
194	let ct = mk loc (Regexp (re,any)) in
195	let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
196	let b = (p, Var id) in
197	exp loc (Op ("flatten", [Map (false,e,[b])]))
198	]
199	\|
200	[ e = expr; "."; l = [LIDENT \| UIDENT \| keyword ] ->
201	exp loc (Dot (e, label l))
202	]
203
204	\|
205	[ op = [ LIDENT "flatten"
206	\| LIDENT "load_xml"
207	\| LIDENT "load_file" \| LIDENT "load_file_utf8"
208	\| LIDENT "load_html"
209	\| LIDENT "print_xml" \| LIDENT "print_xml_utf8"
210	\| LIDENT "print"
211	\| LIDENT "raise"
212	\| LIDENT "int_of"
213	\| LIDENT "string_of"
214	\| LIDENT "atom_of"
215	];
216	e = expr -> exp loc (Op (op,[e]))
217	\| op = [ LIDENT "dump_to_file" \| LIDENT "dump_to_file_utf8" ];
218	e1 = expr LEVEL "no_appl"; e2 = expr -> exp loc (Op (op, [e1;e2]))
219	\| e1 = SELF; LIDENT "div"; e2 = expr -> exp loc (Op ("/", [e1;e2]))
220	\| e1 = SELF; LIDENT "mod"; e2 = expr -> exp loc (Op ("mod", [e1;e2]))
221	\| e1 = SELF; e2 = expr -> exp loc (Apply (e1,e2))
222	]
223
224	\| "no_appl"
225	[ c = const -> exp loc (Cst c)
226	\| "("; l = LIST1 expr SEP ","; ")" -> exp loc (tuple l)
227	\| "["; l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
228	let e = match e with Some e -> e \| None -> cst_nil in
229	let l = List.fold_right
230	(fun x q ->
231	match x with
232	\| `Elems l -> tuple_queue l q
233	\| `Explode x -> Op ("@",[x;q])
234	) l e
235	in
236	exp loc l
237	\| "<"; t = [ "("; e = expr; ")" -> e
238	\| a = [ LIDENT \| UIDENT \| keyword ] ->
239	let a = parse_ident a in
240	exp loc (Cst (Types.Atom (Atoms.mk a))) ];
241	a = expr_attrib_spec; ">"; c = expr ->
242	exp loc (Xml (t, Pair (a,c)))
243	\| "{"; r = [ expr_record_spec \| -> exp loc (RecordLitt LabelMap.empty) ]; "}" -> r
244	\| s = STRING2 ->
245	exp loc (tuple (char_list loc s @ [cst_nil]))
246	\| a = LIDENT -> exp loc (Var (ident (parse_ident a)))
247	]
248
249	];
250
251	seq_elem: [
252	[ x = STRING1 -> `Elems (char_list loc x)
253	\| e = expr LEVEL "no_appl" -> `Elems [e]
254	\| "!"; e = expr LEVEL "no_appl" -> `Explode e
255	]
256	];
257
258	let_binding: [
259	[ "let"; p = pat; "="; e = expr -> (p,e)
260	\| "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, Forget (e,t))
261	\| "let"; "fun"; (f,a,b) = fun_decl ->
262	let p = match f with
263	\| Some x -> mk loc (Capture x)
264	\| _ -> failwith "Function name mandatory in let fun declarations"
265	in
266	let abst = { fun_name = f; fun_iface = a; fun_body = b } in
267	let e = exp loc (Abstraction abst) in
268	(p,e);
269	]
270	];
271
272	fun_decl: [
273	(* need an hack to do this, because both productions would
274	match [ OPT LIDENT; "("; pat ] .... *)
275	[ f = OPT [ x = LIDENT -> ident (parse_ident x)]; "("; p1 = pat LEVEL "no_arrow";
276	res = [ "->"; p2 = pat;
277	a = [ ";"; a = LIST0 arrow SEP ";" -> a \| -> [] ];
278	")"; b = branches -> `Classic (p2,a,b)
279	\| ":"; targ1 = pat;
280	args = LIST0 [ ","; arg = pat; ":"; targ = pat -> (arg,targ) ];
281	")"; ":"; tres = pat ;
282	"="; body = expr ->
283	`Compact (targ1,args,tres,body)
284	] ->
285	match res with
286	\| `Classic (p2,a,b) -> f,(p1,p2)::a,b
287	\| `Compact (targ1,args,tres,body) ->
288	let args = (p1,targ1) :: args in
289	let targ = multi_prod nopos (List.map snd args) in
290	let arg = multi_prod nopos (List.map fst args) in
291	let b = [arg, body] in
292	let a = [targ,tres] in
293	(f,a,b)
294	]
295	];
296
297	arrow: [
298	[ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
299	];
300
301	branches: [
302	[ OPT "\|"; l = LIST1 branch SEP "\|" -> l ]
303	];
304
305	branch: [
306	[ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
307	];
308
309
310	regexp: [
311	[ x = regexp; "\|"; y = regexp ->
312	match (x,y) with
313	\| Elem x, Elem y -> Elem (mk loc (Or (x,y)))
314	\| _ -> Alt (x,y)
315	]
316	\| [ x = regexp; y = regexp -> Seq (x,y) ]
317	\| [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident (parse_ident a),x) ]
318	\| [ x = regexp; "*" -> Star x
319	\| x = regexp; "*?" -> WeakStar x
320	\| x = regexp; "+" -> Seq (x, Star x)
321	\| x = regexp; "+?" -> Seq (x, WeakStar x)
322	\| x = regexp; "?" -> Alt (x, Epsilon)
323	\| x = regexp; "??" -> Alt (Epsilon, x) ]
324	\| [ "("; x = regexp; ")" -> x
325	\| "("; a = LIDENT; ":="; c = const; ")" ->
326	Elem (mk loc (Constant ((ident (parse_ident a),c))))
327	\| UIDENT "PCDATA" -> string_regexp
328	\| i = STRING1; "--"; j = STRING1 ->
329	let i = Chars.mk_int (parse_char loc i)
330	and j = Chars.mk_int (parse_char loc j) in
331	Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
332	\| s = STRING1 ->
333	let s = seq_of_string loc s in
334	List.fold_right
335	(fun (loc,c) accu ->
336	let c = Chars.mk_int c in
337	let c = Chars.atom c in
338	Seq (Elem (mk loc (Internal (Types.char c))), accu))
339	s
340	Epsilon
341	\| e = pat LEVEL "simple" -> Elem e
342	]
343	];
344
345	pat: [
346	[ x = pat; LIDENT "where";
347	b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)
348	\| LIDENT -> error loc "Type/pattern identifiers must be capitalized"
349	] SEP "and"
350	-> mk loc (Recurs (x,b)) ]
351	\| RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
352	\| "no_arrow" [ x = pat; "\|"; y = pat -> mk loc (Or (x,y)) ]
353	\| "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y))
354	\| x = pat; "\\"; y = pat -> mk loc (Diff (x,y)) ]
355	\|
356	[ "{"; r = record_spec; "}" -> mk loc (Record (true,r))
357	\| "{\|"; r = record_spec; "\|}" -> mk loc (Record (false,r))
358	\| LIDENT "_" -> mk loc (Internal Types.any)
359	\| a = LIDENT -> mk loc (Capture (ident (parse_ident a)))
360	\| "("; a = LIDENT; ":="; c = const; ")" ->
361	mk loc (Constant (ident (parse_ident a),c))
362	\| a = UIDENT -> mk loc (PatVar a)
363	\| i = INT ; "--"; j = INT ->
364	let i = Intervals.mk i
365	and j = Intervals.mk j in
366	mk loc (Internal (Types.interval (Intervals.bounded i j)))
367	\| i = INT ->
368	let i = Intervals.mk i in
369	mk loc (Internal (Types.interval (Intervals.atom i)))
370	\| "*"; "--"; j = INT ->
371	let j = Intervals.mk j in
372	mk loc (Internal (Types.interval (Intervals.left j)))
373	\| i = INT; "--"; "*" ->
374	let i = Intervals.mk i in
375	mk loc (Internal (Types.interval (Intervals.right i)))
376	\| i = char ->
377	mk loc (Internal (Types.char (Chars.char_class i i)))
378	\| i = char ; "--"; j = char ->
379	mk loc (Internal (Types.char (Chars.char_class i j)))
380	\| c = const -> mk loc (Internal (Types.constant c))
381	\| "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
382	\| "["; r = [ r = regexp -> r \| -> Epsilon ];
383	q = [ ";"; q = pat -> q
384	\| -> mknoloc (Internal (Sequence.nil_type)) ];
385	"]" -> mk loc (Regexp (r,q))
386	\| "<"; t =
387	[ x = [ LIDENT \| UIDENT \| keyword ] ->
388	let a = if x = "_" then Atoms.any else Atoms.atom (Atoms.mk (parse_ident x)) in
389	mk loc (Internal (Types.atom a))
390	\| "("; t = pat; ")" -> t ];
391	a = attrib_spec; ">"; c = pat ->
392	mk loc (XmlT (t, multi_prod loc [a;c]))
393	\| s = STRING2 ->
394	let s = seq_of_string loc s in
395	let s = List.map
396	(fun (loc,c) ->
397	mk loc (Internal
398	(Types.char
399	(Chars.atom
400	(Chars.mk_int c))))) s in
401	let s = s @ [mk loc (Internal (Sequence.nil_type))] in
402	multi_prod loc s
403	]
404
405	];
406
407	record_spec:
408	[ [ r = LIST0 [ l = [LIDENT \| UIDENT \| keyword ]; "=";
409	o = [ "?" -> true \| -> false];
410	x = pat ->
411	let x = if o then mk loc (Optional x) else x in
412	(label l, x)
413	] SEP ";" ->
414	make_record loc r
415	] ];
416
417	char:
418	[
419	[ c = STRING1 -> Chars.mk_int (parse_char loc c) ]
420	];
421
422
423	const:
424	[
425	[ i = INT -> Types.Integer (Intervals.mk i)
426	\| "`"; a = [LIDENT \| UIDENT \| keyword ] -> Types.Atom (Atoms.mk (parse_ident a))
427	\| c = char -> Types.Char c ]
428	];
429
430
431	attrib_spec:
432	[ [ r = record_spec -> mk loc (Record (true,r))
433	\| "("; t = pat; ")" -> t
434	\| "{"; r = record_spec; "}" -> mk loc (Record (true,r))
435	\| "{\|"; r = record_spec; "\|}" -> mk loc (Record (false,r))
436	] ];
437
438	expr_record_spec:
439	[ [ r = LIST1
440	[ l = [LIDENT \| UIDENT \| keyword ]; "="; x = expr ->
441	(label l,x) ]
442	SEP ";" ->
443	exp loc (RecordLitt (make_record loc r))
444	] ];
445
446	expr_attrib_spec:
447	[ [ r = expr_record_spec -> r ]
448	\| [ e = expr LEVEL "no_appl" -> e
449	\| -> exp loc (RecordLitt (LabelMap.empty))
450	]
451	];
452	END
453
454	let pat' = Grammar.Entry.create gram "type/pattern expression"
455	EXTEND GLOBAL: pat pat';
456	pat': [ [ p = pat; EOI -> p ] ];
457	END
458
459	let pat = Grammar.Entry.parse pat
460	and expr = Grammar.Entry.parse expr
461	and prog = Grammar.Entry.parse prog
462
463
464	module From_string = struct
465	let pat s = Grammar.Entry.parse pat' (Stream.of_string s)
466	let expr s = expr (Stream.of_string s)
467	end
468