pinakes/docs/README.md
NotAShelf c535650f45
docs: clarify wording around API split
Signed-off-by: NotAShelf <raf@notashelf.dev>
Change-Id: I864882c5a22424cd6c8c320627fe9f006a6a6964
2026-03-06 18:29:34 +03:00

185 lines
7.1 KiB
Markdown

# Pinakes
[first known library cataloging system]: https://en.wikipedia.org/wiki/Pinakes
Pinakes, named after the [first known library cataloging system] designed to be
the _last library cataloging system you will ever need_. Pinakes indexes files
across configured directories, extracts metadata from audio, video, document and
text files, and provides full-text search with tagging, collections, roles,
audit logging and more. It supports both SQlite (for easy bootstrapping) and
PostgreSQL (production deployments) as available database backends.
## Building
```bash
# Build all compilable crates
$ cargo build -p pinakes-core -p pinakes-server -p pinakes-tui
# The Dioxus UI requires GTK3 and libsoup system libraries:
# On Debian/Ubuntu: apt install libgtk-3-dev libsoup-3.0-dev libwebkit2gtk-4.1-dev
# On Fedora: dnf install gtk3-devel libsoup3-devel webkit2gtk4.1-devel
# On Nix: Use the dev shell, everything is provided :)
$ cargo build -p pinakes-ui
# Alternatively, while app deps are in PATH, you may simply build the entire
# workspace.
$ cargo build --workspace
```
## Configuration
Pinakes runs with its own built-in configuration file out of the box. While
using the default configuration, you will not be able to edit the configuration
but it will provide the minimum required configuration values to get you going
with Pinakes. If you are more interested in fully configuring Pinakes, you must
create your own configuration. You may copy the example config and edit it to
your needs:
```bash
# Copy the sample config
$ cp pinakes.example.toml pinakes.toml
```
Key settings:
- `storage.backend` - `"sqlite"` or `"postgres"`
- `storage.sqlite.path` - Path to the SQLite database file
- `storage.postgres.*` - PostgreSQL connection parameters
- `directories.roots` - Directories to scan for media files
- `scanning.watch` - Enable filesystem watching for automatic imports
- `scanning.ignore_patterns` - Patterns to skip during scanning (e.g., `".*"`,
`"node_modules"`)
- `server.host` / `server.port` - Server bind address
## Running
### Server
To use Pinakes, you will need the server to be running. The GUI on its own will
work, but it will not be functional without the server.
```sh
# Start the server first
$ cargo run -p pinakes-server -- pinakes.toml
# or:
$ cargo run -p pinakes-server -- --config pinakes.toml
```
The server starts on the configured host:port (default `127.0.0.1:3000`). In a
production scenario you are encouraged to reverse proxy the service, and prefer
SSL.
### TUI
The Pinakes TUI can be used to manage your collections from the comfort of your
terminal. While the server is running you may connect to it using the `--server`
flag.
```bash
# Using defaults
$ cargo run -p pinakes-tui
# or with a custom server URL:
$ cargo run -p pinakes-tui -- --server http://localhost:3000
```
#### Keybindings
The TUI component of Pinakes is designed to be keyboard-centric, as it is
designed for the terminal. The keybindings are as follows:
<!-- markdownlint-disable MD013-->
| Key | Action |
| --------------------- | -------------------------------------------------------- |
| `q` / `Ctrl-C` | Quit |
| `j` / `k` | Navigate down / up |
| `Enter` | Select / confirm |
| `Esc` | Back |
| `/` | Search |
| `i` | Import file |
| `o` | Open file |
| `d` | Delete (media in library, tag/collection in their views) |
| `t` | Tags view |
| `c` | Collections view |
| `a` | Audit log view |
| `s` | Trigger scan |
| `r` | Refresh current view |
| `n` | Create new tag (in tags view) |
| `+` | Tag selected media (in detail view) |
| `-` | Untag selected media (in detail view) |
| `Tab` / `Shift-Tab` | Next / previous tab |
| `PageUp` / `PageDown` | Paginate |
<!-- markdownlint-enable MD013-->
### Desktop/Web UI
Pinakes features a fully fledged Desktop and Web UI powered by Dioxus. Those two
components are meant as a GUI frontend for the Pinakes server, and are
interchangeable in terms of usage.
```bash
# Build the UI
$ cargo run -p pinakes-ui
```
> [!TIP]
> By default Pinakes GUI will assume the server to be running on localhost and
> bound to port 3000. Set `PINAKES_SERVER_URL` to point at the server if it is
> not on `localhost:3000`.
### Extending Pinakes
While Pinakes _does_ aim to be as comprehensive as humanly possible, it is not
feasible to maintain all features in one gigantic repository without taking on
an immense technical debt. To avoid this kind of resource mismanagement while
still allowing for _all_ kinds of extension, Pinakes features two features at
your convenience:
- REST API Routes
- Plugin API
### REST API
[REST API documentation]: ./api.md
There exists a comprehensive UI for the server component that you may query
directly from the `/api/v1` endpoint. All other endpoints are under `/api/v1`.
The server API is, of course, a part of Pinakes core design but it is also how
first-party interfaces like `pinakes-ui` and `pinakes-tui` interact with the
server. You may write your own interfaces for a running Pinakes server with
minimal effort by simply sending requests to the API through your preferred
means. See [REST API documentation] on available routes and tips on how to
interact with the API.
### Plugin API
[Plugin API documentation]: ./plugins.md
The other method, which is by far the most powerful but also perhaps the least
polished as of writing is the **plugin system**. This is designed as a means of
implementing various user-facing features to Pinakes _server_ by writing your
own plugins that can modify certain elements. While this system is not as stable
as the server API, it is generally in good shape and example plugins are
provided. Please see the [Plugin API documentation] for more details, examples
and design.
## Storage Backends
Two storage backends are supported. For convenience, SQLite is the default
backend out of the box but for production deployments you may choose to prefer
PostgreSQL.
### **SQLite** (default)
Single-file database with WAL mode and FTS5 full-text search. Bundled SQLite
guarantees FTS5 availability.
### **PostgreSQL**
Native async with connection pooling (deadpool-postgres). Uses tsvector with
weighted columns for full-text search and pg_trgm for fuzzy matching. Requires
the `pg_trgm` extension.