viewcvs.cgi/parser/parser.ml

open Location
open Ast

  let gram    = Grammar.create (Plexer.make ())
  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)


  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) ]
  ];

  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 = OPT LIDENT; "("; a = LIST1 arrow SEP ";"; ")";
      b = branches -> 
        mk loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
    | "fun";  f = OPT LIDENT; 
      "("; arg = LIDENT; ":"; targ = pat; ")"; ":"; tres = pat ; 
      "="; body = expr ->
        let fun_body = (mk noloc (Capture arg), body) in
        mk loc (Abstraction { fun_name = f; fun_iface = [(targ,tres)];
                              fun_body = [fun_body] })
    | (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
        mk loc (Match (e1,[p,e2]))
    ]

    |
    [ LIDENT "flatten"; e = expr -> mk loc (Op ("flatten",[e]))
    | e1 = expr; e2 = expr -> mk loc (Apply (e1,e2))
    ]
    
    | 
    [ e1 = expr; "+"; e2 = expr -> mk loc (Op ("+",[e1;e2]))  
    | e1 = expr; "@"; e2 = expr -> mk loc (Op ("@",[e1;e2])) ]
    |
    [ e1 = expr; "*"; e2 = expr -> mk loc (Op ("*",[e1;e2]))  ]
    |
    [ e = expr;  "."; l = [LIDENT | UIDENT] -> mk loc (Dot (e,Types.label l)) ]  
 
    | "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
    | "!"; t = pat -> mk loc (DebugTyper t)
    | a = LIDENT -> mk loc (Var a)
    ]

  ];

  seq_elem: [
    [ x = STRING -> 
        let s = seq_of_string (Token.eval_string x) in
        List.map 
          (fun c -> mk loc (Cst (Types.Char (Chars.Unichar.from_char c))))
          s 
    | e = expr LEVEL "no_appl" -> [e]
    ]
  ];
          
  let_binding: [
    [ "let"; p = pat; "="; e = expr -> (p,e)
    | "let"; "fun"; f = LIDENT; "("; a = LIST0 arrow SEP ";"; ")";
      b = branches -> 
        let p = mk loc (Capture f) in
        let abst = { fun_name = Some f; fun_iface = a; fun_body = b } in
        let e = mk loc (Abstraction abst) in
        (p,e)
    ] 
  ];

  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 = STRING ->
        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 i j))
      | 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			let gram = Grammar.create (Plexer.make ())
5	abate	10	let prog = Grammar.Entry.create gram "prog"
6	abate	4	let expr = Grammar.Entry.create gram "expression"
7			let pat = Grammar.Entry.create gram "type/pattern expression"
8			let regexp = Grammar.Entry.create gram "type/pattern regexp"
9			let const = Grammar.Entry.create gram "scalar constant"
10
11			let rec multi_prod loc = function
12			\| [ x ] -> x
13			\| x :: l -> mk loc (Prod (x, multi_prod loc l))
14			\| [] -> assert false
15
16			let rec tuple loc = function
17			\| [ x ] -> x
18	abate	19	\| x :: l -> mk (x.loc) (Pair (x, tuple loc l))
19	abate	4	\| [] -> assert false
20
21	abate	18	let char = mk noloc (Internal (Types.char Chars.any))
22			let string = Star (Elem char)
23
24			let cst_nil = mk noloc (Cst (Types.Atom Sequence.nil_atom))
25
26			let seq_of_string s =
27			let rec aux accu i = if (i = 0) then accu else aux (s.[i-1]::accu) (i-1) in
28			aux [] (String.length s)
29
30
31	abate	4	EXTEND
32	abate	10	GLOBAL: prog expr pat regexp const;
33	abate	4
34	abate	10	prog: [
35	abate	14	[ l = LIST0 [ p = phrase; ";;" -> mk loc p ]; EOI -> l ]
36	abate	10	];
37
38	abate	13	phrase: [
39			[ e = expr -> EvalStatement e
40			\| "type"; x = UIDENT; "="; t = pat -> TypeDecl (x,t) ]
41			];
42
43	abate	4	expr: [
44			"top" RIGHTA
45			[ "match"; e = SELF; "with"; b = branches -> mk loc (Match (e,b))
46			\| "map"; e = SELF; "with"; b = branches -> mk loc (Map (e,b))
47	abate	17	\| "transform"; e = SELF; "with"; b = branches ->
48			mk noloc (Op ("flatten", [mk loc (Map (e,b))]))
49	abate	4	\| "fun"; f = OPT LIDENT; "("; a = LIST1 arrow SEP ";"; ")";
50			b = branches ->
51			mk loc (Abstraction { fun_name = f; fun_iface = a; fun_body = b })
52	abate	24	\| "fun"; f = OPT LIDENT;
53			"("; arg = LIDENT; ":"; targ = pat; ")"; ":"; tres = pat ;
54			"="; body = expr ->
55			let fun_body = (mk noloc (Capture arg), body) in
56			mk loc (Abstraction { fun_name = f; fun_iface = [(targ,tres)];
57			fun_body = [fun_body] })
58	abate	4	\| (p,e1) = let_binding; "in"; e2 = expr LEVEL "top"->
59			mk loc (Match (e1,[p,e2]))
60			]
61
62			\|
63	abate	17	[ LIDENT "flatten"; e = expr -> mk loc (Op ("flatten",[e]))
64			\| e1 = expr; e2 = expr -> mk loc (Apply (e1,e2))
65	abate	4	]
66
67	abate	15	\|
68	abate	17	[ e1 = expr; "+"; e2 = expr -> mk loc (Op ("+",[e1;e2]))
69			\| e1 = expr; "@"; e2 = expr -> mk loc (Op ("@",[e1;e2])) ]
70	abate	16	\|
71			[ e1 = expr; ""; e2 = expr -> mk loc (Op ("",[e1;e2])) ]
72	abate	26	\|
73			[ e = expr; "."; l = [LIDENT \| UIDENT] -> mk loc (Dot (e,Types.label l)) ]
74
75	abate	4	\| "no_appl"
76			[ c = const -> mk loc (Cst c)
77			\| "("; l = LIST1 expr SEP ","; ")" -> tuple loc l
78	abate	18	\| "["; l = LIST0 seq_elem; e = OPT [ ";"; e = expr -> e ]; "]" ->
79			let e = match e with Some e -> e \| None -> cst_nil in
80			let l = List.flatten l in
81	abate	4	tuple loc (l @ [e])
82			\| "<"; t = expr_tag_spec; a = expr_attrib_spec; ">"; c = expr ->
83			tuple loc [t;a;c]
84			\| "{"; r = [ expr_record_spec \| -> mk loc (RecordLitt []) ]; "}" -> r
85	abate	18	\| "!"; t = pat -> mk loc (DebugTyper t)
86	abate	4	\| a = LIDENT -> mk loc (Var a)
87			]
88
89			];
90	abate	18
91			seq_elem: [
92			[ x = STRING ->
93			let s = seq_of_string (Token.eval_string x) in
94			List.map
95			(fun c -> mk loc (Cst (Types.Char (Chars.Unichar.from_char c))))
96			s
97			\| e = expr LEVEL "no_appl" -> [e]
98			]
99			];
100	abate	4
101			let_binding: [
102			[ "let"; p = pat; "="; e = expr -> (p,e)
103			\| "let"; "fun"; f = LIDENT; "("; a = LIST0 arrow SEP ";"; ")";
104			b = branches ->
105			let p = mk loc (Capture f) in
106			let abst = { fun_name = Some f; fun_iface = a; fun_body = b } in
107			let e = mk loc (Abstraction abst) in
108			(p,e)
109			]
110			];
111
112			arrow: [
113	abate	9	[ t1 = pat LEVEL "no_arrow"; "->"; t2 = pat -> (t1,t2)]
114	abate	4	];
115
116			branches: [
117			[ OPT "\|"; l = LIST1 branch SEP "\|" ; OPT "end" -> l ]
118			];
119
120			branch: [
121	abate	6	[ p = pat LEVEL "no_arrow"; "->"; e = expr -> (p,e) ]
122	abate	4	];
123
124
125			regexp: [
126			[ x = regexp; "\|"; y = regexp -> Alt (x,y) ]
127			\| [ x = regexp; y = regexp -> Seq (x,y) ]
128			\| [ a = LIDENT; "::"; x = regexp -> SeqCapture (a,x) ]
129			\| [ x = regexp; "*" -> Star x
130			\| x = regexp; "*?" -> WeakStar x
131			\| x = regexp; "+" -> Seq (x, Star x)
132			\| x = regexp; "+?" -> Seq (x, WeakStar x)
133			\| x = regexp; "?" -> Alt (x, Epsilon)
134			\| x = regexp; "??" -> Alt (Epsilon, x) ]
135			\| [ "("; x = regexp; ")" -> x
136	abate	18	\| UIDENT "String" -> string
137			\| s = STRING ->
138			let s = seq_of_string (Token.eval_string s) in
139			List.fold_right
140			(fun c accu ->
141			let c = Chars.Unichar.from_char c in
142			let c = Chars.atom c in
143			Seq (Elem (mk loc (Internal (Types.char c))), accu))
144			s
145			Epsilon
146	abate	4	\| e = pat LEVEL "simple" -> Elem e
147			]
148			];
149
150			pat: [
151			[ x = pat; "where";
152			b = LIST1 [ a = UIDENT; "="; y = pat -> (a,y)] SEP "and"
153			-> mk loc (Recurs (x,b)) ]
154			\| RIGHTA [ x = pat; "->"; y = pat -> mk loc (Arrow (x,y)) ]
155	abate	6	\| "no_arrow" [ x = pat; "\|"; y = pat -> mk loc (Or (x,y)) ]
156	abate	4	\| "simple" [ x = pat; "&"; y = pat -> mk loc (And (x,y))
157			\| x = pat; "-"; y = pat -> mk loc (Diff (x,y)) ]
158			\|
159			[ "{"; r = record_spec; "}" -> r
160			\| LIDENT "_" -> mk loc (Internal Types.any)
161			\| a = LIDENT -> mk loc (Capture a)
162			\| "("; a = LIDENT; ":="; c = const; ")" -> mk loc (Constant (a,c))
163			\| a = UIDENT -> mk loc (PatVar a)
164			\| i = INT ; "--"; j = INT ->
165	abate	15	let i = Big_int.big_int_of_string i
166			and j = Big_int.big_int_of_string j in
167	abate	4	mk loc (Internal (Types.interval i j))
168	abate	19	\| i = char ->
169			mk loc (Internal (Types.char (Chars.char_class i i)))
170	abate	13	\| i = char ; "--"; j = char ->
171	abate	18	mk loc (Internal (Types.char (Chars.char_class i j)))
172	abate	4	\| c = const -> mk loc (Internal (Types.constant c))
173			\| "("; l = LIST1 pat SEP ","; ")" -> multi_prod loc l
174			\| "["; r = [ r = regexp -> r \| -> Epsilon ];
175			q = [ ";"; q = pat -> q
176	abate	18	\| -> mk noloc (Internal (Sequence.nil_type)) ];
177	abate	4	"]" -> mk loc (Regexp (r,q))
178			\| "<"; t = tag_spec; a = attrib_spec; ">"; c = pat ->
179			multi_prod loc [t;a;c]
180			]
181
182			];
183
184			record_spec:
185			[ [ r = LIST0 [ l = [LIDENT \| UIDENT];
186			o = ["=?" -> true \| "=" -> false];
187			x = pat ->
188			mk loc (Record (Types.label l,o,x))
189			] SEP ";" ->
190			match r with
191	abate	9	\| [] -> mk loc (Internal Types.Record.any)
192			\| h::t -> List.fold_left (fun t1 t2 -> mk loc (And (t1,t2))) h t
193	abate	4	] ];
194
195	abate	13	char:
196			[
197			[ c = CHAR -> Chars.Unichar.from_char (Token.eval_char c)
198			\| "!"; i = INT -> Chars.Unichar.from_int (int_of_string i) ]
199			];
200
201
202	abate	4	const:
203			[
204	abate	15	[ i = INT -> Types.Integer (Big_int.big_int_of_string i)
205	abate	18	\| "`"; a = [LIDENT \| UIDENT] -> Types.Atom (Types.mk_atom a)
206	abate	13	\| c = char -> Types.Char c ]
207	abate	4	];
208
209			tag_spec:
210			[
211	abate	18	[ LIDENT "_" -> mk loc (Internal (Types.atom (Atoms.any))) ]
212			\| [ a = [LIDENT \| UIDENT] ->
213			mk loc (Internal (Types.atom (Atoms.atom (Types.mk_atom a)))) ]
214	abate	4	\| [ t = pat -> t ]
215			];
216
217			attrib_spec:
218			[ [ r = record_spec -> r \| "("; t = pat; ")" -> t ] ];
219
220			expr_record_spec:
221			[ [ r = LIST1
222			[ l = [LIDENT \| UIDENT]; "="; x = expr -> (Types.label l,x) ]
223			SEP ";" ->
224			mk loc (RecordLitt r)
225			] ];
226
227			expr_tag_spec:
228			[
229			[ a = [LIDENT \| UIDENT] ->
230			mk loc (Cst (Types.Atom (Types.mk_atom a))) ]
231			\| [ e = expr LEVEL "no_appl" -> e ]
232			];
233
234			expr_attrib_spec:
235			[ [ r = expr_record_spec -> r ]
236			\| [ e = expr LEVEL "no_appl" -> e
237			\| -> mk loc (RecordLitt [])
238			]
239			];
240			END
241
242	abate	10	let pat = Grammar.Entry.parse pat
243			let expr = Grammar.Entry.parse expr
244			let prog = Grammar.Entry.parse prog
245
246			module From_string = struct
247			let pat s = pat (Stream.of_string s)
248			let expr s = expr (Stream.of_string s)
249	abate	4	end
250