如何使用单个 LINQ 查询将所有父对象选择到 DataContext 中？

Question

我正在寻找使用单个 SELECT 获取整个 LINQ 对象层次结构的特定问题的答案。

起初我试图使用 LoadOptions 填充尽可能多的 LINQ 对象，但 AFAIK 这种方法只允许使用 LoadWith 在一个查询中链接单个表。因此，我发明了一种解决方案，强制设置要获取列表的实体的所有父对象，尽管存在多个 SELECTS 进入数据库的问题 - 单个查询会导致在同一 LINQ 上下文中具有相同参数的两个 SELECTS .

对于这个问题，我已将此查询简化为流行的发票示例：

public static class Extensions
{
        public static IEnumerable<T> ForEach<T>(this IEnumerable<T> collection, Action<T> func)
        {
            foreach(var c in collection)
            {
                func(c);
            }
            return collection;
        }
}

public IEnumerable<Entry> GetResults(AppDataContext context, int CustomerId)
{
    return
    (
        from entry in context.Entries
        join invoice in context.Invoices on entry.EntryInvoiceId equals invoice.InvoiceId
        join period in context.Periods on invoice.InvoicePeriodId equals period.PeriodId
        // LEFT OUTER JOIN, store is not mandatory
        join store in context.Stores on entry.EntryStoreId equals store.StoreId into condStore
        from store in condStore.DefaultIfEmpty()
        where
            (invoice.InvoiceCustomerId = CustomerId)
        orderby entry.EntryPrice descending
        select new
        {
            Entry = entry,
            Invoice = invoice,
            Period = period,
            Store = store
        }
    ).ForEach(x =>
        {
            x.Entry.Invoice = Invoice;
            x.Invoice.Period = Period;
            x.Entry.Store = Store;
        }
    ).Select(x => x.Entry);
}

调用该函数并遍历结果集时，例如：

var entries = GetResults(this.Context);
int withoutStore = 0;
foreach(var k in entries)
{
    if(k.EntryStoreId  == null)
        withoutStore++;
}

对数据库的查询结果如下所示（获取单个结果）：

SELECT
    [t0].[EntryId], 
    [t0].[EntryInvoiceId], 
    [t0].[EntryStoreId],
    [t0].[EntryProductId],
    [t0].[EntryQuantity],
    [t0].[EntryPrice],
    [t1].[InvoiceId], 
    [t1].[InvoiceCustomerId],
    [t1].[InvoiceDate],
    [t1].[InvoicePeriodId],
    [t2].[PeriodId], 
    [t2].[PeriodName], 
    [t2].[PeriodDateFrom],
    [t4].[StoreId],
    [t4].[StoreName]
FROM
    [Entry] AS [t0]
    INNER JOIN [Invoice] AS [t1] ON [t0].[EntryInvoiceId] = [t1].[InvoiceId]
    INNER JOIN [Period] AS [t2] ON [t2].[PeriodId] = [t1].[InvoicePeriodId]
LEFT OUTER JOIN (
    SELECT 1 AS [test], [t3].[StoreId], [t3].[StoreName]
    FROM [Store] AS [t3]
    ) AS [t4] ON [t4].[StoreId] = ([t0].[EntryStoreId])
WHERE (([t1].[InvoiceCustomerId]) = @p0)
ORDER BY [t0].[InvoicePrice] DESC
-- @p0: Input Int (Size = 0; Prec = 0; Scale = 0) [186]
-- Context: SqlProvider(Sql2008) Model: AttributedMetaModel Build: 3.5.30729.1

SELECT
    [t0].[EntryId], 
    [t0].[EntryInvoiceId], 
    [t0].[EntryStoreId],
    [t0].[EntryProductId],
    [t0].[EntryQuantity],
    [t0].[EntryPrice],
    [t1].[InvoiceId], 
    [t1].[InvoiceCustomerId],
    [t1].[InvoiceDate],
    [t1].[InvoicePeriodId],
    [t2].[PeriodId], 
    [t2].[PeriodName], 
    [t2].[PeriodDateFrom],
    [t4].[StoreId],
    [t4].[StoreName]
FROM
    [Entry] AS [t0]
    INNER JOIN [Invoice] AS [t1] ON [t0].[EntryInvoiceId] = [t1].[InvoiceId]
    INNER JOIN [Period] AS [t2] ON [t2].[PeriodId] = [t1].[InvoicePeriodId]
LEFT OUTER JOIN (
    SELECT 1 AS [test], [t3].[StoreId], [t3].[StoreName]
    FROM [Store] AS [t3]
    ) AS [t4] ON [t4].[StoreId] = ([t0].[EntryStoreId])
WHERE (([t1].[InvoiceCustomerId]) = @p0)
ORDER BY [t0].[InvoicePrice] DESC
-- @p0: Input Int (Size = 0; Prec = 0; Scale = 0) [186]
-- Context: SqlProvider(Sql2008) Model: AttributedMetaModel Build: 3.5.30729.1

问题是为什么有两个查询，我怎样才能在没有这种 hack 的情况下获取 LINQ 对象？

Answer 1

为什么不多次拨打LoadWith？

在DataLoadOptions 文档中它说：

每次调用 LoadWith 都会检查循环是否有 [...]

（在避免循环的部分。）

Answer 2

嗯...我认为您正在遍历您的“可查询”两次：在 ForEach 扩展和“命令式” foreach () 块中... 您是否尝试过将 ForEach 的实现更改为...

    public static IEnumerable<T> ForEach<T>(this IEnumerable<T> collection, Action<T> func)
    {
        foreach (var c in collection)
        {
            func(c);
            yield return c;
        }
    }

Answer 3

对于寻求此问题的确切解决方案的其他人，请考虑遵循用于在单个 SELECT 中获取 LINQ 对象层次结构的代码的缩短和工作版本。我已将 GetResults 函数的返回类型更改为 IQueryable，因为 LINQ 更改跟踪机制可以正确跟踪集合，从而允许使用集合中的更改更新数据库。

public void InitContext(AppDataContext context)
{
    DataLoadOptions options = new DataLoadOptions();
    options.LoadWith<Entry>(x => x.Invoice);
    options.LoadWith<Entry>(x => x.Store);
    options.LoadWith<Invoice>(x => x.Period);
    context.DataLoadOptions = options;
}

public IQueryable<Entry> GetResults(AppDataContext context, int customerId)
{
    return
    (
        from entry in context.Entries
        join invoice in context.Invoices on entry.EntryInvoiceId equals invoice.InvoiceId
        join period in context.Periods on invoice.InvoicePeriodId equals period.PeriodId
        // LEFT OUTER JOIN, store is not mandatory
        join store in context.Stores on entry.EntryStoreId equals store.StoreId into condStore
        from store in condStore.DefaultIfEmpty()
        where
            (invoice.InvoiceCustomerId == customerId)
        orderby entry.EntryPrice descending
        select entry
    );
}