parser.ml 22.8 KB
Newer Older
1 2
open Cduce_loc
(* let raise = Pervasives.raise *)
3
open Ast
4
open Ident
5
open Printf
6
open Ulexer
7

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

12 13
let tloc (i,j) = (i,j)
let nopos = (-1,-1)
14 15


16
let mk loc x = Cduce_loc.mk_located (tloc loc) x
17

18
exception Error of string
19
let error (i,j) s = Cduce_loc.raise_loc i j (Error s)
20
let error loc s = error (tloc loc) s
21

22
module Gram = Camlp4.Struct.Grammar.Static.Make(Ulexer)
23

24 25
let id_dummy = U.mk "$$$"

Pietro Abate's avatar
Pietro Abate committed
26
let ident_aux s =
27 28 29 30 31 32 33 34
  let b = Buffer.create (String.length s) in
  let rec aux i =
    if (i = String.length s) then Buffer.contents b
    else match s.[i] with
      | '\\' -> assert (s.[i+1] = '.'); Buffer.add_char b '.'; aux (i+2)
      | c -> Buffer.add_char b c; aux (i+1)
  in
  aux 0
35

Pietro Abate's avatar
Pietro Abate committed
36 37
let label s = U.mk (ident_aux s)
let ident s = U.mk (ident_aux s)
38

39 40 41 42 43 44
let prog    = Gram.Entry.mk "prog"
let top_phrases   = Gram.Entry.mk "toplevel phrases"
let expr    = Gram.Entry.mk "expression"
let pat     = Gram.Entry.mk "type/pattern expression"
let regexp  = Gram.Entry.mk "type/pattern regexp"
let keyword = Gram.Entry.mk "keyword"
45

46
let lop pos = Cduce_loc.loc_of_pos (tloc pos)
47
let exp pos e = LocatedExpr (lop pos,e)
48

49 50 51 52
let rec multi_prod loc = function
  | [ x ] -> x
  | x :: l -> mk loc (Prod (x, multi_prod loc l))
  | [] -> assert false
53

54
let rec tuple = function
55
  | [ x ] -> x
56
  | x :: l -> Pair (x, tuple l)
57
  | [] -> assert false
58

59
let tuple_queue =
60
  List.fold_right (fun x q -> Pair (x, q))
61 62


63
let char = mknoloc (Internal (Types.char Chars.any))
64
let string_regexp = Star (Elem char)
65

66
let seq_of_string s =
67 68
  let s = Encodings.Utf8.mk s in
  let rec aux i j =
69 70
    if Encodings.Utf8.equal_index i j then []
    else let (c,i) = Encodings.Utf8.next s i in c :: (aux i j)
71
  in
72
  aux (Encodings.Utf8.start_index s) (Encodings.Utf8.end_index s)
73

74

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

80 81
let include_stack = ref []

82 83 84 85
let protect_exn f g =
  try let x = f () in g (); x
  with e -> g (); raise e

86
let localize_exn f =
87 88
  try f ()
  with
89 90 91 92 93
  | Ulexer.Loc.Exc_located (_, (Location _ as e)) -> raise e
(*  | Camlp4.PreCast.Loc.Exc_located ((i,j), e) -> raise_loc i j e *)
  | Ulexer.Loc.Exc_located (loc, e) ->
      let i, j = Ulexer.Loc.start_off loc, Ulexer.Loc.stop_off loc in
      raise_loc i j e
94

95
let is_fun_decl =
96
   Gram.Entry.of_parser "[is_fun_decl]"
97 98
     (fun strm ->
       match Stream.npeek 3 strm with
99 100
	 | [ KEYWORD "fun", _; IDENT _, _; KEYWORD "(", _ ]
	 | [ IDENT _, _; KEYWORD "(", _; _ ] -> ()
101 102 103
	 | _ -> raise Stream.Failure
     )

104
let is_capture =
105
   Gram.Entry.of_parser "[is_capture]"
106 107
     (fun strm ->
       match Stream.npeek 2 strm with
108
	 | [ IDENT _, _; KEYWORD "::", _; _ ] -> ()
109 110 111 112
	 | _ -> raise Stream.Failure
     )


113 114 115
let if_then_else cond e1 e2 =
  Match (cond, [(mk (0,0) (Cst (Atom (ident "true")))),e1;
		(mk (0,0) (Cst (Atom (ident "false")))),e2])
116

117 118 119
let logical_and e1 e2 = if_then_else e1 e2 cst_false
let logical_or e1 e2 = if_then_else e1 cst_true e2
let logical_not e = if_then_else e cst_false cst_true
120

121
let apply_op2_noloc op e1 e2 = Apply (Apply (Var (ident op), e1), e2)
122
let apply_op2 loc op e1 e2 = exp loc (apply_op2_noloc op e1 e2)
123

124 125
let set_ref e1 e2 = Apply (Dot (e1, U.mk "set"), e2)
let get_ref e = Apply (Dot (e, U.mk "get"), cst_nil)
126 127 128
let let_in e1 p e2 =  Match (e1, [p,e2])
let seq e1 e2 = let_in e1 pat_nil e2
let concat e1 e2 = apply_op2_noloc "@" e1 e2
129

130
EXTEND Gram
131
  GLOBAL: top_phrases prog expr pat regexp keyword;
132 133

  top_phrases: [
134
    [ l = LIST0 phrase; ";;" -> List.flatten l  ]
135
  ];
136 137

  prog: [
138
    [ l = LIST0 [ p = phrase ; OPT ";;" -> p ]; `EOI -> List.flatten l ]
139 140 141
  ];

  phrase: [
142
    [ (f,p,e) = let_binding ->
143 144
	if f then [ mk _loc (FunDecl e) ] else
	  [ mk _loc (LetDecl (p,e)) ]
145
    | (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
146 147
        [ mk _loc (EvalStatement (exp _loc (let_in e1 p e2))) ]
    | "type"; x = located_ident; "="; t = pat -> [ mk _loc (TypeDecl (x,t)) ]
148
    | "using"; name = IDENT; "="; cu = [ x = IDENT -> x | x = STRING -> x ] ->
149
	[ mk _loc (Using (U.mk name, U.mk cu)) ]
150
    | "open"; ids = LIST1 ident_or_keyword SEP "." ->
151
	let ids = List.map (fun x -> ident x) ids in
152
	[ mk _loc (Open ids) ]
153
    | "schema"; name = IDENT; "="; uri = STRING ->
154
	protect_op "schema";
155
        [ mk _loc (SchemaDecl (U.mk name, uri)) ]
156 157 158 159
    | n = namespace_binding ->
	let d = match n with
	    | `Prefix (name,ns) ->  Namespace (name, ns)
	    | `Keep b -> KeepNs b in
160
	[ mk _loc d ]
161
    | n = namespace_binding; "in"; e2 = expr LEVEL "top" ->
162
	let e =
163 164 165 166
	  match n with
	    | `Prefix (name,ns) -> NamespaceIn (name, ns, e2)
	    | `Keep b -> KeepNsIn (b,e2)
	in
167 168 169 170
	[ mk _loc (EvalStatement (exp _loc e)) ]
    | "debug"; d = debug_directive -> [ mk _loc (Directive (`Debug d)) ]
    | "#"; IDENT "verbose" -> [ mk _loc (Directive `Verbose) ]
    | "#"; IDENT "silent" -> [ mk _loc (Directive `Silent) ]
171 172 173
    | "#"; IDENT "utf8" -> Ulexer.enc := Ulexing.Utf8; [ ]
    | "#"; IDENT "latin1" -> Ulexer.enc := Ulexing.Latin1; [ ]
    | "#"; IDENT "ascii" -> Ulexer.enc := Ulexing.Ascii; [ ]
174 175
    | "#"; IDENT "quit" -> [ mk _loc (Directive `Quit) ]
    | "#"; IDENT "env" -> [ mk _loc (Directive `Env) ]
176
    | "#"; IDENT "print_type"; t = pat ->
177 178 179
        [ mk _loc (Directive (`Print_type t)) ]
    | "#"; IDENT "dump_value"; e = expr -> [ mk _loc (Directive (`Dump e)) ]
    | "#"; IDENT "reinit_ns" -> [ mk _loc (Directive `Reinit_ns) ]
180 181
    | "#"; IDENT "help" -> [ mk _loc (Directive (`Help None)) ]
    | "#"; IDENT "help"; "debug" -> [ mk _loc (Directive (`Help (Some "debug"))) ]
182
    | "#"; IDENT "builtins" -> [ mk _loc (Directive `Builtins) ]
183
    | "include"; s = STRING ->
184
	protect_op "File inclusion";
185
	let s = Cduce_loc.resolve_filename s in
186
	(* avoid looping; should issue an error ? *)
187 188
	(* it is possible to have looping with x/../x/../x/.. ....
	   Need to canonicalize filename *)
189
	if List.mem s !include_stack then []
190 191
	else (
	  include_stack := s :: !include_stack;
192
	  Cduce_loc.push_source (`File s);
193 194
	  let saved_enc = !Ulexer.enc in
	  Ulexer.enc := Ulexing.Latin1;
195 196 197
	  protect_exn
	    (fun () ->
	       let chan = open_in s in
198 199 200
	       protect_exn
		 (fun () ->
		    let input = Stream.of_channel chan in
201
		    localize_exn (fun () -> Gram.parse prog Ulexer.Loc.ghost input))
202
		 (fun () -> close_in chan))
203
	    (fun () ->
204
	       Ulexer.enc := saved_enc;
205
	       Cduce_loc.pop_source ();
206
	       include_stack := List.tl !include_stack)
207
	)
208
    ] |
209
    [ e = expr -> [ mk _loc (EvalStatement e) ]
210 211 212 213
    ]
  ];

  debug_directive: [
214 215 216 217 218
    [ IDENT "filter"; t = pat; p = pat -> `Filter(t,p)
    | IDENT "accept"; p = pat -> `Accept p
    | IDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
    | IDENT "sample"; t = pat -> `Sample t
    | IDENT "subtype"; t1 = pat; t2 = pat -> `Subtype (t1,t2)
219
    | IDENT "single"; t = pat -> `Single t
220 221
    | IDENT "typed"; e = expr -> `Typed e
    | IDENT "lambda"; e = expr -> `Lambda e
222
    ]
223
  ];
224

225
  keyword: [
226
    [ a =
227 228 229
	[ "map" | "match" | "with" | "try" | "xtransform"
	| "if" | "then"	| "else"
	| "transform" | "fun" | "in"
230
	| "let" | "type" | "debug" | "include"
231
        | "and" | "or" | "validate" | "schema" | "namespace" | "ref" | "alias"
232
	| "not" | "as" | "where" | "select" | "from" | "open"
233 234 235 236 237
	]
	-> a
    ]
  ];

238 239 240 241 242
  ident_or_keyword: [
    [ s = IDENT -> s
    | s = keyword -> s ]
  ];

243 244
  expr: [
    "top" RIGHTA
245
    [ "match"; e = SELF; "with"; b = branches ->
246
	exp _loc (Match (e,b))
247
    | "try"; e = SELF; "with"; b = branches ->
248
	exp _loc (Try (e,b))
249
    | "map"; e = SELF; "with"; b = branches ->
250
	exp _loc (Map (e,b))
251
    | "xtransform"; e = SELF; "with"; b = branches ->
252
	exp _loc (Xtrans (e,b))
253 254
    | "if"; e = SELF; "then"; e1 = SELF ->
	exp _loc (if_then_else e e1 cst_nil)
255
    | "if"; e = SELF; "then"; e1 = SELF; "else"; e2 = SELF ->
256
	exp _loc (if_then_else e e1 e2)
257
    | "transform"; e = SELF; "with"; b = branches ->
258
	exp _loc (Transform (e,b))
259
    | "validate"; e = SELF; "with"; (schema, typ) = schema_ref ->
260
        exp _loc (Validate (e, [schema;typ]))
261
    | "select"; e = SELF; "from";
262
      l = LIST1 [ x = pat ; "in"; e = expr -> (x,e)] SEP "," ;
263 264
      cond = [ "where"; c = LIST1 [ expr ] SEP "and" -> c
	     | -> [] ] -> exp _loc (SelectFW (e,l,cond))
265
    | "fun"; (f,a,b) = fun_decl ->
266
	exp _loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
267
    | (_,p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
268
        exp _loc (let_in e1 p e2)
269 270
    | n = namespace_binding; "in"; e2 = expr LEVEL "top" ->
	(match n with
271 272
	   | `Prefix (name,ns) -> exp _loc (NamespaceIn (name, ns, e2))
	   | `Keep f -> exp _loc (KeepNsIn (f,e2)))
273
    | e = expr; ":"; p = pat ->
274
	exp _loc (Forget (e,p))
275
    | e = expr; ":"; "?"; p = pat ->
276
	exp _loc (Check (e,p))
277
    | e1 = expr; ";"; e2 = expr ->
278
	exp _loc (seq e1 e2)
279
    | "ref"; p = pat; e = expr ->
280 281
	exp _loc (Ref (e,p))
    | "not"; e = expr -> exp _loc (logical_not e)
282 283
    ]
    |
284
    [ e1 = expr; ":="; e2 = expr -> exp _loc  (set_ref e1 e2)
285
    ]
286 287
    |
    [ e1 = expr; op = ["=" | "<=" | "<<" | ">>" | ">=" ]; e2 = expr ->
288 289 290 291
	let op = match op with
	  | "<<" -> "<"
	  | ">>" -> ">"
	  | s -> s in
292
	apply_op2 _loc op e1 e2
293 294
    ]

295
    |
296 297
    [ e1 = expr; op = ["+" | "-" | "@" ]; e2 = expr -> apply_op2 _loc op e1 e2
    | e1 = expr; ["||" | "or"]; e2 = expr -> exp _loc (logical_or e1 e2)
298 299
    | e = expr; "\\"; l = ident_or_keyword ->
	exp _loc (RemoveField (e, label l))
300 301
    ]
    |
302 303
    [ e1 = expr; op = ["*"]; e2 = expr -> apply_op2 _loc op e1 e2
    | e1 = expr; "&&"; e2 = expr -> exp _loc (logical_and e1 e2)
304
    | e = expr; op = "/"; p = pat LEVEL "simple" ->
305
	(* transform e with <(Atom)>[($$$::t|_)*] -> [$$$] *)
306
	let tag = mk _loc (Internal (Types.atom Atoms.any)) in
307 308
	let att = mk _loc (Internal Types.Record.any) in
	let any = mk _loc (Internal Types.any) in
309
	let re = Star(Alt(SeqCapture(noloc,id_dummy,Elem p), Elem any)) in
310 311 312
	let ct = mk _loc (Regexp re) in
        let p = mk _loc (XmlT (tag, multi_prod _loc [att;ct])) in
	exp _loc (Transform (e,[p, Var id_dummy]))
313
    | e = expr; "/@"; a = ident_or_keyword ->
314
	(* transform e with <(Atom) {a=$$$}>_ -> [$$$] *)
315
        let tag = mk _loc (Internal (Types.atom Atoms.any)) in
316 317
        let any = mk _loc (Internal Types.any) in
        let att = mk _loc (Record
318
			    (true, [(label a,
319
				     (mk _loc (PatVar [id_dummy]),
320
				      None))])) in
321
        let p = mk _loc (XmlT (tag, multi_prod _loc [att;any])) in
322
        let t = (p, Pair (Var id_dummy,cst_nil)) in
323
        exp _loc (Transform (e,[t]))
324
     | e = expr; "//" ; p = pat ->
325 326
	 (* let $xstack=ref [p*] [] in
            let f  ( x : [Any*]) : [Any*] =
327 328 329
                     xtransform x with $$$ & p & <_ ..>[y::Any* ] ->  $stack := !$stack @ [$$$] ; f y
            in let _ =f e in !$stack *)

330
	 let stk = U.mk "$stack" in
331 332 333
	 let y = U.mk "y" in
	 let x = U.mk "x" in
	 let f = U.mk "f" in
334 335
	 let assign =
	   set_ref
336
	     (Var stk)
337
	     (concat (get_ref (Var stk)) (Pair (Var id_dummy,cst_nil))) in
338
        	let tag = mknoloc (Internal (Types.atom Atoms.any)) in
339 340
	        let att = mknoloc (Internal Types.Record.any) in
	        let any = mknoloc (Internal Types.any) in
341
	        let re = (SeqCapture(noloc,y,Star(Elem(any)))) in
342 343
	        let ct = mknoloc (Regexp re) in
                let children = mknoloc (XmlT (tag, multi_prod _loc [att;ct])) in
344
         let capt = mknoloc (And (mknoloc (And (mknoloc (PatVar [id_dummy]),p)),children)) in
345
	 let assign = seq assign ( (Apply(Var(f) , Var(y) ) ) ) in
346 347
	 let xt = Xtrans ((Var x),[capt,assign]) in
	 let rf = Ref (cst_nil, mknoloc (Regexp (Star(Elem p)))) in
348
         let targ = mknoloc (Regexp(Star(Elem(any)))) in
349
	 let tres = targ in
350
	 let arg = mknoloc(PatVar [x]) in
351
         let abst = {fun_name = Some (lop _loc,ident "f") ; fun_iface = [(targ, tres)] ;fun_body = [(arg,xt)] } in
352
	 let body =
353
	   let_in rf (mknoloc (PatVar [stk]))
354
	    (let_in ((Abstraction abst)) (mknoloc (PatVar[ident "f"]))
355
              (let_in ((Apply(Var(f) , e) )  ) (mknoloc (Internal Types.any)) (get_ref (Var stk))))
356
	 in
357
	 exp _loc body
358
    ]
359
    | [
360 361 362
      e1 = SELF; IDENT "div"; e2 = expr -> apply_op2 _loc "/" e1 e2
    | e1 = SELF; IDENT "mod"; e2 = expr -> apply_op2 _loc "mod" e1 e2
    | e1 = SELF; e2 = expr -> exp _loc (Apply (e1,e2))
363 364
    ]

365
    | "no_appl"
366

367
    [ e = expr;  "."; l = ident_or_keyword;
368 369 370
      tyargs = [ "with"; "{"; tyargs = LIST0 pat; "}" -> Some tyargs
      | -> None ] ->
	let e = Dot (e,label l) in
371
	match tyargs with None -> exp _loc e | Some tyargs ->
372
	  exp _loc (TyArgs (e,tyargs))
373 374
    ]
    | [
375
      "("; l = LIST1 expr SEP ","; ")" -> exp _loc (tuple l)
376
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ];
377
      loc_end = ["]" -> _loc] ->
378
	let e = match e with Some e -> e | None -> cst_nil in
379 380
        let e = exp loc_end e in
        let (_,loc_end) = loc_end in
381
	let l = List.fold_right
382 383
		  (fun x q ->
		     match x with
384
		       | `String (loc,i,j,s) -> exp loc (String (i,j,s,q))
385
		       | `Elems ((loc,_),x) -> exp (loc,loc_end) (Pair(x,q))
386
		       | `Explode x -> concat x q
387 388
		  ) l e
	in
389
	exp _loc l
390
    | "<"; t = [ "("; e = expr; ")" -> e
391
	       | a = tag -> exp _loc a
392
	       ];
393
	a = expr_attrib_spec; ">"; c = expr ->
394
	  exp _loc (Xml (t, Pair (a,c)))
395
    | "{"; r = expr_record_spec; "}" -> r
396
    | s = STRING ->
397
	let s = U.mk s in
398 399 400 401
	exp _loc (String (U.start_index s, U.end_index s, s, cst_nil))
    | a = IDENT -> exp _loc (Var (ident a))
    | "!"; e = expr -> exp _loc (get_ref e)
    | i = INT -> exp _loc (Integer (Intervals.V.mk i))
402
    | "`"; a = tag -> a
403
    | c = char -> exp _loc (Char c)
404 405 406
    ]

  ];
407

408
  tag: [ [ a = ident_or_keyword -> exp _loc (Atom (ident a)) ] ];
409 410

  tag_type: [
411
    [ "_" ->  mk _loc (Internal (Types.atom Atoms.any))
412
    | a = ident_or_keyword -> mk _loc (Cst (Atom (ident a)))
413
    | t = ANY_IN_NS -> mk _loc (NsT (ident t))
414 415 416
    ]
  ];

417
  seq_elem: [
418
    [ x = STRING2 ->
419 420
      let s = U.mk x in
      `String (_loc, U.start_index s, U.end_index s, s)
421
    | e = expr LEVEL "no_appl" -> `Elems (_loc,e)
422
    | "!"; e = expr LEVEL "no_appl" -> `Explode e
423 424
    ]
  ];
425

426
  namespace_binding: [
427
    [ "namespace"; r = [
428
	[ name =
429
	    [ name = ident_or_keyword; "=" -> ident name
430
	    | -> U.mk "" ];
431
	  ns = ns_expr -> `Prefix (name,ns)
432 433 434
	| IDENT "on" -> `Keep true
	| IDENT "off" -> `Keep false ]
      ] -> r ]
435 436
  ];

437
  ns_expr: [
438 439
    [ uri = STRING -> `Uri (Ns.Uri.mk (ident uri))
    | ids = LIST1 ident_or_keyword SEP "." ->
440 441 442 443
	let ids = List.map (fun x -> ident x) ids in
	`Path ids ]
  ];

444

445
  let_binding: [
446 447
    [ "let"; is_fun_decl; OPT "fun"; (f,a,b) = fun_decl ->
	let f = match f with Some x -> x | None -> assert false in
448
	let p = mk _loc (PatVar [snd f]) in
449
	let abst = { fun_name = Some f; fun_iface = a; fun_body = b } in
450
        let e = exp _loc (Abstraction abst) in
451
        (true,p,e)
452 453
    | "let"; p = pat; "="; e = expr -> (false,p,e)
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (false,p, Forget (e,t))
454
    | "let"; p = pat; ":"; "?"; t = pat; "="; e = expr ->
455
	(false,p, Check (e,t))
456
    ]
457 458
  ];

459
 fun_decl_after_lparen: [
460
(* need an hack to do this, because both productions would
461
   match   [ OPT IDENT; "("; pat ] .... *)
462 463 464 465 466
   [ p1 = pat LEVEL "no_arrow";
     res = [ "->"; p2 = pat;
	     a = [ ";"; a = LIST0 arrow SEP ";" -> a | -> [] ];
	     ")"; b = branches -> `Classic (p2,a,b)
	   | ":"; targ1 = pat;
467
	     args = LIST0 [ ","; arg = pat; ":"; targ = pat -> (arg,targ) ];
468
	     ")";
469
	     others = LIST0
470
			[ "(";
471 472
			  args =
			    LIST1
473 474 475
			      [ arg = pat; ":"; targ = pat -> (arg,targ) ]
			      SEP ",";
			  ")" -> args ];
476
	     ":"; tres = pat ;
477
	     "="; body = expr ->
478
	       `Compact (targ1,args,others,tres,body)
479 480 481
	   ] ->
       match res with
	 | `Classic (p2,a,b) -> (p1,p2)::a,b
482 483 484 485 486 487 488 489 490
	 | `Compact (targ1,args,others,tres,body) ->
	     let mkfun args =
	       multi_prod nopos (List.map snd args),
	       multi_prod nopos (List.map fst args)
	     in
	     let (tres,body) =
	       List.fold_right
		 (fun args (tres,body) ->
		    let (targ,arg) = mkfun args in
491 492
		    let e = Abstraction
			      { fun_name = None; fun_iface = [targ,tres];
493 494 495 496 497 498 499
				fun_body = [arg,body] } in
		    let t = mknoloc (Arrow (targ,tres)) in
		    (t,e)
		 )
		 others (tres,body) in
	     let (targ,arg) = mkfun ((p1,targ1) :: args) in
	     [(targ,tres)],[(arg,body)]
500
	      ] ];
501 502 503


 fun_decl: [
504
   [ f = OPT located_ident; "("; (a,b) = fun_decl_after_lparen ->
505
       (f,a,b)
506
   ]
507 508
 ];

509
 arrow: [
510
    [ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
511 512 513
  ];

  branches: [
514
    [ OPT "|"; l = LIST1 branch SEP "|" -> l ]
515 516 517
  ];

  branch: [
518
    [ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
519 520
  ];

521 522 523

  regexp: [
    [ x = regexp; "|"; y = regexp ->
524
	match (x,y) with
525
	  | Elem x, Elem y -> Elem (mk _loc (Or (x,y)))
526
	  | _ -> Alt (x,y)
527
    ]
528
  | [ x = regexp; y = regexp -> Seq (x,y) ]
529 530
  | [ x = regexp; "&"; y = regexp ->
	match (x,y) with
531 532
	  | Elem x, Elem y -> Elem (mk _loc (And (x,y)))
	  | _ -> error _loc "Conjunction not allowed in regular expression"
533
    ]
534
  | [ a = IDENT; "::"; x = regexp -> SeqCapture (lop _loc,ident a,x) ]
535 536 537 538
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
539
    | x = regexp; "?" ->  Alt (x, Epsilon)
540
    | x = regexp; "??" -> Alt (Epsilon, x)
541 542 543 544 545
    | x = regexp; "**"; i = INT ->
	let rec aux i accu =
	  if (i = 0) then accu
	  else aux (pred i) (Seq (x, accu))
	in
546 547
	let i =
	  try
548 549 550
	    let i = int_of_string i in
	    if (i > 1024) then raise Exit else i
	      (* We cannot handle type that huge... *)
551
	  with Failure _ | Exit -> error _loc "Repetition number too large"
552
	in
553
	if (i <= 0) then
554
	  error _loc "Repetition number must be a positive integer";
555 556
	aux i Epsilon
    ]
557
  | [ "("; x = LIST1 regexp SEP ","; ")" ->
558 559 560
	(match x with
	  | [ x ] -> x
	  | _ ->
561 562
	      let x =
		List.map
563 564
		  (function
		     | Elem x -> x
565
		     | _ -> error _loc
566 567
			 "Mixing regular expressions and products")
		  x in
568
	      Elem (multi_prod _loc x))
569
    | "("; a = IDENT; ":="; c = expr; ")" ->
570
	Elem (mk _loc (Constant ((ident a,c))))
571
    | "/"; p = pat LEVEL "simple" -> Guard p
572
    | IDENT "PCDATA" -> string_regexp
573
    | i = STRING2; "--"; j = STRING2 ->
574 575 576
      let i = Chars.V.mk_int (parse_char _loc i)
      and j = Chars.V.mk_int (parse_char _loc j) in
      Elem (mk _loc (Internal (Types.char (Chars.char_class i j))))
577
    | s = STRING2 ->
578
	List.fold_right
579
	  (fun c accu ->
580
	     let c = Chars.V.mk_int c in
581
	     let c = Chars.atom c in
582
	     Seq (Elem (mknoloc (Internal (Types.char c))), accu))
583
	  (seq_of_string s)
584 585
	  Epsilon ]
    | [ e = pat LEVEL "simple" -> Elem e
586 587 588
    ]
  ];

589
  schema_ref: [
590
    [ schema = IDENT; "."; typ = ident_or_keyword -> (U.mk schema, ident typ)
591 592 593
    ]
  ];

594
  located_ident: [ [ a = ident_or_keyword -> (lop _loc,ident a) ] ];
595

596
  pat: [
Pietro Abate's avatar
Pietro Abate committed
597 598 599
    [ x = pat; "where";
      b = LIST1 [ (la,a) = located_ident; "="; y = pat -> (la,a,y) ] SEP "and" -> 
        mk _loc (Recurs (x,b)) ]
600 601 602
    | RIGHTA [ x = pat; "->"; y = pat -> mk _loc (Arrow (x,y))
             | x = pat; "@"; y = pat -> mk _loc (Concat (x,y))
             | x = pat; "+"; y = pat -> mk _loc (Merge (x,y)) ]
603 604
    | "no_arrow" [ x = pat; "|"; y = pat -> mk _loc (Or (x,y)) ]
    | "simple" [ x = pat; "&"; y = pat -> mk _loc (And (x,y))
605
	       | x = pat; "\\"; y = pat -> mk _loc (Diff (x,y)) ]
Pietro Abate's avatar
Pietro Abate committed
606 607
    | "var_typ" [ x = PTYPE -> 
        mk _loc (Internal (Types.var (Var.mk (ident_aux x)))) ]
608
    |
609
      [ "{"; r = record_spec; "}" -> r
610
      | "ref"; p = pat ->
611
	  let get_fun = mk _loc (Arrow (pat_nil, p))
612
	  and set_fun = mk _loc (Arrow (p, pat_nil))in
613
	  let fields =
614
	    [ label "get", (get_fun, None); label "set", (set_fun, None) ] in
615 616
	  mk _loc (Record (false, fields))
      | "_" -> mk _loc (Internal Types.any)
617
      | "("; a = IDENT; ":="; c = expr; ")" ->