feat: check and install firmware updates through Prusa Connect

[ynezz]

hi, my clankers found this nice project during research, I liked it so decided to make you aware about the changes I've made for my use cases, so everyone can profit from that, but feel free to close it or implement it differently, happy to help with testing, cheers, Petr

---

Without a USB stick and a browser, prusactl could not tell which firmware the printer runs, and could not update it.

• prusactl firmware (MCP: get_firmware_status) shows the running firmware and the latest firmware that Prusa Connect offers for the model of the printer.

• prusactl firmware update [--version X.Y.Z] (MCP: update_firmware) does what the update button of the Connect web app does:

  1. Connect copies its hosted .bbf file to the USB drive of the printer through a download-queue entry.
  2. prusactl waits for the copy.
  3. It installs the firmware with the FLASH command. This restarts the printer.

  The command does nothing when the printer already runs the latest version and no --version is given.

• The printer accepts FLASH only when it is idle, ready, finished or stopped. During a print, the file stays on the drive and the result says so.

• A preparatory refactor makes defaultStorage prefer a USB drive and skip corrupted storages. It also gives the refused-state error a type. A table test covers both.

The API flow comes from the Connect web app. The firmware files of Prusa belong to team_id 0. path is the 8.3 short name on the drive, for example /usb/COREON~3.BBF.

Testing

• Unit tests with a mocked Connect: status, the up-to-date case that does nothing, copy then poll then FLASH, FLASH refused during a print (the file stays on the drive), and a timeout.
• Live on a CORE One L:
  • status (6.8.1 running, 7.0.0 offered)
  • copy of 6.5.7 to USB during a print, with FLASH correctly refused while PRINTING and the file left on the drive
  • end-to-end update from 6.8.1 to 7.0.0 with one prusactl firmware update after a print had finished. The copy to the USB drive took less than 1 minute. The printer installed the file, restarted and came back idle on 7.0.0+16903 about 4 minutes after FLASH. Connect then reported the firmware as up to date. Nobody had to touch the printer during the update.

[trevin-lee]

The flow is right and the mocked tests are good. Three bugs I'll fix:

• tools_firmware.go:250 sets installed: true regardless of res.State. A sync command the printer REJECTED comes back as installed, with a note saying it restarts. runCommand returns no error for a rejected sync command (I had the same bug in cmd send and fixed it in 0.2.1). installed will be false unless the event is FINISHED.
• upToDate() at :56 is string equality. A printer on 7.0.0-RC1 with Connect offering 6.8.1 reports update_available: true, and firmware update with no flags flashes 6.8.1. I'll decide on Connect's support.state and refuse a downgrade unless --version is explicit.
• The state check is before FLASH only, so the download is enqueued in any state. Mid-print it writes the .bbf to the USB, polls for 2 minutes, and exits 0. Moving the check before the queue POST, the way upload_file then=print does.

Smaller: require transferring == nil before FLASH, and "pass version" in the error becomes --version for the terminal reader. I'll also add one line to the tool description: install only when the user asked for it.

After the fixes, an end-to-end flash on your L would be the real test. Mine is on 6.8.1 with 7.0.0 offered, so I may run it here too.

[trevin-lee]

Merged in v0.3.0. Two things changed from what I said, after a second review of my own fixes:

• "Up to date" is decided only on support.state: outdated installs, supported does nothing, and anything else (or no state) refuses to decide and asks for --version. On my printer Connect reports outdated for 6.8.1 to 7.0.0, which matches. I haven't seen the string Connect sends for a printer that is current. Yours is on 7.0.0 now, so prusactl firmware --json would tell me whether it's supported or something else.
• installed is true only when the sync FLASH ends with FINISHED. REJECTED says the printer refused; CREATED or EMIT says it wasn't confirmed, rather than claiming either way.

One more thing only a live run can answer: the copy loop now accepts path only once transferring is gone, on the assumption that Connect never reports both at once. If your next update stalls with "made no progress on the download" while the file is clearly on the drive, that assumption is wrong and I'll relax it.