This is a Telegram bot that tests your ability to understand spoken language.

Intended users are constructed languages enthusiasts who rarely have an opportunity to hear their conlangs spoken.

The bot will send you voice messages and await your transcription. If the transcription is correct, the bot will give you experience points.

Disclaimer: My second bot, there's probably massive antipatterns everywhere.

Let's say you want to deploy an instance for Newspeak.

Do the following steps.

Talk to BotFather and make a new bot. Bot's name can be anything and is not important, we really need the token.

Download or clone this repository to your server.

Create file language-newspeak.xml inside botvars/ folder.

There are examples of XML files for Elefen, Esperanto and Interslavic.

<alphabet> section should list letters that are worth experience points (case of letters is not important). For Newspeak it will just contain latin.

<transform> section is needed if your language has several alphabets or alternative spellings for letters, and it explains how to map those alternative letters to letters in <alphabet>. For Newspeak this section will be empty.

Next step is to translate bot's messages from (my bad) English to Newspeak.

First, generate .po file:

pygettext3 -o newspeak.po code/

Open it and write translations. See examples in elefen.po.

After that, convert it to .mo format which the bot can find and read:

mkdir -p locales/newspeak/LC_MESSAGES/
msgfmt -o locales/newspeak/LC_MESSAGES/cmpdbot.mo newspeak.po

What matters here is the resulting cmpdbot.mo file in a correct directory.

Now inside botvars/ folder create a .env file. Actual name is not super important - it can be newspeak.env, bot.env, etc. The examples below are assuming you called it just .env.

cp -n botvars/example-.env botvars/.env
chmod 0600 botvars/.env

Edit the variables inside it:

There are probably better, more efficient ways to do this. Anyway...

Create environment variable DOTENV_FILE holding your .env file name - just .env here in the examples.

export DOTENV_FILE=.env

Now invoke docker-compose up:

docker-compose --env-file botvars/$DOTENV_FILE up --scale manage=0 --build --remove-orphans

It will produce a lot of output while downloading and starting everything.

You know that it (probably) worked when you see INFO:aiogram.dispatcher.dispatcher:Start polling.. Now stop it with Ctrl+C and relaunch everything in a background with -d.

docker-compose --env-file botvars/$DOTENV_FILE up --scale manage=0 --build --remove-orphans -d

Next step is adding some initial Newspeak phrases to the bot's database.

Inside manage container there's a command for it: botaddphrases. It takes a file where each line contains one phrase and adds them all in one go.

Create file phrases.txt inside the Docker volume folder (CMPDBOT_EXCHANGE_DIR_LOCAL):

cat > phrases.txt <<1984
War is peace.
Freedom is slavery.
Ignorance is strength.
1984

Now launch the manage container:

docker-compose --env-file botvars/$DOTENV_FILE run --rm manage bash

And inside the container invoke the command:

botaddphrases $CMPDBOT_EXCHANGE_DIR_CONTAINER/phrases.txt

Sometimes the command will complain about similar phrases that are already in the database. Type y if you wish to insert the phrase nonetheless.

Almost there. Now it's time to add voice recordings.

It's easy to do through the bot itself. Start conversation with it in Telegram and it will send you voice recording tasks.

Pay attention though that if you've set CHOOSE_MIN_XP_VOICE to anything greater than 0, the bot will send you nothing because your XP is 0.

To bypass it you can cheat your XP in database:

docker exec -ti $( docker ps -f "ancestor=postgres" -q ) bash
psql -U $POSTGRES_USER
UPDATE person SET xp = 9000 WHERE id = <YOUR ID>;

Or you can temporarily set CHOOSE_MIN_XP_VOICE to 0 and restart the bot:

docker-compose restart

Let me know about your instance so I could update my table of existing bots. Thank you.

The bot operates with the following entities each stored in their own table in Postgres:

To create data backups do pg_dump for Postgres and mc mirror for S3 voices bucket.

The meaning of is_active column:

Setting is_active = false for a user doesn't block already existing active phrases and voices. If desired, they should be blocked individually or by issuing a command botblockuser <PERSON_ID> inside manage container that blocks user's all existing content.

Things that also should be added soon

• Testing
• Easy monitoring
• Admin site
• In-bot reporting
• Show best XP
• Show generated XP
• Async Sqlalchemy
• In-bot submissions management (delete user data)
• Delete old/blocked content
• Sql indicies
• In-bot management