Filter Out All Media Files from Microsoft Word Documents to Improve Indexing in Confluence

Problem Definition

Currently Confluence only strips out a limited number of media files embedded in a Microsoft Word document before indexing it:

• .png
• .emf
• .wmf
• .jpg
• .jpeg
• .gif

If a document has an embedded file not listed here, it may not get indexed if it is too large.

Background

Currently Confluence does not index files if the content with the removed media listed above is greater than 16Mb. There's a system property that can be set to make this larger, but this isn't used. See CONF-40176 for more details.

Suggested Solution

In com.atlassian.confluence.extra.officeconnector.index.word#WordXMLTextExtractor it has

                    if (!(name.contains("/media/") ||
                            processedName.endsWith(".png") || processedName.endsWith(".emf") || processedName.endsWith(".wmf") ||
                            processedName.endsWith(".jpg") || processedName.endsWith(".jpeg") ||
                            processedName.endsWith(".gif")
                       ))

Either

1. Strip out all content in the /media folder
   or
2. Add all media types that are possible to add to a Word document. See Types of media files you can add.

Notes

Similar issues occur with other Microsoft Office documents (e.g. PowerPoint).