theory ReStar

  imports "Main" 

begin

section {* Sequential Composition of Languages *}

definition

  Sequ :: "string set \<Rightarrow> string set \<Rightarrow> string set" ("_ ;; _" [100,100] 100)

where 

  "A ;; B = {s1 @ s2 | s1 s2. s1 \<in> A \<and> s2 \<in> B}"

text {* Two Simple Properties about Sequential Composition *}

lemma seq_empty [simp]:

  shows "A ;; {[]} = A"

  and   "{[]} ;; A = A"

by (simp_all add: Sequ_def)

lemma seq_null [simp]:

  shows "A ;; {} = {}"

  and   "{} ;; A = {}"

by (simp_all add: Sequ_def)

section {* Semantic Derivative of Languages *}

definition

  Der :: "char \<Rightarrow> string set \<Rightarrow> string set"

where

  "Der c A \<equiv> {s. [c] @ s \<in> A}"

lemma Der_null [simp]:

  shows "Der c {} = {}"

unfolding Der_def

by auto

lemma Der_empty [simp]:

  shows "Der c {[]} = {}"

unfolding Der_def

by auto

lemma Der_char [simp]:

  shows "Der c {[d]} = (if c = d then {[]} else {})"

unfolding Der_def

by auto

lemma Der_union [simp]:

  shows "Der c (A \<union> B) = Der c A \<union> Der c B"

unfolding Der_def

by auto

lemma Der_Sequ [simp]:

  shows "Der c (A ;; B) = (Der c A) ;; B \<union> (if [] \<in> A then Der c B else {})"

unfolding Der_def Sequ_def

by (auto simp add: Cons_eq_append_conv)

section {* Kleene Star for Languages *}

inductive_set

  Star :: "string set \<Rightarrow> string set" ("_\<star>" [101] 102)

  for A :: "string set"

where

  start[intro]: "[] \<in> A\<star>"

| step[intro]:  "\<lbrakk>s1 \<in> A; s2 \<in> A\<star>\<rbrakk> \<Longrightarrow> s1 @ s2 \<in> A\<star>"

lemma star_cases:

  shows "A\<star> = {[]} \<union> A ;; A\<star>"

unfolding Sequ_def

by (auto) (metis Star.simps)

lemma star_decomp: 

  assumes a: "c # x \<in> A\<star>" 

  shows "\<exists>a b. x = a @ b \<and> c # a \<in> A \<and> b \<in> A\<star>"

using a

by (induct x\<equiv>"c # x" rule: Star.induct) 

   (auto simp add: append_eq_Cons_conv)

lemma Der_star [simp]:

  shows "Der c (A\<star>) = (Der c A) ;; A\<star>"

proof -    

  have "Der c (A\<star>) = Der c ({[]} \<union> A ;; A\<star>)"

    by (simp only: star_cases[symmetric])

  also have "... = Der c (A ;; A\<star>)"

    by (simp only: Der_union Der_empty) (simp)

  also have "... = (Der c A) ;; A\<star> \<union> (if [] \<in> A then Der c (A\<star>) else {})"

    by simp

  also have "... =  (Der c A) ;; A\<star>"

    unfolding Sequ_def Der_def

    by (auto dest: star_decomp)

  finally show "Der c (A\<star>) = (Der c A) ;; A\<star>" .

qed

section {* Regular Expressions *}

datatype rexp =

  ZERO

| ONE

| CHAR char

| SEQ rexp rexp

| ALT rexp rexp

| STAR rexp

section {* Semantics of Regular Expressions *}

fun

  L :: "rexp \<Rightarrow> string set"

where

  "L (ZERO) = {}"

| "L (ONE) = {[]}"

| "L (CHAR c) = {[c]}"

| "L (SEQ r1 r2) = (L r1) ;; (L r2)"

| "L (ALT r1 r2) = (L r1) \<union> (L r2)"

| "L (STAR r) = (L r)\<star>"

section {* Nullable, Derivatives *}

fun

 nullable :: "rexp \<Rightarrow> bool"

where

  "nullable (ZERO) = False"

| "nullable (ONE) = True"

| "nullable (CHAR c) = False"

| "nullable (ALT r1 r2) = (nullable r1 \<or> nullable r2)"

| "nullable (SEQ r1 r2) = (nullable r1 \<and> nullable r2)"

| "nullable (STAR r) = True"

fun

 der :: "char \<Rightarrow> rexp \<Rightarrow> rexp"

where

  "der c (ZERO) = ZERO"

| "der c (ONE) = ZERO"

| "der c (ALT r1 r2) = ALT (der c r1) (der c r2)"

| "der c (SEQ r1 r2) = 

     (if nullable r1

      then ALT (SEQ (der c r1) r2) (der c r2)

      else SEQ (der c r1) r2)"

| "der c (STAR r) = SEQ (der c r) (STAR r)"

fun 

 ders :: "string \<Rightarrow> rexp \<Rightarrow> rexp"

where

  "ders [] r = r"

| "ders (c # s) r = ders s (der c r)"

lemma nullable_correctness:

  shows "nullable r  \<longleftrightarrow> [] \<in> (L r)"

by (induct r) (auto simp add: Sequ_def) 

lemma der_correctness:

  shows "L (der c r) = Der c (L r)"

apply(induct r) 

apply(simp_all add: nullable_correctness)

done

section {* Values *}

datatype val = 

  Void

| Char char

| Seq val val

| Right val

| Left val

| Stars "val list"

datatype_compat val

section {* The string behind a value *}

fun 

  flat :: "val \<Rightarrow> string"

where

  "flat (Void) = []"

| "flat (Char c) = [c]"

| "flat (Left v) = flat v"

| "flat (Right v) = flat v"

| "flat (Seq v1 v2) = (flat v1) @ (flat v2)"

| "flat (Stars []) = []"

| "flat (Stars (v#vs)) = (flat v) @ (flat (Stars vs))" 

lemma flat_Stars [simp]:

 "flat (Stars vs) = concat (map flat vs)"

by (induct vs) (auto)

section {* Relation between values and regular expressions *}

inductive 

  Prf :: "val \<Rightarrow> rexp \<Rightarrow> bool" ("\<turnstile> _ : _" [100, 100] 100)

where

 "\<lbrakk>\<turnstile> v1 : r1; \<turnstile> v2 : r2\<rbrakk> \<Longrightarrow> \<turnstile> Seq v1 v2 : SEQ r1 r2"

| "\<turnstile> v1 : r1 \<Longrightarrow> \<turnstile> Left v1 : ALT r1 r2"

| "\<turnstile> v2 : r2 \<Longrightarrow> \<turnstile> Right v2 : ALT r1 r2"

| "\<turnstile> Void : ONE"

| "\<turnstile> Char c : CHAR c"

| "\<turnstile> Stars [] : STAR r"

| "\<lbrakk>\<turnstile> v : r; \<turnstile> Stars vs : STAR r\<rbrakk> \<Longrightarrow> \<turnstile> Stars (v # vs) : STAR r"

lemma not_nullable_flat:

  assumes "\<turnstile> v : r" "\<not> nullable r"

  shows "flat v \<noteq> []"

using assms

by (induct) (auto)

lemma Prf_flat_L:

  assumes "\<turnstile> v : r" shows "flat v \<in> L r"

using assms

apply(induct v r rule: Prf.induct)

apply(auto simp add: Sequ_def)

done

lemma Prf_Stars:

  assumes "\<forall>v \<in> set vs. \<turnstile> v : r"

  shows "\<turnstile> Stars vs : STAR r"

using assms

apply(induct vs)

apply (metis Prf.intros(6))

by (metis Prf.intros(7) insert_iff set_simps(2))

lemma Star_string:

  assumes "s \<in> A\<star>"

  shows "\<exists>ss. concat ss = s \<and> (\<forall>s \<in> set ss. s \<in> A)"

using assms

apply(induct rule: Star.induct)

apply(auto)

apply(rule_tac x="[]" in exI)

apply(simp)

apply(rule_tac x="s1#ss" in exI)

apply(simp)

done

lemma Star_val:

  assumes "\<forall>s\<in>set ss. \<exists>v. s = flat v \<and> \<turnstile> v : r"

  shows "\<exists>vs. concat (map flat vs) = concat ss \<and> (\<forall>v\<in>set vs. \<turnstile> v : r)"

using assms

apply(induct ss)

apply(auto)

apply (metis empty_iff list.set(1))

by (metis concat.simps(2) list.simps(9) set_ConsD)

lemma L_flat_Prf:

  "L(r) = {flat v | v. \<turnstile> v : r}"

apply(induct r)

apply(auto dest: Prf_flat_L simp add: Sequ_def)

apply (metis Prf.intros(4) flat.simps(1))

apply (metis Prf.intros(5) flat.simps(2))

apply (metis Prf.intros(1) flat.simps(5))

apply (metis Prf.intros(2) flat.simps(3))

apply (metis Prf.intros(3) flat.simps(4))

apply(erule Prf.cases)

apply(auto)

apply(subgoal_tac "\<exists>vs::val list. concat (map flat vs) = x \<and> (\<forall>v \<in> set vs. \<turnstile> v : r)")

apply(auto)[1]

apply(rule_tac x="Stars vs" in exI)

apply(simp)

apply(rule Prf_Stars)

apply(simp)

apply(drule Star_string)

apply(auto)

apply(rule Star_val)

apply(simp)

done

section {* Sulzmann functions *}

fun 

  mkeps :: "rexp \<Rightarrow> val"

where

  "mkeps(ONE) = Void"

| "mkeps(SEQ r1 r2) = Seq (mkeps r1) (mkeps r2)"

| "mkeps(ALT r1 r2) = (if nullable(r1) then Left (mkeps r1) else Right (mkeps r2))"

| "mkeps(STAR r) = Stars []"

fun injval :: "rexp \<Rightarrow> char \<Rightarrow> val \<Rightarrow> val"

where

  "injval (CHAR d) c Void = Char d"

| "injval (ALT r1 r2) c (Left v1) = Left(injval r1 c v1)"

| "injval (ALT r1 r2) c (Right v2) = Right(injval r2 c v2)"

| "injval (SEQ r1 r2) c (Seq v1 v2) = Seq (injval r1 c v1) v2"

| "injval (SEQ r1 r2) c (Left (Seq v1 v2)) = Seq (injval r1 c v1) v2"

| "injval (SEQ r1 r2) c (Right v2) = Seq (mkeps r1) (injval r2 c v2)"

| "injval (STAR r) c (Seq v (Stars vs)) = Stars ((injval r c v) # vs)" 

section {* Matcher *}

fun 

  matcher :: "rexp \<Rightarrow> string \<Rightarrow> val option"

where

  "matcher r [] = (if nullable r then Some(mkeps r) else None)"

| "matcher r (c#s) = (case (matcher (der c r) s) of  

                    None \<Rightarrow> None

                  | Some(v) \<Rightarrow> Some(injval r c v))"

fun 

  matcher2 :: "rexp \<Rightarrow> string \<Rightarrow> val"

where

  "matcher2 r [] = mkeps r"

| "matcher2 r (c#s) = injval r c (matcher2 (der c r) s)"

section {* Mkeps, injval *}

lemma mkeps_nullable:

  assumes "nullable(r)" 

  shows "\<turnstile> mkeps r : r"

using assms

apply(induct rule: nullable.induct)

apply(auto intro: Prf.intros)

done

lemma mkeps_flat:

  assumes "nullable(r)" 

  shows "flat (mkeps r) = []"

using assms

apply(induct rule: nullable.induct)

apply(auto)

done

lemma Prf_injval:

  assumes "\<turnstile> v : der c r" 

  shows "\<turnstile> (injval r c v) : r"

using assms

apply(induct r arbitrary: c v rule: rexp.induct)

apply(simp_all)

(* ZERO *)

apply(erule Prf.cases)

apply(simp_all)[7]

(* ONE *)

apply(erule Prf.cases)

apply(simp_all)[7]

(* CHAR *)

apply(case_tac "c = x")

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(rule Prf.intros(5))

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

(* SEQ *)

apply(case_tac "nullable x1")

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(auto)[1]

apply(erule Prf.cases)

apply(simp_all)[7]

apply(auto)[1]

apply(rule Prf.intros)

apply(auto)[2]

apply (metis Prf.intros(1) mkeps_nullable)

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(auto)[1]

apply(rule Prf.intros)

apply(auto)[2]

(* ALT *)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(clarify)

apply (metis Prf.intros(2))

apply (metis Prf.intros(3))

(* STAR *)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(clarify)

apply(rotate_tac 2)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(auto)

apply (metis Prf.intros(6) Prf.intros(7))

by (metis Prf.intros(7))

lemma Prf_injval_flat:

  assumes "\<turnstile> v : der c r" 

  shows "flat (injval r c v) = c # (flat v)"

using assms

apply(induct arbitrary: v rule: der.induct)

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(simp)

apply(simp)

apply(auto)[1]

apply(erule Prf.cases)

apply(simp_all)[7]

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(simp)

apply(case_tac "nullable r1")

apply(simp)

apply(erule Prf.cases)

apply(simp_all (no_asm_use))[7]

apply(auto)[1]

apply(erule Prf.cases)

apply(simp_all)[7]

apply(clarify)

apply(simp only: injval.simps flat.simps)

apply(auto)[1]

apply (metis mkeps_flat)

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(simp)

apply(erule Prf.cases)

apply(simp_all)[7]

apply(auto)

apply(rotate_tac 2)

apply(erule Prf.cases)

apply(simp_all)[7]

done

section {* Our Alternative Posix definition *}

inductive 

  PMatch :: "string \<Rightarrow> rexp \<Rightarrow> val \<Rightarrow> bool" ("_ \<in> _ \<rightarrow> _" [100, 100, 100] 100)

where

  "[] \<in> ONE \<rightarrow> Void"

| "[c] \<in> (CHAR c) \<rightarrow> (Char c)"

| "s \<in> r1 \<rightarrow> v \<Longrightarrow> s \<in> (ALT r1 r2) \<rightarrow> (Left v)"

| "\<lbrakk>s \<in> r2 \<rightarrow> v; s \<notin> L(r1)\<rbrakk> \<Longrightarrow> s \<in> (ALT r1 r2) \<rightarrow> (Right v)"

| "\<lbrakk>s1 \<in> r1 \<rightarrow> v1; s2 \<in> r2 \<rightarrow> v2;

    \<not>(\<exists>s\<^sub>3 s\<^sub>4. s\<^sub>3 \<noteq> [] \<and> s\<^sub>3 @ s\<^sub>4 = s2 \<and> (s1 @ s\<^sub>3) \<in> L r1 \<and> s\<^sub>4 \<in> L r2)\<rbrakk> \<Longrightarrow> 

    (s1 @ s2) \<in> (SEQ r1 r2) \<rightarrow> (Seq v1 v2)"

| "\<lbrakk>s1 \<in> r \<rightarrow> v; s2 \<in> STAR r \<rightarrow> Stars vs; flat v \<noteq> [];

    \<not>(\<exists>s\<^sub>3 s\<^sub>4. s\<^sub>3 \<noteq> [] \<and> s\<^sub>3 @ s\<^sub>4 = s2 \<and> (s1 @ s\<^sub>3) \<in> L r \<and> s\<^sub>4 \<in> L (STAR r))\<rbrakk>

    \<Longrightarrow> (s1 @ s2) \<in> STAR r \<rightarrow> Stars (v # vs)"

| "[] \<in> STAR r \<rightarrow> Stars []"

lemma PMatch1:

  assumes "s \<in> r \<rightarrow> v"

  shows "\<turnstile> v : r" "flat v = s"

using assms

apply(induct s r v rule: PMatch.induct)

apply(auto intro: Prf.intros)

done

lemma PMatch_mkeps:

  assumes "nullable r"

  shows "[] \<in> r \<rightarrow> mkeps r"

using assms

apply(induct r)

apply(auto)

apply (metis PMatch.intros(1))

apply(subst append.simps(1)[symmetric])

apply (rule PMatch.intros)

apply(simp)

apply(simp)

apply(auto)[1]

apply (rule PMatch.intros)

apply(simp)

apply (rule PMatch.intros)

apply(simp)

apply (rule PMatch.intros)

apply(simp)

apply (metis nullable_correctness)

apply(metis PMatch.intros(7))

done

find_theorems Stars

thm compat_val_list.induct compat_val.induct

lemma PMatch_determ:

  shows "\<lbrakk>s \<in> r \<rightarrow> v1; s \<in> r \<rightarrow> v2\<rbrakk> \<Longrightarrow> v1 = v2"

  and   "\<lbrakk>s \<in> (STAR r) \<rightarrow> Stars vs1; s \<in> (STAR r) \<rightarrow> Stars vs2\<rbrakk> \<Longrightarrow> vs1 = vs2"

apply(induct v1 and vs1 arbitrary: s r v2 and s r vs2 rule: compat_val.induct compat_val_list.induct)

apply(erule PMatch.cases)

apply(simp_all)[7]

apply(erule PMatch.cases)

apply(simp_all)[7]

apply(erule PMatch.cases)

apply(simp_all)[7]

apply(erule PMatch.cases)