#!/usr/bin/env bash
# Build and push one architecture-specific development image (main push with [CI]) to moshiro-dev.
set -euo pipefail

[[ "${CI_PIPELINE_EVENT:-}" == push && "${CI_COMMIT_BRANCH:-}" == main && "${CI_COMMIT_MESSAGE:-}" == *"[CI]"* ]] || {
  echo "development image builds require a main push with [CI] in the commit message" >&2
  exit 2
}

component="${1:?usage: build-dev-image.sh <backend|agent|probe> [amd64|arm64]}"
arch="${2:-amd64}"
case "$component" in
  backend|agent|probe) ;;
  *) echo "unsupported component: $component" >&2; exit 2 ;;
esac
case "$arch" in
  amd64|arm64) ;;
  *) echo "unsupported architecture: $arch" >&2; exit 2 ;;
esac
if [[ "$component" == backend && "$arch" != amd64 ]]; then
  echo "the main development pipeline currently builds Backend for amd64 only" >&2
  exit 2
fi
[[ -n "${ZOT_USERNAME:-}" && -n "${ZOT_SECRET:-}" ]] || {
  echo "Zot credentials are required" >&2
  exit 2
}
[[ "${CI_COMMIT_SHA:-}" =~ ^[0-9a-fA-F]{40}$ ]] || {
  echo "CI_COMMIT_SHA must be a full commit SHA" >&2
  exit 2
}

active_command_pid=""
cancel_active_command() {
  local pid="${active_command_pid:-}"
  [[ -n "$pid" ]] || return 0

  kill -TERM -- "-$pid" 2>/dev/null || true
  kill -TERM "$pid" 2>/dev/null || true
  for _ in {1..10}; do
    kill -0 -- "-$pid" 2>/dev/null || break
    sleep 1
  done
  kill -KILL -- "-$pid" 2>/dev/null || true
  kill -KILL "$pid" 2>/dev/null || true
  wait "$pid" 2>/dev/null || true
  active_command_pid=""
}

handle_cancel() {
  trap - TERM INT HUP
  cancel_active_command
  exit 143
}

trap handle_cancel TERM INT HUP

run_bounded() {
  local duration="$1" status=0
  shift

  # setsid gives the command its own process group so a cancel signal can
  # terminate Buildah and its children even when the local backend only signals
  # the shell process. timeout is the fallback if cancellation is lost entirely.
  setsid timeout --foreground --signal=TERM --kill-after=30s "$duration" "$@" <&0 &
  active_command_pid=$!
  if wait "$active_command_pid"; then
    status=0
  else
    status=$?
  fi
  active_command_pid=""
  return "$status"
}

root="$(cd "$(dirname "${BASH_SOURCE[0]}")/.." && pwd)"
# 版本：v<下一个补丁版本>-dev.<自最近 tag 的提交数>-<提交前 7 位>（见 ci/version.sh）；写进工作区的 Directory.Build.props，镜像标签、构建信息、签名清单都用它
source "$root/ci/version.sh"
moshiro_apply_version
registry="${ZOT_REGISTRY:-registry.oeo.one}"
repository="$registry/moshiro-dev/moshiro-$component"
work="$root/.release/dev-image-build/$component-$arch"
base_home="/home/justme/woodpecker/buildah-home/dev-image-build-noble-aot-v1-$component-$arch"
# 镜像标签不带 v（和正式版一致）：latest 和 <版本>；构建信息和清单里的 tag 带 v
image_version="$MOSHIRO_VERSION"
mkdir -p "$work" "$base_home"
export REGISTRY_AUTH_FILE="$work/$component-auth.json"
if [[ ! -s "$REGISTRY_AUTH_FILE" ]]; then
  printf '{"auths":{}}\n' > "$REGISTRY_AUTH_FILE"
  chmod 600 "$REGISTRY_AUTH_FILE"
fi

if [[ "$component" == backend ]]; then
  BUILDAH_HOME="$base_home" \
    ZOT_REGISTRY="$registry" \
    ZOT_IMAGE_REPOSITORY="$repository" \
    IMAGE_TAGS="latest,$image_version" \
    IMAGE_PUBLISH=true \
    run_bounded 30m bash "$root/ci/release-linux.sh" "$arch" backend
  exit 0
fi

export HOME="$base_home"
export BUILDAH_HOME="$base_home"
source "$root/ci/buildah-env.sh"
moshiro_setup_buildah "$base_home"
context="$work/$component"
local_image="localhost/moshiro-$component:dev-$arch-check"
node_rid="linux-musl-x64"
[[ "$arch" == arm64 ]] && node_rid="linux-musl-arm64"
# Build a dedicated .NET 10 NativeAOT/AotAnywhere environment in this isolated
# Buildah store. Keeping NuGet inside image layers avoids a shared writable
# package mount being corrupted by canceled or parallel builds.
builder_image="localhost/moshiro-dotnet10-nativeaot-aotanywhere:ci"
source "$root/ci/nativeaot-toolchain.sh"
moshiro_nativeaot_settings "$root"
builder_cache="${MOSHIRO_RUNNER_IMAGE_CACHE:-/home/justme/woodpecker/cache/runner-image}"
cached_key="$(cat "$builder_cache/nativeaot-builder.key" 2>/dev/null || true)"
builder_id="$(cat "$builder_cache/nativeaot-builder.id" 2>/dev/null || true)"
if [[ "$cached_key" == "$MOSHIRO_NATIVEAOT_BUILDER_KEY" && "$builder_id" =~ ^[0-9a-f]{64}$ && -f "$builder_cache/nativeaot-builder-$builder_id.oci" ]]; then
  local_builder_id="$(buildah images --noheading --no-trunc --format '{{.ID}}' "$builder_image" 2>/dev/null || true)"
  if [[ "${local_builder_id#sha256:}" != "$builder_id" ]]; then
    run_bounded 10m buildah pull "oci-archive:$builder_cache/nativeaot-builder-$builder_id.oci"
    buildah tag "$builder_id" "$builder_image"
  fi
  echo "using cached NativeAOT toolchain $builder_id"
else
  builder_context="$work/nativeaot-builder"
  mkdir -p "$builder_context"
  tr -d '\r' < "$root/global.json" > "$builder_context/global.json"
  run_bounded 30m buildah bud --layers --network host --tls-verify=true --arch amd64 \
    --build-arg "AOTANYWHERE_VERSION=$MOSHIRO_AOTANYWHERE_VERSION" \
    --build-arg "ZIG_TOOLSET_VERSION=$MOSHIRO_ZIG_TOOLSET_VERSION" \
    -f "$root/packaging/container/NativeAot.Builder.Containerfile" \
    -t "$builder_image" "$builder_context"
fi

# A stable timestamp lets a retry of the same commit reuse the publish layer.
commit_epoch="$(git -C "$root" show -s --format=%ct "$CI_COMMIT_SHA")"
built_at="$(date -u -d "@$commit_epoch" +%Y-%m-%dT%H:%M:%SZ)"

python3 "$root/ci/stage-node-image.py" "$component" "$context"
# Zig compiles musl libc by spawning many compiler processes. Buildah's default
# nofile limit (1024) makes Zig fail with ProcessFdQuotaExceeded while linking.
run_bounded 30m buildah bud --layers --network host --tls-verify=true --ulimit nofile=65536:65536 --arch "$arch" \
  --build-arg "NATIVEAOT_BUILDER_IMAGE=$builder_image" \
  --build-arg "APP_VERSION=$image_version" \
  --build-arg "NODE_RID=$node_rid" \
  --build-arg "MOSHIRO_TAG=$MOSHIRO_TAG" --build-arg "MOSHIRO_COMMIT=${CI_COMMIT_SHA:0:10}" --build-arg "MOSHIRO_BUILT_AT=$built_at" \
  -f "$context/Containerfile" \
  -t "$local_image" "$context"

logged_in=false
for attempt in 1 2 3 4; do
  if run_bounded 2m buildah login --tls-verify=true --username "$ZOT_USERNAME" --password-stdin "$registry" <<< "$ZOT_SECRET" >/dev/null; then
    logged_in=true
    break
  fi
  if [[ "$attempt" -lt 4 ]]; then
    echo "Zot login attempt $attempt failed; retrying in 3 seconds" >&2
    sleep 3
  fi
done
[[ "$logged_in" == true ]] || { echo "could not authenticate to Zot after 4 attempts" >&2; exit 1; }

for tag in "$image_version-$arch"; do
  pushed=false
  for attempt in 1 2 3; do
    if run_bounded 10m buildah push --tls-verify=true --digestfile "$work/$component.digest" "$local_image" "docker://$repository:$tag"; then
      pushed=true
      break
    fi
    if [[ "$attempt" -lt 3 ]]; then
      echo "Zot push for $component:$tag attempt $attempt failed; retrying in 5 seconds" >&2
      sleep 5
    fi
  done
  [[ "$pushed" == true ]] || { echo "could not push $repository:$tag after 3 attempts" >&2; exit 1; }
  echo "pushed development image $repository:$tag"
done

# Agent and Probe manifests are signed after both architecture-specific images
# have been pushed. Backend is still a single-architecture development image.
if [[ "$component" == backend && -n "${RELEASE_SIGNING_KEY:-}" ]]; then
  bash "$root/ci/publish-release-offer.sh" "$component" "$repository" "$(cat "$work/$component.digest")" "$MOSHIRO_TAG" dev "$work/$component-release"
fi