示例#1
0
  protected void toXml(Record record) throws SAXException {
    if (!MarcFactory.newInstance().validateRecord(record)) {
      throw new MarcException("Marc record didn't validate");
    }
    char temp[];
    AttributesImpl atts = new AttributesImpl();
    if (indent) handler.ignorableWhitespace("\n  ".toCharArray(), 0, 3);

    handler.startElement(Constants.MARCXML_NS_URI, RECORD, RECORD, atts);

    if (indent) handler.ignorableWhitespace("\n    ".toCharArray(), 0, 5);

    handler.startElement(Constants.MARCXML_NS_URI, LEADER, LEADER, atts);
    Leader leader = record.getLeader();
    temp = leader.toString().toCharArray();
    handler.characters(temp, 0, temp.length);
    handler.endElement(Constants.MARCXML_NS_URI, LEADER, LEADER);

    for (ControlField field : record.getControlFields()) {
      atts = new AttributesImpl();
      atts.addAttribute("", "tag", "tag", "CDATA", field.getTag());

      if (indent) handler.ignorableWhitespace("\n    ".toCharArray(), 0, 5);

      handler.startElement(Constants.MARCXML_NS_URI, CONTROL_FIELD, CONTROL_FIELD, atts);
      temp = getDataElement(field.getData());
      handler.characters(temp, 0, temp.length);
      handler.endElement(Constants.MARCXML_NS_URI, CONTROL_FIELD, CONTROL_FIELD);
    }

    for (DataField field : record.getDataFields()) {
      atts = new AttributesImpl();
      atts.addAttribute("", "tag", "tag", "CDATA", field.getTag());
      atts.addAttribute("", "ind1", "ind1", "CDATA", String.valueOf(field.getIndicator1()));
      atts.addAttribute("", "ind2", "ind2", "CDATA", String.valueOf(field.getIndicator2()));

      if (indent) handler.ignorableWhitespace("\n    ".toCharArray(), 0, 5);

      handler.startElement(Constants.MARCXML_NS_URI, DATA_FIELD, DATA_FIELD, atts);
      for (Subfield subfield : field.getSubfields()) {
        atts = new AttributesImpl();
        atts.addAttribute("", "code", "code", "CDATA", String.valueOf(subfield.getCode()));

        if (indent) handler.ignorableWhitespace("\n      ".toCharArray(), 0, 7);

        handler.startElement(Constants.MARCXML_NS_URI, SUBFIELD, SUBFIELD, atts);
        temp = getDataElement(subfield.getData());
        handler.characters(temp, 0, temp.length);
        handler.endElement(Constants.MARCXML_NS_URI, SUBFIELD, SUBFIELD);
      }

      if (indent) handler.ignorableWhitespace("\n    ".toCharArray(), 0, 5);

      handler.endElement(Constants.MARCXML_NS_URI, DATA_FIELD, DATA_FIELD);
    }

    if (indent) handler.ignorableWhitespace("\n  ".toCharArray(), 0, 3);

    handler.endElement(Constants.MARCXML_NS_URI, RECORD, RECORD);
  }
示例#2
0
  /**
   * Tests {@link Record#find(String)} and {@link Record#find(String, String)} and {@link
   * Record#find(String[], String)}.
   *
   * @throws Exception
   */
  @Test
  public void testFind() throws Exception {
    VariableField field = record.getVariableField("245");
    assertEquals(true, field.find("Summerland"));
    assertEquals(true, field.find("Sum*erland"));
    assertEquals(true, field.find("[Cc]habo[a-z]"));

    field = record.getVariableField("008");
    assertEquals(true, field.find("eng"));

    List<? extends VariableField> result = record.find("Summerland");
    assertEquals(1, result.size());
    field = (VariableField) result.get(0);
    assertEquals("245", field.getTag());

    result = record.find("Chabon");
    assertEquals(2, result.size());

    result = record.find("100", "Chabon");
    assertEquals(1, result.size());

    String[] tags = {"100", "260", "300"};
    result = record.find(tags, "Chabon");
    assertEquals(1, result.size());

    result = record.find("040", "DLC");
    assertTrue(result.size() > 0);

    DataField df = (DataField) result.get(0);
    String agency = df.getSubfield('a').getData();
    assertTrue(agency.matches("DLC"));
  }
示例#3
0
 public boolean execute() {
   List dataFields = record.getDataFieldList();
   if (dataFields == null) return false;
   Iterator dit = dataFields.iterator(); // to make sure remove() operation is supported
   boolean ret = false;
   while (dit.hasNext()) {
     DataField dataField = (DataField) dit.next();
     if (fieldMatches(dataField)) {
       List subfields = dataField.getSubfieldList();
       if (subfields == null) continue;
       Iterator sit = subfields.iterator();
       while (sit.hasNext()) {
         Subfield subfield = (Subfield) sit.next();
         if (subfieldMatches(subfield)) {
           sit.remove();
           ret = true;
         }
       }
       if (subfields.size() == 0) {
         dit.remove();
       }
     }
   }
   return ret;
 }
示例#4
0
  /**
   * Tests creating a new {@link Record}.
   *
   * @throws Exception
   */
  @Test
  public void testCreateRecord() throws Exception {
    MarcFactory factory = MarcFactory.newInstance();
    Record record = factory.newRecord("00000cam a2200000 a 4500");
    assertEquals("00000cam a2200000 a 4500", record.getLeader().marshal());

    record.addVariableField(factory.newControlField("001", "12883376"));

    DataField df = factory.newDataField("245", '1', '0');
    df.addSubfield(factory.newSubfield('a', "Summerland /"));
    df.addSubfield(factory.newSubfield('c', "Michael Chabon."));
    record.addVariableField(df);
  }
 public String getPublishDateDisplay(MarcFunctionContext ctx) {
   for (String tag : new String[] {"260", "264"}) {
     for (DataField df : ctx.record().getDataFields(tag)) {
       if (tag.equals("264") && df.getIndicator2() != '1') {
         continue;
       }
       if (df.getSubfield('c') != null) {
         String subC = df.getSubfield('c').getData();
         if (SINGLE_YEAR_PATTERN.matcher(subC).matches()) {
           return subC;
         }
       }
     }
   }
   return null;
 }
示例#6
0
 /**
  * Creates a new data field with the given tag and indicators and subfields and returns the
  * instance.
  *
  * @return DataField
  */
 @Override
 public DataField newDataField(
     final String tag, final char ind1, final char ind2, final String... subfieldCodesAndData) {
   final DataField df = new DataFieldImpl(tag, ind1, ind2);
   if (subfieldCodesAndData.length % 2 == 1) {
     throw new MarcException(
         "Error: must provide even number of parameters for subfields: code, data, code, data, ...");
   }
   for (int i = 0; i < subfieldCodesAndData.length; i += 2) {
     if (subfieldCodesAndData[i].length() != 1) {
       throw new MarcException("Error: subfieldCode must be a single character");
     }
     final Subfield sf =
         newSubfield(subfieldCodesAndData[i].charAt(0), subfieldCodesAndData[i + 1]);
     df.addSubfield(sf);
   }
   return (df);
 }
  public static void main(String args[]) throws Exception {

    InputStream input = ReadMarcExample.class.getResourceAsStream("resources/summerland.mrc");

    MarcReader reader = new MarcStreamReader(input);
    while (reader.hasNext()) {
      Record record = reader.next();

      // check if the cataloging agency is DLC
      List result = record.find("040", "DLC");
      if (result.size() > 0) System.out.println("Agency for this record is DLC");

      // there is no specific find for a specific subfield
      // so to check if it is the orignal cataloging agency
      DataField field = (DataField) result.get(0);
      String agency = field.getSubfield('a').getData();
      if (agency.matches("DLC")) System.out.println("DLC is the original agency");
    }
  }
  public Set<Integer> getPublishDate(MarcFunctionContext ctx) {
    MarcRecord record = ctx.record();
    Set<Integer> years = new TreeSet<Integer>();

    for (DataField datafield : record.getDataFields("264")) {
      if (datafield.getIndicator2() == '1') {
        if (datafield.getSubfield('c') != null) {
          years.addAll(getPublishDateFromItem(datafield.getSubfield('c').getData()));
        }
      }
    }
    years.addAll(getPublishDateFromItems(ctx, "260", 'c'));
    years.addAll(getPublishDateFromItems(ctx, "773", '9'));
    years.addAll(getPublishDateFromItems(ctx, "996", 'y'));

    String field008 = record.getControlField("008");
    years.addAll(parsePublishDateFrom008(field008));

    return years;
  }
  // file import rules
  public ActionResult importFile(File selectedFile) throws IOException {
    Timer timer = new Timer();
    TreeMap<String, Stats> types = new TreeMap<String, Stats>();

    InputStream in = new FileInputStream(selectedFile);
    MarcReader reader = new MarcPermissiveStreamReader(in, true, true);

    while (reader.hasNext()) {
      Record record = reader.next();

      String bib_id = "";
      DataField df907 = (DataField) record.getVariableField("907");

      if (df907 != null) {
        Subfield df907a = df907.getSubfield('a');
        if (df907a != null) {
          bib_id = df907a.getData();
        }
      }
      if (bib_id.startsWith(".b")) {
        bib_id = bib_id.substring(2, 9);
      }

      String key = bib_id;
      Stats stat = Generator.INSTANCE.create(key);
      types.put(stat.key, stat);

      String checksum = getChecksum(record.toString());
      stat.setVal(InventoryStatsItem.Hash_Code, checksum);
    } // end while loop

    return new ActionResult(
        selectedFile,
        selectedFile.getName(),
        this.toString(),
        details,
        types,
        true,
        timer.getDuration());
  }
示例#10
0
 @Override
 public String[] getValues(VariableField vf) {
   DataField df = (DataField) vf;
   String value = df.getSubfield('a').getData();
   String tag = df.getTag();
   if (tag.equals("100")) {
     Subfield csf = df.getSubfield('c');
     if (csf != null) {
       String data = csf.getData();
       int dlen = data.length();
       StringBuffer sb = new StringBuffer(value.length() + data.length() + 1);
       sb.append(value);
       sb.append(' ');
       char c = data.charAt(dlen - 1);
       if (c == ',' || c == '.') {
         sb.append(data, 0, dlen - 1);
       } else sb.append(data);
       value = sb.toString();
     }
   } else { // 110 & 710
     List<Subfield> sfs = df.getSubfields('b');
     if (sfs.size() > 0) {
       StringBuffer buf = new StringBuffer(value.length() + sfs.size() * 5);
       buf.append(value);
       Iterator<Subfield> sfi = sfs.iterator();
       while (sfi.hasNext()) {
         buf.append(' ');
         buf.append(sfi.next().getData());
       }
       int last = buf.length() - 1;
       char c = buf.charAt(last);
       if (c == ',' || c == '.') {
         buf.setLength(last);
       }
       value = buf.toString();
     }
   }
   return new String[] {trim(value).toString()};
 }
  @Override
  public Set<SkatKey> process(Long item) throws Exception {
    Set<SkatKey> result = new HashSet<>();
    HarvestedRecord hr = harvestedRecordDao.get(item);
    if (hr.getRawRecord() == null) {
      return result;
    }

    MarcRecord marc = null;
    InputStream is = new ByteArrayInputStream(hr.getRawRecord());
    try {
      marc = marcXmlParser.parseRecord(is);
    } catch (Exception e) {
      return result;
    }

    for (DataField df : marc.getDataFields("996")) {
      if (df.getSubfield('e') == null) {
        continue;
      }
      if (df.getSubfield('w') == null) {
        continue;
      }
      String sigla = df.getSubfield('e').getData();
      String recordId = df.getSubfield('w').getData();

      if (recordId.length() > 100 || sigla.length() > 20) {
        // ignore garbage
        continue;
      }

      SkatKey key = new SkatKey(new SkatKeyCompositeId(hr.getId(), sigla, recordId));
      result.add(key);
    }

    // ignore records having not enough information
    return result.size() < 2 ? Collections.emptySet() : result;
  }
  private void makeMaregdRecord(
      int databaseType, Record record, String controlNumber, CustomStorageRecord storageRecord) {
    // making latin record
    String entryTag = StorageUtils.findEntryTagForAuthority((AuthorityRecord) storageRecord);
    String entryRelatedTag = "7" + entryTag.substring(1);
    record = storageRecord.getRecord();
    Record engRecord = MarcUtils.parseContent(storageRecord.getContent());
    engRecord.getControlField(Tags.ID).setData("0");

    // exchanging fields
    DataField df200 = engRecord.getDataField(entryTag);
    DataField df700 = engRecord.getDataField(entryRelatedTag);
    if (df700 != null) {
      df700.setTag(entryTag);
      if (df200 != null) {
        df200.setTag(entryRelatedTag);
        df200.add(new Subfield('3', String.valueOf(storageRecord.getId())));
      }
      CustomStorageRecord engStorageRecord =
          StorageUtils.createNewStorageRecord(
              StorageUtils.getStorageClassByClassType(databaseType));
      engStorageRecord.setId(0);
      engStorageRecord.setContent(MarcUtils.marshalContent(engRecord));
      engStorageRecord.setLanguageOfCataloging(MarcConsts.LANG_ENGLISH);
      engStorageRecord.setCharacterSets("50");
      engStorageRecord.setSubfieldRangeData(Tags.GENERAL_PROCESSING_DATA, 'a', "ba", 21, 22);
      engStorageRecord.setSubfieldRangeData(Tags.GENERAL_PROCESSING_DATA, 'a', "0", 23, 23);
      engStorageRecord.setSubfieldData(Tags.LANGUAGE_OF_THE_ITEM, 'a', MarcConsts.LANG_PERSIAN);

      try {
        engStorageRecord = save(engStorageRecord);
        logger.info("      ^^ english record added.", null);
        df700 = record.getDataField(entryRelatedTag);
        if (df700 != null) {
          df700.add(new Subfield('3', String.valueOf(engStorageRecord.getId())));
          storageRecord.setContent(MarcUtils.marshalContent(record));
          storageRecord = save(storageRecord);
          logger.info("      ^^ farsi record changed.", null);
        }

      } catch (Exception ex) {
        logger.error("error: RecordNo = " + controlNumber + "  english making error!! **", ex);
      }
    }
  }
  /**
   * @param study - study that we are exporting
   * @param out
   * @throws IOException
   */
  @TransactionAttribute(TransactionAttributeType.NOT_SUPPORTED)
  public void exportStudy(Study study, OutputStream out) throws IOException {
    if (study.getReleasedVersion() == null) {
      throw new IllegalArgumentException(
          "Study does not have released version, study.id = " + study.getId());
    }
    Metadata metadata = study.getReleasedVersion().getMetadata();

    MarcFactory factory = MarcFactory.newInstance();
    Record record = factory.newRecord();

    DataField title = factory.newDataField("245", '0', ' ');
    title.addSubfield(factory.newSubfield('a', metadata.getTitle()));
    if (!StringUtil.isEmpty(metadata.getAuthorsStr())) {
      title.addSubfield(factory.newSubfield('c', metadata.getAuthorsStr()));
    }
    if (!StringUtil.isEmpty(metadata.getDistributionDate())) {
      title.addSubfield(factory.newSubfield('s', metadata.getDistributionDate()));
    }
    record.addVariableField(title);

    DataField globalId = factory.newDataField("440", ' ', ' ');
    globalId.addSubfield(factory.newSubfield('v', study.getGlobalId()));
    record.addVariableField(globalId);

    for (StudyOtherId studyOtherId : metadata.getStudyOtherIds()) {
      DataField otherId = factory.newDataField("440", ' ', ' ');
      otherId.addSubfield(factory.newSubfield('v', studyOtherId.getOtherId()));
      record.addVariableField(otherId);
    }
    for (StudyAbstract studyAbstract : metadata.getStudyAbstracts()) {
      DataField abstractField = factory.newDataField("520", ' ', ' ');
      abstractField.addSubfield(factory.newSubfield('a', studyAbstract.getText()));
      record.addVariableField(abstractField);
    }

    DataField handle = factory.newDataField("856", ' ', ' ');
    handle.addSubfield(factory.newSubfield('u', study.getHandleURL()));
    record.addVariableField(handle);

    DataField dataverseUrl = factory.newDataField("535", ' ', ' ');
    dataverseUrl.addSubfield(
        factory.newSubfield(
            'd',
            "http://" + PropertyUtil.getHostUrl() + "/dvn/study?globalId=" + study.getGlobalId()));
    record.addVariableField(dataverseUrl);

    for (StudyTopicClass studyTopicClass : metadata.getStudyTopicClasses()) {
      DataField topicClass = factory.newDataField("650", ' ', ' ');
      topicClass.addSubfield(factory.newSubfield('a', studyTopicClass.getValue()));
      record.addVariableField(topicClass);
    }
    MarcStreamWriter writer = new MarcStreamWriter(out);

    writer.write(record);

    out.flush();
  }