Better handling of multiple author relators in MARC (#772)

import/index_java/src/org/solrmarc/index/VuFindIndexer.java

@@ -1806,31 +1806,60 @@ public class VuFindIndexer extends SolrIndexer
      */
     protected Boolean authorHasAppropriateRelator(DataField authorField,
         String[] noRelatorAllowed, String relatorConfig
+    ) {
+        return getValidRelators(authorField, noRelatorAllowed, relatorConfig).size() > 0;
+    }
+
+    /**
+     * Extract all valid relator terms from a list of subfields using a whitelist.
+     * @param subfields      List of subfields to check
+     * @param permittedRoles Whitelist to check against
+     * @return Set of valid relator terms
+     */
+    public Set<String> getValidRelatorsFromSubfields(List<Subfield> subfields, List<String> permittedRoles)
+    {
+        Set<String> relators = new LinkedHashSet<String>();
+        for (int j = 0; j < subfields.size(); j++) {
+            String current = normalizeRelatorString(subfields.get(j).getData());
+            if (permittedRoles.contains(current)) {
+                relators.add(current);
+            }
+        }
+        return relators;
+    }
+
+    /**
+     * Extract all values that meet the specified relator requirements.
+     * @param authorField      Field to analyze
+     * @param noRelatorAllowed Array of tag names which are allowed to be used with
+     * no declared relator.
+     * @param relatorConfig    The setting in author-classification.ini which
+     * defines which relator terms are acceptable (or a colon-delimited list)
+     * @return Set
+     */
+    public Set<String> getValidRelators(DataField authorField,
+        String[] noRelatorAllowed, String relatorConfig
     ) {
         // get tag number from Field
         String tag = authorField.getTag();
-        List<Subfield> subfieldE = normalizeRelatorSubfieldList(authorField.getSubfields('e'));
-        List<Subfield> subfield4 = normalizeRelatorSubfieldList(authorField.getSubfields('4'));
+        List<Subfield> subfieldE = authorField.getSubfields('e');
+        List<Subfield> subfield4 = authorField.getSubfields('4');
+
+        Set<String> relators = new LinkedHashSet<String>();
 
         // if no relator is found, check to see if the current tag is in the "no
         // relator allowed" list.
         if (subfieldE.size() == 0 && subfield4.size() == 0) {
-            return Arrays.asList(noRelatorAllowed).contains(tag);
-        }
-
-        // If we got this far, we need to figure out what type of relation they have
-        List permittedRoles = normalizeRelatorStringList(Arrays.asList(loadRelatorConfig(relatorConfig)));
-        for (int j = 0; j < subfield4.size(); j++) {
-            if (permittedRoles.contains(subfield4.get(j).getData())) {
-                return true;
-            }
-        }
-        for (int j = 0; j < subfieldE.size(); j++) {
-            if (permittedRoles.contains(subfieldE.get(j).getData())) {
-                return true;
+            if (Arrays.asList(noRelatorAllowed).contains(tag)) {
+                relators.add("");
             }
+        } else {
+            // If we got this far, we need to figure out what type of relation they have
+            List permittedRoles = normalizeRelatorStringList(Arrays.asList(loadRelatorConfig(relatorConfig)));
+            relators.addAll(getValidRelatorsFromSubfields(subfieldE, permittedRoles));
+            relators.addAll(getValidRelatorsFromSubfields(subfield4, permittedRoles));
         }
-        return false;
+        return relators;
     }
 
     /**
@@ -1886,8 +1915,8 @@ public class VuFindIndexer extends SolrIndexer
             DataField authorField;
             while (fieldsIter.hasNext()){
                 authorField = (DataField) fieldsIter.next();
-                //add all author types to the result set
-                if (authorHasAppropriateRelator(authorField, noRelatorAllowed, relatorConfig)) {
+                // add all author types to the result set; if we have multiple relators, repeat the authors
+                for (String iterator: getValidRelators(authorField, noRelatorAllowed, relatorConfig)) {
                     for (String subfields : parsedTagList.get(authorField.getTag())) {
                         String current = this.getDataFromVariableField(authorField, "["+subfields+"]", " ", false);
                         // TODO: we may eventually be able to use this line instead,
@@ -1981,30 +2010,7 @@ public class VuFindIndexer extends SolrIndexer
             while (fieldsIter.hasNext()){
                 authorField = (DataField) fieldsIter.next();
                 //add all author types to the result set
-                if (authorHasAppropriateRelator(authorField, noRelatorAllowed, relatorConfig)) {
-                    List<Subfield> subfieldE = normalizeRelatorSubfieldList(authorField.getSubfields('e'));
-                    List<Subfield> subfield4 = normalizeRelatorSubfieldList(authorField.getSubfields('4'));
-
-                    // get the first non-empty subfield
-                    String relator = defaultRelator;
-
-                    // try subfield E first
-                    for (int j = 0; j < subfieldE.size(); j++) {
-                        if (!subfieldE.get(j).getData().isEmpty()) {
-                            relator = subfieldE.get(j).getData();
-                            continue;
-                        }
-                    }
-                    // try subfield 4 now and overwrite relator as subfield 4 is most important
-                    for (int j = 0; j < subfield4.size(); j++) {
-                        if (!subfield4.get(j).getData().isEmpty()) {
-                            relator = subfield4.get(j).getData();
-                            continue;
-                        }
-                    }
-
-                    result.add(relator);
-                }
+                result.addAll(getValidRelators(authorField, noRelatorAllowed, relatorConfig));
             }
         }
         return result;
@@ -2079,22 +2085,6 @@ public class VuFindIndexer extends SolrIndexer
         return stringList;
     }
 
-    /**
-     * Normalizes the strings in a list of subfields.
-     *
-     * @param subfieldList List of subfields to be normalized
-     * @return subfieldList Normalized List of subfields
-     */
-    protected List<Subfield> normalizeRelatorSubfieldList(List<Subfield> subfieldList)
-    {
-        for (int j = 0; j < subfieldList.size(); j++) {
-            subfieldList.get(j).setData(
-                normalizeRelatorString(subfieldList.get(j).getData())
-            );
-        }
-        return subfieldList;
-    }
-
     /**
      * Normalizes a string
      *

import/index_scripts/author.bsh

@@ -19,31 +19,60 @@ org.solrmarc.index.SolrIndexer indexer = null;
  */
 public Boolean authorHasAppropriateRelator(DataField authorField,
     String[] noRelatorAllowed, String relatorConfig
+) {
+    return getValidRelators(authorField, noRelatorAllowed, relatorConfig).size() > 0;
+}
+
+/**
+ * Extract all valid relator terms from a list of subfields using a whitelist.
+ * @param subfields      List of subfields to check
+ * @param permittedRoles Whitelist to check against
+ * @return Set of valid relator terms
+ */
+public Set getValidRelatorsFromSubfields(List subfields, List permittedRoles)
+{
+    Set relators = new LinkedHashSet();
+    for (int j = 0; j < subfields.size(); j++) {
+        String current = normalizeRelatorString(subfields.get(j).getData());
+        if (permittedRoles.contains(current)) {
+            relators.add(current);
+        }
+    }
+    return relators;
+}
+
+/**
+ * Extract all values that meet the specified relator requirements.
+ * @param authorField      Field to analyze
+ * @param noRelatorAllowed Array of tag names which are allowed to be used with
+ * no declared relator.
+ * @param relatorConfig    The setting in author-classification.ini which
+ * defines which relator terms are acceptable (or a colon-delimited list)
+ * @return Set
+ */
+public Set getValidRelators(DataField authorField,
+    String[] noRelatorAllowed, String relatorConfig
 ) {
     // get tag number from Field
     String tag = authorField.getTag();
-    List subfieldE = normalizeRelatorSubfieldList(authorField.getSubfields('e'));
-    List subfield4 = normalizeRelatorSubfieldList(authorField.getSubfields('4'));
+    List subfieldE = authorField.getSubfields('e');
+    List subfield4 = authorField.getSubfields('4');
+
+    Set relators = new LinkedHashSet();
 
     // if no relator is found, check to see if the current tag is in the "no
     // relator allowed" list.
     if (subfieldE.size() == 0 && subfield4.size() == 0) {
-        return Arrays.asList(noRelatorAllowed).contains(tag);
-    }
-
-    // If we got this far, we need to figure out what type of relation they have
-    List permittedRoles = normalizeRelatorStringList(Arrays.asList(loadRelatorConfig(relatorConfig)));
-    for (int j = 0; j < subfield4.size(); j++) {
-        if (permittedRoles.contains(subfield4.get(j).getData())) {
-            return true;
-        }
-    }
-    for (int j = 0; j < subfieldE.size(); j++) {
-        if (permittedRoles.contains(subfieldE.get(j).getData())) {
-            return true;
+        if (Arrays.asList(noRelatorAllowed).contains(tag)) {
+            relators.add("");
         }
+    } else {
+        // If we got this far, we need to figure out what type of relation they have
+        List permittedRoles = normalizeRelatorStringList(Arrays.asList(loadRelatorConfig(relatorConfig)));
+        relators.addAll(getValidRelatorsFromSubfields(subfieldE, permittedRoles));
+        relators.addAll(getValidRelatorsFromSubfields(subfield4, permittedRoles));
     }
-    return false;
+    return relators;
 }
 
 /**
@@ -99,8 +128,8 @@ public List getAuthorsFilteredByRelator(Record record, String tagList,
         DataField authorField;
         while (fieldsIter.hasNext()){
             authorField = (DataField) fieldsIter.next();
-            //add all author types to the result set
-            if (authorHasAppropriateRelator(authorField, noRelatorAllowed, relatorConfig)) {
+            // add all author types to the result set; if we have multiple relators, repeat the authors
+            for (String iterator: getValidRelators(authorField, noRelatorAllowed, relatorConfig)) {
                 for (String subfields : parsedTagList.get(authorField.getTag())) {
                     String current = indexer.getDataFromVariableField(authorField, "["+subfields+"]", " ", false);
                     // TODO: we may eventually be able to use this line instead,
@@ -194,30 +223,7 @@ public List getRelatorsFilteredByRelator(Record record, String tagList,
         while (fieldsIter.hasNext()){
             authorField = (DataField) fieldsIter.next();
             //add all author types to the result set
-            if (authorHasAppropriateRelator(authorField, noRelatorAllowed, relatorConfig)) {
-                List subfieldE = normalizeRelatorSubfieldList(authorField.getSubfields('e'));
-                List subfield4 = normalizeRelatorSubfieldList(authorField.getSubfields('4'));
-    
-                // get the first non-empty subfield
-                String relator = defaultRelator;
-
-                // try subfield E first
-                for (int j = 0; j < subfieldE.size(); j++) {
-                    if (!subfieldE.get(j).getData().isEmpty()) {
-                        relator = subfieldE.get(j).getData();
-                        continue;
-                    }
-                }
-                // try subfield 4 now and overwrite relator as subfield 4 is most important
-                for (int j = 0; j < subfield4.size(); j++) {
-                    if (!subfield4.get(j).getData().isEmpty()) {
-                        relator = subfield4.get(j).getData();
-                        continue;
-                    }
-                }
-
-                result.add(relator);
-            }
+            result.addAll(getValidRelators(authorField, noRelatorAllowed, relatorConfig));
         }
     }
     return result;
@@ -292,22 +298,6 @@ public List normalizeRelatorStringList(List stringList)
     return stringList;
 }
 
-/**
- * Normalizes the strings in a list of subfields.
- *
- * @param subfieldList List of subfields to be normalized
- * @return subfieldList Normalized List of subfields
- */
-public List normalizeRelatorSubfieldList(List subfieldList)
-{
-    for (int j = 0; j < subfieldList.size(); j++) {
-        subfieldList.get(j).setData(
-            normalizeRelatorString(subfieldList.get(j).getData())
-        );
-    }
-    return subfieldList;
-}
-
 /**
  * Normalizes a string
  *