Merge branch 'feature/work_100' into develop

src/main/java/org/ld4l/bib2lod/entity/Entity.java

@@ -12,6 +12,7 @@
 import org.apache.logging.log4j.LogManager;
 import org.apache.logging.log4j.Logger;
 import org.ld4l.bib2lod.datatypes.Datatype;
+import org.ld4l.bib2lod.datatypes.Ld4lCustomDatatypes.BibDatatype;
 import org.ld4l.bib2lod.ontology.DatatypeProp;
 import org.ld4l.bib2lod.ontology.NamedIndividual;
 import org.ld4l.bib2lod.ontology.ObjectProp;
@@ -285,6 +286,11 @@ public void addAttribute(DatatypeProp prop, String string, Datatype type) {
         addAttribute(prop, new Attribute(string, type));
     }
 
+    public void addLegacySourceDataAttribute(
+                DatatypeProp prop, String string) {
+        addAttribute(prop, string, BibDatatype.LEGACY_SOURCE_DATA);
+    }
+
     public void addAttribute(DatatypeProp prop, int i) {
         addAttribute(prop, new Attribute(i));                
     }

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/MarcxmlEntityBuilder.java

@@ -19,15 +19,6 @@
 
 public class MarcxmlEntityBuilder extends BaseEntityBuilder {
 
-/*
- * TODO
- * Record as a whole - done
- * Field as a whole
- * Field with one or more specific subfields
- * Field iterating through each subfield (e.g., each subfield generates a distinct entity
- * Field iterating through all subfields  (or is that just field as a whole?)
- */
-
     @SuppressWarnings("unused")
     private static final Logger LOGGER = LogManager.getLogger();
 
@@ -194,8 +185,5 @@ protected Entity buildFromString(
 
         return entity;
     }
-
-
-
-
+
 }

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/ld4l/AdminMetadataBuilder.java

@@ -41,10 +41,9 @@ public Entity build(BuildParams params) throws EntityBuilderException {
         buildChildFromControlField(Ld4lIdentifierType.defaultType(), 
                 adminMetadata, record, "001");
 
-        convert_040();      
+        convert040();      
 
-        convert_005();
-
+        convert005();
 
         if (adminMetadata.isEmpty()) {
             return null;
@@ -78,7 +77,7 @@ private void parseBuildParams(BuildParams params)
         }
     }
 
-    private void convert_040() throws EntityBuilderException {
+    private void convert040() throws EntityBuilderException {
 
         MarcxmlDataField field = record.getDataField("040");
 
@@ -174,15 +173,15 @@ private void addDescriptionConventions(MarcxmlDataField field)
         }       
     }
 
-    private void convert_005() throws EntityBuilderException {
+    private void convert005() throws EntityBuilderException {
 
-        MarcxmlControlField field_005 = record.getControlField("005");
+        MarcxmlControlField field = record.getControlField("005");
 
-        if (field_005 == null) {
+        if (field == null) {
             return;
         }    
 
-        String value = field_005.getTextValue();
+        String value = field.getTextValue();
         // Convert format 20130330145647.0 to 2013-03-30T14:56:47
         if (! PATTERN_005.matcher(value).matches()) {
             throw new EntityBuilderException(

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/ld4l/AgentBuilder.java

@@ -18,10 +18,8 @@ public class AgentBuilder extends MarcxmlEntityBuilder {
     private static ObjectProp DEFAULT_RELATIONSHIP = 
             Ld4lObjectProp.HAS_AGENT;
 
-    private Entity agent;
     private MarcxmlDataField field;
     private Entity grandparent;
-    private String name;
     private Entity parent;
     private ObjectProp relationship;
     private MarcxmlSubfield subfield;
@@ -34,17 +32,17 @@ public Entity build(BuildParams params) throws EntityBuilderException {
 
         parseBuildParams(params);
 
-        buildAgent();
+        Entity agent = buildAgent();
 
-        parent.addRelationship(relationship, agent);
+        if (agent != null) {
+            parent.addRelationship(relationship, agent);
+        }
 
         return agent;
     }
 
     private void reset() {
-        this.agent = null;
         this.field = null;
-        this.name = null;
         this.parent = null;  
         this.relationship = null;
         this.subfield = null;     
@@ -63,13 +61,12 @@ private void parseBuildParams(BuildParams params)
         if (type != null && ! (type instanceof Ld4lAgentType)) {
             throw new EntityBuilderException("Invalid agent type");
         } 
-
-        this.name = params.getValue();                
+
         this.subfield = (MarcxmlSubfield) params.getSubfield(); 
         this.field = (MarcxmlDataField) params.getField();       
-        if (name == null && subfield == null && field == null) {
-            throw new EntityBuilderException("A name value, subfield, or " +
-                    "field is required to build an agent.");
+        if (subfield == null && field == null) {
+            throw new EntityBuilderException("A subfield or field is " +
+                    "required to build an agent.");
         }
 
         this.relationship = params.getRelationship();       
@@ -80,18 +77,75 @@ private void parseBuildParams(BuildParams params)
         this.grandparent = params.getGrandparent();
     }
 
+
+    private Entity buildAgent() { 
+
+        Entity agent = null;
+
+        // Subfield only
+        if (field == null) {
+            agent = new Entity(Ld4lAgentType.defaultType());
+            agent.addAttribute(Ld4lDatatypeProp.NAME, 
+                    subfield.getTrimmedTextValue());
+
+        } else  {
+            switch (field.getTag()) {
+            case "100":
+                agent = convert100();
+                break;
+            case "260":
+                agent = convert260();
+                break;
+            default: 
+                break;
+            }   
+        }
+
+        return agent;
+    }
+
+    private Entity convert100() {
+
+        // Person or Family type
+        Type type = field.getFirstIndicator() == 3 ? 
+                Ld4lAgentType.FAMILY : Ld4lAgentType.PERSON; 
+        Entity agent = new Entity(type);
+
+        // Name
+        agent.addAttribute(Ld4lDatatypeProp.NAME, 
+                field.getSubfield('a').getTrimmedTextValue());
+
+        return agent;      
+    }
+
+    private Entity convert260() {
+
+        Entity agent = new Entity(Ld4lAgentType.defaultType());
+
+        if (subfield != null) {
+            agent.addAttribute(Ld4lDatatypeProp.NAME, 
+                    subfield.getTrimmedTextValue());
+        } 
+
+        agent = dedupeAgent(agent);
+
+        return agent;       
+    }
+
     /**
      * If this agent duplicates an agent of another activity of the same
      * type for the same bib resource, use that agent rather than creating a 
      * new one. Current deduping is based only on the agent name strings, 
      * since that is what is available in, e.g., MARC 260$b.
      */
-    private void dedupeAgent() {
+    private Entity dedupeAgent(Entity agent) {
 
         if (grandparent == null) {
-            return;
+            return agent;
         }
 
+        String name = agent.getValue(Ld4lDatatypeProp.NAME);
+
         List<Entity> activities = grandparent.getChildren(
                 Ld4lObjectProp.HAS_ACTIVITY, parent.getType());
         for (Entity activity : activities) {
@@ -101,77 +155,12 @@ private void dedupeAgent() {
                 String agentName = 
                         existingAgent.getValue(Ld4lDatatypeProp.NAME);
                 if (name.equals(agentName)) {
-                    agent = existingAgent;
+                    return existingAgent;
                 }                
             }
-        }       
-    }
-
-    private void buildAgent() { 
-
-        // Use type specified in build params, if any.
-        if (type == null) {
-            // Otherwise determine type from input data.
-            type = getType();
-        }
-
-        this.agent = new Entity(type);
-
-        addAgentName();
+        }  
 
-        dedupeAgent();
-    }
-
-    private void addAgentName() {
-
-        if (name == null) {
-            if (subfield != null) {
-                this.name = subfield.getTrimmedTextValue();
-            } else {
-                this.name = buildNameFromDataField(agent);
-            }           
-        } 
-
-        if (name != null) {
-            agent.addAttribute(Ld4lDatatypeProp.NAME, name);
-        }       
-    }
-
-    /**
-     * Determines type from input data. Defaults to Ld4lAgentType default
-     * type. Never returns null.
-     */
-    private Type getType() {
-
-        if (field != null && field.getTag().equals("100")) {
-            if (field.getFirstIndicator() == 3) {
-                return Ld4lAgentType.FAMILY;
-            } else {
-                return Ld4lAgentType.PERSON;
-            }
-        } 
-
-        return Ld4lAgentType.defaultType();
-    }
-
-    private String buildNameFromDataField(Entity agent) {
-
-        /*
-         * Note that in field 100, ind1 value 0 means given name first, 
-         * value 1 means family name first, but since for now we are not  
-         * parsing the individual name parts we just set the generic name 
-         * property.
-         * 
-         * TODO: if first indicator == 1, split on comma, assign first part
-         * to family name and last part to given name.
-         * 
-         * TODO Add other agent attributes from other subfields.
-         */
-        if (field != null && field.getTag().equals("100")) {
-            return field.getSubfield('a').getTrimmedTextValue();
-        }
-
-        return null;
-    }
+        return agent;
+    }   
 
 }

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/ld4l/AuthorActivityBuilder.java

@@ -25,15 +25,15 @@ public void build() throws EntityBuilderException {
 
         switch (field.getTag()) {
         case "100": 
-            convert_100();
+            convert100();
             break;
         default:
             break;
         }
     }
 
 
-    private void convert_100() throws EntityBuilderException {
+    private void convert100() throws EntityBuilderException {
 
         this.activity = new Entity(TYPE);
         EntityBuilder builder = getBuilder(Ld4lAgentType.defaultType());

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/ld4l/IdentifierBuilder.java

@@ -162,7 +162,7 @@ private Entity build(MarcxmlDataField field, BuildParams params)
                 (MarcxmlSubfield) params.getSubfields().get(0);
 
         if (field.getTag().equals("035")) {
-            identifier = convert_035(subfield);
+            identifier = convert035(subfield);
         }
 
         return identifier;
@@ -172,7 +172,7 @@ private Entity build(MarcxmlDataField field, BuildParams params)
      * Builds an identifier from field 035. Returns null if the identifier 
      * value is already attached to the resource's AdminMetadata object.
      */
-    private Entity convert_035(MarcxmlSubfield subfield) 
+    private Entity convert035(MarcxmlSubfield subfield) 
             throws EntityBuilderException {
 
         Entity identifier;

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/ld4l/InstanceBuilder.java

@@ -82,10 +82,10 @@ private void parseBuildParams(BuildParams params)
     }
 
     private void buildIdentifiers() throws EntityBuilderException {
-        convert_035();
+        convert035();
     }
 
-    private void convert_035() throws EntityBuilderException {
+    private void convert035() throws EntityBuilderException {
 
         // 035 is a repeating field
         List<MarcxmlDataField> fields = record.getDataFields("035");
@@ -119,7 +119,6 @@ private void buildTitles() throws EntityBuilderException {
 
     private void buildPhysicalDescriptions() throws EntityBuilderException {
 
-        // TODO Not sure yet if there are others. 
         buildExtent();
 
     }

src/main/java/org/ld4l/bib2lod/entitybuilders/marcxml/ld4l/LegacySourceDataEntityBuilder.java

@@ -1,15 +1,15 @@
 package org.ld4l.bib2lod.entitybuilders.marcxml.ld4l;
 
-import org.ld4l.bib2lod.datatypes.Ld4lCustomDatatypes.BibDatatype;
 import org.ld4l.bib2lod.entity.Entity;
 import org.ld4l.bib2lod.entitybuilders.BuildParams;
 import org.ld4l.bib2lod.entitybuilders.marcxml.MarcxmlEntityBuilder;
-import org.ld4l.bib2lod.ontology.Type;
 import org.ld4l.bib2lod.ontology.ld4l.Ld4lDatatypeProp;
 
 /**
- * Builds an entity used to store unparsed, unnormalized legacy data. May or
- * may not have a specific type assigned.
+ * Builds an entity used to store unparsed, unnormalized legacy data. Use
+ * when no specific type is assigned; if there is a specific type, use the
+ * builder for that type and add the datatype to the appropriate literal 
+ * value.
  */
 public class LegacySourceDataEntityBuilder extends MarcxmlEntityBuilder {
 
@@ -24,13 +24,7 @@ public Entity build(BuildParams params) throws EntityBuilderException {
 
         Entity entity = new Entity();
 
-        entity.addAttribute(Ld4lDatatypeProp.LABEL, label, 
-                BibDatatype.LEGACY_SOURCE_DATA);
-
-        Type type = params.getType();
-        if (type != null) {
-            entity.addType(type);
-        }
+        entity.addLegacySourceDataAttribute(Ld4lDatatypeProp.LABEL, label);
 
         return entity;
     }