Bulk import: scan a folder and import everything matched (manual import is one-path-at-a-time)

[vavallee]

Problem

Manual Import / Migrate "Look Up" is strictly one path → one book, which makes importing an existing library brutal. A Discord user with 591 unmatched files has to repeat: expand unmatched files → copy a folder path → paste into Import/Migrate → Look Up → pick the book → import. 591 times.

Two specific rough edges feed this:

1. Pointing Look Up at an author/collection folder returns one bogus match, not the books inside it. Scanner.Lookup only parses filepath.Base(path) (internal/importer/lookup.go:36 → parser.go ParseFilename) and never recurses. Pasting /Data/Media/Books/Terry Pratchett/ parses the string "Terry Pratchett" as a single title and fuzzy-matches it to whatever one catalogue title is closest. It never sees the book subfolders. The user has to paste each individual book folder.

2. No "import everything matched under here" path. POST /api/v1/queue/manual-import takes a single path + single bookId (internal/api/manual_import.go:87). The tree-walking importers don't fill this gap: ABS import is remote-API-only, and library scan only reconciles existing catalogue books (it lists unmatched files but doesn't import them).

Proposed design (sketch)

A directory-scan manual import:

• New endpoint, e.g. GET /queue/manual-import/scan?path=... (admin-only, library-root contained like the existing lookup): walk one or two levels under the pasted folder, group files into book units (reuse the importer's folder grouping), ParseFilename + catalogue-match each, and return a list of per-unit results: { path, parsedTitle/Author, match: confident|ambiguous|none, book|candidates }.
• Frontend (web/src/pages/settings/ImportTab.tsx): render the list with a row per book — confident pre-selected, ambiguous with a candidate picker, none skipped — and a "Import N selected" button.
• Batch import endpoint (or loop the existing single import server-side) to commit the selected (path, bookId) pairs in one request, with a summary.
• Stretch: a dry-run/preview count and an "auto-import all confident" shortcut.

Related

• Discord report (Terry Pratchett library, 591 files).
• The "it might be the dash" half of that report is a separate, already-fixed parsing bug: fix(importer): normalize Unicode dashes so em-dash folders parse #1291.

Filing as a feature (not folding into the dash fix) because it needs a new response shape, batch endpoint, and UI. Happy to implement on the green light.

[vavallee]

Second report, from the download-client side: discussion #1225 (Kedryn) migrated from Readarr with ~500 completed torrents in qBittorrent's book category that Bindery ignores.

Root cause confirmed: Bindery imports only downloads it grabbed itself. A downloads row is created at grab time tied to a BookID (internal/scheduler/scheduler.go:589), and the poll loop ranges over those rows and looks each up in the client by hash — it never enumerates the client's torrents (internal/importer/scanner_poll.go:339). Untracked items have no row, so they're never visited (not surfaced as unmatched either). Manual import is filesystem-path + a required BookID, one at a time.

This adds a dimension to the feature:

• (a) filesystem folder scan (this issue) — covers Kedryn too, since their qbit download folder is already mapped into Bindery's download dir, so the files are on a path Bindery can see.
• (b) the migration gap: matching runs against the existing catalogue, so a Readarr migrant whose books aren't in Bindery yet would get everything as "unmatched." Worth considering a metadata-lookup-and-create step (parse file -> resolve via provider -> create book -> attach), not just match-against-existing. At minimum the bulk UI should make the "add the authors/books first" prerequisite obvious.

Two independent users (jiggs ~591 files via Discord, Kedryn ~500 via #1225), both Readarr migrants, is a strong signal this is the top migration pain point.

[vavallee]

MVP implemented in #1293: GET /queue/manual-import/scan + POST /queue/manual-import/batch plus a "Bulk folder import" UI in Settings → Import (scan a folder, confident pre-checked, ambiguous picker, import the selection in one click). Matches against the existing catalogue. Lookup-and-create for cold migrations remains the follow-up tracked here.