parser.ml 14 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
let parse_ident = Encodings.Utf8.mk_latin1

let label s = LabelPool.mk (parse_ident s)

16 17 18 19 20 21
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"
		
22 23
let exp pos e = LocatedExpr (loc_of_pos pos,e)

24 25 26 27 28
let rec multi_prod loc = function
  | [ x ] -> x
  | x :: l -> mk loc (Prod (x, multi_prod loc l))
  | [] -> assert false
      
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 47 48
  let rec aux pos i j =
    if Encodings.Utf8.equal_index i j then []
    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 56 57 58 59 60 61 62 63 64 65 66 67 68
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

69
exception Error of string
70
let error (i,j) s = Location.raise_loc i j (Error s)
71

72 73 74
let make_record loc r =
  LabelMap.from_list (fun _ _ -> error loc "Duplicated record field") r

75
let parse_char loc s =
76 77
  let s = seq_of_string loc s in
  match s with
78
    | [ loc,c ] -> c
79
    | _ -> error loc "Character litteral must have length 1"
80

81 82
let char_list loc s = 
  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 87
let include_stack = ref []

88
EXTEND
89 90 91
  GLOBAL: prog expr pat regexp const;

  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" ]
101 102
    | LIDENT "debug"; d = debug_directive -> [ mk loc (Debug d) ]
    | LIDENT "include"; s = STRING2 ->
103
	let s = get_string s in
104 105 106 107 108 109 110 111 112 113 114 115 116 117
	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
	)
118
    ] | 
119
    [ e = expr -> [ mk loc (EvalStatement e) ]
120 121 122 123 124
    ]
  ];

  debug_directive: [
    [ 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

133 134 135 136 137 138 139 140 141 142 143
  keyword: [
    [ a = 
	[ "map" | "match" | "with" | "try" | "xtransform"
	| "if" | "then"	| "else"
	| "transform" | "fun" | "in"
	| "let"
	]
	-> a
    ]
  ];

144 145
  expr: [
    "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 = 
150 151
	  mknoloc (Capture id), 
	  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))
155 156 157
    | "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
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
    ]

    
170 171 172 173 174 175
    | 
    [ e1 = expr; op = ["=" | "<=" | "<<" | ">>" | ">=" ]; e2 = expr -> 
	let op = match op with
	  | "<<" -> "<"
	  | ">>" -> ">"
	  | 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]))  
187 188 189 190 191
    | 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
192 193
	let id = ident (U.mk "x") in
	let re = Star(Alt(SeqCapture(id,Elem p), Elem any)) in
194 195
	let ct = mk loc (Regexp (re,any)) in
        let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
196 197
	let b = (p,  Var id) in
	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
    |
205 206
    [ op = [ LIDENT "flatten"
	   | 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"
211 212
	   | LIDENT "raise" 
	   | 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" ];
218 219 220 221
      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))
222 223
    ]

224
    | "no_appl" 
225 226
    [ c = const -> exp loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> exp loc (tuple l)
227 228
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
	let e = match e with Some e -> e | None -> cst_nil in
229 230 231 232 233 234 235 236
	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
237 238
    | "<"; t = [ "("; e = expr; ")" -> e
	       | a = [ LIDENT | UIDENT | keyword ] -> 
239 240
		   let a = parse_ident a in
		   exp loc (Cst (Types.Atom (Atoms.mk a))) ];
241
	a = expr_attrib_spec; ">"; c = expr ->
242 243
	exp loc (Xml (t, Pair (a,c)))
    | "{"; 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: [
252 253 254
    [ x = STRING1 -> `Elems (char_list loc x)
    | e = expr LEVEL "no_appl" -> `Elems [e]
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
255 256
    ]
  ];
257 258 259
	  
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
260
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, Forget (e,t))
261 262 263 264 265 266
    | "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
267
        let e = exp loc (Abstraction abst) in
268
        (p,e);
269 270 271
    ] 
  ];

272
 fun_decl: [
273 274
(* need an hack to do this, because both productions would
   match   [ OPT LIDENT; "("; pat ] .... *)
275
   [ f = OPT [ x = LIDENT -> ident (parse_ident x)]; "("; p1 = pat LEVEL "no_arrow";
276 277 278 279 280 281 282 283 284 285 286 287 288
    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
289 290
	    let targ = multi_prod nopos (List.map snd args) in
	    let arg = multi_prod nopos (List.map fst args) in
291 292 293
	    let b = [arg, body] in
	    let a = [targ,tres] in
	    (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: [ 
311 312 313 314 315
    [ x = regexp; "|"; y = regexp -> 
	match (x,y) with
	  | Elem x, Elem y -> Elem (mk loc (Or (x,y)))
	  | _ -> Alt (x,y) 
    ]
316
  | [ x = regexp; y = regexp -> Seq (x,y) ]
317
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident (parse_ident a),x) ]
318 319 320 321
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
322
    | x = regexp; "?" ->  Alt (x, Epsilon)
323 324
    | x = regexp; "??" -> Alt (Epsilon, x) ]
  | [ "("; 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 ->
329 330
	let i = Chars.mk_int (parse_char loc i)
	and j = Chars.mk_int (parse_char loc j) in
331
        Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
332 333
    |  s = STRING1 ->
	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
337 338 339
	     let c = Chars.atom c in
	     Seq (Elem (mk loc (Internal (Types.char c))), accu))
	  s
340
	  Epsilon 
341 342 343 344 345
    | e = pat LEVEL "simple" -> Elem e
    ]
  ];

  pat: [ 
346
      [ x = pat; LIDENT "where"; 
347 348 349
        b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)
		  | LIDENT -> error loc "Type/pattern identifiers must be capitalized"
		  ] SEP "and"
350 351
            -> mk loc (Recurs (x,b)) ]
    | 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
    | 
356 357
      [ "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; 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))
362 363
      | a = UIDENT -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
364 365
          let i =  Intervals.mk i 
	  and j =  Intervals.mk j in
366 367
          mk loc (Internal (Types.interval (Intervals.bounded i j)))
      | 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)))
376 377
      | i = char ->
          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)))
380 381 382 383
      | 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 
384
                 | -> mknoloc (Internal (Sequence.nil_type)) ]; 
385
             "]" -> mk loc (Regexp (r,q))
386 387
      | "<"; t =
        [ x = [ LIDENT | UIDENT | keyword ] ->
388
	    let a = if x = "_" then Atoms.any else Atoms.atom (Atoms.mk (parse_ident x)) in
389 390
	    mk loc (Internal (Types.atom a))
	| "("; t = pat; ")" -> t ];
391
	a = attrib_spec; ">"; c = pat ->
392
          mk loc (XmlT (t, multi_prod loc [a;c]))
393 394
      | s = STRING2 ->
	  let s = seq_of_string loc s in
395
	  let s = List.map 
396
		    (fun (loc,c) -> 
397 398 399
		       mk loc (Internal
				 (Types.char
				    (Chars.atom
400
				       (Chars.mk_int c))))) s in
401 402
	  let s = s @ [mk loc (Internal (Sequence.nil_type))] in
	  multi_prod loc s
403 404 405 406 407
      ]
    
  ];

  record_spec:
408
    [ [ r = LIST0 [ l = [LIDENT | UIDENT | keyword ]; "=";
409
                  o = [ "?" -> true | -> false]; 
410 411
                  x = pat -> 
		    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
      ] ];
  
417 418
  char:
    [ 
419
      [ c = STRING1 -> Chars.mk_int (parse_char loc c) ]
420 421 422
    ];
     

423 424
  const:
    [ 
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)) 
433 434 435 436
      | "("; t = pat; ")" -> t 
      | "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
      ] ];
437 438 439

  expr_record_spec:
    [ [ r = LIST1
440
	      [ l = [LIDENT | UIDENT | keyword ]; "="; x = expr -> 
441
		  (label l,x) ] 
442
	      SEP ";" ->
443
	  exp loc (RecordLitt (make_record loc r))
444 445 446 447 448
      ] ];
  
  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
449
      | -> exp loc (RecordLitt (LabelMap.empty)) 
450 451 452 453
      ] 
    ];
END

454 455 456 457 458
let pat'  = Grammar.Entry.create gram "type/pattern expression"
EXTEND GLOBAL: pat pat';
  pat': [  [ p = pat; EOI -> p ] ];
END

459
let pat = Grammar.Entry.parse pat
460 461
and expr = Grammar.Entry.parse expr
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)
466 467
  let expr s = expr (Stream.of_string s)
end
468