util/org/cacert/gigi/util/HighFinancialValueFetcher.java

   1 package org.cacert.gigi.util;
   2
   3 import java.io.BufferedReader;
   4 import java.io.File;
   5 import java.io.IOException;
   6 import java.io.InputStreamReader;
   7 import java.io.PrintWriter;
   8 import java.net.URL;
   9 import java.util.zip.ZipEntry;
  10 import java.util.zip.ZipInputStream;
  11
  12 public class HighFinancialValueFetcher {
  13
  14     public static void main(String[] args) throws IOException {
  15         int max = 1000;
  16         if (args.length > 1) {
  17             max = Integer.parseInt(args[1]);
  18         }
  19         try (PrintWriter fos = new PrintWriter(new File(args[0]), "UTF-8"); ZipInputStream zis = new ZipInputStream(new URL("https://s3.amazonaws.com/alexa-static/top-1m.csv.zip").openStream())) {
  20             ZipEntry ze;
  21             outer:
  22             while ((ze = zis.getNextEntry()) != null) {
  23                 System.out.println(ze.getName());
  24                 BufferedReader br = new BufferedReader(new InputStreamReader(zis, "UTF-8"));
  25                 String line;
  26                 while ((line = br.readLine()) != null) {
  27                     String[] parts = line.split(",");
  28                     int i = Integer.parseInt(parts[0]);
  29                     if (i > max) {
  30                         zis.close();
  31                         break outer;
  32                     }
  33                     fos.println(parts[1]);
  34                     System.out.println(line);
  35                 }
  36             }
  37         }
  38     }
  39
  40 }