vavallee/bindery v1.19.0 on GitHub

{Genre} file-naming token — organise your library into top-level genre folders (e.g. {Genre}/{Author}/{Title}). Genres are sourced from Hardcover's curated taxonomy when Hardcover is enabled; without it the token falls back to OpenLibrary's noisier subject data. New books pick up genres immediately; refresh an author to backfill genres onto books imported before this release.
{Token:default} fallback syntax in naming templates — any token can specify a fallback used when it renders empty, mirroring Calibre's ifempty(...). For example {Genre:Unsorted}/{Author}/{Title} routes un-tagged books to an Unsorted/ folder instead of dropping the folder level.

Genres now prefer Hardcover's taxonomy over OpenLibrary subjects — when a Hardcover match is available, its curated genres (Fantasy, Science Fiction) replace OpenLibrary's raw subject bag (Fiction, American literature, Large type books) for display and for the new {Genre} token. Non-Hardcover enrichers (e.g. Google Books BISAC categories) no longer overwrite genres.
OpenLibrary search results now carry ISBNs (#1121) — the ISBN data OpenLibrary already returns is no longer discarded, so cross-provider de-duplication can collapse OpenLibrary and Hardcover hits for the same edition.

DNB titles no longer carry promotional/genre bloat (#1115, #1114) — German DNB records often append marketing text and genre markers (e.g. : Roman | SPIEGEL-Bestsellerautorin … / Author). These are now stripped, so titles are clean and indexer searches match instead of failing with "title/author keywords did not match release name". (Books added before this release keep their stored title until re-added.)
Deluge grabs no longer fail in VPN/namespaced setups (#1116, #1110) — Bindery now fetches the .torrent itself and submits the bytes to the daemon, instead of having the Deluge daemon fetch the indexer URL (which fails when the daemon can't reach the indexer). The infohash is returned directly, so the add no longer relies on hash polling.
Naming templates drop a dangling separator when a leading token is empty (#1128) — a template like {SeriesNumber} - {Title} now yields Title rather than - Title for a book with no series number. Interior/trailing separators are unchanged.
CSV author import no longer mis-parses long first lines (#1120) — the parser read past its buffer when a CSV's first line exceeded 4 KB, producing phantom rows; found by the nightly fuzzer.

{Genre} uses the first genre Hardcover lists. If Hardcover later re-categorises a book, new grabs follow the new genre, but already-imported files are not relocated.
Per-book genre editing (your own vocabulary, like a Calibre custom column) is not yet supported; this release sources clean genres automatically but does not let you override them per book.