final SitemapSet sitemapSet = new SitemapSet();
        sitemapSet.setType(SitemapSet.URLSET);

        try {
            final BufferedReader br = new BufferedReader(new InputStreamReader(in, Constants.UTF_8));
            String line;
            while ((line = br.readLine()) != null) {
                final String url = line.trim();

        final StringBuilder sb = new StringBuilder();
        try (BufferedReader br = new BufferedReader(
                new InputStreamReader(this.getClass().getClassLoader().getResourceAsStream("suggest_indices/suggest_settings.json")));) {

            String line;
            while ((line = br.readLine()) != null) {
                sb.append(line);
            }
        }

        final StringBuilder sb = new StringBuilder();
        try (BufferedReader br = new BufferedReader(new InputStreamReader(
                this.getClass().getClassLoader().getResourceAsStream("suggest_indices/suggest_settings_array.json")))) {

            String line;
            while ((line = br.readLine()) != null) {
                sb.append(line);
            }
        }

<!-- tweet text --> 猫を愛する漫画家10人の展覧会「ねこまんてん」<br />「猫の額」（東京・高円寺）にて7/31～8/12開催予定。<br />みんにゃ来てね！<br /><a href="https://t.co/redirect?url=http%3A%2F%2Ft.co%2Ft0dkECemdh%3Ft%3D1%26cn%3DZmxleGlibGVfcmVjc18y%26sig%3Dfbf95178e460b4251cbcef23539c05b1ee5033a3%26al%3D1%26iid%3D7c2e8db7f2e0464eb1fb51f268eeba54%26autoactions%3D14382...

        assertTrue(content.contains("テスト"));
    }

    public void test_getTika_xml_broken() throws UnsupportedEncodingException {
        final InputStream in = new ByteArrayInputStream("<?xml encoding=\"UTF-8\"/><hoge>テスト<br></hoge>".getBytes(Constants.UTF_8));
        final ExtractData extractData = tikaExtractor.getText(in, null);
        final String content = extractData.getContent();
        CloseableUtil.closeQuietly(in);

0),c===!1&&(c=$),this.each(function(){n.event.remove(this,a,c,b)})},trigger:function(a,b){return this.each(function(){n.event.trigger(a,b,this)})},triggerHandler:function(a,b){var c=this[0];return c?n.event.trigger(a,b,c,!0):void 0}});var ab=/<(?!area|br|col|embed|hr|img|input|link|meta|param)(([\w:]+)[^>]*)\/>/gi,bb=/<([\w:]+)/,cb=/<|&#?\w+;/,db=/<(?:script|style|link)/i,eb=/checked\s*(?:[^=]|=\s*.checked.)/i,fb=/^$|\/(?:java|ecma)script/i,gb=/^true\/(.*)/,hb=/^\s*<!(?:\[CDATA\[|--)|(?:\]\]|--)>\s...

        <match value="\x07" offset="4" type="string"/>
      </match>
    </magic>
  </mime-type>

  <mime-type type="application/x-brotli">
    <glob pattern="*.br" />
    <glob pattern="*.brotli" />
  </mime-type>

  <mime-type type="application/x-bzip">
    <magic priority="40">
      <match value="BZ0" type="string" offset="0"/>
    </magic>