我正在尝试查询两个不同的表格（CALL_HISTORY和HUB_DIRECTORY），以查找“集线器商店”之间的所有通话记录。和一个＆＃39;辐射商店＆＃39;。每个调用都有一个CallID字段，并使用发起调用的商店的id创建一个条目，然后为接收调用的每个商店创建一个单独的条目，这些条目都具有该商店的id收到他们。所以它们都具有相同的CallID，但商店ID（DID）对于每个都不同。
问题在于并非所有呼叫都在集线器与其辐条之间，因此我需要将其过滤以仅查找这些记录。
RecordID | CallID | DID | CallDirection | StartTime
--------------------------------------------------------
1563486 | 255429 | 492 | Initiated | 1520870539
1563487 | 255429 | 849 | Received | 1520870539
1563484 | 255430 | 1098 | Initiated | 1520870562
1563485 | 255430 | 1098 | Received | 1520870562
1563482 | 255431 | 307 | Initiated | 1520870567
1563483 | 255431 | 1013 | Received | 1520870567
1563506 | 255432 | 1108 | Initiated | 1520870580
1563509 | 255432 | 1108 | Received | 1520870580
在这里，您可以看到一个呼叫示例，突出显示的CallID组位于集线器和其辐条之间，其余部分则不是。集线器和辐条在HUB_DIRECTORY中链接在一起，如此：
HubStore | HubDID | SpokeStore | SpokeDID
-----------------------------------------
4 | 37 | Store0004 | 37
4 | 37 | Store0522 | 470
7 | 1083 | Store0007 | 1083
7 | 1083 | Store1000 | 714
7 | 1083 | Store1055 | 759
12 | 38 | Store0012 | 38
12 | 38 | Store1063 | 758
13 | 45 | Store0013 | 45
13 | 45 | Store0337 | 296
13 | 45 | Store1012 | 724
HubDID和SpokeDID字段与DID中的CALL_HISTORY相同。因此，我希望查询DID表格中存在的已启动的来电HUB_DIRECTORY，以及HubDID或SpokeDID及CallID的来电。 1}}还有一个DID的记录，它与相应的hub / spoke相匹配。
我的最终目标如下：
HUB | Spoke | Initiated | Received
-----------------------------------------------
Store.0004 | Store.0522 | 304 | 723
我相信我需要使用UNION来获取带有轮毂或轮辐的行，但我无法理解如何完成此操作。
答案 0 :(得分：1)
我认为此查询会为您提供所需的结果。它适用于您提供的有限样本数据。
select h1.hubstore, h1.hubdid,
h1.spokestore, h1.spokedid,
count(distinct if(c2.recordid is null or c1.did!=h1.hubdid, null, c1.recordid)) as initiated,
count(distinct if(c2.did!=h1.hubdid, null, c2.recordid)) as received
from hub_directory h1
left join (select * from call_history where calldirection='Initiated') c1
on c1.did=h1.hubdid or c1.did=h1.spokedid
left join (select * from call_history where calldirection='Received') c2
on c2.callid = c1.callid and c2.did=if(c1.did=h1.hubdid, h1.spokedid, h1.hubdid)
group by h1.hubstore, h1.spokestore
根据您fiddle的新示例数据，此查询提供
hubstore spokestore initiated received
355 Store0355 0 0
355 Store0362 0 0
355 Store0655 0 0
357 Store0233 1 2
357 Store0357 0 0
360 Store0360 0 0
360 Store0868 0 0
360 Store1091 0 0
363 Store0363 0 0
363 Store1462 1 0
363 Store1507 1 0
363 Store2507 0 0
答案 1 :(得分：0)
截至目前我所理解的是，您希望获得的数据包括“通过集线器接收的呼叫数量”和“集线器发起的呼叫数”。
select
a.hubDID, a.CallDirection, count(a.CallDirection) ,
hb.SpokeDID, ch.CallDirection, count(ch.CallDirection)
from CALL_HISTORY ch inner join
(
select RecordID, CallID, DID, CallDirection, StartTime, h.HubStore , h.hubDID
from CALL_HISTORY c inner join HUB_DIRECTORY h on (c.DID = h.HubDID)
)
as a
on ch.CallID = a.CallID
and a.RecordID <> ch.RecordID
inner join HUB_DIRECTORY hb on hb.SpokeDID = ch.DID
在此尝试Demo
分组数据查询：
select
a.hubDID, a.CallDirection, count(a.CallDirection) ,
count(hb.SpokeDID),
ch.CallDirection, count(ch.CallDirection)
from CALL_HISTORY ch inner join
(
select distinct c.RecordID, c.CallID, c.DID, c.CallDirection, c.StartTime, c.DID as hubDID
from CALL_HISTORY c inner join HUB_DIRECTORY h on (c.DID = h.HubDID)
)
as a
on ch.CallID = a.CallID
and a.RecordID <> ch.RecordID
inner join HUB_DIRECTORY hb on hb.SpokeDID = ch.DID
group by a.hubDID, a.CallDirection,
ch.CallDirection
;
没有组数据查询：
select
a.hubDID, a.CallDirection, count(a.CallDirection) ,
hb.SpokeDID,hb.SpokeDID,
ch.CallDirection, count(ch.CallDirection)
from CALL_HISTORY ch inner join
(
select distinct c.RecordID, c.CallID, c.DID, c.CallDirection, c.StartTime, c.DID as hubDID
from CALL_HISTORY c inner join HUB_DIRECTORY h on (c.DID = h.HubDID)
)
as a
on ch.CallID = a.CallID
and a.RecordID <> ch.RecordID
inner join HUB_DIRECTORY hb on hb.SpokeDID = ch.DID
group by a.hubDID, a.CallDirection,
hb.SpokeDID,
ch.CallDirection
;