Repository navigation
Feature request: Auto mapping for books added to Storyteller, and handle multiple hashes for plain/readaloud epubs #519
Replies: 1 comment 7 replies
|
Thanks for the request — both parts are now in dev. Both EPUB versions count as the same book. When a Storyteller book is linked in BookBridge, whether through Add Book or automatically, BookBridge now also recognizes the plain EPUB that Storyteller serves over OPDS and the read-aloud EPUB. KOReader or Crosspoint can sync from either file without any manual linking. Books you've already linked are picked up by an hourly check, and re-aligning a book in Storyteller is handled too. Automatic linking. There's a new option under Account → Integrations → Storyteller: Automatically link aligned Storyteller books (off by default). Once a book finishes aligning in Storyteller, BookBridge links it within the hour. No collection is needed. If a book's title matches one you've already mapped, it's left alone to avoid creating a duplicate; use Add Book for those. It'll be in the next |

Uh oh!
There was an error while loading. Please reload this page.
I am using just Storyteller (no Grimmory, ABS, etc) as my book repository to store 3 files for each book I have: the plain epub, the audiobook, and a Storyteller-generated readaloud epub. I don't really want to add another management system/layer to that.
Whenever I add a new book to Storyteller by first adding the epub and m4b/mp3 audiobook and then aligning them to create a readaloud epub, I have to then manually map them in BookBridge by selecting "Add Book" and then searching for the title and selecting the second option for Storyteller (preferred), then select add to queue, and finally click the "match all" button on the queue.
However, BookBridge seems to already be aware of the books I have in Storyteller, because on that Add Book page the search only produces results that match the books I already have in Storyteller.
There's actually two requests here: One would be to have a way to have BookBridge automatically map new books as they are added to Storyteller. The Storyteller API is already configured as an integration, which I assume is how the search is finding the titles. I would be fine with adding them to a specific collection within Storyteller if that lets you do something similar to the auto mapping functionality that already exists for BookOrbit.
The secondary request is to map the hash of both the readaloud epub and also the plain epub, because for some e-ink devices running clients like KOReader or Crosspoint, I pull the plain epub directly from Storyteller via its OPDS server and this will have a different hash than the larger readaloud epub with embedded audio that I use via other clients on phones/tablets. If both flavors of epub get mapped as a single book (which is how they are represented in Storyteller), then I wouldn't need to separately manually link the plain epub hash later, which is something I have to do currently when I've mapped a book from Storyteller but then Crosspoint sends a progress sync from the plain epub.
I do also have my Storyteller
/data/assetsfolder mapped to Bookbridge in the settings as an "Extra Ebook Library" called/storyteller-booksin case that helps with handling the multiple file hashing scenario.All reactions