package_bundle.sh

#!/usr/bin/env bash
# Create a shareable copy of the bundle (no venv, no caches, no sample outputs).
set -euo pipefail

ROOT="$(cd "$(dirname "${BASH_SOURCE[0]}")" && pwd)"
PARENT="$(dirname "$ROOT")"
BUNDLE_NAME="query_entity_export_bundle"
OUT_DIR="${1:-$PARENT/${BUNDLE_NAME}_share}"
ARCHIVE="${OUT_DIR}.zip"

rm -rf "$OUT_DIR"
mkdir -p "$OUT_DIR"

rsync -a \
  --exclude '.venv-quickumls/' \
  --exclude '.venv/' \
  --exclude '.git/' \
  --exclude '.vscode/' \
  --exclude '__pycache__/' \
  --exclude '*.pyc' \
  --exclude 'results/query_entity_detections_*.json' \
  --exclude "${BUNDLE_NAME}_share/" \
  --exclude '*.zip' \
  "$ROOT/" "$OUT_DIR/"

# Ensure results dir exists but stays empty of prior runs
mkdir -p "$OUT_DIR/results"
touch "$OUT_DIR/results/.gitkeep"

cat > "$OUT_DIR/SHARE_README.txt" <<'EOF'
Query Entity Export Bundle — shared copy
======================================

This folder does NOT include .venv-quickumls (machine-specific).
Your colleague must create it locally:

  cd query_entity_export_bundle_share   # or whatever this folder is named
  ./setup.sh
  export ICD_QUICKUMLS_DATA_DIR="/path/to/quickumls_data"
  ./run_export.sh --query "Type II Diabetes Mellitus Uncontrolled"

See README.md and SETUP.md for full instructions.

QuickUMLS data (~5 GB) is NOT bundled — each site must use their own
UMLS-licensed quickumls_data/ directory.
EOF

echo "Created shareable bundle: $OUT_DIR"
du -sh "$OUT_DIR"

if command -v zip >/dev/null 2>&1; then
  rm -f "$ARCHIVE"
  (cd "$PARENT" && zip -rq "$(basename "$ARCHIVE")" "$(basename "$OUT_DIR")")
  echo "Created archive: $ARCHIVE ($(du -sh "$ARCHIVE" | cut -f1))"
else
  echo "zip not found — folder only (no archive)."
fi