issue #154: subject indexing analysis

pkiraly · Aug 17, 2022 · 264aa0d · 264aa0d
1 parent 3c876bf
commit 264aa0d
Showing 1 changed file with 49 additions and 1 deletion.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/analysis/ClassificationAnalyzer.java b/src/main/java/de/gwdg/metadataqa/marc/analysis/ClassificationAnalyzer.java
@@ -103,7 +103,7 @@ public int process() {
       total = processFieldsWithoutSource(total);
       total = processFieldsWithScheme(total, MARC21_FIELD_WITH_SCHEMES);
     } else if (marcRecord.getSchemaType().equals(SchemaType.PICA)) {
-      total = processFieldsWithScheme(total, PICA_FIELDS_WITH_SCHEME);
+      total = processFieldsWithSchemePica(total, PICA_FIELDS_WITH_SCHEME);
     }
 
     increaseCounters(total);
@@ -135,6 +135,15 @@ private int processFieldsWithScheme(int total, List<FieldWithScheme> fieldsWithS
     return total;
   }
 
+  private int processFieldsWithSchemePica(int total, List<FieldWithScheme> fieldsWithScheme) {
+    for (FieldWithScheme fieldWithScheme : fieldsWithScheme) {
+      var count = processFieldWithSchemePica(marcRecord, fieldWithScheme);
+      if (count > 0)
+        total += count;
+    }
+    return total;
+  }
+
   private int processFieldsWithoutSource(int total) {
     for (String tag : fieldsWithoutSource) {
       var count = processFieldWithoutSource(marcRecord, tag);
@@ -213,6 +222,45 @@ private int processFieldWithScheme(MarcRecord marcRecord,
     return count;
   }
 
+  private int processFieldWithSchemePica(MarcRecord marcRecord,
+                                         FieldWithScheme fieldEntry) {
+    var count = 0;
+    final String tag = fieldEntry.getTag();
+    if (!marcRecord.hasDatafield(tag))
+      return count;
+
+    List<DataField> fields = marcRecord.getDatafield(tag);
+    List<Schema> schemas = new ArrayList<>();
+    for (DataField field : fields) {
+      String firstSubfield = null;
+      // String alt = null;
+      if (field.getSubfield("a") != null) {
+        firstSubfield = "$a";
+      } else {
+        for (MarcSubfield subfield : field.getSubfields()) {
+          String code = subfield.getCode();
+          if (!code.equals("A")) {
+            firstSubfield = "$" + code;
+            break;
+          }
+        }
+      }
+      if (firstSubfield != null) {
+        var scheme = fieldEntry.getSchemaName();
+        var currentSchema = new Schema(tag, firstSubfield, classificationSchemes.resolve(scheme), scheme);
+        schemas.add(currentSchema);
+        updateSchemaSubfieldStatistics(field, currentSchema);
+        count++;
+      } else {
+        logger.severe(String.format("undetected subfield in record %s %s", marcRecord.getId(), field.toString()));
+      }
+    }
+
+    registerSchemas(schemas);
+
+    return count;
+  }
+
   private void registerSchemas(List<Schema> schemas) {
     addSchemasToStatistics(statistics.getInstances(), schemas);