Esempio n. 1
0
  @Override
  public Result handle(DBObject docWrapper) {
    try {
      Set<String> organizationSet = new HashSet<String>();
      List<String> orgStrings = new ArrayList<String>(10);
      DBObject originalDoc = (DBObject) docWrapper.get("OriginalDoc");
      DBObject siDBO = (DBObject) docWrapper.get("SourceInfo");
      String srcId = siDBO.get("SourceID").toString();
      String sourceName = siDBO.get("Name").toString();
      String primaryKey = docWrapper.get("primaryKey").toString();
      JSONObject json = JSONUtils.toJSON((BasicDBObject) originalDoc, false);
      JSONPathProcessor jpp = new JSONPathProcessor();
      List<Object> objects =
          jpp.find(
              "$..'gmd:contact'.'gmd:CI_ResponsibleParty'.'gmd:organisationName'.'gco:CharacterString'.'_$'",
              json);
      if (objects != null) {
        for (Object o : objects) {
          orgStrings.add(o.toString());
        }
      }
      jpp = new JSONPathProcessor();
      objects =
          jpp.find(
              "$..'gmd:citation'.'gmd:CI_Citation'.'gmd:citedResponsibleParty'.'gmd:organisationName'.'gco:CharacterString'.'_$'",
              json);
      if (objects != null) {
        for (Object o : objects) {
          orgStrings.add(o.toString());
        }
      }
      // points of contacts
      jpp = new JSONPathProcessor();
      try {
        objects =
            jpp.find(
                "$..'gmd:identificationInfo'.'gmd:MD_DataIdentification'.'gmd:pointOfContact'[*].'gmd:CI_ResponsibleParty'.'gmd:organisationName'.'gco:CharacterString'.'_$'",
                json);
      } catch (JSONException je) {
        logger.info(je.getMessage());
        try {
          objects =
              jpp.find(
                  "$..'gmd:identificationInfo'.'gmd:MD_DataIdentification'.'gmd:pointOfContact'.'gmd:CI_ResponsibleParty'.'gmd:organisationName'.'gco:CharacterString'.'_$'",
                  json);
        } catch (JSONException jex) {
          logger.error(je);
        }
      }
      if (objects != null) {
        for (Object o : objects) {
          orgStrings.add(o.toString());
        }
      }
      List<Organization> organizations = new ArrayList<Organization>();
      for (String orgStr : orgStrings) {
        String[] toks = orgStr.split("[^a-zA-Z\\s\\d:.\\-]");
        for (String tok : toks) {
          tok = tok.trim();
          if (tok.startsWith(":")) {
            tok = tok.substring(1);
          }
          if (!organizationSet.contains(tok)) {
            organizationSet.add(tok);
            String orgName = Organization.normalizeOrgName(tok);
            logger.info("Organization: " + orgName);
            Organization organization = Organization.validateInViaf(orgName, lenient);
            if (organization != null) {
              logger.info(
                  "**** Validated Organization: "
                      + organization.getName()
                      + " ["
                      + organization.getUri()
                      + "]");
              organizations.add(organization);
            }
          }
        }
      }
      logger.info(
          "==================== end of organization enhancer for "
              + primaryKey
              + " ["
              + sourceName
              + "] =========================");
      //
      String category = "Organization > Institution";
      JSONArray jsArr = new JSONArray();
      List<KeywordInfo> kwiList = new ArrayList<KeywordInfo>(10);
      for (Organization org : organizations) {
        Keyword kw = new Keyword(org.getName());
        kw.addEntityInfo(new EntityInfo("", org.getUri(), -1, -1, category));
        jsArr.put(kw.toJSON());
        kwiList.add(
            new KeywordInfo(
                org.getUri(),
                org.getName(),
                category,
                null,
                CinergiXMLUtils.KeywordType.Organization));
      }
      if (jsArr.length() > 0) {
        DBObject data = (DBObject) docWrapper.get("Data");
        data.put("orgKeywords", JSONUtils.encode(jsArr));
        ProvenanceHelper.ProvData provData =
            new ProvenanceHelper.ProvData(primaryKey, ProvenanceHelper.ModificationType.Added);
        Map<String, List<KeywordInfo>> category2KWIListMap =
            new HashMap<String, List<KeywordInfo>>(7);
        category2KWIListMap.put(category, kwiList);

        provData.setSourceName(sourceName).setSrcId(srcId);
        EnhancerUtils.prepKeywordsProv(category2KWIListMap, provData);
        ProvenanceHelper.saveEnhancerProvenance("organizationEnhancer", provData, docWrapper);
        return new Result(docWrapper, Result.Status.OK_WITH_CHANGE);
      } else {
        return new Result(docWrapper, Result.Status.OK_WITHOUT_CHANGE);
      }
    } catch (Throwable t) {
      t.printStackTrace();
      Result r = new Result(docWrapper, Result.Status.ERROR);
      r.setErrMessage(t.getMessage());
      return r;
    }
  }
Esempio n. 2
0
 public void insertDoc(JSONObject doc) {
   DB db = mongoClient.getDB(this.dbName);
   DBCollection records = db.getCollection(this.collectionName);
   DBObject dbObject = JSONUtils.encode(doc);
   records.insert(dbObject);
 }