blanchet@38293
|
1 |
(* Title: HOL/Tools/ATP/atp_problem.ML
|
blanchet@38261
|
2 |
Author: Jia Meng, Cambridge University Computer Laboratory and NICTA
|
blanchet@37509
|
3 |
Author: Jasmin Blanchette, TU Muenchen
|
blanchet@37509
|
4 |
|
blanchet@39692
|
5 |
Abstract representation of ATP problems and TPTP syntax.
|
blanchet@37509
|
6 |
*)
|
blanchet@37509
|
7 |
|
blanchet@38253
|
8 |
signature ATP_PROBLEM =
|
blanchet@37509
|
9 |
sig
|
blanchet@38226
|
10 |
datatype 'a fo_term = ATerm of 'a * 'a fo_term list
|
blanchet@38226
|
11 |
datatype quantifier = AForall | AExists
|
blanchet@38226
|
12 |
datatype connective = ANot | AAnd | AOr | AImplies | AIf | AIff | ANotIff
|
blanchet@43402
|
13 |
datatype ('a, 'b, 'c) formula =
|
blanchet@43402
|
14 |
AQuant of quantifier * ('a * 'b option) list * ('a, 'b, 'c) formula |
|
blanchet@43402
|
15 |
AConn of connective * ('a, 'b, 'c) formula list |
|
blanchet@43402
|
16 |
AAtom of 'c
|
blanchet@38228
|
17 |
|
blanchet@43778
|
18 |
datatype format = UEQ | FOF | TFF
|
blanchet@43396
|
19 |
datatype formula_kind = Axiom | Definition | Lemma | Hypothesis | Conjecture
|
blanchet@43398
|
20 |
datatype 'a problem_line =
|
blanchet@43414
|
21 |
Decl of string * 'a * 'a list * 'a |
|
blanchet@43448
|
22 |
Formula of string * formula_kind * ('a, 'a, 'a fo_term) formula
|
blanchet@43400
|
23 |
* string fo_term option * string fo_term option
|
blanchet@38251
|
24 |
type 'a problem = (string * 'a problem_line list) list
|
blanchet@38226
|
25 |
|
blanchet@43587
|
26 |
(* official TPTP syntax *)
|
blanchet@43617
|
27 |
val tptp_special_prefix : string
|
blanchet@43617
|
28 |
val tptp_false : string
|
blanchet@43617
|
29 |
val tptp_true : string
|
blanchet@43587
|
30 |
val tptp_tff_type_of_types : string
|
blanchet@43587
|
31 |
val tptp_tff_bool_type : string
|
blanchet@43587
|
32 |
val tptp_tff_individual_type : string
|
blanchet@38253
|
33 |
val timestamp : unit -> string
|
blanchet@43438
|
34 |
val hashw : word * word -> word
|
blanchet@43438
|
35 |
val hashw_string : string * word -> word
|
blanchet@39692
|
36 |
val is_atp_variable : string -> bool
|
blanchet@43580
|
37 |
val tptp_strings_for_atp_problem : format -> string problem -> string list
|
blanchet@39692
|
38 |
val nice_atp_problem :
|
blanchet@38251
|
39 |
bool -> ('a * (string * string) problem_line list) list
|
blanchet@38251
|
40 |
-> ('a * string problem_line list) list
|
blanchet@38251
|
41 |
* (string Symtab.table * string Symtab.table) option
|
blanchet@37509
|
42 |
end;
|
blanchet@37509
|
43 |
|
blanchet@38253
|
44 |
structure ATP_Problem : ATP_PROBLEM =
|
blanchet@37509
|
45 |
struct
|
blanchet@37509
|
46 |
|
blanchet@37643
|
47 |
(** ATP problem **)
|
blanchet@37643
|
48 |
|
blanchet@38198
|
49 |
datatype 'a fo_term = ATerm of 'a * 'a fo_term list
|
blanchet@38198
|
50 |
datatype quantifier = AForall | AExists
|
blanchet@38226
|
51 |
datatype connective = ANot | AAnd | AOr | AImplies | AIf | AIff | ANotIff
|
blanchet@43402
|
52 |
datatype ('a, 'b, 'c) formula =
|
blanchet@43402
|
53 |
AQuant of quantifier * ('a * 'b option) list * ('a, 'b, 'c) formula |
|
blanchet@43402
|
54 |
AConn of connective * ('a, 'b, 'c) formula list |
|
blanchet@43402
|
55 |
AAtom of 'c
|
blanchet@38198
|
56 |
|
blanchet@43778
|
57 |
datatype format = UEQ | FOF | TFF
|
blanchet@43396
|
58 |
datatype formula_kind = Axiom | Definition | Lemma | Hypothesis | Conjecture
|
blanchet@43398
|
59 |
datatype 'a problem_line =
|
blanchet@43414
|
60 |
Decl of string * 'a * 'a list * 'a |
|
blanchet@43448
|
61 |
Formula of string * formula_kind * ('a, 'a, 'a fo_term) formula
|
blanchet@43400
|
62 |
* string fo_term option * string fo_term option
|
blanchet@37643
|
63 |
type 'a problem = (string * 'a problem_line list) list
|
blanchet@37643
|
64 |
|
blanchet@43587
|
65 |
(* official TPTP syntax *)
|
blanchet@43617
|
66 |
val tptp_special_prefix = "$"
|
blanchet@43617
|
67 |
val tptp_false = "$false"
|
blanchet@43617
|
68 |
val tptp_true = "$true"
|
blanchet@43587
|
69 |
val tptp_tff_type_of_types = "$tType"
|
blanchet@43587
|
70 |
val tptp_tff_bool_type = "$o"
|
blanchet@43587
|
71 |
val tptp_tff_individual_type = "$i"
|
blanchet@43587
|
72 |
|
blanchet@38253
|
73 |
val timestamp = Date.fmt "%Y-%m-%d %H:%M:%S" o Date.fromTimeLocal o Time.now
|
blanchet@38253
|
74 |
|
blanchet@43438
|
75 |
(* This hash function is recommended in Compilers: Principles, Techniques, and
|
blanchet@43438
|
76 |
Tools, by Aho, Sethi, and Ullman. The "hashpjw" function, which they
|
blanchet@43438
|
77 |
particularly recommend, triggers a bug in versions of Poly/ML up to 4.2.0. *)
|
blanchet@43438
|
78 |
fun hashw (u, w) = Word.+ (u, Word.* (0w65599, w))
|
blanchet@43438
|
79 |
fun hashw_char (c, w) = hashw (Word.fromInt (Char.ord c), w)
|
blanchet@43438
|
80 |
fun hashw_string (s : string, w) = CharVector.foldl hashw_char w s
|
blanchet@43438
|
81 |
|
blanchet@38854
|
82 |
fun string_for_kind Axiom = "axiom"
|
blanchet@42640
|
83 |
| string_for_kind Definition = "definition"
|
blanchet@42640
|
84 |
| string_for_kind Lemma = "lemma"
|
blanchet@38854
|
85 |
| string_for_kind Hypothesis = "hypothesis"
|
blanchet@38854
|
86 |
| string_for_kind Conjecture = "conjecture"
|
blanchet@38854
|
87 |
|
blanchet@38198
|
88 |
fun string_for_term (ATerm (s, [])) = s
|
blanchet@38334
|
89 |
| string_for_term (ATerm ("equal", ts)) =
|
blanchet@38334
|
90 |
space_implode " = " (map string_for_term ts)
|
blanchet@42640
|
91 |
| string_for_term (ATerm ("[]", ts)) =
|
blanchet@42640
|
92 |
(* used for lists in the optional "source" field of a derivation *)
|
blanchet@42640
|
93 |
"[" ^ commas (map string_for_term ts) ^ "]"
|
blanchet@38198
|
94 |
| string_for_term (ATerm (s, ts)) =
|
blanchet@38334
|
95 |
s ^ "(" ^ commas (map string_for_term ts) ^ ")"
|
blanchet@38198
|
96 |
fun string_for_quantifier AForall = "!"
|
blanchet@38198
|
97 |
| string_for_quantifier AExists = "?"
|
blanchet@38198
|
98 |
fun string_for_connective ANot = "~"
|
blanchet@38198
|
99 |
| string_for_connective AAnd = "&"
|
blanchet@38198
|
100 |
| string_for_connective AOr = "|"
|
blanchet@38198
|
101 |
| string_for_connective AImplies = "=>"
|
blanchet@38226
|
102 |
| string_for_connective AIf = "<="
|
blanchet@38198
|
103 |
| string_for_connective AIff = "<=>"
|
blanchet@38226
|
104 |
| string_for_connective ANotIff = "<~>"
|
blanchet@43778
|
105 |
fun string_for_bound_var TFF (s, ty) =
|
blanchet@43587
|
106 |
s ^ " : " ^ (ty |> the_default tptp_tff_individual_type)
|
blanchet@43778
|
107 |
| string_for_bound_var _ (s, _) = s
|
blanchet@43587
|
108 |
fun string_for_formula format (AQuant (q, xs, phi)) =
|
blanchet@43397
|
109 |
"(" ^ string_for_quantifier q ^
|
blanchet@43587
|
110 |
"[" ^ commas (map (string_for_bound_var format) xs) ^ "] : " ^
|
blanchet@43587
|
111 |
string_for_formula format phi ^ ")"
|
blanchet@43587
|
112 |
| string_for_formula _ (AConn (ANot, [AAtom (ATerm ("equal", ts))])) =
|
blanchet@38239
|
113 |
space_implode " != " (map string_for_term ts)
|
blanchet@43587
|
114 |
| string_for_formula format (AConn (c, [phi])) =
|
blanchet@43587
|
115 |
"(" ^ string_for_connective c ^ " " ^ string_for_formula format phi ^ ")"
|
blanchet@43587
|
116 |
| string_for_formula format (AConn (c, phis)) =
|
blanchet@38198
|
117 |
"(" ^ space_implode (" " ^ string_for_connective c ^ " ")
|
blanchet@43587
|
118 |
(map (string_for_formula format) phis) ^ ")"
|
blanchet@43587
|
119 |
| string_for_formula _ (AAtom tm) = string_for_term tm
|
blanchet@38198
|
120 |
|
blanchet@43399
|
121 |
fun string_for_symbol_type [] res_ty = res_ty
|
blanchet@43399
|
122 |
| string_for_symbol_type [arg_ty] res_ty = arg_ty ^ " > " ^ res_ty
|
blanchet@43399
|
123 |
| string_for_symbol_type arg_tys res_ty =
|
blanchet@43399
|
124 |
string_for_symbol_type ["(" ^ space_implode " * " arg_tys ^ ")"] res_ty
|
blanchet@43399
|
125 |
|
blanchet@43510
|
126 |
val default_source =
|
blanchet@43510
|
127 |
ATerm ("inference", ATerm ("isabelle", []) :: replicate 2 (ATerm ("[]", [])))
|
blanchet@43510
|
128 |
|
blanchet@43580
|
129 |
fun string_for_problem_line _ (Decl (ident, sym, arg_tys, res_ty)) =
|
blanchet@43404
|
130 |
"tff(" ^ ident ^ ", type,\n " ^ sym ^ " : " ^
|
blanchet@43399
|
131 |
string_for_symbol_type arg_tys res_ty ^ ").\n"
|
blanchet@43580
|
132 |
| string_for_problem_line format
|
blanchet@43448
|
133 |
(Formula (ident, kind, phi, source, useful_info)) =
|
blanchet@43778
|
134 |
(case format of UEQ => "cnf" | FOF => "fof" | TFF => "tff") ^
|
blanchet@43580
|
135 |
"(" ^ ident ^ ", " ^ string_for_kind kind ^ ",\n (" ^
|
blanchet@43587
|
136 |
string_for_formula format phi ^ ")" ^
|
blanchet@43580
|
137 |
(case (source, useful_info) of
|
blanchet@43580
|
138 |
(NONE, NONE) => ""
|
blanchet@43580
|
139 |
| (SOME tm, NONE) => ", " ^ string_for_term tm
|
blanchet@43580
|
140 |
| (_, SOME tm) =>
|
blanchet@43580
|
141 |
", " ^ string_for_term (source |> the_default default_source) ^
|
blanchet@43580
|
142 |
", " ^ string_for_term tm) ^ ").\n"
|
blanchet@43580
|
143 |
fun tptp_strings_for_atp_problem format problem =
|
blanchet@37643
|
144 |
"% This file was generated by Isabelle (most likely Sledgehammer)\n\
|
blanchet@37643
|
145 |
\% " ^ timestamp () ^ "\n" ::
|
blanchet@37643
|
146 |
maps (fn (_, []) => []
|
blanchet@37643
|
147 |
| (heading, lines) =>
|
wenzelm@41739
|
148 |
"\n% " ^ heading ^ " (" ^ string_of_int (length lines) ^ ")\n" ::
|
blanchet@43580
|
149 |
map (string_for_problem_line format) lines)
|
blanchet@38854
|
150 |
problem
|
blanchet@37643
|
151 |
|
blanchet@39692
|
152 |
fun is_atp_variable s = Char.isUpper (String.sub (s, 0))
|
blanchet@38251
|
153 |
|
blanchet@37643
|
154 |
|
blanchet@37643
|
155 |
(** Nice names **)
|
blanchet@37643
|
156 |
|
blanchet@37624
|
157 |
fun empty_name_pool readable_names =
|
blanchet@37643
|
158 |
if readable_names then SOME (Symtab.empty, Symtab.empty) else NONE
|
blanchet@37624
|
159 |
|
blanchet@37624
|
160 |
fun pool_fold f xs z = pair z #> fold_rev (fn x => uncurry (f x)) xs
|
blanchet@37624
|
161 |
fun pool_map f xs =
|
blanchet@37624
|
162 |
pool_fold (fn x => fn ys => fn pool => f x pool |>> (fn y => y :: ys)) xs []
|
blanchet@37624
|
163 |
|
blanchet@43088
|
164 |
val no_qualifiers =
|
blanchet@43088
|
165 |
let
|
blanchet@43088
|
166 |
fun skip [] = []
|
blanchet@43088
|
167 |
| skip (#"." :: cs) = skip cs
|
blanchet@43088
|
168 |
| skip (c :: cs) = if Char.isAlphaNum c then skip cs else c :: keep cs
|
blanchet@43088
|
169 |
and keep [] = []
|
blanchet@43088
|
170 |
| keep (#"." :: cs) = skip cs
|
blanchet@43088
|
171 |
| keep (c :: cs) = c :: keep cs
|
blanchet@43088
|
172 |
in String.explode #> rev #> keep #> rev #> String.implode end
|
blanchet@43088
|
173 |
|
blanchet@43626
|
174 |
(* Long names can slow down the ATPs. *)
|
blanchet@43589
|
175 |
val max_readable_name_size = 20
|
blanchet@43438
|
176 |
|
blanchet@39355
|
177 |
(* "op" is also reserved, to avoid the unreadable "op_1", "op_2", etc., in the
|
blanchet@39355
|
178 |
problem files. "equal" is reserved by some ATPs. "eq" is reserved to ensure
|
blanchet@39355
|
179 |
that "HOL.eq" is correctly mapped to equality. *)
|
blanchet@39355
|
180 |
val reserved_nice_names = ["op", "equal", "eq"]
|
blanchet@37624
|
181 |
fun readable_name full_name s =
|
blanchet@37643
|
182 |
if s = full_name then
|
blanchet@37643
|
183 |
s
|
blanchet@37643
|
184 |
else
|
blanchet@43438
|
185 |
s |> no_qualifiers
|
blanchet@43438
|
186 |
|> Name.desymbolize (Char.isUpper (String.sub (full_name, 0)))
|
blanchet@43454
|
187 |
(* SNARK doesn't like sort (type) names that end with digits. We make
|
blanchet@43454
|
188 |
an effort to avoid this here. *)
|
blanchet@43454
|
189 |
|> (fn s => if Char.isDigit (String.sub (s, size s - 1)) then s ^ "_"
|
blanchet@43454
|
190 |
else s)
|
blanchet@43454
|
191 |
|> (fn s =>
|
blanchet@43530
|
192 |
if size s > max_readable_name_size then
|
blanchet@43530
|
193 |
String.substring (s, 0, max_readable_name_size div 2 - 4) ^
|
blanchet@43454
|
194 |
Word.toString (hashw_string (full_name, 0w0)) ^
|
blanchet@43530
|
195 |
String.extract (s, size s - max_readable_name_size div 2 + 4,
|
blanchet@43530
|
196 |
NONE)
|
blanchet@43454
|
197 |
else
|
blanchet@43454
|
198 |
s)
|
blanchet@43438
|
199 |
|> (fn s => if member (op =) reserved_nice_names s then full_name else s)
|
blanchet@37624
|
200 |
|
blanchet@37624
|
201 |
fun nice_name (full_name, _) NONE = (full_name, NONE)
|
blanchet@37624
|
202 |
| nice_name (full_name, desired_name) (SOME the_pool) =
|
blanchet@39630
|
203 |
if String.isPrefix "$" full_name then
|
blanchet@39630
|
204 |
(full_name, SOME the_pool)
|
blanchet@39630
|
205 |
else case Symtab.lookup (fst the_pool) full_name of
|
blanchet@37624
|
206 |
SOME nice_name => (nice_name, SOME the_pool)
|
blanchet@37624
|
207 |
| NONE =>
|
blanchet@37624
|
208 |
let
|
blanchet@37624
|
209 |
val nice_prefix = readable_name full_name desired_name
|
blanchet@37624
|
210 |
fun add j =
|
blanchet@37624
|
211 |
let
|
blanchet@43515
|
212 |
(* The trailing "_" is for SNARK (cf. comment above). *)
|
blanchet@43515
|
213 |
val nice_name =
|
blanchet@43515
|
214 |
nice_prefix ^ (if j = 0 then "" else "_" ^ string_of_int j ^ "_")
|
blanchet@37624
|
215 |
in
|
blanchet@37624
|
216 |
case Symtab.lookup (snd the_pool) nice_name of
|
blanchet@37624
|
217 |
SOME full_name' =>
|
blanchet@37624
|
218 |
if full_name = full_name' then (nice_name, the_pool)
|
blanchet@37624
|
219 |
else add (j + 1)
|
blanchet@37624
|
220 |
| NONE =>
|
blanchet@37624
|
221 |
(nice_name,
|
blanchet@37624
|
222 |
(Symtab.update_new (full_name, nice_name) (fst the_pool),
|
blanchet@37624
|
223 |
Symtab.update_new (nice_name, full_name) (snd the_pool)))
|
blanchet@37624
|
224 |
end
|
blanchet@37624
|
225 |
in add 0 |> apsnd SOME end
|
blanchet@37624
|
226 |
|
blanchet@38198
|
227 |
fun nice_term (ATerm (name, ts)) =
|
blanchet@38198
|
228 |
nice_name name ##>> pool_map nice_term ts #>> ATerm
|
blanchet@38198
|
229 |
fun nice_formula (AQuant (q, xs, phi)) =
|
blanchet@43397
|
230 |
pool_map nice_name (map fst xs)
|
blanchet@43397
|
231 |
##>> pool_map (fn NONE => pair NONE
|
blanchet@43402
|
232 |
| SOME ty => nice_name ty #>> SOME) (map snd xs)
|
blanchet@43397
|
233 |
##>> nice_formula phi
|
blanchet@43397
|
234 |
#>> (fn ((ss, ts), phi) => AQuant (q, ss ~~ ts, phi))
|
blanchet@38198
|
235 |
| nice_formula (AConn (c, phis)) =
|
blanchet@38198
|
236 |
pool_map nice_formula phis #>> curry AConn c
|
blanchet@38280
|
237 |
| nice_formula (AAtom tm) = nice_term tm #>> AAtom
|
blanchet@43414
|
238 |
fun nice_problem_line (Decl (ident, sym, arg_tys, res_ty)) =
|
blanchet@43399
|
239 |
nice_name sym
|
blanchet@43399
|
240 |
##>> pool_map nice_name arg_tys
|
blanchet@43399
|
241 |
##>> nice_name res_ty
|
blanchet@43414
|
242 |
#>> (fn ((sym, arg_tys), res_ty) => Decl (ident, sym, arg_tys, res_ty))
|
blanchet@43448
|
243 |
| nice_problem_line (Formula (ident, kind, phi, source, useful_info)) =
|
blanchet@43400
|
244 |
nice_formula phi
|
blanchet@43448
|
245 |
#>> (fn phi => Formula (ident, kind, phi, source, useful_info))
|
blanchet@38171
|
246 |
fun nice_problem problem =
|
blanchet@37643
|
247 |
pool_map (fn (heading, lines) =>
|
blanchet@38171
|
248 |
pool_map nice_problem_line lines #>> pair heading) problem
|
blanchet@39692
|
249 |
fun nice_atp_problem readable_names problem =
|
blanchet@38251
|
250 |
nice_problem problem (empty_name_pool readable_names)
|
blanchet@37509
|
251 |
|
blanchet@37509
|
252 |
end;
|