parser.ml 11.6 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 32


33
let char = mk noloc (Internal (Types.char Chars.any))
34
let string_regexp = Star (Elem char)
35 36 37
	       
let cst_nil =  mk noloc (Cst (Types.Atom Sequence.nil_atom))
		 
38 39 40 41 42 43
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
44 45
  aux [] (String.length s)

46 47 48
exception Error of string
let error loc s = raise (Location (loc, Error s))

49 50 51
let make_record loc r =
  LabelMap.from_list (fun _ _ -> error loc "Duplicated record field") r

52 53 54 55 56
let parse_char loc s =
  (* TODO: Unicode *)
  if String.length s <> 1 then
    error loc "Character litteral must have length 1";
  s.[0]
57

58
let char_list pos s = 
59
  let s = seq_of_string pos s in
60
  List.map (fun (loc,c) -> mk loc (Cst (Types.Char (Chars.mk_char c)))) s
61 62 63


EXTEND
64 65 66
  GLOBAL: prog expr pat regexp const;

  prog: [
67
    [ l = LIST0 [ p = phrase; ";;" -> mk loc p ]; EOI -> l ]
68 69 70
  ];

  phrase: [
71 72 73
    [ (p,e) = let_binding -> LetDecl (p,e)
    | (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        EvalStatement (mk loc (Match (e1,[p,e2])))
74 75
    | LIDENT "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t)
    | LIDENT "debug"; d = debug_directive -> Debug d
76 77
    ] | 
    [ e = expr -> EvalStatement e 
78 79 80 81 82
    ]
  ];

  debug_directive: [
    [ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
83
    | LIDENT "accept"; p = pat -> `Accept p
84
    | LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
85
    | LIDENT "normal_record"; t = pat -> `Normal_record t
86
    | LIDENT "compile2"; t = pat; p = LIST1 pat -> `Compile2 (t,p)
87
    | LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
88
    ]
89
  ];
90 91 92 93

  expr: [
    "top" RIGHTA
    [ "match"; e = SELF; "with"; b = branches -> mk loc (Match (e,b))
94 95
    | "try"; e = SELF; "with"; b = branches -> 
	let default = 
96 97
	  (mk noloc (Capture (ident "x")),
	   mk noloc (Op ("raise",[mk noloc (Var (ident "x"))]))) in
98
	mk loc (Try (e,b@[default]))
99
    | "map"; e = SELF; "with"; b = branches -> mk loc (Map (e,b))
100
    | "transform"; e = SELF; "with"; b = branches -> 
101
	let default = mk noloc (Capture (ident "x")), cst_nil in
102
	mk loc (Op ("flatten", [mk loc (Map (e,b@[default]))]))
103
    | "fun"; (f,a,b) = fun_decl ->
104 105 106
	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]))
107 108
    | e = expr; ":"; p = pat ->
	mk loc (Forget (e,p))
109 110 111
    ]

    
112
    | 
113 114
    [ e1 = expr; op = ["+" | "-" | "@"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
    ]
115
    |
116 117 118 119
    [ e1 = expr; "++"; "{"; l = [LIDENT | UIDENT]; "="; e = expr; "}" ->
	assert false
    ]
    |
120 121 122 123 124 125
    [ 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
126
	let re = Star(Alt(SeqCapture(ident "x",Elem p), Elem any)) in
127 128
	let ct = mk loc (Regexp (re,any)) in
        let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
129
	let b = (p,  mk loc (Var (ident "x"))) in
130
	mk loc (Op ("flatten", [mk loc (Map (e,[b]))]))
131
    ]
132
    |
133
    [ e = expr;  "."; l = [LIDENT | UIDENT] -> 
134
	mk loc (Dot (e,LabelPool.mk l)) 
135
    ]  
136
 
137
    |
138 139
    [ op = [ LIDENT "flatten"
	   | LIDENT "load_xml"
140
	   | LIDENT "load_html"
141
	   | LIDENT "print_xml"
142
	   | LIDENT "print"
143 144
	   | LIDENT "raise" 
	   | LIDENT "int_of"
145
	   | LIDENT "string_of"
146 147
	   ]; 
      e = expr -> mk loc (Op (op,[e]))
148 149
    | op = [ LIDENT "dump_to_file" ];
      e1 = expr LEVEL "no_appl"; e2 = expr -> mk loc (Op (op, [e1;e2])) 
150
    | e1 = SELF; LIDENT "div"; e2 = expr -> mk loc (Op ("/", [e1;e2]))
151
    | e1 = SELF; LIDENT "mod"; e2 = expr -> mk loc (Op ("mod", [e1;e2]))
152
    | e1 = SELF; e2 = expr -> mk loc (Apply (e1,e2))
153 154
    ]

155 156 157
    | "no_appl" 
    [ c = const -> mk loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
158 159
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
	let e = match e with Some e -> e | None -> cst_nil in
160 161 162 163 164 165
	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
166
    | t = [ a = TAG -> 
167
	      mk loc (Cst (Types.Atom (Atoms.mk a)))
168 169
	  | "<"; e = expr LEVEL "no_appl" -> e ];
	a = expr_attrib_spec; ">"; c = expr ->
170
	mk loc (Xml (t, mk loc (Pair (a,c))))
171
    | "{"; r = [ expr_record_spec | -> mk loc (RecordLitt LabelMap.empty) ]; "}" -> r
172
    | s = STRING2 ->
173
	tuple loc (char_list loc s @ [cst_nil])
174
    | a = LIDENT -> mk loc (Var (ident a))
175 176 177
    ]

  ];
178 179

  seq_elem: [
180 181 182
    [ x = STRING1 -> `Elems (char_list loc x)
    | e = expr LEVEL "no_appl" -> `Elems [e]
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
183 184
    ]
  ];
185 186 187
	  
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
188
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, mk noloc (Forget (e,t)))
189 190 191 192 193 194
    | "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
195
        let e = mk loc (Abstraction abst) in
196
        (p,e);
197 198 199
    ] 
  ];

200
 fun_decl: [
201 202
(* need an hack to do this, because both productions would
   match   [ OPT LIDENT; "("; pat ] .... *)
203
   [ f = OPT [ x = LIDENT -> ident x]; "("; p1 = pat LEVEL "no_arrow";
204 205 206 207 208 209 210 211 212 213 214 215 216 217 218 219 220 221
    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)
222 223 224
  ]
 ];

225
 arrow: [
226
    [ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
227 228 229
  ];

  branches: [
230
    [ OPT "|"; l = LIST1 branch SEP "|" -> l ]
231 232 233
  ];

  branch: [
234
    [ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
235 236 237 238 239 240
  ];

	  
  regexp: [ 
    [ x = regexp; "|"; y = regexp -> Alt (x,y) ]
  | [ x = regexp; y = regexp -> Seq (x,y) ]
241
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident a,x) ]
242 243 244 245
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
246
    | x = regexp; "?" ->  Alt (x, Epsilon)
247 248
    | x = regexp; "??" -> Alt (Epsilon, x) ]
  | [ "("; x = regexp; ")" -> x
249 250
    | "("; a = LIDENT; ":="; c = const; ")" -> 
	Elem (mk loc (Constant ((ident a,c))))
251
    | UIDENT "PCDATA" -> string_regexp
252
    | i = STRING1; "--"; j = STRING1 ->
253 254
	let i = Chars.mk_char (parse_char loc i)
	and j = Chars.mk_char (parse_char loc j) in
255
        Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
256 257
    |  s = STRING1 ->
	let s = seq_of_string loc s in
258
	List.fold_right
259
	  (fun (loc,c) accu -> 
260
	     let c = Chars.mk_char c in
261 262 263
	     let c = Chars.atom c in
	     Seq (Elem (mk loc (Internal (Types.char c))), accu))
	  s
264
	  Epsilon 
265 266 267 268 269 270 271 272 273
    | 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)) ]
274
    | "no_arrow" [ x = pat; "|"; y = pat -> mk loc (Or (x,y)) ] 
275
    | "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y)) 
276
	       | x = pat; "\\"; y = pat -> mk loc (Diff (x,y)) ]
277
    | 
278 279
      [ "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
280
      | LIDENT "_" -> mk loc (Internal Types.any)
281 282 283
      | a = LIDENT -> mk loc (Capture (ident a))
      | "("; a = LIDENT; ":="; c = const; ")" -> 
	  mk loc (Constant (ident a,c))
284 285
      | a = UIDENT -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
286 287
          let i =  Intervals.mk i 
	  and j =  Intervals.mk j in
288 289
          mk loc (Internal (Types.interval (Intervals.bounded i j)))
      | i = INT -> 
290
          let i =  Intervals.mk i  in
291
          mk loc (Internal (Types.interval (Intervals.atom i)))
292
      | "*"; "--"; j = INT ->
293
	  let j =  Intervals.mk j in
294
          mk loc (Internal (Types.interval (Intervals.left j)))
295
      | i = INT; "--"; "*" ->
296
	  let i = Intervals.mk i in
297
          mk loc (Internal (Types.interval (Intervals.right i)))
298 299
      | i = char ->
          mk loc (Internal (Types.char (Chars.char_class i i)))
300
      | i = char ; "--"; j = char ->
301
          mk loc (Internal (Types.char (Chars.char_class i j)))
302 303 304 305
      | 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 
306
                 | -> mk noloc (Internal (Sequence.nil_type)) ]; 
307
             "]" -> mk loc (Regexp (r,q))
308 309 310 311
      | t = [
	  [ "<"; LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any)))
	  | a = TAG -> 
	      mk loc 
312
	      (Internal (Types.atom (Atoms.atom (Atoms.mk a)))) ]
313 314 315
	| [ "<"; t = pat -> t ]
	];
	a = attrib_spec; ">"; c = pat ->
316
          mk loc (XmlT (t, multi_prod loc [a;c]))
317 318
      | s = STRING2 ->
	  let s = seq_of_string loc s in
319
	  let s = List.map 
320
		    (fun (loc,c) -> 
321 322 323
		       mk loc (Internal
				 (Types.char
				    (Chars.atom
324
				       (Chars.mk_char c))))) s in
325 326
	  let s = s @ [mk loc (Internal (Sequence.nil_type))] in
	  multi_prod loc s
327 328 329 330 331
      ]
    
  ];

  record_spec:
332 333
    [ [ r = LIST0 [ l = [LIDENT | UIDENT]; "=";
                  o = [ "?" -> true | -> false]; 
334 335
                  x = pat -> 
		    let x = if o then mk loc (Optional x) else x in
336
		    (LabelPool.mk l, x)
337
                ] SEP ";" ->
338
	  make_record loc r
339 340
      ] ];
  
341 342
  char:
    [ 
343 344
      [ c = STRING1 -> Chars.mk_char (parse_char loc c)
      | "!"; i = INT -> Chars.mk_int (int_of_string i) ]
345 346 347
    ];
     

348 349
  const:
    [ 
350 351
      [ i = INT -> Types.Integer (Intervals.mk i)
      | "`"; a = [LIDENT | UIDENT] -> Types.Atom (Atoms.mk a)
352
      | c = char -> Types.Char c ]
353 354 355 356
    ];


  attrib_spec:
357 358
    [ [ r = record_spec -> mk loc (Record (true,r)) 
      | "("; t = pat; ")" -> t ] ];
359 360 361

  expr_record_spec:
    [ [ r = LIST1
362
	      [ l = [LIDENT | UIDENT]; "="; x = expr -> 
363
		  (LabelPool.mk l,x) ] 
364
	      SEP ";" ->
365
	  mk loc (RecordLitt (make_record loc r))
366 367 368 369 370
      ] ];
  
  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
371
      | -> mk loc (RecordLitt (LabelMap.empty)) 
372 373 374 375
      ] 
    ];
END

376 377 378 379 380
let pat'  = Grammar.Entry.create gram "type/pattern expression"
EXTEND GLOBAL: pat pat';
  pat': [  [ p = pat; EOI -> p ] ];
END

381
let pat = Grammar.Entry.parse pat
382 383
and expr = Grammar.Entry.parse expr
and prog = Grammar.Entry.parse prog
384

385
module From_string = struct
386
  let pat s = Grammar.Entry.parse pat' (Stream.of_string s)
387 388
  let expr s = expr (Stream.of_string s)
end
389