Task/add datatype error indexing

warehouse/ingest-core/src/main/java/datawave/ingest/data/config/DataTypeHelperImpl.java

@@ -59,7 +59,7 @@ public void setup(Configuration config) {
         aliaser.setup(getType(), config);
     }
 
-    private void initType(final Configuration config) {
+    protected void initType(final Configuration config) {
         if (type != null && TypeRegistry.hasInstance())
             return;
 

warehouse/ingest-core/src/main/java/datawave/ingest/data/config/FieldConfigHelper.java

@@ -12,4 +12,13 @@ public interface FieldConfigHelper {
     boolean isTokenizedField(String fieldName);
 
     boolean isReverseTokenizedField(String fieldName);
+
+    /*
+     * SETH NOTE Should this be documented? Not sure what the consensus is for abstract methods. These also force other classes to implement these methods like
+     * XMLFieldConfigHelper. Should they be added to those classes as well, or should we take another approach to adding ErrorIndexedFields?
+     */
+    boolean isErrorIndexedField(String fieldName);
+
+    boolean isErrorReverseIndexedField(String fieldName);
+
 }

warehouse/ingest-core/src/main/java/datawave/ingest/data/config/XMLFieldConfigHelper.java

@@ -204,6 +204,22 @@ public boolean isReverseTokenizedField(String fieldName) {
         return isNoMatchReverseTokenized();
     }
 
+    /*
+     * TODO: Complete these once you get the OK on keeping these in the FieldConfigHelper. Keeps the compiler nice and quiet :)
+     */
+    @Override
+    public boolean isErrorIndexedField(String fieldName) {
+        return false;
+    }
+
+    /*
+     * TODO: Complete these once you get the OK on keeping these in the FieldConfigHelper. Keeps the compiler nice and quiet :)
+     */
+    @Override
+    public boolean isErrorReverseIndexedField(String fieldName) {
+        return false;
+    }
+
     public boolean isNoMatchStored() {
         return noMatchStored;
     }

warehouse/ingest-core/src/main/java/datawave/ingest/data/config/ingest/AbstractIngestHelper.java

@@ -27,6 +27,7 @@ public abstract class AbstractIngestHelper extends DataTypeHelperImpl implements
     /* Map of field names to normalizers, null key is the default normalizer */
     protected MaskedFieldHelper mfHelper = null;
     protected Set<String> shardExclusions = new HashSet<>();
+
     protected boolean hasIndexDisallowlist = false;
     protected boolean hasReverseIndexDisallowlist = false;
 

warehouse/ingest-core/src/main/java/datawave/ingest/data/config/ingest/BaseIngestHelper.java

@@ -140,6 +140,7 @@ public abstract class BaseIngestHelper extends AbstractIngestHelper implements C
     public static final String FIELD_CONFIG_FILE = ".data.category.field.config.file";
 
     private static final String PROPERTY_MALFORMED = " property malformed: ";
+
     private static final Logger log = LoggerFactory.getLogger(BaseIngestHelper.class);
 
     private Multimap<String,datawave.data.type.Type<?>> typeFieldMap = null;
@@ -262,6 +263,12 @@ public void setup(Configuration config) {
             this.fieldConfigHelper = XMLFieldConfigHelper.load(fieldConfigFile, this);
         }
 
+        // --- INDEX_FIELDS ---
+
+        /*
+         * SETH NOTE This is most likely the start of the chunk that needs to be cloned for the error index stuff.
+         */
+
         // Process the indexed fields
         if (config.get(this.getType().typeName() + DISALLOWLIST_INDEX_FIELDS) != null) {
             if (log.isDebugEnabled()) {
@@ -295,6 +302,12 @@ public void setup(Configuration config) {
             }
         }
 
+        // --- REVERSE INDEX FIELDS ---
+
+        /*
+         * SETH NOTE This is what Laura was talking about-- the Allow/Disallow is mutually exclusive. I haven't seen this same block above for the non-reverse
+         * index fields. Maybe I need to take another look.
+         */
         // Ensure that we have only an allowlist or a disallowlist of fields to
         // reverse index
         if (config.get(this.getType().typeName() + DISALLOWLIST_REVERSE_INDEX_FIELDS) != null
@@ -343,6 +356,10 @@ public void setup(Configuration config) {
 
         }
 
+        /*
+         * SETH NOTE Not sure if I'll need what's after this. I'll start with the above block and add to it as needed.
+         */
+
         // gather the list of all indexed fields across all types
         // this list is only used for generating warnings if we are not indexing
         // something that
@@ -453,7 +470,7 @@ public void setup(Configuration config) {
         }
     }
 
-    private void moveToPatternMap(Set<String> in, Map<String,Pattern> out) {
+    protected void moveToPatternMap(Set<String> in, Map<String,Pattern> out) {
         for (Iterator<String> itr = in.iterator(); itr.hasNext();) {
             String str = itr.next();
             if (str.indexOf('*') != -1) {
@@ -1225,4 +1242,5 @@ public void updateDatawaveTypes(String fieldName, String typeClasses) {
             }
         }
     }
+
 }