Exemple #1
0
  public static IRecord processOAIRecord(
      Node node, String baseWikiUri, String oaiUri, String oaiPrefix) throws Exception {
    if (isRecordDeleted(node)) {
      String datestamp = XPathUtil.evalToString(node, DBPediaXPathUtil.getDatestampExpr());
      String oaiId = XPathUtil.evalToString(node, DBPediaXPathUtil.getOAIIdentifierExpr());

      DeletionRecord result = new DeletionRecord(oaiId, datestamp);
      return result;
    } else return exportToRecord(node, baseWikiUri, oaiUri, oaiPrefix);
  }
Exemple #2
0
  public static Record exportToRecord(
      Node node, String baseWikiUri, String oaiUri, String oaiPrefix) throws Exception {
    String datestamp = XPathUtil.evalToString(node, DBPediaXPathUtil.getDatestampExpr());
    // System.out.println("DATESTAMP = " + datestamp);

    // System.out.println(XMLUtil.toString(node));

    String language = XPathUtil.evalToString(node, DBPediaXPathUtil.getLanguageExpr());
    String t = XPathUtil.evalToString(node, DBPediaXPathUtil.getTitleExpr());

    // http://en.wikipedia.org/wiki/Special:OAIRepository
    // MediawikiTitle title = MediawikiHelper.parseTitle(domainUri
    //		+ "/wiki/Special:OAIRepository", t);
    MediawikiTitle title = MediawikiHelper.parseTitle(oaiUri, t);

    String oaiId = oaiPrefix + XPathUtil.evalToString(node, DBPediaXPathUtil.getPageIdExpr());
    // String wikipediaUri = domainUri + "/wiki/" + URLEncoder.encode(title.getFullTitle(),
    // "UTF-8");
    String wikipediaUri = baseWikiUri + URLEncoder.encode(title.getFullTitle(), "UTF-8");
    String revision = XPathUtil.evalToString(node, DBPediaXPathUtil.getRevisionExpr());
    String username = XPathUtil.evalToString(node, DBPediaXPathUtil.getContributorNameExpr());
    String ip = XPathUtil.evalToString(node, DBPediaXPathUtil.getContributorIpExpr());
    String userId = XPathUtil.evalToString(node, DBPediaXPathUtil.getContributorIdExpr());

    String text = XPathUtil.evalToString(node, DBPediaXPathUtil.getTextExpr());

    RecordMetadata metadata =
        new RecordMetadata(
            language, title, oaiId, IRI.create(wikipediaUri), revision, username, ip, userId);

    RecordContent content = new RecordContent(text, revision, XMLUtil.toString(node));

    return new Record(metadata, content);
  }
Exemple #3
0
  private static boolean isRecordDeleted(Node node) throws XPathExpressionException {
    String value = XPathUtil.evalToString(node, DBPediaXPathUtil.getIsRecordDeletedExpr());

    return "deleted".equalsIgnoreCase(value);
  }