# Requirements

* S3 (minio can be used for a local instance).
* Redis
* Postgresql
* Python3.6

# Batch process

To remove the initial roundtrip when looking for books, I keep a cache in postgres.  
To populate this cache, I triggered each of the bot's `LIST` command, at `#ebooks`

I executed:

```
@pondering42
@dv8
@shytot
@dragnbreaker
@Xon-new
```

Then fetched all of the files, unarchived and concatenated them together.
On the resulting file I ran `parse.py` which inserts it into postgres.

# Services

## Installation

```
sudo cp services/bookworm@.service /etc/systemd/system/
sudo systemctl daemon-reload
sudo systemctl start bookworm@ircclient.service
sudo systemctl start bookworm@file_fetcher.service
sudo systemctl start bookworm@unpacker.service
sudo systemctl start bookworm@web.service
```

## IRC Client
In charge of processing received commands and sending them via IRC.  
Will forward details from DCC to the file fetcher via redis.

## File fetcher
In charge of fetching files specified via DCC.  
Will store the file in S3 and notify file details to the unpacker.

## Unpacker
In charge of taking a file (as provided from IRC), unpacking it and, if necessary, converting it to `mobi`.

## Web

### API

* Lists jobs in progress (`/books/status`)
* Serves available books (Anything in the S3 bucket)
* Exposes a search endpont `/book/search?terms=...`
* Fetches books via `/book/fetch/`


### Web interface
--

### Basic (Kindle) Web interface
The kindle has a very basic webbrowser (I believe it renders up to HTML4, CSS2.1), which can be used to download (available) books directly.  
At the moment I re-route at nginx based on user agent

To test, you can use `dillo` which has a set of features similar to the kindle's web browser.

```
if ($http_user_agent ~* "armv7l") {
    rewrite ^/(.*)$ /books/kindle;
}
```

# CLI
To be re-implemented