New findings in the Viomi map format

[halfbakery]

I'm a Viomi V2 Pro (viomi.vacuum.v6) owner, and I created a Viomi map parser for https://github.com/PiotrMachowski/Home-Assistant-custom-components-Xiaomi-Cloud-Map-Extractor/tree/dev_new_api based on your excellent documentation. During the implementation I've found some new information which I'd like to share:

• The map definetly has a structure: as you realized, the map starts with the featureFlags word or dword, followed by the mapId field. It seems the mapId value is the section header magic. I don't know whether it's version specific or random (my findings are based on just some maps, but I've never ever seen the same value). All of the time I've found the featureFlags 0x78ff and the map exactly has 12 sections. The mapId at offset 4 is the magic of the robot status section.
• map section: these are the values I've encountered:
  • 0 / 255: the same as you found
  • 1: scanned but unreachable area
  • 2: I couldn't find its meaning, but found on many maps
  • [10 ... 59]: room_id
  • [60 ... 109]: rooms selected/under cleanup, room_id = value - 50
• history section: the first byte is taxiing (0x00) or working (0x01). I've tested it in both vacuuming or vacuuming+mopping mode.
• room section: in each room entry (id, name, unknown) the unknown 8 bytes is a coordinate pair (two float32) of the room center/room name position on the map.

[Hypfer]

Great!

It would be awesome if you could update the kaitai struct with your findings:
https://github.com/Hypfer/Valetudo/blob/master/assets/kaitai_structs/viomi_map_file.ksy

For testing, just throw that + a sample map file into https://ide.kaitai.io/
You should then get a nice interactive dissection