viewcvs.cgi/parser/parser.ml

open Location
open Ast
open Ident
open Printf

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


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 = Encodings.Utf8.mk_latin1

let id_dummy = ident (U.mk "$$$")
let atom s = Atoms.mk (parse_ident s)
let label s = LabelPool.mk (parse_ident s)
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 (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 []

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 ]
  ];

  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 ->
        let schema_doc = Schema_xml.pxp_tree_of (get_string uri) in
        let schema = Schema_parser.parse_schema schema_doc in
        [ mk loc (SchemaDecl (name, schema))]
    | "debug"; d = debug_directive -> [ mk loc (Debug d) ]
    | DIRECTIVE "#quit" -> [ mk loc (Directive `Quit) ]
    | DIRECTIVE "#env" -> [ mk loc (Directive `Env) ]
    | "include"; s = STRING2 ->
        let s = get_string s in
        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"
        ]
        -> 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]))
    | 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
                       | `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 ] -> 
                   exp loc (Cst (Types.Atom (atom 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 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"; 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 ->
        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 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 = 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 (atom 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 (atom 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.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 ()
1	abate	4	open Location
2			open Ast
3	abate	225	open Ident
4	abate	501	open Printf
5	abate	4
6	abate	161	(*
7	abate	151	let () = Grammar.error_verbose := true
8	abate	161	*)
9	abate	81
10	abate	469
11	abate	81	let gram = Grammar.gcreate (Wlexer.lexer Wlexer.token Wlexer.latin1_engine)
12
13	abate	421	let true_atom = Atoms.mk_ascii "true"
14			let false_atom = Atoms.mk_ascii "false"
15			let true_type = Types.atom (Atoms.atom true_atom)
16			let false_type = Types.atom (Atoms.atom false_atom)
17	abate	81
18	abate	374	let parse_ident = Encodings.Utf8.mk_latin1
19
20	abate	375	let id_dummy = ident (U.mk "$$$")
21			let atom s = Atoms.mk (parse_ident s)
22	abate	374	let label s = LabelPool.mk (parse_ident s)
23	abate	375	let ident s = ident (parse_ident s)
24	abate	374
25	abate	38	let prog = Grammar.Entry.create gram "prog"
26	abate	431	let top_phrases = Grammar.Entry.create gram "toplevel phrases"
27	abate	38	let expr = Grammar.Entry.create gram "expression"
28			let pat = Grammar.Entry.create gram "type/pattern expression"
29			let regexp = Grammar.Entry.create gram "type/pattern regexp"
30			let const = Grammar.Entry.create gram "scalar constant"
31
32	abate	316	let exp pos e = LocatedExpr (loc_of_pos pos,e)
33
34	abate	38	let rec multi_prod loc = function
35			\| [ x ] -> x
36			\| x :: l -> mk loc (Prod (x, multi_prod loc l))
37			\| [] -> assert false
38
39	abate	316	let rec tuple = function
40	abate	38	\| [ x ] -> x
41	abate	316	\| x :: l -> Pair (x, tuple l)
42	abate	38	\| [] -> assert false
43	abate	89
44			let tuple_queue =
45	abate	316	List.fold_right (fun x q -> Pair (x, q))
46	abate	89
47	abate	233
48	abate	249	let char = mknoloc (Internal (Types.char Chars.any))
49	abate	66	let string_regexp = Star (Elem char)
50	abate	38
51	abate	316	let cst_nil = Cst (Types.Atom Sequence.nil_atom)
52	abate	38
53	abate	66	let seq_of_string pos s =
54	abate	310	let s = Encodings.Utf8.mk s in
55	abate	66	let (pos,_) = pos in
56	abate	310	let rec aux pos i j =
57			if Encodings.Utf8.equal_index i j then []
58			else
59	abate	325	let (len,i) = Encodings.Utf8.next s i in
60	abate	310	let (c,i) = Encodings.Utf8.next s i in
61	abate	325	((pos,pos+len),c) :: (aux (pos + len) i j)
62	abate	310	in
63	abate	325	aux (pos+1) (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s)
64	abate	4
65	abate	326	let get_string s =
66			let s = Encodings.Utf8.mk s in
67			let b = Buffer.create 32 in
68			let rec aux i j =
69			if Encodings.Utf8.equal_index i j then ()
70			else
71			let (len,i) = Encodings.Utf8.next s i in
72			let (c,i) = Encodings.Utf8.next s i in
73			Encodings.Utf8.store b c;
74			aux i j
75			in
76			aux (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s);
77			Buffer.contents b
78
79	abate	81	exception Error of string
80	abate	249	let error (i,j) s = Location.raise_loc i j (Error s)
81	abate	18
82	abate	233	let make_record loc r =
83			LabelMap.from_list (fun _ _ -> error loc "Duplicated record field") r
84
85	abate	81	let parse_char loc s =
86	abate	310	let s = seq_of_string loc s in
87			match s with
88	abate	325	\| [ loc,c ] -> c
89	abate	310	\| _ -> error loc "Character litteral must have length 1"
90	abate	81
91	abate	325	let char_list loc s =
92			let s = seq_of_string loc s in
93	abate	316	List.map (fun (loc,c) -> exp loc (Cst (Types.Char (Chars.mk_int c)))) s
94	abate	18
95	abate	249	let include_stack = ref []
96
97	abate	448	let protect_exn f g =
98			try let x = f () in g (); x
99			with e -> g (); raise e
100
101	abate	470	let is_fun_decl =
102			Grammar.Entry.of_parser gram "[is_fun_decl]"
103			(fun strm ->
104			match Stream.npeek 3 strm with
105			\| [ ("", "fun"); ("LIDENT", _); ("", "(") ]
106			\| [ ("LIDENT", _) ; ("", "(") ; _ ] -> ()
107			\| _ -> raise Stream.Failure
108			)
109
110	abate	501	let dot_RE = Pcre.regexp "\\."
111	abate	470
112	abate	38	EXTEND
113	abate	431	GLOBAL: top_phrases prog expr pat regexp const;
114	abate	4
115	abate	431	top_phrases: [
116	abate	446	[ l = LIST0 phrase; ";;" -> List.flatten l ]
117	abate	431	];
118
119	abate	10	prog: [
120	abate	431	[ l = LIST0 [ p = phrase ; OPT ";;" -> p ]; EOI -> List.flatten l ]
121	abate	10	];
122
123	abate	13	phrase: [
124	abate	431	[ (f,p,e) = let_binding ->
125			if f then [ mk loc (FunDecl e) ] else
126			[ mk loc (LetDecl (p,e)) ]
127			\| (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
128	abate	316	[ mk loc (EvalStatement (exp loc (Match (e1,[p,e2])))) ]
129	abate	431	\| "type"; x = UIDENT; "="; t = pat -> [ mk loc (TypeDecl (x,t)) ]
130			\| "type"; x = LIDENT -> error loc "Type identifiers must be capitalized"
131	abate	501	\| "schema"; name = UIDENT; "="; uri = STRING2 ->
132			let schema_doc = Schema_xml.pxp_tree_of (get_string uri) in
133			let schema = Schema_parser.parse_schema schema_doc in
134			[ mk loc (SchemaDecl (name, schema))]
135	abate	431	\| "debug"; d = debug_directive -> [ mk loc (Debug d) ]
136	abate	446	\| DIRECTIVE "#quit" -> [ mk loc (Directive `Quit) ]
137			\| DIRECTIVE "#env" -> [ mk loc (Directive `Env) ]
138	abate	431	\| "include"; s = STRING2 ->
139	abate	326	let s = get_string s in
140	abate	471	let s =
141			if Filename.is_relative s
142			then Filename.concat (Location.current_dir ()) s
143			else s in
144	abate	249	protect_op "File inclusion";
145			(* avoid looping; should issue an error ? *)
146	abate	471	(* it is possible to have looping with x/../x/../x/.. ....
147			Need to canonicalize filename *)
148	abate	249	if List.mem s !include_stack then []
149			else (
150			include_stack := s :: !include_stack;
151			Location.push_source (`File s);
152	abate	448	protect_exn
153			(fun () ->
154			let chan = open_in s in
155	abate	471	protect_exn
156			(fun () ->
157			let input = Stream.of_channel chan in
158			Grammar.Entry.parse prog input)
159			(fun () -> close_in chan))
160	abate	448	(fun () ->
161			Location.pop_source ();
162			include_stack := List.tl !include_stack)
163	abate	249	)
164	abate	66	] \|
165	abate	249	[ e = expr -> [ mk loc (EvalStatement e) ]
166	abate	43	]
167	abate	13	];
168
169	abate	43	debug_directive: [
170			[ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
171	abate	75	\| LIDENT "accept"; p = pat -> `Accept p
172	abate	43	\| LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
173	abate	407	\| LIDENT "sample"; t = pat -> `Sample t
174	abate	224	\| LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
175	abate	43	]
176			];
177
178	abate	332	keyword: [
179			[ a =
180			[ "map" \| "match" \| "with" \| "try" \| "xtransform"
181			\| "if" \| "then" \| "else"
182			\| "transform" \| "fun" \| "in"
183	abate	431	\| "let" \| "type" \| "debug" \| "include"
184	abate	501	\| "and" \| "validate" \| "schema"
185	abate	332	]
186			-> a
187			]
188			];
189
190	abate	4	expr: [
191			"top" RIGHTA
192	abate	421	[ "match"; e = SELF; "with"; b = branches ->
193			exp loc (Match (e,b))
194	abate	64	\| "try"; e = SELF; "with"; b = branches ->
195	abate	425	exp loc (Try (e,b))
196	abate	421	\| "map"; e = SELF; "with"; b = branches ->
197			exp loc (Map (e,b))
198			\| "xtransform"; e = SELF; "with"; b = branches ->
199			exp loc (Xtrans (e,b))
200	abate	237	\| "if"; e = SELF; "then"; e1 = SELF; "else"; e2 = SELF ->
201	abate	421	let p1 = mk loc (Internal true_type)
202			and p2 = mk loc (Internal false_type) in
203	abate	316	exp loc (Match (e, [p1,e1; p2,e2]))
204	abate	17	\| "transform"; e = SELF; "with"; b = branches ->
205	abate	421	exp loc (Transform (e,b))
206	abate	501	\| "validate"; e = SELF; "with"; schema = UIDENT; "#";
207			typ = [ UIDENT \| LIDENT \| keyword ] ->
208			exp loc (Validate (e, schema, typ))
209	abate	48	\| "fun"; (f,a,b) = fun_decl ->
210	abate	316	exp loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
211	abate	431	\| (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
212	abate	316	exp loc (Match (e1,[p,e2]))
213	abate	54	\| e = expr; ":"; p = pat ->
214	abate	316	exp loc (Forget (e,p))
215	abate	4	]
216
217
218	abate	15	\|
219	abate	237	[ e1 = expr; op = ["=" \| "<=" \| "<<" \| ">>" \| ">=" ]; e2 = expr ->
220			let op = match op with
221			\| "<<" -> "<"
222			\| ">>" -> ">"
223			\| s -> s in
224	abate	316	exp loc (Op (op,[e1;e2]))
225	abate	237	]
226
227			\|
228	abate	242	[ e1 = expr; op = ["+" \| "-" \| "@" ]; e2 = expr ->
229	abate	316	exp loc (Op (op,[e1;e2]))
230	abate	339	\| e = expr; "\\"; l = [LIDENT \| UIDENT \| keyword ] ->
231	abate	374	exp loc (RemoveField (e, label l))
232	abate	51	]
233	abate	16	\|
234	abate	316	[ e1 = expr; op = ["*"]; e2 = expr -> exp loc (Op (op,[e1;e2]))
235	abate	151	\| e = expr; op = "/"; p = pat ->
236
237			let tag = mk loc (Internal (Types.atom (Atoms.any))) in
238			let att = mk loc (Internal Types.Record.any) in
239			let any = mk loc (Internal (Types.any)) in
240	abate	375	let re = Star(Alt(SeqCapture(id_dummy,Elem p), Elem any)) in
241	abate	151	let ct = mk loc (Regexp (re,any)) in
242			let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
243	abate	375	let b = (p, Var id_dummy) in
244	abate	421	exp loc (Transform (e,[b]))
245	abate	51	]
246	abate	26	\|
247	abate	339	[ e = expr; "."; l = [LIDENT \| UIDENT \| keyword ] ->
248	abate	374	exp loc (Dot (e, label l))
249	abate	51	]
250	abate	26
251	abate	52	\|
252	abate	66	[ op = [ LIDENT "flatten"
253			\| LIDENT "load_xml"
254	abate	374	\| LIDENT "load_file" \| LIDENT "load_file_utf8"
255	abate	188	\| LIDENT "load_html"
256	abate	374	\| LIDENT "print_xml" \| LIDENT "print_xml_utf8"
257	abate	124	\| LIDENT "print"
258	abate	66	\| LIDENT "int_of"
259	abate	133	\| LIDENT "string_of"
260	abate	329	\| LIDENT "atom_of"
261	abate	421	\| LIDENT "raise"
262	abate	66	];
263	abate	316	e = expr -> exp loc (Op (op,[e]))
264	abate	374	\| op = [ LIDENT "dump_to_file" \| LIDENT "dump_to_file_utf8" ];
265	abate	316	e1 = expr LEVEL "no_appl"; e2 = expr -> exp loc (Op (op, [e1;e2]))
266			\| e1 = SELF; LIDENT "div"; e2 = expr -> exp loc (Op ("/", [e1;e2]))
267			\| e1 = SELF; LIDENT "mod"; e2 = expr -> exp loc (Op ("mod", [e1;e2]))
268			\| e1 = SELF; e2 = expr -> exp loc (Apply (e1,e2))
269	abate	52	]
270
271	abate	4	\| "no_appl"
272	abate	316	[ c = const -> exp loc (Cst c)
273			\| "("; l = LIST1 expr SEP ","; ")" -> exp loc (tuple l)
274	abate	18	\| "["; l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
275			let e = match e with Some e -> e \| None -> cst_nil in
276	abate	321	let l = List.fold_right
277			(fun x q ->
278			match x with
279			\| `Elems l -> tuple_queue l q
280			\| `Explode x -> Op ("@",[x;q])
281			) l e
282			in
283			exp loc l
284	abate	332	\| "<"; t = [ "("; e = expr; ")" -> e
285			\| a = [ LIDENT \| UIDENT \| keyword ] ->
286	abate	375	exp loc (Cst (Types.Atom (atom a))) ];
287	abate	81	a = expr_attrib_spec; ">"; c = expr ->
288	abate	316	exp loc (Xml (t, Pair (a,c)))
289			\| "{"; r = [ expr_record_spec \| -> exp loc (RecordLitt LabelMap.empty) ]; "}" -> r
290	abate	81	\| s = STRING2 ->
291	abate	316	exp loc (tuple (char_list loc s @ [cst_nil]))
292	abate	375	\| a = LIDENT -> exp loc (Var (ident a))
293	abate	4	]
294
295			];
296	abate	18
297			seq_elem: [
298	abate	89	[ x = STRING1 -> `Elems (char_list loc x)
299			\| e = expr LEVEL "no_appl" -> `Elems [e]
300			\| "!"; e = expr LEVEL "no_appl" -> `Explode e
301	abate	18	]
302			];
303	abate	4
304			let_binding: [
305	abate	470	[ "let"; is_fun_decl; OPT "fun"; (f,a,b) = fun_decl ->
306			let f = match f with Some x -> x \| None -> assert false in
307			let p = mk loc (Capture f) in
308			let abst = { fun_name = Some f; fun_iface = a; fun_body = b } in
309	abate	316	let e = exp loc (Abstraction abst) in
310	abate	431	(true,p,e)
311	abate	470	\| "let"; p = pat; "="; e = expr -> (false,p,e)
312			\| "let"; p = pat; ":"; t = pat; "="; e = expr -> (false,p, Forget (e,t))
313	abate	4	]
314			];
315
316	abate	470	fun_decl_after_lparen: [
317	abate	85	(* need an hack to do this, because both productions would
318			match [ OPT LIDENT; "("; pat ] .... *)
319	abate	470	[ p1 = pat LEVEL "no_arrow";
320			res = [ "->"; p2 = pat;
321			a = [ ";"; a = LIST0 arrow SEP ";" -> a \| -> [] ];
322			")"; b = branches -> `Classic (p2,a,b)
323			\| ":"; targ1 = pat;
324			args = LIST0 [ ","; arg = pat; ":"; targ = pat -> (arg,targ) ];
325			")"; ":"; tres = pat ;
326			"="; body = expr ->
327			`Compact (targ1,args,tres,body)
328			] ->
329			match res with
330			\| `Classic (p2,a,b) -> (p1,p2)::a,b
331			\| `Compact (targ1,args,tres,body) ->
332			let args = (p1,targ1) :: args in
333			let targ = multi_prod nopos (List.map snd args) in
334			let arg = multi_prod nopos (List.map fst args) in
335			let b = [arg, body] in
336			let a = [targ,tres] in
337			(a,b) ] ];
338
339
340			fun_decl: [
341			[ f = OPT [ x = LIDENT -> ident x]; "("; (a,b) = fun_decl_after_lparen ->
342	abate	85	(f,a,b)
343	abate	470	]
344	abate	48	];
345
346	abate	85	arrow: [
347	abate	9	[ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
348	abate	4	];
349
350			branches: [
351	abate	81	[ OPT "\|"; l = LIST1 branch SEP "\|" -> l ]
352	abate	4	];
353
354			branch: [
355	abate	6	[ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
356	abate	4	];
357
358
359			regexp: [
360	abate	236	[ x = regexp; "\|"; y = regexp ->
361			match (x,y) with
362			\| Elem x, Elem y -> Elem (mk loc (Or (x,y)))
363			\| _ -> Alt (x,y)
364			]
365	abate	4	\| [ x = regexp; y = regexp -> Seq (x,y) ]
366	abate	375	\| [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident a,x) ]
367	abate	4	\| [ x = regexp; "*" -> Star x
368			\| x = regexp; "*?" -> WeakStar x
369			\| x = regexp; "+" -> Seq (x, Star x)
370			\| x = regexp; "+?" -> Seq (x, WeakStar x)
371	abate	71	\| x = regexp; "?" -> Alt (x, Epsilon)
372	abate	4	\| x = regexp; "??" -> Alt (Epsilon, x) ]
373			\| [ "("; x = regexp; ")" -> x
374	abate	225	\| "("; a = LIDENT; ":="; c = const; ")" ->
375	abate	375	Elem (mk loc (Constant ((ident a,c))))
376	abate	66	\| UIDENT "PCDATA" -> string_regexp
377	abate	81	\| i = STRING1; "--"; j = STRING1 ->
378	abate	310	let i = Chars.mk_int (parse_char loc i)
379			and j = Chars.mk_int (parse_char loc j) in
380	abate	66	Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
381	abate	81	\| s = STRING1 ->
382			let s = seq_of_string loc s in
383	abate	18	List.fold_right
384	abate	66	(fun (loc,c) accu ->
385	abate	310	let c = Chars.mk_int c in
386	abate	18	let c = Chars.atom c in
387			Seq (Elem (mk loc (Internal (Types.char c))), accu))
388			s
389	abate	66	Epsilon
390	abate	4	\| e = pat LEVEL "simple" -> Elem e
391			]
392			];
393
394			pat: [
395	abate	332	[ x = pat; LIDENT "where";
396	abate	338	b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)
397			\| LIDENT -> error loc "Type/pattern identifiers must be capitalized"
398			] SEP "and"
399	abate	4	-> mk loc (Recurs (x,b)) ]
400			\| RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
401	abate	6	\| "no_arrow" [ x = pat; "\|"; y = pat -> mk loc (Or (x,y)) ]
402	abate	121	\| "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y))
403	abate	99	\| x = pat; "\\"; y = pat -> mk loc (Diff (x,y)) ]
404	abate	4	\|
405	abate	159	[ "{"; r = record_spec; "}" -> mk loc (Record (true,r))
406			\| "{\|"; r = record_spec; "\|}" -> mk loc (Record (false,r))
407	abate	4	\| LIDENT "_" -> mk loc (Internal Types.any)
408	abate	375	\| a = LIDENT -> mk loc (Capture (ident a))
409	abate	225	\| "("; a = LIDENT; ":="; c = const; ")" ->
410	abate	375	mk loc (Constant (ident a,c))
411	abate	501	\| schema = UIDENT; "#"; typ = [ UIDENT \| LIDENT \| keyword ];
412			k = OPT [ "as"; k = [ "element" \| "type" \| "attribute" ] -> k ] ->
413			let kind =
414			match k with
415			\| None -> `Any
416			\| Some "element" -> `Element
417			\| Some "type" -> `Type
418			\| Some "attribute" -> `Attribute
419			\| _ -> assert false
420			in
421			mk loc (SchemaVar (kind, schema, typ))
422	abate	4	\| a = UIDENT -> mk loc (PatVar a)
423			\| i = INT ; "--"; j = INT ->
424	abate	222	let i = Intervals.mk i
425			and j = Intervals.mk j in
426	abate	52	mk loc (Internal (Types.interval (Intervals.bounded i j)))
427			\| i = INT ->
428	abate	222	let i = Intervals.mk i in
429	abate	52	mk loc (Internal (Types.interval (Intervals.atom i)))
430	abate	81	\| "*"; "--"; j = INT ->
431	abate	222	let j = Intervals.mk j in
432	abate	52	mk loc (Internal (Types.interval (Intervals.left j)))
433	abate	81	\| i = INT; "--"; "*" ->
434	abate	222	let i = Intervals.mk i in
435	abate	52	mk loc (Internal (Types.interval (Intervals.right i)))
436	abate	19	\| i = char ->
437			mk loc (Internal (Types.char (Chars.char_class i i)))
438	abate	13	\| i = char ; "--"; j = char ->
439	abate	18	mk loc (Internal (Types.char (Chars.char_class i j)))
440	abate	4	\| c = const -> mk loc (Internal (Types.constant c))
441			\| "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
442			\| "["; r = [ r = regexp -> r \| -> Epsilon ];
443			q = [ ";"; q = pat -> q
444	abate	249	\| -> mknoloc (Internal (Sequence.nil_type)) ];
445	abate	4	"]" -> mk loc (Regexp (r,q))
446	abate	332	\| "<"; t =
447			[ x = [ LIDENT \| UIDENT \| keyword ] ->
448	abate	375	let a = if x = "_" then Atoms.any else Atoms.atom (atom x) in
449	abate	332	mk loc (Internal (Types.atom a))
450			\| "("; t = pat; ")" -> t ];
451	abate	81	a = attrib_spec; ">"; c = pat ->
452	abate	110	mk loc (XmlT (t, multi_prod loc [a;c]))
453	abate	81	\| s = STRING2 ->
454			let s = seq_of_string loc s in
455	abate	63	let s = List.map
456	abate	66	(fun (loc,c) ->
457	abate	63	mk loc (Internal
458			(Types.char
459			(Chars.atom
460	abate	310	(Chars.mk_int c))))) s in
461	abate	63	let s = s @ [mk loc (Internal (Sequence.nil_type))] in
462			multi_prod loc s
463	abate	4	]
464
465			];
466
467			record_spec:
468	abate	339	[ [ r = LIST0 [ l = [LIDENT \| UIDENT \| keyword ]; "=";
469	abate	81	o = [ "?" -> true \| -> false];
470	abate	229	x = pat ->
471			let x = if o then mk loc (Optional x) else x in
472	abate	374	(label l, x)
473	abate	4	] SEP ";" ->
474	abate	233	make_record loc r
475	abate	4	] ];
476
477	abate	13	char:
478			[
479	abate	310	[ c = STRING1 -> Chars.mk_int (parse_char loc c) ]
480	abate	13	];
481
482
483	abate	4	const:
484			[
485	abate	222	[ i = INT -> Types.Integer (Intervals.mk i)
486	abate	375	\| "`"; a = [LIDENT \| UIDENT \| keyword ] -> Types.Atom (atom a)
487	abate	13	\| c = char -> Types.Char c ]
488	abate	4	];
489
490
491			attrib_spec:
492	abate	159	[ [ r = record_spec -> mk loc (Record (true,r))
493	abate	252	\| "("; t = pat; ")" -> t
494			\| "{"; r = record_spec; "}" -> mk loc (Record (true,r))
495			\| "{\|"; r = record_spec; "\|}" -> mk loc (Record (false,r))
496			] ];
497	abate	4
498			expr_record_spec:
499			[ [ r = LIST1
500	abate	339	[ l = [LIDENT \| UIDENT \| keyword ]; "="; x = expr ->
501	abate	374	(label l,x) ]
502	abate	4	SEP ";" ->
503	abate	316	exp loc (RecordLitt (make_record loc r))
504	abate	4	] ];
505
506			expr_attrib_spec:
507			[ [ r = expr_record_spec -> r ]
508			\| [ e = expr LEVEL "no_appl" -> e
509	abate	316	\| -> exp loc (RecordLitt (LabelMap.empty))
510	abate	4	]
511			];
512			END
513
514	abate	10	let pat = Grammar.Entry.parse pat
515	abate	81	and expr = Grammar.Entry.parse expr
516			and prog = Grammar.Entry.parse prog
517	abate	431	and top_phrases = Grammar.Entry.parse top_phrases
518	abate	10
519	abate	495	let sync () =
520			match !Wlexer.lexbuf with
521			\| None -> ()
522			\| Some lb ->
523			let rec aux () =
524			match !Wlexer.last_tok with
525			\| ("",";;") \| ("EOI","") -> ()
526			\| _ ->
527			Wlexer.last_tok := Wlexer.token Wlexer.latin1_engine lb;
528			aux ()
529			in
530			aux ()