* - Tab (0x9)
     * - Line feed (0xA)
     * - Carriage return (0xD)
     * - Any character between 0x20 and 0xD7FF
     * - Any character between 0xE000 and 0xFFFD
     * - Any character between 0x10000 and 0x10FFFF
     *
     * @param in the input string to be processed
     * @return a new string with invalid XML characters removed, or the original
     *         string if it is empty

        <match value="0x0100" type="string" offset="26">
      	  <match value="0x0000" type="string" offset="28"/>
      	  <match value="0x0100" type="string" offset="28"/>
      	  <match value="0x0400" type="string" offset="28"/>
      	  <match value="0x0800" type="string" offset="28"/>
      	  <match value="0x1000" type="string" offset="28"/>
      	  <match value="0x1800" type="string" offset="28"/>

      <lastmod>2004-12-23</lastmod>
      <changefreq>weekly</changefreq>
   </url>
   <url>
      <loc>http://www.example.com/catalog?item=74&amp;desc=vacation_newfoundland</loc>
      <lastmod>2004-12-23T18:00:15+00:00</lastmod>
      <priority>0.3</priority>
   </url>
   <url>
      <loc>http://www.example.com/catalog?item=83&amp;desc=vacation_usa</loc>
      <lastmod>2004-11-23</lastmod>
   </url>

    @Test
    public void test_Read() throws Exception {
        String indexName = "test-index";
        Client client = runner.client();
        SuggestSettings settings = suggester.settings();
        int num = 10000;

        addDocument(indexName, client, num);

        ESSourceReader reader = new ESSourceReader(client, settings, indexName);
        reader.setScrollSize(1000);
        int count = 0;

        assertEquals("weekly", ((SitemapUrl) sitemaps[2]).getChangefreq());
        assertNull(((SitemapUrl) sitemaps[2]).getPriority());

        assertEquals("2004-12-23T18:00:15+00:00", sitemaps[3].getLastmod());
        assertEquals("http://www.example.com/catalog?item=74&desc=vacation_newfoundland", sitemaps[3].getLoc());
        assertNull(((SitemapUrl) sitemaps[3]).getChangefreq());

        assertNull(smbAuthenticationHolder.get("smb://hoge:1000/"));
        assertNull(smbAuthenticationHolder.get("smb://foo/"));
        assertNull(smbAuthenticationHolder.get("smb://foo:10000/"));
    }

        // Add URLs up to capacity (10000)
        for (int i = 0; i < 10000; i++) {
            urlSet.add("http://example" + i + ".com/robots.txt");
        }
        assertEquals(10000, urlSet.size());

        // Add one more should maintain size at 10000 (LRU eviction)
        urlSet.add("http://overflow.com/robots.txt");
        assertEquals(10000, urlSet.size());

 *   <li>secretKey - The secret key for authentication</li>
 *   <li>connectTimeout - Connection timeout in milliseconds (default: 10000)</li>
 *   <li>writeTimeout - Write timeout in milliseconds (default: 10000)</li>
 *   <li>readTimeout - Read timeout in milliseconds (default: 10000)</li>
 * </ul>
 *
 * <p>The client supports URLs in the format: {@code storage://bucket-name/object-path}
 *
 * <p>Features:
 * <ul>

        assertNull(smbAuthenticationHolder.get("smb1://hoge:1000/"));
        assertNull(smbAuthenticationHolder.get("smb1://foo/"));
        assertNull(smbAuthenticationHolder.get("smb1://foo:10000/"));
    }

    protected List<CrawlerClientFactory> clientFactoryList = new LinkedList<>();

    /**
     * The maximum size of the client factory list.
     */
    protected int maxClientFactorySize = 10000;

    /**
     * Registers a CrawlerClientFactory with this creator.
     * All existing client mappings will be loaded into the new factory.
     * @param crawlerClientFactory The CrawlerClientFactory to register.
     */