simplex-chat/scripts/ci/desktop-linux/ui

#!/bin/bash

# Drives the desktop app on a virtual X display, for the scripted cases and for an agent on the runner.
# Compose for Desktop has no accessibility tree on Linux, so text is found on screenshots by OCR (tesseract);
# coordinates are screen pixels, as in the screenshots. Screenshots, the text read from them and
# a log of every command (steps.log) go to UI_OUT; the app's data stays in UI_STATE between starts.
#
#   ui start <app dir>            start Xvfb, openbox and <app dir>/bin/simplex, and wait for its window
#   ui shot <name>                save UI_OUT/NN-<name>.png and NN-<name>.txt (OCR's input and output in UI_OUT/ocr), and print the text on screen
#   ui text                       print the text on screen, one phrase per line after its centre x,y
#   ui wait <text> [secs]         wait up to secs (30) until the text is on screen (case and punctuation ignored)
#   ui click <x> <y>              click at x,y
#   ui click-text <text> [secs]   wait up to secs (20) for the text and click its centre
#   ui type <text>                type into the focused field
#   ui key <key>...               press keys named as in xdotool (Return, Escape, ctrl+a)
#   ui stop                       stop the app, openbox and Xvfb

set -euo pipefail

: "${UI_OUT:?UI_OUT must be the output directory}"
export DISPLAY=${UI_DISPLAY:-:99}
state=${UI_STATE:-${RUNNER_TEMP:-/tmp}/ui}
here=$(cd "$(dirname "$0")" && pwd)
mkdir -p "$UI_OUT" "$state/ocr"

log() {
  echo "$(date -u +%H:%M:%S) $*" | tee -a "$UI_OUT/steps.log"
}

running() {
  [ -f "$state/$1.pid" ] && kill -0 "$(cat "$state/$1.pid")" 2> /dev/null
}

stop_process() {
  local name=$1 secs=$2 pid
  if ! running "$name"; then
    rm -f "$state/$name.pid"
    return
  fi
  pid=$(cat "$state/$name.pid")
  kill "$pid" 2> /dev/null || true
  for _ in $(seq "$secs"); do
    kill -0 "$pid" 2> /dev/null || break
    sleep 1
  done
  kill -9 "$pid" 2> /dev/null || true
  rm -f "$state/$name.pid"
}

start() {
  local app_dir
  app_dir=$(cd "$1" && pwd)
  if running app; then
    echo "error: the app is already running"
    return 1
  fi
  stop_process openbox 2
  stop_process xvfb 2

  setsid Xvfb "$DISPLAY" -screen 0 1440x900x24 -nolisten tcp < /dev/null > "$UI_OUT/xvfb.log" 2>&1 &
  echo $! > "$state/xvfb.pid"
  for _ in $(seq 20); do
    xdotool getdisplaygeometry > /dev/null 2>&1 && break
    sleep 0.5
  done
  xdotool getdisplaygeometry

  setsid openbox --sm-disable < /dev/null > "$UI_OUT/openbox.log" 2>&1 &
  echo $! > "$state/openbox.pid"

  mkdir -p "$state/data" "$state/config" "$state/tmp"
  XDG_DATA_HOME="$state/data" \
    XDG_CONFIG_HOME="$state/config" \
    JAVA_TOOL_OPTIONS="-Djava.io.tmpdir=$state/tmp -Dskiko.renderApi=SOFTWARE" \
    LD_LIBRARY_PATH="$app_dir/lib/app/resources" \
    setsid "$app_dir/bin/simplex" < /dev/null >> "$UI_OUT/app.log" 2>&1 &
  echo $! > "$state/app.pid"

  if ! timeout 120 xdotool search --sync --onlyvisible --name '^SimpleX$' > "$state/window"; then
    echo "error: no app window after 120 s"
    return 1
  fi
  xdotool getwindowgeometry "$(head -n 1 "$state/window")"
}

# ocr <png> lines | ocr <png> find <text>
# Text of either polarity becomes dark on white: the difference from a morphological close (which removes dark strokes) and
# from an open (which removes light ones). A mean-based local threshold rings coloured buttons, and tesseract skips text in rings.
ocr() {
  local png=$1 o=$state/ocr
  shift
  convert "$png" -colorspace Gray "$o/gray.png"
  convert "$o/gray.png" \( +clone -morphology Dilate Rectangle:9x1 -morphology Dilate Rectangle:1x9 -morphology Erode Rectangle:9x1 -morphology Erode Rectangle:1x9 \) -compose Difference -composite "$o/dark.png"
  convert "$o/gray.png" \( +clone -morphology Erode Rectangle:9x1 -morphology Erode Rectangle:1x9 -morphology Dilate Rectangle:9x1 -morphology Dilate Rectangle:1x9 \) -compose Difference -composite "$o/light.png"
  convert "$o/dark.png" "$o/light.png" -compose Lighten -composite -negate -resize 200% "$o/text.png"
  tesseract "$o/text.png" "$o/text" --psm 11 tsv > /dev/null 2>&1
  python3 "$here/screen-text.py" 2 "$@" "$o/text.tsv"
}

screen_ocr() {
  import -window root "$state/ocr/screen.png"
  ocr "$state/ocr/screen.png" "$@"
}

# prints "x y" of the text once it is on screen
wait_for() {
  local text=$1 deadline=$((SECONDS + $2))
  until screen_ocr find "$text"; do
    [ "$SECONDS" -lt "$deadline" ] || return 1
    sleep 1
  done
}

shot() {
  local n f
  n=$(find "$UI_OUT" -maxdepth 1 -name '[0-9][0-9]-*.png' | wc -l)
  f=$(printf '%s/%02d-%s' "$UI_OUT" $((n + 1)) "$1")
  import -window root "$f.png"
  ocr "$f.png" lines > "$f.txt"
  mkdir -p "$UI_OUT/ocr"
  cp "$state/ocr/text.png" "$UI_OUT/ocr/$(basename "$f").png"
  cp "$state/ocr/text.tsv" "$UI_OUT/ocr/$(basename "$f").tsv"
  echo "::group::text on screen: $(basename "$f")"
  cat "$f.txt"
  echo "::endgroup::"
  log "shot $(basename "$f").png"
}

click() {
  xdotool mousemove --sync "$1" "$2" click 1
  sleep 1
}

case "${1:-}" in
  start)
    log "start $2"
    start "$2"
    ;;
  shot) shot "$2" ;;
  text) screen_ocr lines ;;
  wait)
    secs=${3:-30}
    if ! pos=$(wait_for "$2" "$secs"); then
      log "wait \"$2\": not on screen after $secs s"
      exit 1
    fi
    log "wait \"$2\": at ${pos/ /,}"
    ;;
  click)
    click "$2" "$3"
    log "click $2,$3"
    ;;
  click-text)
    secs=${3:-20}
    if ! pos=$(wait_for "$2" "$secs"); then
      log "click-text \"$2\": not on screen after $secs s"
      exit 1
    fi
    read -r x y <<< "$pos"
    click "$x" "$y"
    log "click-text \"$2\" at $x,$y"
    ;;
  type)
    xdotool type --delay 50 -- "$2"
    log "type \"$2\""
    ;;
  key)
    shift
    xdotool key --delay 100 -- "$@"
    log "key $*"
    ;;
  stop)
    stop_process app 10
    stop_process openbox 2
    stop_process xvfb 2
    log "stop"
    ;;
  *)
    sed -n '3,16s/^# \{0,1\}//p' "$0"
    exit 1
    ;;
esac