<!DOCTYPE html>
<html>
<!-- Default style for pretty-print of text files -->
<head><title>shonaiben_translator.rb</title>
<style type="text/css">
pre {margin-left: 6ex;}
pre span {counter-increment: linenum; text-align: right; position: relative;}
pre span:before {
position: absolute; right: 0px;
background: #ddd; text-align: right; padding: 0 0.2ex;
width: 5.5ex; margin-right: 0.5ex;
content: counter(linenum) ":";
}
</style>
</head>
<body>
<p>[[ <a href="?showattc+article_m+280+raw">RAW TEXT</a>(DOWNLOAD here) ]]</p>
<pre lang="en">
<span></span>#!/usr/koeki/bin/ruby
<span></span># -*- coding: utf-8 -*-
<span></span>
<span></span>def levenshtein(string1, string2, memo={}) # レーベンシュタイン距離を計算するメソッド
<span></span> return memo[[string1, string2]] if memo[[string1, string2]]
<span></span> return string2.size if string1.empty?
<span></span> return string1.size if string2.empty?
<span></span> return levenshtein(string1[1..], string2[1..], memo) if string1[0] == string2[0]
<span></span> min_dist = 1 + [levenshtein(string1[1..], string2, memo),
<span></span> levenshtein(string1, string2[1..], memo),
<span></span> levenshtein(string1[1..], string2[1..], memo)
<span></span> ].min
<span></span> memo[[string1, string2]] = min_dist
<span></span> min_dist
<span></span>end
<span></span>
<span></span>source = File.readlines("shounaiben.txt") # 庄内弁の文章をファイルから読み込み、配列に保存
<span></span>candidates = File.readlines("kyoutsuugo.txt") # 共通語の訳文をファイルから読み込み、配列に保存
<span></span>
<span></span>i = 0
<span></span>while i < source.length # 庄内弁の文章ごとに同じ処理を繰り返す
<span></span> dist2target = levenshtein(source[i].chomp, candidates[i].chomp) # 正解訳文への編集距離の計算
<span></span> min_dist = dist2target # 最短編集距離の初期値（=正解訳文への編集距離）
<span></span> count_same_or_lower = 0 # 編集距離が正解への距離以下である文章の総数の初期値
<span></span> best_cand = candidates[i].chomp # 最短編集距離の訳文の初期値（=正解訳文）
<span></span> for cand in candidates # それぞれの共通語の文章への編集距離を調べる
<span></span> dist = levenshtein(source[i].chomp, cand.chomp)
<span></span> if dist < min_dist
<span></span> min_dist = dist
<span></span> best_cand = cand
<span></span> end
<span></span> if dist <= dist2target
<span></span> count_same_or_lower += 1
<span></span> end
<span></span> end
<span></span> printf("庄内弁の文章: %s\n", source[i].chomp)
<span></span> printf("共通語の訳文（正解）: %s\n", candidates[i].chomp)
<span></span> printf("最短編集距離の訳文: %s\n", best_cand.chomp)
<span></span> printf("正解訳文への編集距離: %d\n", dist2target)
<span></span> printf("最短編集距離: %d\n", min_dist)
<span></span> printf("編集距離が正解訳文への距離以下である文章の総数: %d\n", count_same_or_lower)
<span></span> puts
<span></span> i += 1
<span></span>end
</pre>
</body>
</html>
KONNO Ayato