BookStatistics.java

package ethanjones.statistics;

import org.jsoup.Jsoup;
import org.jsoup.nodes.Document;
import org.jsoup.nodes.Element;
import org.jsoup.select.Elements;

import java.io.File;
import java.io.FileWriter;
import java.io.PrintWriter;
import java.io.Writer;
import java.net.URL;

public class BookStatistics {

  public static void main(String[] args) throws Exception {
    get_statistics("fiction", "http://www.whsmith.co.uk/chart/books/fiction-02x08974?results=200");
    get_statistics("nonfiction", "http://www.whsmith.co.uk/chart/books/non-fiction-02x09005?results=200");
    get_statistics("childrens", "http://www.whsmith.co.uk/chart/books/childrens-02x09004?results=200&filters=FILTER_format_fg%3aHardback%2cPaperback");
  }

  private static void get_statistics(String file, String page) throws Exception {
      File f = new File("output/" + file + ".csv");
      Writer w = new PrintWriter(new FileWriter(f));

      Document document = Jsoup.parse(new URL(page), 1000);
      Elements books = document.select("li.chart_product");
      for (Element book : books) {
        try {
          String num = book.select("span.chart_number").first().ownText();
          Element a = book.select("a").first();
          String title = a.attr("title");
          String productPage = a.attr("href");
          String author = book.select("span.product_second").first().ownText();
          String price = book.select("span.price").first().ownText().replace("£", "");

          Document product = Jsoup.parse(new URL(productPage), 1000);
          Element attributes = product.select("ul.product_attribute_list").first();
          String pages = attributes.select("li[itemprop*=Pages]").first().ownText().trim();
          String format = attributes.select("li[itemprop*=format]").first().ownText().trim().toLowerCase();

          w.write("\"" + num +  "\",\"" + title + "\",\"" + author  + "\",\"" + price  + "\",\"" + pages + "\",\"" + format + "\"\n");
        } catch (Exception ignored) {}
      }
    w.close();
  }
}