src/HOL/Tools/Sledgehammer/sledgehammer_filter_mash.ML
author blanchet
Wed, 18 Jul 2012 08:44:04 +0200
changeset 49336 c552d7f1720b
parent 49335 891a24a48155
child 49337 8a8d71e34297
permissions -rw-r--r--
learn from minimized ATP proofs
blanchet@49263
     1
(*  Title:      HOL/Tools/Sledgehammer/sledgehammer_filter_mash.ML
blanchet@49263
     2
    Author:     Jasmin Blanchette, TU Muenchen
blanchet@49263
     3
blanchet@49263
     4
Sledgehammer's machine-learning-based relevance filter (MaSh).
blanchet@49263
     5
*)
blanchet@49263
     6
blanchet@49263
     7
signature SLEDGEHAMMER_FILTER_MASH =
blanchet@49263
     8
sig
blanchet@49266
     9
  type status = ATP_Problem_Generate.status
blanchet@49266
    10
  type stature = ATP_Problem_Generate.stature
blanchet@49311
    11
  type fact = Sledgehammer_Fact.fact
blanchet@49311
    12
  type fact_override = Sledgehammer_Fact.fact_override
blanchet@49266
    13
  type params = Sledgehammer_Provers.params
blanchet@49303
    14
  type relevance_fudge = Sledgehammer_Provers.relevance_fudge
blanchet@49266
    15
  type prover_result = Sledgehammer_Provers.prover_result
blanchet@49266
    16
blanchet@49323
    17
  val trace : bool Config.T
blanchet@49334
    18
  val MaShN : string
blanchet@49329
    19
  val meshN : string
blanchet@49329
    20
  val iterN : string
blanchet@49329
    21
  val mashN : string
blanchet@49329
    22
  val fact_filters : string list
blanchet@49318
    23
  val escape_meta : string -> string
blanchet@49318
    24
  val escape_metas : string list -> string
blanchet@49323
    25
  val unescape_meta : string -> string
blanchet@49323
    26
  val unescape_metas : string -> string list
blanchet@49326
    27
  val extract_query : string -> string * string list
blanchet@49336
    28
  val suggested_facts : string list -> ('a * thm) list -> ('a * thm) list
blanchet@49336
    29
  val mesh_facts :
blanchet@49336
    30
    int -> (('a * thm) list * ('a * thm) list) list -> ('a * thm) list
blanchet@49333
    31
  val is_likely_tautology : Proof.context -> string -> thm -> bool
blanchet@49330
    32
  val is_too_meta : thm -> bool
blanchet@49266
    33
  val theory_ord : theory * theory -> order
blanchet@49266
    34
  val thm_ord : thm * thm -> order
blanchet@49333
    35
  val features_of :
blanchet@49333
    36
    Proof.context -> string -> theory -> status -> term list -> string list
blanchet@49331
    37
  val isabelle_dependencies_of : unit Symtab.table -> thm -> string list
blanchet@49266
    38
  val goal_of_thm : theory -> thm -> thm
blanchet@49336
    39
  val run_prover_for_mash :
blanchet@49333
    40
    Proof.context -> params -> string -> fact list -> thm -> prover_result
blanchet@49323
    41
  val mash_RESET : Proof.context -> unit
blanchet@49331
    42
  val mash_INIT :
blanchet@49331
    43
    Proof.context -> bool
blanchet@49331
    44
    -> (string * string list * string list * string list) list -> unit
blanchet@49323
    45
  val mash_ADD :
blanchet@49331
    46
    Proof.context -> bool
blanchet@49331
    47
    -> (string * string list * string list * string list) list -> unit
blanchet@49331
    48
  val mash_QUERY :
blanchet@49333
    49
    Proof.context -> bool -> int -> string list * string list -> string list
blanchet@49323
    50
  val mash_reset : Proof.context -> unit
blanchet@49333
    51
  val mash_could_suggest_facts : unit -> bool
blanchet@49336
    52
  val mash_can_suggest_facts : Proof.context -> bool
blanchet@49313
    53
  val mash_suggest_facts :
blanchet@49336
    54
    Proof.context -> params -> string -> int -> term list -> term
blanchet@49336
    55
    -> ('a * thm) list -> ('a * thm) list * ('a * thm) list
blanchet@49334
    56
  val mash_learn_thy :
blanchet@49334
    57
    Proof.context -> params -> theory -> Time.time -> fact list -> string
blanchet@49331
    58
  val mash_learn_proof :
blanchet@49336
    59
    Proof.context -> params -> term -> ('a * thm) list -> thm list -> unit
blanchet@49303
    60
  val relevant_facts :
blanchet@49307
    61
    Proof.context -> params -> string -> int -> fact_override -> term list
blanchet@49311
    62
    -> term -> fact list -> fact list
blanchet@49334
    63
  val kill_learners : unit -> unit
blanchet@49334
    64
  val running_learners : unit -> unit
blanchet@49263
    65
end;
blanchet@49263
    66
blanchet@49263
    67
structure Sledgehammer_Filter_MaSh : SLEDGEHAMMER_FILTER_MASH =
blanchet@49263
    68
struct
blanchet@49264
    69
blanchet@49266
    70
open ATP_Util
blanchet@49266
    71
open ATP_Problem_Generate
blanchet@49266
    72
open Sledgehammer_Util
blanchet@49266
    73
open Sledgehammer_Fact
blanchet@49266
    74
open Sledgehammer_Filter_Iter
blanchet@49266
    75
open Sledgehammer_Provers
blanchet@49333
    76
open Sledgehammer_Minimize
blanchet@49266
    77
blanchet@49323
    78
val trace =
blanchet@49323
    79
  Attrib.setup_config_bool @{binding sledgehammer_filter_mash_trace} (K false)
blanchet@49323
    80
fun trace_msg ctxt msg = if Config.get ctxt trace then tracing (msg ()) else ()
blanchet@49323
    81
blanchet@49334
    82
val MaShN = "MaSh"
blanchet@49334
    83
blanchet@49329
    84
val meshN = "mesh"
blanchet@49329
    85
val iterN = "iter"
blanchet@49329
    86
val mashN = "mash"
blanchet@49329
    87
blanchet@49329
    88
val fact_filters = [meshN, iterN, mashN]
blanchet@49329
    89
blanchet@49329
    90
fun mash_home () = getenv "MASH_HOME"
blanchet@49329
    91
fun mash_state_dir () =
blanchet@49324
    92
  getenv "ISABELLE_HOME_USER" ^ "/mash"
blanchet@49331
    93
  |> tap (Isabelle_System.mkdir o Path.explode)
blanchet@49329
    94
fun mash_state_path () = mash_state_dir () ^ "/state" |> Path.explode
blanchet@49266
    95
blanchet@49266
    96
(*** Isabelle helpers ***)
blanchet@49266
    97
blanchet@49323
    98
fun meta_char c =
blanchet@49266
    99
  if Char.isAlphaNum c orelse c = #"_" orelse c = #"." orelse c = #"(" orelse
blanchet@49266
   100
     c = #")" orelse c = #"," then
blanchet@49266
   101
    String.str c
blanchet@49266
   102
  else
blanchet@49266
   103
    (* fixed width, in case more digits follow *)
blanchet@49266
   104
    "\\" ^ stringN_of_int 3 (Char.ord c)
blanchet@49266
   105
blanchet@49323
   106
fun unmeta_chars accum [] = String.implode (rev accum)
blanchet@49323
   107
  | unmeta_chars accum (#"\\" :: d1 :: d2 :: d3 :: cs) =
blanchet@49323
   108
    (case Int.fromString (String.implode [d1, d2, d3]) of
blanchet@49323
   109
       SOME n => unmeta_chars (Char.chr n :: accum) cs
blanchet@49323
   110
     | NONE => "" (* error *))
blanchet@49323
   111
  | unmeta_chars _ (#"\\" :: _) = "" (* error *)
blanchet@49323
   112
  | unmeta_chars accum (c :: cs) = unmeta_chars (c :: accum) cs
blanchet@49323
   113
blanchet@49323
   114
val escape_meta = String.translate meta_char
blanchet@49318
   115
val escape_metas = map escape_meta #> space_implode " "
blanchet@49330
   116
val unescape_meta = String.explode #> unmeta_chars []
blanchet@49330
   117
val unescape_metas =
blanchet@49330
   118
  space_explode " " #> filter_out (curry (op =) "") #> map unescape_meta
blanchet@49266
   119
blanchet@49326
   120
fun extract_query line =
blanchet@49326
   121
  case space_explode ":" line of
blanchet@49330
   122
    [goal_name, suggs] => (unescape_meta goal_name, unescape_metas suggs)
blanchet@49327
   123
  | _ => ("", [])
blanchet@49326
   124
blanchet@49326
   125
fun find_suggested facts sugg =
blanchet@49326
   126
  find_first (fn (_, th) => Thm.get_name_hint th = sugg) facts
blanchet@49326
   127
fun suggested_facts suggs facts = map_filter (find_suggested facts) suggs
blanchet@49326
   128
blanchet@49335
   129
fun sum_avg _ [] = 1000000000 (* big number *)
blanchet@49335
   130
  | sum_avg n xs = fold (Integer.add o Integer.mult n) xs 0 div (length xs)
blanchet@49328
   131
blanchet@49335
   132
fun mesh_facts max_facts [(selected, unknown)] =
blanchet@49335
   133
    take max_facts selected @ take (max_facts - length selected) unknown
blanchet@49329
   134
  | mesh_facts max_facts mess =
blanchet@49329
   135
    let
blanchet@49335
   136
      val mess = mess |> map (apfst (`length))
blanchet@49329
   137
      val n = length mess
blanchet@49329
   138
      val fact_eq = Thm.eq_thm o pairself snd
blanchet@49335
   139
      fun score_in fact ((sel_len, sels), unks) =
blanchet@49335
   140
        case find_index (curry fact_eq fact) sels of
blanchet@49335
   141
          ~1 => (case find_index (curry fact_eq fact) unks of
blanchet@49335
   142
                   ~1 => SOME sel_len
blanchet@49335
   143
                 | _ => NONE)
blanchet@49329
   144
        | j => SOME j
blanchet@49329
   145
      fun score_of fact = mess |> map_filter (score_in fact) |> sum_avg n
blanchet@49335
   146
      val facts = fold (union fact_eq o take max_facts o snd o fst) mess []
blanchet@49329
   147
    in
blanchet@49329
   148
      facts |> map (`score_of) |> sort (int_ord o pairself fst) |> map snd
blanchet@49329
   149
            |> take max_facts
blanchet@49329
   150
    end
blanchet@49327
   151
blanchet@49318
   152
val thy_feature_prefix = "y_"
blanchet@49266
   153
blanchet@49318
   154
val thy_feature_name_of = prefix thy_feature_prefix
blanchet@49318
   155
val const_name_of = prefix const_prefix
blanchet@49318
   156
val type_name_of = prefix type_const_prefix
blanchet@49318
   157
val class_name_of = prefix class_prefix
blanchet@49266
   158
blanchet@49266
   159
local
blanchet@49266
   160
blanchet@49266
   161
fun has_bool @{typ bool} = true
blanchet@49266
   162
  | has_bool (Type (_, Ts)) = exists has_bool Ts
blanchet@49266
   163
  | has_bool _ = false
blanchet@49266
   164
blanchet@49266
   165
fun has_fun (Type (@{type_name fun}, _)) = true
blanchet@49266
   166
  | has_fun (Type (_, Ts)) = exists has_fun Ts
blanchet@49266
   167
  | has_fun _ = false
blanchet@49266
   168
blanchet@49266
   169
val is_conn = member (op =)
blanchet@49266
   170
  [@{const_name Trueprop}, @{const_name HOL.conj}, @{const_name HOL.disj},
blanchet@49266
   171
   @{const_name HOL.implies}, @{const_name Not},
blanchet@49266
   172
   @{const_name All}, @{const_name Ex}, @{const_name Ball}, @{const_name Bex},
blanchet@49266
   173
   @{const_name HOL.eq}]
blanchet@49266
   174
blanchet@49266
   175
val has_bool_arg_const =
blanchet@49266
   176
  exists_Const (fn (c, T) =>
blanchet@49266
   177
                   not (is_conn c) andalso exists has_bool (binder_types T))
blanchet@49266
   178
blanchet@49331
   179
fun higher_inst_const thy (s, T) =
blanchet@49266
   180
  case binder_types T of
blanchet@49266
   181
    [] => false
blanchet@49331
   182
  | Ts => length (binder_types (Sign.the_const_type thy s)) <> length Ts
blanchet@49331
   183
  handle TYPE _ => false
blanchet@49266
   184
blanchet@49266
   185
val binders = [@{const_name All}, @{const_name Ex}]
blanchet@49266
   186
blanchet@49266
   187
in
blanchet@49266
   188
blanchet@49266
   189
fun is_fo_term thy t =
blanchet@49266
   190
  let
blanchet@49266
   191
    val t =
blanchet@49266
   192
      t |> Envir.beta_eta_contract
blanchet@49266
   193
        |> transform_elim_prop
blanchet@49266
   194
        |> Object_Logic.atomize_term thy
blanchet@49266
   195
  in
blanchet@49266
   196
    Term.is_first_order binders t andalso
blanchet@49266
   197
    not (exists_subterm (fn Var (_, T) => has_bool T orelse has_fun T
blanchet@49266
   198
                          | _ => false) t orelse
blanchet@49266
   199
         has_bool_arg_const t orelse exists_Const (higher_inst_const thy) t)
blanchet@49266
   200
  end
blanchet@49266
   201
blanchet@49266
   202
end
blanchet@49266
   203
blanchet@49333
   204
fun interesting_terms_types_and_classes ctxt prover term_max_depth
blanchet@49333
   205
                                        type_max_depth ts =
blanchet@49266
   206
  let
blanchet@49266
   207
    val bad_types = [@{type_name prop}, @{type_name bool}, @{type_name fun}]
blanchet@49333
   208
    fun is_bad_const (x as (s, _)) args =
blanchet@49333
   209
      member (op =) atp_logical_consts s orelse
blanchet@49333
   210
      fst (is_built_in_const_for_prover ctxt prover x args)
blanchet@49319
   211
    fun add_classes @{sort type} = I
blanchet@49319
   212
      | add_classes S = union (op =) (map class_name_of S)
blanchet@49266
   213
    fun do_add_type (Type (s, Ts)) =
blanchet@49266
   214
        (not (member (op =) bad_types s) ? insert (op =) (type_name_of s))
blanchet@49266
   215
        #> fold do_add_type Ts
blanchet@49319
   216
      | do_add_type (TFree (_, S)) = add_classes S
blanchet@49319
   217
      | do_add_type (TVar (_, S)) = add_classes S
blanchet@49266
   218
    fun add_type T = type_max_depth >= 0 ? do_add_type T
blanchet@49266
   219
    fun mk_app s args =
blanchet@49266
   220
      if member (op <>) args "" then s ^ "(" ^ space_implode "," args ^ ")"
blanchet@49266
   221
      else s
blanchet@49266
   222
    fun patternify ~1 _ = ""
blanchet@49266
   223
      | patternify depth t =
blanchet@49266
   224
        case strip_comb t of
blanchet@49266
   225
          (Const (s, _), args) =>
blanchet@49266
   226
          mk_app (const_name_of s) (map (patternify (depth - 1)) args)
blanchet@49266
   227
        | _ => ""
blanchet@49266
   228
    fun add_term_patterns ~1 _ = I
blanchet@49266
   229
      | add_term_patterns depth t =
blanchet@49266
   230
        insert (op =) (patternify depth t)
blanchet@49266
   231
        #> add_term_patterns (depth - 1) t
blanchet@49266
   232
    val add_term = add_term_patterns term_max_depth
blanchet@49266
   233
    fun add_patterns t =
blanchet@49266
   234
      let val (head, args) = strip_comb t in
blanchet@49266
   235
        (case head of
blanchet@49333
   236
           Const (x as (_, T)) =>
blanchet@49333
   237
           not (is_bad_const x args) ? (add_term t #> add_type T)
blanchet@49266
   238
         | Free (_, T) => add_type T
blanchet@49266
   239
         | Var (_, T) => add_type T
blanchet@49266
   240
         | Abs (_, T, body) => add_type T #> add_patterns body
blanchet@49266
   241
         | _ => I)
blanchet@49266
   242
        #> fold add_patterns args
blanchet@49266
   243
      end
blanchet@49317
   244
  in [] |> fold add_patterns ts |> sort string_ord end
blanchet@49266
   245
blanchet@49333
   246
fun is_likely_tautology ctxt prover th =
blanchet@49333
   247
  null (interesting_terms_types_and_classes ctxt prover 0 ~1 [prop_of th])
blanchet@49333
   248
  andalso not (Thm.eq_thm_prop (@{thm ext}, th))
blanchet@49266
   249
blanchet@49330
   250
(* ### FIXME: optimize *)
blanchet@49330
   251
fun is_too_meta th =
blanchet@49330
   252
  fastype_of (Object_Logic.atomize_term (theory_of_thm th) (prop_of th))
blanchet@49330
   253
  <> @{typ bool}
blanchet@49266
   254
blanchet@49266
   255
fun theory_ord p =
blanchet@49330
   256
  if Theory.eq_thy p then
blanchet@49330
   257
    EQUAL
blanchet@49330
   258
  else if Theory.subthy p then
blanchet@49330
   259
    LESS
blanchet@49330
   260
  else if Theory.subthy (swap p) then
blanchet@49330
   261
    GREATER
blanchet@49330
   262
  else case int_ord (pairself (length o Theory.ancestors_of) p) of
blanchet@49330
   263
    EQUAL => string_ord (pairself Context.theory_name p)
blanchet@49330
   264
  | order => order
blanchet@49266
   265
blanchet@49266
   266
val thm_ord = theory_ord o pairself theory_of_thm
blanchet@49266
   267
blanchet@49266
   268
fun is_exists (s, _) = (s = @{const_name Ex} orelse s = @{const_name Ex1})
blanchet@49266
   269
blanchet@49312
   270
val term_max_depth = 1
blanchet@49312
   271
val type_max_depth = 1
blanchet@49266
   272
blanchet@49266
   273
(* TODO: Generate type classes for types? *)
blanchet@49333
   274
fun features_of ctxt prover thy status ts =
blanchet@49318
   275
  thy_feature_name_of (Context.theory_name thy) ::
blanchet@49333
   276
  interesting_terms_types_and_classes ctxt prover term_max_depth type_max_depth
blanchet@49333
   277
                                      ts
blanchet@49317
   278
  |> exists (not o is_lambda_free) ts ? cons "lambdas"
blanchet@49317
   279
  |> exists (exists_Const is_exists) ts ? cons "skolems"
blanchet@49317
   280
  |> exists (not o is_fo_term thy) ts ? cons "ho"
blanchet@49317
   281
  |> (case status of
blanchet@49317
   282
        General => I
blanchet@49317
   283
      | Induction => cons "induction"
blanchet@49317
   284
      | Intro => cons "intro"
blanchet@49317
   285
      | Inductive => cons "inductive"
blanchet@49317
   286
      | Elim => cons "elim"
blanchet@49317
   287
      | Simp => cons "simp"
blanchet@49317
   288
      | Def => cons "def")
blanchet@49266
   289
blanchet@49266
   290
fun isabelle_dependencies_of all_facts =
blanchet@49318
   291
  thms_in_proof (SOME all_facts) #> sort string_ord
blanchet@49266
   292
blanchet@49266
   293
val freezeT = Type.legacy_freeze_type
blanchet@49266
   294
blanchet@49266
   295
fun freeze (t $ u) = freeze t $ freeze u
blanchet@49266
   296
  | freeze (Abs (s, T, t)) = Abs (s, freezeT T, freeze t)
blanchet@49266
   297
  | freeze (Var ((s, _), T)) = Free (s, freezeT T)
blanchet@49266
   298
  | freeze (Const (s, T)) = Const (s, freezeT T)
blanchet@49266
   299
  | freeze (Free (s, T)) = Free (s, freezeT T)
blanchet@49266
   300
  | freeze t = t
blanchet@49266
   301
blanchet@49266
   302
fun goal_of_thm thy = prop_of #> freeze #> cterm_of thy #> Goal.init
blanchet@49266
   303
blanchet@49336
   304
fun run_prover_for_mash ctxt params prover facts goal =
blanchet@49266
   305
  let
blanchet@49266
   306
    val problem =
blanchet@49266
   307
      {state = Proof.init ctxt, goal = goal, subgoal = 1, subgoal_count = 1,
blanchet@49304
   308
       facts = facts |> map (apfst (apfst (fn name => name ())))
blanchet@49333
   309
                     |> map Untranslated_Fact}
blanchet@49336
   310
    val prover = get_minimizing_prover ctxt Normal (K ()) prover
blanchet@49266
   311
  in prover params (K (K (K ""))) problem end
blanchet@49266
   312
blanchet@49266
   313
blanchet@49317
   314
(*** Low-level communication with MaSh ***)
blanchet@49317
   315
blanchet@49333
   316
fun write_file write file =
blanchet@49333
   317
  let val path = Path.explode file in
blanchet@49333
   318
    File.write path ""; write (File.append path)
blanchet@49333
   319
  end
blanchet@49331
   320
blanchet@49331
   321
fun mash_info overlord =
blanchet@49331
   322
  if overlord then (getenv "ISABELLE_HOME_USER", "")
blanchet@49331
   323
  else (getenv "ISABELLE_TMP", serial_string ())
blanchet@49331
   324
blanchet@49331
   325
fun run_mash ctxt (temp_dir, serial) core =
blanchet@49326
   326
  let
blanchet@49331
   327
    val log_file = temp_dir ^ "/mash_log" ^ serial
blanchet@49331
   328
    val err_file = temp_dir ^ "/mash_err" ^ serial
blanchet@49326
   329
    val command =
blanchet@49331
   330
      mash_home () ^ "/mash.py --quiet --outputDir " ^ mash_state_dir () ^
blanchet@49331
   331
      " --log " ^ log_file ^ " " ^ core ^ " 2>&1 > " ^ err_file
blanchet@49331
   332
  in
blanchet@49331
   333
    trace_msg ctxt (fn () => "Running " ^ command);
blanchet@49333
   334
    write_file (K ()) log_file;
blanchet@49333
   335
    write_file (K ()) err_file;
blanchet@49331
   336
    Isabelle_System.bash command; ()
blanchet@49331
   337
  end
blanchet@49326
   338
blanchet@49331
   339
fun run_mash_init ctxt overlord write_access write_feats write_deps =
blanchet@49331
   340
  let
blanchet@49331
   341
    val info as (temp_dir, serial) = mash_info overlord
blanchet@49331
   342
    val in_dir = temp_dir ^ "/mash_init" ^ serial
blanchet@49331
   343
                 |> tap (Isabelle_System.mkdir o Path.explode)
blanchet@49331
   344
  in
blanchet@49331
   345
    write_file write_access (in_dir ^ "/mash_accessibility");
blanchet@49331
   346
    write_file write_feats (in_dir ^ "/mash_features");
blanchet@49331
   347
    write_file write_deps (in_dir ^ "/mash_dependencies");
blanchet@49331
   348
    run_mash ctxt info ("--init --inputDir " ^ in_dir)
blanchet@49331
   349
  end
blanchet@49331
   350
blanchet@49333
   351
fun run_mash_commands ctxt overlord save max_suggs write_cmds read_suggs =
blanchet@49331
   352
  let
blanchet@49331
   353
    val info as (temp_dir, serial) = mash_info overlord
blanchet@49333
   354
    val sugg_file = temp_dir ^ "/mash_suggs" ^ serial
blanchet@49331
   355
    val cmd_file = temp_dir ^ "/mash_commands" ^ serial
blanchet@49331
   356
  in
blanchet@49333
   357
    write_file (K ()) sugg_file;
blanchet@49331
   358
    write_file write_cmds cmd_file;
blanchet@49331
   359
    run_mash ctxt info
blanchet@49333
   360
             ("--inputFile " ^ cmd_file ^ " --predictions " ^ sugg_file ^
blanchet@49333
   361
              " --numberOfPredictions " ^ string_of_int max_suggs ^
blanchet@49331
   362
              (if save then " --saveModel" else ""));
blanchet@49333
   363
    read_suggs (fn () => File.read_lines (Path.explode sugg_file))
blanchet@49331
   364
  end
blanchet@49331
   365
blanchet@49331
   366
fun str_of_update (name, parents, feats, deps) =
blanchet@49331
   367
  "! " ^ escape_meta name ^ ": " ^ escape_metas parents ^ "; " ^
blanchet@49326
   368
  escape_metas feats ^ "; " ^ escape_metas deps ^ "\n"
blanchet@49326
   369
blanchet@49331
   370
fun str_of_query (parents, feats) =
blanchet@49331
   371
  "? " ^ escape_metas parents ^ "; " ^ escape_metas feats
blanchet@49331
   372
blanchet@49331
   373
fun init_str_of_update get (upd as (name, _, _, _)) =
blanchet@49331
   374
  escape_meta name ^ ": " ^ escape_metas (get upd) ^ "\n"
blanchet@49326
   375
blanchet@49323
   376
fun mash_RESET ctxt =
blanchet@49329
   377
  let val path = mash_state_dir () |> Path.explode in
blanchet@49324
   378
    trace_msg ctxt (K "MaSh RESET");
blanchet@49324
   379
    File.fold_dir (fn file => fn () =>
blanchet@49324
   380
                      File.rm (Path.append path (Path.basic file)))
blanchet@49324
   381
                  path ()
blanchet@49324
   382
  end
blanchet@49317
   383
blanchet@49331
   384
fun mash_INIT ctxt _ [] = mash_RESET ctxt
blanchet@49331
   385
  | mash_INIT ctxt overlord upds =
blanchet@49331
   386
    (trace_msg ctxt (fn () => "MaSh INIT " ^
blanchet@49331
   387
         elide_string 1000 (space_implode " " (map #1 upds)));
blanchet@49331
   388
     run_mash_init ctxt overlord
blanchet@49331
   389
         (fn append => List.app (append o init_str_of_update #2) upds)
blanchet@49331
   390
         (fn append => List.app (append o init_str_of_update #3) upds)
blanchet@49331
   391
         (fn append => List.app (append o init_str_of_update #4) upds))
blanchet@49317
   392
blanchet@49331
   393
fun mash_ADD _ _ [] = ()
blanchet@49331
   394
  | mash_ADD ctxt overlord upds =
blanchet@49331
   395
    (trace_msg ctxt (fn () => "MaSh ADD " ^
blanchet@49331
   396
         elide_string 1000 (space_implode " " (map #1 upds)));
blanchet@49333
   397
     run_mash_commands ctxt overlord true 0
blanchet@49331
   398
         (fn append => List.app (append o str_of_update) upds) (K ()))
blanchet@49317
   399
blanchet@49333
   400
fun mash_QUERY ctxt overlord max_suggs (query as (_, feats)) =
blanchet@49329
   401
  (trace_msg ctxt (fn () => "MaSh QUERY " ^ space_implode " " feats);
blanchet@49333
   402
   run_mash_commands ctxt overlord false max_suggs
blanchet@49331
   403
       (fn append => append (str_of_query query))
blanchet@49333
   404
       (fn suggs => snd (extract_query (List.last (suggs ()))))
blanchet@49326
   405
   handle List.Empty => [])
blanchet@49317
   406
blanchet@49317
   407
blanchet@49266
   408
(*** High-level communication with MaSh ***)
blanchet@49266
   409
blanchet@49336
   410
fun try_graph ctxt when def f =
blanchet@49336
   411
  f ()
blanchet@49336
   412
  handle Graph.CYCLES (cycle :: _) =>
blanchet@49336
   413
         (trace_msg ctxt (fn () =>
blanchet@49336
   414
              "Cycle involving " ^ commas cycle ^ " when " ^ when); def)
blanchet@49336
   415
       | Graph.UNDEF name =>
blanchet@49336
   416
         (trace_msg ctxt (fn () =>
blanchet@49336
   417
              "Unknown fact " ^ quote name ^ " when " ^ when); def)
blanchet@49336
   418
blanchet@49316
   419
type mash_state =
blanchet@49331
   420
  {thys : bool Symtab.table,
blanchet@49331
   421
   fact_graph : unit Graph.T}
blanchet@49264
   422
blanchet@49331
   423
val empty_state = {thys = Symtab.empty, fact_graph = Graph.empty}
blanchet@49316
   424
blanchet@49316
   425
local
blanchet@49316
   426
blanchet@49336
   427
fun mash_load _ (state as (true, _)) = state
blanchet@49336
   428
  | mash_load ctxt _ =
blanchet@49324
   429
    let val path = mash_state_path () in
blanchet@49317
   430
      (true,
blanchet@49317
   431
       case try File.read_lines path of
blanchet@49331
   432
         SOME (comp_thys :: incomp_thys :: fact_lines) =>
blanchet@49317
   433
         let
blanchet@49331
   434
           fun add_thy comp thy = Symtab.update (thy, comp)
blanchet@49331
   435
           fun add_fact_line line =
blanchet@49331
   436
             case extract_query line of
blanchet@49331
   437
               ("", _) => I (* shouldn't happen *)
blanchet@49331
   438
             | (name, parents) =>
blanchet@49331
   439
               Graph.default_node (name, ())
blanchet@49331
   440
               #> fold (fn par => Graph.add_edge (par, name)) parents
blanchet@49331
   441
           val thys =
blanchet@49331
   442
             Symtab.empty |> fold (add_thy true) (unescape_metas comp_thys)
blanchet@49331
   443
                          |> fold (add_thy false) (unescape_metas incomp_thys)
blanchet@49336
   444
           val fact_graph =
blanchet@49336
   445
             try_graph ctxt "loading state file" Graph.empty (fn () =>
blanchet@49336
   446
                 Graph.empty |> fold add_fact_line fact_lines)
blanchet@49331
   447
         in {thys = thys, fact_graph = fact_graph} end
blanchet@49319
   448
       | _ => empty_state)
blanchet@49317
   449
    end
blanchet@49316
   450
blanchet@49331
   451
fun mash_save ({thys, fact_graph, ...} : mash_state) =
blanchet@49316
   452
  let
blanchet@49324
   453
    val path = mash_state_path ()
blanchet@49331
   454
    val thys = Symtab.dest thys
blanchet@49333
   455
    val line_for_thys = escape_metas o AList.find (op =) thys
blanchet@49333
   456
    fun fact_line_for name parents =
blanchet@49333
   457
      escape_meta name ^ ": " ^ escape_metas parents
blanchet@49331
   458
    val append_fact = File.append path o suffix "\n" oo fact_line_for
blanchet@49316
   459
  in
blanchet@49331
   460
    File.write path (line_for_thys true ^ "\n" ^ line_for_thys false ^ "\n");
blanchet@49331
   461
    Graph.fold (fn (name, ((), (parents, _))) => fn () =>
blanchet@49331
   462
                   append_fact name (Graph.Keys.dest parents))
blanchet@49331
   463
        fact_graph ()
blanchet@49316
   464
  end
blanchet@49316
   465
blanchet@49317
   466
val global_state =
blanchet@49319
   467
  Synchronized.var "Sledgehammer_Filter_MaSh.global_state" (false, empty_state)
blanchet@49316
   468
blanchet@49316
   469
in
blanchet@49316
   470
blanchet@49336
   471
fun mash_map ctxt f =
blanchet@49336
   472
  Synchronized.change global_state (mash_load ctxt ##> (f #> tap mash_save))
blanchet@49316
   473
blanchet@49336
   474
fun mash_get ctxt =
blanchet@49336
   475
  Synchronized.change_result global_state (mash_load ctxt #> `snd)
blanchet@49317
   476
blanchet@49323
   477
fun mash_reset ctxt =
blanchet@49317
   478
  Synchronized.change global_state (fn _ =>
blanchet@49324
   479
      (mash_RESET ctxt; File.write (mash_state_path ()) "";
blanchet@49323
   480
       (true, empty_state)))
blanchet@49316
   481
blanchet@49316
   482
end
blanchet@49316
   483
blanchet@49333
   484
fun mash_could_suggest_facts () = mash_home () <> ""
blanchet@49336
   485
fun mash_can_suggest_facts ctxt =
blanchet@49336
   486
  not (Graph.is_empty (#fact_graph (mash_get ctxt)))
blanchet@49264
   487
blanchet@49336
   488
fun parents_wrt_facts ctxt facts fact_graph =
blanchet@49331
   489
  let
blanchet@49333
   490
    val graph_facts = Symtab.make (map (rpair ()) (Graph.keys fact_graph))
blanchet@49331
   491
    val facts =
blanchet@49336
   492
      try_graph ctxt "when computing ancestor facts" [] (fn () =>
blanchet@49336
   493
          [] |> fold (cons o Thm.get_name_hint o snd) facts
blanchet@49336
   494
             |> filter (Symtab.defined graph_facts)
blanchet@49336
   495
             |> Graph.all_preds fact_graph)
blanchet@49333
   496
    val facts = Symtab.empty |> fold (fn name => Symtab.update (name, ())) facts
blanchet@49336
   497
  in
blanchet@49336
   498
    try_graph ctxt "when computing parent facts" [] (fn () =>
blanchet@49336
   499
        fact_graph |> Graph.restrict (Symtab.defined facts) |> Graph.maximals)
blanchet@49336
   500
  end
blanchet@49331
   501
blanchet@49333
   502
(* Generate more suggestions than requested, because some might be thrown out
blanchet@49333
   503
   later for various reasons and "meshing" gives better results with some
blanchet@49333
   504
   slack. *)
blanchet@49333
   505
fun max_suggs_of max_facts = max_facts + Int.min (200, max_facts)
blanchet@49333
   506
blanchet@49335
   507
fun is_fact_in_graph fact_graph (_, th) =
blanchet@49335
   508
  can (Graph.get_node fact_graph) (Thm.get_name_hint th)
blanchet@49335
   509
blanchet@49333
   510
fun mash_suggest_facts ctxt ({overlord, ...} : params) prover max_facts hyp_ts
blanchet@49333
   511
                       concl_t facts =
blanchet@49316
   512
  let
blanchet@49317
   513
    val thy = Proof_Context.theory_of ctxt
blanchet@49336
   514
    val fact_graph = #fact_graph (mash_get ctxt)
blanchet@49336
   515
    val parents = parents_wrt_facts ctxt facts fact_graph
blanchet@49333
   516
    val feats = features_of ctxt prover thy General (concl_t :: hyp_ts)
blanchet@49333
   517
    val suggs =
blanchet@49335
   518
      if Graph.is_empty fact_graph then []
blanchet@49335
   519
      else mash_QUERY ctxt overlord (max_suggs_of max_facts) (parents, feats)
blanchet@49335
   520
    val selected = facts |> suggested_facts suggs
blanchet@49335
   521
    val unknown = facts |> filter_out (is_fact_in_graph fact_graph)
blanchet@49335
   522
  in (selected, unknown) end
blanchet@49264
   523
blanchet@49331
   524
fun add_thys_for thy =
blanchet@49331
   525
  let fun add comp thy = Symtab.update (Context.theory_name thy, comp) in
blanchet@49331
   526
    add false thy #> fold (add true) (Theory.ancestors_of thy)
blanchet@49331
   527
  end
blanchet@49319
   528
blanchet@49331
   529
fun update_fact_graph ctxt (name, parents, feats, deps) (upds, graph) =
blanchet@49331
   530
  let
blanchet@49331
   531
    fun maybe_add_from from (accum as (parents, graph)) =
blanchet@49336
   532
      try_graph ctxt "updating graph" accum (fn () =>
blanchet@49336
   533
          (from :: parents, Graph.add_edge_acyclic (from, name) graph))
blanchet@49336
   534
    val graph = graph |> Graph.default_node (name, ())
blanchet@49331
   535
    val (parents, graph) = ([], graph) |> fold maybe_add_from parents
blanchet@49331
   536
    val (deps, graph) = ([], graph) |> fold maybe_add_from deps
blanchet@49331
   537
  in ((name, parents, feats, deps) :: upds, graph) end
blanchet@49321
   538
blanchet@49333
   539
val pass1_learn_timeout_factor = 0.5
blanchet@49333
   540
blanchet@49333
   541
(* The timeout is understood in a very slack fashion. *)
blanchet@49334
   542
fun mash_learn_thy ctxt ({provers, verbose, overlord, ...} : params) thy timeout
blanchet@49334
   543
                   facts =
blanchet@49319
   544
  let
blanchet@49333
   545
    val timer = Timer.startRealTimer ()
blanchet@49333
   546
    val prover = hd provers
blanchet@49333
   547
    fun timed_out frac =
blanchet@49333
   548
      Time.> (Timer.checkRealTimer timer, time_mult frac timeout)
blanchet@49336
   549
    val {fact_graph, ...} = mash_get ctxt
blanchet@49335
   550
    val new_facts =
blanchet@49335
   551
      facts |> filter_out (is_fact_in_graph fact_graph)
blanchet@49335
   552
            |> sort (thm_ord o pairself snd)
blanchet@49323
   553
  in
blanchet@49323
   554
    if null new_facts then
blanchet@49334
   555
      ""
blanchet@49323
   556
    else
blanchet@49319
   557
      let
blanchet@49323
   558
        val ths = facts |> map snd
blanchet@49330
   559
        val all_names =
blanchet@49333
   560
          ths |> filter_out (is_likely_tautology ctxt prover orf is_too_meta)
blanchet@49331
   561
              |> map (rpair () o Thm.get_name_hint)
blanchet@49331
   562
              |> Symtab.make
blanchet@49333
   563
        fun do_fact _ (accum as (_, true)) = accum
blanchet@49333
   564
          | do_fact ((_, (_, status)), th) ((parents, upds), false) =
blanchet@49333
   565
            let
blanchet@49333
   566
              val name = Thm.get_name_hint th
blanchet@49333
   567
              val feats = features_of ctxt prover thy status [prop_of th]
blanchet@49333
   568
              val deps = isabelle_dependencies_of all_names th
blanchet@49333
   569
              val upd = (name, parents, feats, deps)
blanchet@49333
   570
            in (([name], upd :: upds), timed_out pass1_learn_timeout_factor) end
blanchet@49336
   571
        val parents = parents_wrt_facts ctxt facts fact_graph
blanchet@49333
   572
        val ((_, upds), _) =
blanchet@49333
   573
          ((parents, []), false) |> fold do_fact new_facts |>> apsnd rev
blanchet@49333
   574
        val n = length upds
blanchet@49331
   575
        fun trans {thys, fact_graph} =
blanchet@49331
   576
          let
blanchet@49331
   577
            val mash_INIT_or_ADD =
blanchet@49331
   578
              if Graph.is_empty fact_graph then mash_INIT else mash_ADD
blanchet@49331
   579
            val (upds, fact_graph) =
blanchet@49331
   580
              ([], fact_graph) |> fold (update_fact_graph ctxt) upds
blanchet@49331
   581
          in
blanchet@49331
   582
            (mash_INIT_or_ADD ctxt overlord (rev upds);
blanchet@49331
   583
             {thys = thys |> add_thys_for thy,
blanchet@49331
   584
              fact_graph = fact_graph})
blanchet@49331
   585
          end
blanchet@49333
   586
      in
blanchet@49336
   587
        mash_map ctxt trans;
blanchet@49334
   588
        if verbose then
blanchet@49334
   589
          "Processed " ^ string_of_int n ^ " proof" ^ plural_s n ^
blanchet@49334
   590
          (if verbose then
blanchet@49334
   591
             " in " ^ string_from_time (Timer.checkRealTimer timer)
blanchet@49334
   592
           else
blanchet@49334
   593
             "") ^ "."
blanchet@49334
   594
        else
blanchet@49334
   595
          ""
blanchet@49333
   596
      end
blanchet@49323
   597
  end
blanchet@49319
   598
blanchet@49336
   599
fun mash_learn_proof ctxt ({provers, overlord, ...} : params) t facts used_ths =
blanchet@49331
   600
  let
blanchet@49331
   601
    val thy = Proof_Context.theory_of ctxt
blanchet@49333
   602
    val prover = hd provers
blanchet@49333
   603
    val name = ATP_Util.timestamp () ^ serial_string () (* fresh enough *)
blanchet@49333
   604
    val feats = features_of ctxt prover thy General [t]
blanchet@49331
   605
    val deps = used_ths |> map Thm.get_name_hint
blanchet@49331
   606
  in
blanchet@49336
   607
    mash_map ctxt (fn {thys, fact_graph} =>
blanchet@49324
   608
        let
blanchet@49336
   609
          val parents = parents_wrt_facts ctxt facts fact_graph
blanchet@49331
   610
          val upds = [(name, parents, feats, deps)]
blanchet@49331
   611
          val (upds, fact_graph) =
blanchet@49331
   612
            ([], fact_graph) |> fold (update_fact_graph ctxt) upds
blanchet@49324
   613
        in
blanchet@49331
   614
          mash_ADD ctxt overlord upds;
blanchet@49331
   615
          {thys = thys, fact_graph = fact_graph}
blanchet@49331
   616
        end)
blanchet@49331
   617
  end
blanchet@49264
   618
blanchet@49333
   619
(* The threshold should be large enough so that MaSh doesn't kick in for Auto
blanchet@49333
   620
   Sledgehammer and Try. *)
blanchet@49333
   621
val min_secs_for_learning = 15
blanchet@49333
   622
val short_learn_timeout_factor = 0.2
blanchet@49333
   623
val long_learn_timeout_factor = 4.0
blanchet@49333
   624
blanchet@49336
   625
fun relevant_facts ctxt (params as {learn, fact_filter, timeout, ...}) prover
blanchet@49336
   626
        max_facts ({add, only, ...} : fact_override) hyp_ts concl_t facts =
blanchet@49329
   627
  if not (subset (op =) (the_list fact_filter, fact_filters)) then
blanchet@49329
   628
    error ("Unknown fact filter: " ^ quote (the fact_filter) ^ ".")
blanchet@49329
   629
  else if only then
blanchet@49304
   630
    facts
blanchet@49336
   631
  else if max_facts <= 0 orelse null facts then
blanchet@49303
   632
    []
blanchet@49303
   633
  else
blanchet@49303
   634
    let
blanchet@49333
   635
      val thy = Proof_Context.theory_of ctxt
blanchet@49333
   636
      fun maybe_learn can_suggest =
blanchet@49336
   637
        if not learn orelse Async_Manager.has_running_threads MaShN then
blanchet@49334
   638
          ()
blanchet@49334
   639
        else if Time.toSeconds timeout >= min_secs_for_learning then
blanchet@49334
   640
          let
blanchet@49334
   641
            val factor =
blanchet@49334
   642
              if can_suggest then short_learn_timeout_factor
blanchet@49334
   643
              else long_learn_timeout_factor
blanchet@49334
   644
            val soft_timeout = time_mult factor timeout
blanchet@49334
   645
            val hard_timeout = time_mult 2.0 soft_timeout
blanchet@49334
   646
            val birth_time = Time.now ()
blanchet@49334
   647
            val death_time = Time.+ (birth_time, hard_timeout)
blanchet@49334
   648
            val desc = ("machine learner for Sledgehammer", "")
blanchet@49334
   649
          in
blanchet@49334
   650
            Async_Manager.launch MaShN birth_time death_time desc
blanchet@49334
   651
                (fn () =>
blanchet@49334
   652
                    (true, mash_learn_thy ctxt params thy soft_timeout facts))
blanchet@49334
   653
          end
blanchet@49333
   654
        else
blanchet@49333
   655
          ()
blanchet@49329
   656
      val fact_filter =
blanchet@49329
   657
        case fact_filter of
blanchet@49333
   658
          SOME ff =>
blanchet@49336
   659
          (if ff <> iterN then maybe_learn (mash_can_suggest_facts ctxt)
blanchet@49336
   660
           else (); ff)
blanchet@49333
   661
        | NONE =>
blanchet@49336
   662
          if mash_can_suggest_facts ctxt then (maybe_learn true; meshN)
blanchet@49333
   663
          else if mash_could_suggest_facts () then (maybe_learn false; iterN)
blanchet@49333
   664
          else iterN
blanchet@49303
   665
      val add_ths = Attrib.eval_thms ctxt add
blanchet@49307
   666
      fun prepend_facts ths accepts =
blanchet@49303
   667
        ((facts |> filter (member Thm.eq_thm_prop ths o snd)) @
blanchet@49307
   668
         (accepts |> filter_out (member Thm.eq_thm_prop ths o snd)))
blanchet@49308
   669
        |> take max_facts
blanchet@49329
   670
      fun iter () =
blanchet@49313
   671
        iterative_relevant_facts ctxt params prover max_facts NONE hyp_ts
blanchet@49313
   672
                                 concl_t facts
blanchet@49329
   673
      fun mash () =
blanchet@49335
   674
        mash_suggest_facts ctxt params prover max_facts hyp_ts concl_t facts
blanchet@49329
   675
      val mess =
blanchet@49335
   676
        [] |> (if fact_filter <> mashN then cons (iter (), []) else I)
blanchet@49329
   677
           |> (if fact_filter <> iterN then cons (mash ()) else I)
blanchet@49303
   678
    in
blanchet@49328
   679
      mesh_facts max_facts mess
blanchet@49303
   680
      |> not (null add_ths) ? prepend_facts add_ths
blanchet@49303
   681
    end
blanchet@49303
   682
blanchet@49334
   683
fun kill_learners () = Async_Manager.kill_threads MaShN "learner"
blanchet@49334
   684
fun running_learners () = Async_Manager.running_threads MaShN "learner"
blanchet@49334
   685
blanchet@49263
   686
end;