[stefanodangelo]

Adds Control, the first tab of the Print Hub group from #184. Print Hub was hidden until a feature gave it a child, so this is the PR that makes it appear.

Nothing here is new functionality. The camera, the console and the print controls already exist in the printer's Fluidd/Moonraker interface. Control only routes them into the multiACE page, so everything is handy from one place. It replaces nothing in Fluidd.

What it adds

• Webcam. GET /api/webcam/info, /stream (MJPEG) and /snapshot, plus POST /api/webcam/webrtc-signal for WebRTC cameras such as camera-streamer. The camera is looked up through Moonraker, and our own Fluidd iframe entry is skipped. The browser never sees the camera's address: media goes browser-to-camera, and only the SDP handshake goes through us, because camera-streamer sends no CORS headers. The lookup retries on its own while no camera is found, and a connection that comes up without delivering video fails visibly and retries.
• Print panel. Speed, flow and fan sliders, nozzle and bed targets, Z-offset steps, Pause / Resume / Cancel (Cancel asks first). Values are read back from the printer every 2 s while the tab is open.
• Chamber light. A switch for led cavity_led, shown only when the printer has that LED. While a print is running or paused with defect detection enabled, the light stays on: stock turns it on itself because the AI camera check needs it, so the server refuses dimming or switching it off then, and the switch is disabled with the reason.
• Chamber temperature and enclosure. The chamber temperature comes from the cavity sensor and reads without an enclosure. With an enclosure attached (power_detected) the tab also shows the active mode and its target temperature, two sliders for the exhaust and inner fan, and "Fans off". See below.
• Console. Read-only, incremental via since_id, using the Moonraker websocket the backend already listens to, so there is no extra polling. History is loaded once from Moonraker's gcode store.
• Restart Firmware / Restart Printer. POST /api/restart with one of two fixed behaviours. See below.

Safety. POST /api/print-control takes a fixed verb and a number, never a G-code string. Values are clamped on the server:

• speed 25–300 %, flow 75–125 %, fan, light and the enclosure fans 0–100 %;
• heaters are capped by Klipper's own max_temp (read from its config) and nothing tighter: stock loads and unloads above a material's printing range, so a per-material cap would stop someone heating up to clear a clog. A target below a printing range is not raised, and 0 means off;
• Z offset in steps of at most 0.05 mm, with a ±0.5 mm total per print that resets when a print starts;
• the status poll is homing-gated like _query_state_gated, and a command that times out on its way back is reported as queued, with the matching control kept disabled, so a swap that holds Klipper's queue cannot make a second press send it twice.

Unknown verbs return 400 and send nothing. The nozzle follows the active extruder. When Moonraker rejects a command, the panel shows its error message, not the traceback that comes with it.

Restarts are refused while a print is running or paused (read fresh from Moonraker) and under MULTIACE_MANAGED, where the buttons are hidden. A downed Klipper (503) does not block a firmware restart, which is when it is needed; an unreachable Moonraker does. firmware_restart is what the config save already sends.

Restart Printer asks Moonraker (/machine/reboot) first. That runs sudo systemctl reboot, and the Snapmaker U1 has neither sudo nor systemd (busybox init), so it always fails there. If it fails, and only when the backend runs as root on the same machine as Moonraker, the backend runs reboot itself. A backend on a dev box talking to a remote printer, one that is not root, or one on Windows never takes that path, and the firmware restart never does. reboot is looked up in /usr/sbin as well, because the service's own PATH does not include it.

Enclosure controls use the stock commands: SET_PURIFIER FAN=exhaust|inner SPEED=0..1 and SET_PURIFIER_MODE MODE=0 EXHAUST_FAN_DELAY_OFF=0 INNER_FAN_DELAY_OFF=0 for "Fans off". The server refuses the fan speeds unless an enclosure is attached, no print is running and no mode is running, because stock drives the fans itself in modes 1–3 and sets the mode from the sliced file during a print. "Fans off" stays available in a mode as the way out. The modes themselves (cool, preheat, hot chamber) are not driven from here.

Navigation. One navItems entry with group: "printhub", control added to _validTabs, and ui.tabs.control plus ui.control.* and ui.restart.* in en, de and zh. The top bar and .subnav rules are untouched. Panel mode is unaffected.

Not included. A console input box, the enclosure modes, and Retry controls. Polling and the camera connection only run while the Control tab is showing.

Mock mode. Fixtures: mock_webcam.json (shows "camera unavailable"), mock_console.json, mock_print_control.json. The stream, snapshot and signalling endpoints and the restarts refuse or skip Moonraker in mock mode, and ?mock= still only works while mock mode is on.

Testing

• pytest tests/unit: 242 passed, 226 of them new (camera lookup, stream proxy, WebRTC relay, console buffer, every clamp and guard, restarts including the local reboot fallback, enclosure, light and defect detection).
• Tried on a real Snapmaker U1: live 1920×1080 WebRTC video in dark, light and classic, the page does not widen at 390 px, the light switch, the console and the readbacks. Restart Printer fails on it through Moonraker as described above; with the fallback, the new code run on the printer rebooted it and it came back with Klipper ready and multiACE intact.
• Not exercised on hardware: the enclosure controls (no enclosure), and the heater, fan, babystep and firmware-restart commands, which are covered by unit tests and mock mode.

Screenshots. See the comments.

[stefanodangelo]

Here are the screenshots:

[stefanodangelo]

If you like it, then I can also make the console send commands rather than be just read-only. I use it on my printer and it's pretty handy to have.
Part of that, I also have 2 buttons that send macros to the console to restart the firmware and the printer. Eventually we could potentially think of adding other buttons too.

[decay71]

Thanks, three points

1. Nozzle clamp: _heater_limits intersects the material ranges of every head with filament_detected, not just the active one (filament_detected is also true for an unloaded head with a spool at the ACE gate). PLA on one head and PETG on another gives 210-230 on both; PLA and PC inverts the range (240/230), so any nozzle target on the PLA head lands at 240. Please use only the toolhead whose idx matches the active extruder.

2. Homing gate: the 2 s status poll (_print_status) calls Moonraker directly. Please check _homing_active() first, like _query_state_gated does, and skip the poll or serve the last value while the flag is fresh. Polling during bed mesh/homing is what that gate prevents.

3. Swaps: /printer/gcode/script only returns once the command has run, and a multiACE swap holds the queue for 60-150 s. The 30 s timeout then reports 502 although the command is still queued and runs after the swap, so a second press sends it twice. Please treat a timeout as "queued" in the UI, or raise the timeout above a swap.

One question on top: would you like to add controls for the Snapmaker enclosure as well? Stock drives it through the purifier object (exhaust and inner fan, with its own attach detection), and the cavity temperature sensor gives the matching readout. A toggle for the chamber light (led cavity_led, part of the printer itself) would fit next to it and also fix your dark camera picture. Same pattern as the rest: fixed verbs, values clamped on the server.

[stefanodangelo]

Thanks for the review, I'll make those three changes and add the chamber light toggle.
However, I don't have the enclosure, so I have no way to test the purifier fan controls or the cavity temperature readout.

[decay71]

Thank you, i will test that here!

[stefanodangelo]

Pushed the fixes.

Here are the details:

• Nozzle clamp: with PLA on head 0 and PC on head 1, a target of 200 now stays 200 instead of landing on 240.
• Homing gate: besides the 2 s poll, I also cached the Klipper config query (5 min) and skip it during homing, because it hit Moonraker on every nozzle or bed call.
• Swaps: a connect timeout is still reported as a failure, since nothing was queued then. Only a read timeout counts as "queued". While something is queued, the matching control (Pause, Resume, Cancel, Z steps) stays disabled for 3 minutes.
• Chamber light: it's led cavity_led, driven with SET_LED LED=cavity_led WHITE=<0..1>. I tested it on my printer, and the camera picture is no longer black.
• Enclosure: I don't have the enclosure, so I could only verify what the printer itself reports. I added the chamber temperature and, only when power_detected is true, the exhaust and inner fan speeds. The one control is "Fans off", which sends SET_PURIFIER_MODE MODE=0, the command stock sends in PRINT_END and CANCEL_PRINT. I couldn't find a command for turning the fans on or setting a speed, so I didn't guess one. If you tell me the stock command and its mode values, I'll add it, but someone with the enclosure would have to test it.
• Restart buttons: I also added Restart Firmware and Restart Printer, as in the plan. They're refused while a print is running or paused, and hidden under MULTIACE_MANAGED. A downed Klipper doesn't block a firmware restart, since that's when it's needed. I know index.html has comments saying a restart from the UI "worked only sometimes" and that Fluidd can do it. If you'd rather not have them here, they're one commit and easy to drop.

Other fixes found while testing on the printer:

• The console history only loaded when the buffer was empty, so it never loaded once any live line had arrived.
• A WebRTC connection that came up but delivered no video now shows an error and retries.
• The camera lookup retries on its own if it fails once at load.

Here are the screenshots from my printer:

[decay71]

Thanks, this reads well and the safety design (fixed verbs, server-side clamps, print-state gate on restarts) is exactly right. Three things:

1. Heater clamp raises low targets. A nozzle target below the material range is lifted to the floor: PLA 150 becomes 170, PC 200 becomes 240. Someone setting a standby temperature gets a hotter nozzle than they typed. Please cap only the top (max_temp and the material maximum) and let values below the material floor through, the same way 0 already passes as "off".

2. MATERIAL_TEMP_LIMITS only knows base names. PLA-CF, PETG-CF, PETG-HF and similar find no entry and fall back to max_temp alone. That is safe, just inconsistent. Either match on the base material before the first "-" or leave it as is and say so in the comment.

3. The babystep allowance resets only when the filename changes. Printing the same file again keeps the used total from the previous job. Resetting it on print start would fit better than keying on the filename.

The restart buttons can stay: firmware_restart is what our own config save sends, the warning in index.html is about the bare Klipper restart, which you do not use.

On your enclosure question, this is what stock purifier.py (2.0.0) offers, take whatever you think fits:

• SET_PURIFIER FAN=exhaust|inner SPEED=<0..1> sets one fan directly. SPEED=0 starts the delayed turn-off (default 180 s) rather than stopping at once. Speed > 0 is refused without an attached enclosure.
• SET_PURIFIER_MODE MODE=0..3: 0 idle, 1 cool chamber (exhaust, FAN_SPEED default 0.6, DESIRE_TEMP default 42, ALARM_TEMP), 2 preheat chamber (inner fan, waits for DESIRE_TEMP), 3 hot chamber (inner fan, exhaust off). get_status reports the active mode and desired_chamber_temp, so the panel could show what is running.
• Your current "Fans off" (MODE=0) stops the fans only after that delay. MODE=0 EXHAUST_FAN_DELAY_OFF=0 INNER_FAN_DELAY_OFF=0 stops them now.

Things worth keeping in mind if you go further: in modes 1 to 3 stock drives the fans itself (dynamically in cool mode), so a direct speed control would compete with it; and during a print the mode usually comes from the sliced file, which is also what a power-loss resume restores. Direct fan control in idle and mode buttons outside a print would be one way, but it is your call.

I will test the enclosure readout, the fan controls and the chamber light here after the merge.

[stefanodangelo]

Pushed two more commits.

The first one targets your 3 points. The nozzle target is now only capped at the top, so lower targets pass through like 0 does. The material ceiling matches on the base name, so PLA-CF and PETG-HF get their base material's ceiling. The babystep allowance resets when a print starts, and pause then resume is still the same print.

The second and last one instead features the following:

• "Fans off" now stops both fans at once (MODE=0 with both delays at 0).
• Two sliders set the exhaust and inner fan directly with SET_PURIFIER. The server refuses them unless an enclosure is attached, no print is running and no mode is running. The sliders are disabled and say why, and "Fans off" stays available as the way out of a mode.
• The tab shows the active mode and its target temperature.

I left the modes themselves (cool, preheat, hot chamber) out, since they take extra parameters and stock sets them during a print. I couldn't run any of the enclosure part without an enclosure, so it's covered by unit tests and mock mode only. I'd appreciate your test on yours.

I also updated the PR description to reflect the new entries. Mock-mode screenshots of the enclosure sliders are below. I couldn't take screenshots from the real frontend because I have no enclosure.

In Idle:

In Cool mode:

During a print:

[decay71]

Thanks, the three fixes are in and the enclosure commands match stock purifier.py. Three notes:

1. Light off during a print: stock turns cavity_led on itself at DEFECT_DETECTION_START because the AI camera check needs it. Switching it off from the tab during a print with defect detection enabled leaves the camera dark, so the detection can misfire or go blind. Please refuse (or at least warn) light-off while printing or paused with defect detection on.
2. One more on the material table from my last review: the ceilings are stricter than stock. PLA stops at 230 here, while the Snapmaker filament database gives PLA a load temperature of 250 and stock loads and unloads at that, so heating up to clear a clog is not possible from the tab. Please cap at max_temp only, or use more generous ceilings.
3. Restart Printer fails on my U1: /machine/reboot answers "500 Error running shell command: 'sudo systemctl reboot'". Did you try the button on your printer?

[stefanodangelo]

Pushed four more commits for your three notes.

Light during a print (22e9137): the server now refuses dimming or switching off the chamber light while a print is running or paused and defect detection is enabled (the defect_detection master switch plus at least one of its checks). Keeping it fully on is still allowed, and without detection, outside a print, or without that object nothing changes. The tab reports a detection flag, and the switch is disabled with the reason.

Ceilings (b71af85): the nozzle is now capped at Klipper's max_temp only. I removed the material table, and with it the base-name matching from last round, which only existed to feed it.

Restart Printer (7e1d0f2): you were right, and I hadn't tried it on a printer before. I did now, and it fails on mine with the same sudo systemctl reboot error. The U1 ships without sudo and without systemctl (busybox init), so Moonraker's /machine/reboot can never work there.

It now tries Moonraker first, so printers where that works are unchanged. Only if it fails, and only when the backend runs as root on the same machine as Moonraker, it runs reboot itself. A backend on a dev box talking to a remote printer, or one that isn't root, never takes that path. The print, managed-install and mock-mode guards still apply first, and the firmware restart never takes it. reboot lives in /usr/sbin, which the service's own PATH doesn't include, so it is looked up there as well. I only found that by running it on the printer.

I ran the new code on my U1 from a scratch copy, without touching the installed service. The printer went down and came back with Klipper ready and multiACE intact (uptime reset from 20742 s to 48 s). I haven't been able to try it on yours, so it relies on your web service running as root, which is the default.

9f174d4: a failed command used to put Moonraker's whole traceback into the panel. It now shows only Moonraker's error message.

[decay71]

Merged, thanks for the careful work and for testing it on your U1. I will run the enclosure controls and the remaining commands on my printer and report back here if anything turns up.