Support for Dreame vacuums

Vacuums confirmed to use this API:

• dreame.vacuum.p2008 (Dreame F9)
• dreame.vacuum.mc1808 (Xiaomi Mi Mop/Xiaomi Miija 1C)
• Other ones mentioned in comments below

You can help in implementation of this feature by sending your map file and a screenshot from Xiaomi app at

piotr.machowski.dev [at] gmail [dot] com.

To get raw map file you have to use following config:

camera:
  - platform: xiaomi_cloud_map_extractor
    host: !secret xiaomi_vacuum_host
    token: !secret xiaomi_vacuum_token
    username: !secret xiaomi_cloud_username
    password: !secret xiaomi_cloud_password
    store_map_raw: true
    force_api: dreame

and then follow steps provided in documentation: Retrieving map.

Existing implementation of map decoder

[Neonox31]

Hello 👋

I’ve got a dreame.vacuum.p2008 model and I’m currently working on transcribing Valetudo’s javascript code for map parsing of Dreame models.

I’m quite comfortable with Javascript, but I’m just discovering Python so it’s taking me a little while. Here is the actual map I can generate with my model : You can see floor, walls, charger and vaccum position as well as no go areas. I also implemented path parsing, but the dreame map contains several paths and the MapData can only contain one if I understand correctly. I tried to flatten the list but it causes unwanted interconnections between paths :

I have a lot of work to do and I will submit a P.R if you’re intesrested @PiotrMachowski

[PiotrMachowski]

Nice 👍 I think I can add support for multiple paths without any problems. You can create a PR, I will check it out and adjust if needed 😉