issue #174: PICA FRBR functions

pkiraly · Oct 6, 2022 · 9c83f4a · 9c83f4a
1 parent 6f8712d
commit 9c83f4a
Show file tree

Hide file tree

Showing 3 changed files with 31 additions and 12 deletions.
diff --git a/src/main/java/de/gwdg/metadataqa/marc/cli/FunctionalAnalysis.java b/src/main/java/de/gwdg/metadataqa/marc/cli/FunctionalAnalysis.java
@@ -125,8 +125,11 @@ private void countDataFields(Map<FRBRFunction, FunctionValue> recordCounter,
             if (subfield.getDefinition() != null && subfield.getDefinition().getFrbrFunctions() != null)
               FrbrFunctionLister.countFunctions(subfield.getDefinition().getFrbrFunctions(), recordCounter);
         } else if (schemaType.equals(SchemaType.PICA)) {
-          if (frbrFunctionLister.getFunctionByPicaPath().containsKey(dataField.getTag()))
-            FrbrFunctionLister.countFunctions(frbrFunctionLister.getFunctionByPicaPath().get(dataField.getTag()), recordCounter);
+          for (MarcSubfield subfield : dataField.getSubfields()) {
+            String key = dataField.getTag() + "$" + subfield.getCode();
+            if (frbrFunctionLister.getFunctionByPicaPath().containsKey(key))
+              FrbrFunctionLister.countFunctions(frbrFunctionLister.getFunctionByPicaPath().get(key), recordCounter);
+          }
         }
       }
     }
@@ -205,7 +208,7 @@ private void saveMapping(String fileExtension,
     if (parameters.getSchemaType().equals(SchemaType.MARC21))
       functions = frbrFunctionLister.getMarcPathByFunction();
     else if (parameters.getSchemaType().equals(SchemaType.PICA))
-      functions = frbrFunctionLister.getPicaPathByFunction();
+      functions = frbrFunctionLister.getPicaPathByFunctionConcensed();
 
     var path = Paths.get(parameters.getOutputDir(), "functional-analysis-mapping" + fileExtension);
     try (var writer = Files.newBufferedWriter(path)) {

diff --git a/src/main/java/de/gwdg/metadataqa/marc/utils/FrbrFunctionLister.java b/src/main/java/de/gwdg/metadataqa/marc/utils/FrbrFunctionLister.java
@@ -38,7 +38,8 @@ public class FrbrFunctionLister {
   private Map<String, List<FRBRFunction>> functionByMarcPath;
   private Map<String, List<FRBRFunction>> functionByPicaPath;
   private AppendableHashMap<FRBRFunction, String> marcPathByFunction;
-  private Map<FRBRFunction, List<String>> picaPathByFunction;
+  private Map<FRBRFunction, List<String>> picaPathByFunctionCondensed;
+  private Map<FRBRFunction, Map<String, List<String>>> picaPathByFunction;
 
   public FrbrFunctionLister(SchemaType schemaType, MarcVersion marcVersion) {
     this.schemaType = schemaType == null ? SchemaType.MARC21 : schemaType;
@@ -194,7 +195,7 @@ public Map<FRBRFunction, List<String>> getMarcPathByFunction() {
     return marcPathByFunction.getMap();
   }
 
-  public Map<FRBRFunction, List<String>> getPicaPathByFunction() {
+  public Map<FRBRFunction, Map<String, List<String>>> getPicaPathByFunction() {
     if (picaPathByFunction == null) {
       initializePica();
     }
@@ -208,8 +209,16 @@ public Map<String, List<FRBRFunction>> getFunctionByPicaPath() {
     return functionByPicaPath;
   }
 
+  public Map<FRBRFunction, List<String>> getPicaPathByFunctionConcensed() {
+    if (picaPathByFunctionCondensed == null) {
+      initializePica();
+    }
+    return picaPathByFunctionCondensed;
+  }
+
   private void initializePica() {
     picaPathByFunction = new HashMap<>();
+    picaPathByFunctionCondensed = new HashMap<>();
     functionByPicaPath = new HashMap<>();
     for (Map.Entry<FRBRFunction, List<String>> entry : marcPathByFunction.entrySet()) {
       for (String address : entry.getValue()) {
@@ -219,15 +228,22 @@ private void initializePica() {
           for (Crosswalk crosswalk : PicaMarcCrosswalkReader.lookupMarc21(key)) {
             String pica = crosswalk.getPica();
             if (!picaPathByFunction.containsKey(function))
-              picaPathByFunction.put(function, new ArrayList<>());
-            picaPathByFunction.get(function).add(pica);
-
-            if (!functionByPicaPath.containsKey(pica))
-              functionByPicaPath.put(pica, new ArrayList<>());
-            functionByPicaPath.get(pica).add(function);
+              picaPathByFunction.put(function, new HashMap<>());
+            if (!picaPathByFunction.get(function).containsKey(pica))
+              picaPathByFunction.get(function).put(pica, new ArrayList<>());
+            picaPathByFunction.get(function).get(pica).add(crosswalk.getPicaUf());
+
+            if (!picaPathByFunctionCondensed.containsKey(function))
+              picaPathByFunctionCondensed.put(function, new ArrayList<>());
+            picaPathByFunctionCondensed.get(function).add(pica + crosswalk.getPicaUf());
+
+            if (!functionByPicaPath.containsKey(pica + crosswalk.getPicaUf()))
+              functionByPicaPath.put(pica + crosswalk.getPicaUf(), new ArrayList<>());
+            functionByPicaPath.get(pica + crosswalk.getPicaUf()).add(function);
           }
         }
       }
     }
   }
+
 }
diff --git a/src/test/java/de/gwdg/metadataqa/marc/utils/pica/crosswalk/PicaMarcCrosswalkReaderTest.java b/src/test/java/de/gwdg/metadataqa/marc/utils/pica/crosswalk/PicaMarcCrosswalkReaderTest.java
@@ -21,7 +21,7 @@ public void test() {
   @Test
   public void testLoad() {
     FrbrFunctionLister lister = new FrbrFunctionLister(MarcVersion.MARC21);
-    Map<FRBRFunction, List<String>> picaFunctions = lister.getPicaPathByFunction();
+    Map<FRBRFunction, List<String>> picaFunctions = lister.getPicaPathByFunctionConcensed();
     assertEquals(11, picaFunctions.size());
     assertEquals(167, picaFunctions.get(FRBRFunction.DiscoveryObtain).size());
     assertEquals(178, picaFunctions.get(FRBRFunction.DiscoverySearch).size());