src/HOL/Tools/ATP/atp_proof_reconstruct.ML

more refactoring to accommodate SMT proofs

(*  Title:      HOL/Tools/ATP/atp_proof_reconstruct.ML
    Author:     Lawrence C. Paulson, Cambridge University Computer Laboratory
    Author:     Claire Quigley, Cambridge University Computer Laboratory
    Author:     Jasmin Blanchette, TU Muenchen

Basic proof reconstruction from ATP proofs.
*)

signature ATP_PROOF_RECONSTRUCT =
sig
  type ('a, 'b) atp_term = ('a, 'b) ATP_Problem.atp_term
  type ('a, 'b, 'c, 'd) atp_formula = ('a, 'b, 'c, 'd) ATP_Problem.atp_formula
  type stature = ATP_Problem_Generate.stature
  type ('a, 'b) atp_step = ('a, 'b) ATP_Proof.atp_step
  type 'a atp_proof = 'a ATP_Proof.atp_proof

  val metisN : string
  val full_typesN : string
  val partial_typesN : string
  val no_typesN : string
  val really_full_type_enc : string
  val full_type_enc : string
  val partial_type_enc : string
  val no_type_enc : string
  val full_type_encs : string list
  val partial_type_encs : string list
  val default_metis_lam_trans : string
  val metis_call : string -> string -> string
  val forall_of : term -> term -> term
  val exists_of : term -> term -> term
  val unalias_type_enc : string -> string list
  val term_of_atp :
    Proof.context -> bool -> int Symtab.table -> typ option ->
    (string, string) atp_term -> term
  val prop_of_atp :
    Proof.context -> bool -> int Symtab.table ->
    (string, string, (string, string) atp_term, string) atp_formula -> term

  val used_facts_in_atp_proof :
    Proof.context -> (string * stature) list vector -> string atp_proof ->
    (string * stature) list
  val used_facts_in_unsound_atp_proof :
    Proof.context -> (string * stature) list vector -> 'a atp_proof ->
    string list option
  val lam_trans_of_atp_proof : string atp_proof -> string -> string
  val is_typed_helper_used_in_atp_proof : string atp_proof -> bool
  val termify_atp_proof : Proof.context -> string Symtab.table -> (string * term) list ->
    int Symtab.table -> string atp_proof -> (term, string) atp_step list
  val factify_atp_proof : (string * 'a) list vector -> term list -> term ->
    (term, string) atp_step list -> (term, string) atp_step list
end;

structure ATP_Proof_Reconstruct : ATP_PROOF_RECONSTRUCT =
struct

open ATP_Util
open ATP_Problem
open ATP_Proof
open ATP_Problem_Generate

val metisN = "metis"

val full_typesN = "full_types"
val partial_typesN = "partial_types"
val no_typesN = "no_types"

val really_full_type_enc = "mono_tags"
val full_type_enc = "poly_guards_query"
val partial_type_enc = "poly_args"
val no_type_enc = "erased"

val full_type_encs = [full_type_enc, really_full_type_enc]
val partial_type_encs = partial_type_enc :: full_type_encs

val type_enc_aliases =
  [(full_typesN, full_type_encs),
   (partial_typesN, partial_type_encs),
   (no_typesN, [no_type_enc])]

fun unalias_type_enc s =
  AList.lookup (op =) type_enc_aliases s |> the_default [s]

val default_metis_lam_trans = combsN

fun metis_call type_enc lam_trans =
  let
    val type_enc =
      case AList.find (fn (enc, encs) => enc = hd encs) type_enc_aliases
                      type_enc of
        [alias] => alias
      | _ => type_enc
    val opts = [] |> type_enc <> partial_typesN ? cons type_enc
                  |> lam_trans <> default_metis_lam_trans ? cons lam_trans
  in metisN ^ (if null opts then "" else " (" ^ commas opts ^ ")") end

fun term_name' (Var ((s, _), _)) = perhaps (try Name.dest_skolem) s
  | term_name' _ = ""

fun lambda' v = Term.lambda_name (term_name' v, v)

fun forall_of v t = HOLogic.all_const (fastype_of v) $ lambda' v t
fun exists_of v t = HOLogic.exists_const (fastype_of v) $ lambda' v t

fun make_tfree ctxt w =
  let val ww = "'" ^ w in
    TFree (ww, the_default HOLogic.typeS (Variable.def_sort ctxt (ww, ~1)))
  end

exception ATP_TERM of (string, string) atp_term list
exception ATP_FORMULA of
    (string, string, (string, string) atp_term, string) atp_formula list
exception SAME of unit

(* Type variables are given the basic sort "HOL.type". Some will later be
   constrained by information from type literals, or by type inference. *)
fun typ_of_atp ctxt (u as ATerm ((a, _), us)) =
  let val Ts = map (typ_of_atp ctxt) us in
    case unprefix_and_unascii type_const_prefix a of
      SOME b => Type (invert_const b, Ts)
    | NONE =>
      if not (null us) then
        raise ATP_TERM [u]  (* only "tconst"s have type arguments *)
      else case unprefix_and_unascii tfree_prefix a of
        SOME b => make_tfree ctxt b
      | NONE =>
        (* Could be an Isabelle variable or a variable from the ATP, say "X1"
           or "_5018". Sometimes variables from the ATP are indistinguishable
           from Isabelle variables, which forces us to use a type parameter in
           all cases. *)
        (a |> perhaps (unprefix_and_unascii tvar_prefix), HOLogic.typeS)
        |> Type_Infer.param 0
  end

(* Type class literal applied to a type. Returns triple of polarity, class,
   type. *)
fun type_constraint_of_term ctxt (u as ATerm ((a, _), us)) =
  case (unprefix_and_unascii class_prefix a, map (typ_of_atp ctxt) us) of
    (SOME b, [T]) => (b, T)
  | _ => raise ATP_TERM [u]

(* Accumulate type constraints in a formula: negative type literals. *)
fun add_var (key, z)  = Vartab.map_default (key, []) (cons z)
fun add_type_constraint false (cl, TFree (a ,_)) = add_var ((a, ~1), cl)
  | add_type_constraint false (cl, TVar (ix, _)) = add_var (ix, cl)
  | add_type_constraint _ _ = I

fun repair_var_name s =
  let
    fun subscript_name s n = s ^ nat_subscript n
    val s = s |> String.map Char.toLower
  in
    case space_explode "_" s of
      [_] => (case take_suffix Char.isDigit (String.explode s) of
                (cs1 as _ :: _, cs2 as _ :: _) =>
                subscript_name (String.implode cs1)
                               (the (Int.fromString (String.implode cs2)))
              | (_, _) => s)
    | [s1, s2] => (case Int.fromString s2 of
                     SOME n => subscript_name s1 n
                   | NONE => s)
    | _ => s
  end

(* The number of type arguments of a constant, zero if it's monomorphic. For
   (instances of) Skolem pseudoconstants, this information is encoded in the
   constant name. *)
fun robust_const_num_type_args thy s =
  if String.isPrefix skolem_const_prefix s then
    s |> Long_Name.explode |> List.last |> Int.fromString |> the
  else if String.isPrefix lam_lifted_prefix s then
    if String.isPrefix lam_lifted_poly_prefix s then 2 else 0
  else
    (s, Sign.the_const_type thy s) |> Sign.const_typargs thy |> length

fun slack_fastype_of t = fastype_of t handle TERM _ => HOLogic.typeT

(* Cope with "tt(X) = X" atoms, where "X" is existentially quantified. *)
fun loose_aconv (Free (s, _), Free (s', _)) = s = s'
  | loose_aconv (t, t') = t aconv t'

val vampire_skolem_prefix = "sK"

(* First-order translation. No types are known for variables. "HOLogic.typeT"
   should allow them to be inferred. *)
fun term_of_atp ctxt textual sym_tab =
  let
    val thy = Proof_Context.theory_of ctxt
    (* For Metis, we use 1 rather than 0 because variable references in clauses
       may otherwise conflict with variable constraints in the goal. At least,
       type inference often fails otherwise. See also "axiom_inference" in
       "Metis_Reconstruct". *)
    val var_index = if textual then 0 else 1
    fun do_term extra_ts opt_T u =
      case u of
        ATerm ((s, _), us) =>
        if s = ""
          then error "Isar proof reconstruction failed because the ATP proof \
                     \contains unparsable material."
        else if String.isPrefix native_type_prefix s then
          @{const True} (* ignore TPTP type information *)
        else if s = tptp_equal then
          let val ts = map (do_term [] NONE) us in
            if textual andalso length ts = 2 andalso
               loose_aconv (hd ts, List.last ts) then
              @{const True}
            else
              list_comb (Const (@{const_name HOL.eq}, HOLogic.typeT), ts)
          end
        else case unprefix_and_unascii const_prefix s of
          SOME s' =>
          let
            val ((s', s''), mangled_us) =
              s' |> unmangled_const |>> `invert_const
          in
            if s' = type_tag_name then
              case mangled_us @ us of
                [typ_u, term_u] =>
                do_term extra_ts (SOME (typ_of_atp ctxt typ_u)) term_u
              | _ => raise ATP_TERM us
            else if s' = predicator_name then
              do_term [] (SOME @{typ bool}) (hd us)
            else if s' = app_op_name then
              let val extra_t = do_term [] NONE (List.last us) in
                do_term (extra_t :: extra_ts)
                        (case opt_T of
                           SOME T => SOME (slack_fastype_of extra_t --> T)
                         | NONE => NONE)
                        (nth us (length us - 2))
              end
            else if s' = type_guard_name then
              @{const True} (* ignore type predicates *)
            else
              let
                val new_skolem = String.isPrefix new_skolem_const_prefix s''
                val num_ty_args =
                  length us - the_default 0 (Symtab.lookup sym_tab s)
                val (type_us, term_us) =
                  chop num_ty_args us |>> append mangled_us
                val term_ts = map (do_term [] NONE) term_us
                val T =
                  (if not (null type_us) andalso
                      robust_const_num_type_args thy s' = length type_us then
                     let val Ts = type_us |> map (typ_of_atp ctxt) in
                       if new_skolem then
                         SOME (Type_Infer.paramify_vars (tl Ts ---> hd Ts))
                       else if textual then
                         try (Sign.const_instance thy) (s', Ts)
                       else
                         NONE
                     end
                   else
                     NONE)
                  |> (fn SOME T => T
                       | NONE => map slack_fastype_of term_ts --->
                                 (case opt_T of
                                    SOME T => T
                                  | NONE => HOLogic.typeT))
                val t =
                  if new_skolem then
                    Var ((new_skolem_var_name_of_const s'', var_index), T)
                  else
                    Const (unproxify_const s', T)
              in list_comb (t, term_ts @ extra_ts) end
          end
        | NONE => (* a free or schematic variable *)
          let
            (* This assumes that distinct names are mapped to distinct names by
               "Variable.variant_frees". This does not hold in general but
               should hold for ATP-generated Skolem function names, since these
               end with a digit and "variant_frees" appends letters. *)
            fun fresh_up s =
              [(s, ())] |> Variable.variant_frees ctxt [] |> hd |> fst
            val term_ts =
              map (do_term [] NONE) us
              (* Vampire (2.6) passes arguments to Skolem functions in reverse
                 order *)
              |> String.isPrefix vampire_skolem_prefix s ? rev
            val ts = term_ts @ extra_ts
            val T =
              case opt_T of
                SOME T => map slack_fastype_of term_ts ---> T
              | NONE => map slack_fastype_of ts ---> HOLogic.typeT
            val t =
              case unprefix_and_unascii fixed_var_prefix s of
                SOME s => Free (s, T)
              | NONE =>
                case unprefix_and_unascii schematic_var_prefix s of
                  SOME s => Var ((s, var_index), T)
                | NONE =>
                  if textual andalso not (is_tptp_variable s) then
                    Free (s |> textual ? (repair_var_name #> fresh_up), T)
                  else
                    Var ((s |> textual ? repair_var_name, var_index), T)
          in list_comb (t, ts) end
  in do_term [] end

fun term_of_atom ctxt textual sym_tab pos (u as ATerm ((s, _), _)) =
  if String.isPrefix class_prefix s then
    add_type_constraint pos (type_constraint_of_term ctxt u)
    #> pair @{const True}
  else
    pair (term_of_atp ctxt textual sym_tab (SOME @{typ bool}) u)

(* Update schematic type variables with detected sort constraints. It's not
   totally clear whether this code is necessary. *)
fun repair_tvar_sorts (t, tvar_tab) =
  let
    fun do_type (Type (a, Ts)) = Type (a, map do_type Ts)
      | do_type (TVar (xi, s)) =
        TVar (xi, the_default s (Vartab.lookup tvar_tab xi))
      | do_type (TFree z) = TFree z
    fun do_term (Const (a, T)) = Const (a, do_type T)
      | do_term (Free (a, T)) = Free (a, do_type T)
      | do_term (Var (xi, T)) = Var (xi, do_type T)
      | do_term (t as Bound _) = t
      | do_term (Abs (a, T, t)) = Abs (a, do_type T, do_term t)
      | do_term (t1 $ t2) = do_term t1 $ do_term t2
  in t |> not (Vartab.is_empty tvar_tab) ? do_term end

fun quantify_over_var quant_of var_s t =
  let
    val vars = [] |> Term.add_vars t |> filter (fn ((s, _), _) => s = var_s)
    val normTs = vars |> AList.group (op =) |> map (apsnd hd)
    fun norm_var_types (Var (x, T)) =
        Var (x, case AList.lookup (op =) normTs x of
                  NONE => T
                | SOME T' => T')
      | norm_var_types t = t
  in t |> map_aterms norm_var_types |> fold_rev quant_of (map Var normTs) end

(* Interpret an ATP formula as a HOL term, extracting sort constraints as they
   appear in the formula. *)
fun prop_of_atp ctxt textual sym_tab phi =
  let
    fun do_formula pos phi =
      case phi of
        AQuant (_, [], phi) => do_formula pos phi
      | AQuant (q, (s, _) :: xs, phi') =>
        do_formula pos (AQuant (q, xs, phi'))
        (* FIXME: TFF *)
        #>> quantify_over_var
              (case q of AForall => forall_of | AExists => exists_of)
              (s |> textual ? repair_var_name)
      | AConn (ANot, [phi']) => do_formula (not pos) phi' #>> s_not
      | AConn (c, [phi1, phi2]) =>
        do_formula (pos |> c = AImplies ? not) phi1
        ##>> do_formula pos phi2
        #>> (case c of
               AAnd => s_conj
             | AOr => s_disj
             | AImplies => s_imp
             | AIff => s_iff
             | ANot => raise Fail "impossible connective")
      | AAtom tm => term_of_atom ctxt textual sym_tab pos tm
      | _ => raise ATP_FORMULA [phi]
  in repair_tvar_sorts (do_formula true phi Vartab.empty) end

fun find_first_in_list_vector vec key =
  Vector.foldl (fn (ps, NONE) => AList.lookup (op =) ps key
                 | (_, value) => value) NONE vec

val unprefix_fact_number = space_implode "_" o tl o space_explode "_"

fun resolve_one_named_fact fact_names s =
  case try (unprefix fact_prefix) s of
    SOME s' =>
    let val s' = s' |> unprefix_fact_number |> unascii_of in
      s' |> find_first_in_list_vector fact_names |> Option.map (pair s')
    end
  | NONE => NONE
fun resolve_fact fact_names = map_filter (resolve_one_named_fact fact_names)
fun is_fact fact_names = not o null o resolve_fact fact_names

fun resolve_one_named_conjecture s =
  case try (unprefix conjecture_prefix) s of
    SOME s' => Int.fromString s'
  | NONE => NONE

val resolve_conjecture = map_filter resolve_one_named_conjecture
val is_conjecture = not o null o resolve_conjecture

fun is_axiom_used_in_proof pred =
  exists (fn ((_, ss), _, _, _, []) => exists pred ss | _ => false)

fun add_non_rec_defs fact_names accum =
  Vector.foldl (fn (facts, facts') =>
      union (op =) (filter (fn (_, (_, status)) => status = Non_Rec_Def) facts)
            facts')
    accum fact_names

val isa_ext = Thm.get_name_hint @{thm ext}
val isa_short_ext = Long_Name.base_name isa_ext

fun ext_name ctxt =
  if Thm.eq_thm_prop (@{thm ext},
       singleton (Attrib.eval_thms ctxt) (Facts.named isa_short_ext, [])) then
    isa_short_ext
  else
    isa_ext

val leo2_extcnf_equal_neg_rule = "extcnf_equal_neg"
val leo2_unfold_def_rule = "unfold_def"

fun add_fact ctxt fact_names ((_, ss), _, _, rule, deps) =
  (if rule = leo2_extcnf_equal_neg_rule then
     insert (op =) (ext_name ctxt, (Global, General))
   else if rule = leo2_unfold_def_rule then
     (* LEO 1.3.3 does not record definitions properly, leading to missing
        dependencies in the TSTP proof. Remove the next line once this is
        fixed. *)
     add_non_rec_defs fact_names
   else if rule = agsyhol_coreN orelse rule = satallax_coreN then
     (fn [] =>
         (* agsyHOL and Satallax don't include definitions in their
            unsatisfiable cores, so we assume the worst and include them all
            here. *)
         [(ext_name ctxt, (Global, General))] |> add_non_rec_defs fact_names
       | facts => facts)
   else
     I)
  #> (if null deps then union (op =) (resolve_fact fact_names ss) else I)

fun used_facts_in_atp_proof ctxt fact_names atp_proof =
  if null atp_proof then Vector.foldl (uncurry (union (op =))) [] fact_names
  else fold (add_fact ctxt fact_names) atp_proof []

fun used_facts_in_unsound_atp_proof _ _ [] = NONE
  | used_facts_in_unsound_atp_proof ctxt fact_names atp_proof =
    let val used_facts = used_facts_in_atp_proof ctxt fact_names atp_proof in
      if forall (fn (_, (sc, _)) => sc = Global) used_facts andalso
         not (is_axiom_used_in_proof (is_conjecture o single) atp_proof) then
        SOME (map fst used_facts)
      else
        NONE
    end

val ascii_of_lam_fact_prefix = ascii_of lam_fact_prefix

(* overapproximation (good enough) *)
fun is_lam_lifted s =
  String.isPrefix fact_prefix s andalso
  String.isSubstring ascii_of_lam_fact_prefix s

val is_combinator_def = String.isPrefix (helper_prefix ^ combinator_prefix)

fun lam_trans_of_atp_proof atp_proof default =
  case (is_axiom_used_in_proof is_combinator_def atp_proof,
        is_axiom_used_in_proof is_lam_lifted atp_proof) of
    (false, false) => default
  | (false, true) => liftingN
(*  | (true, true) => combs_and_liftingN -- not supported by "metis" *)
  | (true, _) => combsN

val is_typed_helper_name =
  String.isPrefix helper_prefix andf String.isSuffix typed_helper_suffix

fun is_typed_helper_used_in_atp_proof atp_proof =
  is_axiom_used_in_proof is_typed_helper_name atp_proof

fun repair_name "$true" = "c_True"
  | repair_name "$false" = "c_False"
  | repair_name "$$e" = tptp_equal (* seen in Vampire proofs *)
  | repair_name s =
    if is_tptp_equal s orelse
       (* seen in Vampire proofs *)
       (String.isPrefix "sQ" s andalso String.isSuffix "_eqProxy" s) then
      tptp_equal
    else
      s

fun infer_formula_types ctxt =
  Type.constraint HOLogic.boolT
  #> Syntax.check_term
         (Proof_Context.set_mode Proof_Context.mode_schematic ctxt)

val combinator_table =
  [(@{const_name Meson.COMBI}, @{thm Meson.COMBI_def [abs_def]}),
   (@{const_name Meson.COMBK}, @{thm Meson.COMBK_def [abs_def]}),
   (@{const_name Meson.COMBB}, @{thm Meson.COMBB_def [abs_def]}),
   (@{const_name Meson.COMBC}, @{thm Meson.COMBC_def [abs_def]}),
   (@{const_name Meson.COMBS}, @{thm Meson.COMBS_def [abs_def]})]

fun uncombine_term thy =
  let
    fun aux (t1 $ t2) = betapply (pairself aux (t1, t2))
      | aux (Abs (s, T, t')) = Abs (s, T, aux t')
      | aux (t as Const (x as (s, _))) =
        (case AList.lookup (op =) combinator_table s of
           SOME thm => thm |> prop_of |> specialize_type thy x
                           |> Logic.dest_equals |> snd
         | NONE => t)
      | aux t = t
  in aux end

fun unlift_term lifted =
  map_aterms (fn t as Const (s, _) =>
                 if String.isPrefix lam_lifted_prefix s then
                   case AList.lookup (op =) lifted s of
                     SOME t =>
                     (* FIXME: do something about the types *)
                     unlift_term lifted t
                   | NONE => t
                 else
                   t
               | t => t)

fun decode_line ctxt lifted sym_tab (name, role, u, rule, deps) =
  let
    val thy = Proof_Context.theory_of ctxt
    val t =
      u |> prop_of_atp ctxt true sym_tab
        |> uncombine_term thy
        |> unlift_term lifted
        |> infer_formula_types ctxt
  in (name, role, t, rule, deps) end

val waldmeister_conjecture_num = "1.0.0.0"

fun repair_waldmeister_endgame arg =
  let
    fun do_tail (name, _, t, rule, deps) =
      (name, Negated_Conjecture, s_not t, rule, deps)
    fun do_body [] = []
      | do_body ((line as ((num, _), _, _, _, _)) :: lines) =
        if num = waldmeister_conjecture_num then map do_tail (line :: lines)
        else line :: do_body lines
  in do_body arg end

fun termify_atp_proof ctxt pool lifted sym_tab =
  clean_up_atp_proof_dependencies
  #> nasty_atp_proof pool
  #> map_term_names_in_atp_proof repair_name
  #> map (decode_line ctxt lifted sym_tab)
  #> repair_waldmeister_endgame

fun factify_atp_proof fact_names hyp_ts concl_t atp_proof =
  let
    fun factify_step ((num, ss), role, t, rule, deps) =
      let
        val (ss', role', t') =
          (case resolve_conjecture ss of
            [j] =>
            if j = length hyp_ts then ([], Conjecture, concl_t) else ([], Hypothesis, nth hyp_ts j)
           | _ =>
             (case resolve_fact fact_names ss of
               [] => (ss, Plain, t)
             | facts => (map fst facts, Axiom, t)))
      in
        ((num, ss'), role', t', rule, deps)
      end

    val atp_proof = map factify_step atp_proof
    val names = map #1 atp_proof

    fun repair_dep (num, ss) = (num, the_default ss (AList.lookup (op =) names num))
    fun repair_deps (name, role, t, rule, deps) = (name, role, t, rule, map repair_dep deps)

  in map repair_deps atp_proof end

end;