MODINV-1044 Additional Requirements - Update Data Import logic to nor…

…malize OCLC 035 values (#630)

* MODINV-1044 Additional Requirements - Update Data Import logic to normalize OCLC 035 values

* Update AdditionalFieldsUtilTest.java

NEWS.md

@@ -3,6 +3,7 @@
 * [MODSOURCE-756](https://issues.folio.org/browse/MODSOURCE-756) After setting an instance as marked for deletion it is no longer editable in quickmarc
 * [MODSOURCE-753](https://folio-org.atlassian.net/browse/MODSOURCE-753) Change SQL query parameters for MARC Search
 * [MODSOURCE-773](https://folio-org.atlassian.net/browse/MODSOURCE-773) MARC Search omits suppressed from discovery records in default search
+* [MODINV-1044](https://folio-org.atlassian.net/browse/MODINV-1044) Additional Requirements - Update Data Import logic to normalize OCLC 035 values
 
 ## 2024-03-20 5.8.0
 * [MODSOURCE-733](https://issues.folio.org/browse/MODSOURCE-733) Reduce Memory Allocation of Strings

mod-source-record-storage-server/src/main/java/org/folio/services/util/AdditionalFieldsUtil.java

@@ -72,7 +72,7 @@ public final class AdditionalFieldsUtil {
   private static final char HR_ID_FIELD_IND = ' ';
   private static final String ANY_STRING = "*";
   private static final String OCLC = "OCoLC";
-  private static final String OCLC_PATTERN = "\\((" + OCLC + ")\\)((ocm|ocn)?0*|([a-zA-Z]+)0*)(\\d+\\w*)";
+  private static final String OCLC_PATTERN = "\\((" + OCLC + ")\\)((ocm|ocn|on)?0*|([a-zA-Z]+)0*)(\\d+\\w*)";
 
 
   public static final DateTimeFormatter dateTime005Formatter = DateTimeFormatter.ofPattern("yyyyMMddHHmmss.S");
@@ -460,7 +460,7 @@ private static Set<String> formatOclc(List<Subfield> subFields) {
     Pattern pattern = Pattern.compile(OCLC_PATTERN);
 
     for (Subfield subfield : subFields) {
-      String data = subfield.getData();
+      String data = subfield.getData().replaceAll("[.\\s]", "");
       var code = subfield.getCode();
       Matcher matcher = pattern.matcher(data);
 
@@ -469,7 +469,7 @@ private static Set<String> formatOclc(List<Subfield> subFields) {
         String numericAndTrailing = matcher.group(5); // Numeric part and any characters that follow
         String prefix = matcher.group(2); // Entire prefix including letters and potentially leading zeros
 
-        if (prefix != null && (prefix.startsWith("ocm") || prefix.startsWith("ocn"))) {
+        if (prefix != null && (prefix.startsWith("ocm") || prefix.startsWith("ocn") || prefix.startsWith("on"))) {
           // If "ocm" or "ocn", strip entirely from the prefix
           processedSet.add(code + "&(" + oclcTag + ")" + numericAndTrailing);
         } else {

mod-source-record-storage-server/src/test/java/org/folio/services/AdditionalFieldsUtilTest.java

@@ -359,6 +359,28 @@ public void shouldReturnSubfieldIfOclcExist() {
     Assert.assertEquals(expectedSubfields.get(0), subfields.get(0));
   }
 
+  @Test
+  public void shouldRemovePeriodsAndSpacesAfterNormalization() {
+    // given
+    var parsedContent = "{\"leader\":\"00120nam  22000731a 4500\",\"fields\":[{\"001\":\"in001\"}," +
+      "{\"035\":{\"subfields\":[{\"a\":\"(OCoLC)on. 607TST .001\"}],\"ind1\":\" \",\"ind2\":\" \"}}," +
+      "{\"500\":{\"subfields\":[{\"a\":\"data\"}],\"ind1\":\" \",\"ind2\":\" \"}}]}";
+
+    var expectedParsedContent = "{\"leader\":\"00098nam  22000611a 4500\",\"fields\":[{\"001\":\"in001\"}," +
+      "{\"035\":{\"subfields\":[{\"a\":\"(OCoLC)607TST001\"}],\"ind1\":\" \",\"ind2\":\" \"}}," +
+      "{\"500\":{\"subfields\":[{\"a\":\"data\"}],\"ind1\":\" \",\"ind2\":\" \"}}]}";
+    ParsedRecord parsedRecord = new ParsedRecord().withContent(parsedContent);
+
+    Record record = new Record().withId(UUID.randomUUID().toString())
+      .withParsedRecord(parsedRecord)
+      .withGeneration(0)
+      .withState(Record.State.ACTUAL)
+      .withExternalIdsHolder(new ExternalIdsHolder().withInstanceId("001").withInstanceHrid("in001"));
+    // when
+    AdditionalFieldsUtil.normalize035(record);
+    Assert.assertEquals(expectedParsedContent, parsedRecord.getContent());
+  }
+
   @Test
   public void shouldNotReturnSubfieldIfOclcNotExist() {
     // given