# HG changeset patch # User Chengsong # Date 1643933112 0 # Node ID f71df68776bb8969985a20a3439dfd50f04cb42e # Parent 01d1285b08ed61a7107f80ead6a4633460a7dd42 5ct diff -r 01d1285b08ed -r f71df68776bb thys2/SizeBound4.thy --- a/thys2/SizeBound4.thy Wed Feb 02 22:30:41 2022 +0000 +++ b/thys2/SizeBound4.thy Fri Feb 04 00:05:12 2022 +0000 @@ -1444,6 +1444,40 @@ using bsimp_ASEQ_idem apply presburger oops +lemma neg: + shows " \(\r2. r1 \ r2 \ (r2 \* bsimp r1) )" + apply(rule notI) + apply(erule exE) + apply(erule conjE) + oops + + + + +lemma reduction_always_in_bsimp: + shows " \ r1 \ r2 ; \(r2 \* bsimp r1)\ \ False" + apply(erule rrewrite.cases) + apply simp + apply auto + + oops + +(* +AALTs [] [AZERO, AALTs(bs1, [a, b]) ] +rewrite seq 1: \ AALTs [] [ AALTs(bs1, [a, b]) ] \ +fuse [] (AALTs bs1, [a, b]) +rewrite seq 2: \ AALTs [] [AZERO, (fuse bs1 a), (fuse bs1 b)]) ] + +*) + +lemma normal_bsimp: + shows "\r'. bsimp r \ r'" + oops + + (*r' size bsimp r > size r' + r' \* bsimp bsimp r +size bsimp r > size r' \ size bsimp bsimp r*) + export_code blexer_simp blexer lexer bders bders_simp in Scala module_name VerifiedLexers diff -r 01d1285b08ed -r f71df68776bb thys2/SizeBound4CT.thy --- a/thys2/SizeBound4CT.thy Wed Feb 02 22:30:41 2022 +0000 +++ b/thys2/SizeBound4CT.thy Fri Feb 04 00:05:12 2022 +0000 @@ -1156,6 +1156,16 @@ using bsimp_ASEQ_idem apply presburger oops + +lemma normal_form: + shows "\r. \ r'. bsimp r \ r'" + + oops + +lemma another_normal: + shows "\r'. bsimp r \ r'" + oops + export_code blexer_simp blexer lexer bders bders_simp in Scala module_name VerifiedLexers diff -r 01d1285b08ed -r f71df68776bb thys2/SizeBound5CT.thy --- /dev/null Thu Jan 01 00:00:00 1970 +0000 +++ b/thys2/SizeBound5CT.thy Fri Feb 04 00:05:12 2022 +0000 @@ -0,0 +1,1493 @@ + +theory SizeBound5CT + imports "Lexer" "PDerivs" +begin + +section \Bit-Encodings\ + +datatype bit = Z | S + +fun code :: "val \ bit list" +where + "code Void = []" +| "code (Char c) = []" +| "code (Left v) = Z # (code v)" +| "code (Right v) = S # (code v)" +| "code (Seq v1 v2) = (code v1) @ (code v2)" +| "code (Stars []) = [S]" +| "code (Stars (v # vs)) = (Z # code v) @ code (Stars vs)" + + +fun + Stars_add :: "val \ val \ val" +where + "Stars_add v (Stars vs) = Stars (v # vs)" + +function + decode' :: "bit list \ rexp \ (val * bit list)" +where + "decode' bs ZERO = (undefined, bs)" +| "decode' bs ONE = (Void, bs)" +| "decode' bs (CH d) = (Char d, bs)" +| "decode' [] (ALT r1 r2) = (Void, [])" +| "decode' (Z # bs) (ALT r1 r2) = (let (v, bs') = decode' bs r1 in (Left v, bs'))" +| "decode' (S # bs) (ALT r1 r2) = (let (v, bs') = decode' bs r2 in (Right v, bs'))" +| "decode' bs (SEQ r1 r2) = (let (v1, bs') = decode' bs r1 in + let (v2, bs'') = decode' bs' r2 in (Seq v1 v2, bs''))" +| "decode' [] (STAR r) = (Void, [])" +| "decode' (S # bs) (STAR r) = (Stars [], bs)" +| "decode' (Z # bs) (STAR r) = (let (v, bs') = decode' bs r in + let (vs, bs'') = decode' bs' (STAR r) + in (Stars_add v vs, bs''))" +by pat_completeness auto + +lemma decode'_smaller: + assumes "decode'_dom (bs, r)" + shows "length (snd (decode' bs r)) \ length bs" +using assms +apply(induct bs r) +apply(auto simp add: decode'.psimps split: prod.split) +using dual_order.trans apply blast +by (meson dual_order.trans le_SucI) + +termination "decode'" +apply(relation "inv_image (measure(%cs. size cs) <*lex*> measure(%s. size s)) (%(ds,r). (r,ds))") +apply(auto dest!: decode'_smaller) +by (metis less_Suc_eq_le snd_conv) + +definition + decode :: "bit list \ rexp \ val option" +where + "decode ds r \ (let (v, ds') = decode' ds r + in (if ds' = [] then Some v else None))" + +lemma decode'_code_Stars: + assumes "\v\set vs. \ v : r \ (\x. decode' (code v @ x) r = (v, x)) \ flat v \ []" + shows "decode' (code (Stars vs) @ ds) (STAR r) = (Stars vs, ds)" + using assms + apply(induct vs) + apply(auto) + done + +lemma decode'_code: + assumes "\ v : r" + shows "decode' ((code v) @ ds) r = (v, ds)" +using assms + apply(induct v r arbitrary: ds) + apply(auto) + using decode'_code_Stars by blast + +lemma decode_code: + assumes "\ v : r" + shows "decode (code v) r = Some v" + using assms unfolding decode_def + by (smt append_Nil2 decode'_code old.prod.case) + + +section {* Annotated Regular Expressions *} + +datatype arexp = + AZERO +| AONE "bit list" +| ACHAR "bit list" char +| ASEQ "bit list" arexp arexp +| AALTs "bit list" "arexp list" +| ASTAR "bit list" arexp + +abbreviation + "AALT bs r1 r2 \ AALTs bs [r1, r2]" + +fun asize :: "arexp \ nat" where + "asize AZERO = 1" +| "asize (AONE cs) = 1" +| "asize (ACHAR cs c) = 1" +| "asize (AALTs cs rs) = Suc (sum_list (map asize rs))" +| "asize (ASEQ cs r1 r2) = Suc (asize r1 + asize r2)" +| "asize (ASTAR cs r) = Suc (asize r)" + +fun + erase :: "arexp \ rexp" +where + "erase AZERO = ZERO" +| "erase (AONE _) = ONE" +| "erase (ACHAR _ c) = CH c" +| "erase (AALTs _ []) = ZERO" +| "erase (AALTs _ [r]) = (erase r)" +| "erase (AALTs bs (r#rs)) = ALT (erase r) (erase (AALTs bs rs))" +| "erase (ASEQ _ r1 r2) = SEQ (erase r1) (erase r2)" +| "erase (ASTAR _ r) = STAR (erase r)" + + +fun fuse :: "bit list \ arexp \ arexp" where + "fuse bs AZERO = AZERO" +| "fuse bs (AONE cs) = AONE (bs @ cs)" +| "fuse bs (ACHAR cs c) = ACHAR (bs @ cs) c" +| "fuse bs (AALTs cs rs) = AALTs (bs @ cs) rs" +| "fuse bs (ASEQ cs r1 r2) = ASEQ (bs @ cs) r1 r2" +| "fuse bs (ASTAR cs r) = ASTAR (bs @ cs) r" + +lemma fuse_append: + shows "fuse (bs1 @ bs2) r = fuse bs1 (fuse bs2 r)" + apply(induct r) + apply(auto) + done + + +fun intern :: "rexp \ arexp" where + "intern ZERO = AZERO" +| "intern ONE = AONE []" +| "intern (CH c) = ACHAR [] c" +| "intern (ALT r1 r2) = AALT [] (fuse [Z] (intern r1)) + (fuse [S] (intern r2))" +| "intern (SEQ r1 r2) = ASEQ [] (intern r1) (intern r2)" +| "intern (STAR r) = ASTAR [] (intern r)" + + +fun retrieve :: "arexp \ val \ bit list" where + "retrieve (AONE bs) Void = bs" +| "retrieve (ACHAR bs c) (Char d) = bs" +| "retrieve (AALTs bs [r]) v = bs @ retrieve r v" +| "retrieve (AALTs bs (r#rs)) (Left v) = bs @ retrieve r v" +| "retrieve (AALTs bs (r#rs)) (Right v) = bs @ retrieve (AALTs [] rs) v" +| "retrieve (ASEQ bs r1 r2) (Seq v1 v2) = bs @ retrieve r1 v1 @ retrieve r2 v2" +| "retrieve (ASTAR bs r) (Stars []) = bs @ [S]" +| "retrieve (ASTAR bs r) (Stars (v#vs)) = + bs @ [Z] @ retrieve r v @ retrieve (ASTAR [] r) (Stars vs)" + + + +fun + bnullable :: "arexp \ bool" +where + "bnullable (AZERO) = False" +| "bnullable (AONE bs) = True" +| "bnullable (ACHAR bs c) = False" +| "bnullable (AALTs bs rs) = (\r \ set rs. bnullable r)" +| "bnullable (ASEQ bs r1 r2) = (bnullable r1 \ bnullable r2)" +| "bnullable (ASTAR bs r) = True" + +abbreviation + bnullables :: "arexp list \ bool" +where + "bnullables rs \ (\r \ set rs. bnullable r)" + +fun + bmkeps :: "arexp \ bit list" and + bmkepss :: "arexp list \ bit list" +where + "bmkeps(AONE bs) = bs" +| "bmkeps(ASEQ bs r1 r2) = bs @ (bmkeps r1) @ (bmkeps r2)" +| "bmkeps(AALTs bs rs) = bs @ (bmkepss rs)" +| "bmkeps(ASTAR bs r) = bs @ [S]" +| "bmkepss [] = []" +| "bmkepss (r # rs) = (if bnullable(r) then (bmkeps r) else (bmkepss rs))" + +lemma bmkepss1: + assumes "\ bnullables rs1" + shows "bmkepss (rs1 @ rs2) = bmkepss rs2" + using assms + by (induct rs1) (auto) + +lemma bmkepss2: + assumes "bnullables rs1" + shows "bmkepss (rs1 @ rs2) = bmkepss rs1" + using assms + by (induct rs1) (auto) + + +fun + bder :: "char \ arexp \ arexp" +where + "bder c (AZERO) = AZERO" +| "bder c (AONE bs) = AZERO" +| "bder c (ACHAR bs d) = (if c = d then AONE bs else AZERO)" +| "bder c (AALTs bs rs) = AALTs bs (map (bder c) rs)" +| "bder c (ASEQ bs r1 r2) = + (if bnullable r1 + then AALT bs (ASEQ [] (bder c r1) r2) (fuse (bmkeps r1) (bder c r2)) + else ASEQ bs (bder c r1) r2)" +| "bder c (ASTAR bs r) = ASEQ bs (fuse [Z] (bder c r)) (ASTAR [] r)" + + +fun + bders :: "arexp \ string \ arexp" +where + "bders r [] = r" +| "bders r (c#s) = bders (bder c r) s" + +lemma bders_append: + "bders c (s1 @ s2) = bders (bders c s1) s2" + apply(induct s1 arbitrary: c s2) + apply(simp_all) + done + +lemma bnullable_correctness: + shows "nullable (erase r) = bnullable r" + apply(induct r rule: erase.induct) + apply(simp_all) + done + +lemma erase_fuse: + shows "erase (fuse bs r) = erase r" + apply(induct r rule: erase.induct) + apply(simp_all) + done + +lemma erase_intern [simp]: + shows "erase (intern r) = r" + apply(induct r) + apply(simp_all add: erase_fuse) + done + +lemma erase_bder [simp]: + shows "erase (bder a r) = der a (erase r)" + apply(induct r rule: erase.induct) + apply(simp_all add: erase_fuse bnullable_correctness) + done + +lemma erase_bders [simp]: + shows "erase (bders r s) = ders s (erase r)" + apply(induct s arbitrary: r ) + apply(simp_all) + done + +lemma bnullable_fuse: + shows "bnullable (fuse bs r) = bnullable r" + apply(induct r arbitrary: bs) + apply(auto) + done + +lemma retrieve_encode_STARS: + assumes "\v\set vs. \ v : r \ code v = retrieve (intern r) v" + shows "code (Stars vs) = retrieve (ASTAR [] (intern r)) (Stars vs)" + using assms + apply(induct vs) + apply(simp_all) + done + +lemma retrieve_fuse2: + assumes "\ v : (erase r)" + shows "retrieve (fuse bs r) v = bs @ retrieve r v" + using assms + apply(induct r arbitrary: v bs) + apply(auto elim: Prf_elims)[4] + apply(case_tac x2a) + apply(simp) + using Prf_elims(1) apply blast + apply(case_tac x2a) + apply(simp) + apply(simp) + apply(case_tac list) + apply(simp) + apply(simp) + apply (smt (verit, best) Prf_elims(3) append_assoc retrieve.simps(4) retrieve.simps(5)) + apply(simp) + using retrieve_encode_STARS + apply(auto elim!: Prf_elims)[1] + apply(case_tac vs) + apply(simp) + apply(simp) + done + +lemma retrieve_fuse: + assumes "\ v : r" + shows "retrieve (fuse bs (intern r)) v = bs @ retrieve (intern r) v" + using assms + by (simp_all add: retrieve_fuse2) + + +lemma retrieve_code: + assumes "\ v : r" + shows "code v = retrieve (intern r) v" + using assms + apply(induct v r ) + apply(simp_all add: retrieve_fuse retrieve_encode_STARS) + done + + +lemma retrieve_AALTs_bnullable1: + assumes "bnullable r" + shows "retrieve (AALTs bs (r # rs)) (mkeps (erase (AALTs bs (r # rs)))) + = bs @ retrieve r (mkeps (erase r))" + using assms + apply(case_tac rs) + apply(auto simp add: bnullable_correctness) + done + +lemma retrieve_AALTs_bnullable2: + assumes "\bnullable r" "bnullables rs" + shows "retrieve (AALTs bs (r # rs)) (mkeps (erase (AALTs bs (r # rs)))) + = retrieve (AALTs bs rs) (mkeps (erase (AALTs bs rs)))" + using assms + apply(induct rs arbitrary: r bs) + apply(auto) + using bnullable_correctness apply blast + apply(case_tac rs) + apply(auto) + using bnullable_correctness apply blast + apply(case_tac rs) + apply(auto) + done + +lemma bmkeps_retrieve_AALTs: + assumes "\r \ set rs. bnullable r \ bmkeps r = retrieve r (mkeps (erase r))" + "bnullables rs" + shows "bs @ bmkepss rs = retrieve (AALTs bs rs) (mkeps (erase (AALTs bs rs)))" + using assms + apply(induct rs arbitrary: bs) + apply(auto) + using retrieve_AALTs_bnullable1 apply presburger + apply (metis retrieve_AALTs_bnullable2) + apply (simp add: retrieve_AALTs_bnullable1) + by (metis retrieve_AALTs_bnullable2) + + +lemma bmkeps_retrieve: + assumes "bnullable r" + shows "bmkeps r = retrieve r (mkeps (erase r))" + using assms + apply(induct r) + apply(auto) + using bmkeps_retrieve_AALTs by auto + +lemma bder_retrieve: + assumes "\ v : der c (erase r)" + shows "retrieve (bder c r) v = retrieve r (injval (erase r) c v)" + using assms + apply(induct r arbitrary: v rule: erase.induct) + using Prf_elims(1) apply auto[1] + using Prf_elims(1) apply auto[1] + apply(auto)[1] + apply (metis Prf_elims(4) injval.simps(1) retrieve.simps(1) retrieve.simps(2)) + using Prf_elims(1) apply blast + (* AALTs case *) + apply(simp) + apply(erule Prf_elims) + apply(simp) + apply(simp) + apply(rename_tac "r\<^sub>1" "r\<^sub>2" rs v) + apply(erule Prf_elims) + apply(simp) + apply(simp) + apply(case_tac rs) + apply(simp) + apply(simp) + using Prf_elims(3) apply fastforce + (* ASEQ case *) + apply(simp) + apply(case_tac "nullable (erase r1)") + apply(simp) + apply(erule Prf_elims) + using Prf_elims(2) bnullable_correctness apply force + apply (simp add: bmkeps_retrieve bnullable_correctness retrieve_fuse2) + apply (simp add: bmkeps_retrieve bnullable_correctness retrieve_fuse2) + using Prf_elims(2) apply force + (* ASTAR case *) + apply(rename_tac bs r v) + apply(simp) + apply(erule Prf_elims) + apply(clarify) + apply(erule Prf_elims) + apply(clarify) + by (simp add: retrieve_fuse2) + + +lemma MAIN_decode: + assumes "\ v : ders s r" + shows "Some (flex r id s v) = decode (retrieve (bders (intern r) s) v) r" + using assms +proof (induct s arbitrary: v rule: rev_induct) + case Nil + have "\ v : ders [] r" by fact + then have "\ v : r" by simp + then have "Some v = decode (retrieve (intern r) v) r" + using decode_code retrieve_code by auto + then show "Some (flex r id [] v) = decode (retrieve (bders (intern r) []) v) r" + by simp +next + case (snoc c s v) + have IH: "\v. \ v : ders s r \ + Some (flex r id s v) = decode (retrieve (bders (intern r) s) v) r" by fact + have asm: "\ v : ders (s @ [c]) r" by fact + then have asm2: "\ injval (ders s r) c v : ders s r" + by (simp add: Prf_injval ders_append) + have "Some (flex r id (s @ [c]) v) = Some (flex r id s (injval (ders s r) c v))" + by (simp add: flex_append) + also have "... = decode (retrieve (bders (intern r) s) (injval (ders s r) c v)) r" + using asm2 IH by simp + also have "... = decode (retrieve (bder c (bders (intern r) s)) v) r" + using asm by (simp_all add: bder_retrieve ders_append) + finally show "Some (flex r id (s @ [c]) v) = + decode (retrieve (bders (intern r) (s @ [c])) v) r" by (simp add: bders_append) +qed + +definition blexer where + "blexer r s \ if bnullable (bders (intern r) s) then + decode (bmkeps (bders (intern r) s)) r else None" + +lemma blexer_correctness: + shows "blexer r s = lexer r s" +proof - + { define bds where "bds \ bders (intern r) s" + define ds where "ds \ ders s r" + assume asm: "nullable ds" + have era: "erase bds = ds" + unfolding ds_def bds_def by simp + have mke: "\ mkeps ds : ds" + using asm by (simp add: mkeps_nullable) + have "decode (bmkeps bds) r = decode (retrieve bds (mkeps ds)) r" + using bmkeps_retrieve + using asm era + using bnullable_correctness by force + also have "... = Some (flex r id s (mkeps ds))" + using mke by (simp_all add: MAIN_decode ds_def bds_def) + finally have "decode (bmkeps bds) r = Some (flex r id s (mkeps ds))" + unfolding bds_def ds_def . + } + then show "blexer r s = lexer r s" + unfolding blexer_def lexer_flex + by (auto simp add: bnullable_correctness[symmetric]) +qed + + +fun distinctBy :: "'a list \ ('a \ 'b) \ 'b set \ 'a list" + where + "distinctBy [] f acc = []" +| "distinctBy (x#xs) f acc = + (if (f x) \ acc then distinctBy xs f acc + else x # (distinctBy xs f ({f x} \ acc)))" + + + +fun flts :: "arexp list \ arexp list" + where + "flts [] = []" +| "flts (AZERO # rs) = flts rs" +| "flts ((AALTs bs rs1) # rs) = (map (fuse bs) rs1) @ flts rs" +| "flts (r1 # rs) = r1 # flts rs" + + + +fun bsimp_ASEQ :: "bit list \ arexp \ arexp \ arexp" + where + "bsimp_ASEQ _ AZERO _ = AZERO" +| "bsimp_ASEQ _ _ AZERO = AZERO" +| "bsimp_ASEQ bs1 (AONE bs2) r2 = fuse (bs1 @ bs2) r2" +| "bsimp_ASEQ bs1 r1 r2 = ASEQ bs1 r1 r2" + +lemma bsimp_ASEQ0[simp]: + shows "bsimp_ASEQ bs r1 AZERO = AZERO" + by (case_tac r1)(simp_all) + +lemma bsimp_ASEQ1: + assumes "r1 \ AZERO" "r2 \ AZERO" "\bs. r1 = AONE bs" + shows "bsimp_ASEQ bs r1 r2 = ASEQ bs r1 r2" + using assms + apply(induct bs r1 r2 rule: bsimp_ASEQ.induct) + apply(auto) + done + +lemma bsimp_ASEQ2[simp]: + shows "bsimp_ASEQ bs1 (AONE bs2) r2 = fuse (bs1 @ bs2) r2" + by (case_tac r2) (simp_all) + + +fun bsimp_AALTs :: "bit list \ arexp list \ arexp" + where + "bsimp_AALTs _ [] = AZERO" +| "bsimp_AALTs bs1 [r] = fuse bs1 r" +| "bsimp_AALTs bs1 rs = AALTs bs1 rs" + + +fun bsimp :: "arexp \ arexp" + where + "bsimp (ASEQ bs1 r1 r2) = bsimp_ASEQ bs1 (bsimp r1) (bsimp r2)" +| "bsimp (AALTs bs1 rs) = bsimp_AALTs bs1 (distinctBy (flts (map bsimp rs)) erase {}) " +| "bsimp r = r" + + +fun + bders_simp :: "arexp \ string \ arexp" +where + "bders_simp r [] = r" +| "bders_simp r (c # s) = bders_simp (bsimp (bder c r)) s" + +definition blexer_simp where + "blexer_simp r s \ if bnullable (bders_simp (intern r) s) then + decode (bmkeps (bders_simp (intern r) s)) r else None" + + + +lemma bders_simp_append: + shows "bders_simp r (s1 @ s2) = bders_simp (bders_simp r s1) s2" + apply(induct s1 arbitrary: r s2) + apply(simp_all) + done + + +lemma bmkeps_fuse: + assumes "bnullable r" + shows "bmkeps (fuse bs r) = bs @ bmkeps r" + using assms + by (metis bmkeps_retrieve bnullable_correctness erase_fuse mkeps_nullable retrieve_fuse2) + +lemma bmkepss_fuse: + assumes "bnullables rs" + shows "bmkepss (map (fuse bs) rs) = bs @ bmkepss rs" + using assms + apply(induct rs arbitrary: bs) + apply(auto simp add: bmkeps_fuse bnullable_fuse) + done + +lemma bder_fuse: + shows "bder c (fuse bs a) = fuse bs (bder c a)" + apply(induct a arbitrary: bs c) + apply(simp_all) + done + + + + +inductive + rrewrite:: "arexp \ arexp \ bool" ("_ \ _" [99, 99] 99) +and + srewrite:: "arexp list \ arexp list \ bool" (" _ s\ _" [100, 100] 100) +where + bs1: "ASEQ bs AZERO r2 \ AZERO" +| bs2: "ASEQ bs r1 AZERO \ AZERO" +| bs3: "ASEQ bs1 (AONE bs2) r \ fuse (bs1@bs2) r" +| bs4: "r1 \ r2 \ ASEQ bs r1 r3 \ ASEQ bs r2 r3" +| bs5: "r3 \ r4 \ ASEQ bs r1 r3 \ ASEQ bs r1 r4" +| bs6: "AALTs bs [] \ AZERO" +| bs7: "AALTs bs [r] \ fuse bs r" +| bs8: "rs1 s\ rs2 \ AALTs bs rs1 \ AALTs bs rs2" +(*| ss1: "[] s\ []"*) +| ss2: "rs1 s\ rs2 \ (r # rs1) s\ (r # rs2)" +| ss3: "r1 \ r2 \ (r1 # rs) s\ (r2 # rs)" +| ss4: "(AZERO # rs) s\ rs" +| ss5: "(AALTs bs1 rs1 # rsb) s\ ((map (fuse bs1) rs1) @ rsb)" +| ss6: "erase a1 = erase a2 \ (rsa@[a1]@rsb@[a2]@rsc) s\ (rsa@[a1]@rsb@rsc)" + + +inductive + rrewrites:: "arexp \ arexp \ bool" ("_ \* _" [100, 100] 100) +where + rs1[intro, simp]:"r \* r" +| rs2[intro]: "\r1 \* r2; r2 \ r3\ \ r1 \* r3" + +inductive + srewrites:: "arexp list \ arexp list \ bool" ("_ s\* _" [100, 100] 100) +where + sss1[intro, simp]:"rs s\* rs" +| sss2[intro]: "\rs1 s\ rs2; rs2 s\* rs3\ \ rs1 s\* rs3" + + +lemma r_in_rstar: + shows "r1 \ r2 \ r1 \* r2" + using rrewrites.intros(1) rrewrites.intros(2) by blast + +lemma rrewrites_trans[trans]: + assumes a1: "r1 \* r2" and a2: "r2 \* r3" + shows "r1 \* r3" + using a2 a1 + apply(induct r2 r3 arbitrary: r1 rule: rrewrites.induct) + apply(auto) + done + +lemma srewrites_trans[trans]: + assumes a1: "r1 s\* r2" and a2: "r2 s\* r3" + shows "r1 s\* r3" + using a1 a2 + apply(induct r1 r2 arbitrary: r3 rule: srewrites.induct) + apply(auto) + done + + + +lemma contextrewrites0: + "rs1 s\* rs2 \ AALTs bs rs1 \* AALTs bs rs2" + apply(induct rs1 rs2 rule: srewrites.inducts) + apply simp + using bs8 r_in_rstar rrewrites_trans by blast + +lemma contextrewrites1: + "r \* r' \ AALTs bs (r # rs) \* AALTs bs (r' # rs)" + apply(induct r r' rule: rrewrites.induct) + apply simp + using bs8 ss3 by blast + +lemma srewrite1: + shows "rs1 s\ rs2 \ (rs @ rs1) s\ (rs @ rs2)" + apply(induct rs) + apply(auto) + using ss2 by auto + +lemma srewrites1: + shows "rs1 s\* rs2 \ (rs @ rs1) s\* (rs @ rs2)" + apply(induct rs1 rs2 rule: srewrites.induct) + apply(auto) + using srewrite1 by blast + +lemma srewrite2: + shows "r1 \ r2 \ True" + and "rs1 s\ rs2 \ (rs1 @ rs) s\* (rs2 @ rs)" + apply(induct rule: rrewrite_srewrite.inducts) + apply(auto) + apply (metis append_Cons append_Nil srewrites1) + apply(meson srewrites.simps ss3) + apply (meson srewrites.simps ss4) + apply (meson srewrites.simps ss5) + by (metis append_Cons append_Nil srewrites.simps ss6) + + +lemma srewrites3: + shows "rs1 s\* rs2 \ (rs1 @ rs) s\* (rs2 @ rs)" + apply(induct rs1 rs2 arbitrary: rs rule: srewrites.induct) + apply(auto) + by (meson srewrite2(2) srewrites_trans) + +(* +lemma srewrites4: + assumes "rs3 s\* rs4" "rs1 s\* rs2" + shows "(rs1 @ rs3) s\* (rs2 @ rs4)" + using assms + apply(induct rs3 rs4 arbitrary: rs1 rs2 rule: srewrites.induct) + apply (simp add: srewrites3) + using srewrite1 by blast +*) + +lemma srewrites6: + assumes "r1 \* r2" + shows "[r1] s\* [r2]" + using assms + apply(induct r1 r2 rule: rrewrites.induct) + apply(auto) + by (meson srewrites.simps srewrites_trans ss3) + +lemma srewrites7: + assumes "rs3 s\* rs4" "r1 \* r2" + shows "(r1 # rs3) s\* (r2 # rs4)" + using assms + by (smt (verit, del_insts) append.simps srewrites1 srewrites3 srewrites6 srewrites_trans) + +lemma ss6_stronger_aux: + shows "(rs1 @ rs2) s\* (rs1 @ distinctBy rs2 erase (set (map erase rs1)))" + apply(induct rs2 arbitrary: rs1) + apply(auto) + apply (smt (verit, best) append.assoc append.right_neutral append_Cons append_Nil split_list srewrite2(2) srewrites_trans ss6) + apply(drule_tac x="rs1 @ [a]" in meta_spec) + apply(simp) + done + +lemma ss6_stronger: + shows "rs1 s\* distinctBy rs1 erase {}" + using ss6_stronger_aux[of "[]" _] by auto + +lemma rewrite_preserves_fuse: + shows "r2 \ r3 \ fuse bs r2 \ fuse bs r3" + and "rs2 s\ rs3 \ map (fuse bs) rs2 s\ map (fuse bs) rs3" +proof(induct rule: rrewrite_srewrite.inducts) + case (bs3 bs1 bs2 r) + then show "fuse bs (ASEQ bs1 (AONE bs2) r) \ fuse bs (fuse (bs1 @ bs2) r)" + by (metis fuse.simps(5) fuse_append rrewrite_srewrite.bs3) +next + case (bs7 bs1 r) + then show "fuse bs (AALTs bs1 [r]) \ fuse bs (fuse bs1 r)" + by (metis fuse.simps(4) fuse_append rrewrite_srewrite.bs7) +next + case (ss2 rs1 rs2 r) + then show "map (fuse bs) (r # rs1) s\ map (fuse bs) (r # rs2)" + by (simp add: rrewrite_srewrite.ss2) +next + case (ss3 r1 r2 rs) + then show "map (fuse bs) (r1 # rs) s\ map (fuse bs) (r2 # rs)" + by (simp add: rrewrite_srewrite.ss3) +next + case (ss5 bs1 rs1 rsb) + have "map (fuse bs) (AALTs bs1 rs1 # rsb) = AALTs (bs @ bs1) rs1 # (map (fuse bs) rsb)" by simp + also have "... s\ ((map (fuse (bs @ bs1)) rs1) @ (map (fuse bs) rsb))" + by (simp add: rrewrite_srewrite.ss5) + finally show "map (fuse bs) (AALTs bs1 rs1 # rsb) s\ map (fuse bs) (map (fuse bs1) rs1 @ rsb)" + by (simp add: comp_def fuse_append) +next + case (ss6 a1 a2 rsa rsb rsc) + then show "map (fuse bs) (rsa @ [a1] @ rsb @ [a2] @ rsc) s\ map (fuse bs) (rsa @ [a1] @ rsb @ rsc)" + apply(simp) + apply(rule rrewrite_srewrite.ss6[simplified]) + apply(simp add: erase_fuse) + done +qed (auto intro: rrewrite_srewrite.intros) + +lemma rewrites_fuse: + assumes "r1 \* r2" + shows "fuse bs r1 \* fuse bs r2" +using assms +apply(induction r1 r2 arbitrary: bs rule: rrewrites.induct) +apply(auto intro: rewrite_preserves_fuse) +done + + +lemma star_seq: + assumes "r1 \* r2" + shows "ASEQ bs r1 r3 \* ASEQ bs r2 r3" +using assms +apply(induct r1 r2 arbitrary: r3 rule: rrewrites.induct) +apply(auto intro: rrewrite_srewrite.intros) +done + +lemma star_seq2: + assumes "r3 \* r4" + shows "ASEQ bs r1 r3 \* ASEQ bs r1 r4" + using assms +apply(induct r3 r4 arbitrary: r1 rule: rrewrites.induct) +apply(auto intro: rrewrite_srewrite.intros) +done + +lemma continuous_rewrite: + assumes "r1 \* AZERO" + shows "ASEQ bs1 r1 r2 \* AZERO" +using assms bs1 star_seq by blast + +(* +lemma continuous_rewrite2: + assumes "r1 \* AONE bs" + shows "ASEQ bs1 r1 r2 \* (fuse (bs1 @ bs) r2)" + using assms by (meson bs3 rrewrites.simps star_seq) +*) + +lemma bsimp_aalts_simpcases: + shows "AONE bs \* bsimp (AONE bs)" + and "AZERO \* bsimp AZERO" + and "ACHAR bs c \* bsimp (ACHAR bs c)" + by (simp_all) + +lemma bsimp_AALTs_rewrites: + shows "AALTs bs1 rs \* bsimp_AALTs bs1 rs" + by (smt (verit) bs6 bs7 bsimp_AALTs.elims rrewrites.simps) + +lemma trivialbsimp_srewrites: + assumes "\x. x \ set rs \ x \* f x" + shows "rs s\* (map f rs)" +using assms + apply(induction rs) + apply(simp_all add: srewrites7) + done + +lemma fltsfrewrites: "rs s\* flts rs" + apply(induction rs rule: flts.induct) + apply(auto intro: rrewrite_srewrite.intros) + apply (meson srewrites.simps srewrites1 ss5) + using rs1 srewrites7 apply presburger + using srewrites7 apply force + apply (simp add: srewrites7) + by (simp add: srewrites7) + +lemma bnullable0: +shows "r1 \ r2 \ bnullable r1 = bnullable r2" + and "rs1 s\ rs2 \ bnullables rs1 = bnullables rs2" + apply(induct rule: rrewrite_srewrite.inducts) + apply(auto simp add: bnullable_fuse) + apply (meson UnCI bnullable_fuse imageI) + by (metis bnullable_correctness) + + +lemma rewrites_bnullable_eq: + assumes "r1 \* r2" + shows "bnullable r1 = bnullable r2" +using assms + apply(induction r1 r2 rule: rrewrites.induct) + apply simp + using bnullable0(1) by auto + +lemma rewrite_bmkeps_aux: + shows "r1 \ r2 \ bnullable r1 \ bmkeps r1 = bmkeps r2" + and "rs1 s\ rs2 \ bnullables rs1 \ bmkepss rs1 = bmkepss rs2" +proof (induct rule: rrewrite_srewrite.inducts) + case (bs3 bs1 bs2 r) + have IH2: "bnullable (ASEQ bs1 (AONE bs2) r)" by fact + then show "bmkeps (ASEQ bs1 (AONE bs2) r) = bmkeps (fuse (bs1 @ bs2) r)" + by (simp add: bmkeps_fuse) +next + case (bs7 bs r) + have IH2: "bnullable (AALTs bs [r])" by fact + then show "bmkeps (AALTs bs [r]) = bmkeps (fuse bs r)" + by (simp add: bmkeps_fuse) +next + case (ss3 r1 r2 rs) + have IH1: "bnullable r1 \ bmkeps r1 = bmkeps r2" by fact + have as: "r1 \ r2" by fact + from IH1 as show "bmkepss (r1 # rs) = bmkepss (r2 # rs)" + by (simp add: bnullable0) +next + case (ss5 bs1 rs1 rsb) + have "bnullables (AALTs bs1 rs1 # rsb)" by fact + then show "bmkepss (AALTs bs1 rs1 # rsb) = bmkepss (map (fuse bs1) rs1 @ rsb)" + by (simp add: bmkepss1 bmkepss2 bmkepss_fuse bnullable_fuse) +next + case (ss6 a1 a2 rsa rsb rsc) + have as1: "erase a1 = erase a2" by fact + have as3: "bnullables (rsa @ [a1] @ rsb @ [a2] @ rsc)" by fact + show "bmkepss (rsa @ [a1] @ rsb @ [a2] @ rsc) = bmkepss (rsa @ [a1] @ rsb @ rsc)" using as1 as3 + by (smt (verit, best) append_Cons bmkeps.simps(3) bmkepss.simps(2) bmkepss1 bmkepss2 bnullable_correctness) +qed (auto) + +lemma rewrites_bmkeps: + assumes "r1 \* r2" "bnullable r1" + shows "bmkeps r1 = bmkeps r2" + using assms +proof(induction r1 r2 rule: rrewrites.induct) + case (rs1 r) + then show "bmkeps r = bmkeps r" by simp +next + case (rs2 r1 r2 r3) + then have IH: "bmkeps r1 = bmkeps r2" by simp + have a1: "bnullable r1" by fact + have a2: "r1 \* r2" by fact + have a3: "r2 \ r3" by fact + have a4: "bnullable r2" using a1 a2 by (simp add: rewrites_bnullable_eq) + then have "bmkeps r2 = bmkeps r3" + using a3 bnullable0(1) rewrite_bmkeps_aux(1) by blast + then show "bmkeps r1 = bmkeps r3" using IH by simp +qed + + +lemma rewrites_to_bsimp: + shows "r \* bsimp r" +proof (induction r rule: bsimp.induct) + case (1 bs1 r1 r2) + have IH1: "r1 \* bsimp r1" by fact + have IH2: "r2 \* bsimp r2" by fact + { assume as: "bsimp r1 = AZERO \ bsimp r2 = AZERO" + with IH1 IH2 have "r1 \* AZERO \ r2 \* AZERO" by auto + then have "ASEQ bs1 r1 r2 \* AZERO" + by (metis bs2 continuous_rewrite rrewrites.simps star_seq2) + then have "ASEQ bs1 r1 r2 \* bsimp (ASEQ bs1 r1 r2)" using as by auto + } + moreover + { assume "\bs. bsimp r1 = AONE bs" + then obtain bs where as: "bsimp r1 = AONE bs" by blast + with IH1 have "r1 \* AONE bs" by simp + then have "ASEQ bs1 r1 r2 \* fuse (bs1 @ bs) r2" using bs3 star_seq by blast + with IH2 have "ASEQ bs1 r1 r2 \* fuse (bs1 @ bs) (bsimp r2)" + using rewrites_fuse by (meson rrewrites_trans) + then have "ASEQ bs1 r1 r2 \* bsimp (ASEQ bs1 (AONE bs) r2)" by simp + then have "ASEQ bs1 r1 r2 \* bsimp (ASEQ bs1 r1 r2)" by (simp add: as) + } + moreover + { assume as1: "bsimp r1 \ AZERO" "bsimp r2 \ AZERO" and as2: "(\bs. bsimp r1 = AONE bs)" + then have "bsimp_ASEQ bs1 (bsimp r1) (bsimp r2) = ASEQ bs1 (bsimp r1) (bsimp r2)" + by (simp add: bsimp_ASEQ1) + then have "ASEQ bs1 r1 r2 \* bsimp_ASEQ bs1 (bsimp r1) (bsimp r2)" using as1 as2 IH1 IH2 + by (metis rrewrites_trans star_seq star_seq2) + then have "ASEQ bs1 r1 r2 \* bsimp (ASEQ bs1 r1 r2)" by simp + } + ultimately show "ASEQ bs1 r1 r2 \* bsimp (ASEQ bs1 r1 r2)" by blast +next + case (2 bs1 rs) + have IH: "\x. x \ set rs \ x \* bsimp x" by fact + then have "rs s\* (map bsimp rs)" by (simp add: trivialbsimp_srewrites) + also have "... s\* flts (map bsimp rs)" by (simp add: fltsfrewrites) + also have "... s\* distinctBy (flts (map bsimp rs)) erase {}" by (simp add: ss6_stronger) + finally have "AALTs bs1 rs \* AALTs bs1 (distinctBy (flts (map bsimp rs)) erase {})" + using contextrewrites0 by blast + also have "... \* bsimp_AALTs bs1 (distinctBy (flts (map bsimp rs)) erase {})" + by (simp add: bsimp_AALTs_rewrites) + finally show "AALTs bs1 rs \* bsimp (AALTs bs1 rs)" by simp +qed (simp_all) + + +lemma to_zero_in_alt: + shows "AALT bs (ASEQ [] AZERO r) r2 \ AALT bs AZERO r2" + by (simp add: bs1 bs8 ss3) + + + +lemma bder_fuse_list: + shows "map (bder c \ fuse bs1) rs1 = map (fuse bs1 \ bder c) rs1" + apply(induction rs1) + apply(simp_all add: bder_fuse) + done + +lemma rewrite_preserves_bder: + shows "r1 \ r2 \ bder c r1 \* bder c r2" + and "rs1 s\ rs2 \ map (bder c) rs1 s\* map (bder c) rs2" +proof(induction rule: rrewrite_srewrite.inducts) + case (bs1 bs r2) + show "bder c (ASEQ bs AZERO r2) \* bder c AZERO" + by (simp add: continuous_rewrite) +next + case (bs2 bs r1) + show "bder c (ASEQ bs r1 AZERO) \* bder c AZERO" + apply(auto) + apply (meson bs6 contextrewrites0 rrewrite_srewrite.bs2 rs2 ss3 ss4 sss1 sss2) + by (simp add: r_in_rstar rrewrite_srewrite.bs2) +next + case (bs3 bs1 bs2 r) + show "bder c (ASEQ bs1 (AONE bs2) r) \* bder c (fuse (bs1 @ bs2) r)" + apply(simp) + by (metis (no_types, lifting) bder_fuse bs8 bs7 fuse_append rrewrites.simps ss4 to_zero_in_alt) +next + case (bs4 r1 r2 bs r3) + have as: "r1 \ r2" by fact + have IH: "bder c r1 \* bder c r2" by fact + from as IH show "bder c (ASEQ bs r1 r3) \* bder c (ASEQ bs r2 r3)" + by (metis bder.simps(5) bnullable0(1) contextrewrites1 rewrite_bmkeps_aux(1) star_seq) +next + case (bs5 r3 r4 bs r1) + have as: "r3 \ r4" by fact + have IH: "bder c r3 \* bder c r4" by fact + from as IH show "bder c (ASEQ bs r1 r3) \* bder c (ASEQ bs r1 r4)" + apply(simp) + apply(auto) + using contextrewrites0 r_in_rstar rewrites_fuse srewrites6 srewrites7 star_seq2 apply presburger + using star_seq2 by blast +next + case (bs6 bs) + show "bder c (AALTs bs []) \* bder c AZERO" + using rrewrite_srewrite.bs6 by force +next + case (bs7 bs r) + show "bder c (AALTs bs [r]) \* bder c (fuse bs r)" + by (simp add: bder_fuse r_in_rstar rrewrite_srewrite.bs7) +next + case (bs8 rs1 rs2 bs) + have IH1: "map (bder c) rs1 s\* map (bder c) rs2" by fact + then show "bder c (AALTs bs rs1) \* bder c (AALTs bs rs2)" + using contextrewrites0 by force +(*next + case ss1 + show "map (bder c) [] s\* map (bder c) []" by simp*) +next + case (ss2 rs1 rs2 r) + have IH1: "map (bder c) rs1 s\* map (bder c) rs2" by fact + then show "map (bder c) (r # rs1) s\* map (bder c) (r # rs2)" + by (simp add: srewrites7) +next + case (ss3 r1 r2 rs) + have IH: "bder c r1 \* bder c r2" by fact + then show "map (bder c) (r1 # rs) s\* map (bder c) (r2 # rs)" + by (simp add: srewrites7) +next + case (ss4 rs) + show "map (bder c) (AZERO # rs) s\* map (bder c) rs" + using rrewrite_srewrite.ss4 by fastforce +next + case (ss5 bs1 rs1 rsb) + show "map (bder c) (AALTs bs1 rs1 # rsb) s\* map (bder c) (map (fuse bs1) rs1 @ rsb)" + apply(simp) + using bder_fuse_list map_map rrewrite_srewrite.ss5 srewrites.simps by blast +next + case (ss6 a1 a2 bs rsa rsb) + have as: "erase a1 = erase a2" by fact + show "map (bder c) (bs @ [a1] @ rsa @ [a2] @ rsb) s\* map (bder c) (bs @ [a1] @ rsa @ rsb)" + apply(simp only: map_append) + by (smt (verit, best) erase_bder list.simps(8) list.simps(9) as rrewrite_srewrite.ss6 srewrites.simps) +qed + +lemma rewrites_preserves_bder: + assumes "r1 \* r2" + shows "bder c r1 \* bder c r2" +using assms +apply(induction r1 r2 rule: rrewrites.induct) +apply(simp_all add: rewrite_preserves_bder rrewrites_trans) +done + + +lemma central: + shows "bders r s \* bders_simp r s" +proof(induct s arbitrary: r rule: rev_induct) + case Nil + then show "bders r [] \* bders_simp r []" by simp +next + case (snoc x xs) + have IH: "\r. bders r xs \* bders_simp r xs" by fact + have "bders r (xs @ [x]) = bders (bders r xs) [x]" by (simp add: bders_append) + also have "... \* bders (bders_simp r xs) [x]" using IH + by (simp add: rewrites_preserves_bder) + also have "... \* bders_simp (bders_simp r xs) [x]" using IH + by (simp add: rewrites_to_bsimp) + finally show "bders r (xs @ [x]) \* bders_simp r (xs @ [x])" + by (simp add: bders_simp_append) +qed + +lemma main_aux: + assumes "bnullable (bders r s)" + shows "bmkeps (bders r s) = bmkeps (bders_simp r s)" +proof - + have "bders r s \* bders_simp r s" by (rule central) + then + show "bmkeps (bders r s) = bmkeps (bders_simp r s)" using assms + by (rule rewrites_bmkeps) +qed + + +theorem main_blexer_simp: + shows "blexer r s = blexer_simp r s" + unfolding blexer_def blexer_simp_def + by (metis central main_aux rewrites_bnullable_eq) + + +theorem blexersimp_correctness: + shows "lexer r s = blexer_simp r s" + using blexer_correctness main_blexer_simp by simp + + +(* some tests *) + +lemma asize_fuse: + shows "asize (fuse bs r) = asize r" + apply(induct r arbitrary: bs) + apply(auto) + done + +lemma asize_rewrite2: + shows "r1 \ r2 \ asize r1 \ asize r2" + and "rs1 s\ rs2 \ (sum_list (map asize rs1)) \ (sum_list (map asize rs2))" + apply(induct rule: rrewrite_srewrite.inducts) + apply(auto simp add: asize_fuse comp_def) + done + +lemma asize_rrewrites: + assumes "r1 \* r2" + shows "asize r1 \ asize r2" + using assms + apply(induct rule: rrewrites.induct) + apply(auto) + using asize_rewrite2(1) le_trans by blast + + + +fun asize2 :: "arexp \ nat" where + "asize2 AZERO = 1" +| "asize2 (AONE cs) = 1" +| "asize2 (ACHAR cs c) = 1" +| "asize2 (AALTs cs rs) = Suc (Suc (sum_list (map asize2 rs)))" +| "asize2 (ASEQ cs r1 r2) = Suc (asize2 r1 + asize2 r2)" +| "asize2 (ASTAR cs r) = Suc (asize2 r)" + + +lemma asize2_fuse: + shows "asize2 (fuse bs r) = asize2 r" + apply(induct r arbitrary: bs) + apply(auto) + done + +lemma asize2_not_zero: + shows "0 < asize2 r" + apply(induct r) + apply(auto) + done + +lemma asize_rewrite: + shows "r1 \ r2 \ asize2 r1 > asize2 r2" + and "rs1 s\ rs2 \ (sum_list (map asize2 rs1)) > (sum_list (map asize2 rs2))" + apply(induct rule: rrewrite_srewrite.inducts) + apply(auto simp add: asize2_fuse comp_def) + apply(simp add: asize2_not_zero) + done + +lemma asize2_bsimp_ASEQ: + shows "asize2 (bsimp_ASEQ bs r1 r2) \ Suc (asize2 r1 + asize2 r2)" + apply(induct bs r1 r2 rule: bsimp_ASEQ.induct) + apply(auto) + done + +lemma asize2_bsimp_AALTs: + shows "asize2 (bsimp_AALTs bs rs) \ Suc (Suc (sum_list (map asize2 rs)))" + apply(induct bs rs rule: bsimp_AALTs.induct) + apply(auto simp add: asize2_fuse) + done + +lemma distinctBy_asize2: + shows "sum_list (map asize2 (distinctBy rs f acc)) \ sum_list (map asize2 rs)" + apply(induct rs f acc rule: distinctBy.induct) + apply(auto) + done + +lemma flts_asize2: + shows "sum_list (map asize2 (flts rs)) \ sum_list (map asize2 rs)" + apply(induct rs rule: flts.induct) + apply(auto simp add: comp_def asize2_fuse) + done + +lemma sumlist_asize2: + assumes "\x. x \ set rs \ asize2 (f x) \ asize2 x" + shows "sum_list (map asize2 (map f rs)) \ sum_list (map asize2 rs)" + using assms + apply(induct rs) + apply(auto simp add: comp_def) + by (simp add: add_le_mono) + +lemma test0: + assumes "r1 \* r2" + shows "r1 = r2 \ (\r3. r1 \ r3 \ r3 \* r2)" + using assms + apply(induct r1 r2 rule: rrewrites.induct) + apply(auto) + done + +lemma test2: + assumes "r1 \* r2" + shows "asize2 r1 \ asize2 r2" +using assms + apply(induct r1 r2 rule: rrewrites.induct) + apply(auto) + using asize_rewrite(1) by fastforce + + +lemma test3: + shows "r = bsimp r \ (asize2 (bsimp r) < asize2 r)" +proof - + have "r \* bsimp r" + by (simp add: rewrites_to_bsimp) + then have "r = bsimp r \ (\r3. r \ r3 \ r3 \* bsimp r)" + using test0 by blast + then show ?thesis + by (meson asize_rewrite(1) dual_order.strict_trans2 test2) +qed + +lemma test3Q: + shows "r = bsimp r \ (asize (bsimp r) \ asize r)" +proof - + have "r \* bsimp r" + by (simp add: rewrites_to_bsimp) + then have "r = bsimp r \ (\r3. r \ r3 \ r3 \* bsimp r)" + using test0 by blast + then show ?thesis + using asize_rewrite2(1) asize_rrewrites le_trans by blast +qed + +lemma test4: + shows "asize2 (bsimp (bsimp r)) \ asize2 (bsimp r)" + apply(induct r rule: bsimp.induct) + apply(auto) + using rewrites_to_bsimp test2 apply fastforce + using rewrites_to_bsimp test2 by presburger + +lemma test4Q: + shows "asize (bsimp (bsimp r)) \ asize (bsimp r)" + apply(induct r rule: bsimp.induct) + apply(auto) + apply (metis order_refl test3Q) + by (metis le_refl test3Q) + + + +lemma testb0: + shows "fuse bs1 (bsimp_ASEQ bs r1 r2) = bsimp_ASEQ (bs1 @ bs) r1 r2" + apply(induct bs r1 r2 arbitrary: bs1 rule: bsimp_ASEQ.induct) + apply(auto) + done + +lemma testb1: + shows "fuse bs1 (bsimp_AALTs bs rs) = bsimp_AALTs (bs1 @ bs) rs" + apply(induct bs rs arbitrary: bs1 rule: bsimp_AALTs.induct) + apply(auto simp add: fuse_append) + done + +lemma testb2: + shows "bsimp (bsimp_ASEQ bs r1 r2) = bsimp_ASEQ bs (bsimp r1) (bsimp r2)" + apply(induct bs r1 r2 rule: bsimp_ASEQ.induct) + apply(auto simp add: testb0 testb1) + done + +lemma testb3: + shows "\r'. (bsimp r \ r') \ asize2 (bsimp r) > asize2 r'" +apply(induct r rule: bsimp.induct) + apply(auto) + defer + defer + using rrewrite.cases apply blast + using rrewrite.cases apply blast + using rrewrite.cases apply blast + using rrewrite.cases apply blast + oops + +lemma testb4: + assumes "sum_list (map asize rs1) \ sum_list (map asize rs2)" + shows "asize (bsimp_AALTs bs1 rs1) \ Suc (asize (bsimp_AALTs bs1 rs2))" + using assms +apply(induct bs1 rs2 arbitrary: rs1 rule: bsimp_AALTs.induct) + apply(auto) + apply(case_tac rs1) + apply(auto) + using asize2.elims apply auto[1] + apply (metis One_nat_def Zero_not_Suc asize.elims) + apply(case_tac rs1) + apply(auto) + apply(case_tac list) + apply(auto) + using asize_fuse apply force + apply (simp add: asize_fuse) + by (smt (verit, ccfv_threshold) One_nat_def add.right_neutral asize.simps(1) asize.simps(4) asize_fuse bsimp_AALTs.elims le_Suc_eq list.map(1) list.map(2) not_less_eq_eq sum_list_simps(1) sum_list_simps(2)) + +lemma flts_asize: + shows "sum_list (map asize (flts rs)) \ sum_list (map asize rs)" + apply(induct rs rule: flts.induct) + apply(auto simp add: comp_def asize_fuse) + done + + +lemma test5: + shows "asize2 r \ asize2 (bsimp r)" + apply(induct r rule: bsimp.induct) + apply(auto) + apply (meson Suc_le_mono add_le_mono asize2_bsimp_ASEQ order_trans) + apply(rule order_trans) + apply(rule asize2_bsimp_AALTs) + apply(simp) + apply(rule order_trans) + apply(rule distinctBy_asize2) + apply(rule order_trans) + apply(rule flts_asize2) + using sumlist_asize2 by force + + +fun awidth :: "arexp \ nat" where + "awidth AZERO = 1" +| "awidth (AONE cs) = 1" +| "awidth (ACHAR cs c) = 1" +| "awidth (AALTs cs rs) = (sum_list (map awidth rs))" +| "awidth (ASEQ cs r1 r2) = (awidth r1 + awidth r2)" +| "awidth (ASTAR cs r) = (awidth r)" + + + +lemma + shows "s \ L r \ blexer_simp r s = None" + by (simp add: blexersimp_correctness lexer_correct_None) + +lemma g1: + "bders_simp AZERO s = AZERO" + apply(induct s) + apply(simp) + apply(simp) + done + +lemma g2: + "s \ Nil \ bders_simp (AONE bs) s = AZERO" + apply(induct s) + apply(simp) + apply(simp) + apply(case_tac s) + apply(simp) + apply(simp) + done + +lemma finite_pder: + shows "finite (pder c r)" + apply(induct c r rule: pder.induct) + apply(auto) + done + + + +lemma awidth_fuse: + shows "awidth (fuse bs r) = awidth r" + apply(induct r arbitrary: bs) + apply(auto) + done + +lemma pders_SEQs: + assumes "finite A" + shows "card (SEQs A (STAR r)) \ card A" + using assms + by (simp add: SEQs_eq_image card_image_le) + +lemma binullable_intern: + shows "bnullable (intern r) = nullable r" + apply(induct r) + apply(auto simp add: bnullable_fuse) + done + +lemma + "card (pder c r) \ awidth (bder c (intern r))" + apply(induct c r rule: pder.induct) + apply(simp) + apply(simp) + apply(simp) + apply(simp) + apply(rule order_trans) + apply(rule card_Un_le) + apply (simp add: awidth_fuse bder_fuse) + defer + apply(simp) + apply(rule order_trans) + apply(rule pders_SEQs) + using finite_pder apply presburger + apply (simp add: awidth_fuse) + apply(auto) + apply(rule order_trans) + apply(rule card_Un_le) + apply(simp add: awidth_fuse) + defer + using binullable_intern apply blast + using binullable_intern apply blast + apply (smt (verit, best) SEQs_eq_image add.commute add_Suc_right card_image_le dual_order.trans finite_pder trans_le_add2) + apply(subgoal_tac "card (SEQs (pder c r1) r2) \ card (pder c r1)") + apply(linarith) + by (simp add: UNION_singleton_eq_range card_image_le finite_pder) + +lemma + "card (pder c r) \ asize (bder c (intern r))" + apply(induct c r rule: pder.induct) + apply(simp) + apply(simp) + apply(simp) + apply(simp) + apply (metis add_mono_thms_linordered_semiring(1) asize_fuse bder_fuse card_Un_le le_Suc_eq order_trans) + defer + apply(simp) + apply(rule order_trans) + apply(rule pders_SEQs) + using finite_pder apply presburger + apply (simp add: asize_fuse) + apply(simp) + apply(auto) + apply(rule order_trans) + apply(rule card_Un_le) + apply (smt (z3) SEQs_eq_image add.commute add_Suc_right add_mono_thms_linordered_semiring(1) asize_fuse card_image_le dual_order.trans finite_pder le_add1) + apply(rule order_trans) + apply(rule card_Un_le) + using binullable_intern apply blast + using binullable_intern apply blast + by (smt (verit, best) SEQs_eq_image add.commute add_Suc_right card_image_le dual_order.trans finite_pder trans_le_add2) + +lemma + "card (pder c r) \ asize (bsimp (bder c (intern r)))" + apply(induct c r rule: pder.induct) + apply(simp) + apply(simp) + apply(simp) + apply(simp) + apply(rule order_trans) + apply(rule card_Un_le) + prefer 3 + apply(simp) + apply(rule order_trans) + apply(rule pders_SEQs) + using finite_pder apply blast + oops + + +(* below is the idempotency of bsimp *) + +lemma bsimp_ASEQ_fuse: + shows "fuse bs1 (bsimp_ASEQ bs2 r1 r2) = bsimp_ASEQ (bs1 @ bs2) r1 r2" + apply(induct r1 r2 arbitrary: bs1 bs2 rule: bsimp_ASEQ.induct) + apply(auto) + done + +lemma bsimp_AALTs_fuse: + assumes "\r \ set rs. fuse bs1 (fuse bs2 r) = fuse (bs1 @ bs2) r" + shows "fuse bs1 (bsimp_AALTs bs2 rs) = bsimp_AALTs (bs1 @ bs2) rs" + using assms + apply(induct bs2 rs arbitrary: bs1 rule: bsimp_AALTs.induct) + apply(auto) + done + +lemma bsimp_fuse: + shows "fuse bs (bsimp r) = bsimp (fuse bs r)" + apply(induct r arbitrary: bs) + apply(simp_all add: bsimp_ASEQ_fuse bsimp_AALTs_fuse fuse_append) + done + +lemma bsimp_ASEQ_idem: + assumes "bsimp (bsimp r1) = bsimp r1" "bsimp (bsimp r2) = bsimp r2" + shows "bsimp (bsimp_ASEQ x1 (bsimp r1) (bsimp r2)) = bsimp_ASEQ x1 (bsimp r1) (bsimp r2)" + using assms + apply(case_tac "bsimp r1 = AZERO") + apply(simp) + apply(case_tac "bsimp r2 = AZERO") + apply(simp) + apply(case_tac "\bs. bsimp r1 = AONE bs") + apply(auto)[1] + apply (metis bsimp_fuse) + apply(simp add: bsimp_ASEQ1) + done + +lemma bsimp_AALTs_idem: + assumes "\r \ set rs. bsimp (bsimp r) = bsimp r" + shows "bsimp (bsimp_AALTs bs rs) = bsimp_AALTs bs (map bsimp rs)" + using assms + apply(induct bs rs rule: bsimp_AALTs.induct) + apply(simp) + apply(simp) + using bsimp_fuse apply presburger + oops + +lemma bsimp_idem_rev: + shows "\r2. bsimp r1 \ r2" + apply(induct r1 rule: bsimp.induct) + apply(auto) + defer + defer + using rrewrite.simps apply blast + using rrewrite.cases apply blast + using rrewrite.simps apply blast + using rrewrite.cases apply blast + apply(case_tac "bsimp r1 = AZERO") + apply(simp) + apply(case_tac "bsimp r2 = AZERO") + apply(simp) + apply(case_tac "\bs. bsimp r1 = AONE bs") + apply(auto)[1] + prefer 2 + apply (smt (verit, best) arexp.distinct(25) arexp.inject(3) bsimp_ASEQ1 rrewrite.simps) + defer + oops + +lemma bsimp_idem: + shows "bsimp (bsimp r) = bsimp r" + apply(induct r rule: bsimp.induct) + apply(auto) + using bsimp_ASEQ_idem apply presburger + oops + +lemma neg: + shows " \(\r2. r1 \ r2 \ (r2 \* bsimp r1) )" + apply(rule notI) + apply(erule exE) + apply(erule conjE) + oops + + + + +lemma reduction_always_in_bsimp: + shows " \ r1 \ r2 ; \(r2 \* bsimp r1)\ \ False" + apply(erule rrewrite.cases) + apply simp + apply auto + + oops + +(* +AALTs [] [AZERO, AALTs(bs1, [a, b]) ] +rewrite seq 1: \ AALTs [] [ AALTs(bs1, [a, b]) ] \ +fuse [] (AALTs bs1, [a, b]) +rewrite seq 2: \ AALTs [] [AZERO, (fuse bs1 a), (fuse bs1 b)]) ] + +*) + +lemma normal_bsimp: + shows "\r'. bsimp r \ r'" + oops + + (*r' size bsimp r > size r' + r' \* bsimp bsimp r +size bsimp r > size r' \ size bsimp bsimp r*) + +export_code blexer_simp blexer lexer bders bders_simp in Scala module_name VerifiedLexers + + +unused_thms + + +inductive aggressive:: "arexp \ arexp \ bool" ("_ \? _" [99, 99] 99) + where + "ASEQ bs (AALTs bs1 rs) r \? AALTs (bs@bs1) (map (\r'. ASEQ [] r' r) rs) " + + + +end diff -r 01d1285b08ed -r f71df68776bb thys2/blexer1.sc --- a/thys2/blexer1.sc Wed Feb 02 22:30:41 2022 +0000 +++ b/thys2/blexer1.sc Fri Feb 04 00:05:12 2022 +0000 @@ -361,7 +361,7 @@ case AALTS(bs1, rs) => { val rs_simp = rs.map(bsimp(_)) val flat_res = flats(rs_simp) - val dist_res = strongDistinctBy(flat_res)//distinctBy(flat_res, erase) + val dist_res = distinctBy(flat_res, erase)//strongDB(flat_res)//distinctBy(flat_res, erase) dist_res match { case Nil => AZERO case s :: Nil => fuse(bs1, s) @@ -372,6 +372,30 @@ case r => r } } + def strongBsimp(r: ARexp): ARexp = + { + r match { + case ASEQ(bs1, r1, r2) => (strongBsimp(r1), strongBsimp(r2)) match { + case (AZERO, _) => AZERO + case (_, AZERO) => AZERO + case (AONE(bs2), r2s) => fuse(bs1 ++ bs2, r2s) + case (r1s, r2s) => ASEQ(bs1, r1s, r2s) + } + case AALTS(bs1, rs) => { + val rs_simp = rs.map(strongBsimp(_)) + val flat_res = flats(rs_simp) + val dist_res = strongDB(flat_res)//distinctBy(flat_res, erase) + dist_res match { + case Nil => AZERO + case s :: Nil => fuse(bs1, s) + case rs => AALTS(bs1, rs) + } + + } + case r => r + } + } + def bders (s: List[Char], r: ARexp) : ARexp = s match { case Nil => r case c::s => bders(s, bder(c, r)) @@ -487,13 +511,13 @@ } - def strongDistinctBy(xs: List[ARexp], + def strongDB(xs: List[ARexp], acc1: List[Rexp] = Nil, acc2 : List[(List[Rexp], Rexp)] = Nil): List[ARexp] = xs match { case Nil => Nil case (x::xs) => if(acc1.contains(erase(x))) - strongDistinctBy(xs, acc1, acc2) + strongDB(xs, acc1, acc2) else{ x match { case ASTAR(bs0, r0) => @@ -502,7 +526,7 @@ r2stl => {val (r2s, tl) = r2stl; tl == erase(r0) } ) if(i == -1){ - x::strongDistinctBy( + x::strongDB( xs, erase(x)::acc1, (ONE::Nil, erase(r0))::acc2 ) } @@ -513,10 +537,10 @@ newHeads match{ case newHead::Nil => ASTAR(bs0, r0) :: - strongDistinctBy(xs, erase(x)::acc1, + strongDB(xs, erase(x)::acc1, acc2.updated(i, (oldHeadsUpdated, headListAlready._2)) )//TODO: acc2 already contains headListAlready case Nil => - strongDistinctBy(xs, erase(x)::acc1, + strongDB(xs, erase(x)::acc1, acc2) } } @@ -526,7 +550,7 @@ r2stl => {val (r2s, tl) = r2stl; tl == erase(r0) } ) if(i == -1){ - x::strongDistinctBy( + x::strongDB( xs, erase(x)::acc1, (headList.map(erase(_)), erase(r0))::acc2 ) } @@ -537,18 +561,18 @@ newHeads match{ case newHead::Nil => ASEQ(bs, newHead, ASTAR(bs0, r0)) :: - strongDistinctBy(xs, erase(x)::acc1, + strongDB(xs, erase(x)::acc1, acc2.updated(i, (oldHeadsUpdated, headListAlready._2)) )//TODO: acc2 already contains headListAlready case Nil => - strongDistinctBy(xs, erase(x)::acc1, + strongDB(xs, erase(x)::acc1, acc2) case hds => val AALTS(bsp, rsp) = r1 ASEQ(bs, AALTS(bsp, hds), ASTAR(bs0, r0)) :: - strongDistinctBy(xs, erase(x)::acc1, + strongDB(xs, erase(x)::acc1, acc2.updated(i, (oldHeadsUpdated, headListAlready._2))) } } - case rPrime => x::strongDistinctBy(xs, erase(x)::acc1, acc2) + case rPrime => x::strongDB(xs, erase(x)::acc1, acc2) } } @@ -586,12 +610,12 @@ // @arg(doc = "small tests") -val STARREG = ("a" | "aa").% +val STARREG = ((STAR("a") | STAR("aa") ).%).% @main def small() = { - val prog0 = """aaa""" + val prog0 = """aaaaaaaaa""" println(s"test: $prog0") // println(lexing_simp(NOTREG, prog0)) // val v = lex_simp(NOTREG, prog0.toList)