chdorner / epubinfo Goto Github PK

We have a couple of Epubs that have dots in the cover image item ID, e.g.

<item id="cover.image" properties="cover.image" href="orm_front_cover.jpg" media-type="image/jpeg"/>

It fails to load the cover image file, because Nokogiri blows up when you try to get the item:

manifest.css("item [id = cover.image]")

We tried escaping the CSS selector, but that rendered no results instead of blowing up.

Nokogiri 1.6.4 / 2014-11-04 release announcement mentions this fix:

• Fix a bug with CSS attribute selector without any prefix where "foo [bar]" was treated as "foo[bar]". (#1174)

Since this fix, cover parsing fails in /models/cover.rb.
Spaces should be removed in CSS attribute selectors in epub_cover_item method (lines 78 and 82 to 85).
Ex:
manifest.css("item [id = \"#{cover_id}\"]")
should become
manifest.css("item[id = \"#{cover_id}\"]")

I did not see any other impact of this Nokogiri fix in epubinfo.
Thanks.

Some epubs have dc:subject which are subject areas or "tags". I might look into a fork to add this, but it will probably be a while.