#!/usr/bin/env bash
# Verify a CloudFront SPA deploy from deploy-web.yaml.
#
# Fail fast when any origin_path is still non-empty. Poll while the
# distribution is InProgress or the served index.html hash does not match
# the build. Then smoke-check caching headers, hashed assets, the baked
# API URL, and CORS against the target API.
set -euo pipefail

DISTRIBUTION_ID="${DISTRIBUTION_ID:-}"
EXPECTED_INDEX_SHA256="${EXPECTED_INDEX_SHA256:-}"
SITE_URL="${SITE_URL:-}"
API_URL="${API_URL:-https://api.dev.seahaven.com/api}"
BUDGET="${BUDGET:-40}"
INTERVAL="${INTERVAL:-15}"

if [[ -z "${DISTRIBUTION_ID}" || -z "${EXPECTED_INDEX_SHA256}" || -z "${SITE_URL}" ]]; then
  echo "Usage: DISTRIBUTION_ID EXPECTED_INDEX_SHA256 SITE_URL must be set." >&2
  exit 2
fi

SITE_URL="${SITE_URL%/}"
API_URL="${API_URL%/}"

sha256_of() {
  python3 -c "import hashlib,sys; print(hashlib.sha256(sys.stdin.buffer.read()).hexdigest())"
}

read_distribution_json() {
  aws cloudfront get-distribution --id "${DISTRIBUTION_ID}" --output json
}

parse_distribution() {
  python3 -c '
import json, sys
payload = json.load(sys.stdin)
dist = payload.get("Distribution") or payload
status = dist.get("Status") or "Unknown"
config = dist.get("DistributionConfig") or {}
origins = ((config.get("Origins") or {}).get("Items")) or []
paths = [origin.get("OriginPath") or "" for origin in origins]
nonempty = [path for path in paths if path]
print(status)
print("\x1f".join(paths))
print("yes" if nonempty else "no")
'
}

last_status="Unknown"
last_paths="Unknown"
last_hash="Unknown"
last_path_nonempty="no"

observe() {
  local parsed
  parsed="$(read_distribution_json | parse_distribution)"
  last_status="$(printf '%s\n' "${parsed}" | sed -n '1p')"
  last_paths="$(printf '%s\n' "${parsed}" | sed -n '2p' | tr '\037' ' ')"
  last_path_nonempty="$(printf '%s\n' "${parsed}" | sed -n '3p')"
  local hash
  if hash="$(curl -fsS --max-time 30 "${SITE_URL}/" | sha256_of)" && [[ -n "${hash}" ]]; then
    last_hash="${hash}"
  else
    last_hash="unreachable"
  fi
}

report_state() {
  echo "last observed: status=${last_status} origins=${last_paths} served_sha256=${last_hash}"
}

fail_fast_if_origin_path() {
  if [[ "${last_path_nonempty}" == "yes" ]]; then
    echo "FAIL: live origin_path values are '${last_paths}'; expected empty bucket-root origins." >&2
    report_state >&2
    exit 1
  fi
}

observe
fail_fast_if_origin_path

attempt=0
while [[ "${attempt}" -lt "${BUDGET}" ]]; do
  attempt=$((attempt + 1))
  echo "poll ${attempt}/${BUDGET}: status=${last_status} served_sha256=${last_hash}"
  fail_fast_if_origin_path
  if [[ "${last_status}" == "Deployed" && "${last_hash}" == "${EXPECTED_INDEX_SHA256}" ]]; then
    break
  fi
  sleep "${INTERVAL}"
  observe
done

if [[ "${last_status}" != "Deployed" || "${last_hash}" != "${EXPECTED_INDEX_SHA256}" ]]; then
  echo "FAIL: release did not converge within the budget." >&2
  report_state >&2
  exit 1
fi

write_asset_paths() {
  python3 -c '
import re, sys
html = open(sys.argv[1], encoding="utf-8").read()
seen = []
for path in re.findall(r"(?:src|href)=\"(/assets/[^\"]+\.(?:js|css))\"", html):
    if path not in seen:
        seen.append(path)
        print(path)
' "$1"
}

api_host() {
  python3 -c 'import os,urllib.parse; print(urllib.parse.urlparse(os.environ["API_URL"]).hostname or "")'
}

assert_baked_api_url() {
  local tmp="$1"
  if [[ ! -s "${tmp}/asset-paths.txt" ]]; then
    echo "FAIL: served index.html has no /assets/ URL to check immutable caching." >&2
    exit 1
  fi
  : > "${tmp}/assets.txt"
  local immutable_ok="no"
  local asset_path
  while IFS= read -r asset_path; do
    curl -fsS --max-time 30 "${SITE_URL}${asset_path}" \
      -o "${tmp}/asset-body" -D "${tmp}/asset.headers"
    cat "${tmp}/asset-body" >> "${tmp}/assets.txt"
    if [[ "${asset_path}" == *.js && "${immutable_ok}" == "no" ]]; then
      if ! grep -qiE 'cache-control:.*immutable' "${tmp}/asset.headers"; then
        echo "FAIL: hashed asset is missing Cache-Control immutable." >&2
        exit 1
      fi
      immutable_ok="yes"
    fi
  done < "${tmp}/asset-paths.txt"
  if [[ "${immutable_ok}" != "yes" ]]; then
    echo "FAIL: served index.html has no hashed JS asset to check immutable caching." >&2
    exit 1
  fi
  cat "${tmp}/index.html" "${tmp}/assets.txt" > "${tmp}/served.txt"
  local host
  host="$(api_host)"
  local forbidden=""
  case "${host}" in
    api.dev.seahaven.com) forbidden="api.staging.seahaven.com" ;;
    api.staging.seahaven.com) forbidden="api.dev.seahaven.com" ;;
    *)
      echo "FAIL: API_URL host '${host}' is not a known SHOC API." >&2
      exit 1
      ;;
  esac
  if grep -Fq "${forbidden}" "${tmp}/served.txt"; then
    echo "FAIL: served assets contain forbidden URL ${forbidden}." >&2
    exit 1
  fi
  if grep -Fq "localhost:5141" "${tmp}/served.txt"; then
    echo "FAIL: served assets contain forbidden URL localhost:5141." >&2
    exit 1
  fi
  if ! grep -Fq "${host}" "${tmp}/served.txt"; then
    echo "FAIL: served JS assets are missing the baked API URL ${host}." >&2
    exit 1
  fi
}

tmp="$(mktemp -d)"
trap 'rm -rf "${tmp}"' EXIT

curl -fsS --max-time 30 "${SITE_URL}/" -o "${tmp}/index.html" -D "${tmp}/index.headers"
curl -fsS --max-time 30 "${SITE_URL}/login" -o "${tmp}/login.html"
curl -fsS --max-time 30 "${SITE_URL}/work-orders" -o "${tmp}/route.html"
if ! grep -qiE 'cache-control:.*no-store' "${tmp}/index.headers"; then
  echo "FAIL: HTML Cache-Control is missing no-store." >&2
  exit 1
fi
write_asset_paths "${tmp}/index.html" > "${tmp}/asset-paths.txt"
assert_baked_api_url "${tmp}"

cors_code="$(curl -sS --max-time 30 -o /dev/null -D "${tmp}/cors.headers" -w '%{http_code}' -X OPTIONS "${API_URL}" \
  -H "Origin: ${SITE_URL}" \
  -H "Access-Control-Request-Method: GET")"
if [[ "${cors_code}" != "200" && "${cors_code}" != "204" ]]; then
  echo "FAIL: CORS preflight returned HTTP ${cors_code}." >&2
  exit 1
fi
if ! grep -qi 'access-control-allow-origin' "${tmp}/cors.headers"; then
  echo "FAIL: CORS preflight is missing Access-Control-Allow-Origin." >&2
  exit 1
fi

echo "PASS: CloudFront release is Deployed, hash-matched, and smoke-clean."
report_state