Porte42 - Hall Voice

Bonjour la porte

Requirements

Studs

• Buy Hall Voice Change on intra shop
• Make a Pull Request to add your files
  • Include your login in the PR name
  • Add a .json file with your configuration (Also see custom/_Example.json)
  • No sound during more than 10 seconds
  • No more than 20 sounds
  • No sound over 1 Mo
  • No troll, insults, too loud sound, etc

PR might be refused if one of those requirements are missing

You can find an example PR here

Staff

• Buy one RPI and a pretty good audio speaker
• In Raspbian, clone this repository
• Goto the repository and type
  • pip3 install -r requirements.txt
• Install a Redis server
  • sudo apt-get install redis
• How to run?
  • You can use systemctl
    • sudo cp hallvoice.service /etc/systemd/system
    • sudo systemctl enable --now hallvoice.service
  • You can use Docker (WIP DO NOT USE ME, OR THE RPI WILL BURN)
    • docker compose build --no-cache
    • docker compose up -d
    • To look at logs: docker compose logs -f
• Copy and fill-up the config file in
  • cp config/config.ini.example config/config.ini
• Kafka:
  • The RPI must be in bocal's network so it can access Redpanda
  • You messages must be JSONised like so
    • {"building":"<where>","firstname":"<name>","kind":"<in/out>","login":"<login>"}
• You can pre-fill the redis cache with
  • goinfre-firstname-redis.py
    • to get all firstname from list of logins in login.list
  • goinfre-tts-redis.py
    • to cache all gTTS from a list of firstname in firstname.list
      • This can take a very long time, generate one tts per second to avoid Google rate-limit
• All data stored in redis have a TTL of 6 months

QnA

Studs

• Q: Will the hallvoice use Usualname if I have one?
  • A: Yes! Hallvoice use 42API to get your Usualname from your login, if there is no Usualname find, legal firstname will be used and cached
• Q: I juste set a Usualname but the old legal firstname is still in use, how to "force" the hallvoice to use the my Usualname?
  • A: Firstname is cached in Redis, you need to ask the bocal in Slack #42paris_staff_it or Discord #staff_it to purge it.
• Q: I bought hallvoice, how to use it?
  • A: As written in Requirements: You can find an example PR here, also see custom/_Example.json
• Q: I want to change a song, do I have to buy hallvoice again?
  • A: No
• Q: What's the TTL (Time to live) for the cache?
  • A: TTL for redis cache is 6 month, the staff can change it in the config file
• Q: Code is garbage, can I improve it?
  • I guess... Yes you can? You can try to do a PR

Staff

• Q: I am the bocal, how to purge cache from a login?
  • A: From bocal's network, make GET call at http://<rpi-IP>:8000/?purge=<login>
• Q: How can I query cache ?
  • A: From bocal's network, make GET call at http://<rpi-IP>:8000/?get=<login>
• Q: Something's f*cky, need to debug!
  • A: Logs can be found in folder logs/<date>-hallvoice.log
• Q: How to update?
  • A: Hmmm, Are you pisciner? A simple git pull should do the trick :)

Fun fact

• When redis cache is fully pre-filled:
  • it consume ~300M of disk space
  • 39888 TTS cached (2493 firstname * 16 welcome and goodbye message)
  • ~5300 login is used to cache firstnames
• Volume is at maximum in alsamixer

Change logs

• Hallvoice V2
  • Now works with kafka
• Hallvoice V3
  • Still in python but now code is object oriented
  • Redis cache to store the firstname from 42API
  • Redis cache to store
  • Norm and sanitize JSON files
• Hallvoice V3.1
  • Dockerized!
• Hallvoice V3.2
  • Improvement in dockerization!
  • Added get param in PurgeAPI
  • PurgeAPI code Improvement
• Hallvoice V3.2.1
  • WIP, Token authentification for PurgeAPI
• Hallvoice V3.3
  • Removed unused files
  • Moved CI scripts to dedicated folder
  • Moved conf file to folder
  • Updated some path in conf files
• Hallvoice V3.3.1
  • Updated volumes for redis container
• Hallvoice V3.3.2
  • Added the very most important folders in hallvoice container
  • Custom path if run in docker mode
• Hallvoice V3.3.3
  • Path for custom MP3 and JSON is now in config file
  • Custom MP3 played is now logged
  • Added print() and TTS for helping debugging
  • Added one Welcome and Goodbye message
  • docker-compose.yaml
    • added depends_on for hallvoice and purgeapi containers
    • Explicit read-only on folder: config, mp3, custom
    • Forcing TZ
    • Ports for purgeAPI
  • More print() and TTS for debugging
• Hallvoice V3.3.4
  • Moved and modified print()

Made with love by 42Paris team SI

Soon, Hallvoice V4 in Golang!