use underscore und non-empty names for tags, indicators, subfield IDs

src/main/java/org/xbib/marc/MarcField.java

@@ -43,6 +43,11 @@ public class MarcField implements Comparable<MarcField> {
 
     private static final String BLANK_STRING = " ";
 
+    private static final char UNDERSCORE = '_';
+
+    private static final String UNDERSCORE_STRING = "_";
+    private static final String EMPTY_TAG = "___";
+
     private static final Set<Character> ASCII_GRAPHICS = new HashSet<>(Arrays.asList(
             '\u0020', '\u0021', '\u0022', '\u0023', '\u0024', '\u0025', '\u0026', '\'',
             '\u0028', '\u0029', '\u002A', '\u002B', '\u002C', '\u002D', '\u002E', '\u002F',
@@ -207,8 +212,10 @@ public class MarcField implements Comparable<MarcField> {
 
     public boolean isTagValid() {
         if (tag == null) {
+            // we allow no tag
             return true;
         }
+        // only tags of length 3 are supposed to be valid
         return tag.length() == 3
                 && ((tag.charAt(0) >= '0' && tag.charAt(0) <= '9')
                 || (tag.charAt(0) >= 'A' && tag.charAt(0) <= 'Z'))
@@ -220,18 +227,20 @@ public class MarcField implements Comparable<MarcField> {
 
     public boolean isIndicatorValid() {
         if (isControl()) {
+            // ignore indicator check for control fields
             return true;
         }
         if (indicator == null) {
+            // we allow no indicator
             return true;
         }
         boolean b = indicator.length() <= 9;
         for (int i = 0; i < indicator.length(); i++) {
-            b = indicator.charAt(i) == ' '
+            b = indicator.charAt(i) == UNDERSCORE
                     || (indicator.charAt(i) >= '0' && indicator.charAt(i) <= '9')
                     || (indicator.charAt(i) >= 'a' && indicator.charAt(i) <= 'z')
                     || (indicator.charAt(i) >= 'A' && indicator.charAt(i) <= 'Z')
-                    || indicator.charAt(i) == '@'; // for our PICA hack
+                    || indicator.charAt(i) == '@'; // must be valid, for PICA dialect
             if (!b) {
                 break;
             }
@@ -241,9 +250,11 @@ public class MarcField implements Comparable<MarcField> {
 
     public boolean isSubfieldValid() {
         if (isControl()) {
+            // for control fields, there are no subfields, disable check
             return true;
         }
         if (subfieldIds == null) {
+            // we allow no subfield ids
             return true;
         }
         boolean b = true;
@@ -256,6 +267,10 @@ public class MarcField implements Comparable<MarcField> {
         return b;
     }
 
+    /**
+     * Checks if the field has a valid tag (if present), a valid indicator (if present), and a valid subfield (if present).
+     * @return true if valid
+     */
     public boolean isValid() {
         return isTagValid() && isIndicatorValid() && isSubfieldValid();
     }
@@ -381,7 +396,19 @@ public class MarcField implements Comparable<MarcField> {
          * @return this builder
          */
         public Builder tag(String tag) {
-            this.tag = tag;
+            if (tag != null) {
+                // We have inconsistent use of tag symbols as placeholders for a "blank space"
+                // and we need to fix it here for consistency.
+                this.tag = tag
+                        .replace('-', UNDERSCORE)
+                        .replace('#', UNDERSCORE)
+                        .replace('.', UNDERSCORE)
+                        .replace(' ', UNDERSCORE);
+            }
+            // do not allow null or empty tags
+            if (this.tag == null || this.tag.isEmpty()) {
+                this.tag = EMPTY_TAG;
+            }
             return this;
         }
 
@@ -400,8 +427,21 @@ public class MarcField implements Comparable<MarcField> {
          */
         public Builder indicator(String indicator) {
             if (indicator != null) {
-                // check if indicators are visible replacements like "-" or "#" . Replace with blank.
+                // We have inconsistent use of indicator symbols as placeholders for a "blank space"
-                this.indicator = indicator.replace('-', ' ').replace('#', ' ');
+                // and we need to fix it here for consistency.
+                // Check if indicators are artifacts like "-" or "#" or '.' or ' '.
+                // Replace with underscore. This preserves better usage and visibility.
+                // Especially the dot. The dot will break Elasticsearch indexing with an exception.
+                // If you do not like underscores, replace them by a space character in your application.
+                this.indicator = indicator
+                        .replace('-', UNDERSCORE)
+                        .replace('#', UNDERSCORE)
+                        .replace('.', UNDERSCORE)
+                        .replace(' ', UNDERSCORE);
+            }
+            // we do not allow an empty indicator. Elasticsearch field names require a length > 0.
+            if (this.indicator != null && this.indicator.isEmpty()) {
+                this.indicator = UNDERSCORE_STRING;
             }
             return this;
         }
@@ -612,7 +652,7 @@ public class MarcField implements Comparable<MarcField> {
          */
         public boolean isControl() {
             if (isControl == null) {
-                this.isControl = tag != null && tag.charAt(0) == '0' && tag.charAt(1) == '0';
+                this.isControl = tag != null && tag.length() >= 2 && tag.charAt(0) == '0' && tag.charAt(1) == '0';
             }
             return isControl;
         }
@@ -638,18 +678,13 @@ public class MarcField implements Comparable<MarcField> {
          * @return the built MARC field.
          */
         public MarcField build() {
-            if (isControl == null) {
-                this.isControl = tag != null && tag.charAt(0) == '0' && tag.charAt(1) == '0';
-            }
             return new MarcField(tag, indicator, position, length,
-                    value, subfields, subfieldIds.toString(), isControl);
+                    value, subfields, subfieldIds.toString(), isControl());
         }
 
         @Override
         public String toString() {
-            return "tag=" + tag + ",indicator=" + indicator +
+            return "tag=" + tag + ",indicator=" + indicator + ",value=" + value + ",subfields=" + subfields;
-                    ",value=" + value + ",subfields=" +
-                    subfields;
         }
     }
 
@@ -658,12 +693,24 @@ public class MarcField implements Comparable<MarcField> {
      */
     public static class Subfield {
 
-        private final String id;
+        private String id;
 
         private final String value;
 
         private Subfield(String id, String value) {
-            this.id = id;
+            if (id != null) {
+                // We have inconsistent use of subfield id symbols as placeholders for a "blank space"
+                // and we need to fix it here for consistency.
+                this.id = id
+                        .replace('-', UNDERSCORE)
+                        .replace('#', UNDERSCORE)
+                        .replace('.', UNDERSCORE)
+                        .replace(' ', UNDERSCORE);
+            }
+            // we do not allow an empty subfield id. Elasticsearch field names require a length > 0.
+            if (this.id != null && this.id.isEmpty()) {
+                this.id = UNDERSCORE_STRING;
+            }
             this.value = value;
         }
 
@@ -689,10 +736,9 @@ public class MarcField implements Comparable<MarcField> {
         }
     }
 
+    @SuppressWarnings("serial")
     private static class SubfieldIds extends LinkedList<String> {
 
-        private static final long serialVersionUID = 7016733919690084153L;
-
         /**
          * Insertion sort. This is considered faster than sorting afterwards,
          * especially for short lists (we have << 10 subfields at average in a field).

src/main/java/org/xbib/marc/MarcFieldDirectory.java

@@ -24,10 +24,9 @@ import java.util.TreeMap;
 /**
  *
  */
+@SuppressWarnings("serial")
 public class MarcFieldDirectory extends TreeMap<Integer, MarcField.Builder> {
 
-    private static final long serialVersionUID = 4339262603982720001L;
-
     public MarcFieldDirectory(RecordLabel label, String encodedDirectory) throws IOException {
         super();
         if (label == null) {

src/main/java/org/xbib/marc/MarcRecord.java

@@ -33,12 +33,11 @@ import java.util.stream.Collectors;
 /**
  * A MARC record. This is an extended MARC record augmented with MarcXchange information.
  */
+@SuppressWarnings("serial")
 public class MarcRecord extends LinkedHashMap<String, Object> {
 
     private static final MarcRecord EMPTY = Marc.builder().buildRecord();
 
-    private static final long serialVersionUID = 5305809148724342653L;
-
     private final String format;
 
     private final String type;
@@ -184,7 +183,6 @@ public class MarcRecord extends LinkedHashMap<String, Object> {
             }
             String indicator = marcField.getIndicator();
             if (indicator != null && !indicator.isEmpty()) {
-                indicator = indicator.replace(' ', '_');
                 Map<String, Object> indicators = new LinkedHashMap<>();
                 repeatMap.put(Integer.toString(repeat), indicators);
                 if (!indicators.containsKey(indicator)) {
@@ -197,7 +195,6 @@ public class MarcRecord extends LinkedHashMap<String, Object> {
                 } else {
                     for (MarcField.Subfield subfield : marcField.getSubfields()) {
                         String code = subfield.getId();
-                        code = code.replace(' ', '_');
                         Object subfieldValue = subfields.get(code);
                         if (subfieldValue instanceof List) {
                             List<String> list = (List<String>) subfieldValue;

src/main/java/org/xbib/marc/json/JsonException.java

@@ -16,10 +16,9 @@
  */
 package org.xbib.marc.json;
 
+@SuppressWarnings("serial")
 public class JsonException extends RuntimeException {
 
-    private static final long serialVersionUID = -3386151672072419281L;
-
     JsonException(Throwable throwable) {
         super(throwable);
     }

src/main/java/org/xbib/marc/transformer/field/MarcFieldTransformer.java

@@ -32,10 +32,9 @@ import java.util.regex.Pattern;
 /**
  *
  */
+@SuppressWarnings("serial")
 public class MarcFieldTransformer extends LinkedHashMap<String, MarcField> {
 
-    private static final long serialVersionUID = -3250616818472683245L;
-
     private static final Logger logger = Logger.getLogger(MarcFieldTransformer.class.getName());
 
     // the repeat counter pattern, simple integer
@@ -47,7 +46,7 @@ public class MarcFieldTransformer extends LinkedHashMap<String, MarcField> {
     private transient MarcField lastReceived;
     private transient MarcField lastBuilt;
     private int repeatCounter;
-    private Operator operator;
+    private final Operator operator;
 
     private MarcFieldTransformer(Map<String, MarcField> map, boolean ignoreIndicator, boolean ignoreSubfieldIds,
                                  Operator operator) {

src/main/java/org/xbib/marc/xml/MarcContentHandler.java

@@ -268,7 +268,7 @@ public class MarcContentHandler
 
     @Override
     public void startDocument() throws SAXException {
-        stack.clear();
+        stack = new LinkedList<>();
     }
 
     @Override

src/main/java/org/xbib/marc/xml/MarcXchangeWriter.java

@@ -514,11 +514,6 @@ public class MarcXchangeWriter extends MarcContentHandler implements Flushable,
 
     private void setupEventConsumer(Writer writer, boolean indent) throws IOException {
         XMLOutputFactory outputFactory = XMLOutputFactory.newInstance();
-        try {
-            outputFactory.setProperty("com.ctc.wstx.useDoubleQuotesInXmlDecl", Boolean.TRUE);
-        } catch (IllegalArgumentException e) {
-            logger.log(Level.FINEST, e.getMessage(), e);
-        }
         try {
             this.xmlEventConsumer = indent ?
                     new IndentingXMLEventWriter(outputFactory.createXMLEventWriter(writer)) :

src/test/java/org/xbib/marc/ConcurrencyTest.java

@@ -43,10 +43,10 @@ public class ConcurrencyTest {
             writer.startDocument();
             writer.beginCollection();
             for (int i = 0; i < n; i++) {
-                InputStream in = getClass().getResource(s).openStream();
+                InputStream inputStream = getClass().getResource(s).openStream();
                 executorService.submit(() -> {
                     Marc.builder()
-                            .setInputStream(in)
+                            .setInputStream(inputStream)
                             .setMarcRecordListener(writer)
                             .build()
                             .writeRecords();

src/test/java/org/xbib/marc/MarcFieldTest.java

@@ -160,7 +160,7 @@ public class MarcFieldTest {
                 .subfield("d", null)
                 .subfield("e", null)
                 .build();
-        assertEquals("901$  $ade", marcField.toKey());
+        assertEquals("901$__$ade", marcField.toKey());
     }
 
     @Test

src/test/java/org/xbib/marc/xml/MarcEventConsumerTest.java

@@ -7,7 +7,10 @@ import org.xbib.marc.Marc;
 import org.xmlunit.matchers.CompareMatcher;
 import java.io.InputStream;
 import java.io.StringWriter;
+import java.io.Writer;
 import java.nio.charset.StandardCharsets;
+import java.nio.file.Files;
+import java.nio.file.Paths;
 import javax.xml.stream.XMLInputFactory;
 
 /**
@@ -34,11 +37,9 @@ public class MarcEventConsumerTest {
         writer.setFormat("AlephXML").setType("Bibliographic");
         writer.startDocument();
         writer.beginCollection();
-
         MarcXchangeEventConsumer consumer = new MarcXchangeEventConsumer();
         consumer.setMarcListener(writer);
         consumer.addNamespace("http://www.ddb.de/professionell/mabxml/mabxml-1.xsd");
-
         Marc.builder()
                 .setInputStream(in)
                 .setCharset(StandardCharsets.UTF_8)

src/test/resources/org/xbib/marc/chabon.mrc.xml

@@ -1 +1 @@
-<?xml version="1.0" encoding="UTF-8"?><collection xmlns="info:lc/xmlns/marcxchange-v2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="info:lc/xmlns/marcxchange-v2 http://www.loc.gov/standards/iso25577/marcxchange-2-0.xsd"><record format="MARC21" type="Bibliographic"><leader>00759cam a2200229 a 4500</leader><controlfield tag="001">11939876</controlfield><controlfield tag="005">20041229190604.0</controlfield><controlfield tag="008">000313s2000    nyu           000 1 eng  </controlfield><datafield tag="020" ind1=" " ind2=" "><subfield code="a">0679450041 (acid-free paper)</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Chabon, Michael.</subfield></datafield><datafield tag="245" ind1="1" ind2="4"><subfield code="a">The amazing adventures of Kavalier and Clay :</subfield><subfield code="b">a novel /</subfield><subfield code="c">Michael Chabon.</subfield></datafield><datafield tag="260" ind1=" " ind2=" "><subfield code="a">New York :</subfield><subfield code="b">Random House,</subfield><subfield code="c">c2000.</subfield></datafield><datafield tag="300" ind1=" " ind2=" "><subfield code="a">639 p. ;</subfield><subfield code="c">25 cm.</subfield></datafield><datafield tag="650" ind1=" " ind2="0"><subfield code="a">Comic books, strips, etc.</subfield><subfield code="x">Authorship</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="650" ind1=" " ind2="0"><subfield code="a">Heroes in mass media</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="650" ind1=" " ind2="0"><subfield code="a">Czech Americans</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="651" ind1=" " ind2="0"><subfield code="a">New York (N.Y.)</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="650" ind1=" " ind2="0"><subfield code="a">Young men</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="650" ind1=" " ind2="0"><subfield code="a">Cartoonists</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="655" ind1=" " ind2="7"><subfield code="a">Humorous stories.</subfield><subfield code="2">gsafd</subfield></datafield><datafield tag="655" ind1=" " ind2="7"><subfield code="a">Bildungsromane.</subfield><subfield code="2">gsafd</subfield></datafield></record><record format="MARC21" type="Bibliographic"><leader>00714cam a2200205 a 4500</leader><controlfield tag="001">12883376</controlfield><controlfield tag="005">20030616111422.0</controlfield><controlfield tag="008">020805s2002    nyu    j      000 1 eng  </controlfield><datafield tag="020" ind1=" " ind2=" "><subfield code="a">0786808772</subfield></datafield><datafield tag="020" ind1=" " ind2=" "><subfield code="a">0786816155 (pbk.)</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Chabon, Michael.</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">Summerland /</subfield><subfield code="c">Michael Chabon.</subfield></datafield><datafield tag="250" ind1=" " ind2=" "><subfield code="a">1st ed.</subfield></datafield><datafield tag="260" ind1=" " ind2=" "><subfield code="a">New York :</subfield><subfield code="b">Miramax Books/Hyperion Books for Children,</subfield><subfield code="c">c2002.</subfield></datafield><datafield tag="300" ind1=" " ind2=" "><subfield code="a">500 p. ;</subfield><subfield code="c">22 cm.</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Ethan Feld, the worst baseball player in the history of the game, finds himself recruited by a 100-year-old scout to help a band of fairies triumph over an ancient enemy.</subfield></datafield><datafield tag="650" ind1=" " ind2="1"><subfield code="a">Fantasy.</subfield></datafield><datafield tag="650" ind1=" " ind2="1"><subfield code="a">Baseball</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="650" ind1=" " ind2="1"><subfield code="a">Magic</subfield><subfield code="v">Fiction.</subfield></datafield></record></collection>
+<?xml version="1.0" encoding="UTF-8"?><collection xmlns="info:lc/xmlns/marcxchange-v2" xsi:schemaLocation="info:lc/xmlns/marcxchange-v2 http://www.loc.gov/standards/iso25577/marcxchange-2-0.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"><record format="MARC21" type="Bibliographic"><leader>00759cam a2200229 a 4500</leader><controlfield tag="001">11939876</controlfield><controlfield tag="005">20041229190604.0</controlfield><controlfield tag="008">000313s2000    nyu           000 1 eng  </controlfield><datafield ind2="_" ind1="_" tag="020"><subfield code="a">0679450041 (acid-free paper)</subfield></datafield><datafield ind2="_" ind1="_" tag="040"><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield ind2="_" ind1="1" tag="100"><subfield code="a">Chabon, Michael.</subfield></datafield><datafield ind2="4" ind1="1" tag="245"><subfield code="a">The amazing adventures of Kavalier and Clay :</subfield><subfield code="b">a novel /</subfield><subfield code="c">Michael Chabon.</subfield></datafield><datafield ind2="_" ind1="_" tag="260"><subfield code="a">New York :</subfield><subfield code="b">Random House,</subfield><subfield code="c">c2000.</subfield></datafield><datafield ind2="_" ind1="_" tag="300"><subfield code="a">639 p. ;</subfield><subfield code="c">25 cm.</subfield></datafield><datafield ind2="0" ind1="_" tag="650"><subfield code="a">Comic books, strips, etc.</subfield><subfield code="x">Authorship</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="0" ind1="_" tag="650"><subfield code="a">Heroes in mass media</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="0" ind1="_" tag="650"><subfield code="a">Czech Americans</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="0" ind1="_" tag="651"><subfield code="a">New York (N.Y.)</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="0" ind1="_" tag="650"><subfield code="a">Young men</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="0" ind1="_" tag="650"><subfield code="a">Cartoonists</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="7" ind1="_" tag="655"><subfield code="a">Humorous stories.</subfield><subfield code="2">gsafd</subfield></datafield><datafield ind2="7" ind1="_" tag="655"><subfield code="a">Bildungsromane.</subfield><subfield code="2">gsafd</subfield></datafield></record><record format="MARC21" type="Bibliographic"><leader>00714cam a2200205 a 4500</leader><controlfield tag="001">12883376</controlfield><controlfield tag="005">20030616111422.0</controlfield><controlfield tag="008">020805s2002    nyu    j      000 1 eng  </controlfield><datafield ind2="_" ind1="_" tag="020"><subfield code="a">0786808772</subfield></datafield><datafield ind2="_" ind1="_" tag="020"><subfield code="a">0786816155 (pbk.)</subfield></datafield><datafield ind2="_" ind1="_" tag="040"><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield ind2="_" ind1="1" tag="100"><subfield code="a">Chabon, Michael.</subfield></datafield><datafield ind2="0" ind1="1" tag="245"><subfield code="a">Summerland /</subfield><subfield code="c">Michael Chabon.</subfield></datafield><datafield ind2="_" ind1="_" tag="250"><subfield code="a">1st ed.</subfield></datafield><datafield ind2="_" ind1="_" tag="260"><subfield code="a">New York :</subfield><subfield code="b">Miramax Books/Hyperion Books for Children,</subfield><subfield code="c">c2002.</subfield></datafield><datafield ind2="_" ind1="_" tag="300"><subfield code="a">500 p. ;</subfield><subfield code="c">22 cm.</subfield></datafield><datafield ind2="_" ind1="_" tag="520"><subfield code="a">Ethan Feld, the worst baseball player in the history of the game, finds himself recruited by a 100-year-old scout to help a band of fairies triumph over an ancient enemy.</subfield></datafield><datafield ind2="1" ind1="_" tag="650"><subfield code="a">Fantasy.</subfield></datafield><datafield ind2="1" ind1="_" tag="650"><subfield code="a">Baseball</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="1" ind1="_" tag="650"><subfield code="a">Magic</subfield><subfield code="v">Fiction.</subfield></datafield></record></collection>

src/test/resources/org/xbib/marc/diacritic4.mrc.xml

@@ -1 +1 @@
-<?xml version="1.0" encoding="UTF-8"?><collection xmlns="info:lc/xmlns/marcxchange-v2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="info:lc/xmlns/marcxchange-v2 http://www.loc.gov/standards/iso25577/marcxchange-2-0.xsd"><record format="MARC21" type="Bibliographic"><leader>03059cam  2200301 i 4500</leader><controlfield tag="001">   77123332 </controlfield><controlfield tag="003">DLC</controlfield><controlfield tag="005">20051218154744.0</controlfield><controlfield tag="008">981008b2001    ilu           000 0 eng  </controlfield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield tag="010" ind1=" " ind2=" "><subfield code="a">   77123332 </subfield></datafield><datafield tag="245" ind1="0" ind2="0"><subfield code="a">OCLC diacritic and special character test record.</subfield></datafield><datafield tag="260" ind1=" " ind2=" "><subfield code="a">ny :</subfield><subfield code="b">ny,</subfield><subfield code="c">2001.</subfield></datafield><datafield tag="300" ind1=" " ind2=" "><subfield code="a">100 p. ;</subfield><subfield code="c">12 cm.</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 0  (NEW):  Degree sign (°);  Phono Copyright mark (℗);  Copyright mark (©);  Sharp (♯);  Inverted Question mark (¿);  Inverted Exclamation mark (¡); Eszett (ß); Euro (€).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 1:  Script L (ℓ);  Polish L (Ł);  Scandanavian O (Ø);  D with Crossbar (Đ);  Icelandic Thorn (Þ);  AE Digraph (Æ); OE Digraph (Œ);  Miagkii Znak (ʹ);  Dot at Midline (·).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 2:  Musical Flat (♭);  Patent Mark (®);  Plus or Minus (±);  O Hook (Ơ);  U Hook (Ư);  Alif (ʼ);  alpha (α);  Ayn (ʻ);  Polish l (ł).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 3:  Scandanavian o (ø);  d with crossbar (đ);  Icelandic Thorn (þ);  ae Digraph (æ);  oe Digraph (œ);  Tverdii Znak (ʺ);  Turkish i (ı);  British Pound (£);  eth (ð).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 4:  Dagger (DO NOT USE);  o Hook (ơ);  u Hook (ư);  Beta (β);  Gamma (γ);  Superscript 0 (⁰);  Superscript 1 (¹);  Superscript 2 (²);  Superscript 3 (³).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 5:  Superscript 4 (⁴);  Superscript 5 (⁵);  Superscript 6 (⁶);  Superscript 7 (⁷);  Superscript 8 (⁸);  Superscript 9 (⁹);  Superscript + (⁺);  Superscript - (⁻);  Superscript ( (⁽).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 6:  Superscript ) (⁾);  Subscript 0 (₀);  Subscript  1 (₁);  Subscript 2 (₂);  Subscript 3 (₃);  Subscript 4 (₄);  Subscript 5 (₅);  Subscript 6 (₆);  Subscript 7 (₇).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 7:  Subscript 8 (₈);  Subscript 9 (₉);  Subscript + (₊);  Subscript - (₋);  Subscript ( (₍);  Subscript ) (₎);  Pseudo Question Mark (ỏ);  Grave (ò);  Acute (ó).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 8:  Circumflex (ô);  Tilde (õ);  Macron (ō);  Breve (ŏ);  Superior Dot (ȯ);  Umlaut (ö);  Hacek (ǒ);  Circle Above (o̊);  Ligature left (o͡).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 9:  Ligature right (o) ;  High Comma off center (o̕);  Double Acute (ő);  Candrabindu (o̐);  Cedilla (o̧);  Right Hook (ǫ);  Dot Below (ọ);  Double Dot Below (o̤);  Circle Below (o̥).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER COLUMN 10:  Double Underscore (o̳);  Underscore (o̲);  Left Hook (o̦);  Right Cedilla (o̜);  Upadhmaniya (o̮);  Double Tilde 1st half (o͠);  Double Tilde 2nd half (o) ;  High Comma centered (o̓).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">VOYAGER PC Keyboard:  Spacing Circumflex (^); Spacing Underscore (_); Spacing Grave (`); Open Curly Bracket ({); Close Curly Bracket (}); Spacing Tilde (~).</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">Standard PC Keyboard:   1234567890-=   !@#$%^&amp;*()_+   qwertyuiop[]\   QWERTYUIOP{}|   asdfghjkl;'  ASDFGHJKL:"   zxcvbnm,./   ZXCVBNM&lt;&gt;?</subfield></datafield><datafield tag="500" ind1=" " ind2=" "><subfield code="a">Double Tilde, 1st and 2nd halves (o͠o) ; Ligature, both halves (o͡o).</subfield></datafield></record></collection>
+<?xml version="1.0" encoding="UTF-8"?><collection xmlns="info:lc/xmlns/marcxchange-v2" xsi:schemaLocation="info:lc/xmlns/marcxchange-v2 http://www.loc.gov/standards/iso25577/marcxchange-2-0.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"><record format="MARC21" type="Bibliographic"><leader>03059cam  2200301 i 4500</leader><controlfield tag="001">   77123332 </controlfield><controlfield tag="003">DLC</controlfield><controlfield tag="005">20051218154744.0</controlfield><controlfield tag="008">981008b2001    ilu           000 0 eng  </controlfield><datafield ind2="_" ind1="_" tag="040"><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield ind2="_" ind1="_" tag="010"><subfield code="a">   77123332 </subfield></datafield><datafield ind2="0" ind1="0" tag="245"><subfield code="a">OCLC diacritic and special character test record.</subfield></datafield><datafield ind2="_" ind1="_" tag="260"><subfield code="a">ny :</subfield><subfield code="b">ny,</subfield><subfield code="c">2001.</subfield></datafield><datafield ind2="_" ind1="_" tag="300"><subfield code="a">100 p. ;</subfield><subfield code="c">12 cm.</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 0  (NEW):  Degree sign (°);  Phono Copyright mark (℗);  Copyright mark (©);  Sharp (♯);  Inverted Question mark (¿);  Inverted Exclamation mark (¡); Eszett (ß); Euro (€).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 1:  Script L (ℓ);  Polish L (Ł);  Scandanavian O (Ø);  D with Crossbar (Đ);  Icelandic Thorn (Þ);  AE Digraph (Æ); OE Digraph (Œ);  Miagkii Znak (ʹ);  Dot at Midline (·).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 2:  Musical Flat (♭);  Patent Mark (®);  Plus or Minus (±);  O Hook (Ơ);  U Hook (Ư);  Alif (ʼ);  alpha (α);  Ayn (ʻ);  Polish l (ł).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 3:  Scandanavian o (ø);  d with crossbar (đ);  Icelandic Thorn (þ);  ae Digraph (æ);  oe Digraph (œ);  Tverdii Znak (ʺ);  Turkish i (ı);  British Pound (£);  eth (ð).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 4:  Dagger (DO NOT USE);  o Hook (ơ);  u Hook (ư);  Beta (β);  Gamma (γ);  Superscript 0 (⁰);  Superscript 1 (¹);  Superscript 2 (²);  Superscript 3 (³).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 5:  Superscript 4 (⁴);  Superscript 5 (⁵);  Superscript 6 (⁶);  Superscript 7 (⁷);  Superscript 8 (⁸);  Superscript 9 (⁹);  Superscript + (⁺);  Superscript - (⁻);  Superscript ( (⁽).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 6:  Superscript ) (⁾);  Subscript 0 (₀);  Subscript  1 (₁);  Subscript 2 (₂);  Subscript 3 (₃);  Subscript 4 (₄);  Subscript 5 (₅);  Subscript 6 (₆);  Subscript 7 (₇).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 7:  Subscript 8 (₈);  Subscript 9 (₉);  Subscript + (₊);  Subscript - (₋);  Subscript ( (₍);  Subscript ) (₎);  Pseudo Question Mark (ỏ);  Grave (ò);  Acute (ó).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 8:  Circumflex (ô);  Tilde (õ);  Macron (ō);  Breve (ŏ);  Superior Dot (ȯ);  Umlaut (ö);  Hacek (ǒ);  Circle Above (o̊);  Ligature left (o͡).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 9:  Ligature right (o) ;  High Comma off center (o̕);  Double Acute (ő);  Candrabindu (o̐);  Cedilla (o̧);  Right Hook (ǫ);  Dot Below (ọ);  Double Dot Below (o̤);  Circle Below (o̥).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER COLUMN 10:  Double Underscore (o̳);  Underscore (o̲);  Left Hook (o̦);  Right Cedilla (o̜);  Upadhmaniya (o̮);  Double Tilde 1st half (o͠);  Double Tilde 2nd half (o) ;  High Comma centered (o̓).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">VOYAGER PC Keyboard:  Spacing Circumflex (^); Spacing Underscore (_); Spacing Grave (`); Open Curly Bracket ({); Close Curly Bracket (}); Spacing Tilde (~).</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">Standard PC Keyboard:   1234567890-=   !@#$%^&amp;*()_+   qwertyuiop[]\   QWERTYUIOP{}|   asdfghjkl;'  ASDFGHJKL:"   zxcvbnm,./   ZXCVBNM&lt;&gt;?</subfield></datafield><datafield ind2="_" ind1="_" tag="500"><subfield code="a">Double Tilde, 1st and 2nd halves (o͠o) ; Ligature, both halves (o͡o).</subfield></datafield></record></collection>

src/test/resources/org/xbib/marc/summerland.mrc.xml

@@ -1 +1 @@
-<?xml version="1.0" encoding="UTF-8"?><collection xmlns="info:lc/xmlns/marcxchange-v2" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance" xsi:schemaLocation="info:lc/xmlns/marcxchange-v2 http://www.loc.gov/standards/iso25577/marcxchange-2-0.xsd"><record format="MARC21" type="Bibliographic"><leader>00714cam a2200205 a 4500</leader><controlfield tag="001">12883376</controlfield><controlfield tag="005">20030616111422.0</controlfield><controlfield tag="008">020805s2002    nyu    j      000 1 eng  </controlfield><datafield tag="020" ind1=" " ind2=" "><subfield code="a">0786808772</subfield></datafield><datafield tag="020" ind1=" " ind2=" "><subfield code="a">0786816155 (pbk.)</subfield></datafield><datafield tag="040" ind1=" " ind2=" "><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield tag="100" ind1="1" ind2=" "><subfield code="a">Chabon, Michael.</subfield></datafield><datafield tag="245" ind1="1" ind2="0"><subfield code="a">Summerland /</subfield><subfield code="c">Michael Chabon.</subfield></datafield><datafield tag="250" ind1=" " ind2=" "><subfield code="a">1st ed.</subfield></datafield><datafield tag="260" ind1=" " ind2=" "><subfield code="a">New York :</subfield><subfield code="b">Miramax Books/Hyperion Books for Children,</subfield><subfield code="c">c2002.</subfield></datafield><datafield tag="300" ind1=" " ind2=" "><subfield code="a">500 p. ;</subfield><subfield code="c">22 cm.</subfield></datafield><datafield tag="520" ind1=" " ind2=" "><subfield code="a">Ethan Feld, the worst baseball player in the history of the game, finds himself recruited by a 100-year-old scout to help a band of fairies triumph over an ancient enemy.</subfield></datafield><datafield tag="650" ind1=" " ind2="1"><subfield code="a">Fantasy.</subfield></datafield><datafield tag="650" ind1=" " ind2="1"><subfield code="a">Baseball</subfield><subfield code="v">Fiction.</subfield></datafield><datafield tag="650" ind1=" " ind2="1"><subfield code="a">Magic</subfield><subfield code="v">Fiction.</subfield></datafield></record></collection>
+<?xml version="1.0" encoding="UTF-8"?><collection xmlns="info:lc/xmlns/marcxchange-v2" xsi:schemaLocation="info:lc/xmlns/marcxchange-v2 http://www.loc.gov/standards/iso25577/marcxchange-2-0.xsd" xmlns:xsi="http://www.w3.org/2001/XMLSchema-instance"><record format="MARC21" type="Bibliographic"><leader>00714cam a2200205 a 4500</leader><controlfield tag="001">12883376</controlfield><controlfield tag="005">20030616111422.0</controlfield><controlfield tag="008">020805s2002    nyu    j      000 1 eng  </controlfield><datafield ind2="_" ind1="_" tag="020"><subfield code="a">0786808772</subfield></datafield><datafield ind2="_" ind1="_" tag="020"><subfield code="a">0786816155 (pbk.)</subfield></datafield><datafield ind2="_" ind1="_" tag="040"><subfield code="a">DLC</subfield><subfield code="c">DLC</subfield><subfield code="d">DLC</subfield></datafield><datafield ind2="_" ind1="1" tag="100"><subfield code="a">Chabon, Michael.</subfield></datafield><datafield ind2="0" ind1="1" tag="245"><subfield code="a">Summerland /</subfield><subfield code="c">Michael Chabon.</subfield></datafield><datafield ind2="_" ind1="_" tag="250"><subfield code="a">1st ed.</subfield></datafield><datafield ind2="_" ind1="_" tag="260"><subfield code="a">New York :</subfield><subfield code="b">Miramax Books/Hyperion Books for Children,</subfield><subfield code="c">c2002.</subfield></datafield><datafield ind2="_" ind1="_" tag="300"><subfield code="a">500 p. ;</subfield><subfield code="c">22 cm.</subfield></datafield><datafield ind2="_" ind1="_" tag="520"><subfield code="a">Ethan Feld, the worst baseball player in the history of the game, finds himself recruited by a 100-year-old scout to help a band of fairies triumph over an ancient enemy.</subfield></datafield><datafield ind2="1" ind1="_" tag="650"><subfield code="a">Fantasy.</subfield></datafield><datafield ind2="1" ind1="_" tag="650"><subfield code="a">Baseball</subfield><subfield code="v">Fiction.</subfield></datafield><datafield ind2="1" ind1="_" tag="650"><subfield code="a">Magic</subfield><subfield code="v">Fiction.</subfield></datafield></record></collection>