TIKA-4219 -- clean up...do not include font names in main package

apache · Mar 26, 2024 · 2d585b8 · 2d585b8
1 parent 88b582f
commit 2d585b8
Showing 1 changed file with 5 additions and 1 deletion.
diff --git a/...s/tika-parser-miscoffice-module/src/main/java/org/apache/tika/parser/epub/EpubParser.java b/...s/tika-parser-miscoffice-module/src/main/java/org/apache/tika/parser/epub/EpubParser.java
@@ -452,9 +452,13 @@ private void handleEmbedded(ZipFile zipFile, String relativePath, HRefMediaPair
 
         xhtml.startElement("div", "class", "embedded");
         try {
+            boolean outputHtml = true;
+            if (hRefMediaPair.media.contains("font") || hRefMediaPair.href.startsWith("fonts")) {
+                outputHtml = false;
+            }
             embeddedDocumentExtractor
                     .parseEmbedded(stream, new EmbeddedContentHandler(xhtml), embeddedMetadata,
-                            true);
+                            outputHtml);
 
         } finally {
             IOUtils.closeQuietly(stream);