#!/usr/bin/env bash

set -euo pipefail

# 创建 syslog-consumer-rule 统计物化 Topic。
# 默认值对应当前客户现场，直接执行即可：
#   bash 03-create-stats-materialize-topic.sh
#
# 其他环境可在执行前覆盖：BOOTSTRAP、RAW_TOPIC、STATS_TOPIC、KAFKA_BIN、COMMAND_CONFIG。

SCRIPT_DIR=$(cd -- "$(dirname -- "${BASH_SOURCE[0]}")" && pwd)
BOOTSTRAP="${BOOTSTRAP:-10.141.121.35:9092,10.141.121.50:9092,10.141.121.59:9092}"
RAW_TOPIC="${RAW_TOPIC:-agent-01-syslog-topic}"
STATS_TOPIC="${STATS_TOPIC:-${RAW_TOPIC}.STATS.MATERIALIZE}"
RETENTION_MS="${RETENTION_MS:-604800000}"

if [[ -z "${KAFKA_BIN:-}" ]]; then
  for CANDIDATE_BIN in \
    "${SCRIPT_DIR}" \
    "/opt/kafka/kafka_2.13-3.6.1/bin" \
    "/opt/kafka/bin"; do
    if [[ -x "${CANDIDATE_BIN}/kafka-topics.sh" ]] \
        && [[ -x "${CANDIDATE_BIN}/kafka-configs.sh" ]]; then
      KAFKA_BIN="${CANDIDATE_BIN}"
      break
    fi
  done
fi

if [[ -z "${KAFKA_BIN:-}" ]] && command -v kafka-topics.sh >/dev/null 2>&1; then
  KAFKA_BIN=$(dirname -- "$(command -v kafka-topics.sh)")
fi

if [[ -z "${KAFKA_BIN:-}" ]]; then
  echo "ERROR: 未找到 Kafka CLI；现场预期路径为 /opt/kafka/kafka_2.13-3.6.1/bin" >&2
  exit 1
fi

TOPICS_CMD="${KAFKA_BIN}/kafka-topics.sh"
CONFIGS_CMD="${KAFKA_BIN}/kafka-configs.sh"
COMMON_ARGS=(--bootstrap-server "${BOOTSTRAP}")

if [[ -n "${COMMAND_CONFIG:-}" ]]; then
  if [[ ! -r "${COMMAND_CONFIG}" ]]; then
    echo "ERROR: 无法读取认证配置 ${COMMAND_CONFIG}" >&2
    exit 1
  fi
  COMMON_ARGS+=(--command-config "${COMMAND_CONFIG}")
fi

echo "Kafka：${BOOTSTRAP}"
echo "原始 Topic：${RAW_TOPIC}"
echo "统计 Topic：${STATS_TOPIC}"

echo "[1/4] 读取原始 Topic 的分区数和副本数"
RAW_INFO=$("${TOPICS_CMD}" "${COMMON_ARGS[@]}" --describe --topic "${RAW_TOPIC}")
printf '%s\n' "${RAW_INFO}"

PARTITIONS=$(printf '%s\n' "${RAW_INFO}" | awk '
  /PartitionCount:/ {
    for (i=1; i<=NF; i++) if ($i=="PartitionCount:") { print $(i+1); exit }
  }
')
REPLICAS=$(printf '%s\n' "${RAW_INFO}" | awk '
  /ReplicationFactor:/ {
    for (i=1; i<=NF; i++) if ($i=="ReplicationFactor:") { print $(i+1); exit }
  }
')

if [[ ! "${PARTITIONS}" =~ ^[1-9][0-9]*$ ]] || [[ ! "${REPLICAS}" =~ ^[1-9][0-9]*$ ]]; then
  echo "ERROR: 无法读取 ${RAW_TOPIC} 的分区数或副本数" >&2
  exit 1
fi
echo "将沿用：分区数=${PARTITIONS}，副本数=${REPLICAS}"

echo "[2/4] 创建统计 Topic（已存在时不重复创建）"
"${TOPICS_CMD}" "${COMMON_ARGS[@]}" \
  --create --if-not-exists \
  --topic "${STATS_TOPIC}" \
  --partitions "${PARTITIONS}" \
  --replication-factor "${REPLICAS}" \
  --config cleanup.policy=delete \
  --config "retention.ms=${RETENTION_MS}"

echo "[3/4] 统一保留策略"
if ! "${CONFIGS_CMD}" "${COMMON_ARGS[@]}" \
  --alter --entity-type topics --entity-name "${STATS_TOPIC}" \
  --add-config "cleanup.policy=delete,retention.ms=${RETENTION_MS}"; then
  echo "WARN: Kafka 配置更新请求超时或连接中断，将继续验证 Topic。" >&2
  echo "WARN: 新建命令已经携带相同配置，不影响 syslog-consumer-rule 使用该 Topic。" >&2
fi

echo "[4/4] 验证统计 Topic"
STATS_INFO=$("${TOPICS_CMD}" "${COMMON_ARGS[@]}" --describe --topic "${STATS_TOPIC}")
printf '%s\n' "${STATS_INFO}"

ACTUAL_PARTITIONS=$(printf '%s\n' "${STATS_INFO}" | awk '
  /PartitionCount:/ {
    for (i=1; i<=NF; i++) if ($i=="PartitionCount:") { print $(i+1); exit }
  }
')
ACTUAL_REPLICAS=$(printf '%s\n' "${STATS_INFO}" | awk '
  /ReplicationFactor:/ {
    for (i=1; i<=NF; i++) if ($i=="ReplicationFactor:") { print $(i+1); exit }
  }
')

if [[ "${ACTUAL_PARTITIONS}" != "${PARTITIONS}" ]]; then
  echo "ERROR: ${STATS_TOPIC} 已存在，但分区数=${ACTUAL_PARTITIONS}，与原始 Topic 的 ${PARTITIONS} 不一致" >&2
  echo "为避免同一 key 重新映射，脚本不会自动增加分区，请先联系研发确认。" >&2
  exit 1
fi
if [[ "${ACTUAL_REPLICAS}" != "${REPLICAS}" ]]; then
  echo "ERROR: ${STATS_TOPIC} 副本数=${ACTUAL_REPLICAS}，预期=${REPLICAS}" >&2
  exit 1
fi

echo "PASS: ${STATS_TOPIC} 已创建并验证成功。"
echo "分区数=${ACTUAL_PARTITIONS}，副本数=${ACTUAL_REPLICAS}，保留时间=${RETENTION_MS}ms。"
echo "脚本未删除消息，未修改任何消费组 offset。"