added a bunch of typing for beets AlbumInfo, TrackInfo and managed (poorly) to serialize them.

…tarting new session.

added types to frontend. candidates and prompts communication works, with some rough edges.

…is consistent.

…n-functional.

the websocket routes into their own folder (Terminal and Importer).

Added a text input to define the path of the import, for debugging.

No output in frontend thus far, but connection and init seems to work.

Frontend still broken

Front-end needs fixing according to new types tho. Also status updates
are not reimplemented yet.

…oss reloads.

…nmounting after we are done.

Moved the files for candidate formatting into its own folder
and removed some unnecessary css declarations.

Enhanced the grid layout of the trackChanges class
for proper number alignment.

Adjusted font sizes for trackChanges to overall fit more content.

Added a temporary component to view the current candidate changes.

… knows beets better

…now checks

multidisc consistent with beets. `all_album_folders` (used by inbox!) is some 20% faster
but still ~75 sec to scan ~250gb of muisc folders via lan from spinning discs.
we still need to handle this better - large inboxes are still not doable.

….walk.

some 10% faster than glob. also sessions should now work on multi-disk albums.

… have `import/duplicates = ask`

showing duplicates works, todo: handle them and tidy up.

… 5...

I attached the duplicate action to the selection state (instead of candidate).
Also was lazy and re-used the existing socket-event (candidatechoice).
we should probably tidy this up a bit more. but it seems to work.

…vention:

for dataclasses, define all attributes with default values (that should not be set)
in the __post__ init, so it remains easy to see whats needed for the constructor,
since we do not code an __init__ function.

…on selected

(todo: check that default duplicate action from config works)

…lso added badge,

and reordered the display. badge needs better styling: fix height and width
and maybe create some character padding / centering.

Not quite there yet but feels like a good point to do another high-level planning session before finalizing

…d mutation

socket hook.

…a bit

annoyed by typing the path again and again...

I had to update mui for this as I wanted to use autocomplete. Hope this
produces no issues.

searchparams.

- Added abort session ws endpoint
- added more details to status messages, fully typed now
- added abort button in frontend depending on current state
- fixed issues with initial state set by clients, this produces a deadlock
 if multiple clients are connected and all want to reset the candidate

WIP: Apply button still needs some work i.e. it is broken atm and does
not work, also styling of the apply btn is not too nice.

…zontal alignement with 'asis'

…status

but then realized we likely dont need it yet.
Layout should be more streamlined for the (common) case that we only import one selection.
I'd group all buttons at the end (check length of the selectionstates
to either place apply+abort outside, or inline with duplicate and search)

…viour together.

(currently scroll only takes place on inner container, which looks weird.
i think we need a helper css to get limited-width for wide windows to avoid the nested containers!)

…rminalcontext

is not global anymore. any ideas for workarounds?

Library-Search still needs work.
And Search and Library List views seem a bit broken for me currently - no items showing.
(sporadically)
Maybe since MUI upgrade?

useQueryParamsState hook.
Updated tanstack to latest version.

candidate.

auto parses svgs as react components.

rq. Needs some more testing tho.