TIKA-4808 -- fix extension calculations for embedded files, further refinements

Author: tballison

tika-core/src/main/java/org/apache/tika/extractor/EmbeddedDocumentUtil.java

@@ -148,23 +148,27 @@ public String getExtension(TikaInputStream is, Metadata metadata) {
         //use the buffered mimetypes as default
         MimeTypes localMimeTypes = getMimeTypes();
 
-        MimeType mimeType = null;
-        if (mimeString != null) {
-            mimeType = getRegisteredMimeType(localMimeTypes, mimeString);
+        //a parseable declared type wins, even if we have no glob for it. Don't
+        //detect just because the registry lookup came back empty -- that would
+        //overwrite a type the calling parser set deliberately.
+        if (mimeString != null && MediaType.parse(mimeString) != null) {
+            return extensionOf(getRegisteredMimeType(localMimeTypes, mimeString));
         }
-        if (mimeType == null) {
-            try {
-                MediaType mediaType = getDetector().detect(is, metadata, context);
-                is.reset();
-                //set or correct the mime type. Record what was detected, not the
-                //registry match, which may have fallen back to the base type.
-                metadata.set(Metadata.CONTENT_TYPE, mediaType.toString());
-                mimeType = getRegisteredMimeType(localMimeTypes, mediaType.toString());
-            } catch (IOException e) {
-                //swallow
-            }
+        try {
+            MediaType mediaType = getDetector().detect(is, metadata, context);
+            is.reset();
+            //set or correct the mime type. Record what was detected, not the
+            //registry match, which may have fallen back to the base type.
+            metadata.set(Metadata.CONTENT_TYPE, mediaType.toString());
+            return extensionOf(getRegisteredMimeType(localMimeTypes, mediaType.toString()));
+        } catch (IOException e) {
+            //swallow
         }
-        return mimeType == null ? ".bin" : mimeType.getExtension();
+        return ".bin";
+    }
+
+    private static String extensionOf(MimeType mimeType) {
+        return mimeType == null ? "" : mimeType.getExtension();
     }
 
     /**

tika-core/src/test/java/org/apache/tika/extractor/EmbeddedDocumentUtilExtensionTest.java

@@ -16,12 +16,16 @@
  */
 package org.apache.tika.extractor;
 
+import static java.nio.charset.StandardCharsets.UTF_8;
 import static org.junit.jupiter.api.Assertions.assertEquals;
 
 import org.junit.jupiter.api.Test;
 
+import org.apache.tika.io.TikaInputStream;
+import org.apache.tika.metadata.Metadata;
 import org.apache.tika.mime.MediaType;
 import org.apache.tika.mime.MimeTypes;
+import org.apache.tika.parser.ParseContext;
 
 /**
  * TIKA-4808 -- a media type carrying parameters must still resolve to the
@@ -107,6 +111,32 @@ public void testGeneratedResourceNameKeepsExtension() {
                 EmbeddedDocumentUtil.EmbeddedResourcePrefix.IMAGE, 0, "image/png"));
     }
 
+    /**
+     * A declared type that simply has no glob must not trigger detection -- that would
+     * overwrite a CONTENT_TYPE the calling parser set deliberately.
+     */
+    @Test
+    public void testDeclaredUnregisteredTypeIsNotOverwritten() throws Exception {
+        Metadata metadata = new Metadata();
+        metadata.set(Metadata.CONTENT_TYPE, "application/tika-bogus-xyz");
+        EmbeddedDocumentUtil util = new EmbeddedDocumentUtil(new ParseContext());
+        try (TikaInputStream tis = TikaInputStream.get("%PDF-1.4\n".getBytes(UTF_8))) {
+            assertEquals("", util.getExtension(tis, metadata));
+            assertEquals("application/tika-bogus-xyz", metadata.get(Metadata.CONTENT_TYPE));
+        }
+    }
+
+    @Test
+    public void testDeclaredParameterizedTypeResolvesAndIsPreserved() throws Exception {
+        Metadata metadata = new Metadata();
+        metadata.set(Metadata.CONTENT_TYPE, "text/plain; charset=UTF-8");
+        EmbeddedDocumentUtil util = new EmbeddedDocumentUtil(new ParseContext());
+        try (TikaInputStream tis = TikaInputStream.get("hello".getBytes(UTF_8))) {
+            assertEquals(".txt", util.getExtension(tis, metadata));
+            assertEquals("text/plain; charset=UTF-8", metadata.get(Metadata.CONTENT_TYPE));
+        }
+    }
+
     /**
      * Guard the assumption the fix rests on: normalize() deliberately preserves
      * parameters, which is why forName() misses the registry for parameterized names.

tika-parsers/tika-parsers-standard/tika-parsers-standard-modules/tika-parser-mail-module/src/test/java/org/apache/tika/parser/mail/RFC822ParserTest.java

@@ -404,7 +404,7 @@ public void testExtractAttachments() throws Exception {
         assertEquals(null, metadataList.get(1).get(Metadata.CONTENT_DISPOSITION));
         assertEquals("attachment; filename=\"testPNG.png\"",
                 metadataList.get(2).get(Metadata.CONTENT_DISPOSITION));
-        assertEquals("/Test Attachment Email.eml/embedded-1",
+        assertEquals("/Test Attachment Email.eml/embedded-1.txt",
                 metadataList.get(1).get(TikaCoreProperties.FINAL_EMBEDDED_RESOURCE_PATH));
         assertEquals("/Test Attachment Email.eml/testPNG.png",
                 metadataList.get(2).get(TikaCoreProperties.FINAL_EMBEDDED_RESOURCE_PATH));