新漢字を旧漢字に変換するスクリプト

home 　ScriptMenu 　　　□	新旧漢字対照表　新→旧漢字　旧→新漢字　青空文庫の※　説明

新漢字→旧漢字（basic version）　パールスクリプト

shift-jis で表示できる漢字のみを変換します。

まず、パールをインストールして、パスを通しておいてください。
下のテキストボックスを全選択、コピーして、メモ帳などで、ファイル名を kyukanji_basic.pl などとして保存します。文字コードは utf-8 です。
変換元となるテキストファイルが utf-8 の場合は文字コード変数 $mojicode を書き換えてください。

plファイルを置いたフォルダの下にworksというフォルダを作って、そこに変換元となるファイルを置いてください。

［例］　C:\data\text\perl に kyukanji_basic.pl を置いて、
　　　　その下層のフォルダ　C:\data\text\perl\works に aozora.txt を置く。

エクスプローラから kyukanji_basic.pl を選んでエンターキーを押すと、worksフォルダに kanji_kyu_ba_aozora.txt という変換済みのファイルが生成されます。

kyukanji_basic.pl


# 新漢字を旧漢字に変換
# shift-jis でも表示できる漢字のみを変換
# 入力ファイルの文字コードと $mojicode の一致を確認すべし
# kanji から始まるファイルは対象から除外。対象ファイルの拡張子は txt または dat
# 出力ファイルの文字コードは $mojicode 、ファイル名は kanji_kyu_ba_$file
# 元ファイルが aozora.txt ならば kanji_kyu_ba_aozora.txt


	use utf8;
	use Encode qw/encode decode/;	

	$mojicode = "SHIFT-JIS";
#	$mojicode = "UTF-8";


# ディレクトリを開く

	$crdr = "./works/";
	opendir(DIR,"$crdr");
	chdir $crdr;

# ファイル名を配列に格納
	@files = readdir(DIR);


# テキストファイルを順次開く
	foreach $file (@files){

		if ($file =~ /^kanji/ ){ next;}
		if ($file =~ /txt$/i || $file =~ /dat$/i){
			open (IN,"$file") || &error ("file not found");
			open(OUT, "> tmp.txt");
			while(<IN>){
				$_ = decode($mojicode,$_);
				
				$_ =~ s/桜/櫻/g;
				$_ =~ s/岳/嶽/g;
				$_ =~ s/灯/燈/g;
				$_ =~ s/仏/佛/g;
				$_ =~ s/万/萬/g;
				$_ =~ s/竜/龍/g;
				$_ =~ s/缶/罐/g;
				$_ =~ s/亜/亞/g;
				$_ =~ s/悪/惡/g;
				$_ =~ s/圧/壓/g;
				$_ =~ s/囲/圍/g;
				$_ =~ s/医/醫/g;
				$_ =~ s/為/爲/g;
				$_ =~ s/壱/壹/g;
				$_ =~ s/飲/飮/g;
				$_ =~ s/隠/隱/g;
				$_ =~ s/栄/榮/g;
				$_ =~ s/営/營/g;
				$_ =~ s/衛/衞/g;
				$_ =~ s/駅/驛/g;
				$_ =~ s/円/圓/g;
				$_ =~ s/塩/鹽/g;
				$_ =~ s/艶/艷/g;
				$_ =~ s/応/應/g;
				$_ =~ s/欧/歐/g;
				$_ =~ s/殴/毆/g;
				$_ =~ s/奥/奧/g;
				$_ =~ s/穏/穩/g;
				$_ =~ s/仮/假/g;
				$_ =~ s/価/價/g;
				$_ =~ s/画/畫/g;
				$_ =~ s/会/會/g;
				$_ =~ s/絵/繪/g;
				$_ =~ s/壊/壞/g;
				$_ =~ s/懐/懷/g;
				$_ =~ s/拡/擴/g;
				$_ =~ s/殻/殼/g;
				$_ =~ s/覚/覺/g;
				$_ =~ s/学/學/g;
				$_ =~ s/楽/樂/g;
				$_ =~ s/巻/卷/g;
				$_ =~ s/陥/陷/g;
				$_ =~ s/勧/勸/g;
				$_ =~ s/関/關/g;
				$_ =~ s/歓/歡/g;
				$_ =~ s/観/觀/g;
				$_ =~ s/顔/顏/g;
				$_ =~ s/巌/巖/g;
				$_ =~ s/気/氣/g;
				$_ =~ s/帰/歸/g;
				$_ =~ s/亀/龜/g;
				$_ =~ s/偽/僞/g;
				$_ =~ s/戯/戲/g;
				$_ =~ s/犠/犧/g;
				$_ =~ s/却/卻/g;
				$_ =~ s/旧/舊/g;
				$_ =~ s/拠/據/g;
				$_ =~ s/挙/擧/g;
				$_ =~ s/峡/峽/g;
				$_ =~ s/挟/挾/g;
				$_ =~ s/狭/狹/g;
				$_ =~ s/尭/堯/g;
				$_ =~ s/暁/曉/g;
				$_ =~ s/区/區/g;
				$_ =~ s/駆/驅/g;
				$_ =~ s/勲/勳/g;
				$_ =~ s/径/徑/g;
				$_ =~ s/茎/莖/g;
				$_ =~ s/恵/惠/g;
				$_ =~ s/渓/溪/g;
				$_ =~ s/経/經/g;
				$_ =~ s/蛍/螢/g;
				$_ =~ s/軽/輕/g;
				$_ =~ s/継/繼/g;
				$_ =~ s/鶏/鷄/g;
				$_ =~ s/芸/藝/g;
				$_ =~ s/欠/缺/g;
				$_ =~ s/県/縣/g;
				$_ =~ s/倹/儉/g;
				$_ =~ s/剣/劍/g;
				$_ =~ s/険/險/g;
				$_ =~ s/圏/圈/g;
				$_ =~ s/検/檢/g;
				$_ =~ s/献/獻/g;
				$_ =~ s/権/權/g;
				$_ =~ s/顕/顯/g;
				$_ =~ s/験/驗/g;
				$_ =~ s/厳/嚴/g;
				$_ =~ s/広/廣/g;
				$_ =~ s/効/效/g;
				$_ =~ s/恒/恆/g;
				$_ =~ s/鉱/鑛/g;
				$_ =~ s/号/號/g;
				$_ =~ s/国/國/g;
				$_ =~ s/砕/碎/g;
				$_ =~ s/済/濟/g;
				$_ =~ s/斎/齋/g;
				$_ =~ s/剤/劑/g;
				$_ =~ s/冊/册/g;
				$_ =~ s/雑/雜/g;
				$_ =~ s/参/參/g;
				$_ =~ s/桟/棧/g;
				$_ =~ s/蚕/蠶/g;
				$_ =~ s/惨/慘/g;
				$_ =~ s/賛/贊/g;
				$_ =~ s/残/殘/g;
				$_ =~ s/糸/絲/g;
				$_ =~ s/歯/齒/g;
				$_ =~ s/児/兒/g;
				$_ =~ s/辞/辭/g;
				$_ =~ s/湿/濕/g;
				$_ =~ s/実/實/g;
				$_ =~ s/写/寫/g;
				$_ =~ s/舎/舍/g;
				$_ =~ s/釈/釋/g;
				$_ =~ s/寿/壽/g;
				$_ =~ s/収/收/g;
				$_ =~ s/従/從/g;
				$_ =~ s/渋/澁/g;
				$_ =~ s/獣/獸/g;
				$_ =~ s/縦/縱/g;
				$_ =~ s/粛/肅/g;
				$_ =~ s/処/處/g;
				$_ =~ s/叙/敍/g;
				$_ =~ s/将/將/g;
				$_ =~ s/称/稱/g;
				$_ =~ s/焼/燒/g;
				$_ =~ s/証/證/g;
				$_ =~ s/奨/奬/g;
				$_ =~ s/条/條/g;
				$_ =~ s/乗/乘/g;
				$_ =~ s/浄/淨/g;
				$_ =~ s/剰/剩/g;
				$_ =~ s/畳/疊/g;
				$_ =~ s/縄/繩/g;
				$_ =~ s/壌/壤/g;
				$_ =~ s/嬢/孃/g;
				$_ =~ s/穣/穰/g;
				$_ =~ s/譲/讓/g;
				$_ =~ s/醸/釀/g;
				$_ =~ s/触/觸/g;
				$_ =~ s/嘱/囑/g;
				$_ =~ s/晋/晉/g;
				$_ =~ s/真/眞/g;
				$_ =~ s/寝/寢/g;
				$_ =~ s/慎/愼/g;
				$_ =~ s/尽/盡/g;
				$_ =~ s/図/圖/g;
				$_ =~ s/粋/粹/g;
				$_ =~ s/酔/醉/g;
				$_ =~ s/穂/穗/g;
				$_ =~ s/随/隨/g;
				$_ =~ s/髄/髓/g;
				$_ =~ s/枢/樞/g;
				$_ =~ s/数/數/g;
				$_ =~ s/声/聲/g;
				$_ =~ s/斉/齊/g;
				$_ =~ s/静/靜/g;
				$_ =~ s/窃/竊/g;
				$_ =~ s/摂/攝/g;
				$_ =~ s/専/專/g;
				$_ =~ s/浅/淺/g;
				$_ =~ s/戦/戰/g;
				$_ =~ s/践/踐/g;
				$_ =~ s/銭/錢/g;
				$_ =~ s/潜/潛/g;
				$_ =~ s/繊/纖/g;
				$_ =~ s/禅/禪/g;
				$_ =~ s/双/雙/g;
				$_ =~ s/壮/壯/g;
				$_ =~ s/争/爭/g;
				$_ =~ s/荘/莊/g;
				$_ =~ s/捜/搜/g;
				$_ =~ s/挿/插/g;
				$_ =~ s/曽/曾/g;
				$_ =~ s/装/裝/g;
				$_ =~ s/総/總/g;
				$_ =~ s/聡/聰/g;
				$_ =~ s/騒/騷/g;
				$_ =~ s/蔵/藏/g;
				$_ =~ s/臓/臟/g;
				$_ =~ s/属/屬/g;
				$_ =~ s/続/續/g;
				$_ =~ s/堕/墮/g;
				$_ =~ s/対/對/g;
				$_ =~ s/体/體/g;
				$_ =~ s/帯/帶/g;
				$_ =~ s/滞/滯/g;
				$_ =~ s/台/臺/g;
				$_ =~ s/滝/瀧/g;
				$_ =~ s/択/擇/g;
				$_ =~ s/沢/澤/g;
				$_ =~ s/担/擔/g;
				$_ =~ s/単/單/g;
				$_ =~ s/胆/膽/g;
				$_ =~ s/団/團/g;
				$_ =~ s/断/斷/g;
				$_ =~ s/弾/彈/g;
				$_ =~ s/遅/遲/g;
				$_ =~ s/痴/癡/g;
				$_ =~ s/虫/蟲/g;
				$_ =~ s/昼/晝/g;
				$_ =~ s/鋳/鑄/g;
				$_ =~ s/庁/廳/g;
				$_ =~ s/聴/聽/g;
				$_ =~ s/勅/敕/g;
				$_ =~ s/鎮/鎭/g;
				$_ =~ s/逓/遞/g;
				$_ =~ s/鉄/鐵/g;
				$_ =~ s/点/點/g;
				$_ =~ s/転/轉/g;
				$_ =~ s/伝/傳/g;
				$_ =~ s/当/當/g;
				$_ =~ s/党/黨/g;
				$_ =~ s/盗/盜/g;
				$_ =~ s/稲/稻/g;
				$_ =~ s/闘/鬪/g;
				$_ =~ s/独/獨/g;
				$_ =~ s/読/讀/g;
				$_ =~ s/届/屆/g;
				$_ =~ s/弐/貳/g;
				$_ =~ s/悩/惱/g;
				$_ =~ s/脳/腦/g;
				$_ =~ s/覇/霸/g;
				$_ =~ s/拝/拜/g;
				$_ =~ s/廃/廢/g;
				$_ =~ s/売/賣/g;
				$_ =~ s/麦/麥/g;
				$_ =~ s/発/發/g;
				$_ =~ s/髪/髮/g;
				$_ =~ s/抜/拔/g;
				$_ =~ s/蛮/蠻/g;
				$_ =~ s/秘/祕/g;
				$_ =~ s/浜/濱/g;
				$_ =~ s/払/拂/g;
				$_ =~ s/並/竝/g;
				$_ =~ s/餅/餠/g;
				$_ =~ s/辺/邊/g;
				$_ =~ s/変/變/g;
				$_ =~ s/弁/辯/g;
				$_ =~ s/宝/寶/g;
				$_ =~ s/豊/豐/g;
				$_ =~ s/褒/襃/g;
				$_ =~ s/没/沒/g;
				$_ =~ s/翻/飜/g;
				$_ =~ s/槙/槇/g;
				$_ =~ s/満/滿/g;
				$_ =~ s/黙/默/g;
				$_ =~ s/弥/彌/g;
				$_ =~ s/訳/譯/g;
				$_ =~ s/薬/藥/g;
				$_ =~ s/与/與/g;
				$_ =~ s/予/豫/g;
				$_ =~ s/余/餘/g;
				$_ =~ s/誉/譽/g;
				$_ =~ s/揺/搖/g;
				$_ =~ s/様/樣/g;
				$_ =~ s/謡/謠/g;
				$_ =~ s/遥/遙/g;
				$_ =~ s/瑶/瑤/g;
				$_ =~ s/来/來/g;
				$_ =~ s/乱/亂/g;
				$_ =~ s/覧/覽/g;
				$_ =~ s/両/兩/g;
				$_ =~ s/猟/獵/g;
				$_ =~ s/隣/鄰/g;
				$_ =~ s/塁/壘/g;
				$_ =~ s/礼/禮/g;
				$_ =~ s/励/勵/g;
				$_ =~ s/霊/靈/g;
				$_ =~ s/隷/隸/g;
				$_ =~ s/齢/齡/g;
				$_ =~ s/恋/戀/g;
				$_ =~ s/炉/爐/g;
				$_ =~ s/労/勞/g;
				$_ =~ s/楼/樓/g;
				$_ =~ s/禄/祿/g;
				$_ =~ s/湾/灣/g;
				$_ =~ s/舗/舖/g;
				
#新旧関係ではないが追加
				$_ =~ s/付/附/g;
				
				$_ = encode($mojicode,$_);
				print OUT $_;
			}
			close(OUT);
			close(IN);
			rename("tmp.txt","kanji_kyu_ba_" . $file);
#			rename($file,$file . ".ori");
#			unlink($file);
#			rename("tmp.txt",$file);

		}
	}


sub error{
	($sec, $min,$hour, $mday, $mon, $year, $wday) = localtime;
	$mon++;
	$year += 1900;
		open(OUT, "> error.log");
		print OUT "$_[0]\n";
		print OUT "$year/$mon/$mday  $hour:$min \n";
		close (OUT);
	
	exit(0);
	
}