Kindle Scribe notebook format

[jhowell]

It would be useful for me to see a notebook after being manipulated by the new cut/paste and resize tools. That may have introduced some new variants in the data format.

[jhowell]

I am curious about how Scribes handle conflicting changes to a notebook made on different devices. For example putting two Scribes in airplane mode, adding a new page to each, and then syncing. Do both new pages appear in the same order on both devices in the end. Or making different alterations to the same stroke while offline.

[tomsem]

Originally Posted by jhowell

I am curious about how Scribes handle conflicting changes to a notebook made on different devices. For example putting two Scribes in airplane mode, adding a new page to each, and then syncing. Do both new pages appear in the same order on both devices in the end. Or making different alterations to the same stroke while offline.

I looked at this some: strokes 'merge': you get the union of all adds and deletions.

The only question concerns updates.

It may well be possible to update strokes by using eraser to erase part of an existing one. I'll have to test this. But I think it is going to be same situation as updating a text note for a book: the last change to the note 'wins' (sync time, or edit time? there's a case to be made for either approach).

I have not tried new, deleted, and re-ordered page scenarios. Adds and deletes will merge, but which re-ordering 'wins'?

Will do more testing.

[tomsem]

Originally Posted by jhowell

It would be useful for me to see a notebook after being manipulated by the new cut/paste and resize tools. That may have introduced some new variants in the data format.

- initial version
- after copy paste (to page 2) and cut paste (to page 1
- resized page 1 circle

[jhowell]

Originally Posted by tomsem

- initial version
- after copy paste (to page 2) and cut paste (to page 1
- resized page 1 circle

Umm. Did you forget something?

[tomsem]

Originally Posted by jhowell

I am curious about how Scribes handle conflicting changes to a notebook made on different devices. For example putting two Scribes in airplane mode, adding a new page to each, and then syncing. Do both new pages appear in the same order on both devices in the end. Or making different alterations to the same stroke while offline.

First to set baseline expectation:

I created a Note in a Kindle book. Synced & took both Kindles (both Scribes) offline.

On Kindle #1 edited the note at 3:30pm.
On Kindle #2 edited the note at 3:33pm, took it online, and synced.
Took Kindle #1 online and synced. The edit remained unchanged (did not pick up updated note).
Synced Kindle #2. It picked up the edit made on Kindle #1.

So un-synced changes are treated as 'new', regardless of when the actual edit was made.

I found this somewhat disturbing: if you make note edits on a Kindle and leave it offline for some days or weeks, meanwhile dutifully updating and syncing on some other one, any notes shared on both will wind up with the 'old' data when the other Kindle syncs. But probably not a 'real world' scenario.

On to Notebooks:

Scenario 1
- I drew a box (one continuous stroke), synced to both Kindles and took both offline
- used eraser to erase one side of the box on one and the opposite side on the other
- added scribbles on both
- added a page with both, deleted a page on one, and changed order on the other
- went online with both and synced (do not remember which order)

Result:
The added scribbles from both show up
The deleted page is gone
The added pages are there
The page re-order happened.
But: only one of the box sides went away, I assume with the 'last sync wins' rule.

Scenario 2: change page order on both
- go offline
- re-order pages on each, in a different way, adding labels indicating the resulting order
- go online with one and sync
- bring the other online and sync
- manually sync the other

Result:
After syncing is "complete", neither order 'wins'. I tried bringing them both online at about the same time, and waiting before syncing the second one, and in both cases some new order resulted. I also found that the updated page labels I put in to track the order changes took some time to propagate in both directions, several minutes at least.

It seems to indicate they need to improve things before more people have more than one device that's able to edit Notebooks (e.g. if they add editing feature to iOS or Android).

---

Also, quite oddly, I found I was unable to update one of the pages in this Notebook, and when closing and re-opening the notebook (or just moving to a different page and back) it reflects a history of my attempts to erase and re-draw (as if erase had no effect). Could be unrelated to being offline. I've made a copy of this notebook for further analysis and deleted the 'offending' page and it has not happened again.

[tomsem]

Originally Posted by jhowell

Umm. Did you forget something?

Perhaps that, but I blame Safari. It's been acting up and eating some of my posts.

[tomsem]

It occurred to me that some people might be interested in extracting, converting and applying Print Replica annotations to the original PDF. But then I thought about it more, and have a feeling that Amazon will be improving the export to do exactly that, and we will not have to wait too long for this.

The one thing they have not delivered on is the PDF Markup story, but this week's update was a necessary step towards that goal. I think the next update will be adding at least some other tools from PDF markup palette: Shape, Text, Underline, Strikethrough, etc.

The current export product is only a placeholder while they get these other pieces in place. The end goal has to be the original PDF, with Scribe annotations converted to PDF annotations, or it will be worthless for most requirements. And if the annotation palette is not rich enough, it's also not measuring up.

They're going to want to promote Scribe for Back To School, the Print Replica part is good to go now, so I think we'll see this PDF related stuff by July.

That said, I'm trying to find time to work on a command line tool for preparing PDFs for Send to Kindle.

The main thing is to detect chapters and create PDF bookmarks for them (these become Print Replica ToC entries). Then detect Contents listing book sections and create page links from and to. And for extra credit, create page links for Index entries, Footnote references etc.

[jhowell]

Originally Posted by tomsem

- initial version
- after copy paste (to page 2) and cut paste (to page 1
- resized page 1 circle

Thanks for posting that sample.

Amazon has implemented moving and resizing strokes using the transform attribute. I have added support for that in the KFX Input plugin release I am working on.

[tomsem]

Some people are disappointed that their 'PDF' annotations cannot be sent off for conversion to text.

For example, Amazon sells 'guided journal' books for Scribe: the pages have space to write on so you wind up with a bunch of writing, but there's not much help when it comes to transcribing them (apart from laborious copy paste to a notebook and then sending it for conversion to text).

I am going to try the dead simple trick of taking the personal document nbk and overwriting the one for an existing notebook. It might not like the lack of a template declaration. But if it works, it would be a welcome discovery.

If it only lacks a template declaration, is there some way of editing the .ion file and then re-baking the nbk?

[tomsem]

Originally Posted by tomsem

I am going to try the dead simple trick of taking the personal document nbk and overwriting the one for an existing notebook. It might not like the lack of a template declaration. But if it works, it would be a welcome discovery.

It did not work. But it is an interesting failure: the notebook has a lower case letter 't' in the upper left corner, rather than the nothing I was expecting.

[HackerDude]

Originally Posted by jhowell

More notebook examples would help. Having examples of simple drawings, such as a single dot or straight line, would be useful in decoding how drawing data is stored. Seeing the .notebook files associated with annotated documents would also help fill in the gaps.



I am surprised by what we have found so far. I expected the Scribe to store something like SVG (scalable vector graphics). Instead it is keeping lower level data, like pen tilt and pressure, as binary data.

Spoiler:

      'nmdl.stroke_bounds': [3123, 4448, 3447, 4707],
      'nmdl.color': 0,
      'nmdl.brush_type': 7,
      'nmdl.random_seed': 0,
      'nmdl.thickness': 23.625,
      'nmdl.type': 'nmdl.stroke',
      'nmdl.stroke_points': {
        'nmdl.thickness_adjust_factor': {{
          AQFMAQAAFERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERGQ=
        }},
        'nmdl.density_adjust_factor': {{
          AQFMAQAAFERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERGQ=
        }},
        'nmdl.num_points': 332,
        'nmdl.position_x': {{
          AQFMAQAAHV1ubdRERETW1NVERE1dRdReVE1URNXVTVRU1EXV1EREREREREVNXUVNXURdRURF3V1URERd
          1UXVREXUbURERNVEVNREVNRdVE1URE1URNVNVERNXV1FRdRdRU1dVERUTV1ERF1FRE5dVNRF1F1ETVRN
          VERF1ERdRERdXV1NRF1NXV3VTV1NXVRN1U1U1dVF1ERdXV1FRNVETVTU1URF1UVF1UVF1rQ=
        }},
        'nmdl.position_y': {{
          AQFMAQAAFF1ERU1ERERF1NbURF1ERURERE1V1EXV1ERF5V5URNVF1dXURF1ERF1ERdRF5URERERETVRE
          RERERURF1URERE1E1V1dRNVEXUREXURdXdRdRU1F1ERETW1E1dXV1E1dVE1eVEREXUXV5URdRERERdRE
          1dREXURETV1URE1dVERERF1dRNRd1dVETV1ERF1G1FXVREREXVREXW5U1URE1d1F3V5e3RI=
        }},
        'nmdl.pressure': {{
          AQFMAQAAFV1NcZGU3heUSfGRlJ7vYZTZHmkdVJ9FcWdUT01PV5bpGe9E0fVO1HUZGWTX4e5pHfleZpTd
          3V9JcZdeHVXU1fTUbW1OfWfk15Xk1k1k3lQdTX5dTUREQZTZFGQZVPfk5U1uVuRGQZfZHlcZafHm5W8f
          Zd51VOftTd15VZFlVn0UdEkd7f/V9E517fXdVUfU3Wkd1pbxnVnflGndkfT+mZmRkRERGR13PQofJCgE
          FR4cFgUEBAQEBAQEBAUEBAQEBQUEBAUFBAQEBQQFBQQEBAQGBQgGBQYGBQQJBAcFBAUEBAQEBAQEBgQI
          BAUGBQQFPpcXZwYGCwUvLCAH
        }},
        'nmdl.tilt_x': {{
          AQFMAQAAJEREQZRERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERERERERERERERERERERBlEREREREREREREREREREREREgNZGRkZA==
        }},
        'nmdl.tilt_y': {{
          AQFMAQAAJBlEQZRERERERBlERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          REREkURERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERERE
          RERERERERERERERERERERERERERESRRERERERERERERERBlERERERERERERERERERERERGwHZGTIyGRk
          ZGTIyMjI
        }},
      },

I have never worked on software for digitizing drawings so I am unfamiliar with the data structures and APIs typically used in that field. Perhaps someone familiar with drawing devices could suggest a documented data format that would be suitable for exporting Scribe drawings.

I do know that some work has been done to reverse engineer the internal data formats used by reMarkable tablets. That might be worth looking into.



I could not find anything relevant online about "NMDL". I am assuming that it is an acronym made up by Amazon. The phrase "note model" appears often in the Scribe firmware. That might be the source of "NM".

The actual NMDL data itself:

AQFMAQAAFERERERERERERERERERERERERERERERERERERERERE RERERERERERERERERERERERERERERERERERERERERERERERERE RERERERERERERERERERERERERERERERERERERERERERERERERE RERERERERERERERERERERERERERERERERERERERERERERERERE RERERERERERERERERERERERERERERGQ=

Seems to be base64 encoded binary... unfortunately since it's encoded binary even after decoding it's not exactly easy to understand...

[HackerDude]

Additionally, something I found interesting, KDF files did exist before the scribe... but there's very little information on them

https://www.kdpcommunity.com/s/quest...language=en_US

[HackerDude]

Originally Posted by Bluebotlabs

The actual NMDL data itself:


Seems to be base64 encoded binary... unfortunately since it's encoded binary even after decoding it's not exactly easy to understand...

Perhaps it is ION?
https://amazon-ion.github.io/ion-docs/

[jhowell]

Originally Posted by tomsem

I am going to try the dead simple trick of taking the personal document nbk and overwriting the one for an existing notebook. It might not like the lack of a template declaration. But if it works, it would be a welcome discovery.

Personal notebooks that are based on the blank page template also lack any template content. As far as I can tell they are internally identical to the notebooks used for annotations.

Originally Posted by tomsem

If it only lacks a template declaration, is there some way of editing the .ion file and then re-baking the nbk?

There is currently no software (outside of Amazon's) that can alter the format used by notebooks (KDF). It would be possible to create something but so far the need has not been great enough to be worth the effort involved.

Originally Posted by tomsem

It did not work. But it is an interesting failure: the notebook has a lower case letter 't' in the upper left corner, rather than the nothing I was expecting.

That is an odd result. Perhaps the software is having problems because the internal identifiers and state of the notebook do not match what was expected. The Scribe firmware needs to keep track of the state of each notebook in order to support update syncing. Switching out one notebook with another might break that in various ways.

If the notebook you substituted was not based on the blank template it might be worth trying again with one that is. That would be a closer match for an annotation notebook.

I did notice one change in the latest Scribe firmware, handling of "orphan" notebooks. That appears to refer to notebook files found in the .notebooks folder that are not expected to be on the device. As far as I can tell the firmware will now delete orphan notebooks when they are detected.