viewcvs.cgi/parser/parser.ml

open Location
open Ast

(* let ()  = Grammar.error_verbose := true *)
let gram    = Grammar.gcreate (Lexer.gmake ())
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
  | x :: l -> mk (x.loc) (Pair (x, tuple loc l))
  | [] -> assert false
      
let char = mk noloc (Internal (Types.char Chars.any))
let string = Star (Elem char)
               
let cst_nil =  mk noloc (Cst (Types.Atom Sequence.nil_atom))
                 
let seq_of_string s =
  let rec aux accu i = if (i = 0) then accu else aux (s.[i-1]::accu) (i-1) in
  aux [] (String.length s)


let char_list loc s = 
  let s = seq_of_string (Token.eval_string s) in
  List.map (fun c -> mk loc (Cst (Types.Char (Chars.Unichar.from_char c)))) s


EXTEND
  GLOBAL: prog expr pat regexp const;

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

  phrase: [
    [ e = expr -> EvalStatement e
    | "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t)
    | "debug"; d = debug_directive -> Debug d
    ]
  ];

  debug_directive: [
    [ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
    | LIDENT "accept"; p = pat -> `Accept p;
    | LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
    ]
  ];

  expr: [
    "top" RIGHTA
    [ "match"; e = SELF; "with"; b = branches -> mk loc (Match (e,b))
    | "map"; e = SELF; "with"; b = branches -> mk loc (Map (e,b))
    | "transform"; e = SELF; "with"; b = branches -> 
        mk noloc (Op ("flatten", [mk loc (Map (e,b))]))
    | "fun"; (f,a,b) = fun_decl ->
        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]))
    | e = expr; ":"; p = pat ->
        mk loc (Forget (e,p))
    ]

    
    | 
    [ e1 = expr; op = ["+" | "-" | "@"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
    ]
    |
    [ e1 = expr; op = ["*" | "/"]; e2 = expr -> mk loc (Op (op,[e1;e2]))  
    ]
    |
    [ e = expr;  "."; l = [LIDENT | UIDENT] -> mk loc (Dot (e,Types.label l)) 
    ]  
 
    |
    [ LIDENT "flatten"; e = expr -> mk loc (Op ("flatten",[e]))
    | LIDENT "load_xml"; e = expr -> mk loc (Op ("load_xml",[e]))
    | e1 = expr; e2 = expr -> mk loc (Apply (e1,e2))
    ]

    | "no_appl" 
    [ c = const -> mk loc (Cst c)
    | "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
    | "[";  l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
        let e = match e with Some e -> e | None -> cst_nil in
        let l = List.flatten l in
        tuple loc (l @ [e])
    | "<"; t = expr_tag_spec; a = expr_attrib_spec; ">"; c = expr ->
        tuple loc [t;a;c]
    | "{"; r = [ expr_record_spec | -> mk loc (RecordLitt []) ]; "}" -> r
    | s = STRING ->
        tuple loc (char_list loc s @ [cst_nil])
    | "!"; t = pat -> mk loc (DebugTyper t)
    | a = LIDENT -> mk loc (Var a)
    ]

  ];

  seq_elem: [
    [ x = [CHAR | STRING] -> char_list loc x
    | e = expr LEVEL "no_appl" -> [e]
    ]
  ];
          
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
    | "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, mk noloc (Forget (e,t)))
    | "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
        let e = mk loc (Abstraction abst) in
        (p,e);
    ] 
  ];

 fun_decl: [
  [ f = OPT LIDENT; "("; a = LIST0 arrow SEP ";"; ")"; b = branches -> 
      (f,a,b)
  | f = OPT LIDENT; "("; arg = LIDENT; ":"; targ = pat; ")"; ":"; tres = pat ; 
      "="; body = expr ->
        let b = [mk noloc (Capture arg), body] in
        let a = [targ,tres] in
        (f,a,b)
  ]
 ];

  arrow: [
    [ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
  ];

  branches: [
    [ OPT "|"; l = LIST1 branch SEP "|" ; OPT "end" -> l ]
  ];

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

          
  regexp: [ 
    [ x = regexp; "|"; y = regexp -> Alt (x,y) ]
  | [ x = regexp; y = regexp -> Seq (x,y) ]
  | [ a = LIDENT; "::"; x = regexp -> SeqCapture (a,x) ]
  | [ x = regexp; "*" -> Star x
    | x = regexp; "*?" -> WeakStar x
    | x = regexp; "+" -> Seq (x, Star x)
    | x = regexp; "+?" -> Seq (x, WeakStar x)
    | x = regexp; "?" -> Alt (x, Epsilon)
    | x = regexp; "??" -> Alt (Epsilon, x) ]
  | [ "("; x = regexp; ")" -> x
    | UIDENT "String" -> string
    | s = CHAR ->
        let s = seq_of_string (Token.eval_string s) in
        List.fold_right
          (fun c accu -> 
             let c = Chars.Unichar.from_char c in
             let c = Chars.atom c in
             Seq (Elem (mk loc (Internal (Types.char c))), accu))
          s
          Epsilon
    | 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)) ]
    | "no_arrow" [ x = pat; "|"; y = pat -> mk loc (Or (x,y)) ] 
    | "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y,true)) 
(*             | x = pat; ":"; y = pat -> mk loc (And (x,y,false))   *)
               | x = pat; "-"; y = pat -> mk loc (Diff (x,y)) ]
    | 
      [ "{"; r = record_spec; "}" -> r
      | LIDENT "_" -> mk loc (Internal Types.any)
      | a = LIDENT -> mk loc (Capture a)
      | "("; a = LIDENT; ":="; c = const; ")" -> mk loc (Constant (a,c))
      | a = UIDENT -> mk loc (PatVar a)
      | i = INT ; "--"; j = INT -> 
          let i = Big_int.big_int_of_string i 
          and j = Big_int.big_int_of_string j in
          mk loc (Internal (Types.interval (Intervals.bounded i j)))
      | i = INT -> 
          let i = Big_int.big_int_of_string i  in
          mk loc (Internal (Types.interval (Intervals.atom i)))
      | "*--"; j = INT ->
          let j = Big_int.big_int_of_string j in
          mk loc (Internal (Types.interval (Intervals.left j)))
      | i = INT; "--*" ->
          let i = Big_int.big_int_of_string i in
          mk loc (Internal (Types.interval (Intervals.right i)))
      | i = char ->
          mk loc (Internal (Types.char (Chars.char_class i i)))
      | i = char ; "--"; j = char ->
          mk loc (Internal (Types.char (Chars.char_class i j)))
      | 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 
                 | -> mk noloc (Internal (Sequence.nil_type)) ]; 
             "]" -> mk loc (Regexp (r,q))
      | "<"; t = tag_spec; a = attrib_spec; ">"; c = pat ->
          multi_prod loc [t;a;c]
      ]
    
  ];

  record_spec:
    [ [ r = LIST0 [ l = [LIDENT | UIDENT]; 
                  o = ["=?" -> true | "=" -> false]; 
                  x = pat ->
                    mk loc (Record (Types.label l,o,x))
                ] SEP ";" ->
        match r with
          | [] -> mk loc (Internal Types.Record.any)
          | h::t -> List.fold_left (fun t1 t2 -> mk loc (And (t1,t2,true))) h t
      ] ];
  
  char:
    [ 
      [ c = CHAR -> Chars.Unichar.from_char (Token.eval_char c)
      | "!"; i = INT -> Chars.Unichar.from_int (int_of_string i) ]
    ];
     

  const:
    [ 
      [ i = INT -> Types.Integer (Big_int.big_int_of_string i)
      | "`"; a = [LIDENT | UIDENT] -> Types.Atom (Types.mk_atom a)
      | c = char -> Types.Char c ]
    ];

  tag_spec:
    [
      [ LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any))) ]
    | [ a = [LIDENT | UIDENT] -> 
          mk loc (Internal (Types.atom (Atoms.atom (Types.mk_atom a)))) ]
    | [ t = pat -> t ]
    ];

  attrib_spec:
    [ [ r = record_spec -> r | "("; t = pat; ")" -> t ] ];

  expr_record_spec:
    [ [ r = LIST1
              [ l = [LIDENT | UIDENT]; "="; x = expr -> (Types.label l,x) ] 
              SEP ";" ->
          mk loc (RecordLitt r)
      ] ];
  
  expr_tag_spec:
    [
      [ a = [LIDENT | UIDENT] -> 
          mk loc (Cst (Types.Atom (Types.mk_atom a))) ]
    | [ e = expr LEVEL "no_appl" -> e ]
    ];

  expr_attrib_spec:
    [ [ r = expr_record_spec -> r ]
    | [ e = expr LEVEL "no_appl" -> e 
      | -> mk loc (RecordLitt []) 
      ] 
    ];
END

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

let pat = Grammar.Entry.parse pat
let expr = Grammar.Entry.parse expr
let prog = Grammar.Entry.parse prog

module From_string = struct
  let pat s = Grammar.Entry.parse pat' (Stream.of_string s)
  let expr s = expr (Stream.of_string s)
end

1	open Location
2	open Ast
3
4	(* let () = Grammar.error_verbose := true *)
5	let gram = Grammar.gcreate (Lexer.gmake ())
6	let prog = Grammar.Entry.create gram "prog"
7	let expr = Grammar.Entry.create gram "expression"
8	let pat = Grammar.Entry.create gram "type/pattern expression"
9	let regexp = Grammar.Entry.create gram "type/pattern regexp"
10	let const = Grammar.Entry.create gram "scalar constant"
11
12	let rec multi_prod loc = function
13	\| [ x ] -> x
14	\| x :: l -> mk loc (Prod (x, multi_prod loc l))
15	\| [] -> assert false
16
17	let rec tuple loc = function
18	\| [ x ] -> x
19	\| x :: l -> mk (x.loc) (Pair (x, tuple loc l))
20	\| [] -> assert false
21
22	let char = mk noloc (Internal (Types.char Chars.any))
23	let string = Star (Elem char)
24
25	let cst_nil = mk noloc (Cst (Types.Atom Sequence.nil_atom))
26
27	let seq_of_string s =
28	let rec aux accu i = if (i = 0) then accu else aux (s.[i-1]::accu) (i-1) in
29	aux [] (String.length s)
30
31
32	let char_list loc s =
33	let s = seq_of_string (Token.eval_string s) in
34	List.map (fun c -> mk loc (Cst (Types.Char (Chars.Unichar.from_char c)))) s
35
36
37	EXTEND
38	GLOBAL: prog expr pat regexp const;
39
40	prog: [
41	[ l = LIST0 [ p = phrase; ";;" -> mk loc p ]; EOI -> l ]
42	];
43
44	phrase: [
45	[ e = expr -> EvalStatement e
46	\| "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t)
47	\| "debug"; d = debug_directive -> Debug d
48	]
49	];
50
51	debug_directive: [
52	[ LIDENT "filter"; t = pat; p = pat -> `Filter(t,p)
53	\| LIDENT "accept"; p = pat -> `Accept p;
54	\| LIDENT "compile"; t = pat; p = LIST1 pat -> `Compile (t,p)
55	]
56	];
57
58	expr: [
59	"top" RIGHTA
60	[ "match"; e = SELF; "with"; b = branches -> mk loc (Match (e,b))
61	\| "map"; e = SELF; "with"; b = branches -> mk loc (Map (e,b))
62	\| "transform"; e = SELF; "with"; b = branches ->
63	mk noloc (Op ("flatten", [mk loc (Map (e,b))]))
64	\| "fun"; (f,a,b) = fun_decl ->
65	mk loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
66	\| (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
67	mk loc (Match (e1,[p,e2]))
68	\| e = expr; ":"; p = pat ->
69	mk loc (Forget (e,p))
70	]
71
72
73	\|
74	[ e1 = expr; op = ["+" \| "-" \| "@"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
75	]
76	\|
77	[ e1 = expr; op = ["*" \| "/"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
78	]
79	\|
80	[ e = expr; "."; l = [LIDENT \| UIDENT] -> mk loc (Dot (e,Types.label l))
81	]
82
83	\|
84	[ LIDENT "flatten"; e = expr -> mk loc (Op ("flatten",[e]))
85	\| LIDENT "load_xml"; e = expr -> mk loc (Op ("load_xml",[e]))
86	\| e1 = expr; e2 = expr -> mk loc (Apply (e1,e2))
87	]
88
89	\| "no_appl"
90	[ c = const -> mk loc (Cst c)
91	\| "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
92	\| "["; l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
93	let e = match e with Some e -> e \| None -> cst_nil in
94	let l = List.flatten l in
95	tuple loc (l @ [e])
96	\| "<"; t = expr_tag_spec; a = expr_attrib_spec; ">"; c = expr ->
97	tuple loc [t;a;c]
98	\| "{"; r = [ expr_record_spec \| -> mk loc (RecordLitt []) ]; "}" -> r
99	\| s = STRING ->
100	tuple loc (char_list loc s @ [cst_nil])
101	\| "!"; t = pat -> mk loc (DebugTyper t)
102	\| a = LIDENT -> mk loc (Var a)
103	]
104
105	];
106
107	seq_elem: [
108	[ x = [CHAR \| STRING] -> char_list loc x
109	\| e = expr LEVEL "no_appl" -> [e]
110	]
111	];
112
113	let_binding: [
114	[ "let"; p = pat; "="; e = expr -> (p,e)
115	\| "let"; p = pat; ":"; t = pat; "="; e = expr -> (p, mk noloc (Forget (e,t)))
116	\| "let"; "fun"; (f,a,b) = fun_decl ->
117	let p = match f with
118	\| Some x -> mk loc (Capture x)
119	\| _ -> failwith "Function name mandatory in let fun declarations"
120	in
121	let abst = { fun_name = f; fun_iface = a; fun_body = b } in
122	let e = mk loc (Abstraction abst) in
123	(p,e);
124	]
125	];
126
127	fun_decl: [
128	[ f = OPT LIDENT; "("; a = LIST0 arrow SEP ";"; ")"; b = branches ->
129	(f,a,b)
130	\| f = OPT LIDENT; "("; arg = LIDENT; ":"; targ = pat; ")"; ":"; tres = pat ;
131	"="; body = expr ->
132	let b = [mk noloc (Capture arg), body] in
133	let a = [targ,tres] in
134	(f,a,b)
135	]
136	];
137
138	arrow: [
139	[ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
140	];
141
142	branches: [
143	[ OPT "\|"; l = LIST1 branch SEP "\|" ; OPT "end" -> l ]
144	];
145
146	branch: [
147	[ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
148	];
149
150
151	regexp: [
152	[ x = regexp; "\|"; y = regexp -> Alt (x,y) ]
153	\| [ x = regexp; y = regexp -> Seq (x,y) ]
154	\| [ a = LIDENT; "::"; x = regexp -> SeqCapture (a,x) ]
155	\| [ x = regexp; "*" -> Star x
156	\| x = regexp; "*?" -> WeakStar x
157	\| x = regexp; "+" -> Seq (x, Star x)
158	\| x = regexp; "+?" -> Seq (x, WeakStar x)
159	\| x = regexp; "?" -> Alt (x, Epsilon)
160	\| x = regexp; "??" -> Alt (Epsilon, x) ]
161	\| [ "("; x = regexp; ")" -> x
162	\| UIDENT "String" -> string
163	\| s = CHAR ->
164	let s = seq_of_string (Token.eval_string s) in
165	List.fold_right
166	(fun c accu ->
167	let c = Chars.Unichar.from_char c in
168	let c = Chars.atom c in
169	Seq (Elem (mk loc (Internal (Types.char c))), accu))
170	s
171	Epsilon
172	\| e = pat LEVEL "simple" -> Elem e
173	]
174	];
175
176	pat: [
177	[ x = pat; "where";
178	b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)] SEP "and"
179	-> mk loc (Recurs (x,b)) ]
180	\| RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
181	\| "no_arrow" [ x = pat; "\|"; y = pat -> mk loc (Or (x,y)) ]
182	\| "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y,true))
183	(* \| x = pat; ":"; y = pat -> mk loc (And (x,y,false)) *)
184	\| x = pat; "-"; y = pat -> mk loc (Diff (x,y)) ]
185	\|
186	[ "{"; r = record_spec; "}" -> r
187	\| LIDENT "_" -> mk loc (Internal Types.any)
188	\| a = LIDENT -> mk loc (Capture a)
189	\| "("; a = LIDENT; ":="; c = const; ")" -> mk loc (Constant (a,c))
190	\| a = UIDENT -> mk loc (PatVar a)
191	\| i = INT ; "--"; j = INT ->
192	let i = Big_int.big_int_of_string i
193	and j = Big_int.big_int_of_string j in
194	mk loc (Internal (Types.interval (Intervals.bounded i j)))
195	\| i = INT ->
196	let i = Big_int.big_int_of_string i in
197	mk loc (Internal (Types.interval (Intervals.atom i)))
198	\| "*--"; j = INT ->
199	let j = Big_int.big_int_of_string j in
200	mk loc (Internal (Types.interval (Intervals.left j)))
201	\| i = INT; "--*" ->
202	let i = Big_int.big_int_of_string i in
203	mk loc (Internal (Types.interval (Intervals.right i)))
204	\| i = char ->
205	mk loc (Internal (Types.char (Chars.char_class i i)))
206	\| i = char ; "--"; j = char ->
207	mk loc (Internal (Types.char (Chars.char_class i j)))
208	\| c = const -> mk loc (Internal (Types.constant c))
209	\| "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
210	\| "["; r = [ r = regexp -> r \| -> Epsilon ];
211	q = [ ";"; q = pat -> q
212	\| -> mk noloc (Internal (Sequence.nil_type)) ];
213	"]" -> mk loc (Regexp (r,q))
214	\| "<"; t = tag_spec; a = attrib_spec; ">"; c = pat ->
215	multi_prod loc [t;a;c]
216	]
217
218	];
219
220	record_spec:
221	[ [ r = LIST0 [ l = [LIDENT \| UIDENT];
222	o = ["=?" -> true \| "=" -> false];
223	x = pat ->
224	mk loc (Record (Types.label l,o,x))
225	] SEP ";" ->
226	match r with
227	\| [] -> mk loc (Internal Types.Record.any)
228	\| h::t -> List.fold_left (fun t1 t2 -> mk loc (And (t1,t2,true))) h t
229	] ];
230
231	char:
232	[
233	[ c = CHAR -> Chars.Unichar.from_char (Token.eval_char c)
234	\| "!"; i = INT -> Chars.Unichar.from_int (int_of_string i) ]
235	];
236
237
238	const:
239	[
240	[ i = INT -> Types.Integer (Big_int.big_int_of_string i)
241	\| "`"; a = [LIDENT \| UIDENT] -> Types.Atom (Types.mk_atom a)
242	\| c = char -> Types.Char c ]
243	];
244
245	tag_spec:
246	[
247	[ LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any))) ]
248	\| [ a = [LIDENT \| UIDENT] ->
249	mk loc (Internal (Types.atom (Atoms.atom (Types.mk_atom a)))) ]
250	\| [ t = pat -> t ]
251	];
252
253	attrib_spec:
254	[ [ r = record_spec -> r \| "("; t = pat; ")" -> t ] ];
255
256	expr_record_spec:
257	[ [ r = LIST1
258	[ l = [LIDENT \| UIDENT]; "="; x = expr -> (Types.label l,x) ]
259	SEP ";" ->
260	mk loc (RecordLitt r)
261	] ];
262
263	expr_tag_spec:
264	[
265	[ a = [LIDENT \| UIDENT] ->
266	mk loc (Cst (Types.Atom (Types.mk_atom a))) ]
267	\| [ e = expr LEVEL "no_appl" -> e ]
268	];
269
270	expr_attrib_spec:
271	[ [ r = expr_record_spec -> r ]
272	\| [ e = expr LEVEL "no_appl" -> e
273	\| -> mk loc (RecordLitt [])
274	]
275	];
276	END
277
278	let pat' = Grammar.Entry.create gram "type/pattern expression"
279	EXTEND GLOBAL: pat pat';
280	pat': [ [ p = pat; EOI -> p ] ];
281	END
282
283	let pat = Grammar.Entry.parse pat
284	let expr = Grammar.Entry.parse expr
285	let prog = Grammar.Entry.parse prog
286
287	module From_string = struct
288	let pat s = Grammar.Entry.parse pat' (Stream.of_string s)
289	let expr s = expr (Stream.of_string s)
290	end
291