在AWS Amplify和DynamoDB中实现帖子及表情反应的最优方案

方案一：使用AppSync自定义VTL Resolver（推荐，最简单） #

1. 修改GraphQL Schema #

将Post模型中原有关联的reactions字段替换为自定义聚合类型，避免返回原始的单用户反应列表：

# 定义聚合后的反应类型
type ReactionAggregate {
  emojiUnicode: String!
  customerIds: [ID!]!
  count: Int! # 可选，直接返回反应数量，无需前端计算
}

type Post {
  id: ID!
  channelId: ID @index(
      name: "byChannel", sortKeyFields: ["createdAt"],
      queryField: "listPostsByChannel"
  )
  customerId: ID @index(
      name: "byCustomer", sortKeyFields: ["postType", "createdAt"]
  )
  text: String!
  postTempId: String
  postType: String
  reactions: [ReactionAggregate] # 去掉@hasMany，改为自定义字段
  createdAt: AWSDateTime
  updatedAt: AWSDateTime
}

# PostReaction模型保持原有定义不变
type PostReaction {
  postId: ID! @primaryKey(sortKeyFields: ["customerId", "emojiUnicode"])
  customerId: String!
  customerMeta: CustomerMeta
  emojiUnicode: String!
  createdAt: AWSDateTime
  updatedAt: AWSDateTime
}

2. 配置自定义VTL Resolver #

为Post的reactions字段添加VTL解析器，在后端完成查询与聚合：

请求模板（查询当前帖子的所有反应）

{
  "version": "2017-02-28",
  "operation": "Query",
  "query": {
    "expression": "postId = :postId",
    "expressionValues": {
      ":postId": $util.dynamodb.toDynamoDBJson($ctx.source.id)
    }
  }
}

响应模板（按表情分组聚合用户ID）

# 初始化分组容器
#set $aggregated = {}

#foreach($reaction in $ctx.result.items)
  #set $emoji = $reaction.emojiUnicode
  #if(!$aggregated.containsKey($emoji))
    #set $_ = $aggregated.put($emoji, {
      "emojiUnicode": $emoji,
      "customerIds": [],
      "count": 0
    })
  #end
  #set $_ = $aggregated.get($emoji).customerIds.add($reaction.customerId)
  #set $aggregated.get($emoji).count = $aggregated.get($emoji).count + 1
#end

#return $util.toJson($aggregated.values())

注：PostReaction的主键已将postId作为分区键，因此该查询效率极高，无需额外创建索引。

方案二：使用Lambda Resolver（逻辑更灵活） #

如果VTL的分组逻辑无法满足复杂需求（比如过滤特定用户、按时间排序），可以用Lambda函数实现聚合：

1. 编写Lambda聚合逻辑（Node.js示例）：

const AWS = require('aws-sdk');
const docClient = new AWS.DynamoDB.DocumentClient();

exports.handler = async (event) => {
  const postId = event.source.id;
  const params = {
    TableName: 'PostReaction-XXXXXX', // 替换为你的PostReaction表名
    KeyConditionExpression: 'postId = :postId',
    ExpressionAttributeValues: { ':postId': postId }
  };
  
  const result = await docClient.query(params).promise();
  const aggregated = {};
  
  result.Items.forEach(reaction => {
    const { emojiUnicode, customerId } = reaction;
    if (!aggregated[emojiUnicode]) {
      aggregated[emojiUnicode] = {
        emojiUnicode,
        customerIds: [],
        count: 0
      };
    }
    aggregated[emojiUnicode].customerIds.push(customerId);
    aggregated[emojiUnicode].count++;
  });
  
  return Object.values(aggregated);
};

2. 在AppSync中配置该Lambda作为Post的reactions字段解析器，查询时自动触发聚合逻辑。

方案三：前端聚合（不推荐） #

若暂时无法修改后端，可在前端获取数据后自行分组，但会增加前端逻辑开销：

// 前端聚合示例（React）
const aggregateReactions = (reactions) => {
  return Object.values(reactions.reduce((acc, reaction) => {
    if (!acc[reaction.emojiUnicode]) {
      acc[reaction.emojiUnicode] = {
        emojiUnicode: reaction.emojiUnicode,
        customerIds: [],
        count: 0
      };
    }
    acc[reaction.emojiUnicode].customerIds.push(reaction.customerId);
    acc[reaction.emojiUnicode].count++;
    return acc;
  }, {}));
};

// 使用方式
const post = await API.graphql(graphqlOperation(getPost, { id: postId }));
const aggregatedReactions = aggregateReactions(post.data.getPost.reactions);

内容的提问来源于stack exchange，提问作者Robson