Script for e-braille production (EPUB 3 with braille rendition)

[bertfrees]

I've started working on this a while back. See snaekobbi/pipeline-mod-braille#6. I'm continuing the work now on branch sbs-ebraille.

[bertfrees]

Some parts have been dropped for now with respect to snaekobbi/pipeline-mod-braille#6 (see branch "braille-in-epub3"):

• Input-output synchronization is only done on (HTML) file level. I want to rewrite the implementation of fine-level synchronization.
• Move <dc:type>edupub</dc:type> to metadata.xml if present: too specific; need something more general?
• Translate alt attributes to braille: this required a private version of "liblouis-block-translate.xsl", but because the script has been made more generic, I need to come up with a generic solution for translating alt attributes. Maybe the query features (input:html)(output:html) should enforce this.

Some additional things I would like to do:

• Option to pick default rendition
• Reindent/justify HTML after text nodes have been translated to braille.
• More efficient copying of resources. Currently the whole ZIP is loaded into memory which can make things slow if there are a lot of MP3 files.
• Add rendition:media: this is blocked on standardization work being done in the TIES working group. Currently the media attribute in EPUB3 expects a CSS 3 Media Query which does not support braille/embossed.
• ...

[bertfrees]

Also, there is the question of whether we want an SBS version of this script, like is the case with "sbs:dtbook-to-pef" (and "sbs:epub3-to-pef"). The main reason for the existence of the SBS specific scripts is to hide some options and provide new ones.

[bertfrees]

I'm putting this in "review" for now, that is it shouldn't be deployed yet, because I can't run the full test suite (same issue as for sbsdev/pipeline#1 and #9).

[bertfrees]

Test suite works again (#52 (comment)). To do: rebase and merge branch sbs-ebraille.

[bertfrees]

Has been merged in sbsdev/pipeline@25212c2.

I have sent Mischa a test version of Readium with support for multi-rendition which should make it easier for him to test the epub3-to-epub3 script.

[bertfrees]

After some testing we have discovered some problems:

• The script fails to convert big books with audio files due to a OutOfMemoryException. I already mentioned above that copying of resources is slow because the whole ZIP is loaded into memory. I'm not sure whether this also has to do with the OutOfMemoryException.
• The resulting EPUBs are not valid according to epubcheck.
• The resulting EPUBs can be opened in Readium, but after changing the rendition in the settings the waiting symbol is shown and the rendition is not loaded. (After deleting the book and opening it again the new rendition is rendered fine though.)

[bertfrees]

It looks like multiple rendition publications might be dropped from EPUB v3.2. Will verify.

[bertfrees]

To do:

• Make an SBS version of this script that supports EPUBs with brl:* elements.

[mixa72]

Is this deployed already, i.e., does the e-braille script translate brl:* inline elements as in the PEF now?

[bertfrees]

Yes.

[mixa72]

Translation looks good now, I'll send a sample book to Petra, so she can check accessibility etc. Thanks!