动态LINQ GroupBy多列

时间:2010-10-14 00:09:05

标签: c# .net linq entity-framework dynamic-linq

我需要将以下LINQ查询转换为动态LINQ,它根据用户输入接受多个分组列。基本上我有一堆应用分组的下拉列表,我不想枚举每个分组组合。如果动态LINQ失败,我可能必须手动构建SQL查询,没有人想要它。

var grouping = ( from entry in ObjectContext.OmniturePageModules
    where entry.StartOfWeek >= startDate && entry.StartOfWeek <= endDate &&
        ( section == "Total" || section == "All" || entry.Section == section ) &&
        ( page == "Total" || page == "All" || entry.Page == page ) &&
        ( module == "Total" || module == "All" || entry.Module == module ) 
    group entry by new
    {
        entry.Page, // I want to be able to tell this anonymous type
        entry.Module, // which columns to group by
        entry.StartOfWeek // at runtime
    }
    into entryGroup
    select new
    {
        SeriesName = section + ":" + entryGroup.Key.Page + ":" + entryGroup.Key.Module,
        Week = entryGroup.Key.StartOfWeek,
        Clicks = entryGroup.Sum( p => p.Clicks )
    } );

我不知道如何做到这一点,因为动态LINQ在“hello world”之外完全没有文档记录!选择/ where / orderby案例。我只是无法弄清楚语法。

类似:(?)

var grouping = ObjectContext.OmniturePageModules.Where(entry => entry.StartOfWeek >= startDate && entry.StartOfWeek <= endDate &&
                                           ( section == "Total" || section == "All" || entry.Section == section ) &&
                                           ( page == "Total" || page == "All" || entry.Page == page ) &&
                                           ( module == "Total" || module == "All" || entry.Module == module ))
                                           .GroupBy("new (StartOfWeek,Page,Module)", "it")
                                           .Select("new (Sum(Clicks) as Clicks, SeriesName = section + key.Page + Key.Module, Week = it.Key.StartOfWeek)");

我正在使用System.Linq.Dynamic中的DynamicQueryable类。请参阅:http://weblogs.asp.net/scottgu/archive/2008/01/07/dynamic-linq-part-1-using-the-linq-dynamic-query-library.aspx

随访: Enigmativity的解决方案主要用于 。由于某种原因,它不希望按日期时间“StartOfWeek”列进行分组 - 解决方法只是进行二级分组:

var entries = ( from entry in ObjectContext.OmniturePageModules
                            where entry.StartOfWeek >= startDate
                                && entry.StartOfWeek <= endDate
                                && ( section == "Total" || section == "All" || entry.Section == section )
                                && ( page == "Total" || page == "All" || entry.Page == page )
                                && ( module == "Total" || module == "All" || entry.Module == module )
                            select entry ).ToArray(); // Force query execution

            var grouping = from entry in entries
                            let grouper = new EntryGrouper( entry, section, page, module )
                            group entry by grouper into entryGroup
                            select new
                            {
                                entryGroup.Key.SeriesName,
                                entryGroup.Key.Date, 
                                Clicks = entryGroup.Sum( p => p.Clicks ),
                            };

            var grouping2 = (from groups in grouping
                            group groups by new {groups.SeriesName, groups.Date } into entryGroup
                            select new
                            {
                               entryGroup.Key.SeriesName,
                               entryGroup.Key.Date,
                               Clicks = entryGroup.Sum( p => p.Clicks ),
                            } );

但这似乎严重降低了性能... = /

3 个答案:

答案 0 :(得分:8)

这里是动态LINQ - 当然你在运行时构建GroupBy和Select字符串:

var double_grouping = ( ObjectContext.OmniturePageModules.Where( entry => entry.StartOfWeek >= startDate
                     && entry.StartOfWeek <= endDate
                     && ( section == "Total" || section == "All" || entry.Section == section )
                     && ( page == "Total" || page == "All" || entry.Page == page )
                     && ( module == "Total" || module == "All" || entry.Module == module ) )
                     .GroupBy( "new ( it.Section, it.Page, it.StartOfWeek )", "it" ) )
                     .Select( "new ( Sum(Clicks) as Clicks, Key.Section as SeriesSection, Key.Page as SeriesPage, Key.StartOfWeek as Week )" );

这是正常的LINQ方式,直到同事指出它才逃过我 - 这基本上是没有石斑鱼类的Enigmativity解决方案:

var grouping = ( from entry in ObjectContext.OmniturePageModules
    where entry.StartOfWeek >= startDate && entry.StartOfWeek <= endDate &&
        ( section == "Total" || section == "All" || entry.Section == section ) &&
        ( page == "Total" || page == "All" || entry.Page == page ) &&
        ( module == "Total" || module == "All" || entry.Module == module )
    group entry by new
    {
        Section = section == "All" ? entry.Section : section,
        Page = page == "All" ? entry.Page : page,
        Module = module == "All" ? entry.Module : module,
        entry.StartOfWeek
    }
        into entryGroup
        select new
        {
            SeriesName =
            entryGroup.Key.Section + ":" + entryGroup.Key.Page + ":" + entryGroup.Key.Module,
            Week = entryGroup.Key.StartOfWeek,
            Clicks = entryGroup.Sum( p => p.Clicks )
        } );

答案 1 :(得分:3)

如果您明确想要使用LINQ动态查询库,那么我的答案将不是您想要的,但如果您想要您想要的行为并且您很乐意使用常规LINQ,那么我想我可以提供帮助。

基本上我已经创建了一个EntryGrouper类来处理下拉列表中所选值的分组逻辑,我假设变量sectionpage&amp; module持有这些价值观。我还假设ObjectContext.OmniturePageModulesEntry类型的可枚举。

所以你的LINQ查询现在变成了这两个:

var entries = (from entry in ObjectContext.OmniturePageModules
               where entry.StartOfWeek >= startDate
                   && entry.StartOfWeek <= endDate
                   && (section == "Total" || section == "All" || entry.Section == section)
                   && (page == "Total" || page == "All" || entry.Page == page)
                   && (module == "Total" || module == "All" || entry.Module == module)
               select entry).ToArray(); // Force query execution

var grouping = from entry in entries
               let grouper = new EntryGrouper(entry, section, page, module)
               group entry by grouper into entryGroup
               select new
               {
                   SeriesName = entryGroup.Key.SeriesName,
                   Week = entryGroup.Key.StartOfWeek,
                   Clicks = entryGroup.Sum(p => p.Clicks),
               };

第一个查询用于强制对数据库执行简单的选择查询,并仅返回要分组的记录。通常group by个查询会多次调用数据库,因此以这种方式查询通常要快得多。

第二个查询通过创建EntryGrouper类的实例作为分组键来对第一个查询的结果进行分组。

我在SeriesName类中包含了一个EntryGrouper属性,以便在一个地方整齐地定义所有分组逻辑。

现在,EntryGrouper类非常大,为了让分组正常工作,它需要具有StartOfWeekSectionPage和&amp;的属性。 Module,并包含Equals&amp;的重载GetHashCode方法,并实现IEquatable<Entry>接口。

这是:

public class EntryGrouper : IEquatable<Entry>
{
    private Entry _entry;
    private string _section;
    private string _page;
    private string _module;

    public EntryGrouper(Entry entry, string section, string page, string module)
    {
        _entry = entry;
        _section = section;
        _page = page;
        _module = module;
    }

    public string SeriesName
    {
        get
        {
            return String.Format("{0}:{1}:{2}", this.Section, this.Page, this.Module);
        }
    }

    public DateTime StartOfWeek
    {
        get
        {
            return _entry.StartOfWeek;
        }
    }

    public string Section
    {
        get
        {
            if (_section == "Total" || _section == "All")
                return _section;
            return _entry.Section;
        }
    }

    public string Page
    {
        get
        {
            if (_page == "Total" || _page == "All")
                return _page;
            return _entry.Page;
        }
    }

    public string Module
    {
        get
        {
            if (_module == "Total" || _module == "All")
                return _module;
            return _entry.Module;
        }
    }

    public override bool Equals(object other)
    {
        if (other is Entry)
            return this.Equals((Entry)other);
        return false;
    }

    public bool Equals(Entry other)
    {
        if (other == null)
            return false;
        if (!EqualityComparer<DateTime>.Default.Equals(this.StartOfWeek, other.StartOfWeek))
            return false;
        if (!EqualityComparer<string>.Default.Equals(this.Section, other.Section))
            return false;
        if (!EqualityComparer<string>.Default.Equals(this.Page, other.Page))
            return false;
        if (!EqualityComparer<string>.Default.Equals(this.Module, other.Module))
            return false;
        return true;
    }

    public override int GetHashCode()
    {
        var hash = 0;
        hash ^= EqualityComparer<DateTime>.Default.GetHashCode(this.StartOfWeek);
        hash ^= EqualityComparer<string>.Default.GetHashCode(this.Section);
        hash ^= EqualityComparer<string>.Default.GetHashCode(this.Page);
        hash ^= EqualityComparer<string>.Default.GetHashCode(this.Module);
        return hash;
    }

    public override string ToString()
    {
        var template = "{{ StartOfWeek = {0}, Section = {1}, Page = {2}, Module = {3} }}";
        return String.Format(template, this.StartOfWeek, this.Section, this.Page, this.Module);
    }
}

这个类的分组逻辑看起来像这样:

if (_page == "Total" || _page == "All")
    return _page;
return _entry.Page;

如果我误解了你的下拉值如何打开和关闭分组,那么你应该只需要改变这些方法,但是这个代码的关键是当分组打开时它应该返回一个基于值的组值。条目,否则它应返回所有条目的公共值。如果该值对于所有条目都是通用的,那么它在逻辑上只创建一个与完全不分组相同的组。

如果您有更多的下拉列表,那么您需要向EntryGrouper类添加更多属性。不要忘记将这些新属性添加到Equals&amp; GetHashCode方法也是如此。

因此,此逻辑表示您想要的动态分组。如果我有帮助或者你需要更多细节,请告诉我。

享受!

答案 2 :(得分:0)

我知道自从这个问题发布以来已经有一段时间了,但我最近不得不处理类似的问题(用户在运行时选择的多个列进行动态分组)所以这就是我对它的看法

  1. 用于创建分组lambda的辅助函数

    static Expression<Func<T, Object>> GetGroupBy<T>( string property )
    {
      var data = Expression.Parameter( typeof( T ), "data" );
      var dataProperty = Expression.PropertyOrField( data, property );
      var conversion = Expression.Convert( dataProperty, typeof( object ) );
      return Expression.Lambda<Func<T, Object>>( conversion, data );
    }
    
  2. 执行内存分组的功能。返回组。

    static IEnumerable<IEnumerable<T>> Group<T>( IEnumerable<T> ds, params Func<T, object>[] groupSelectors )
    {
      Func<IEnumerable<T>, Func<T, object>[], IEnumerable<IEnumerable<T>>> inner = null;
      inner = ( d, ss ) => {
        if ( null == ss || ss.Length == 0 ) {
          return new[] { d };
        } else {
          var s = ss.First();
          return d.GroupBy( s ).Select( g => inner( g.Select( x => x ), ss.Skip( 1 ).ToArray() ) ) .SelectMany( x => x );
        }
      };
      return inner( ds, groupSelectors );
    }
    
  3. 如何使用:

    String[] columnsSelectedByUser = ... // contains names of grouping columns selected by user
    var entries = ... // Force query execution i.e. fetch all data
    var groupBys = columnsSelectedByUser.Select( x => GetGroupBy( x ).Compile()).ToArray();
    var grouping = Group(entries, groupBys); // enumerable containing groups of entries
    
  4. 关于降低表现,我不认为这实际上是一个(大)问题。即使您动态构建了一个分组SQL,查询也必须返回与没有分组的查询相同的行数。因此,虽然在此方法中,分组不是由数据库完成的,但强制查询执行返回的行数与具有分组条件的假设SQL查询的行数相同。当然,数据库可能会胜过C#代码完成的内存分组,但流量只取决于必须对多少行(entries)进行分组。