The fetched settings are only written into app state when nothing is
already held there, so a refresh landing while the user has the settings
prompt open cannot overwrite edits in progress.
The fixtures give every field a distinct value. Settings whose fields
all share one value cannot detect two of them being mapped to each
other's keys.
Results come back keyed only by indexer id, so each one is joined
against the cached indexer list to recover a display name, and its
validation failures are flattened into a single line for the results
table.
The endpoint reports failures with a non-2xx status, so the request
ignores the status code and parses the body either way. When the
request itself fails the results table is replaced with an empty one
rather than being left showing the previous run.
Fetch the indexer's full details, then post them back verbatim to the
test endpoint. The payload round-trips through serde_json::Value so the
fields Indexer does not model, such as infoLink and the per-field label
and type metadata, survive the round trip intact.
A successful test returns an empty JSON object and clears
indexer_test_errors; a failure returns an array whose first entry
carries the error message. The test endpoint reports failures with a
non-2xx status, so the request ignores the status code and parses the
body either way.
The delete flags travel as query parameters rather than a request body,
and Readarr spells the exclusion key addImportListExclusion, not the
addImportExclusion that Radarr uses.
The PUT body round-trips through serde_json::Value rather than the typed
Book, so the fields Readarr returns but Book does not declare survive the
toggle instead of being silently dropped. BOOK_JSON now carries a few of
those undeclared fields so a regression to a typed round-trip fails the
tests rather than passing unnoticed.
Add the GetAuthorHistory event, backed by GET /history/author?authorId=N,
which returns a flat array of history records for a single author. The
fetcher mirrors the artist-history equivalent, sorting by id and carrying
the sort-prompt route guard so a refresh landing while the sort prompt is
open cannot clobber an in-progress sort selection.
Expose it as `readarr list author-history --author-id N`.
Posts the RefreshAuthor command to the command endpoint with no author ID, which is how Readarr distinguishes a library wide refresh from a single author refresh. The response is returned verbatim and no app state is mutated.
Posts the RefreshAuthor command to the command endpoint with the author ID, queueing a metadata refresh and a disk scan for that author. The response is returned verbatim and no app state is mutated.
Adds the refresh subcommand group to the Readarr CLI.
Posts the AuthorSearch command to the command endpoint with the author ID, queueing an indexer search for that author. The response is returned verbatim and no app state is mutated.
Introduces ReadarrCommandBody, the request body shared by the author and book command calls. Both ID fields are omitted from the payload when absent rather than sent as null.
Add ReadarrEvent::ListAuthors, mapping to GET /author, and wire it up as
`managarr readarr list authors`.
This is the first Readarr event that touches the library rather than
configuration, so it creates the library/authors module tree mirroring
Lidarr's library/artists layout.
list_authors mirrors Lidarr's list_artists, including the sort-prompt
route guard: a response landing while the user has the sort prompt open
must not clobber their in-progress sort selection. Both directions of
that guard are covered by tests.
Adds the GetRootFolders event backed by GET /rootfolder, storing the result
on the Readarr root folders table, and surfaces it as
"managarr readarr list root-folders".
Root folders reuse the shared RootFolder model. Readarr omits the
unmappedFolders key that the other servarrs return, and the field is
optional, so it deserializes to None without changing the shared type.
Only the read path is included here. Adding and deleting root folders
mutate the library layout, so they are left for separate work.
The system network tests asserted that queued events, disk space, logs,
version, tasks, and updates were empty against a default state that was
already empty. Those assertions passed even with the fetcher body removed,
so they reported coverage that did not exist.
Seeding a stale value first makes the empty response test prove the fetcher
replaces rather than merges, and makes the failure tests prove the update
closure never runs on a non-success response by pinning the exact seeded
value.
The assertion on the deserialized payload in the empty response test is
left alone: it reads the response body rather than app state, so it was
never vacuous.
The empty-response and failure tests asserted the maps were empty against a
default state that was already empty, so they passed even with the fetcher
body removed. Seeding a stale entry first makes the empty-response tests prove
the fetcher replaces rather than merges, and the failure tests prove the update
closure never runs on a non-success response.
Dark-Alex-17