parser.ml 11.4 KB
Newer Older
1 2
open Location
open Ast
3
open Ident
4

5
(*
6
let ()  = Grammar.error_verbose := true
7
*)
8 9 10 11

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


12 13 14 15 16 17 18 19 20 21 22 23 24
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
25
  | x :: l -> mk loc (Pair (x, tuple loc l))
26
  | [] -> assert false
27 28 29 30

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

31
let char = mk noloc (Internal (Types.char Chars.any))
32
let string_regexp = Star (Elem char)
33 34 35
	       
let cst_nil =  mk noloc (Cst (Types.Atom Sequence.nil_atom))
		 
36 37 38 39 40 41
let seq_of_string pos s =
  let (pos,_) = pos in
  let rec aux accu i = 
    if (i = 0) 
    then accu 
    else aux (((pos+i,pos+i+1),s.[i-1])::accu) (i-1) in
42 43
  aux [] (String.length s)

44 45 46 47 48 49 50 51
exception Error of string
let error loc s = raise (Location (loc, Error s))

let parse_char loc s =
  (* TODO: Unicode *)
  if String.length s <> 1 then
    error loc "Character litteral must have length 1";
  s.[0]
52

53
let char_list pos s = 
54
  let s = seq_of_string pos s in
55
  List.map (fun (loc,c) -> mk loc (Cst (Types.Char (Chars.mk_char c)))) s
56 57 58


EXTEND
59 60 61
  GLOBAL: prog expr pat regexp const;

  prog: [
62
    [ l = LIST0 [ p = phrase; ";;" -> mk loc p ]; EOI -> l ]
63 64 65
  ];

  phrase: [
66 67 68
    [ (p,e) = let_binding -> LetDecl (p,e)
    | (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        EvalStatement (mk loc (Match (e1,[p,e2])))
69 70
    | LIDENT "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t)
    | LIDENT "debug"; d = debug_directive -> Debug d
71 72
    ] | 
    [ e = expr -> EvalStatement e 
73 74 75 76 77
    ]
  ];

  debug_directive: [
    [ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
78
    | LIDENT "accept"; p = pat -> `Accept p
79
    | LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
80
    | LIDENT "normal_record"; t = pat -> `Normal_record t
81
    | LIDENT "compile2"; t = pat; p = LIST1 pat -> `Compile2 (t,p)
82
    | LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
83
    ]
84
  ];
85 86 87 88

  expr: [
    "top" RIGHTA
    [ "match"; e = SELF; "with"; b = branches -> mk loc (Match (e,b))
89 90
    | "try"; e = SELF; "with"; b = branches -> 
	let default = 
91 92
	  (mk noloc (Capture (ident "x")),
	   mk noloc (Op ("raise",[mk noloc (Var (ident "x"))]))) in
93
	mk loc (Try (e,b@[default]))
94
    | "map"; e = SELF; "with"; b = branches -> mk loc (Map (e,b))
95
    | "transform"; e = SELF; "with"; b = branches -> 
96
	let default = mk noloc (Capture (ident "x")), cst_nil in
97
	mk loc (Op ("flatten", [mk loc (Map (e,b@[default]))]))
98
    | "fun"; (f,a,b) = fun_decl ->
99 100 101
	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]))
102 103
    | e = expr; ":"; p = pat ->
	mk loc (Forget (e,p))
104 105 106
    ]

    
107
    | 
108 109
    [ e1 = expr; op = ["+" | "-" | "@"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
    ]
110
    |
111 112 113 114 115 116
    [ e1 = expr; op = ["*"]; e2 = expr -> mk 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
117
	let re = Star(Alt(SeqCapture(ident "x",Elem p), Elem any)) in
118 119
	let ct = mk loc (Regexp (re,any)) in
        let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
120
	let b = (p,  mk loc (Var (ident "x"))) in
121
	mk loc (Op ("flatten", [mk loc (Map (e,[b]))]))
122
    ]
123
    |
124 125
    [ e = expr;  "."; l = [LIDENT | UIDENT] -> 
	mk loc (Dot (e,Types.LabelPool.mk l)) 
126
    ]  
127
 
128
    |
129 130
    [ op = [ LIDENT "flatten"
	   | LIDENT "load_xml"
131
	   | LIDENT "load_html"
132
	   | LIDENT "print_xml"
133
	   | LIDENT "print"
134 135
	   | LIDENT "raise" 
	   | LIDENT "int_of"
136
	   | LIDENT "string_of"
137 138
	   ]; 
      e = expr -> mk loc (Op (op,[e]))
139 140
    | op = [ LIDENT "dump_to_file" ];
      e1 = expr LEVEL "no_appl"; e2 = expr -> mk loc (Op (op, [e1;e2])) 
141
    | e1 = SELF; LIDENT "div"; e2 = expr -> mk loc (Op ("/", [e1;e2]))
142
    | e1 = SELF; LIDENT "mod"; e2 = expr -> mk loc (Op ("mod", [e1;e2]))
143
    | e1 = SELF; e2 = expr -> mk loc (Apply (e1,e2))
144 145
    ]

146 147 148
    | "no_appl" 
    [ c = const -> mk loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
149 150
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
	let e = match e with Some e -> e | None -> cst_nil in
151 152 153 154 155 156
	List.fold_right 
	  (fun x q ->
	     match x with
	       | `Elems l -> tuple_queue l q
	       | `Explode x -> mk x.loc (Op ("@",[x;q]))
	  ) l e
157
    | t = [ a = TAG -> 
158
	      mk loc (Cst (Types.Atom (Atoms.mk a)))
159 160
	  | "<"; e = expr LEVEL "no_appl" -> e ];
	a = expr_attrib_spec; ">"; c = expr ->
161
	mk loc (Xml (t, mk loc (Pair (a,c))))
162
    | "{"; r = [ expr_record_spec | -> mk loc (RecordLitt []) ]; "}" -> r
163
    | s = STRING2 ->
164
	tuple loc (char_list loc s @ [cst_nil])
165
    | a = LIDENT -> mk loc (Var (ident a))
166 167 168
    ]

  ];
169 170

  seq_elem: [
171 172 173
    [ x = STRING1 -> `Elems (char_list loc x)
    | e = expr LEVEL "no_appl" -> `Elems [e]
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
174 175
    ]
  ];
176 177 178
	  
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
179
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, mk noloc (Forget (e,t)))
180 181 182 183 184 185
    | "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
186
        let e = mk loc (Abstraction abst) in
187
        (p,e);
188 189 190
    ] 
  ];

191
 fun_decl: [
192 193
(* need an hack to do this, because both productions would
   match   [ OPT LIDENT; "("; pat ] .... *)
194
   [ f = OPT [ x = LIDENT -> ident x]; "("; p1 = pat LEVEL "no_arrow";
195 196 197 198 199 200 201 202 203 204 205 206 207 208 209 210 211 212
    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 noloc (List.map snd args) in
	    let arg = multi_prod noloc (List.map fst args) in
	    let b = [arg, body] in
	    let a = [targ,tres] in
	    (f,a,b)
213 214 215
  ]
 ];

216
 arrow: [
217
    [ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
218 219 220
  ];

  branches: [
221
    [ OPT "|"; l = LIST1 branch SEP "|" -> l ]
222 223 224
  ];

  branch: [
225
    [ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
226 227 228 229 230 231
  ];

	  
  regexp: [ 
    [ x = regexp; "|"; y = regexp -> Alt (x,y) ]
  | [ x = regexp; y = regexp -> Seq (x,y) ]
232
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident a,x) ]
233 234 235 236
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
237
    | x = regexp; "?" ->  Alt (x, Epsilon)
238 239
    | x = regexp; "??" -> Alt (Epsilon, x) ]
  | [ "("; x = regexp; ")" -> x
240 241
    | "("; a = LIDENT; ":="; c = const; ")" -> 
	Elem (mk loc (Constant ((ident a,c))))
242
    | UIDENT "PCDATA" -> string_regexp
243
    | i = STRING1; "--"; j = STRING1 ->
244 245
	let i = Chars.mk_char (parse_char loc i)
	and j = Chars.mk_char (parse_char loc j) in
246
        Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
247 248
    |  s = STRING1 ->
	let s = seq_of_string loc s in
249
	List.fold_right
250
	  (fun (loc,c) accu -> 
251
	     let c = Chars.mk_char c in
252 253 254
	     let c = Chars.atom c in
	     Seq (Elem (mk loc (Internal (Types.char c))), accu))
	  s
255
	  Epsilon 
256 257 258 259 260 261 262 263 264
    | 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)) ]
265
    | "no_arrow" [ x = pat; "|"; y = pat -> mk loc (Or (x,y)) ] 
266
    | "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y)) 
267
	       | x = pat; "\\"; y = pat -> mk loc (Diff (x,y)) ]
268
    | 
269 270
      [ "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
271
      | LIDENT "_" -> mk loc (Internal Types.any)
272 273 274
      | a = LIDENT -> mk loc (Capture (ident a))
      | "("; a = LIDENT; ":="; c = const; ")" -> 
	  mk loc (Constant (ident a,c))
275 276
      | a = UIDENT -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
277 278
          let i =  Intervals.mk i 
	  and j =  Intervals.mk j in
279 280
          mk loc (Internal (Types.interval (Intervals.bounded i j)))
      | i = INT -> 
281
          let i =  Intervals.mk i  in
282
          mk loc (Internal (Types.interval (Intervals.atom i)))
283
      | "*"; "--"; j = INT ->
284
	  let j =  Intervals.mk j in
285
          mk loc (Internal (Types.interval (Intervals.left j)))
286
      | i = INT; "--"; "*" ->
287
	  let i = Intervals.mk i in
288
          mk loc (Internal (Types.interval (Intervals.right i)))
289 290
      | i = char ->
          mk loc (Internal (Types.char (Chars.char_class i i)))
291
      | i = char ; "--"; j = char ->
292
          mk loc (Internal (Types.char (Chars.char_class i j)))
293 294 295 296
      | 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 
297
                 | -> mk noloc (Internal (Sequence.nil_type)) ]; 
298
             "]" -> mk loc (Regexp (r,q))
299 300 301 302
      | t = [
	  [ "<"; LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any)))
	  | a = TAG -> 
	      mk loc 
303
	      (Internal (Types.atom (Atoms.atom (Atoms.mk a)))) ]
304 305 306
	| [ "<"; t = pat -> t ]
	];
	a = attrib_spec; ">"; c = pat ->
307
          mk loc (XmlT (t, multi_prod loc [a;c]))
308 309
      | s = STRING2 ->
	  let s = seq_of_string loc s in
310
	  let s = List.map 
311
		    (fun (loc,c) -> 
312 313 314
		       mk loc (Internal
				 (Types.char
				    (Chars.atom
315
				       (Chars.mk_char c))))) s in
316 317
	  let s = s @ [mk loc (Internal (Sequence.nil_type))] in
	  multi_prod loc s
318 319 320 321 322
      ]
    
  ];

  record_spec:
323 324
    [ [ r = LIST0 [ l = [LIDENT | UIDENT]; "=";
                  o = [ "?" -> true | -> false]; 
325
                  x = pat -> (Types.LabelPool.mk l,o,x)
326
                ] SEP ";" ->
327 328
	  (* TODO: check here uniqueness *)
	  List.sort (fun (l1,_,_) (l2,_,_) -> compare l1 l2) r
329 330
      ] ];
  
331 332
  char:
    [ 
333 334
      [ c = STRING1 -> Chars.mk_char (parse_char loc c)
      | "!"; i = INT -> Chars.mk_int (int_of_string i) ]
335 336 337
    ];
     

338 339
  const:
    [ 
340 341
      [ i = INT -> Types.Integer (Intervals.mk i)
      | "`"; a = [LIDENT | UIDENT] -> Types.Atom (Atoms.mk a)
342
      | c = char -> Types.Char c ]
343 344 345 346
    ];


  attrib_spec:
347 348
    [ [ r = record_spec -> mk loc (Record (true,r)) 
      | "("; t = pat; ")" -> t ] ];
349 350 351

  expr_record_spec:
    [ [ r = LIST1
352 353
	      [ l = [LIDENT | UIDENT]; "="; x = expr -> 
		  (Types.LabelPool.mk l,x) ] 
354 355 356 357 358 359 360 361 362 363 364 365
	      SEP ";" ->
	  mk loc (RecordLitt r)
      ] ];
  
  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
      | -> mk loc (RecordLitt []) 
      ] 
    ];
END

366 367 368 369 370
let pat'  = Grammar.Entry.create gram "type/pattern expression"
EXTEND GLOBAL: pat pat';
  pat': [  [ p = pat; EOI -> p ] ];
END

371
let pat = Grammar.Entry.parse pat
372 373
and expr = Grammar.Entry.parse expr
and prog = Grammar.Entry.parse prog
374

375
module From_string = struct
376
  let pat s = Grammar.Entry.parse pat' (Stream.of_string s)
377 378
  let expr s = expr (Stream.of_string s)
end
379