NavDoc & NCX parsing: when to parse which one about swift-toolkit

[banux]

Hi,

Currently in the go streamer i parse the navigation document first and if i can't populate the TOC i try the NCX. So 1) and 3) i think try to merge the 2 documents will be difficult and not very useful.

[llemeurfr]

I think that this process is good: 1) try the nav doc (EPUB3), and if it does not exist or contains no ToC 2) try the NCX (EPUB2).
This leaves some flexibility if an EPUB3 (by its declared version attribute) contains no nav doc (which is an authoring mistake).
Doing so also avoids mixing info that may be inconsistent between the nav doc and the ncx, e.g. page lists, landmarks ...
We should create a processing model document to describe such things clearly.

[HadrienGardeur]

This should be documented on the main Readium-2 repo in a new document at the following place in the repo: https://github.com/readium/readium-2/tree/master/streamer/parser

[HadrienGardeur]

// There is a reference to toc and a file name (in the href). That's not coherent?
  "landmarks" : [
    {
      "href" : "#toc",
      "title" : "Table of Contents"
    },
    {
      "href" : "p10.xhtml",
      "title" : "Begin Reading"
    }
  ],

We actually fixed self-referencing navigation documents in a recent commit: readium/go-toolkit#21

Here's an output from the live version of the Go streamer at https://readium2.feedbooks.net/Ym9va3MvbW9ieS1kaWNrLmVwdWI=/manifest.json

"landmarks": [
  {
   "href": "toc.xhtml#toc",
   "title": "Table of Contents"
  },
  {
   "href": "chapter_001.xhtml",
   "title": "Begin Reading"
  },
  {
   "href": "copyright.xhtml",
   "title": "Copyright Page"
  }
 ]

Ok great, will update that on the streamer-swift too.