Making occurrence a first class citizen of PICA data fields #234

pkiraly · Feb 26, 2023 · 6d18a6d · 6d18a6d
1 parent 393a71f
commit 6d18a6d
Show file tree

Hide file tree

Showing 6 changed files with 34 additions and 13 deletions.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/MarcSubfield.java b/src/main/java/de/gwdg/metadataqa/marc/MarcSubfield.java
@@ -130,7 +130,10 @@ public Map<String, List<String>> getKeyValuePairs(DataFieldKeyGenerator keyGener
       prefixCache = new HashMap<>();
     }
 
-    String cacheKey = String.format("%s$%s-%s-%s", this.getField().getTag(), code, keyGenerator.getType().getType(), keyGenerator.getMarcVersion());
+    String tag = this.getField().getTag();
+    if (this.getField().getOccurrence() != null)
+      tag += "/" + this.getField().getOccurrence();
+    String cacheKey = String.format("%s$%s-%s-%s", tag, code, keyGenerator.getType().getType(), keyGenerator.getMarcVersion());
     if (!prefixCache.containsKey(cacheKey))
       prefixCache.put(cacheKey, keyGenerator.forSubfield(this));
     String prefix = prefixCache.get(cacheKey);

diff --git a/src/main/java/de/gwdg/metadataqa/marc/dao/DataField.java b/src/main/java/de/gwdg/metadataqa/marc/dao/DataField.java
@@ -323,7 +323,12 @@ public Map<String, List<String>> getKeyValuePairs(SolrFieldType type,
                                                     MarcVersion marcVersion) {
     Map<String, List<String>> pairs = new HashMap<>();
 
-    DataFieldKeyGenerator keyGenerator = new DataFieldKeyGenerator(definition, type, getTag());
+    String tag = getTag();
+    if (getOccurrence() != null)
+        tag += "_" + getOccurrence();
+
+    SchemaType schemaType = marcRecord != null ? marcRecord.getSchemaType() : SchemaType.MARC21;
+    DataFieldKeyGenerator keyGenerator = new DataFieldKeyGenerator(definition, type, tag, schemaType);
     keyGenerator.setMarcVersion(marcVersion);
 
     // ind1

diff --git a/src/main/java/de/gwdg/metadataqa/marc/dao/record/BibliographicRecord.java b/src/main/java/de/gwdg/metadataqa/marc/dao/record/BibliographicRecord.java
@@ -430,9 +430,11 @@ public String asJson() {
         }
         fieldMap.put("subfields", subfields);
 
-        String tag = field.getDefinition() != null
-          ? field.getDefinition().getTag()
-          : field.getTag();
+        String tag = field.getOccurrence() != null
+          ? field.getTag() + "/" + field.getOccurrence()
+          : (field.getDefinition() != null
+            ? field.getDefinition().getTag()
+            : field.getTag());
 
         map.computeIfAbsent(tag, s -> new ArrayList<Map<String, Object>>());
         ((ArrayList)map.get(tag)).add(fieldMap);

diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/keygenerator/DataFieldKeyGenerator.java b/src/main/java/de/gwdg/metadataqa/marc/utils/keygenerator/DataFieldKeyGenerator.java
@@ -2,6 +2,7 @@
 
 import de.gwdg.metadataqa.marc.MarcSubfield;
 import de.gwdg.metadataqa.marc.definition.MarcVersion;
+import de.gwdg.metadataqa.marc.definition.bibliographic.SchemaType;
 import de.gwdg.metadataqa.marc.definition.structure.DataFieldDefinition;
 import de.gwdg.metadataqa.marc.definition.structure.SubfieldDefinition;
 import de.gwdg.metadataqa.marc.model.SolrFieldType;
@@ -28,15 +29,21 @@ public DataFieldKeyGenerator(DataFieldDefinition definition, SolrFieldType type)
 
   public DataFieldKeyGenerator(DataFieldDefinition definition,
                                SolrFieldType type,
-                               String tag) {
+                               String tag,
+                               SchemaType schemaType) {
     this.definition = definition;
     this.type = type;
-    if (definition != null) {
-      this.tag = definition.getTag();
-      indexTag = definition.getIndexTag();
-    } else {
+    if (schemaType.equals(SchemaType.PICA)) {
       this.tag = tag;
-      indexTag = tag;
+      this.indexTag = tag;
+    } else {
+      if (definition != null) {
+        this.tag = definition.getTag();
+        this.indexTag = definition.getIndexTag();
+      } else {
+        this.tag = tag;
+        this.indexTag = tag;
+      }
     }
   }
 
@@ -191,4 +198,8 @@ public SolrFieldType getType() {
   public MarcVersion getMarcVersion() {
     return marcVersion;
   }
+
+  public String getTag() {
+    return tag;
+  }
 }
diff --git a/src/test/java/de/gwdg/metadataqa/marc/cli/MarcToSolrTest.java b/src/test/java/de/gwdg/metadataqa/marc/cli/MarcToSolrTest.java
@@ -44,7 +44,8 @@ public void pica() throws Exception {
     BibliographicRecord marcRecord = MarcFactory.createPicaFromMarc4j(record, schema);
     Map<String, List<String>> map = marcRecord.getKeyValuePairs(SolrFieldType.HUMAN, true, MarcVersion.MARC21);
     // System.err.println(map.keySet());
-    System.err.println(marcRecord.asJson());
+    assertTrue(marcRecord.asJson().contains("036E/01"));
+    assertTrue(map.containsKey("036E_01_a"));
   }
 
   @Test

diff --git a/src/test/java/de/gwdg/metadataqa/marc/cli/ValidatorCliTest.java b/src/test/java/de/gwdg/metadataqa/marc/cli/ValidatorCliTest.java
@@ -2,7 +2,6 @@
 
 import de.gwdg.metadataqa.api.util.FileUtils;
 import de.gwdg.metadataqa.marc.cli.utils.RecordIterator;
-import org.apache.commons.lang3.StringUtils;
 import org.junit.Before;
 import org.junit.Test;