#!/usr/bin/env bash
# Snapshot Public Holidays + School Holidays for all 16 German Bundeslaender
# from openholidaysapi.org. The result is committed to the repo and imported
# at first DB boot by school-service. Re-run yearly (or whenever the next
# school year's data needs to be added).
#
# Usage:  bash scripts/calendar-snapshot.sh [FIRST_YEAR] [LAST_YEAR]
#         defaults: current year .. current year + 2
#
# Output: school-service/internal/seed/calendar_holidays.json
#         shape: [{ region, event_type, name_de, name_en, start_date, end_date }, ...]
set -euo pipefail

ROOT="$(cd "$(dirname "$0")/.." && pwd)"
OUT="$ROOT/school-service/internal/seed/calendar_holidays.json"
mkdir -p "$(dirname "$OUT")"

START_YEAR="${1:-$(date +%Y)}"
END_YEAR="${2:-$((START_YEAR + 2))}"
API="https://openholidaysapi.org"

# DE-XX codes for all 16 Bundeslaender (alphabetical).
REGIONS=(
  "DE-BW" "DE-BY" "DE-BE" "DE-BB" "DE-HB" "DE-HH" "DE-HE" "DE-MV"
  "DE-NI" "DE-NW" "DE-RP" "DE-SL" "DE-SN" "DE-ST" "DE-SH" "DE-TH"
)

if ! command -v jq >/dev/null 2>&1; then
  echo "jq is required (brew install jq)" >&2
  exit 1
fi

TMP=$(mktemp)
trap 'rm -f "$TMP"' EXIT
echo '[]' > "$TMP"

fetch() {
  local endpoint="$1" region="$2" year="$3"
  curl -sf -G "$API/$endpoint" \
    --data-urlencode "countryIsoCode=DE" \
    --data-urlencode "languageIsoCode=DE" \
    --data-urlencode "validFrom=${year}-01-01" \
    --data-urlencode "validTo=${year}-12-31" \
    --data-urlencode "subdivisionCode=$region" \
    || echo '[]'
}

# Map OpenHolidaysAPI shape → our DB schema. The API returns an array of:
#   { id, startDate, endDate, type, name: [{ language, text }], ... }
# We keep DE name as canonical, EN name if present, plus dates and a typed
# event_type discriminator. PublicHolidays and SchoolHolidays come from two
# separate endpoints.
normalise_jq='
  map({
    region: $region,
    event_type: $event_type,
    name_de: ((.name // []) | map(select(.language == "DE")) | .[0].text // ""),
    name_en: ((.name // []) | map(select(.language == "EN")) | .[0].text // null),
    start_date: .startDate,
    end_date: .endDate
  }) | map(select(.name_de != ""))
'

for region in "${REGIONS[@]}"; do
  for year in $(seq "$START_YEAR" "$END_YEAR"); do
    echo "  $region $year — public" >&2
    fetch "PublicHolidays" "$region" "$year" \
      | jq --arg region "$region" --arg event_type "public_holiday" "$normalise_jq" \
      | jq -s --slurpfile existing "$TMP" '$existing[0] + .[0]' > "$TMP.new"
    mv "$TMP.new" "$TMP"

    echo "  $region $year — school" >&2
    fetch "SchoolHolidays" "$region" "$year" \
      | jq --arg region "$region" --arg event_type "school_holiday" "$normalise_jq" \
      | jq -s --slurpfile existing "$TMP" '$existing[0] + .[0]' > "$TMP.new"
    mv "$TMP.new" "$TMP"
  done
done

# Deduplicate (the API sometimes returns overlapping rows for events that
# straddle a year boundary) and sort for a stable diff.
jq 'unique_by({region, event_type, name_de, start_date}) | sort_by([.region, .start_date])' \
   "$TMP" > "$OUT"

echo
echo "Wrote $(jq length "$OUT") events to $OUT"