or1ko's diary

日々を書きます

22

言語処理100本ノック 2015

22. カテゴリ名の抽出
記事のカテゴリ名を(行単位ではなく名前で)抽出せよ.

22.hs

{-# LANGUAGE OverloadedStrings, DeriveGeneric #-}

import Data.Aeson as A
import Control.Applicative
import System.IO.UTF8 as I8
import System.Environment
import Data.Maybe
import Data.ByteString.Lazy.UTF8 as B
import GHC.Generics
import Text.Regex.Posix
import Prelude as P

data Article = Article {
  text :: String,
  title :: String
} deriving (Eq, Show, Generic)
instance FromJSON Article

main = do
  filename <- head <$> getArgs
  body <- fromString <$> I8.readFile filename
  I8.writeFile "22.txt" $ unlines $ map f2 $ filter f $ concatMap P.lines $ map (text . fromJust . A.decode) $ B.lines $ body 
    where
      f = ((=~ ("^[[][[]:?(カテゴリ|Category):" :: String)) :: String -> Bool)
      f2 = tail . ((=~ (":[^]|]*" :: String)) :: String -> String)