EPUB Troubleshooting: Making Kindle Downloads Apple-Compatible

A DRM-free EPUB downloaded from Amazon’s Content & Devices page (the “Download & transfer via USB” export) may fail to open in Apple Books on iOS and macOS, even though it opens fine in other readers. This is a standards-validation mismatch, not a corrupted file — and it’s fixable in a few minutes.

Symptoms

• Apple Books refuses the file during import with a generic error (“This book cannot be opened” or similar), on both iOS and macOS
• The same file opens fine in desktop readers (Calibre, KOReader) and passes zip integrity checks
• The file itself is DRM-free and downloads successfully from Amazon

Root Cause

The EPUB 3 specification requires epub:type attribute values to be at least one XML NMTOKEN. The Amazon conversion pipeline emits a landmark entry with an empty value:

<a epub:type="" href="...">Title Page</a>

Apple Books validates epub:type strictly and rejects the book at import time. The official validator reports this as:

ERROR(RSC-005): nav.xhtml(49,74): value of attribute "epub:type" is invalid;
must be a list of XML NMTOKENS with length at least 1 (actual length was 0)

Diagnostic Steps

1. Download the official validator (IDPF/W3C reference implementation):

   curl -sL https://github.com/w3c/epubcheck/releases/download/v5.2.1/epubcheck-5.2.1.zip -o /tmp/epubcheck.zip
   cd /tmp && unzip -oq epubcheck.zip

2. Run it against the EPUB (requires Java):

   java -jar /tmp/epubcheck-5.2.1/epubcheck.jar path/to/book.epub

3. Read the report — the error line names the file and the exact row/column of every violation.

Repair Steps

1. Extract the archive:

   mkdir -p /tmp/book && unzip -q path/to/book.epub -d /tmp/book

2. Open OEBPS/nav.xhtml and replace the empty attribute with the valid landmark type:

   <!-- before -->
   <a epub:type="" href="xhtml/xhtml-0-1.xhtml#aid_355">Title Page</a>
   <!-- after -->
   <a epub:type="titlepage" href="xhtml/xhtml-0-1.xhtml#aid_355">Title Page</a>

3. Rebuild the EPUB with the correct ZIP layout — mimetype must be the first entry and uncompressed (STORED), everything else deflated:

   cd /tmp/book
   rm -f /tmp/book-fixed.epub
   zip -X0 /tmp/book-fixed.epub mimetype
   zip -Xr9D /tmp/book-fixed.epub META-INF OEBPS

4. Re-validate — expect a fully clean report:

   No errors or warnings detected.
   Messages: 0 fatals / 0 errors / 0 warnings / 0 infos

5. Confirm the ZIP layout is correct:

   unzip -v /tmp/book-fixed.epub | head
   # mimetype must appear first, Method = Stored

Verification on Apple Platforms

• Send the repaired file to the iPhone/iPad (AirDrop, or save to Files and tap → Open in Books)
• Import should now succeed; TOC, landmarks and cover all render from the untouched content

Notes

• Only the navigation document was repaired — content, images, and table of contents are byte-identical after rebuild.
• If Apple Books still refuses after a clean epubcheck run, the remaining culprits are rarer (missing dcterms:modified, malformed META-INF/container.xml, or a non-UTF8 mimetype first entry) — inspect those before rewriting anything.
• Keep the original download as a backup; the repaired copy is a derivative for personal use.