Method readContents

src/test/org/apache/nutch/util/CrawlTestUtil.java:100–129

Read entries from a data store @return list of matching URLWebPage objects @throws Exception

(
      DataStore<String, WebPage> store, Mark requiredMark, String... fields)

   * @throws Exception
   */
  public static ArrayList<URLWebPage> readContents(
      DataStore<String, WebPage> store, Mark requiredMark, String... fields)
          throws Exception {
    ArrayList<URLWebPage> l = new ArrayList<URLWebPage>();

    Query<String, WebPage> query = store.newQuery();
    if (fields != null) {
      query.setFields(fields);
    }

    Result<String, WebPage> results = store.execute(query);
    while (results.next()) {
      try {
        WebPage page = results.get();
        String url = results.getKey();

        if (page == null)
          continue;

        if (requiredMark != null && requiredMark.checkMark(page) == null)
          continue;

        l.add(new URLWebPage(TableUtil.unreverseUrl(url), WebPage.newBuilder(
            page).build()));
      } catch (Exception e) {
        e.printStackTrace();
      }
    }
    return l;
  }

  /**
   * Creates a new JettyServer with one static root context