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]))
    ]

    
    | 
    [ 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]))
    | 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"; "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)) 
      | 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))) 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.parse pat
let expr = Grammar.Entry.parse expr
let prog = Grammar.Entry.parse prog

module From_string = struct
  let pat s = pat (Stream.of_string s)
  let expr s = expr (Stream.of_string s)
end

1	abate	4	open Location
2			open Ast
3
4	abate	38	(* 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	abate	4
31	abate	18
32	abate	38	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	abate	18
36
37	abate	38	EXTEND
38	abate	10	GLOBAL: prog expr pat regexp const;
39	abate	4
40	abate	10	prog: [
41	abate	14	[ l = LIST0 [ p = phrase; ";;" -> mk loc p ]; EOI -> l ]
42	abate	10	];
43
44	abate	13	phrase: [
45			[ e = expr -> EvalStatement e
46	abate	43	\| "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t)
47			\| "debug"; d = debug_directive -> Debug d
48			]
49	abate	13	];
50
51	abate	43	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	abate	4	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	abate	17	\| "transform"; e = SELF; "with"; b = branches ->
63			mk noloc (Op ("flatten", [mk loc (Map (e,b))]))
64	abate	48	\| "fun"; (f,a,b) = fun_decl ->
65	abate	4	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			]
69
70
71	abate	15	\|
72	abate	51	[ e1 = expr; op = ["+" \| "-" \| "@"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
73			]
74	abate	16	\|
75	abate	51	[ e1 = expr; op = ["*" \| "/"]; e2 = expr -> mk loc (Op (op,[e1;e2]))
76			]
77	abate	26	\|
78	abate	51	[ e = expr; "."; l = [LIDENT \| UIDENT] -> mk loc (Dot (e,Types.label l))
79			]
80	abate	26
81	abate	52	\|
82			[ LIDENT "flatten"; e = expr -> mk loc (Op ("flatten",[e]))
83			\| e1 = expr; e2 = expr -> mk loc (Apply (e1,e2))
84			]
85
86	abate	4	\| "no_appl"
87			[ c = const -> mk loc (Cst c)
88			\| "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
89	abate	18	\| "["; l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
90			let e = match e with Some e -> e \| None -> cst_nil in
91			let l = List.flatten l in
92	abate	4	tuple loc (l @ [e])
93			\| "<"; t = expr_tag_spec; a = expr_attrib_spec; ">"; c = expr ->
94			tuple loc [t;a;c]
95			\| "{"; r = [ expr_record_spec \| -> mk loc (RecordLitt []) ]; "}" -> r
96	abate	38	\| s = STRING ->
97			tuple loc (char_list loc s @ [cst_nil])
98	abate	18	\| "!"; t = pat -> mk loc (DebugTyper t)
99	abate	4	\| a = LIDENT -> mk loc (Var a)
100			]
101
102			];
103	abate	18
104			seq_elem: [
105	abate	38	[ x = [CHAR \| STRING] -> char_list loc x
106	abate	18	\| e = expr LEVEL "no_appl" -> [e]
107			]
108			];
109	abate	4
110			let_binding: [
111			[ "let"; p = pat; "="; e = expr -> (p,e)
112	abate	48	\| "let"; "fun"; (f,a,b) = fun_decl ->
113			let p = match f with
114			\| Some x -> mk loc (Capture x)
115			\| _ -> failwith "Function name mandatory in let fun declarations"
116			in
117			let abst = { fun_name = f; fun_iface = a; fun_body = b } in
118	abate	4	let e = mk loc (Abstraction abst) in
119	abate	48	(p,e);
120	abate	4	]
121			];
122
123	abate	48	fun_decl: [
124			[ f = OPT LIDENT; "("; a = LIST0 arrow SEP ";"; ")"; b = branches ->
125			(f,a,b)
126			\| f = OPT LIDENT; "("; arg = LIDENT; ":"; targ = pat; ")"; ":"; tres = pat ;
127			"="; body = expr ->
128			let b = [mk noloc (Capture arg), body] in
129			let a = [targ,tres] in
130			(f,a,b)
131			]
132			];
133
134	abate	4	arrow: [
135	abate	9	[ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
136	abate	4	];
137
138			branches: [
139			[ OPT "\|"; l = LIST1 branch SEP "\|" ; OPT "end" -> l ]
140			];
141
142			branch: [
143	abate	6	[ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
144	abate	4	];
145
146
147			regexp: [
148			[ x = regexp; "\|"; y = regexp -> Alt (x,y) ]
149			\| [ x = regexp; y = regexp -> Seq (x,y) ]
150			\| [ a = LIDENT; "::"; x = regexp -> SeqCapture (a,x) ]
151			\| [ x = regexp; "*" -> Star x
152			\| x = regexp; "*?" -> WeakStar x
153			\| x = regexp; "+" -> Seq (x, Star x)
154			\| x = regexp; "+?" -> Seq (x, WeakStar x)
155			\| x = regexp; "?" -> Alt (x, Epsilon)
156			\| x = regexp; "??" -> Alt (Epsilon, x) ]
157			\| [ "("; x = regexp; ")" -> x
158	abate	18	\| UIDENT "String" -> string
159	abate	38	\| s = CHAR ->
160	abate	18	let s = seq_of_string (Token.eval_string s) in
161			List.fold_right
162			(fun c accu ->
163			let c = Chars.Unichar.from_char c in
164			let c = Chars.atom c in
165			Seq (Elem (mk loc (Internal (Types.char c))), accu))
166			s
167			Epsilon
168	abate	4	\| e = pat LEVEL "simple" -> Elem e
169			]
170			];
171
172			pat: [
173			[ x = pat; "where";
174			b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)] SEP "and"
175			-> mk loc (Recurs (x,b)) ]
176			\| RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
177	abate	6	\| "no_arrow" [ x = pat; "\|"; y = pat -> mk loc (Or (x,y)) ]
178	abate	4	\| "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y))
179			\| x = pat; "-"; y = pat -> mk loc (Diff (x,y)) ]
180			\|
181			[ "{"; r = record_spec; "}" -> r
182			\| LIDENT "_" -> mk loc (Internal Types.any)
183			\| a = LIDENT -> mk loc (Capture a)
184			\| "("; a = LIDENT; ":="; c = const; ")" -> mk loc (Constant (a,c))
185			\| a = UIDENT -> mk loc (PatVar a)
186			\| i = INT ; "--"; j = INT ->
187	abate	15	let i = Big_int.big_int_of_string i
188			and j = Big_int.big_int_of_string j in
189	abate	52	mk loc (Internal (Types.interval (Intervals.bounded i j)))
190			\| i = INT ->
191			let i = Big_int.big_int_of_string i in
192			mk loc (Internal (Types.interval (Intervals.atom i)))
193			\| "*--"; j = INT ->
194			let j = Big_int.big_int_of_string j in
195			mk loc (Internal (Types.interval (Intervals.left j)))
196			\| i = INT; "--*" ->
197			let i = Big_int.big_int_of_string i in
198			mk loc (Internal (Types.interval (Intervals.right i)))
199	abate	19	\| i = char ->
200			mk loc (Internal (Types.char (Chars.char_class i i)))
201	abate	13	\| i = char ; "--"; j = char ->
202	abate	18	mk loc (Internal (Types.char (Chars.char_class i j)))
203	abate	4	\| c = const -> mk loc (Internal (Types.constant c))
204			\| "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
205			\| "["; r = [ r = regexp -> r \| -> Epsilon ];
206			q = [ ";"; q = pat -> q
207	abate	18	\| -> mk noloc (Internal (Sequence.nil_type)) ];
208	abate	4	"]" -> mk loc (Regexp (r,q))
209			\| "<"; t = tag_spec; a = attrib_spec; ">"; c = pat ->
210			multi_prod loc [t;a;c]
211			]
212
213			];
214
215			record_spec:
216			[ [ r = LIST0 [ l = [LIDENT \| UIDENT];
217			o = ["=?" -> true \| "=" -> false];
218			x = pat ->
219			mk loc (Record (Types.label l,o,x))
220			] SEP ";" ->
221			match r with
222	abate	9	\| [] -> mk loc (Internal Types.Record.any)
223			\| h::t -> List.fold_left (fun t1 t2 -> mk loc (And (t1,t2))) h t
224	abate	4	] ];
225
226	abate	13	char:
227			[
228			[ c = CHAR -> Chars.Unichar.from_char (Token.eval_char c)
229			\| "!"; i = INT -> Chars.Unichar.from_int (int_of_string i) ]
230			];
231
232
233	abate	4	const:
234			[
235	abate	15	[ i = INT -> Types.Integer (Big_int.big_int_of_string i)
236	abate	18	\| "`"; a = [LIDENT \| UIDENT] -> Types.Atom (Types.mk_atom a)
237	abate	13	\| c = char -> Types.Char c ]
238	abate	4	];
239
240			tag_spec:
241			[
242	abate	18	[ LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any))) ]
243			\| [ a = [LIDENT \| UIDENT] ->
244			mk loc (Internal (Types.atom (Atoms.atom (Types.mk_atom a)))) ]
245	abate	4	\| [ t = pat -> t ]
246			];
247
248			attrib_spec:
249			[ [ r = record_spec -> r \| "("; t = pat; ")" -> t ] ];
250
251			expr_record_spec:
252			[ [ r = LIST1
253			[ l = [LIDENT \| UIDENT]; "="; x = expr -> (Types.label l,x) ]
254			SEP ";" ->
255			mk loc (RecordLitt r)
256			] ];
257
258			expr_tag_spec:
259			[
260			[ a = [LIDENT \| UIDENT] ->
261			mk loc (Cst (Types.Atom (Types.mk_atom a))) ]
262			\| [ e = expr LEVEL "no_appl" -> e ]
263			];
264
265			expr_attrib_spec:
266			[ [ r = expr_record_spec -> r ]
267			\| [ e = expr LEVEL "no_appl" -> e
268			\| -> mk loc (RecordLitt [])
269			]
270			];
271			END
272
273	abate	10	let pat = Grammar.Entry.parse pat
274			let expr = Grammar.Entry.parse expr
275			let prog = Grammar.Entry.parse prog
276
277			module From_string = struct
278			let pat s = pat (Stream.of_string s)
279			let expr s = expr (Stream.of_string s)
280	abate	4	end
281