TIKA-4585 -- simplify serialization (#2471)

* TIKA-4585 -- simplify ParseContext serialization -WIP

* TIKA-4585 -- further progress - WIP

* TIKA-4585 -- further progress - WIP

* TIKA-4585 -- further progress - WIP

* TIKA-4585 -- further progress

* TIKA-4585 -- fix unit test, add comment

* TIKA-4585 -- rm unused mixins

* TIKA-4585 -- bring back in some design features from main

Author: tballison

tika-app/src/test/java/org/apache/tika/cli/TikaCLITest.java

@@ -38,8 +38,6 @@
 import java.util.List;
 import java.util.Set;
 
-import org.jetbrains.annotations.NotNull;
-import org.jetbrains.annotations.Nullable;
 import org.junit.jupiter.api.AfterEach;
 import org.junit.jupiter.api.BeforeEach;
 import org.junit.jupiter.api.Disabled;
@@ -237,7 +235,7 @@ public void testJsonMetadataOutput() throws Exception {
     public void testJsonMetadataPrettyPrintOutput() throws Exception {
         String json = getParamOutContent("--json", "-r", resourcePrefix + "testJsonMultipleInts.html");
 
-        assertTrue(json.contains("org.apache.tika.parser.CompositeParser\", \"org.apache.tika.parser.html.JSoupParser"));
+        assertTrue(json.contains("org.apache.tika.parser.DefaultParser\", \"org.apache.tika.parser.html.JSoupParser"));
         //test pretty-print alphabetic sort of keys
         int enc = json.indexOf("\"Content-Encoding\"");
         int fb = json.indexOf("fb:admins");
@@ -413,22 +411,23 @@ private Set<String> getFileNames(Path extractDir) throws IOException {
         final Set<String> names = new HashSet<>();
         Files.walkFileTree(extractDir, new FileVisitor<Path>() {
             @Override
-            public @NotNull FileVisitResult preVisitDirectory(Path path, @NotNull BasicFileAttributes basicFileAttributes) throws IOException {
+            public FileVisitResult preVisitDirectory(Path path, BasicFileAttributes basicFileAttributes) throws IOException {
                 return FileVisitResult.CONTINUE;
             }
 
-            public @NotNull FileVisitResult visitFile(Path path, @NotNull BasicFileAttributes basicFileAttributes) throws IOException {
+            @Override
+            public FileVisitResult visitFile(Path path, BasicFileAttributes basicFileAttributes) throws IOException {
                 names.add(extractDir.relativize(path).toString().replace('\\', '/'));
                 return FileVisitResult.CONTINUE;
             }
 
             @Override
-            public @NotNull FileVisitResult visitFileFailed(Path path, @NotNull IOException e) throws IOException {
+            public FileVisitResult visitFileFailed(Path path, IOException e) throws IOException {
                 return FileVisitResult.CONTINUE;
             }
 
             @Override
-            public @NotNull FileVisitResult postVisitDirectory(Path path, @Nullable IOException e) throws IOException {
+            public FileVisitResult postVisitDirectory(Path path, IOException e) throws IOException {
                 return FileVisitResult.CONTINUE;
             }
         });

tika-core/src/main/java/org/apache/tika/config/ConfigContainer.java

@@ -24,13 +24,13 @@
 import org.apache.tika.parser.ParseContext;
 
 /**
- * Facade for accessing runtime configuration from ParseContext's ConfigContainer.
+ * Facade for accessing runtime configuration from ParseContext's jsonConfigs.
  * <p>
  * This wrapper provides a safe way for parsers to access runtime configuration
  * without directly depending on tika-serialization. It performs these critical checks:
  * <ul>
- *   <li>If ConfigContainer has config for the requested key but ConfigDeserializer
- *       is not on the classpath, throws IllegalStateException with a clear error message</li>
+ *   <li>If ParseContext has JSON config for the requested key but ConfigDeserializer
+ *       is not on the classpath, throws TikaConfigException with a clear error message</li>
  *   <li>If ConfigDeserializer is available, delegates to it for deserialization</li>
  *   <li>If no config is present, returns the default config</li>
  * </ul>
@@ -55,7 +55,7 @@ public class ParseContextConfig {
         Method hasMethod = null;
         try {
             clazz = Class.forName("org.apache.tika.serialization.ConfigDeserializer");
-            getMethod = clazz.getMethod("getConfig", 
+            getMethod = clazz.getMethod("getConfig",
                 ParseContext.class, String.class, Class.class, Object.class);
             hasMethod = clazz.getMethod("hasConfig", ParseContext.class, String.class);
         } catch (ClassNotFoundException | NoSuchMethodException e) {
@@ -74,10 +74,11 @@ public class ParseContextConfig {
      * re-deserializing. This is efficient for embedded documents where the config
      * was already deserialized for the parent document.
      * <p>
-     * If not found, it checks ConfigContainer for the config key and deserializes
-     * the JSON. The deserialized config is also set in ParseContext for future lookups.
+     * If not found, it checks jsonConfigs for the config key and deserializes
+     * the JSON. The deserialized config is cached in resolvedConfigs and also
+     * set in the main ParseContext for future lookups.
      * <p>
-     * This method performs defensive checking: if the ConfigContainer has configuration
+     * This method performs defensive checking: if the ParseContext has JSON configuration
      * for the requested key but the ConfigDeserializer is not available on the classpath,
      * it throws TikaConfigException to prevent silent failures.
      *
@@ -87,7 +88,7 @@ public class ParseContextConfig {
      * @param defaultConfig the default configuration to use if no runtime config exists
      * @param <T> the configuration type
      * @return the runtime config merged with defaults, or the default config if no runtime config
-     * @throws TikaConfigException if ConfigContainer has config but ConfigDeserializer is not on classpath
+     * @throws TikaConfigException if ParseContext has JSON config but ConfigDeserializer is not on classpath
      * @throws IOException if deserialization fails
      */
     public static <T> T getConfig(ParseContext context, String configKey,
@@ -104,25 +105,18 @@ public static <T> T getConfig(ParseContext context, String configKey,
             return existingConfig;
         }
 
-        ConfigContainer configContainer = context.get(ConfigContainer.class);
-        if (configContainer == null) {
+        // Check for JSON config
+        if (!context.hasJsonConfig(configKey)) {
             return defaultConfig;
         }
 
-        // Check if there's config for this specific key
-        boolean hasConfigForKey = configContainer.get(configKey).isPresent();
-        if (!hasConfigForKey) {
-            return defaultConfig;
-        }
-
-        // Config exists for this key - ConfigDeserializer MUST be available
+        // JSON config exists for this key - ConfigDeserializer MUST be available
         if (CONFIG_DESERIALIZER_CLASS == null) {
             throw new TikaConfigException(String.format(Locale.ROOT,
-                "ParseContext contains ConfigContainer with configuration for '%s' " +
+                "ParseContext contains JSON configuration for '%s' " +
                 "but org.apache.tika.serialization.ConfigDeserializer is not on the classpath. " +
                 "This means your runtime configuration will be ignored. " +
-                "To fix: add tika-serialization as a dependency, or remove the ConfigContainer " +
-                "from ParseContext if runtime configuration via ConfigContainer is not needed.",
+                "To fix: add tika-serialization as a dependency.",
                 configKey));
         }
 
@@ -150,19 +144,13 @@ public static <T> T getConfig(ParseContext context, String configKey,
      *
      * @param context the parse context
      * @param configKey the configuration key
-     * @return true if config exists for this key
+     * @return true if JSON config exists for this key
      */
     public static boolean hasConfig(ParseContext context, String configKey) {
         if (context == null) {
             return false;
         }
-
-        ConfigContainer configContainer = context.get(ConfigContainer.class);
-        if (configContainer == null) {
-            return false;
-        }
-
-        return configContainer.get(configKey).isPresent();
+        return context.hasJsonConfig(configKey);
     }
 
     /**

tika-core/src/main/java/org/apache/tika/config/ParseContextConfig.java

@@ -18,12 +18,12 @@
 
 /**
  * Marker interface indicating that a component reads its own configuration
- * from {@link ConfigContainer} inside the {@link org.apache.tika.parser.ParseContext} at runtime.
+ * from {@link org.apache.tika.parser.ParseContext}'s jsonConfigs at runtime.
  * <p>
  * Components implementing this interface will NOT be automatically resolved
  * by ParseContextUtils. Instead, the JSON configuration will remain in
- * ConfigContainer, and the component is responsible for reading and applying
- * its own configuration during execution.
+ * ParseContext's jsonConfigs, and the component is responsible for reading
+ * and applying its own configuration during execution.
  * <p>
  * This is typically used by parsers and other components that need fine-grained
  * control over how their configuration is loaded and merged with defaults.
@@ -36,7 +36,7 @@
  *     private final PDFParserConfig defaultConfig;
  *
  *     public void parse(..., ParseContext context) {
- *         // Component reads its own config from ConfigContainer
+ *         // Component reads its own config from ParseContext
  *         PDFParserConfig config = ParseContextConfig.getConfig(
  *             context, "pdf-parser", PDFParserConfig.class, defaultConfig);
  *         // use config...
@@ -48,7 +48,6 @@
  * automatically deserialized and added to ParseContext by ParseContextUtils.
  *
  * @since Apache Tika 4.0
- * @see ConfigContainer
  * @see ParseContextConfig
  */
 public interface SelfConfiguring {

tika-core/src/main/java/org/apache/tika/config/SelfConfiguring.java

@@ -22,6 +22,9 @@
 
 /**
  * Configuration class for specifying parse task timeout.
+ * <p>
+ * This is a config POJO (not a component like Parser/Detector), so it uses
+ * standard Jackson format rather than compact component format:
  * <pre>
  * {
  *   "parse-context": {

tika-core/src/main/java/org/apache/tika/config/TikaTaskTimeout.java

@@ -16,12 +16,14 @@
  */
 package org.apache.tika.detect;
 
+import java.util.ArrayList;
 import java.util.Collection;
 import java.util.Collections;
 import java.util.List;
 import javax.imageio.spi.ServiceRegistry;
 
 import org.apache.tika.config.ServiceLoader;
+import org.apache.tika.config.TikaComponent;
 import org.apache.tika.mime.MimeTypes;
 import org.apache.tika.utils.ServiceLoaderUtils;
 
@@ -37,18 +39,23 @@
  *
  * @since Apache Tika 0.9
  */
+@TikaComponent(spi = false)
 public class DefaultDetector extends CompositeDetector {
 
     /**
      * Serial version UID
      */
     private static final long serialVersionUID = -8170114575326908027L;
     private transient final ServiceLoader loader;
+    private final Collection<Class<? extends Detector>> excludedClasses;
 
     public DefaultDetector(MimeTypes types, ServiceLoader loader,
                            Collection<Class<? extends Detector>> excludeDetectors) {
         super(types.getMediaTypeRegistry(), getDefaultDetectors(types, loader, excludeDetectors));
         this.loader = loader;
+        this.excludedClasses = excludeDetectors != null ?
+                Collections.unmodifiableCollection(new ArrayList<>(excludeDetectors)) :
+                Collections.emptySet();
     }
 
     public DefaultDetector(MimeTypes types, ServiceLoader loader) {
@@ -124,4 +131,13 @@ public List<Detector> getDetectors() {
         }
     }
 
+    /**
+     * Returns the classes that were explicitly excluded when constructing this detector.
+     * Used for round-trip serialization to preserve exclusion configuration.
+     *
+     * @return unmodifiable collection of excluded detector classes, never null
+     */
+    public Collection<Class<? extends Detector>> getExcludedClasses() {
+        return excludedClasses;
+    }
 }

tika-core/src/main/java/org/apache/tika/detect/DefaultDetector.java

@@ -33,8 +33,12 @@
  * when parsing data that is extremely compressible and resembles a ZIP
  * bomb. Null values will be ignored and will not affect the default values
  * in SecureContentHandler.
+ * <p>
+ * This is a config POJO. It uses standard Jackson deserialization for its
+ * primitive fields, but component fields (like embeddedDocumentExtractorFactory)
+ * use compact format.
  */
-@TikaComponent
+@TikaComponent(spi = false)
 public class AutoDetectParserConfig implements Serializable {
 
     private static ContentHandlerDecoratorFactory NOOP_CONTENT_HANDLER_DECORATOR_FACTORY =