Mysql 使用同一表中的数据更新行

编写一个sql的正确方法是什么？该sql将使用行总数更新所有

type=1

，其中parent=id为

type=1

简言之： update likesd set totals=所有总计的总和，其中父项=行的id，其中类型=1

"id"    "type"  "parent"    "country"   "totals"
"3"     "1"     "1"         "US"        "6"
"4"     "2"     "3"         "US"        "6"
"5"     "3"     "3"         "US"        "5"

期望的结果

"id"    "type"  "parent"    "country"   "totals"
"3"     "1"     "1"         "US"        "17" ->6+6+5=17
"4"     "2"     "3"         "US"        "6"
"5"     "3"     "3"         "US"        "5"

我尝试过（但失败了）

---

请尝试以下查询：

with update_cte(parent,totals)
as(select parent, sum(totals)totalsNew
FROM likesd where type=1 group by parent)
update a
set a.totals=b.totals
from likesd a join update_cte b on a.id=b.parent

---

编辑：根据，使用隐式临时表允许对嵌套SELECT语句中使用的相同表进行更新。

以下是执行所需操作的命令

update likesd as upTbl
        inner join
    (select 
        tbl.id, tbl.totals + sum(tbl2.totals) as totals
    from
        likesd tbl
    inner join likesd tbl2 ON tbl2.parent = tbl.id
    where
        tbl.type = 1
    group by tbl.id) as results ON upTbl.id = results.id 
set 
    upTbl.totals = results.totals;

---

在MySql 5.5上进行了测试，您可以使用以下描述的多表语法执行此操作：

查询将更新一行并生成以下结果：

+------+------+--------+---------+--------+
| id   | type | parent | country | totals |
+------+------+--------+---------+--------+
|    3 |    1 |      1 | US      |     17 |
|    4 |    2 |      3 | US      |      6 |
|    5 |    3 |      3 | US      |      5 |
+------+------+--------+---------+--------+

---

你在“失败”中得到了什么结果？我会说，保持原样，根据孩子/父母的关系，显示总数。这样，在某些地方就没有“计算”字段，而在其他地方就没有得到计算字段的数学。例如，将其留在表示层。@Elias。我在“where子句”中得到“Unknown column”b.parent“ID 3是否总是得到总数，或者如果说ID 5是父项，那么应该只得到父项为5的ID的总数吗？

Type=1

将得到总数。因此，如果有许多

type=1

，那么所有这些都需要使用各自子级的sums（）进行更新。MySQL还不支持CTE。这就给出了：

您不能在FROM子句中为update指定目标表'likesd'

如何？对不起，我的机器上只运行MS-SQL，所以我在猜测/阅读MySQL语义。

update likesd as upTbl
        inner join
    (select 
        tbl.id, tbl.totals + sum(tbl2.totals) as totals
    from
        likesd tbl
    inner join likesd tbl2 ON tbl2.parent = tbl.id
    where
        tbl.type = 1
    group by tbl.id) as results ON upTbl.id = results.id 
set 
    upTbl.totals = results.totals;

update likesd a, (select parent, sum(totals) as tsum
       from likesd group by parent) b
set    a.totals = a.totals + b.tsum
where  a.type = 1 and b.parent = a.id;

+------+------+--------+---------+--------+
| id   | type | parent | country | totals |
+------+------+--------+---------+--------+
|    3 |    1 |      1 | US      |     17 |
|    4 |    2 |      3 | US      |      6 |
|    5 |    3 |      3 | US      |      5 |
+------+------+--------+---------+--------+