我继承了一个设计不完全优化的数据库，我需要处理一些数据。让我对我必须做的事情做一个更常见的类比：
假设我们有一张Student桌子，一张StudentClass记录他参加过的所有课程的StudentTeacher表格，还有一张存储所有教过这个学生的老师的表格。是的，我知道这是一个愚蠢的设计，将教师存储在 Class table 上会更有意义——但这就是我们正在使用的。
我现在想清理数据，我想找到一个学生有老师但没有班级，或者有班级但没有老师的所有地方。SQL因此：
select *
from StudentClass sc
full outer join StudentTeacher st on st.StudentID = sc.StudentID
where st.id is null or sc.id is null
你如何在 Linq 中做到这一点？
我想我在这里有答案，虽然没有我希望的那么优雅，但它应该可以解决问题：
var studentIDs = StudentClasses.Select(sc => sc.StudentID)
.Union(StudentTeachers.Select(st => st.StudentID);
//.Distinct(); -- Distinct not necessary after Union
var q =
from id in studentIDs
join sc in StudentClasses on id equals sc.StudentID into jsc
from sc in jsc.DefaultIfEmpty()
join st in StudentTeachers on id equals st.StudentID into jst
from st in jst.DefaultIfEmpty()
where st == null ^ sc == null
select new { sc, st };
您可能可以将这两个语句合二为一，但我认为您会牺牲代码的清晰度。
扩展方法：
public static IEnumerable<TResult> FullOuterJoin<TOuter, TInner, TKey, TResult>(this IEnumerable<TOuter> outer, IEnumerable<TInner> inner, Func<TOuter,TKey> outerKeySelector, Func<TInner,TKey> innerKeySelector, Func<TOuter,TInner,TResult> resultSelector)
where TInner : class
where TOuter : class
{
var innerLookup = inner.ToLookup(innerKeySelector);
var outerLookup = outer.ToLookup(outerKeySelector);
var innerJoinItems = inner
.Where(innerItem => !outerLookup.Contains(innerKeySelector(innerItem)))
.Select(innerItem => resultSelector(null, innerItem));
return outer
.SelectMany(outerItem =>
{
var innerItems = innerLookup[outerKeySelector(outerItem)];
return innerItems.Any() ? innerItems : new TInner[] { null };
}, resultSelector)
.Concat(innerJoinItems);
}
测试：
[Test]
public void CanDoFullOuterJoin()
{
var list1 = new[] {"A", "B"};
var list2 = new[] { "B", "C" };
list1.FullOuterJoin(list2, x => x, x => x, (x1, x2) => (x1 ?? "") + (x2 ?? ""))
.ShouldCollectionEqual(new [] { "A", "BB", "C"} );
}
对于给定的 2 个集合a和b，所需的完全外连接可能如下：
a.Union(b).Except(a.Intersect(b));
如果 a 和 b 不是同一类型，则需要 2 个单独的左外连接：
var studentsWithoutTeachers =
from sc in studentClasses
join st in studentTeachers on sc.StudentId equals st.StudentId into g
from st in g.DefaultIfEmpty()
where st == null
select sc;
var teachersWithoutStudents =
from st in studentTeachers
join sc in studentClasses on st.StudentId equals sc.StudentId into g
from sc in g.DefaultIfEmpty()
where sc == null
select st;
这是使用 Concat() 的单行选项：
(from l in left
join r in right on l.Id equals r.Id into g
from r in g.DefaultIfEmpty()
where r == null
select new {l, r})
.Concat(
from r in right
join sc in left on r.Id equals sc.Id into g
from l in g.DefaultIfEmpty()
where l == null
select new {l, r});
一个开始...
var q = from sc in StudentClass
join st in StudentTeachers on sc.StudentID equals st.StudentID into g
from st in g.DefaultIfEmpty()
select new {StudentID = sc.StudentID, StudentIDParent = st == null ? "(no StudentTeacher)" : st.StudentID...........};
有关更多示例，另请参见http://www.linqpad.net/很好的工具
基于 Shaul 的回答，但有一点精简：
var q =
from id in studentIDs
join sc in StudentClasses on id equals sc.StudentID into jsc
join st in StudentTeachers on id equals st.StudentID into jst
where jst.Any() ^ jsc.Any() //exclusive OR, so one must be empty
//this will return the group with the student's teachers, and an empty group
// for the student's classes -
// or group of classes, and empty group of teachers
select new { classes = jsc, teachers = jst };
//or, if you know that the non-empty group will always have only one element:
select new { class = jsc.DefaultIfEmpty(), teacher = jst.DefaultIfEmpty() };
请注意，对于完全外部连接，这也可以工作。省略where子句并使用select上面的第一个，而不是第二个。