ambry-app / ambry Goto Github PK

View Code? Open in Web Editor NEW

69.0 2.0 8.0 3.42 MB

Self-hosted audiobook streaming server

Home Page: https://ambry.app

License: MIT License

Elixir 12.16% Dockerfile 0.02% JavaScript 0.39% Shell 0.01% HTML 87.41% PLpgSQL 0.01%

audiobooks elixir phoenix liveview self-hosted streaming hacktoberfest

ambry's Introduction

Intro

Ambry is your personal audiobook shelf. Upload your books to a self hosted server and stream to any device over the web.

Running the server

NOTE: This README reflects the main branch of this project and is not necessarily accurate to the latest stable release.

The easiest way to get up and running quickly is to use the container image which is hosted on the GitHub container registry: https://github.com/doughsay/ambry/pkgs/container/ambry

The only external requirement is a PostgreSQL database.

You can optionally also supply a headless FireFox instance running marionette for web-scraping to import metadata from external sources such as GoodReads.

Compose example

Here is an example Docker Compose file that could be used to run Ambry and the required PostgreSQL database:

---
version: "3"
services:
  postgres:
    image: postgres:alpine
    container_name: postgres
    environment:
      - POSTGRES_HOST_AUTH_METHOD=trust
    volumes:
      - pgdata:/var/lib/postgresql/data
    restart: unless-stopped

  firefox:
    image: ghcr.io/ambry-app/firefox-headless-marionette:latest
    container_name: firefox
    restart: unless-stopped

  ambry:
    image: ghcr.io/ambry-app/ambry:latest
    container_name: ambry
    environment:
      - DATABASE_URL=postgres://postgres:postgres@postgres/postgres
      - MARIONETTE_URL=tcp://firefox:2828
      - SECRET_KEY_BASE=FpmsgoGanxtwT6/M9/LbP2vFQP70dVqz2G/lC23lzOo2cmGkl82lW18Q01Av3RGV
      - BASE_URL=http://localhost:9000
      - PORT=9000
    ports:
      - 9000:9000
    volumes:
      - uploads:/app/uploads
    restart: unless-stopped

volumes:
  pgdata:
  uploads:

WARNING: The secret key above is only an example, do not use a publicly available key!

Configuration

The following environment variables are used for configuration:

Variable	Description	Default	Required?
`BASE_URL`	The url at which you will be serving Ambry. e.g. `https://ambry.mydomain.com`	N/A	Yes
`DATABASE_URL`	A postgresql URL. e.g. `postgresql://username:password@host/database_name`	N/A	Yes
`SECRET_KEY_BASE`	A secret key string of at least 64 bytes, used for signing secrets like session cookies.	N/A	Yes
`PORT`	The port you wish the server to listen on.	`80`	No
`POOL_SIZE`	The number of postgresql database connections to open.	`10`	No
`USER_REGISTRATION_ENABLED`	Whether or not users are allowed to register themselves with the server.	`no`	No
`MAIL_PROVIDER`	Valid values: mailjet	not-set	No
`MAIL_FROM_ADDRESS`	The email address that transactional emails are sent from	`noreply@<HOST>`	No
`MARIONETTE_URL`	A tcp URL to a marionette enabled FireFox. e.g. `tcp://hostname:2828`	not-set	No

Based on which mail provider you choose, you will need to supply provider specific configuration:

Mailjet

Variable	Description
`MAILJET_API_KEY`	The API key provided to you by Mailjet
`MAILJET_SECRET`	The API secret provided to you by Mailjet

The mail provider is only used for sending registration emails and forgotten password emails. If you don't need or want this functionality, you can just leave the MAIL_PROVIDER variable unset.

Setting up a fully working email provider requires a domain name that you control that you can configure correctly with your chosen provider. Currently the only provider that's working with Ambry is Mailjet, but if there's interest in others, they can be very easily added.

First time setup

The first time Ambry is booted up, it will walk you through setting up your initial admin user account. Just visit the URL at which you're hosting Ambry to get started:

http(s)://your-ambry-domain/

Once that's done, the server will restart and you can log into your new account and get started using Ambry!

Local development

Ambry is a Phoenix LiveView application, so to run the server on your machine for local development follow standard steps for phoenix applications. To be able to transcode audio files, you'll also need ffmpeg and shaka-packager available in your path.

Requirements

A PostgreSQL server running on localhost (you can customize the details in ./config/dev.exs)
Elixir and Erlang/OTP installed
FFmpeg installed
shaka-packager installed
(optional) FireFox running in headless mode with marionette enabled

For Elixir/Erlang you can easily install all the right versions using asdf by running asdf install from within the root directory. The versions are defined in .tool-versions.

# download hex and npm dependencies
mix deps.get
mix npm_deps.get

# create and migrate the database
mix ecto.setup

# run the server
iex -S mix phx.server

Seeds & example files

To add some example books and media to your local database, you can run:

mix seed

This will populate the database with some example books & media, and download & extract the example books & media files into your local uploads/ folder.

NOTE: The example files omit the source files for a smaller download-size, so visiting the admin audit page (http://localhost:4000/admin/audit) will show them all as missing their sources.

ambry's People

Contributors

Stargazers

Watchers

Forkers

froseph oasis256 honeyankit lubien kianmeng mohammedzeglam-pg rxwatcher kerbymart

ambry's Issues

Improve empty homepage experience

When starting the server for the first time, the homepage is empty. Not sure what to put there, but doing something seems better than nothing...

Add bookmarks feature

There's already an icon in the drop-down-player. Clicking it should open a modal or something similar with an interface for creating, deleting and seeking to bookmarks.

Rebuild bookmarks feature

Allow state deletion / abandon book

A user should be allowed to "abandon" a book. i.e. delete a given player-state.

If they started listening to a book and for whatever reason don't want to continue and don't care about losing their saved progress, they can just request to delete the player-state.

Not sure where to put the button though...

Support chapters

This is a bit of an open-ended or "hard" problem to solve, since there's no guarantee any chapter information will be present in any uploaded source files.

sometimes audiobooks are split into mp3 files by chapter
sometimes m4b files contain embedded chapter metadata
sometimes no chapter information is present

So the problem is two fold:

If chapter information is present in some way and we can reliably know that it is present, then we need to implement a data-model to store the information
Once we do have chapter information, we need to build a UI to view/select chapters, and to change the "next"/"previous" buttons to allow skipping to chapter boundaries

Additionally, it would then also be nice to improve the sleep-timer feature to allow for "play until end of chapter".

Increase test coverage

Tests are still a bit lacking, would be nice to round it out.

Allow media file replacement

If you make a mistake (e.g. forget a file in a multi file upload), or upload corrupted files or something, it would be nice if you could just replace a media's files and re-process them.

Caveats: if a user has already started listening to it, it might cause issues for them.

Add PubSub to admin media page

It would help if the admin media page auto-updated when background processing jobs change state.

Improve media upload status feedback

Currently when uploading media, there's no indication if the processor is working or has crashed.

Add navigation links to admin sections

If you're an admin user, there is currently no way to reach the admin section without manually typing in the URL. Would be nice to have a way to get there other than that.

Update readme with new logo

Show more book covers in series tiles

Series tiles currently only show the cover of the first book in the series. It would be kinda cool if it maybe showed the first 3, in some kind of nice fanout or something.

Implement mp4 uploader/processor

mp4, m4a and m4b files are currently not supported.

When updating a media, the status resets to `pending`

When updating media from the form, don't allow updating the status field at all.

Consider removing surface in favor of vanilla LiveView 0.17

How much work will this be?
What are the benefits / drawbacks?

Processing a book longer than 24 hours breaks the progress tracker

The progress tracker will crash when parsing timestamps over 24 hours because it's using Time.from_iso8601!/2, which is assuming a "time of day" not an interval of time.

This also apparently stops the processing altogether, which is a separate issue: #402

Allow replacing media

Allow uploading an entirely new media source that replaces the current one. This might cause downstream issues with users who have already started (or finished) listening to the media, but we can just add a warning that this is a concern but still allow you to do it.

Add tagging for books

Allow for tagging of books so we can do rough categorization with things like genres.

Allow specifying series when creating/editing a book

For convenience, it would be nice to allow specifying zero-or-more series to add a book to on the new books / edit book form.

Add typespecs to all public functions

MP4 concat re-encode processor

Sometimes, mp4s are just busted. The MP4 concat processor should support re-encoding the mp4s while it's concating them in this case.

GraphQL returns media that is still processing

The GraphQL API returns media even though it's in the middle of being processed, and if you try and resolve the duration it crashes because the duration is nil.

Add CI (github actions)

check formatted
enforce no compile warnings
enforce no credo warnings
dialyzer maybe?

(no tests yet since there are none... 😭)

Build a users admin page

Features:

Create - emails temporary password to new user
Delete - also signs out
Invalidate session - signs user out
Flip admin bit
List sessions per user

Configure swoosh using env vars

Enable email sending by exposing env vars to properly configure swoosh.

Chapterized mp4 chapter strategy

Chapter extraction strategy for individual chapterized mp4 files.

Loading book into player sometimes doesn't replace metadata and artwork

Sometimes when starting a book (loading into the player) it will load the audio, but the artwork and metadata will be for the previous book.

Disable form save button while uploading media

Add abridged & full cast inputs on media form

We have fields for abridged and full_cast but there are no inputs present for them.

Add some caching to CI

Caching is really tricky to get right with Elixir + GitHub actions, that's why it hasn't been added yet. But at some point it should be added...

Doesn't work on iOS (any browser) or Safari on macOS

This is because iOS and Safari do not support DASH, they only support HLS. Ambry is built around Dash.js, which only supports DASH, and also, Ambry only packages uploaded books into DASH streams.

So to fix this requires two major changes:

Use something other than Dash.js (e.g. Google's shaka player)
When uploading media files, convert them to two streams: one DASH and one HLS. (shaka packager already supports this)

Allow deleting chapters

Add tests

At least some tests, we don't need 100% out the gate

Explore media sessions api

Possibly implementing the Media Sessions API can improve the mobile experience on Android.

Update README for 0.4.0 release

Indicate which media has chapters in admin media list

When the websocket connection is lost and re-connects, it stops playback

Would be much nicer if it seamlessly re-connected and just kept playing.

Implement a "full-screen" player UI

Instead of just providing the drop-down player, it might be nice (at least on mobile devices), to have a "full-screen" player.

Update Favicon

Add better filter options to admin list pages

As the list of books/media grows, it would be nice to be able to filter the list with more than just a simple search
- filter by series
- filter by author
- filter by narrator
- status

Add sleep timer feature

The ability to set a timer for when to stop playing is a common audiobook player feature.

Be less strict on counting when books are finished

Right now, you have to listen to within the last 10 seconds of the book for it to count as having been finished. This is usually within the closing credits and people stop listening before then, so a lot of users have books that are technically still in-progress, even though they "finished".

Control user sign up via env var

Introduce new env var called USER_SIGN_UP_MODE
Allow 3 possible values:
- setup - First time setup. The first user that signs up is automatically made an admin.
- enabled - Regular user sign-up is enabled and there are links to the sign-up page
- disabled - User sign-up is disabled and there are no links to the sign-up page

List views that should be affected:

Library recent books list
Series books list
Search results list

Lists that need some consideration:

These lists are embedded in pages with multiple lists and don't really have a good "scroll bottom".

Continue Listening
Authored / Narrated books