parser.ml 15 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 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)
16

17
18
let parse_ident = Encodings.Utf8.mk_latin1

19
20
let id_dummy = ident (U.mk "$$$")
let atom s = Atoms.mk (parse_ident s)
21
let label s = LabelPool.mk (parse_ident s)
22
let ident s = ident (parse_ident s)
23

24
let prog    = Grammar.Entry.create gram "prog"
25
let top_phrases   = Grammar.Entry.create gram "toplevel phrases"
26
27
28
29
30
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"
		
31
32
let exp pos e = LocatedExpr (loc_of_pos pos,e)

33
34
35
36
37
let rec multi_prod loc = function
  | [ x ] -> x
  | x :: l -> mk loc (Prod (x, multi_prod loc l))
  | [] -> assert false
      
38
let rec tuple = function
39
  | [ x ] -> x
40
  | x :: l -> Pair (x, tuple l)
41
  | [] -> assert false
42
43

let tuple_queue = 
44
  List.fold_right (fun x q -> Pair (x, q))
45
46


47
let char = mknoloc (Internal (Types.char Chars.any))
48
let string_regexp = Star (Elem char)
49
	       
50
let cst_nil =  Cst (Types.Atom Sequence.nil_atom)
51
		 
52
let seq_of_string pos s =
53
  let s =  Encodings.Utf8.mk s in
54
  let (pos,_) = pos in
55
56
57
  let rec aux pos i j =
    if Encodings.Utf8.equal_index i j then []
    else
58
      let (len,i) = Encodings.Utf8.next s i in
59
      let (c,i) = Encodings.Utf8.next s i in
60
      ((pos,pos+len),c) :: (aux (pos + len) i j)
61
  in
62
  aux (pos+1) (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s)
63

64
65
66
67
68
69
70
71
72
73
74
75
76
77
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

78
exception Error of string
79
let error (i,j) s = Location.raise_loc i j (Error s)
80

81
82
83
let make_record loc r =
  LabelMap.from_list (fun _ _ -> error loc "Duplicated record field") r

84
let parse_char loc s =
85
86
  let s = seq_of_string loc s in
  match s with
87
    | [ loc,c ] -> c
88
    | _ -> error loc "Character litteral must have length 1"
89

90
91
let char_list loc s = 
  let s = seq_of_string loc s in
92
  List.map (fun (loc,c) -> exp loc (Cst (Types.Char (Chars.mk_int c)))) s
93
94


95
96
let include_stack = ref []

97
98
99
100
let protect_exn f g =
  try let x = f () in g (); x
  with e -> g (); raise e

101
102
103
104
105
106
107
108
109
110
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
     )


111
EXTEND
112
113
114
  GLOBAL: top_phrases prog expr pat regexp const;

  top_phrases: [
115
    [ l = LIST0 phrase; ";;" -> List.flatten l  ]
116
  ];
117
118

  prog: [
119
    [ l = LIST0 [ p = phrase ; OPT ";;" -> p ]; EOI -> List.flatten l ]
120
121
122
  ];

  phrase: [
123
124
125
126
    [ (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"->
127
        [ mk loc (EvalStatement (exp loc (Match (e1,[p,e2])))) ]
128
129
130
    | "type"; x = UIDENT; "="; t = pat -> [ mk loc (TypeDecl (x,t)) ]
    | "type"; x = LIDENT -> error loc "Type identifiers must be capitalized"
    | "debug"; d = debug_directive -> [ mk loc (Debug d) ]
131
132
    | DIRECTIVE "#quit" -> [ mk loc (Directive `Quit) ]
    | DIRECTIVE "#env" -> [ mk loc (Directive `Env) ]
133
    | "include"; s = STRING2 ->
134
	let s = get_string s in
135
136
137
138
	let s = 
	  if Filename.is_relative s 
	  then Filename.concat (Location.current_dir ()) s
	  else s in
139
140
	protect_op "File inclusion";
	(* avoid looping; should issue an error ? *)
141
142
	(* it is possible to have looping with x/../x/../x/.. ....
	   Need to canonicalize filename *)
143
144
145
146
	if List.mem s !include_stack then [] 
	else (
	  include_stack := s :: !include_stack;
	  Location.push_source (`File s);
147
148
149
	  protect_exn
	    (fun () ->
	       let chan = open_in s in
150
151
152
153
154
	       protect_exn
		 (fun () ->
		    let input = Stream.of_channel chan in
		    Grammar.Entry.parse prog input)
		 (fun () -> close_in chan))
155
156
157
	    (fun () ->
	       Location.pop_source ();
	       include_stack := List.tl !include_stack)
158
	)
159
    ] | 
160
    [ e = expr -> [ mk loc (EvalStatement e) ]
161
162
163
164
165
    ]
  ];

  debug_directive: [
    [ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
166
    | LIDENT "accept"; p = pat -> `Accept p
167
    | LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
168
    | LIDENT "sample"; t = pat -> `Sample t
169
    | LIDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
170
    ]
171
  ];
172

173
174
175
176
177
  keyword: [
    [ a = 
	[ "map" | "match" | "with" | "try" | "xtransform"
	| "if" | "then"	| "else"
	| "transform" | "fun" | "in"
178
	| "let" | "type" | "debug" | "include"
179
        | "and"
180
181
182
183
184
	]
	-> a
    ]
  ];

185
186
  expr: [
    "top" RIGHTA
187
188
    [ "match"; e = SELF; "with"; b = branches -> 
	exp loc (Match (e,b))
189
    | "try"; e = SELF; "with"; b = branches -> 
190
	exp loc (Try (e,b))
191
192
193
194
    | "map"; e = SELF; "with"; b = branches -> 
	exp loc (Map (e,b))
    | "xtransform"; e = SELF; "with"; b = branches -> 
	exp loc (Xtrans (e,b))
195
    | "if"; e = SELF; "then"; e1 = SELF; "else"; e2 = SELF ->
196
197
	let p1 = mk loc (Internal true_type)
	and p2 = mk loc (Internal false_type) in
198
	exp loc (Match (e, [p1,e1; p2,e2]))
199
    | "transform"; e = SELF; "with"; b = branches -> 
200
	exp loc (Transform (e,b))
201
    | "fun"; (f,a,b) = fun_decl ->
202
	exp loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
203
    | (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
204
        exp loc (Match (e1,[p,e2]))
205
    | e = expr; ":"; p = pat ->
206
	exp loc (Forget (e,p))
207
208
209
    ]

    
210
211
212
213
214
215
    | 
    [ e1 = expr; op = ["=" | "<=" | "<<" | ">>" | ">=" ]; e2 = expr -> 
	let op = match op with
	  | "<<" -> "<"
	  | ">>" -> ">"
	  | s -> s in
216
	exp loc (Op (op,[e1;e2]))
217
218
    ]

219
    | 
220
    [ e1 = expr; op = ["+" | "-" | "@" ]; e2 = expr -> 
221
	exp loc (Op (op,[e1;e2]))
222
    | e = expr; "\\"; l = [LIDENT | UIDENT | keyword ] -> 
223
	exp loc (RemoveField (e, label l)) 
224
225
    ]
    |
226
    [ e1 = expr; op = ["*"]; e2 = expr -> exp loc (Op (op,[e1;e2]))  
227
228
229
230
231
    | 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
232
	let re = Star(Alt(SeqCapture(id_dummy,Elem p), Elem any)) in
233
234
	let ct = mk loc (Regexp (re,any)) in
        let p = mk loc (XmlT (tag, multi_prod loc [att;ct])) in
235
	let b = (p,  Var id_dummy) in
236
	exp loc (Transform (e,[b]))
237
    ]
238
    |
239
    [ e = expr;  "."; l = [LIDENT | UIDENT | keyword ] -> 
240
	exp loc (Dot (e, label l)) 
241
    ]  
242
 
243
    |
244
245
    [ op = [ LIDENT "flatten"
	   | LIDENT "load_xml"
246
	   | LIDENT "load_file" | LIDENT "load_file_utf8"
247
	   | LIDENT "load_html"
248
	   | LIDENT "print_xml" | LIDENT "print_xml_utf8"
249
	   | LIDENT "print"
250
	   | LIDENT "int_of"
251
	   | LIDENT "string_of"
252
	   | LIDENT "atom_of"
253
	   | LIDENT "raise"
254
	   ]; 
255
      e = expr -> exp loc (Op (op,[e]))
256
    | op = [ LIDENT "dump_to_file" | LIDENT "dump_to_file_utf8" ];
257
258
259
260
      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))
261
262
    ]

263
    | "no_appl" 
264
265
    [ c = const -> exp loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> exp loc (tuple l)
266
267
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
	let e = match e with Some e -> e | None -> cst_nil in
268
269
270
271
272
273
274
275
	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
276
277
    | "<"; t = [ "("; e = expr; ")" -> e
	       | a = [ LIDENT | UIDENT | keyword ] -> 
278
		   exp loc (Cst (Types.Atom (atom a))) ];
279
	a = expr_attrib_spec; ">"; c = expr ->
280
281
	exp loc (Xml (t, Pair (a,c)))
    | "{"; r = [ expr_record_spec | -> exp loc (RecordLitt LabelMap.empty) ]; "}" -> r
282
    | s = STRING2 ->
283
	exp loc (tuple (char_list loc s @ [cst_nil]))
284
    | a = LIDENT -> exp loc (Var (ident a))
285
286
287
    ]

  ];
288
289

  seq_elem: [
290
291
292
    [ x = STRING1 -> `Elems (char_list loc x)
    | e = expr LEVEL "no_appl" -> `Elems [e]
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
293
294
    ]
  ];
295
296
	  
  let_binding: [
297
298
299
300
    [ "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
301
        let e = exp loc (Abstraction abst) in
302
        (true,p,e)
303
304
    | "let"; p = pat; "="; e = expr -> (false,p,e)
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (false,p, Forget (e,t))
305
306
307
    ] 
  ];

308
 fun_decl_after_lparen: [
309
310
(* need an hack to do this, because both productions would
   match   [ OPT LIDENT; "("; pat ] .... *)
311
312
313
314
315
316
317
318
319
320
321
322
323
324
325
326
327
328
329
330
331
332
333
   [ 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 ->
334
	    (f,a,b)
335
   ]
336
337
 ];

338
 arrow: [
339
    [ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
340
341
342
  ];

  branches: [
343
    [ OPT "|"; l = LIST1 branch SEP "|" -> l ]
344
345
346
  ];

  branch: [
347
    [ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
348
349
350
351
  ];

	  
  regexp: [ 
352
353
354
355
356
    [ x = regexp; "|"; y = regexp -> 
	match (x,y) with
	  | Elem x, Elem y -> Elem (mk loc (Or (x,y)))
	  | _ -> Alt (x,y) 
    ]
357
  | [ x = regexp; y = regexp -> Seq (x,y) ]
358
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (ident a,x) ]
359
360
361
362
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
363
    | x = regexp; "?" ->  Alt (x, Epsilon)
364
365
    | x = regexp; "??" -> Alt (Epsilon, x) ]
  | [ "("; x = regexp; ")" -> x
366
    | "("; a = LIDENT; ":="; c = const; ")" -> 
367
	Elem (mk loc (Constant ((ident a,c))))
368
    | UIDENT "PCDATA" -> string_regexp
369
    | i = STRING1; "--"; j = STRING1 ->
370
371
	let i = Chars.mk_int (parse_char loc i)
	and j = Chars.mk_int (parse_char loc j) in
372
        Elem (mk loc (Internal (Types.char (Chars.char_class i j))))
373
374
    |  s = STRING1 ->
	let s = seq_of_string loc s in
375
	List.fold_right
376
	  (fun (loc,c) accu -> 
377
	     let c = Chars.mk_int c in
378
379
380
	     let c = Chars.atom c in
	     Seq (Elem (mk loc (Internal (Types.char c))), accu))
	  s
381
	  Epsilon 
382
383
384
385
386
    | e = pat LEVEL "simple" -> Elem e
    ]
  ];

  pat: [ 
387
      [ x = pat; LIDENT "where"; 
388
389
390
        b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)
		  | LIDENT -> error loc "Type/pattern identifiers must be capitalized"
		  ] SEP "and"
391
392
            -> mk loc (Recurs (x,b)) ]
    | RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
393
    | "no_arrow" [ x = pat; "|"; y = pat -> mk loc (Or (x,y)) ] 
394
    | "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y)) 
395
	       | x = pat; "\\"; y = pat -> mk loc (Diff (x,y)) ]
396
    | 
397
398
      [ "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
399
      | LIDENT "_" -> mk loc (Internal Types.any)
400
      | a = LIDENT -> mk loc (Capture (ident a))
401
      | "("; a = LIDENT; ":="; c = const; ")" -> 
402
	  mk loc (Constant (ident a,c))
403
404
      | a = UIDENT -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
405
406
          let i =  Intervals.mk i 
	  and j =  Intervals.mk j in
407
408
          mk loc (Internal (Types.interval (Intervals.bounded i j)))
      | i = INT -> 
409
          let i =  Intervals.mk i  in
410
          mk loc (Internal (Types.interval (Intervals.atom i)))
411
      | "*"; "--"; j = INT ->
412
	  let j =  Intervals.mk j in
413
          mk loc (Internal (Types.interval (Intervals.left j)))
414
      | i = INT; "--"; "*" ->
415
	  let i = Intervals.mk i in
416
          mk loc (Internal (Types.interval (Intervals.right i)))
417
418
      | i = char ->
          mk loc (Internal (Types.char (Chars.char_class i i)))
419
      | i = char ; "--"; j = char ->
420
          mk loc (Internal (Types.char (Chars.char_class i j)))
421
422
423
424
      | 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 
425
                 | -> mknoloc (Internal (Sequence.nil_type)) ]; 
426
             "]" -> mk loc (Regexp (r,q))
427
428
      | "<"; t =
        [ x = [ LIDENT | UIDENT | keyword ] ->
429
	    let a = if x = "_" then Atoms.any else Atoms.atom (atom x) in
430
431
	    mk loc (Internal (Types.atom a))
	| "("; t = pat; ")" -> t ];
432
	a = attrib_spec; ">"; c = pat ->
433
          mk loc (XmlT (t, multi_prod loc [a;c]))
434
435
      | s = STRING2 ->
	  let s = seq_of_string loc s in
436
	  let s = List.map 
437
		    (fun (loc,c) -> 
438
439
440
		       mk loc (Internal
				 (Types.char
				    (Chars.atom
441
				       (Chars.mk_int c))))) s in
442
443
	  let s = s @ [mk loc (Internal (Sequence.nil_type))] in
	  multi_prod loc s
444
445
446
447
448
      ]
    
  ];

  record_spec:
449
    [ [ r = LIST0 [ l = [LIDENT | UIDENT | keyword ]; "=";
450
                  o = [ "?" -> true | -> false]; 
451
452
                  x = pat -> 
		    let x = if o then mk loc (Optional x) else x in
453
		    (label l, x)
454
                ] SEP ";" ->
455
	  make_record loc r
456
457
      ] ];
  
458
459
  char:
    [ 
460
      [ c = STRING1 -> Chars.mk_int (parse_char loc c) ]
461
462
463
    ];
     

464
465
  const:
    [ 
466
      [ i = INT -> Types.Integer (Intervals.mk i)
467
      | "`"; a = [LIDENT | UIDENT | keyword ] -> Types.Atom (atom a)
468
      | c = char -> Types.Char c ]
469
470
471
472
    ];


  attrib_spec:
473
    [ [ r = record_spec -> mk loc (Record (true,r)) 
474
475
476
477
      | "("; t = pat; ")" -> t 
      | "{"; r = record_spec; "}" -> mk loc (Record (true,r))
      | "{|"; r = record_spec; "|}" -> mk loc (Record (false,r))
      ] ];
478
479
480

  expr_record_spec:
    [ [ r = LIST1
481
	      [ l = [LIDENT | UIDENT | keyword ]; "="; x = expr -> 
482
		  (label l,x) ] 
483
	      SEP ";" ->
484
	  exp loc (RecordLitt (make_record loc r))
485
486
487
488
489
      ] ];
  
  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
490
      | -> exp loc (RecordLitt (LabelMap.empty)) 
491
492
493
494
      ] 
    ];
END

495
let pat = Grammar.Entry.parse pat
496
497
and expr = Grammar.Entry.parse expr
and prog = Grammar.Entry.parse prog
498
499
and top_phrases = Grammar.Entry.parse top_phrases

500
501
502
503
504
505
506
507
508
509
510
511
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 ()