inpxer

OPDS 1.1 and web server for .inpx libraries with full-text search.

Usage

Standalone

Download the latest release. Download inpxer-example.toml, rename to inpxer.toml, put next to executable (or current working directory) and edit to your liking.

Import data:

./inpxer import ./file.inpx

You can specify --partial flag to import only new records and keep old ones. Otherwise, old index data will be deleted (the whole folder specified in index_path) and reindex from scratch.

Start server:

./inpxer serve

Web interface will be available on http://localhost:8080/ and OPDS will be on http://localhost:8080/opds by default.

Docker

Download inpxer-example.toml, rename to inpxer.toml and edit to your liking.

inpxer expects config file to be at /data/inpxer.toml.

Images are published to GitHub Container Registry (GHCR): ghcr.io/hedger/inpxer.

Quick start: auto-import on first run

Set INPX_FILE to the path of your .inpx file (mounted inside the container). The entrypoint will:

• Check if /data/index exists and matches the .inpx version;
• Import (reindex) if missing or mismatched;
• Start the server.

docker run --rm -it \
	-e INPX_FILE=/import/flibusta_fb2_local.inpx \
	-v ./import/:/import \
	-v ./data/:/data \
	-p 8080:8080 \
	ghcr.io/hedger/inpxer:latest
	# If you built locally instead, use: inpxer:latest

Optional flags via env vars:

• PARTIAL_IMPORT=true to only add new records (do not delete old ones)
• KEEP_DELETED=true to keep records marked as Deleted in INP

Manual import, then serve

You can still run import explicitly and then start the server:

# Import (may delete existing index unless PARTIAL_IMPORT is used)
docker run --rm -it \
	-v ./import/:/import \
	-v ./data/:/data \
	ghcr.io/hedger/inpxer:latest \
	inpxer import /import/file.inpx

# Start server
docker run -it \
	-p 8080:8080 \
	-v ./data/:/data \
	ghcr.io/hedger/inpxer:latest

Data volumes and layout

The container expects a single writable volume mounted at /data that contains:

• /data/inpxer.toml — configuration file. See inpxer-example.toml for all options. Key fields:
  • index_path = "/data/index" (default in example)
  • library_path = "/data/library" — where your actual book files are stored
• /data/index — the index directory created by inpxer (bleve and badger/bolt subfolders). This can be large.
• /data/library — your library files (fb2, fb2.zip, epub, etc.). Required for downloads to work via OPDS/web.

Additionally, for first-run auto-import, mount a read-only helper volume at /import pointing to the directory that contains your .inpx file and set INPX_FILE=/import/yourfile.inpx.

On subsequent runs (when you don’t want to re-index), you can omit the /import volume and the INPX_FILE environment variable.

Permissions note

• The container runs as a non-root user (app) and will ensure /data/index is writable at startup. If /data is a bind mount owned by root on the host, the entrypoint will adjust ownership/permissions of /data/index inside the container.
• To avoid a chown on startup (useful with large existing indexes), you can pre-chown the host directory to UID/GID 10000:10000 or run with an explicit user:

docker run -it -p 8080:8080 \
	--user 10000:10000 \
	-v <path to data storage>:/data \
	ghcr.io/hedger/inpxer:latest

docker-compose example (optional)

services:
	inpxer:
		image: ghcr.io/hedger/inpxer:latest
		environment:
			INPX_FILE: /import/file.inpx
			# PARTIAL_IMPORT: "true"
			# KEEP_DELETED: "true"
		ports:
			- "8080:8080"
		volumes:
				- ./data:/data # contains inpxer.toml, index/, and library/ (as configured)
				- ./import:/import # folder with your .inpx file (for first run)
		# Uncomment to avoid chown inside container if host dir is pre-owned by 10000:10000
		# user: "10000:10000"