#!/usr/bin/env bash

set -euo pipefail

# --- Configuration (override with env vars in CI) ---
DIFF_OUTPUT="${DIFF_OUTPUT:-deleted_workflows.txt}"
BASE_REF="${BASE_REF:-${GITHUB_BASE_REF:-main}}"
HEAD_REF="${HEAD_REF:-${GITHUB_HEAD_REF:-HEAD}}"
REPOSITORY_FILE="${REPOSITORY_FILE:-src/template_code_location/repository.py}"
FAIL_ON_DELETION="${FAIL_ON_DELETION:-true}"

TMP_DIR="$(mktemp -d)"
BASE_DIR="${TMP_DIR}/base"
HEAD_DIR="${TMP_DIR}/head"
BASE_JOBS_FILE="${TMP_DIR}/base_jobs.txt"
HEAD_JOBS_FILE="${TMP_DIR}/head_jobs.txt"

export DAGSTER_HOME="$(mktemp -d)"

cleanup() {
    git worktree remove -f "${BASE_DIR}" >/dev/null 2>&1 || true
    git worktree remove -f "${HEAD_DIR}" >/dev/null 2>&1 || true
    rm -rf "${TMP_DIR}" "${DAGSTER_HOME}"
}
trap cleanup EXIT

get_jobs_for_ref() {
    local workdir="$1"

    (
        cd "${workdir}"
        PYTHONPATH="${workdir}/src${PYTHONPATH:+:${PYTHONPATH}}" \
            dagster job list -f "${REPOSITORY_FILE}" 2>/dev/null | \
            grep '^Job: ' | \
            awk '{print $2}' | \
            sort -u
    )
}

echo "Fetching refs from origin..."
git fetch origin --quiet

if ! git rev-parse --verify "${BASE_REF}" >/dev/null 2>&1; then
    git fetch origin --quiet "${BASE_REF}:${BASE_REF}"
fi

if [ "${HEAD_REF}" != "HEAD" ] && ! git rev-parse --verify "${HEAD_REF}" >/dev/null 2>&1; then
    git fetch origin --quiet "${HEAD_REF}:${HEAD_REF}"
fi

echo "Preparing worktrees for ${BASE_REF} and ${HEAD_REF}..."
git worktree add --quiet "${BASE_DIR}" "${BASE_REF}"
git worktree add --quiet "${HEAD_DIR}" "${HEAD_REF}"

echo "Collecting workflows/jobs from ${BASE_REF}..."
get_jobs_for_ref "${BASE_DIR}" > "${BASE_JOBS_FILE}" || true

echo "Collecting workflows/jobs from ${HEAD_REF}..."
get_jobs_for_ref "${HEAD_DIR}" > "${HEAD_JOBS_FILE}" || true

# comm -23: items present in base but missing from head
comm -23 "${BASE_JOBS_FILE}" "${HEAD_JOBS_FILE}" > "${DIFF_OUTPUT}"

if [ -s "${DIFF_OUTPUT}" ]; then
    echo "------------------------------------------------"
    echo "DELETED WORKFLOWS DETECTED"
    echo "The following workflows/jobs exist in ${BASE_REF} but are missing in ${HEAD_REF}:"
    cat "${DIFF_OUTPUT}"
    echo "------------------------------------------------"

    if [ "${FAIL_ON_DELETION}" = "true" ]; then
        exit 1
    fi
else
    echo "No workflows/jobs were deleted in ${HEAD_REF} compared to ${BASE_REF}."
    rm -f "${DIFF_OUTPUT}"
fi