src/HOL/Tools/Sledgehammer/sledgehammer_filter_mash.ML
author blanchet
Wed, 18 Jul 2012 08:44:04 +0200
changeset 49334 340187063d84
parent 49333 325c8fd0d762
child 49335 891a24a48155
permissions -rw-r--r--
use async manager to manage MaSh learners to make sure they get killed cleanly
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@49327
    28
  val suggested_facts : string list -> fact list -> fact list
blanchet@49328
    29
  val mesh_facts : int -> (fact list * int option) list -> fact list
blanchet@49333
    30
  val is_likely_tautology : Proof.context -> string -> thm -> bool
blanchet@49330
    31
  val is_too_meta : thm -> bool
blanchet@49266
    32
  val theory_ord : theory * theory -> order
blanchet@49266
    33
  val thm_ord : thm * thm -> order
blanchet@49333
    34
  val features_of :
blanchet@49333
    35
    Proof.context -> string -> theory -> status -> term list -> string list
blanchet@49331
    36
  val isabelle_dependencies_of : unit Symtab.table -> thm -> string list
blanchet@49266
    37
  val goal_of_thm : theory -> thm -> thm
blanchet@49333
    38
  val run_prover :
blanchet@49333
    39
    Proof.context -> params -> string -> fact list -> thm -> prover_result
blanchet@49323
    40
  val mash_RESET : Proof.context -> unit
blanchet@49331
    41
  val mash_INIT :
blanchet@49331
    42
    Proof.context -> bool
blanchet@49331
    43
    -> (string * string list * string list * string list) list -> unit
blanchet@49323
    44
  val mash_ADD :
blanchet@49331
    45
    Proof.context -> bool
blanchet@49331
    46
    -> (string * string list * string list * string list) list -> unit
blanchet@49331
    47
  val mash_QUERY :
blanchet@49333
    48
    Proof.context -> bool -> int -> string list * string list -> string list
blanchet@49323
    49
  val mash_reset : Proof.context -> unit
blanchet@49333
    50
  val mash_could_suggest_facts : unit -> bool
blanchet@49333
    51
  val mash_can_suggest_facts : unit -> bool
blanchet@49313
    52
  val mash_suggest_facts :
blanchet@49333
    53
    Proof.context -> params -> string -> int -> term list -> term -> fact list
blanchet@49326
    54
    -> fact list
blanchet@49334
    55
  val mash_learn_thy :
blanchet@49334
    56
    Proof.context -> params -> theory -> Time.time -> fact list -> string
blanchet@49331
    57
  val mash_learn_proof :
blanchet@49331
    58
    Proof.context -> params -> term -> thm list -> fact list -> unit
blanchet@49303
    59
  val relevant_facts :
blanchet@49307
    60
    Proof.context -> params -> string -> int -> fact_override -> term list
blanchet@49311
    61
    -> term -> fact list -> fact list
blanchet@49334
    62
  val kill_learners : unit -> unit
blanchet@49334
    63
  val running_learners : unit -> unit
blanchet@49263
    64
end;
blanchet@49263
    65
blanchet@49263
    66
structure Sledgehammer_Filter_MaSh : SLEDGEHAMMER_FILTER_MASH =
blanchet@49263
    67
struct
blanchet@49264
    68
blanchet@49266
    69
open ATP_Util
blanchet@49266
    70
open ATP_Problem_Generate
blanchet@49266
    71
open Sledgehammer_Util
blanchet@49266
    72
open Sledgehammer_Fact
blanchet@49266
    73
open Sledgehammer_Filter_Iter
blanchet@49266
    74
open Sledgehammer_Provers
blanchet@49333
    75
open Sledgehammer_Minimize
blanchet@49266
    76
blanchet@49323
    77
val trace =
blanchet@49323
    78
  Attrib.setup_config_bool @{binding sledgehammer_filter_mash_trace} (K false)
blanchet@49323
    79
fun trace_msg ctxt msg = if Config.get ctxt trace then tracing (msg ()) else ()
blanchet@49323
    80
blanchet@49334
    81
val MaShN = "MaSh"
blanchet@49334
    82
blanchet@49329
    83
val meshN = "mesh"
blanchet@49329
    84
val iterN = "iter"
blanchet@49329
    85
val mashN = "mash"
blanchet@49329
    86
blanchet@49329
    87
val fact_filters = [meshN, iterN, mashN]
blanchet@49329
    88
blanchet@49329
    89
fun mash_home () = getenv "MASH_HOME"
blanchet@49329
    90
fun mash_state_dir () =
blanchet@49324
    91
  getenv "ISABELLE_HOME_USER" ^ "/mash"
blanchet@49331
    92
  |> tap (Isabelle_System.mkdir o Path.explode)
blanchet@49329
    93
fun mash_state_path () = mash_state_dir () ^ "/state" |> Path.explode
blanchet@49266
    94
blanchet@49266
    95
(*** Isabelle helpers ***)
blanchet@49266
    96
blanchet@49323
    97
fun meta_char c =
blanchet@49266
    98
  if Char.isAlphaNum c orelse c = #"_" orelse c = #"." orelse c = #"(" orelse
blanchet@49266
    99
     c = #")" orelse c = #"," then
blanchet@49266
   100
    String.str c
blanchet@49266
   101
  else
blanchet@49266
   102
    (* fixed width, in case more digits follow *)
blanchet@49266
   103
    "\\" ^ stringN_of_int 3 (Char.ord c)
blanchet@49266
   104
blanchet@49323
   105
fun unmeta_chars accum [] = String.implode (rev accum)
blanchet@49323
   106
  | unmeta_chars accum (#"\\" :: d1 :: d2 :: d3 :: cs) =
blanchet@49323
   107
    (case Int.fromString (String.implode [d1, d2, d3]) of
blanchet@49323
   108
       SOME n => unmeta_chars (Char.chr n :: accum) cs
blanchet@49323
   109
     | NONE => "" (* error *))
blanchet@49323
   110
  | unmeta_chars _ (#"\\" :: _) = "" (* error *)
blanchet@49323
   111
  | unmeta_chars accum (c :: cs) = unmeta_chars (c :: accum) cs
blanchet@49323
   112
blanchet@49323
   113
val escape_meta = String.translate meta_char
blanchet@49318
   114
val escape_metas = map escape_meta #> space_implode " "
blanchet@49330
   115
val unescape_meta = String.explode #> unmeta_chars []
blanchet@49330
   116
val unescape_metas =
blanchet@49330
   117
  space_explode " " #> filter_out (curry (op =) "") #> map unescape_meta
blanchet@49266
   118
blanchet@49326
   119
fun extract_query line =
blanchet@49326
   120
  case space_explode ":" line of
blanchet@49330
   121
    [goal_name, suggs] => (unescape_meta goal_name, unescape_metas suggs)
blanchet@49327
   122
  | _ => ("", [])
blanchet@49326
   123
blanchet@49326
   124
fun find_suggested facts sugg =
blanchet@49326
   125
  find_first (fn (_, th) => Thm.get_name_hint th = sugg) facts
blanchet@49326
   126
fun suggested_facts suggs facts = map_filter (find_suggested facts) suggs
blanchet@49326
   127
blanchet@49328
   128
fun sum_avg n xs =
blanchet@49328
   129
  fold (Integer.add o Integer.mult n) xs 0 div (length xs)
blanchet@49328
   130
blanchet@49329
   131
fun mesh_facts max_facts [(facts, _)] = facts |> take max_facts
blanchet@49329
   132
  | mesh_facts max_facts mess =
blanchet@49329
   133
    let
blanchet@49329
   134
      val n = length mess
blanchet@49329
   135
      val fact_eq = Thm.eq_thm o pairself snd
blanchet@49329
   136
      fun score_in fact (facts, def) =
blanchet@49329
   137
        case find_index (curry fact_eq fact) facts of
blanchet@49329
   138
          ~1 => def
blanchet@49329
   139
        | j => SOME j
blanchet@49329
   140
      fun score_of fact = mess |> map_filter (score_in fact) |> sum_avg n
blanchet@49329
   141
      val facts = fold (union fact_eq o take max_facts o fst) mess []
blanchet@49329
   142
    in
blanchet@49329
   143
      facts |> map (`score_of) |> sort (int_ord o pairself fst) |> map snd
blanchet@49329
   144
            |> take max_facts
blanchet@49329
   145
    end
blanchet@49327
   146
blanchet@49318
   147
val thy_feature_prefix = "y_"
blanchet@49266
   148
blanchet@49318
   149
val thy_feature_name_of = prefix thy_feature_prefix
blanchet@49318
   150
val const_name_of = prefix const_prefix
blanchet@49318
   151
val type_name_of = prefix type_const_prefix
blanchet@49318
   152
val class_name_of = prefix class_prefix
blanchet@49266
   153
blanchet@49266
   154
local
blanchet@49266
   155
blanchet@49266
   156
fun has_bool @{typ bool} = true
blanchet@49266
   157
  | has_bool (Type (_, Ts)) = exists has_bool Ts
blanchet@49266
   158
  | has_bool _ = false
blanchet@49266
   159
blanchet@49266
   160
fun has_fun (Type (@{type_name fun}, _)) = true
blanchet@49266
   161
  | has_fun (Type (_, Ts)) = exists has_fun Ts
blanchet@49266
   162
  | has_fun _ = false
blanchet@49266
   163
blanchet@49266
   164
val is_conn = member (op =)
blanchet@49266
   165
  [@{const_name Trueprop}, @{const_name HOL.conj}, @{const_name HOL.disj},
blanchet@49266
   166
   @{const_name HOL.implies}, @{const_name Not},
blanchet@49266
   167
   @{const_name All}, @{const_name Ex}, @{const_name Ball}, @{const_name Bex},
blanchet@49266
   168
   @{const_name HOL.eq}]
blanchet@49266
   169
blanchet@49266
   170
val has_bool_arg_const =
blanchet@49266
   171
  exists_Const (fn (c, T) =>
blanchet@49266
   172
                   not (is_conn c) andalso exists has_bool (binder_types T))
blanchet@49266
   173
blanchet@49331
   174
fun higher_inst_const thy (s, T) =
blanchet@49266
   175
  case binder_types T of
blanchet@49266
   176
    [] => false
blanchet@49331
   177
  | Ts => length (binder_types (Sign.the_const_type thy s)) <> length Ts
blanchet@49331
   178
  handle TYPE _ => false
blanchet@49266
   179
blanchet@49266
   180
val binders = [@{const_name All}, @{const_name Ex}]
blanchet@49266
   181
blanchet@49266
   182
in
blanchet@49266
   183
blanchet@49266
   184
fun is_fo_term thy t =
blanchet@49266
   185
  let
blanchet@49266
   186
    val t =
blanchet@49266
   187
      t |> Envir.beta_eta_contract
blanchet@49266
   188
        |> transform_elim_prop
blanchet@49266
   189
        |> Object_Logic.atomize_term thy
blanchet@49266
   190
  in
blanchet@49266
   191
    Term.is_first_order binders t andalso
blanchet@49266
   192
    not (exists_subterm (fn Var (_, T) => has_bool T orelse has_fun T
blanchet@49266
   193
                          | _ => false) t orelse
blanchet@49266
   194
         has_bool_arg_const t orelse exists_Const (higher_inst_const thy) t)
blanchet@49266
   195
  end
blanchet@49266
   196
blanchet@49266
   197
end
blanchet@49266
   198
blanchet@49333
   199
fun interesting_terms_types_and_classes ctxt prover term_max_depth
blanchet@49333
   200
                                        type_max_depth ts =
blanchet@49266
   201
  let
blanchet@49266
   202
    val bad_types = [@{type_name prop}, @{type_name bool}, @{type_name fun}]
blanchet@49333
   203
    fun is_bad_const (x as (s, _)) args =
blanchet@49333
   204
      member (op =) atp_logical_consts s orelse
blanchet@49333
   205
      fst (is_built_in_const_for_prover ctxt prover x args)
blanchet@49319
   206
    fun add_classes @{sort type} = I
blanchet@49319
   207
      | add_classes S = union (op =) (map class_name_of S)
blanchet@49266
   208
    fun do_add_type (Type (s, Ts)) =
blanchet@49266
   209
        (not (member (op =) bad_types s) ? insert (op =) (type_name_of s))
blanchet@49266
   210
        #> fold do_add_type Ts
blanchet@49319
   211
      | do_add_type (TFree (_, S)) = add_classes S
blanchet@49319
   212
      | do_add_type (TVar (_, S)) = add_classes S
blanchet@49266
   213
    fun add_type T = type_max_depth >= 0 ? do_add_type T
blanchet@49266
   214
    fun mk_app s args =
blanchet@49266
   215
      if member (op <>) args "" then s ^ "(" ^ space_implode "," args ^ ")"
blanchet@49266
   216
      else s
blanchet@49266
   217
    fun patternify ~1 _ = ""
blanchet@49266
   218
      | patternify depth t =
blanchet@49266
   219
        case strip_comb t of
blanchet@49266
   220
          (Const (s, _), args) =>
blanchet@49266
   221
          mk_app (const_name_of s) (map (patternify (depth - 1)) args)
blanchet@49266
   222
        | _ => ""
blanchet@49266
   223
    fun add_term_patterns ~1 _ = I
blanchet@49266
   224
      | add_term_patterns depth t =
blanchet@49266
   225
        insert (op =) (patternify depth t)
blanchet@49266
   226
        #> add_term_patterns (depth - 1) t
blanchet@49266
   227
    val add_term = add_term_patterns term_max_depth
blanchet@49266
   228
    fun add_patterns t =
blanchet@49266
   229
      let val (head, args) = strip_comb t in
blanchet@49266
   230
        (case head of
blanchet@49333
   231
           Const (x as (_, T)) =>
blanchet@49333
   232
           not (is_bad_const x args) ? (add_term t #> add_type T)
blanchet@49266
   233
         | Free (_, T) => add_type T
blanchet@49266
   234
         | Var (_, T) => add_type T
blanchet@49266
   235
         | Abs (_, T, body) => add_type T #> add_patterns body
blanchet@49266
   236
         | _ => I)
blanchet@49266
   237
        #> fold add_patterns args
blanchet@49266
   238
      end
blanchet@49317
   239
  in [] |> fold add_patterns ts |> sort string_ord end
blanchet@49266
   240
blanchet@49333
   241
fun is_likely_tautology ctxt prover th =
blanchet@49333
   242
  null (interesting_terms_types_and_classes ctxt prover 0 ~1 [prop_of th])
blanchet@49333
   243
  andalso not (Thm.eq_thm_prop (@{thm ext}, th))
blanchet@49266
   244
blanchet@49330
   245
(* ### FIXME: optimize *)
blanchet@49330
   246
fun is_too_meta th =
blanchet@49330
   247
  fastype_of (Object_Logic.atomize_term (theory_of_thm th) (prop_of th))
blanchet@49330
   248
  <> @{typ bool}
blanchet@49266
   249
blanchet@49266
   250
fun theory_ord p =
blanchet@49330
   251
  if Theory.eq_thy p then
blanchet@49330
   252
    EQUAL
blanchet@49330
   253
  else if Theory.subthy p then
blanchet@49330
   254
    LESS
blanchet@49330
   255
  else if Theory.subthy (swap p) then
blanchet@49330
   256
    GREATER
blanchet@49330
   257
  else case int_ord (pairself (length o Theory.ancestors_of) p) of
blanchet@49330
   258
    EQUAL => string_ord (pairself Context.theory_name p)
blanchet@49330
   259
  | order => order
blanchet@49266
   260
blanchet@49266
   261
val thm_ord = theory_ord o pairself theory_of_thm
blanchet@49266
   262
blanchet@49266
   263
fun is_exists (s, _) = (s = @{const_name Ex} orelse s = @{const_name Ex1})
blanchet@49266
   264
blanchet@49312
   265
val term_max_depth = 1
blanchet@49312
   266
val type_max_depth = 1
blanchet@49266
   267
blanchet@49266
   268
(* TODO: Generate type classes for types? *)
blanchet@49333
   269
fun features_of ctxt prover thy status ts =
blanchet@49318
   270
  thy_feature_name_of (Context.theory_name thy) ::
blanchet@49333
   271
  interesting_terms_types_and_classes ctxt prover term_max_depth type_max_depth
blanchet@49333
   272
                                      ts
blanchet@49317
   273
  |> exists (not o is_lambda_free) ts ? cons "lambdas"
blanchet@49317
   274
  |> exists (exists_Const is_exists) ts ? cons "skolems"
blanchet@49317
   275
  |> exists (not o is_fo_term thy) ts ? cons "ho"
blanchet@49317
   276
  |> (case status of
blanchet@49317
   277
        General => I
blanchet@49317
   278
      | Induction => cons "induction"
blanchet@49317
   279
      | Intro => cons "intro"
blanchet@49317
   280
      | Inductive => cons "inductive"
blanchet@49317
   281
      | Elim => cons "elim"
blanchet@49317
   282
      | Simp => cons "simp"
blanchet@49317
   283
      | Def => cons "def")
blanchet@49266
   284
blanchet@49266
   285
fun isabelle_dependencies_of all_facts =
blanchet@49318
   286
  thms_in_proof (SOME all_facts) #> sort string_ord
blanchet@49266
   287
blanchet@49266
   288
val freezeT = Type.legacy_freeze_type
blanchet@49266
   289
blanchet@49266
   290
fun freeze (t $ u) = freeze t $ freeze u
blanchet@49266
   291
  | freeze (Abs (s, T, t)) = Abs (s, freezeT T, freeze t)
blanchet@49266
   292
  | freeze (Var ((s, _), T)) = Free (s, freezeT T)
blanchet@49266
   293
  | freeze (Const (s, T)) = Const (s, freezeT T)
blanchet@49266
   294
  | freeze (Free (s, T)) = Free (s, freezeT T)
blanchet@49266
   295
  | freeze t = t
blanchet@49266
   296
blanchet@49266
   297
fun goal_of_thm thy = prop_of #> freeze #> cterm_of thy #> Goal.init
blanchet@49266
   298
blanchet@49333
   299
fun run_prover ctxt params prover facts goal =
blanchet@49266
   300
  let
blanchet@49266
   301
    val problem =
blanchet@49266
   302
      {state = Proof.init ctxt, goal = goal, subgoal = 1, subgoal_count = 1,
blanchet@49304
   303
       facts = facts |> map (apfst (apfst (fn name => name ())))
blanchet@49333
   304
                     |> map Untranslated_Fact}
blanchet@49333
   305
    val prover = get_minimizing_prover ctxt Normal prover
blanchet@49266
   306
  in prover params (K (K (K ""))) problem end
blanchet@49266
   307
blanchet@49266
   308
blanchet@49317
   309
(*** Low-level communication with MaSh ***)
blanchet@49317
   310
blanchet@49333
   311
fun write_file write file =
blanchet@49333
   312
  let val path = Path.explode file in
blanchet@49333
   313
    File.write path ""; write (File.append path)
blanchet@49333
   314
  end
blanchet@49331
   315
blanchet@49331
   316
fun mash_info overlord =
blanchet@49331
   317
  if overlord then (getenv "ISABELLE_HOME_USER", "")
blanchet@49331
   318
  else (getenv "ISABELLE_TMP", serial_string ())
blanchet@49331
   319
blanchet@49331
   320
fun run_mash ctxt (temp_dir, serial) core =
blanchet@49326
   321
  let
blanchet@49331
   322
    val log_file = temp_dir ^ "/mash_log" ^ serial
blanchet@49331
   323
    val err_file = temp_dir ^ "/mash_err" ^ serial
blanchet@49326
   324
    val command =
blanchet@49331
   325
      mash_home () ^ "/mash.py --quiet --outputDir " ^ mash_state_dir () ^
blanchet@49331
   326
      " --log " ^ log_file ^ " " ^ core ^ " 2>&1 > " ^ err_file
blanchet@49331
   327
  in
blanchet@49331
   328
    trace_msg ctxt (fn () => "Running " ^ command);
blanchet@49333
   329
    write_file (K ()) log_file;
blanchet@49333
   330
    write_file (K ()) err_file;
blanchet@49331
   331
    Isabelle_System.bash command; ()
blanchet@49331
   332
  end
blanchet@49326
   333
blanchet@49331
   334
fun run_mash_init ctxt overlord write_access write_feats write_deps =
blanchet@49331
   335
  let
blanchet@49331
   336
    val info as (temp_dir, serial) = mash_info overlord
blanchet@49331
   337
    val in_dir = temp_dir ^ "/mash_init" ^ serial
blanchet@49331
   338
                 |> tap (Isabelle_System.mkdir o Path.explode)
blanchet@49331
   339
  in
blanchet@49331
   340
    write_file write_access (in_dir ^ "/mash_accessibility");
blanchet@49331
   341
    write_file write_feats (in_dir ^ "/mash_features");
blanchet@49331
   342
    write_file write_deps (in_dir ^ "/mash_dependencies");
blanchet@49331
   343
    run_mash ctxt info ("--init --inputDir " ^ in_dir)
blanchet@49331
   344
  end
blanchet@49331
   345
blanchet@49333
   346
fun run_mash_commands ctxt overlord save max_suggs write_cmds read_suggs =
blanchet@49331
   347
  let
blanchet@49331
   348
    val info as (temp_dir, serial) = mash_info overlord
blanchet@49333
   349
    val sugg_file = temp_dir ^ "/mash_suggs" ^ serial
blanchet@49331
   350
    val cmd_file = temp_dir ^ "/mash_commands" ^ serial
blanchet@49331
   351
  in
blanchet@49333
   352
    write_file (K ()) sugg_file;
blanchet@49331
   353
    write_file write_cmds cmd_file;
blanchet@49331
   354
    run_mash ctxt info
blanchet@49333
   355
             ("--inputFile " ^ cmd_file ^ " --predictions " ^ sugg_file ^
blanchet@49333
   356
              " --numberOfPredictions " ^ string_of_int max_suggs ^
blanchet@49331
   357
              (if save then " --saveModel" else ""));
blanchet@49333
   358
    read_suggs (fn () => File.read_lines (Path.explode sugg_file))
blanchet@49331
   359
  end
blanchet@49331
   360
blanchet@49331
   361
fun str_of_update (name, parents, feats, deps) =
blanchet@49331
   362
  "! " ^ escape_meta name ^ ": " ^ escape_metas parents ^ "; " ^
blanchet@49326
   363
  escape_metas feats ^ "; " ^ escape_metas deps ^ "\n"
blanchet@49326
   364
blanchet@49331
   365
fun str_of_query (parents, feats) =
blanchet@49331
   366
  "? " ^ escape_metas parents ^ "; " ^ escape_metas feats
blanchet@49331
   367
blanchet@49331
   368
fun init_str_of_update get (upd as (name, _, _, _)) =
blanchet@49331
   369
  escape_meta name ^ ": " ^ escape_metas (get upd) ^ "\n"
blanchet@49326
   370
blanchet@49323
   371
fun mash_RESET ctxt =
blanchet@49329
   372
  let val path = mash_state_dir () |> Path.explode in
blanchet@49324
   373
    trace_msg ctxt (K "MaSh RESET");
blanchet@49324
   374
    File.fold_dir (fn file => fn () =>
blanchet@49324
   375
                      File.rm (Path.append path (Path.basic file)))
blanchet@49324
   376
                  path ()
blanchet@49324
   377
  end
blanchet@49317
   378
blanchet@49331
   379
fun mash_INIT ctxt _ [] = mash_RESET ctxt
blanchet@49331
   380
  | mash_INIT ctxt overlord upds =
blanchet@49331
   381
    (trace_msg ctxt (fn () => "MaSh INIT " ^
blanchet@49331
   382
         elide_string 1000 (space_implode " " (map #1 upds)));
blanchet@49331
   383
     run_mash_init ctxt overlord
blanchet@49331
   384
         (fn append => List.app (append o init_str_of_update #2) upds)
blanchet@49331
   385
         (fn append => List.app (append o init_str_of_update #3) upds)
blanchet@49331
   386
         (fn append => List.app (append o init_str_of_update #4) upds))
blanchet@49317
   387
blanchet@49331
   388
fun mash_ADD _ _ [] = ()
blanchet@49331
   389
  | mash_ADD ctxt overlord upds =
blanchet@49331
   390
    (trace_msg ctxt (fn () => "MaSh ADD " ^
blanchet@49331
   391
         elide_string 1000 (space_implode " " (map #1 upds)));
blanchet@49333
   392
     run_mash_commands ctxt overlord true 0
blanchet@49331
   393
         (fn append => List.app (append o str_of_update) upds) (K ()))
blanchet@49317
   394
blanchet@49333
   395
fun mash_QUERY ctxt overlord max_suggs (query as (_, feats)) =
blanchet@49329
   396
  (trace_msg ctxt (fn () => "MaSh QUERY " ^ space_implode " " feats);
blanchet@49333
   397
   run_mash_commands ctxt overlord false max_suggs
blanchet@49331
   398
       (fn append => append (str_of_query query))
blanchet@49333
   399
       (fn suggs => snd (extract_query (List.last (suggs ()))))
blanchet@49326
   400
   handle List.Empty => [])
blanchet@49317
   401
blanchet@49317
   402
blanchet@49266
   403
(*** High-level communication with MaSh ***)
blanchet@49266
   404
blanchet@49316
   405
type mash_state =
blanchet@49331
   406
  {thys : bool Symtab.table,
blanchet@49331
   407
   fact_graph : unit Graph.T}
blanchet@49264
   408
blanchet@49331
   409
val empty_state = {thys = Symtab.empty, fact_graph = Graph.empty}
blanchet@49316
   410
blanchet@49316
   411
local
blanchet@49316
   412
blanchet@49316
   413
fun mash_load (state as (true, _)) = state
blanchet@49316
   414
  | mash_load _ =
blanchet@49324
   415
    let val path = mash_state_path () in
blanchet@49317
   416
      (true,
blanchet@49317
   417
       case try File.read_lines path of
blanchet@49331
   418
         SOME (comp_thys :: incomp_thys :: fact_lines) =>
blanchet@49317
   419
         let
blanchet@49331
   420
           fun add_thy comp thy = Symtab.update (thy, comp)
blanchet@49331
   421
           fun add_fact_line line =
blanchet@49331
   422
             case extract_query line of
blanchet@49331
   423
               ("", _) => I (* shouldn't happen *)
blanchet@49331
   424
             | (name, parents) =>
blanchet@49331
   425
               Graph.default_node (name, ())
blanchet@49331
   426
               #> fold (fn par => Graph.add_edge (par, name)) parents
blanchet@49331
   427
           val thys =
blanchet@49331
   428
             Symtab.empty |> fold (add_thy true) (unescape_metas comp_thys)
blanchet@49331
   429
                          |> fold (add_thy false) (unescape_metas incomp_thys)
blanchet@49331
   430
           val fact_graph = Graph.empty |> fold add_fact_line fact_lines
blanchet@49331
   431
         in {thys = thys, fact_graph = fact_graph} end
blanchet@49319
   432
       | _ => empty_state)
blanchet@49317
   433
    end
blanchet@49316
   434
blanchet@49331
   435
fun mash_save ({thys, fact_graph, ...} : mash_state) =
blanchet@49316
   436
  let
blanchet@49324
   437
    val path = mash_state_path ()
blanchet@49331
   438
    val thys = Symtab.dest thys
blanchet@49333
   439
    val line_for_thys = escape_metas o AList.find (op =) thys
blanchet@49333
   440
    fun fact_line_for name parents =
blanchet@49333
   441
      escape_meta name ^ ": " ^ escape_metas parents
blanchet@49331
   442
    val append_fact = File.append path o suffix "\n" oo fact_line_for
blanchet@49316
   443
  in
blanchet@49331
   444
    File.write path (line_for_thys true ^ "\n" ^ line_for_thys false ^ "\n");
blanchet@49331
   445
    Graph.fold (fn (name, ((), (parents, _))) => fn () =>
blanchet@49331
   446
                   append_fact name (Graph.Keys.dest parents))
blanchet@49331
   447
        fact_graph ()
blanchet@49316
   448
  end
blanchet@49316
   449
blanchet@49317
   450
val global_state =
blanchet@49319
   451
  Synchronized.var "Sledgehammer_Filter_MaSh.global_state" (false, empty_state)
blanchet@49316
   452
blanchet@49316
   453
in
blanchet@49316
   454
blanchet@49321
   455
fun mash_map f =
blanchet@49317
   456
  Synchronized.change global_state (mash_load ##> (f #> tap mash_save))
blanchet@49316
   457
blanchet@49319
   458
fun mash_get () = Synchronized.change_result global_state (mash_load #> `snd)
blanchet@49317
   459
blanchet@49323
   460
fun mash_reset ctxt =
blanchet@49317
   461
  Synchronized.change global_state (fn _ =>
blanchet@49324
   462
      (mash_RESET ctxt; File.write (mash_state_path ()) "";
blanchet@49323
   463
       (true, empty_state)))
blanchet@49316
   464
blanchet@49316
   465
end
blanchet@49316
   466
blanchet@49333
   467
fun mash_could_suggest_facts () = mash_home () <> ""
blanchet@49333
   468
fun mash_can_suggest_facts () = not (Graph.is_empty (#fact_graph (mash_get ())))
blanchet@49264
   469
blanchet@49333
   470
fun parents_wrt_facts facts fact_graph =
blanchet@49331
   471
  let
blanchet@49333
   472
    val graph_facts = Symtab.make (map (rpair ()) (Graph.keys fact_graph))
blanchet@49331
   473
    val facts =
blanchet@49333
   474
      [] |> fold (cons o Thm.get_name_hint o snd) facts
blanchet@49333
   475
         |> filter (Symtab.defined graph_facts)
blanchet@49333
   476
         |> Graph.all_preds fact_graph
blanchet@49333
   477
    val facts = Symtab.empty |> fold (fn name => Symtab.update (name, ())) facts
blanchet@49331
   478
  in fact_graph |> Graph.restrict (Symtab.defined facts) |> Graph.maximals end
blanchet@49331
   479
blanchet@49333
   480
(* Generate more suggestions than requested, because some might be thrown out
blanchet@49333
   481
   later for various reasons and "meshing" gives better results with some
blanchet@49333
   482
   slack. *)
blanchet@49333
   483
fun max_suggs_of max_facts = max_facts + Int.min (200, max_facts)
blanchet@49333
   484
blanchet@49333
   485
fun mash_suggest_facts ctxt ({overlord, ...} : params) prover max_facts hyp_ts
blanchet@49333
   486
                       concl_t facts =
blanchet@49316
   487
  let
blanchet@49317
   488
    val thy = Proof_Context.theory_of ctxt
blanchet@49331
   489
    val fact_graph = #fact_graph (mash_get ())
blanchet@49333
   490
    val parents = parents_wrt_facts facts fact_graph
blanchet@49333
   491
    val feats = features_of ctxt prover thy General (concl_t :: hyp_ts)
blanchet@49333
   492
    val suggs =
blanchet@49333
   493
      mash_QUERY ctxt overlord (max_suggs_of max_facts) (parents, feats)
blanchet@49326
   494
  in suggested_facts suggs facts end
blanchet@49264
   495
blanchet@49331
   496
fun add_thys_for thy =
blanchet@49331
   497
  let fun add comp thy = Symtab.update (Context.theory_name thy, comp) in
blanchet@49331
   498
    add false thy #> fold (add true) (Theory.ancestors_of thy)
blanchet@49331
   499
  end
blanchet@49319
   500
blanchet@49331
   501
fun update_fact_graph ctxt (name, parents, feats, deps) (upds, graph) =
blanchet@49331
   502
  let
blanchet@49331
   503
    fun maybe_add_from from (accum as (parents, graph)) =
blanchet@49331
   504
      (from :: parents, Graph.add_edge_acyclic (from, name) graph)
blanchet@49331
   505
      handle Graph.CYCLES _ =>
blanchet@49331
   506
             (trace_msg ctxt (fn () =>
blanchet@49331
   507
                  "Cycle between " ^ quote from ^ " and " ^ quote name); accum)
blanchet@49331
   508
           | Graph.UNDEF _ =>
blanchet@49331
   509
             (trace_msg ctxt (fn () => "Unknown node " ^ quote from); accum)
blanchet@49331
   510
    val graph = graph |> Graph.new_node (name, ())
blanchet@49331
   511
    val (parents, graph) = ([], graph) |> fold maybe_add_from parents
blanchet@49331
   512
    val (deps, graph) = ([], graph) |> fold maybe_add_from deps
blanchet@49331
   513
  in ((name, parents, feats, deps) :: upds, graph) end
blanchet@49321
   514
blanchet@49333
   515
val pass1_learn_timeout_factor = 0.5
blanchet@49333
   516
blanchet@49333
   517
(* The timeout is understood in a very slack fashion. *)
blanchet@49334
   518
fun mash_learn_thy ctxt ({provers, verbose, overlord, ...} : params) thy timeout
blanchet@49334
   519
                   facts =
blanchet@49319
   520
  let
blanchet@49333
   521
    val timer = Timer.startRealTimer ()
blanchet@49333
   522
    val prover = hd provers
blanchet@49333
   523
    fun timed_out frac =
blanchet@49333
   524
      Time.> (Timer.checkRealTimer timer, time_mult frac timeout)
blanchet@49331
   525
    val {fact_graph, ...} = mash_get ()
blanchet@49331
   526
    fun is_old (_, th) = can (Graph.get_node fact_graph) (Thm.get_name_hint th)
blanchet@49331
   527
    val new_facts = facts |> filter_out is_old |> sort (thm_ord o pairself snd)
blanchet@49323
   528
  in
blanchet@49323
   529
    if null new_facts then
blanchet@49334
   530
      ""
blanchet@49323
   531
    else
blanchet@49319
   532
      let
blanchet@49323
   533
        val ths = facts |> map snd
blanchet@49330
   534
        val all_names =
blanchet@49333
   535
          ths |> filter_out (is_likely_tautology ctxt prover orf is_too_meta)
blanchet@49331
   536
              |> map (rpair () o Thm.get_name_hint)
blanchet@49331
   537
              |> Symtab.make
blanchet@49333
   538
        fun do_fact _ (accum as (_, true)) = accum
blanchet@49333
   539
          | do_fact ((_, (_, status)), th) ((parents, upds), false) =
blanchet@49333
   540
            let
blanchet@49333
   541
              val name = Thm.get_name_hint th
blanchet@49333
   542
              val feats = features_of ctxt prover thy status [prop_of th]
blanchet@49333
   543
              val deps = isabelle_dependencies_of all_names th
blanchet@49333
   544
              val upd = (name, parents, feats, deps)
blanchet@49333
   545
            in (([name], upd :: upds), timed_out pass1_learn_timeout_factor) end
blanchet@49333
   546
        val parents = parents_wrt_facts facts fact_graph
blanchet@49333
   547
        val ((_, upds), _) =
blanchet@49333
   548
          ((parents, []), false) |> fold do_fact new_facts |>> apsnd rev
blanchet@49333
   549
        val n = length upds
blanchet@49331
   550
        fun trans {thys, fact_graph} =
blanchet@49331
   551
          let
blanchet@49331
   552
            val mash_INIT_or_ADD =
blanchet@49331
   553
              if Graph.is_empty fact_graph then mash_INIT else mash_ADD
blanchet@49331
   554
            val (upds, fact_graph) =
blanchet@49331
   555
              ([], fact_graph) |> fold (update_fact_graph ctxt) upds
blanchet@49331
   556
          in
blanchet@49331
   557
            (mash_INIT_or_ADD ctxt overlord (rev upds);
blanchet@49331
   558
             {thys = thys |> add_thys_for thy,
blanchet@49331
   559
              fact_graph = fact_graph})
blanchet@49331
   560
          end
blanchet@49333
   561
      in
blanchet@49334
   562
        mash_map trans;
blanchet@49334
   563
        if verbose then
blanchet@49334
   564
          "Processed " ^ string_of_int n ^ " proof" ^ plural_s n ^
blanchet@49334
   565
          (if verbose then
blanchet@49334
   566
             " in " ^ string_from_time (Timer.checkRealTimer timer)
blanchet@49334
   567
           else
blanchet@49334
   568
             "") ^ "."
blanchet@49334
   569
        else
blanchet@49334
   570
          ""
blanchet@49333
   571
      end
blanchet@49323
   572
  end
blanchet@49319
   573
blanchet@49333
   574
fun mash_learn_proof ctxt ({provers, overlord, ...} : params) t used_ths facts =
blanchet@49331
   575
  let
blanchet@49331
   576
    val thy = Proof_Context.theory_of ctxt
blanchet@49333
   577
    val prover = hd provers
blanchet@49333
   578
    val name = ATP_Util.timestamp () ^ serial_string () (* fresh enough *)
blanchet@49333
   579
    val feats = features_of ctxt prover thy General [t]
blanchet@49331
   580
    val deps = used_ths |> map Thm.get_name_hint
blanchet@49331
   581
  in
blanchet@49331
   582
    mash_map (fn {thys, fact_graph} =>
blanchet@49324
   583
        let
blanchet@49333
   584
          val parents = parents_wrt_facts facts fact_graph
blanchet@49331
   585
          val upds = [(name, parents, feats, deps)]
blanchet@49331
   586
          val (upds, fact_graph) =
blanchet@49331
   587
            ([], fact_graph) |> fold (update_fact_graph ctxt) upds
blanchet@49324
   588
        in
blanchet@49331
   589
          mash_ADD ctxt overlord upds;
blanchet@49331
   590
          {thys = thys, fact_graph = fact_graph}
blanchet@49331
   591
        end)
blanchet@49331
   592
  end
blanchet@49264
   593
blanchet@49333
   594
(* The threshold should be large enough so that MaSh doesn't kick in for Auto
blanchet@49333
   595
   Sledgehammer and Try. *)
blanchet@49333
   596
val min_secs_for_learning = 15
blanchet@49333
   597
val short_learn_timeout_factor = 0.2
blanchet@49333
   598
val long_learn_timeout_factor = 4.0
blanchet@49333
   599
blanchet@49333
   600
fun relevant_facts ctxt (params as {fact_filter, timeout, ...}) prover max_facts
blanchet@49313
   601
        ({add, only, ...} : fact_override) hyp_ts concl_t facts =
blanchet@49329
   602
  if not (subset (op =) (the_list fact_filter, fact_filters)) then
blanchet@49329
   603
    error ("Unknown fact filter: " ^ quote (the fact_filter) ^ ".")
blanchet@49329
   604
  else if only then
blanchet@49304
   605
    facts
blanchet@49308
   606
  else if max_facts <= 0 then
blanchet@49303
   607
    []
blanchet@49303
   608
  else
blanchet@49303
   609
    let
blanchet@49333
   610
      val thy = Proof_Context.theory_of ctxt
blanchet@49333
   611
      fun maybe_learn can_suggest =
blanchet@49334
   612
        if Async_Manager.has_running_threads MaShN orelse null facts then
blanchet@49334
   613
          ()
blanchet@49334
   614
        else if Time.toSeconds timeout >= min_secs_for_learning then
blanchet@49334
   615
          let
blanchet@49334
   616
            val factor =
blanchet@49334
   617
              if can_suggest then short_learn_timeout_factor
blanchet@49334
   618
              else long_learn_timeout_factor
blanchet@49334
   619
            val soft_timeout = time_mult factor timeout
blanchet@49334
   620
            val hard_timeout = time_mult 2.0 soft_timeout
blanchet@49334
   621
            val birth_time = Time.now ()
blanchet@49334
   622
            val death_time = Time.+ (birth_time, hard_timeout)
blanchet@49334
   623
            val desc = ("machine learner for Sledgehammer", "")
blanchet@49334
   624
          in
blanchet@49334
   625
            Async_Manager.launch MaShN birth_time death_time desc
blanchet@49334
   626
                (fn () =>
blanchet@49334
   627
                    (true, mash_learn_thy ctxt params thy soft_timeout facts))
blanchet@49334
   628
          end
blanchet@49333
   629
        else
blanchet@49333
   630
          ()
blanchet@49329
   631
      val fact_filter =
blanchet@49329
   632
        case fact_filter of
blanchet@49333
   633
          SOME ff =>
blanchet@49333
   634
          (if ff <> iterN then maybe_learn (mash_can_suggest_facts ()) else ();
blanchet@49333
   635
           ff)
blanchet@49333
   636
        | NONE =>
blanchet@49333
   637
          if mash_can_suggest_facts () then (maybe_learn true; meshN)
blanchet@49333
   638
          else if mash_could_suggest_facts () then (maybe_learn false; iterN)
blanchet@49333
   639
          else iterN
blanchet@49303
   640
      val add_ths = Attrib.eval_thms ctxt add
blanchet@49307
   641
      fun prepend_facts ths accepts =
blanchet@49303
   642
        ((facts |> filter (member Thm.eq_thm_prop ths o snd)) @
blanchet@49307
   643
         (accepts |> filter_out (member Thm.eq_thm_prop ths o snd)))
blanchet@49308
   644
        |> take max_facts
blanchet@49329
   645
      fun iter () =
blanchet@49313
   646
        iterative_relevant_facts ctxt params prover max_facts NONE hyp_ts
blanchet@49313
   647
                                 concl_t facts
blanchet@49329
   648
        |> (fn facts => (facts, SOME (length facts)))
blanchet@49329
   649
      fun mash () =
blanchet@49333
   650
        (mash_suggest_facts ctxt params prover max_facts hyp_ts concl_t facts,
blanchet@49333
   651
         NONE)
blanchet@49329
   652
      val mess =
blanchet@49329
   653
        [] |> (if fact_filter <> mashN then cons (iter ()) else I)
blanchet@49329
   654
           |> (if fact_filter <> iterN then cons (mash ()) else I)
blanchet@49303
   655
    in
blanchet@49328
   656
      mesh_facts max_facts mess
blanchet@49303
   657
      |> not (null add_ths) ? prepend_facts add_ths
blanchet@49303
   658
    end
blanchet@49303
   659
blanchet@49334
   660
fun kill_learners () = Async_Manager.kill_threads MaShN "learner"
blanchet@49334
   661
fun running_learners () = Async_Manager.running_threads MaShN "learner"
blanchet@49334
   662
blanchet@49263
   663
end;