issue #142: calculating PICA completeness

src/main/java/de/gwdg/metadataqa/marc/cli/Completeness.java

@@ -3,6 +3,8 @@
 import de.gwdg.metadataqa.marc.*;
 import de.gwdg.metadataqa.marc.cli.parameters.CommonParameters;
 import de.gwdg.metadataqa.marc.cli.parameters.CompletenessParameters;
+import de.gwdg.metadataqa.marc.cli.plugin.CompletenessFactory;
+import de.gwdg.metadataqa.marc.cli.plugin.CompletenessPlugin;
 import de.gwdg.metadataqa.marc.cli.processor.BibliographicInputProcessor;
 import de.gwdg.metadataqa.marc.cli.utils.RecordIterator;
 import de.gwdg.metadataqa.marc.dao.DataField;
@@ -56,11 +58,13 @@ public class Completeness implements BibliographicInputProcessor, Serializable {
   // private Map<String, Integer> fieldMap = new HashMap<>();
   private Map<String, Map<Integer, Integer>> fieldHistogram = new HashMap<>();
   private boolean readyToProcess;
+  private CompletenessPlugin plugin;
 
   public Completeness(String[] args) throws ParseException {
     parameters = new CompletenessParameters(args);
     options = parameters.getOptions();
     readyToProcess = true;
+    plugin = CompletenessFactory.create(parameters);
   }
 
   public static void main(String[] args) {
@@ -103,13 +107,7 @@ public void processRecord(MarcRecord marcRecord, int recordNumber) throws IOExce
     Map<String, Integer> recordPackageCounter = new TreeMap<>();
     // private Map<String, Map<String, Integer>>
 
-    String documentType = "dummy";
-    if (parameters.isMarc21())
-      documentType = marcRecord.getType().getValue();
-    else if (parameters.isPica()) {
-      String[] parts = parameters.getPicaRecordTypeField().split(Pattern.quote(parameters.getPicaSubfieldSeparator()));
-      documentType = marcRecord.getDatafield(parts[0]).get(0).getSubfield(parts[1]).get(0).getValue();
-    }
+    String documentType = plugin.getDocumentType(marcRecord);
     elementCardinality.computeIfAbsent(documentType, s -> new TreeMap<>());
     elementFrequency.computeIfAbsent(documentType, s -> new TreeMap<>());
 

src/main/java/de/gwdg/metadataqa/marc/cli/plugin/CompletenessFactory.java

@@ -0,0 +1,16 @@
+package de.gwdg.metadataqa.marc.cli.plugin;
+
+import de.gwdg.metadataqa.marc.cli.parameters.CompletenessParameters;
+
+public class CompletenessFactory {
+
+  public static CompletenessPlugin create(CompletenessParameters parameters) {
+    if (parameters.isMarc21()) {
+      return new Marc21CompletenessPlugin(parameters);
+    } else if (parameters.isPica()) {
+      return new PicaCompletenessPlugin(parameters);
+    }
+    return null;
+  }
+
+}

src/main/java/de/gwdg/metadataqa/marc/cli/plugin/CompletenessPlugin.java

@@ -0,0 +1,7 @@
+package de.gwdg.metadataqa.marc.cli.plugin;
+
+import de.gwdg.metadataqa.marc.dao.MarcRecord;
+
+public interface CompletenessPlugin {
+  String getDocumentType(MarcRecord marcRecord);
+}

src/main/java/de/gwdg/metadataqa/marc/cli/plugin/Marc21CompletenessPlugin.java

@@ -0,0 +1,19 @@
+package de.gwdg.metadataqa.marc.cli.plugin;
+
+import de.gwdg.metadataqa.marc.cli.parameters.CompletenessParameters;
+import de.gwdg.metadataqa.marc.dao.MarcRecord;
+
+import java.util.regex.Pattern;
+
+public class Marc21CompletenessPlugin implements CompletenessPlugin {
+  private final CompletenessParameters parameters;
+
+  public Marc21CompletenessPlugin(CompletenessParameters parameters) {
+    this.parameters = parameters;
+  }
+
+  @Override
+  public String getDocumentType(MarcRecord marcRecord) {
+    return marcRecord.getType().getValue();
+  }
+}

src/main/java/de/gwdg/metadataqa/marc/cli/plugin/PicaCompletenessPlugin.java

@@ -0,0 +1,24 @@
+package de.gwdg.metadataqa.marc.cli.plugin;
+
+import de.gwdg.metadataqa.marc.cli.parameters.CompletenessParameters;
+import de.gwdg.metadataqa.marc.dao.MarcRecord;
+
+import java.util.regex.Pattern;
+
+public class PicaCompletenessPlugin implements CompletenessPlugin {
+  private final CompletenessParameters parameters;
+  private final String field;
+  private final String subfield;
+
+  public PicaCompletenessPlugin(CompletenessParameters parameters) {
+    this.parameters = parameters;
+    String[] parts = parameters.getPicaRecordTypeField().split(Pattern.quote(parameters.getPicaSubfieldSeparator()));
+    field = parts[0];
+    subfield = parts[1];
+  }
+
+  @Override
+  public String getDocumentType(MarcRecord marcRecord) {
+    return marcRecord.getDatafield(field).get(0).getSubfield(subfield).get(0).getValue();
+  }
+}

src/main/java/de/gwdg/metadataqa/marc/cli/utils/RecordIterator.java

@@ -65,7 +65,7 @@ public void start() {
     if (parameters.getSchemaType().equals(SchemaType.PICA)) {
       String schemaFile = StringUtils.isNotEmpty(parameters.getPicaSchemaFile())
                         ? parameters.getPicaSchemaFile()
-                        : Paths.get("src/main/resources/pica/k10plus.json").toAbsolutePath().toString();
+                        : Paths.get("src/main/resources/pica/avram-k10plus.json").toAbsolutePath().toString();
       picaSchema = PicaSchemaReader.create(schemaFile);
     }