theory Myhill_2
  imports Myhill_1 Prefix_subtract
          "~~/src/HOL/Library/List_Prefix"
begin

section {* Direction @{text "regular language \<Rightarrow>finite partition"} *}

definition
  str_eq :: "string \<Rightarrow> lang \<Rightarrow> string \<Rightarrow> bool" ("_ \<approx>_ _")
where
  "x \<approx>A y \<equiv> (x, y) \<in> (\<approx>A)"

definition 
   tag_eq_rel :: "(string \<Rightarrow> 'b) \<Rightarrow> (string \<times> string) set" ("=_=")
where
   "=tag= \<equiv> {(x, y) | x y. tag x = tag y}"

lemma finite_eq_tag_rel:
  assumes rng_fnt: "finite (range tag)"
  shows "finite (UNIV // =tag=)"
proof -
  let "?f" =  "\<lambda>X. tag ` X" and ?A = "(UNIV // =tag=)"
  have "finite (?f ` ?A)" 
  proof -
    have "range ?f \<subseteq> (Pow (range tag))" unfolding Pow_def by auto
    moreover 
    have "finite (Pow (range tag))" using rng_fnt by simp
    ultimately 
    have "finite (range ?f)" unfolding image_def by (blast intro: finite_subset)
    moreover
    have "?f ` ?A \<subseteq> range ?f" by auto
    ultimately show "finite (?f ` ?A)" by (rule rev_finite_subset) 
  qed
  moreover
  have "inj_on ?f ?A"
  proof -
    { fix X Y
      assume X_in: "X \<in> ?A"
        and  Y_in: "Y \<in> ?A"
        and  tag_eq: "?f X = ?f Y"
      then obtain x y 
        where "x \<in> X" "y \<in> Y" "tag x = tag y"
        unfolding quotient_def Image_def image_def tag_eq_rel_def
        by (simp) (blast)
      with X_in Y_in 
      have "X = Y"
	unfolding quotient_def tag_eq_rel_def by auto
    } 
    then show "inj_on ?f ?A" unfolding inj_on_def by auto
  qed
  ultimately show "finite (UNIV // =tag=)" by (rule finite_imageD)
qed

lemma refined_partition_finite:
  assumes fnt: "finite (UNIV // R1)"
  and refined: "R1 \<subseteq> R2"
  and eq1: "equiv UNIV R1" and eq2: "equiv UNIV R2"
  shows "finite (UNIV // R2)"
proof -
  let ?f = "\<lambda>X. {R1 `` {x} | x. x \<in> X}" 
    and ?A = "UNIV // R2" and ?B = "UNIV // R1"
  have "?f ` ?A \<subseteq> Pow ?B"
    unfolding image_def Pow_def quotient_def by auto
  moreover
  have "finite (Pow ?B)" using fnt by simp
  ultimately  
  have "finite (?f ` ?A)" by (rule finite_subset)
  moreover
  have "inj_on ?f ?A"
  proof -
    { fix X Y
      assume X_in: "X \<in> ?A" and Y_in: "Y \<in> ?A" and eq_f: "?f X = ?f Y"
      from quotientE [OF X_in]
      obtain x where "X = R2 `` {x}" by blast
      with equiv_class_self[OF eq2] have x_in: "x \<in> X" by simp
      then have "R1 ``{x} \<in> ?f X" by auto
      with eq_f have "R1 `` {x} \<in> ?f Y" by simp
      then obtain y 
        where y_in: "y \<in> Y" and eq_r1_xy: "R1 `` {x} = R1 `` {y}" by auto
      with eq_equiv_class[OF _ eq1] 
      have "(x, y) \<in> R1" by blast
      with refined have "(x, y) \<in> R2" by auto
      with quotient_eqI [OF eq2 X_in Y_in x_in y_in]
      have "X = Y" .
    } 
    then show "inj_on ?f ?A" unfolding inj_on_def by blast 
  qed
  ultimately show "finite (UNIV // R2)" by (rule finite_imageD)
qed

lemma tag_finite_imageD:
  assumes rng_fnt: "finite (range tag)" 
  and same_tag_eqvt: "\<And>m n. tag m = tag n \<Longrightarrow> m \<approx>A n"
  shows "finite (UNIV // \<approx>A)"
proof (rule_tac refined_partition_finite [of "=tag="])
  show "finite (UNIV // =tag=)" by (rule finite_eq_tag_rel[OF rng_fnt])
next
  from same_tag_eqvt
  show "=tag= \<subseteq> \<approx>A" unfolding tag_eq_rel_def str_eq_def
    by auto
next
  show "equiv UNIV =tag="
    unfolding equiv_def tag_eq_rel_def refl_on_def sym_def trans_def
    by auto
next
  show "equiv UNIV (\<approx>A)" 
    unfolding equiv_def str_eq_rel_def sym_def refl_on_def trans_def
    by blast
qed


subsection {* The proof *}

subsubsection {* The base case for @{const "NULL"} *}

lemma quot_null_eq:
  shows "UNIV // \<approx>{} = {UNIV}"
unfolding quotient_def Image_def str_eq_rel_def by auto

lemma quot_null_finiteI [intro]:
  shows "finite (UNIV // \<approx>{})"
unfolding quot_null_eq by simp


subsubsection {* The base case for @{const "EMPTY"} *}

lemma quot_empty_subset:
  shows "UNIV // \<approx>{[]} \<subseteq> {{[]}, UNIV - {[]}}"
proof
  fix x
  assume "x \<in> UNIV // \<approx>{[]}"
  then obtain y where h: "x = {z. (y, z) \<in> \<approx>{[]}}" 
    unfolding quotient_def Image_def by blast
  show "x \<in> {{[]}, UNIV - {[]}}"
  proof (cases "y = []")
    case True with h
    have "x = {[]}" by (auto simp: str_eq_rel_def)
    thus ?thesis by simp
  next
    case False with h
    have "x = UNIV - {[]}" by (auto simp: str_eq_rel_def)
    thus ?thesis by simp
  qed
qed

lemma quot_empty_finiteI [intro]:
  shows "finite (UNIV // \<approx>{[]})"
by (rule finite_subset[OF quot_empty_subset]) (simp)


subsubsection {* The base case for @{const "CHAR"} *}

lemma quot_char_subset:
  "UNIV // (\<approx>{[c]}) \<subseteq> {{[]},{[c]}, UNIV - {[], [c]}}"
proof 
  fix x 
  assume "x \<in> UNIV // \<approx>{[c]}"
  then obtain y where h: "x = {z. (y, z) \<in> \<approx>{[c]}}" 
    unfolding quotient_def Image_def by blast
  show "x \<in> {{[]},{[c]}, UNIV - {[], [c]}}"
  proof -
    { assume "y = []" hence "x = {[]}" using h 
        by (auto simp:str_eq_rel_def) } 
    moreover 
    { assume "y = [c]" hence "x = {[c]}" using h 
        by (auto dest!:spec[where x = "[]"] simp:str_eq_rel_def) } 
    moreover 
    { assume "y \<noteq> []" and "y \<noteq> [c]"
      hence "\<forall> z. (y @ z) \<noteq> [c]" by (case_tac y, auto)
      moreover have "\<And> p. (p \<noteq> [] \<and> p \<noteq> [c]) = (\<forall> q. p @ q \<noteq> [c])" 
        by (case_tac p, auto)
      ultimately have "x = UNIV - {[],[c]}" using h
        by (auto simp add:str_eq_rel_def)
    } 
    ultimately show ?thesis by blast
  qed
qed

lemma quot_char_finiteI [intro]:
  shows "finite (UNIV // \<approx>{[c]})"
by (rule finite_subset[OF quot_char_subset]) (simp)


subsubsection {* The inductive case for @{const ALT} *}

definition 
  tag_str_ALT :: "lang \<Rightarrow> lang \<Rightarrow> string \<Rightarrow> (lang \<times> lang)"
where
  "tag_str_ALT A B \<equiv> (\<lambda>x. (\<approx>A `` {x}, \<approx>B `` {x}))"

lemma quot_union_finiteI [intro]:
  fixes L1 L2::"lang"
  assumes finite1: "finite (UNIV // \<approx>A)"
  and     finite2: "finite (UNIV // \<approx>B)"
  shows "finite (UNIV // \<approx>(A \<union> B))"
proof (rule_tac tag = "tag_str_ALT A B" in tag_finite_imageD)
  have "finite ((UNIV // \<approx>A) \<times> (UNIV // \<approx>B))" 
    using finite1 finite2 by auto
  then show "finite (range (tag_str_ALT A B))"
    unfolding tag_str_ALT_def quotient_def
    by (rule rev_finite_subset) (auto)
next
  show "\<And>x y. tag_str_ALT A B x = tag_str_ALT A B y \<Longrightarrow> x \<approx>(A \<union> B) y"
    unfolding tag_str_ALT_def
    unfolding str_eq_def
    unfolding str_eq_rel_def
    by auto
qed


subsubsection {* The inductive case for @{text "SEQ"}*}

definition 
  tag_str_SEQ :: "lang \<Rightarrow> lang \<Rightarrow> string \<Rightarrow> (lang \<times> lang set)"
where
  "tag_str_SEQ L1 L2 \<equiv>
     (\<lambda>x. (\<approx>L1 `` {x}, {(\<approx>L2 `` {x - xa}) | xa.  xa \<le> x \<and> xa \<in> L1}))"

lemma Seq_in_cases:
  assumes "x @ z \<in> A ;; B"
  shows "(\<exists> x' \<le> x. x' \<in> A \<and> (x - x') @ z \<in> B) \<or> 
         (\<exists> z' \<le> z. (x @ z') \<in> A \<and> (z - z') \<in> B)"
using assms
unfolding Seq_def prefix_def
by (auto simp add: append_eq_append_conv2)

lemma tag_str_SEQ_injI:
  assumes eq_tag: "tag_str_SEQ A B x = tag_str_SEQ A B y" 
  shows "x \<approx>(A ;; B) y"
proof -
  { fix x y z
    assume xz_in_seq: "x @ z \<in> A ;; B"
    and tag_xy: "tag_str_SEQ A B x = tag_str_SEQ A B y"
    have"y @ z \<in> A ;; B" 
    proof -
      { (* first case with x' in A and (x - x') @ z in B *)
        fix x'
        assume h1: "x' \<le> x" and h2: "x' \<in> A" and h3: "(x - x') @ z \<in> B"
        obtain y' 
          where "y' \<le> y" 
          and "y' \<in> A" 
          and "(y - y') @ z \<in> B"
        proof -
          have "{\<approx>B `` {x - x'} |x'. x' \<le> x \<and> x' \<in> A} = 
                {\<approx>B `` {y - y'} |y'. y' \<le> y \<and> y' \<in> A}" (is "?Left = ?Right")
            using tag_xy unfolding tag_str_SEQ_def by simp
          moreover 
	  have "\<approx>B `` {x - x'} \<in> ?Left" using h1 h2 by auto
          ultimately 
	  have "\<approx>B `` {x - x'} \<in> ?Right" by simp
          then obtain y' 
            where eq_xy': "\<approx>B `` {x - x'} = \<approx>B `` {y - y'}" 
            and pref_y': "y' \<le> y" and y'_in: "y' \<in> A"
            by simp blast
	  
	  have "(x - x')  \<approx>B (y - y')" using eq_xy'
            unfolding Image_def str_eq_rel_def str_eq_def by auto
          with h3 have "(y - y') @ z \<in> B" 
	    unfolding str_eq_rel_def str_eq_def by simp
          with pref_y' y'_in 
          show ?thesis using that by blast
        qed
	then have "y @ z \<in> A ;; B" by (erule_tac prefixE) (auto simp: Seq_def)
      } 
      moreover 
      { (* second case with x @ z' in A and z - z' in B *)
        fix z'
        assume h1: "z' \<le> z" and h2: "(x @ z') \<in> A" and h3: "z - z' \<in> B"
	 have "\<approx>A `` {x} = \<approx>A `` {y}" 
           using tag_xy unfolding tag_str_SEQ_def by simp
         with h2 have "y @ z' \<in> A"
            unfolding Image_def str_eq_rel_def str_eq_def by auto
        with h1 h3 have "y @ z \<in> A ;; B"
	  unfolding prefix_def Seq_def
	  by (auto) (metis append_assoc)
      }
      ultimately show "y @ z \<in> A ;; B" 
	using Seq_in_cases [OF xz_in_seq] by blast
    qed
  }
  from this [OF _ eq_tag] and this [OF _ eq_tag [THEN sym]]
    show "x \<approx>(A ;; B) y" unfolding str_eq_def str_eq_rel_def by blast
qed 

lemma quot_seq_finiteI [intro]:
  fixes L1 L2::"lang"
  assumes fin1: "finite (UNIV // \<approx>L1)" 
  and     fin2: "finite (UNIV // \<approx>L2)" 
  shows "finite (UNIV // \<approx>(L1 ;; L2))"
proof (rule_tac tag = "tag_str_SEQ L1 L2" in tag_finite_imageD)
  show "\<And>x y. tag_str_SEQ L1 L2 x = tag_str_SEQ L1 L2 y \<Longrightarrow> x \<approx>(L1 ;; L2) y"
    by (rule tag_str_SEQ_injI)
next
  have *: "finite ((UNIV // \<approx>L1) \<times> (Pow (UNIV // \<approx>L2)))" 
    using fin1 fin2 by auto
  show "finite (range (tag_str_SEQ L1 L2))" 
    unfolding tag_str_SEQ_def
    apply(rule finite_subset[OF _ *])
    unfolding quotient_def
    by auto
qed


subsubsection {* The inductive case for @{const "STAR"} *}

definition 
  tag_str_STAR :: "lang \<Rightarrow> string \<Rightarrow> lang set"
where
  "tag_str_STAR L1 \<equiv> (\<lambda>x. {\<approx>L1 `` {x - xa} | xa. xa < x \<and> xa \<in> L1\<star>})"

text {* A technical lemma. *}
lemma finite_set_has_max: "\<lbrakk>finite A; A \<noteq> {}\<rbrakk> \<Longrightarrow> 
           (\<exists> max \<in> A. \<forall> a \<in> A. f a <= (f max :: nat))"
proof (induct rule:finite.induct)
  case emptyI thus ?case by simp
next
  case (insertI A a)
  show ?case
  proof (cases "A = {}")
    case True thus ?thesis by (rule_tac x = a in bexI, auto)
  next
    case False
    with insertI.hyps and False  
    obtain max 
      where h1: "max \<in> A" 
      and h2: "\<forall>a\<in>A. f a \<le> f max" by blast
    show ?thesis
    proof (cases "f a \<le> f max")
      assume "f a \<le> f max"
      with h1 h2 show ?thesis by (rule_tac x = max in bexI, auto)
    next
      assume "\<not> (f a \<le> f max)"
      thus ?thesis using h2 by (rule_tac x = a in bexI, auto)
    qed
  qed
qed


text {* The following is a technical lemma, which helps to show the range finiteness of tag function. *}

lemma finite_strict_prefix_set: "finite {xa. xa < (x::string)}"
apply (induct x rule:rev_induct, simp)
apply (subgoal_tac "{xa. xa < xs @ [x]} = {xa. xa < xs} \<union> {xs}")
by (auto simp:strict_prefix_def)


lemma tag_str_STAR_injI:
  assumes eq_tag: "tag_str_STAR L\<^isub>1 v = tag_str_STAR L\<^isub>1 w"
  shows "v \<approx>(L\<^isub>1\<star>) w"
proof-
  { fix x y z
    assume xz_in_star: "x @ z \<in> L\<^isub>1\<star>" 
      and tag_xy: "tag_str_STAR L\<^isub>1 x = tag_str_STAR L\<^isub>1 y"
    have "y @ z \<in> L\<^isub>1\<star>"
    proof(cases "x = []")
      case True
      with tag_xy have "y = []" 
        by (auto simp add: tag_str_STAR_def strict_prefix_def)
      thus ?thesis using xz_in_star True by simp
    next
      case False
      let ?S = "{xa. xa < x \<and> xa \<in> L\<^isub>1\<star> \<and> (x - xa) @ z \<in> L\<^isub>1\<star>}"
      have "finite ?S"
        by (rule_tac B = "{xa. xa < x}" in finite_subset, 
          auto simp:finite_strict_prefix_set)
      moreover have "?S \<noteq> {}" using False xz_in_star
        by (simp, rule_tac x = "[]" in exI, auto simp:strict_prefix_def)
      ultimately have "\<exists> xa_max \<in> ?S. \<forall> xa \<in> ?S. length xa \<le> length xa_max" 
        using finite_set_has_max by blast
      then obtain xa_max 
        where h1: "xa_max < x" 
        and h2: "xa_max \<in> L\<^isub>1\<star>" 
        and h3: "(x - xa_max) @ z \<in> L\<^isub>1\<star>" 
        and h4:"\<forall> xa < x. xa \<in> L\<^isub>1\<star> \<and> (x - xa) @ z \<in> L\<^isub>1\<star>  
                                     \<longrightarrow> length xa \<le> length xa_max"
        by blast
      obtain ya 
        where h5: "ya < y" and h6: "ya \<in> L\<^isub>1\<star>" 
        and eq_xya: "(x - xa_max) \<approx>L\<^isub>1 (y - ya)"
      proof-
        from tag_xy have "{\<approx>L\<^isub>1 `` {x - xa} |xa. xa < x \<and> xa \<in> L\<^isub>1\<star>} = 
          {\<approx>L\<^isub>1 `` {y - xa} |xa. xa < y \<and> xa \<in> L\<^isub>1\<star>}" (is "?left = ?right")
          by (auto simp:tag_str_STAR_def)
        moreover have "\<approx>L\<^isub>1 `` {x - xa_max} \<in> ?left" using h1 h2 by auto
        ultimately have "\<approx>L\<^isub>1 `` {x - xa_max} \<in> ?right" by simp
        thus ?thesis using that 
          apply (simp add:Image_def str_eq_rel_def str_eq_def) by blast
      qed 
      have "(y - ya) @ z \<in> L\<^isub>1\<star>" 
      proof-
        obtain za zb where eq_zab: "z = za @ zb" 
          and l_za: "(y - ya)@za \<in> L\<^isub>1" and ls_zb: "zb \<in> L\<^isub>1\<star>"
        proof -
          from h1 have "(x - xa_max) @ z \<noteq> []" 
            by (auto simp:strict_prefix_def elim:prefixE)
          from star_decom [OF h3 this]
          obtain a b where a_in: "a \<in> L\<^isub>1" 
            and a_neq: "a \<noteq> []" and b_in: "b \<in> L\<^isub>1\<star>" 
            and ab_max: "(x - xa_max) @ z = a @ b" by blast
          let ?za = "a - (x - xa_max)" and ?zb = "b"
          have pfx: "(x - xa_max) \<le> a" (is "?P1") 
            and eq_z: "z = ?za @ ?zb" (is "?P2")
          proof -
            have "((x - xa_max) \<le> a \<and> (a - (x - xa_max)) @ b = z) \<or> 
              (a < (x - xa_max) \<and> ((x - xa_max) - a) @ z = b)" 
              using append_eq_dest[OF ab_max] by (auto simp:strict_prefix_def)
            moreover {
              assume np: "a < (x - xa_max)" 
                and b_eqs: "((x - xa_max) - a) @ z = b"
              have "False"
              proof -
                let ?xa_max' = "xa_max @ a"
                have "?xa_max' < x" 
                  using np h1 by (clarsimp simp:strict_prefix_def diff_prefix) 
                moreover have "?xa_max' \<in> L\<^isub>1\<star>" 
                  using a_in h2 by (simp add:star_intro3) 
                moreover have "(x - ?xa_max') @ z \<in> L\<^isub>1\<star>" 
                  using b_eqs b_in np h1 by (simp add:diff_diff_append)
                moreover have "\<not> (length ?xa_max' \<le> length xa_max)" 
                  using a_neq by simp
                ultimately show ?thesis using h4 by blast
              qed }
            ultimately show ?P1 and ?P2 by auto
          qed
          hence "(x - xa_max)@?za \<in> L\<^isub>1" using a_in by (auto elim:prefixE)
          with eq_xya have "(y - ya) @ ?za \<in> L\<^isub>1" 
            by (auto simp:str_eq_def str_eq_rel_def)
           with eq_z and b_in 
          show ?thesis using that by blast
        qed
        have "((y - ya) @ za) @ zb \<in> L\<^isub>1\<star>" using  l_za ls_zb by blast
        with eq_zab show ?thesis by simp
      qed
      with h5 h6 show ?thesis 
        by (drule_tac star_intro1, auto simp:strict_prefix_def elim:prefixE)
    qed
  } 
  from this [OF _ eq_tag] and this [OF _ eq_tag [THEN sym]]
    show  ?thesis unfolding str_eq_def str_eq_rel_def by blast
qed

lemma quot_star_finiteI [intro]:
  assumes finite1: "finite (UNIV // \<approx>L1)"
  shows "finite (UNIV // \<approx>(L1\<star>))"
proof (rule_tac tag = "tag_str_STAR L1" in tag_finite_imageD)
  show "\<And>x y. tag_str_STAR L1 x = tag_str_STAR L1 y \<Longrightarrow> x \<approx>(L1\<star>) y"
    by (rule tag_str_STAR_injI)
next
  have *: "finite (Pow (UNIV // \<approx>L1))" 
    using finite1 by auto
  show "finite (range (tag_str_STAR L1))"
    unfolding tag_str_STAR_def
    apply(rule finite_subset[OF _ *])
    unfolding quotient_def
    by auto
qed

subsubsection{* The conclusion *}

lemma Myhill_Nerode2:
  fixes r::"rexp"
  shows "finite (UNIV // \<approx>(L r))"
by (induct r) (auto)


theorem Myhill_Nerode:
  shows "(\<exists>r::rexp. A = L r) \<longleftrightarrow> finite (UNIV // \<approx>A)"
using Myhill_Nerode1 Myhill_Nerode2 by auto

end