Creating valid epub from command line using zip

[bquinn]

Is there really a way to use zip on the command line to create an epub file that validates properly with epubcheck? I have tried every sequence of command line commands that I can find on the Internet and they all say they work, but I get exactly the same errors no matter how I do it. I suspect that the people who claim that zip creates valid epub files have not validated them with epubcheck. The files this method creates seem to work on a few ereaders, but not all of them, and it always fails epubcheck.

Here is one of the typical ways that I've tried to do it, but it always gives me the same errors.

$ rm ./test-epub-3-0-1.epub 
$ zip -X0 ./test-epub-3-0-1.epub mimetype
  adding: mimetype (stored 0%)
$ zip -rDX9 ./test-epub-3-0-1.epub META-INF -x mimetype
  adding: META-INF/container.xml (deflated 30%)
$ zip -rDX9 ./test-epub-3-0-1.epub OEBPS -x mimetype
  adding: OEBPS/cover.xhtml (deflated 29%)
  adding: OEBPS/title_pages.xhtml (deflated 30%)
  adding: OEBPS/chapter01.xhtml (deflated 30%)
  adding: OEBPS/epub.css (stored 0%)
  adding: OEBPS/toc.xhtml (deflated 32%)
  adding: OEBPS/chapter03.xhtml (deflated 30%)
  adding: OEBPS/notes.xhtml (deflated 30%)
  adding: OEBPS/MyBook.opf (deflated 67%)
  adding: OEBPS/chapter02.xhtml (deflated 33%)
$ cd ..
$ java -jar epubcheck-3.0.1/epubcheck-3.0.1.jar test-epub-3-0-1.epub 
Epubcheck Version 3.0.1

ERROR: test-epub-3-0-1.epub: Mimetype entry must not have an extra field in its ZIP header
Validating against EPUB version 3.0
ERROR: test-epub-3-0-1.epub/mimetype: Mimetype file should contain only the string "application/epub+zip".
ERROR: test-epub-3-0-1.epub/XHTML/MyBook.opf(2,161): element "package" missing required attribute "unique-identifier"
ERROR: test-epub-3-0-1.epub/XHTML/MyBook.opf(2,161): unique-identifier attribute in package element must be present and have a value
ERROR: test-epub-3-0-1.epub/XHTML/MyBook.opf(9,13): element "metadata" incomplete; missing required element "dc:identifier"
ERROR: test-epub-3-0-1.epub/XHTML/MyBook.opf(3,12): assertion failed: package dcterms:modified meta element must occur exactly once
ERROR: test-epub-3-0-1.epub/XHTML/MyBook.opf(10,12): assertion failed: Exactly one manifest item must declare the 'nav' property (number of 'nav' items: 0).
ERROR: test-epub-3-0-1.epub/XHTML/MyBook.opf: unique-identifier attribute in package element must reference an existing identifier element id

I am on Linux and don't have access to many of the utilities and I would much prefer to create my own epub files, but I am not sure there is a real way to create truly valid epub files using zip. I know that epubcheck works because it returns no errors for epub files that I have downloaded, but there does not appear to be a way to make one on the command line that validates. I know that my mimetype file is correct and at least the first two errors should not be there. I haven't looked into the other errors because it doesn't even get past the mimetype file.

Does anyone have any recommendations on how I might be able to do this so that epubcheck will validate the file?

[Jellby]

I use a script like this all the time:

#!/bin/bash

rm -f "$1"
zip -X0 "$1" mimetype
zip -X9Dr "$1" META-INF OEBPS

which is basically the same you do.

Now, the question is, what is the content of your "mimetype" file? It should be:

application/epub+zip

and only that, no newline at the end or anything else.

[Toxaris]

Your problem is not the zip commands, but the content...

[bquinn]

The contents of my mimetype file are exactly what Jellby shows in the example, with no newline character, and I have not been able to eliminate that error. I haven't begun to look at the other errors. Will epubcheck show errors for the mimetype file if the real problem is in some other file?

[j.p.s]

Originally Posted by bquinn

The contents of my mimetype file are exactly what Jellby shows in the example, with no newline character, and I have not been able to eliminate that error. I haven't begun to look at the other errors. Will epubcheck show errors for the mimetype file if the real problem is in some other file?

What is the output from:

od -t x1 mimetype

[eschwartz]

I would suggest using ebook-convert (the command-line interface to calibre's conversion utility).

Also, epubcheck is reporting "ERROR: test-epub-3-0-1.epub/mimetype: Mimetype file should contain only the string "application/epub+zip"." which seems to indicate that there is something fishy there.

[Toxaris]

Originally Posted by bquinn

The contents of my mimetype file are exactly what Jellby shows in the example, with no newline character, and I have not been able to eliminate that error. I haven't begun to look at the other errors. Will epubcheck show errors for the mimetype file if the real problem is in some other file?

Could you look at the mimetype file with a hex-editor? There could be a strange character there. The commands you use should work.
Other than that, your OPF is not correct. You are missing an identifier and it seems not to be complete. Did you create the OPF by hand?

[j.p.s]

Originally Posted by Toxaris

Could you look at the mimetype file with a hex-editor? There could be a strange character there. The commands you use should work.
Other than that, your OPF is not correct. You are missing an identifier and it seems not to be complete. Did you create the OPF by hand?

Wouldn't posting the output of
od -t x1 mimetype
in a message here do the same thing and short circuit all ambiguities and miscommunications that happen when people attempt to discuss technical details in natural language?

[eschwartz]

The obvious solution is to use calibre, whose job is to package ebooks. The command-line interface to the conversion module accepts an OPF as input.

Or use the editor which handles all packaging and leaves you free to focus on the content.

Oh yeah, and I think Kovid finished the plugin system because it's an optional toolbar in the editor now. So extra functionality can be added in new an excitind ways.

[Steve_White]

Create the ebook with the -X option, to stop it adding the "extra field" to the zip (the first error message), and '-Z store' to avoid compressing the mimetype file.

The mimetype file should have length 20 bytes. Check with command 'ls -l'.
If the length is greater, you probably have a newline at the end. But most editors put this on automatically. To make a mimetype file,

echo -n "application/epub+zip" > mimetype

Then create the book like this (change names where appropriate)

zip -Z store -Xr MyBook.epub mimetype
zip -r MyBook.epub META-INF/
zip -r MyBook.epub Content/

Cheers!