issue #99: implementing B3Kat union catalogue local defined data elem…

…ents.

README.md

@@ -244,6 +244,8 @@ options:
    * `BL`, fields available at the British Library
    * `MARC21NO`, fields available at the MARC21 profile for Norwegian public libraries
    * `UVA`, fields available at the University of Amsterdam Library
+   * `B3KAT`, fields available at the B3Kat union catalogue of Bibliotheksverbundes 
+     Bayern (BVB) and Kooperativen Bibliotheksverbundes Berlin-Brandenburg (KOBV)    
 * output parameters:
   * `-t [directory]`, `--outputDir [directory]` specifies the output
     directory where the files will be created

scripts/bayern.sh

@@ -3,8 +3,8 @@
 . ./setdir.sh
 NAME=bayern
 MARC_DIR=${BASE_INPUT_DIR}/bayern/marc
-TYPE_PARAMS="--marcxml"
-MASK=*.xml
+TYPE_PARAMS="--marcxml --marcVersion B3KAT --emptyLargeCollectors"
+MASK=b3kat*.xml
 
 . ./common-script
 

src/main/java/de/gwdg/metadataqa/marc/Utils.java

@@ -72,6 +72,7 @@ public static MarcVersion package2version(String packageName) {
       case "sztetags":    version = MarcVersion.SZTE;    break;
       case "nkcrtags":    version = MarcVersion.NKCR;    break;
       case "uvatags":     version = MarcVersion.UVA;     break;
+      case "b3kattags":   version = MarcVersion.B3KAT;   break;
       default:            version = MarcVersion.MARC21;  break;
     }
     return version;

src/main/java/de/gwdg/metadataqa/marc/dao/Control005.java

@@ -64,33 +64,46 @@ protected void processContent() {
         String cleanContent = content.replaceAll("[\\. ]*$", "").replaceAll("\\.", "");
         // logger.warning(String.format("005 ('%s') does not match the expected pattern", content));
         if (cleanContent.length() >= 4) {
-          year = Integer.parseInt(cleanContent.substring(0, 4));
-          cleanContent = cleanContent.substring(5);
+          year = extractRaw(cleanContent, 4, "year");
+          cleanContent = cleanContent.substring(4);
         }
         if (cleanContent.length() >= 2) {
-          month = Integer.parseInt(cleanContent.substring(0, 2));
-          cleanContent = cleanContent.substring(3);
+          month = extractRaw(cleanContent, 2, "month"); // Integer.parseInt(cleanContent.substring(0, 2));
+          cleanContent = cleanContent.substring(2);
         }
         if (cleanContent.length() >= 2) {
-          day = Integer.parseInt(cleanContent.substring(0, 2));
-          cleanContent = cleanContent.substring(3);
+          day = extractRaw(cleanContent, 2, "day"); // Integer.parseInt(cleanContent.substring(0, 2));
+          cleanContent = cleanContent.substring(2);
         }
         if (cleanContent.length() >= 2) {
-          hour = Integer.parseInt(cleanContent.substring(0, 2));
-          cleanContent = cleanContent.substring(3);
+          hour = extractRaw(cleanContent, 2, "hour"); // Integer.parseInt(cleanContent.substring(0, 2));
+          cleanContent = cleanContent.substring(2);
         }
         if (cleanContent.length() >= 2) {
-          min = Integer.parseInt(cleanContent.substring(0, 2));
-          cleanContent = cleanContent.substring(3);
+          min = extractRaw(cleanContent, 2, "min"); // Integer.parseInt(cleanContent.substring(0, 2));
+          cleanContent = cleanContent.substring(2);
         }
         if (cleanContent.length() >= 2) {
-          sec = Integer.parseInt(cleanContent.substring(0, 2));
-          cleanContent = cleanContent.substring(3);
+          sec = extractRaw(cleanContent, 2, "sec"); // Integer.parseInt(cleanContent.substring(0, 2));
+          cleanContent = cleanContent.substring(2);
         }
       }
     }
   }
 
+  private Integer extractRaw(String cleanContent, int end, String field) {
+    String raw = cleanContent.substring(0, end);
+    Integer data = null;
+    try {
+      data = Integer.parseInt(raw);
+    } catch (NumberFormatException e) {
+      String id = marcRecord != null ? String.format("#%s) ", marcRecord.getId()) : "";
+      logger.severe(String.format("%sBad input for %s: %s", id, field, raw));
+      initializationErrors.add(createError(String.format("invalid %s: %s", field, raw)));
+    }
+    return data;
+  }
+
   @Override
   public boolean validate(MarcVersion marcVersion) {
     validationErrors = new ArrayList();

src/main/java/de/gwdg/metadataqa/marc/definition/MarcVersion.java

@@ -11,7 +11,8 @@ public enum MarcVersion {
   SZTE("SZTE", "Szegedi Tudományegyetem"),
   UNIMARC("UNIMARC", "UNIMARC"),
   MARC21NO("MARC21NO", "MARC21 profile for Norwegian public libraries"),
-  UVA("UVA", "University of Amsterdam")
+  UVA("UVA", "University of Amsterdam"),
+  B3KAT("B3KAT", "B3Kat union catalogue of Bibliotheksverbundes Bayern (BVB) and Kooperativen Bibliotheksverbundes Berlin-Brandenburg (KOBV)"),
   ;
 
   String code;

src/main/java/de/gwdg/metadataqa/marc/definition/TagDefinitionLoader.java

@@ -79,7 +79,10 @@ private static MarcVersion getMarcVersion(Class<? extends DataFieldDefinition> d
       version = MarcVersion.BL;
     } else if (definitionClazz.getCanonicalName().contains(".uvatags.")) {
       version = MarcVersion.UVA;
+    } else if (definitionClazz.getCanonicalName().contains(".b3kattags.")) {
+      version = MarcVersion.B3KAT;
     }
+
     return version;
   }
 

src/main/java/de/gwdg/metadataqa/marc/definition/general/codelist/B3KatIdentifiers.java

@@ -0,0 +1,29 @@
+package de.gwdg.metadataqa.marc.definition.general.codelist;
+
+import de.gwdg.metadataqa.marc.utils.EncodedValueFileReader;
+
+/**
+ * Übersicht über die am B3Kat beteiligten Bibliotheken
+ * https://www.bib-bvb.de/BibList/b3kat-biblist.html
+ */
+public class B3KatIdentifiers extends CodeList {
+
+  private void initialize() {
+    name = "MARC Organization Codes";
+    url = "http://www.loc.gov/marc/organizations/orgshome.html";
+    codes = EncodedValueFileReader.fileToCodeList("marc/b3kat.isil.csv");
+    indexCodes();
+  }
+
+  private static B3KatIdentifiers uniqueInstance;
+
+  private B3KatIdentifiers() {
+    initialize();
+  }
+
+  public static B3KatIdentifiers getInstance() {
+    if (uniqueInstance == null)
+      uniqueInstance = new B3KatIdentifiers();
+    return uniqueInstance;
+  }
+}

src/main/java/de/gwdg/metadataqa/marc/definition/tags/TagCategory.java

@@ -27,6 +27,7 @@ public enum TagCategory {
   nkcr(19, "nkcrtags", "NKCR", "Locally defined tags of NKCR", false),
   bl(20, "bltags", "BL", "Locally defined tags of the British Library", false),
   uva(21, "uvatags", "UvA", "Locally defined tags of University of Amsterdam", false),
+  b3kat(22, "b3kat", "B3Kat", "Locally defined tags of a German union cataogue B3Kat", false),
   other(99, "unknown", "unknown", "unknown origin", false)
   ;
 

src/main/java/de/gwdg/metadataqa/marc/definition/tags/b3kattags/Tag049.java

@@ -0,0 +1,53 @@
+package de.gwdg.metadataqa.marc.definition.tags.b3kattags;
+
+import de.gwdg.metadataqa.marc.codes.B3KatIdentifierReader;
+import de.gwdg.metadataqa.marc.definition.Cardinality;
+import de.gwdg.metadataqa.marc.definition.general.codelist.B3KatIdentifiers;
+import de.gwdg.metadataqa.marc.definition.general.codelist.OrganizationCodes;
+import de.gwdg.metadataqa.marc.definition.general.validator.RegexValidator;
+import de.gwdg.metadataqa.marc.definition.structure.DataFieldDefinition;
+import de.gwdg.metadataqa.marc.definition.structure.Indicator;
+
+/**
+ * Besitznachweis (ISIL)
+ * https://www.bib-bvb.de/web/b3kat/open-data
+ */
+public class Tag049 extends DataFieldDefinition {
+
+  private static Tag049 uniqueInstance;
+
+  private Tag049() {
+    initialize();
+    postCreation();
+  }
+
+  public static Tag049 getInstance() {
+    if (uniqueInstance == null)
+      uniqueInstance = new Tag049();
+    return uniqueInstance;
+  }
+
+  private void initialize() {
+
+    tag = "049";
+    label = "Besitznachweis (ISIL)";
+    mqTag = "BesitznachweisISIL";
+    cardinality = Cardinality.Repeatable;
+    descriptionUrl = "https://www.bib-bvb.de/web/b3kat/open-data";
+    // setCompilanceLevels("O");
+
+    ind1 = new Indicator();
+
+    ind2 = new Indicator();
+
+    setSubfieldsWithCardinality(
+      "a", "Besitznachweis", "R"
+    );
+
+    // TODO: validator: possible ISIL numbers: https://www.bib-bvb.de/BibList/b3kat-biblist.html
+    getSubfield("a")
+      .setMqTag("Besitznachweis")
+      .setCodeList(B3KatIdentifiers.getInstance());
+    ;
+  }
+}

src/main/java/de/gwdg/metadataqa/marc/definition/tags/b3kattags/Tag940.java

@@ -0,0 +1,55 @@
+package de.gwdg.metadataqa.marc.definition.tags.b3kattags;
+
+import de.gwdg.metadataqa.marc.definition.Cardinality;
+import de.gwdg.metadataqa.marc.definition.structure.DataFieldDefinition;
+import de.gwdg.metadataqa.marc.definition.structure.Indicator;
+
+/**
+ * Regionale und lokale Kodierungen (MAB 078)
+ * https://www.bib-bvb.de/web/b3kat/open-data
+ */
+public class Tag940 extends DataFieldDefinition {
+
+  private static Tag940 uniqueInstance;
+
+  private Tag940() {
+    initialize();
+    postCreation();
+  }
+
+  public static Tag940 getInstance() {
+    if (uniqueInstance == null)
+      uniqueInstance = new Tag940();
+    return uniqueInstance;
+  }
+
+  private void initialize() {
+
+    tag = "940";
+    label = "Regionale und lokale Kodierungen (MAB 078)";
+    mqTag = "RegionaleUndLokaleKodierungen";
+    cardinality = Cardinality.Nonrepeatable;
+    descriptionUrl = "https://www.bib-bvb.de/web/b3kat/open-data";
+    // setCompilanceLevels("O");
+
+    ind1 = new Indicator("")
+      .setCodes(
+        "1", "",
+        "2", ""
+      );
+
+    ind2 = new Indicator();
+
+    setSubfieldsWithCardinality(
+      "f", "Selektionskennzeichen Sprachkreis", "NR",
+      "n", "Selektionskennzeichen bibliotheksübergreifende Bibliographien und Projekte", "NR",
+      "q", "Bibliotheksspezifische Selektionskennzeichen", "NR",
+      "r", "ISIL der redigierenden Bibliothek (VD18-Kontext, MAB 088 Unterfeld r)", "NR"
+    );
+
+    getSubfield("f").setMqTag("SelektionskennzeichenSprachkreis");
+    getSubfield("n").setMqTag("SelektionskennzeichenBibliotheksübergreifendeBibliographienUndProjekte");
+    getSubfield("q").setMqTag("BibliotheksspezifischeSelektionskennzeichen");
+    getSubfield("r").setMqTag("ISIL");
+  }
+}

src/main/java/de/gwdg/metadataqa/marc/definition/tags/b3kattags/Tag941.java

@@ -0,0 +1,49 @@
+package de.gwdg.metadataqa.marc.definition.tags.b3kattags;
+
+import de.gwdg.metadataqa.marc.definition.Cardinality;
+import de.gwdg.metadataqa.marc.definition.structure.DataFieldDefinition;
+import de.gwdg.metadataqa.marc.definition.structure.Indicator;
+
+/**
+ * Strukturierte Quellenangabe (unselbständige Publikationen)
+ * https://www.bib-bvb.de/web/b3kat/open-data
+ */
+public class Tag941 extends DataFieldDefinition {
+
+  private static Tag941 uniqueInstance;
+
+  private Tag941() {
+    initialize();
+    postCreation();
+  }
+
+  public static Tag941 getInstance() {
+    if (uniqueInstance == null)
+      uniqueInstance = new Tag941();
+    return uniqueInstance;
+  }
+
+  private void initialize() {
+
+    tag = "941";
+    label = "Strukturierte Quellenangabe (unselbständige Publikationen)";
+    mqTag = "StrukturierteQuellenangabe";
+    cardinality = Cardinality.Nonrepeatable;
+    descriptionUrl = "https://www.bib-bvb.de/web/b3kat/open-data";
+    // setCompilanceLevels("O");
+
+    ind1 = new Indicator();
+
+    ind2 = new Indicator();
+
+    setSubfieldsWithCardinality(
+      "b", "", "NR",
+      "h", "", "NR",
+      "j", "", "NR",
+      "m", "", "NR",
+      "t", "", "NR",
+      "r", "", "NR",
+      "s", "", "NR"
+    );
+  }
+}

src/main/java/de/gwdg/metadataqa/marc/definition/tags/b3kattags/Tag942.java

@@ -0,0 +1,53 @@
+package de.gwdg.metadataqa.marc.definition.tags.b3kattags;
+
+import de.gwdg.metadataqa.marc.definition.Cardinality;
+import de.gwdg.metadataqa.marc.definition.structure.DataFieldDefinition;
+import de.gwdg.metadataqa.marc.definition.structure.Indicator;
+
+/**
+ * DDC (Dewey Decimal Classification) analytisch
+ * https://www.bib-bvb.de/web/b3kat/open-data
+ */
+public class Tag942 extends DataFieldDefinition {
+
+  private static Tag942 uniqueInstance;
+
+  private Tag942() {
+    initialize();
+    postCreation();
+  }
+
+  public static Tag942 getInstance() {
+    if (uniqueInstance == null)
+      uniqueInstance = new Tag942();
+    return uniqueInstance;
+  }
+
+  private void initialize() {
+
+    tag = "942";
+    label = "DDC (Dewey Decimal Classification) analytisch";
+    mqTag = "DDCAnalytisch";
+    cardinality = Cardinality.Nonrepeatable;
+    descriptionUrl = "https://www.bib-bvb.de/web/b3kat/open-data";
+    // setCompilanceLevels("O");
+
+    ind1 = new Indicator("")
+      .setCodes("1", "");
+
+    ind2 = new Indicator("")
+      .setCodes("1", "");
+
+    setSubfieldsWithCardinality(
+      "c", "Grundnotation (Sachaspekt)", "NR",
+      "f", "Notation aus Hilfstafel 1 (Zeitaspekt)", "NR",
+      "g", "Notation aus Hilfstafel 2 (Geographischer Aspekt)", "NR",
+      "e", "Angabe der zugrunde liegenden DDC-Ausgabe", "NR"
+    );
+
+    getSubfield("c").setMqTag("Grundnotation");
+    getSubfield("f").setMqTag("NotationAusHilfstafel1Zeitaspekt");
+    getSubfield("g").setMqTag("NotationAusHilfstafel2GeographischerAspekt");
+    getSubfield("e").setMqTag("AngabeDerZugrundeLiegendenDDCAusgabe");
+  }
+}