如何在带有子查询的列中“更新选择”某些值？

Question

我使用的是 Apache Derby DB (SQL) 版本 10.14，这是 Derby 参考手册：https://db.apache.org/derby/docs/10.14/ref/refderby.pdf

我正在尝试创建星型模式，目前正在处理区域维度表。

这些是我正在使用的表格： DWH_PRICE_PAID_RECORDS

TRANSACTION_ID                          PRICE   DATE_OF_TRANSFER    PROPERTY_TYPE   OLD_NEW     DURATION    TOWN_CITY               DISTRICT                COUNTY                  PPDCATEGORY_TYPE    RECORD_TYPE
{FDD12C8B-5A02-4B3F-8C67-9BC523DC780B}  71000   15.09.2000 00:00    D                   N       F           SCUNTHORPE              NORTH LINCOLNSHIRE      NORTH LINCOLNSHIRE      A                       A
{70F7F480-4A9A-4FEB-A58A-2B964605BFD2}  97000   01.08.2002 00:00    F                   N       L           MANCHESTER              SALFORD                 GREATER MANCHESTER      A                       A
{64D48FA9-8C85-49D6-AF5A-23FABDDB4FEB}  104000  17.01.2006 00:00    S                   N       L           ROCHDALE                ROCHDALE                GREATER MANCHESTER      A                       A
{F0316F65-E375-4DC4-BCDF-3FDC054ADE9C}  188500  18.05.2015 00:00    S                   N       F           KIDDERMINSTER           MALVERN HILLS           WORCESTERSHIRE          A                       A
{2EC5A85B-7BEF-4127-B3D0-6B416899CAEB}  180000  07.05.1999 00:00    S                   N       F           KINGSTON UPON THAMES    KINGSTON UPON THAMES    GREATER LONDON          A                       A
{21E5FEB7-A62E-2439-E050-A8C06205342E}  55000   28.08.2015 00:00    T                   N       F           MOUNTAIN ASH            RHONDDA CYNON TAFF      RHONDDA CYNON TAFF      B                       A
{3E0330F0-0F44-8D89-E050-A8C062052140}  77000   30.08.2016 00:00    T                   N       F           WALLASEY                WIRRAL                  MERSEYSIDE              A                       A
{D43A8B4A-6272-4706-9189-30F8E24EDF13}  210000  23.05.2007 00:00    S                   N       F           BRISTOL                 NORTH SOMERSET          NORTH SOMERSET          A                       A
{3575DAF5-0E80-408F-9970-FDF5D1475E73}  185000  16.11.2007 00:00    S                   N       F           CREWKERNE               SOUTH SOMERSET          SOMERSET                A                       A
{A4246390-61F4-4228-BC82-79D3F369CA34}  32700   12.12.1996 00:00    F                   N       L           SOUTHAMPTON             SOUTHAMPTON             SOUTHAMPTON             A                       A

CREATE TABLE "DWH_PRICE_PAID_RECORDS" ("TRANSACTION_ID" VARCHAR(50) NOT NULL, "PRICE" INTEGER, "DATE_OF_TRANSFER" DATE NOT NULL, "PROPERTY_TYPE" CHAR(1), "OLD_NEW" CHAR(1), "DURATION" CHAR(1), "TOWN_CITY" VARCHAR(50), "DISTRICT" VARCHAR(50), "COUNTY" VARCHAR(50), "PPDCATEGORY_TYPE" CHAR(1), "RECORD_TYPE" CHAR(1));

DWH_POSTCODES

Postcode    Eastings    Northings   Latitude    Longitude           Town            Region          UK_Region   Country_Code    Country_String
AB10        392900      804900      5,713,514   -211,731        Aberdeen            Aberdeen City   Scotland    SCT             Scotland
AB13        385600      801900      5,710,801   -223,776        Milltimber          Aberdeen City   Scotland    SCT             Scotland
AB14        383600      801100      5,710,076   -227,073        Peterculter         Aberdeen City   Scotland    SCT             Scotland
AB21        387900      813200      572,096     -220,033        Aberdeen Airport    Aberdeen City   Scotland    SCT             Scotland
AB22        392800      810700      5,718,724   -211,913        Bridge of Don       Aberdeen City   Scotland    SCT             Scotland
AB30        370900      772900      5,684,678   -247,712        Laurencekirk        Aberdeenshire   Scotland    SCT             Scotland
AB31        368100      798300      5,707,479   -252,623        Banchory            Aberdeenshire   Scotland    SCT             Scotland
AB32        380800      807200      5,715,545   -231,742        Westhill            Aberdeenshire   Scotland    SCT             Scotland
AB33        355200      815100      5,722,464   -274,203        Alford              Aberdeenshire   Scotland    SCT             Scotland
AB34        350800      800600      5,709,393   -281,204        Aboyne              Aberdeenshire   Scotland    SCT             Scotland


CREATE TABLE "DWH_POSTCODES" ("POSTCODE_ID" INTEGER generated always as identity (start with 1 increment by 1), "POSTCODE" VARCHAR(10), "EASTINGS" VARCHAR(50), "NORTHINGS" VARCHAR(50), "LATITUDE" VARCHAR(10), "LONGITUDE" VARCHAR(10), "TOWN" VARCHAR(50) NOT NULL, "REGION" VARCHAR(50), "UK_REGION" VARCHAR(50), "COUNTRY_CODE" VARCHAR(20), "COUNTRY_STRING" VARCHAR(20));

还有 DIM_REGION

CREATE TABLE "DIM_REGION" ("REGION_ID" INTEGER generated always as identity (start with 1 increment by 1), "TOWN" VARCHAR(30), "COUNTY" VARCHAR(30), "DISTRICT" VARCHAR(30), "LATITUDE" VARCHAR(10), "LONGITUDE" VARCHAR(10), "COUNTRY_STRING" VARCHAR(20));

首先我插入城镇（唯一）、县和区：

INSERT INTO DIM_REGION (TOWN, County, District) SELECT town_city, MAX(county), MAX(district) FROM DWH_PRICE_PAID_RECORDS GROUP BY town_city;

Town 必须是唯一的，因为我没有其他任何东西作为标识符来匹配/链接表格。 DIM_REGION 表现在有 938 个唯一的城镇记录，我想“更新”剩余的列“纬度”、“经度”和“国家字符串”。 DWH_POSTCODES 表有 1637 条独特的城镇记录，这意味着城镇是不同的！只有 532 个城镇是匹配的，并且只有那些应该在 DIM_REGION 表中更新。

我只从一列“纬度”开始，这是我希望它在 Update 语句之后的样子：

UPDATE DIM_REGION SET DIM_REGION.LATITUDE = (SELECT DWH_POSTCODES.LATITUDE from dim_region join dwh_postcodes on dim_region.town = dwh_postcodes.town where dim_region.town = dwh_postcodes.town);

但是当我这样做时，我收到了这条错误消息（错误消息是德语，我不得不翻译，抱歉）：

ERROR 21000 a scalar subquery returned more than one value

我不明白为什么。如果我只提交 select 语句：

SELECT DWH_POSTCODES.LATITUDE from dim_region join dwh_postcodes on dim_region.town = dwh_postcodes.town where dim_region.town = dwh_postcodes.town

我得到的正是我期望得到的，这就是所有纬度的整个专栏。

非常感谢！

Answer 1

IBM Db2 和Apache Derby DB 是完全不同的产品。因此，请删除 db2 标签。
您在两个表中都提供了完全不匹配的城镇列表。准备匹配的数据集有那么难吗？

如果您在 dwh_postcodes.town 列中没有重复项，则 update 应如下所示（可能与 WHERE 中的 EXISTS 谓词中的 subselect 相同）：

UPDATE DIM_REGION SET DIM_REGION.LATITUDE = 
(
SELECT 
--MAX (
DWH_POSTCODES.LATITUDE 
--)
from dwh_postcodes 
where dim_region.town = dwh_postcodes.town
-- fetch first 1 row only
)
/*
WHERE EXISTS 
(
SELECT 1
from dwh_postcodes 
where dim_region.town = dwh_postcodes.town
)
*/
;

否则，您必须使用一些聚合函数（或 fetch first 子句，如果 Derby 支持它）来使此 subselect 仅返回 1 行，如上面示例中注释掉的那样。

或者，您可以执行 1 INSERT 而不是 INSERT 和 UPDATE 来填充所有 4 列：

INSERT INTO DIM_REGION (TOWN, County, District, latitude) 
SELECT p.town_city, p.county, p.district, d.latitude
FROM (
SELECT town_city, MAX(county) county, MAX(district) district
FROM DWH_PRICE_PAID_RECORDS 
GROUP BY town_city
) p
join dwh_postcodes d on p.town_city = d.town;

Answer 2

当您需要更新多行时使用下面的查询。

UPDATE A SET A.LATITUDE = B.LATITUDE 
FROM DIM_REGION as A JOIN DWH_POSTCODES B  
ON A.TOWN = B.TOWN 
WHERE A.TOWN = B.TOWN

Answer 3

您遇到的错误是因为您的子查询返回了超过 1 条记录。您需要将子查询的结果限制为 1。否则，它将不起作用。

或者，您可以将子查询与 TOP 1

一起使用

UPDATE DIM_REGION SET DIM_REGION.LATITUDE = (SELECT TOP 1 DWH_POSTCODES.LATITUDE from dim_region join dwh_postcodes on dim_region.town = dwh_postcodes.town where dim_region.town = dwh_postcodes.town);

希望这会有所帮助。

Answer 4

@flippi

在一个块中执行多个 UPDATE 语句的一种方法：

     BEGIN
       UPDATE ...
       UPDATE ...
       UPDATE ...
       ...
     END;