nixpkgs/nixos/tests/lomiri-camera-app.nix
OPNA2608 1f57618cae nixosTests.lomiri-camera-app: Fix OCR, round 4
We're struggling with OCRing gnome-text-editor now. Let's inspect the clipboard via simpler means.
2025-06-04 19:20:35 +02:00

258 lines
8.3 KiB
Nix

let
makeTest = import ./make-test-python.nix;
feedLabel = "Image";
feedQrContent = "Test";
feedImageFile = "feed.png";
makeFeedImage =
pkgs:
pkgs.runCommand feedImageFile
{
nativeBuildInputs = with pkgs; [
(imagemagick.override { ghostscriptSupport = true; }) # add label for OCR
qrtool # generate QR code
];
}
''
qrtool encode '${feedQrContent}' -s 20 -m 10 > qr.png
# Horizontal flip, add text, flip back. Camera displays image mirrored, so need reversed text for OCR
magick qr.png \
-flop \
-pointsize 30 -fill black -annotate +100+100 '${feedLabel}' \
-flop \
$out
'';
in
{
basic = makeTest (
{ lib, ... }:
{
name = "lomiri-camera-app-basic";
meta.maintainers = lib.teams.lomiri.members;
nodes.machine =
{ config, pkgs, ... }:
{
imports = [ ./common/x11.nix ];
services.xserver.enable = true;
environment = {
systemPackages = with pkgs.lomiri; [
suru-icon-theme
lomiri-camera-app
];
variables = {
UITK_ICON_THEME = "suru";
};
};
i18n.supportedLocales = [ "all" ];
fonts = {
packages = with pkgs; [
# Intended font & helps with OCR
ubuntu-classic
];
};
};
enableOCR = true;
testScript =
let
qrLabel = "Feed";
qrContent = "Test";
in
''
machine.wait_for_x()
with subtest("lomiri camera launches"):
machine.succeed("lomiri-camera-app >&2 &")
# emitted twice
machine.wait_for_console_text("updateViewfinderResolution: viewfinder resolutions is not known yet")
machine.wait_for_console_text("updateViewfinderResolution: viewfinder resolutions is not known yet")
machine.sleep(10)
machine.send_key("alt-f10")
machine.sleep(5)
machine.wait_for_text("Cannot access")
machine.screenshot("lomiri-camera_open")
machine.succeed("pgrep -afx lomiri-camera-app >&2")
machine.succeed("pkill -efx lomiri-camera-app >&2")
machine.wait_until_fails("pgrep -afx lomiri-camera-app >&2")
# Sometimes, GStreamer errors out on camera init with: CameraBin error: "Failed to allocate required memory."
# Adding more VM memory didn't affect this. Maybe flaky in general?
# Current assumption: Camera access gets requested in a weird/still-in-use state, so sleep abit
machine.sleep(10)
with subtest("lomiri camera localisation works"):
machine.succeed("env LANG=de_DE.UTF-8 lomiri-camera-app >&2 &")
# emitted twice
machine.wait_for_console_text("updateViewfinderResolution: viewfinder resolutions is not known yet")
machine.wait_for_console_text("updateViewfinderResolution: viewfinder resolutions is not known yet")
machine.sleep(10)
machine.send_key("alt-f10")
machine.sleep(5)
machine.wait_for_text("Zugriff auf")
machine.screenshot("lomiri-camera_localised")
'';
}
);
v4l2-photo = makeTest (
{ lib, ... }:
{
name = "lomiri-camera-app-v4l2-photo";
meta.maintainers = lib.teams.lomiri.members;
nodes.machine =
{ config, pkgs, ... }:
{
imports = [ ./common/x11.nix ];
services.xserver.enable = true;
environment = {
etc."${feedImageFile}".source = makeFeedImage pkgs;
systemPackages =
with pkgs;
[
feh # view photo result
ffmpeg # fake webcam stream
imagemagick # unflip webcam photo
xdotool # clicking on camera button
]
++ (with pkgs.lomiri; [
suru-icon-theme
lomiri-camera-app
]);
variables = {
UITK_ICON_THEME = "suru";
};
};
i18n.supportedLocales = [ "all" ];
fonts = {
packages = with pkgs; [
# Intended font & helps with OCR
ubuntu-classic
];
};
# Fake camera
boot.extraModulePackages = with config.boot.kernelPackages; [ v4l2loopback ];
};
enableOCR = true;
testScript = ''
machine.wait_for_x()
# Setup fake v4l2 camera
machine.succeed("modprobe v4l2loopback video_nr=10 card_label=Video-Loopback exclusive_caps=1")
machine.succeed("ffmpeg -re -loop 1 -i /etc/${feedImageFile} -vf format=yuv420p -f v4l2 /dev/video10 -loglevel fatal >&2 &")
with subtest("lomiri camera uses camera"):
machine.succeed("lomiri-camera-app >&2 &")
# emitted twice
machine.wait_for_console_text("No flash control support")
machine.wait_for_console_text("No flash control support")
machine.sleep(10)
machine.send_key("alt-f10")
machine.sleep(5)
machine.wait_for_text("${feedLabel}")
machine.screenshot("lomiri-camera_feed")
machine.succeed("xdotool mousemove 510 670 click 1") # take photo
machine.wait_until_succeeds("ls /root/Pictures/camera.ubports | grep '\\.jpg$'")
# Check that the image is correct
machine.send_key("ctrl-alt-right")
machine.succeed("magick /root/Pictures/camera.ubports/IMG_00000001.jpg -flop photo_flip.png")
machine.succeed("feh photo_flip.png >&2 &")
machine.sleep(10)
machine.send_key("alt-f10")
machine.sleep(5)
machine.wait_for_text("${feedLabel}")
machine.screenshot("lomiri-camera_photo")
'';
}
);
v4l2-qr = makeTest (
{ lib, ... }:
{
name = "lomiri-camera-app-v4l2-qr";
meta.maintainers = lib.teams.lomiri.members;
nodes.machine =
{ config, pkgs, ... }:
{
imports = [ ./common/x11.nix ];
services.xserver.enable = true;
environment = {
etc."${feedImageFile}".source = makeFeedImage pkgs;
systemPackages =
with pkgs;
[
ffmpeg # fake webcam stream
xclip # inspect QR contents copied into clipboard
xdotool # clicking on QR button
]
++ (with pkgs.lomiri; [
suru-icon-theme
lomiri-camera-app
]);
variables = {
UITK_ICON_THEME = "suru";
};
};
i18n.supportedLocales = [ "all" ];
fonts = {
packages = with pkgs; [
# Intended font & helps with OCR
ubuntu-classic
];
};
# Fake camera
boot.extraModulePackages = with config.boot.kernelPackages; [ v4l2loopback ];
};
enableOCR = true;
testScript = ''
machine.wait_for_x()
# Setup fake v4l2 camera
machine.succeed("modprobe v4l2loopback video_nr=10 card_label=Video-Loopback exclusive_caps=1")
machine.succeed("ffmpeg -re -loop 1 -i /etc/${feedImageFile} -vf format=yuv420p -f v4l2 /dev/video10 -loglevel fatal >&2 &")
with subtest("lomiri barcode scanner uses camera"):
machine.succeed("lomiri-camera-app --mode=barcode-reader >&2 &")
# emitted twice
machine.wait_for_console_text("No flash control support")
machine.wait_for_console_text("No flash control support")
machine.sleep(10)
machine.send_key("alt-f10")
machine.sleep(5)
machine.wait_for_text("${feedLabel}")
machine.succeed("xdotool mousemove 510 670 click 1") # open up QR decode result
# OCR is struggling to recognise the text. Click the clipboard button, check what got copied
machine.sleep(5)
machine.screenshot("lomiri-barcode_decode")
machine.succeed("xdotool mousemove 540 590 click 1")
machine.wait_until_succeeds("env DISPLAY=:0 xclip -selection clipboard -o | grep -q '${feedQrContent}'")
'';
}
);
}