英会話学習用のテキストを自作したくてQuoraのクスレイピングを書いた。なぜScalaなのかは不明…… #CodePiece

CodePiece.scala

import java.io.{File, PrintWriter}

import org.jsoup.Jsoup
import scalaj.http.BaseHttp

object Main extends App {
  def dumpToFile(text: String) = {
    val writer = new PrintWriter(new File("dump.md"))
    writer.write(text)
  }

  val url = "https://www.quora.com/How-is-your-life-after-moving-to-Thailand-as-a-foreigner"
  val body = LoadFromQuora(url).asString.body

  val html = Jsoup.parse(body)
  val title = html.getElementsByTag("h1").first().text()
  println(title)
  println("===============")
  println(url)

  val answers = html.getElementsByClass("pagedlist_item")
  answers.forEach(answer => {
    val person = answer.getElementsByClass("feed_item_answer_user").first()
    if (person != null) {
      println("## " + person.text())
    }

    answer.getElementsByClass("ui_qtext_expanded").forEach(element => {
      element.getElementsByClass("ui_qtext_para").forEach(para => {
        println(para.text())
      })
    })
  })
}

object LoadFromQuora extends BaseHttp(
  userAgent = "curl/7.54.0",
)