I simply can not believe calibre's quality check approves of using an "&" in a file name inside an epub!
Even so urls/uris are supposed to be % encoded not xml escaped. So unless Kovid can point to the epub spec where "&" chars are supposed to xml escaped in urls or uri's in the opf, I consider this a bug in calibre not Sigil.
An & is not a legal char in a uri/url filename or path and you can not xml escape an & without using another &!
percent encoding of spaces, &, #, etc is how the spec says to deal with these chars.
That said, I could try to xml unquote the filename before % decoding but I think that will probably open a kettle of fish for something that is simply not a good idea at all.
Sorry but I am going to need to see where in the opf spec it says you can do that. It should have been % encoded and then no xml quoting or unquoting would be needed.
To workaround this, unzip the epub and use AddExisting to add the missing xhtml files. Sigil should properly % encode these when adding them to the opf.
Quote:
Originally Posted by Doitsu
I was able to reproduce this issue with a file that was originally created with Calibre. If you add an ampersand to a file name in Calibre, it'll escape it as & instead of %26.
Apparently, the ampersand and the semicolon in & are escaped by Sigil and the file isn't found:
Code:
<item id="test.xhtml" href="Text/%26amp%3Btest.xhtml" media-type="application/xhtml+xml"/>
AFAIK, Calibre will sometimes add very long JSON-based metadata entries. I couldn't find an example, because I usually remove these entries. Hopefully, un_pogaz will provide an example. |