Aura H2O Annotation xml format

[Gergely]

I copied the annotation -- .epub.annot -- file from my Kobo Aura H2O device.

It is in xml format, an annotation example is here:

<annotation>
<dc:identifier>urn:uuid:35330bcf-af11-43f4-8eb3-9e346a3472ef</dc:identifier>
<dc:date>2018-04-15T14:04:47Z</dc:date>
<target>
<fragment start="OEBPS/9781118087220c07.xhtml#point(/1/4/2/134/1:2)" end="OEBPS/9781118087220c07.xhtml#point(/1/4/2/134/1:100)" progress="0">
<text> advanced sales technique is for salespeople to overcome the objections posed by their prospects. </text>
</fragment>
</target>
<content>
<dc:date>2018-04-15T14:04:47Z</dc:date>
</content>
</annotation>

Where can I find a description for this xml format? Especially I am interested in understanding

<fragment
start="OEBPS/9781118087220c07.xhtml#point(/1/4/2/134/1:2)"
end="OEBPS/9781118087220c07.xhtml#point(/1/4/2/134/1:100)"
progress="0">

[Doitsu]

Looks like EPUB Canonical Fragment Identifiers 1.1.

[davidfor]

I don't think it is the ePub CFI. It is what Adobe uses for their locations in ADE and the RMSDK. It might be what the ePub CFI is based on.

@Gergely: And it is the same format I described the other day. Kobo's epub reader is based on the Adobe RMSDK and uses these in the bookmark table for epubs and PDFs. I haven't found an official definition, but I also haven't looked very hard.

[Quoth]

I thought annotation is in the database?
I have noticed that recently in Calibre, the annotation is now raw HTML instead of plain text. I don't know if a Calibre update or the 4.8 FW update has caused it. I now have to strip that, as it's of no use in my procedure to use annotations & edit my source while updating what I've done in the annotation text file.
I like how Calibre reads the annotation on a per book basis and how the Kobo has the % in chapter and source text highlighted. I find the annotation export poor on my Kindle DXG, apart from being awkward to create.

[davidfor]

Originally Posted by FrustratedReader

I thought annotation is in the database?

They are in two places. This has been discussed to many times here including in the last few days.

I have noticed that recently in Calibre, the annotation is now raw HTML instead of plain text. I don't know if a Calibre update or the 4.8 FW update has caused it. I now have to strip that, as it's of no use in my procedure to use annotations & edit my source while updating what I've done in the annotation text file.

And have you reported this problem somewhere so that someone can look at it and fix it? And of course, with enough details so they know what you are actually doing.

[Gergely]

I have found something:

http://www.mobileread.mobi/forums/sh...=205062&page=6

/1 indicates the root element (<html...)
even numbers navigate through the elements. So 2 is the first child of the root (<head...), 4 is the second child of the root (<body...), /1/4/10/ is the 5th child of the body (usually the 5th paragraph or header).
: idicates the offset within the element.

However, I am not quite sure what odd numbers mean. They usually occur if the highlight starts at the beginning of the paragraph. So the position before the highlight is "between the paragraphs". But in case of /1/4/9:2 --> what's the second character of "between paragraph 4 and 5? Do you happen to know that, by chance?

Is this reasonable about even and odd numbers? I will write some python code to check this.

[davidfor]

Originally Posted by Gergely

I have found something:

http://www.mobileread.mobi/forums/sh...=205062&page=6



Is this reasonable about even and odd numbers? I will write some python code to check this.

Well, yes. As that is what I said the other day, and said in the following posts in that thread. It will be interesting to see what you produce.

[davidfor]

Originally Posted by FrustratedReader

I thought annotation is in the database?
I have noticed that recently in Calibre, the annotation is now raw HTML instead of plain text. I don't know if a Calibre update or the 4.8 FW update has caused it. I now have to strip that, as it's of no use in my procedure to use annotations & edit my source while updating what I've done in the annotation text file.

This has probably been fixed in calibre 3.27.0. Luckily your rather vague description of the problem reminded me of a problem in the Annotations plugin cause by a recent change in a library calibre uses. The change also affected the Fetch Annotations function. If it doesn't fix the problem, please report it in the calibre forum or open a bug report.