MongoDB聚合分页问题：搜索时无法返回匹配记录总数

我在编写MongoDB聚合通用分页管道时遇到问题：无搜索参数时能正确返回全量文档总数，但传入搜索参数后，返回的总数仍是全量文档数，而非匹配搜索条件的记录数。当前代码如下：

export const paginate = (prevPipeline: any[], options: IOptions) => {
  const dataPipeline = [...prevPipeline];

  if (options.searchBy) {
    const searchCriteria: any = {};
    options.searchBy.split(',').forEach((searchOption: string) => {
      const [field, searchString, searchOptionStr] = searchOption.split(':');
      if (searchString.trim().length > 0) {
        searchCriteria[field] = {
          $regex: searchString,
          $options: searchOptionStr,
        };
      }
    });
    if (Object.keys(searchCriteria).length > 0) {
      dataPipeline.push({ $match: searchCriteria });
    }
  }

  if (options.sortBy) {
    const sortingCriteria: any = {};
    options.sortBy.split(',').forEach((sortOption: string) => {
      const [key, order] = sortOption.split(':');
      sortingCriteria[key] = order === 'desc' ? -1 : 1;
    });
    dataPipeline.push({ $sort: sortingCriteria });
  } else {
    dataPipeline.push({ $sort: { createdAt: 1 } });
  }

  // Project
  if (options.projectBy) {
    const projectionCriteria: any = {};
    options.projectBy.split(',').forEach((projectOption) => {
      const [key, include] = projectOption.split(':');
      projectionCriteria[key] = include === 'hide' ? 0 : 1;
    });
    dataPipeline.push({ $project: projectionCriteria });
  } else {
    dataPipeline.push({ $project: { createdAt: 0, updatedAt: 0 } });
  }

  // Pagination
  const limit =
    options.limit && parseInt(options.limit.toString(), 10) > 0
      ? parseInt(options.limit.toString(), 10)
      : 10;
  const page =
    options.page && parseInt(options.page.toString(), 10) > 0
      ? parseInt(options.page.toString(), 10)
      : 1;
  const skip = (page - 1) * limit;
  dataPipeline.push({ $skip: skip }, { $limit: limit });

  return [
    {
      $facet: {
        metadata: [{ $count: 'totalCount' }],
        data: dataPipeline,
      },
    },
  ];
};

问题分析 #

问题核心在于$facet的metadata管道：它仅执行了$count操作，完全没有应用任何过滤逻辑（包括传入的prevPipeline和生成的搜索$match条件），而搜索条件仅被添加到了dataPipeline中，导致统计的是集合所有文档的数量，而非过滤后的结果。

修复方案 #

将**过滤逻辑（原始prevPipeline + 搜索$match）**单独抽离，同时应用到metadata和dataPipeline中；排序、投影、分页仅保留在dataPipeline内。修复后的代码如下：

export const paginate = (prevPipeline: any[], options: IOptions) => {
  // 1. 单独提取过滤管道：包含原始管道和搜索条件
  const filterPipeline = [...prevPipeline];
  if (options.searchBy) {
    const searchCriteria: any = {};
    options.searchBy.split(',').forEach((searchOption: string) => {
      const [field, searchString, searchOptionStr] = searchOption.split(':');
      if (searchString.trim().length > 0) {
        searchCriteria[field] = {
          $regex: searchString,
          $options: searchOptionStr,
        };
      }
    });
    if (Object.keys(searchCriteria).length > 0) {
      filterPipeline.push({ $match: searchCriteria });
    }
  }

  // 2. 数据管道：基于过滤管道，添加排序、投影、分页
  const dataPipeline = [...filterPipeline];
  if (options.sortBy) {
    const sortingCriteria: any = {};
    options.sortBy.split(',').forEach((sortOption: string) => {
      const [key, order] = sortOption.split(':');
      sortingCriteria[key] = order === 'desc' ? -1 : 1;
    });
    dataPipeline.push({ $sort: sortingCriteria });
  } else {
    dataPipeline.push({ $sort: { createdAt: 1 } });
  }

  // Project
  if (options.projectBy) {
    const projectionCriteria: any = {};
    options.projectBy.split(',').forEach((projectOption) => {
      const [key, include] = projectOption.split(':');
      projectionCriteria[key] = include === 'hide' ? 0 : 1;
    });
    dataPipeline.push({ $project: projectionCriteria });
  } else {
    dataPipeline.push({ $project: { createdAt: 0, updatedAt: 0 } });
  }

  // Pagination
  const limit =
    options.limit && parseInt(options.limit.toString(), 10) > 0
      ? parseInt(options.limit.toString(), 10)
      : 10;
  const page =
    options.page && parseInt(options.page.toString(), 10) > 0
      ? parseInt(options.page.toString(), 10)
      : 1;
  const skip = (page - 1) * limit;
  dataPipeline.push({ $skip: skip }, { $limit: limit });

  // 3. facet中，metadata使用过滤管道后再统计总数
  return [
    {
      $facet: {
        metadata: [...filterPipeline, { $count: 'totalCount' }],
        data: dataPipeline,
      },
    },
  ];
};

关键改动说明：

• 拆分出filterPipeline，集中处理所有过滤逻辑（原始管道+搜索条件）
• metadata管道先执行filterPipeline过滤，再执行$count，确保统计的是匹配后的记录数
• dataPipeline基于filterPipeline扩展后续分页相关操作，保证数据和总数的统计逻辑一致

内容的提问来源于stack exchange，提问作者deepak