TIKA-4289 -- improve serializability for the metadatafilters (#1868)

apache · Jul 30, 2024 · e4d774b · e4d774b
1 parent 137c0d1
commit e4d774b
Show file tree

Hide file tree

Showing 10 changed files with 60 additions and 6 deletions.
diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/CaptureGroupMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/CaptureGroupMetadataFilter.java
@@ -87,6 +87,18 @@ public void setTargetField(String targetField) {
         this.targetField = targetField;
     }
 
+    public String getRegex() {
+        return regexString;
+    }
+
+    public String getSourceField() {
+        return sourceField;
+    }
+
+    public String getTargetField() {
+        return targetField;
+    }
+
     @Override
     public void initialize(Map<String, Param> params) throws TikaConfigException {
         try {

diff --git a/...re/src/main/java/org/apache/tika/metadata/filter/ClearByAttachmentTypeMetadataFilter.java b/...re/src/main/java/org/apache/tika/metadata/filter/ClearByAttachmentTypeMetadataFilter.java
@@ -16,6 +16,7 @@
  */
 package org.apache.tika.metadata.filter;
 
+import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
@@ -82,4 +83,8 @@ public void setTypes(List<String> types) throws TikaConfigException {
         }
         this.types.addAll(types);
     }
+
+    public List<String> getTypes() {
+        return new ArrayList<>(types);
+    }
 }
diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/ClearByMimeMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/ClearByMimeMetadataFilter.java
@@ -16,6 +16,7 @@
  */
 package org.apache.tika.metadata.filter;
 
+import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
@@ -65,4 +66,8 @@ public void filter(Metadata metadata) throws TikaException {
     public void setMimes(List<String> mimes) {
         this.mimes.addAll(mimes);
     }
+
+    public List<String> getMimes() {
+        return new ArrayList<>(mimes);
+    }
 }
diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/CompositeMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/CompositeMetadataFilter.java
@@ -16,19 +16,33 @@
  */
 package org.apache.tika.metadata.filter;
 
+import java.util.ArrayList;
 import java.util.List;
 
 import org.apache.tika.exception.TikaException;
 import org.apache.tika.metadata.Metadata;
 
 public class CompositeMetadataFilter extends MetadataFilter {
 
-    private final List<MetadataFilter> filters;
+    //no longer final to allow for no arg initialization during serialization
+    private List<MetadataFilter> filters;
 
+    public CompositeMetadataFilter() {
+        filters = new ArrayList<>();
+    }
     public CompositeMetadataFilter(List<MetadataFilter> filters) {
         this.filters = filters;
     }
 
+    public void setFilters(List<MetadataFilter> filters) {
+        this.filters.clear();
+        this.filters.addAll(filters);
+    }
+
+    public List<MetadataFilter> getFilters() {
+        return filters;
+    }
+
     @Override
     public void filter(Metadata metadata) throws TikaException {
         for (MetadataFilter filter : filters) {

diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/DateNormalizingMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/DateNormalizingMetadataFilter.java
@@ -87,4 +87,8 @@ public void filter(Metadata metadata) throws TikaException {
     public void setDefaultTimeZone(String timeZoneId) {
         this.defaultTimeZone = TimeZone.getTimeZone(ZoneId.of(timeZoneId));
     }
+
+    public String getDefaultTimeZone() {
+        return this.defaultTimeZone.toZoneId().toString();
+    }
 }
diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/ExcludeFieldMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/ExcludeFieldMetadataFilter.java
@@ -16,6 +16,7 @@
  */
 package org.apache.tika.metadata.filter;
 
+import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
@@ -49,4 +50,8 @@ public void filter(Metadata metadata) throws TikaException {
     public void setExclude(List<String> exclude) {
         this.excludeSet.addAll(exclude);
     }
+
+    public List<String> getExclude() {
+        return new ArrayList<>(excludeSet);
+    }
 }
diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/FieldNameMappingFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/FieldNameMappingFilter.java
@@ -76,6 +76,10 @@ public void setMappings(Map<String, String> mappings) {
         }
     }
 
+    public Map<String, String> getMappins() {
+        return mappings;
+    }
+
     @Override
     public String toString() {
         return "FieldNameMappingFilter{" + "mappings=" + mappings + ", excludeUnmapped=" + excludeUnmapped + '}';

diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/GeoPointMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/GeoPointMetadataFilter.java
@@ -44,6 +44,10 @@ public void setGeoPointFieldName(String geoPointFieldName) {
         this.geoPointFieldName = geoPointFieldName;
     }
 
+    public String getGeoPointFieldName() {
+        return geoPointFieldName;
+    }
+
     @Override
     public void filter(Metadata metadata) throws TikaException {
         String lat = metadata.get(TikaCoreProperties.LATITUDE);

diff --git a/tika-core/src/main/java/org/apache/tika/metadata/filter/IncludeFieldMetadataFilter.java b/tika-core/src/main/java/org/apache/tika/metadata/filter/IncludeFieldMetadataFilter.java
@@ -16,6 +16,7 @@
  */
 package org.apache.tika.metadata.filter;
 
+import java.util.ArrayList;
 import java.util.HashSet;
 import java.util.List;
 import java.util.Set;
@@ -43,6 +44,10 @@ public void setInclude(List<String> include) {
         includeSet.addAll(include);
     }
 
+    public List<String> getInclude() {
+        return new ArrayList<>(includeSet);
+    }
+
     @Override
     public void filter(Metadata metadata) throws TikaException {
 

diff --git a/tika-serialization/src/main/java/org/apache/tika/serialization/TikaJsonSerializer.java b/tika-serialization/src/main/java/org/apache/tika/serialization/TikaJsonSerializer.java
@@ -106,7 +106,7 @@ public static void serializeObject(String fieldName, Object obj, Class superClas
                     .getClass()
                     .getConstructor();
         } catch (NoSuchMethodException e) {
-            throw new IllegalArgumentException("class (" + obj.getClass() + ") doesn't have a no-arg constructor. Respectfully not seralizing.");
+            throw new IllegalArgumentException("class (" + obj.getClass() + ") doesn't have a no-arg constructor. Respectfully not serializing.");
         }
         try {
             if (fieldName != null) {
@@ -221,10 +221,6 @@ private static void serializeCollection(String fieldName, Object obj, JsonGenera
         }
     }
 
-    private static void serializeItem(String field, Object obj, Class<?> generalType, JsonGenerator jsonGenerator) {
-
-    }
-
     private static void serializePrimitiveAndBoxed(String paramName, Object obj, JsonGenerator jsonGenerator) throws IOException {
         Class clazz = obj.getClass();
         if (paramName != null) {