ms-swift/uni/run_with_progress_updates.sh

#!/usr/bin/env bash
#
# run_layoutsam_with_slack.sh
# → Builds the filtered image list, then downloads only those images,
#    posting progress & errors into Slack every 5 min.

# —————————————————————————————————————————————
# 1) Slack webhook setup
# —————————————————————————————————————————————
WEBHOOK="https://hooks.slack.com/services/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"

send_slack_message() {
  local msg="$1"
  curl -s -X POST -H 'Content-type: application/json' \
       --data "{\"text\":\"$msg\"}" \
       "$WEBHOOK"
}

# —————————————————————————————————————————————
# 2) Paths & startup
# —————————————————————————————————————————————
IMG_LIST_SCRIPT="uni/generate_image_list.py"
DOWNLOAD_SCRIPT="uni/download.py"
DATA_DIR="/home/ubuntu/ext-mamba-illinois/UCSD-project/workspace/datasets/SA-1b"
DOWNLOAD_DIR="$DATA_DIR/LayoutImgs"
MASK_DIR="$DATA_DIR/LayoutMasks"
LOG_FILE="./download_progress.log"
START_TIME=$(date +%s)

# —————————————————————————————————————————————
# 3) Check if JSON file exists, skip generation if it does
# —————————————————————————————————————————————
if [ ! -f "./images_to_download.json" ]; then
  send_slack_message "🔍 Generating LayoutSAM image list…"
  python3 "$IMG_LIST_SCRIPT" >> "$LOG_FILE" 2>&1 || {
    send_slack_message "❌ Failed to generate image list – aborting."
    exit 1
  }
else
  send_slack_message "📋 Using existing image list: images_to_download.json"
fi

# —————————————————————————————————————————————
# 4) Compute total images count
# —————————————————————————————————————————————
TOTAL_IMAGES=$(python3 - << 'PYCODE'
import json, sys
try:
    data = json.load(open("images_to_download.json"))
    print(len(data))
except:
    sys.exit(1)
PYCODE
)
if [ -z "$TOTAL_IMAGES" ]; then
  send_slack_message "❌ Could not determine total images count."
  exit 1
fi
send_slack_message "✅ Image list ready: $TOTAL_IMAGES paths."

# —————————————————————————————————————————————
# 5) Generate shard links for downloader
# —————————————————————————————————————————————
send_slack_message "🔗 Generating shard links from image list…"
python3 generate_shard_links.py >> "$LOG_FILE" 2>&1 || {
  send_slack_message "❌ Failed to generate shard links – aborting."
  exit 1
}

# —————————————————————————————————————————————
# 6) Launch the downloader in background
# —————————————————————————————————————————————
send_slack_message "🚀 Starting SA-1B download of LayoutSAM subset…"
python3 "$DOWNLOAD_SCRIPT" \
  --processes 16 \
  --input_file "./layoutsam_shard_links.txt" \
  --raw_dir "$DOWNLOADER_DIR/raw" \
  --images_dir "$DOWNLOAD_DIR" \
  --masks_dir "$MASK_DIR" \
  --images_json "./images_to_download.json" \
  --skip_existing >> "$LOG_FILE" 2>&1 &
SCRIPT_PID=$!

# —————————————————————————————————————————————
# 7) Progress & error‐checking functions
# —————————————————————————————————————————————
get_progress(){
  local now=$(date +%s)
  local elapsed=$(( (now - START_TIME)/60 ))
  local got=$(find "$DOWNLOAD_DIR" -type f | wc -l)
  local usage=$(df -h "$DOWNLOAD_DIR" | tail -1 | awk '{print $5}')
  echo "📊 Progress (${elapsed}m): $got / $TOTAL_IMAGES images downloaded • Disk: $usage"
}

check_errors(){
  if grep -iE "error|exception|traceback" "$LOG_FILE" >/dev/null; then
    echo "⚠️  Last error: $(grep -iE "error|exception|traceback" "$LOG_FILE" | tail -1)"
  fi
}

# —————————————————————————————————————————————
# 8) Monitoring loop
# —————————————————————————————————————————————
while kill -0 $SCRIPT_PID 2>/dev/null; do
  sleep 300
  prog=$(get_progress)
  err=$(check_errors)
  send_slack_message "$prog${err:+\n$err}"
done

# —————————————————————————————————————————————
# 9) Final report
# —————————————————————————————————————————————
wait $SCRIPT_PID
EXIT=$?

if [ $EXIT -eq 0 ]; then
  FINAL=$(find "$DOWNLOAD_DIR" -type f | wc -l)
  DURATION=$(( ( $(date +%s) - START_TIME )/60 ))
  send_slack_message "✅ Done! Downloaded $FINAL/$TOTAL_IMAGES images in ${DURATION}m."
else
  send_slack_message "❌ Download failed (exit code $EXIT). Check log: $LOG_FILE"
  send_slack_message "🔍 Last log lines:\n$(tail -5 "$LOG_FILE")"
fi

# —————————————————————————————————————————————
# 10) Cleanup
# —————————————————————————————————————————————
rm -f "$LOG_FILE"