Make px:epub3-create-package-doc generate a predicable unique-identifier

[kalaspuffar]

Currently, we have a failing generation of EPUBs when we validate them against the nordic guidelines. The failure is based on the fact that unique-identifier should be equal to the pub-identifier value.

Expected Behavior

pub-identifier and unique-identifier need to be present and equal in EPUB to follow nordic guidelines.

Actual Behavior

A pub-identifier is created with a value not equal to the unique-identifier.

Steps to Reproduce

Provide a list of steps taken to reproduce the problem.

1. Checkout the nordic-epub-migrator repository
2. Change the pom file to use 1.14.2 modules-parent dependency
3. Run test cases

Environment

• Linux Gentoo
• DAISY Pipeline 2 version: 1.14.2
• Interface: Test of nordic-epub-migrator

Logs

[opf2] on the package element; the unique-identifier-attribute must be present and equal ‘pub-identifier’.

[bertfrees]

Thanks @kalaspuffar.

Just to be clear: the Nordic guidelines require that unique-identifier has a value of "pub-identifier". This is just some random value, there is no such thing as "a pub-identifier" in EPUB.

What px:epub3-create-package-doc could do is make the unique-identifier attribute more predictable.

For instance, currently when dc:identifier elements are present in the "metadata" input, the first one becomes the unique identifier. We could change px:epub3-create-package-doc so that the id attribute of this dc:identifier element (and therefore the value of the unique-identifier attribute) is preserved if present.

Another solution could be to add an option to set the value of the unique-identifier attribute (and therefore of the id attribute of the corresponding dc:identifier, either generated or coming from the metadata input). But I like this solution less.

[martinpub]

Hi, just wanted to verify @kalaspuffar and @bertfrees, is this affecting all epub3-to-epub3 uses, e.g. including metadata injection?

[bertfrees]

Possibly yes.

@martinpub Also consider relaxing this restriction. Personally I think it makes no sence to enforce an IDREF attribute. Do you know what was the purpose of this requirement?